R. von Mises K. O. Friedrichs Applied Mathematical Sciences ° Fluid Dynamics 5 Springer-Verlag New York - Heidelberg - BerlinApplied Mathematical Sciences EDITORS Fritz John Joseph P. LaSalle Lawrence Sirovich Courant Institute of Division of Division of Mathematical Sciences Applied Mathematics Applied Mathematics New York University Brown University Brown University ‘New York, N.Y. 10003 Providence, R. 1. 02912 Providence, R. I. 02912 EDITORIAL STATEMENT ‘The mathematization of all sciences, the fading of traditional scientific boun- daries, the impact of computer technology, the growing importance of mathematical- computer modelling and the necessity of scientific planning all ereate the need both in education and research for books that are introductory to and abreast of these developments. ‘The purpose of this series is to provide such books, suitable for the user of ‘mathematics, the mathematician interested in applications, and the student scientist. In particular, this series will provide an outlet for material less formally presented and more anticipatory of needs than finished texts or monographs, yet of immediate interest because of the novelty of its treatment of an application or of mathematics being applied or lying elose to applications. ‘The aim of the series is, through rapid publication in an attractive but in- expensive format, to make material of current interest widely accessible. This implies the absence of excessive generality and abstraction, and unrealistic ideali- zation, but with quality of exposition as a goal. Many of the books will originate out of and will stimulate the development ‘of new undergraduate and graduate courses in the applications of mathematics. Some of the books will present introductions to new areas of research, new applications and act as signposts for new directions in the mathematical sciences. This series will often serve as an intermediate stage of the publication of material which, through exposure here, will be further developed and refined and appear later in the Mathe- ‘matics in Science Series of books in applied mathematics also published by Springer- Verlag and in the same spirit as this series. MANUSCRIPTS ‘The Editors welcome all inquiries regarding the submission of manuscripts for the series. Final preparation of all manuscripts will take place in the editorial ‘offices of the series in the Division of Applied Mathematics, Brown University, Prov- idence, Rhode Island. Published by SPRINGER-VERLAG NEW YORK INC., 175 Fifth Avenue, New York, N.Y. 10010. Printed in US.A.Applied Mathematical Sciences| Volume 5R. von Mises K. O. Friedrichs Fluid Dynamics With 216 Illustrations 5 | Springer-Verlag New York - Heidelberg - Berlin ® 1971Richard von Misest Kurt 0. Friedrichs Courant Institute of Mathematical Sciences New York University, New York All rights reserved, No part of this book may be translated or reproduced in any form without written permission from Springer-Verlag. © 1971 by Springer-Verlag New York Inc. Library of Congress Catalog Card Number 73-175242. Printed in the United States of America ISBN 0-387.90028-4 Springer-Verlag New York * Heidelberg * Berlin ISBN 3-540-90028-4 Springer-Verlag Berlin * Heidelberg * New YorkFREPACE In the summer of 19h Brom University undertook s Program of Advanced Instruction and Research in Mechanics, This in fact was the precursor to the present day Division of Applied Mathenstics, Certainly an outstanding feature of this program must have been the Lectures Sn Fluid Dysanics by Professor Friedrichs and the Inte Professor von Mises, Theis notes vere prepared in mimcosraph form and given a wide distribution et that tine, Sinee thelr appearance these lectures have had a strong influence on tesehing and research in the subject. Aas the reader soon learns the votes have lost none of their vitality over the years. Indeed in certain instanees only in the last fev years has the field caught up with the ideas developed in the course of these lectures, Many ideas of value are still to be found Sn these notes and the Editors are most happy to bbe able to include this volume in the series, ‘The corrections which have secunulated over the years have been incor= porated, end in addition an index has been added. With these exceptions all de- sire to revise bas been resisted. Also in this connection we are very grateful to Dr. TH. Chong for carefully overseeing the preparation of the present manue seripts ‘The Baitors August, 1971‘TABLE OF CONTENTS Introduction... ss a ‘CHAPTER I - GRUERAL THEORY OF PERFECT FIATDS CHAPTER IT Equations of motions... ee Bernoulli equation . . . Chrewlation 2 ee Helmholte' vortex theory. . « Relation between vorticity and Bernoulli function... Bquation for vortex-free mot: lon Monentun theorens for perfect fluid motion. . . = MOTTON IN| TWO DIMENSTONS-AIRWING OF TNFINITE SPAN Steady motion of an incompressible fluid. . . Steady, irrotational motion of an incompressible fluid... . fect of untform flow upon inmersed bodies . . Circular cross-section and theorem of Joukowsks Solution of the problem for @ single wing profile... . Exanplee of Airing sections. (a) the Joukowskd profiles to) Kbmbh-trettts profiles... . (c) Generai profiles... The direct problen-theory of thin wings... . Pinal venarke so... ee (a) Experimental verification of the Lift form: () Moment formula... (c) Compressibie flute CHAPTER TIT - MOTTON IN THREE DIMGUSTONS-ATRWING OF FINITE SPAN 1 Vortex Lines and vortex shee: Horseshoe vortex Lines and sheets . Lanchester-Prandtl wing theot 8. ey. vi 8 6 23 2% 8 u 32 ‘r n & a 98 ns cry ne wee aa 1534 5 6. t Ainding of minimm drags ee ee eee General problen 6. ee ee cer en Formal solution of the integral equation... 2. es Application of the theory to the biplane. so. se sees CHAPTER IV - THEORY OF VISCOUS FLUIDS 1 2. 3. a 5 6 1b 8 9% 10. nL zB B. a Be 6 Couette and Poiseuitle flow. . (a) Plane Couette flow... (0) Plane Potseuille flow. . Navier-Stokes equation... . Problems ss see eee ee Similarity. oc ee ee Saal Reynolds number... Unsteady flows eee Flow in convergent and divergent chamels .... 1.5 Flow tovards a plane plate. . . ‘The mathenatical structure of the boundary Laye Problem... se re Boundary Layer equations... « Flow along a flet plate... . Displacenent thickness - momentum equation... 2... ++. Jets and wakes... 2. ee von Mises equations»... . Curves walls and separation... Instability of vortex motion. . CHAPTER V - COMPRESSIBLE FLUIDS 2. 3. 4 Equations of notion for coupressible fluids... 4.4 ++ Introductory concepts of thermodynamics ©. 22. ee Steady flow in one-dimensional treatment. ©... +s oer Compression shocks... +s + Non-steady one-dimensional flow 161 xe 8 awe asi aos 196 199 209 ag 226, 226 236 2 aug, 253 26 266 err 280 285 258 300 303 sor 3B6. Pwo-dinens: 7. Nearly con: onal, steady, adiabatic flow. . stant, parallel flv... 2. 8, Flow in and around corners, Oblique shocke PROBLEMS Problem 1. Problem 2. 9. Won-sdisbatic flow and boundary layer... « Problen 38,556. oe ee eee eee ee Problem 7. = Problem 8,9,10. Problen 2. « Pte 2 et ete 50 Problen 13... Problem 1h,15 . . Problem Boe ee cee eee eee tenes Problem 17,18 . « Problen 29,20... Problem he eee eee ee Problen 22. Problem 23,24... erties ooo ee eee . Problen 26.0... 54 g00000c0000 a Problem 27.0 se ee eer 5 oer Problen 28... Problem 29... + Problem 30. « Problem 3L.. . ax #6 3 En 7 113 we 338 199 201 208 222 226 283 2h 260 266 270 an 280 207 su 337TNTRODUCTION Fluld dynamics, including mechanics of Liquide and of gases, is a part of phyeice. We can divide the processes of any physical theory into four steps! Placing observed data into general physical lave,, assuming the uniform ity of nature, 2, Transferring the physical lave into mathematical form, thus obtaining @ Logically consietent eystem of axicne which ueually coneiste of a ayeten of difver- ential equations 3. Drawing conelusione from these differential equations. Thie ie a purely mathematical step. 4, Check on the resulte obtained in (3) by experiments. It 1s often ascuned that the firet two eteps are ccmpleted for all branches of mechanics and that, therefore, mechanics at present offers only mathematical aifviculties, This, however, is not so for certain facts in fluid dynamics (and elsewhere). The first two steps above have been carried out, se far as fluide are con cerned, in tvo forms, First a theory of "perfect fluids" has been developed, and then it was supplemented by a theory of "viscous fluids". However, these theories do not explain all the know facts in fluid dynamics. They cover the explanation of so-called laminar floy. Here the particles travel along snocth curves, for example the slow motion of a fluid through a pipe of anall cross-section, or the Jet emanat- jing frm the opening of @ tank. On the other hand, motion in larger channels, Like a river, may lock smooth as vhole, but each particle by iteelf has a complicated irregular motion. Tt fluctuates violently and may exhibit chaotic behavior. Such @ motion 1 called turbulent. We may say that the first two steps of the physical investigation have been completed for laninar motion; the theory of perfect fluide was given by Euler about 1760, that for viscous fluide by Navier and by Stokes about 1850. In both cases, wehave @ definite system of partial differential equations. Now the third step con sists of the solution of certain boundary value problems of these partial differentisl equations. This mathematical investigation is still going on. However, in the case of turbulent motion hardly the first step has been completed. Sone general rules have been derived, but at present there ie no complete and consistent set of axioms (principal equations) which would explain ali particular problems of turbulent, motion. In these lectures, we first consider the theory of perfect fluids, as this is the simplest case and the basis for sli further developments. A grest desl of practical reaulte can be obtained within thie theory and ite ueetulnese ie broader ‘than usually assumed, It is not correct to ssy thet there is no resistance to a body moving through a fluid when considered as @ perfect fluid, Thus, we can derive, in ‘the case of an airwing, a ligt force and a drag even if we aseue 9 perfect fluié. Hence the results are better than would be expected and the simpler assumptions make it worthwhile studying.(CHAPTER 1 GENERAL THEORY OF PERFECT FLATDS 1, Equations of Motion We consider an element of volume anywhere ina fluid, Let aA be an ele. ent of area of the surface of this elenent of volume. tet aF be the force exerted upon 4A by the surrounding particles. It is the essentisl hypothesis by which the perfect fluid is defined that this force is alvays acting normal to aA. Then it follows by mathenstical reasoning, which we omit here, that the value of F/AA ts the cane for all clemente of surface passing through a given point. If ve consider two elements aA and aA! belonging to the boundsry surfaces of two different ele- mente of volume, but both having a point P in comon, then the differential quotients aP/@A and aF'/aA' have the cane value. This quotient (of dimension ¥Lr*) te called the hydraulic pressure p. We shall have to coneider p ara funetion of x, y, 2, and eventually of the time, t, but p will not depend upon ‘the orientation of the surface element aA. oF oF" Jn the case of viscous fluids the foree AF fe no longer nomal to aA. In this case, we adit @ tangential component and then the relation between the amount of these normal and tangential canponents and the orientation of the elenent aA vecones complicated. However, in the state of equilibrium, (hydrostatics) we have sn no case to consider shearing stresses. Hence the distinction between per~ fect and viscous fluids vanishes here, In problems of equilibriun the hydraulic pressure Se fimetion of position alone. In order to set up equations of motion for the perfect fluid we have toapply Newton's second 1 of motion and to combine it with the concept of hydraulic preceure. In the mechanice of ensll rigid bodies or particles the second law states ‘that at any instant the acceleration of the body is given by gun of ores: acceleration here Fy, Fyy.s+ are the external forces acting upon the body and m is a positive constant, the mass of the body. uch a material body of mal dinensions has just one velocity and accelers- ton. Bit in « fluid ve may have at each point » different velocity and acceleration ‘Tous the second Law needs a certain adaptation to this cate, Forces umuslly acting upon a fluid element are 1) gravity and 2) the force resulting from the preeeures wich act upon the surface lenents of the elenent of volune under consideration. (for viscous fluids we have also to include shearing forces, but at present we have only to do with the weight force and pressure forces.) Denote the velocity vector of a point P in the fluid at tinet ty @ whose components dg) Gy, dy He shell denote by u,v, w respectively, Then at slightly later tine t+ dt the point P will be at FP! and will have the velocity To ajar at, me derivative ci/at 1s cated “naterial” derivative as tt refers to a determined material point and includes the change in Z due to change in position as well as to change tn time. d@/at 1s then the acceleration at P at tine &. a4 or G+ Ge at‘The fom of Newton's second law as adapted to the case of Fluid mechanics would be gum of fore scceteration = etn oh vol. +0 The Limit is understood to mean that the dimensions of the element of volume, that: includes the point F, are more and more reduced to zero, Assuming thst the Linit of the sum in our expression is the sum of the Limite and the Limit of the product in the following is the product of the Limits, we write hove SB aie Be BL Wow, for any body, we know that Z, the acceleration of gravity, sich te « vector directed vertically downwards and of constant sacunt shout 72 2%./ seo?. he Linit of the quotient BME is called the denelty or apecitic mass of the fluid at the point P and will be denoted by p. Tn general, p Le not & constant, dim Bressure forces eu we may assume here that the Limit In computing vol. +0 exists, no matter how the snall volume shout P is reduced to PL (the existence of the Limit, however, can be proved mathematically.) We take the smell volune about P in the form of right cylinder of base 4A and Length ax and let the generators of the cylinder be parallel to the xaxis, Consider the x component of the pressure forees. The pressures on the conve portion of the cylinder are normal to the x-axis and contribute nothing. ‘he pressures on the two bases are show in ‘the figure, and give the total preseure force pan - (p+ BB axan = - 2 ax an.2 ot oe Since the volume of the cylinder is dx dA the limit of the x component of the pressure forces ratio Ds ts - 2. simtiarty, by taking the generators of the cylinder parallel to the y- and z-axes, the y and 2 comonents are found to be ~2/2% and -B/3e respectively. But 2/3, ®/¥, d/ 2 are the componente of a vector celled the gradient of p and written as grad p. Substituting these reults back into the equation for d@/at we have Euler's equation of motion Hee 2-5 ema. wo A few words have to be said about p and p We cali incompressible fluid fluid in valch p can be considered as con stant, In reel Liquids (water, ofl, mercury, ete.) p is not exactly constant but. almost 80. We call compressible fluid a fluid in which p cannot be treated az con stant. In ell mechanical problems ve shell assume that p is @ function of p. In General we should use the equation of stste for the fluid vhich is an equation of tne fom F(p,0,2) = 0, where T te the temperature, (p is often replaced ty the specific volume" v= 1/pg). As long as ve remain within the domain of mechanics, we may either take T 9 constant, isothermsl case, or elee introduce another Ffote thet thie definition of epectfic volume differs from the customary definition of volume per unit mse,relation between p,p and T which allows ue to eliminate 7, e.g., adiabatic eae. In elther ease p becomes @ function of p_slone. In general, Liquide Like vater, ete. will be considered at incompressible and gnses Like air, etc. as compressible, fut there are exceptions. For instance in the study of the vibrations of a column of vater ina long pipe Line the effect of change In p due to the compressibility te not negligible. On the other hand in the study of airfoils moving through air over a wide range of speeds the sir can bbe aseuned to have a constant p without any great error. We are now going to study sone transformations of Equation (1). Consider first the vector @. We can write it as ~ered (eh) vote componente re -e Bt, eB, og Bt store in the height of the potnt P above an arbitrary Horizontal plane, messured positive vnward. In fast, a8 shoxn fn the figure we have @p/& = cos(x,h) and therefore -g 3 ie the ccaponent of the acceleration of gravity along the maxis If the x and yeaxes are horizonts}, B/& = R/O and B/%= 1 ae expected, h ZA ZA Je As p 1s assumed to be @ function of p, we may consider the integralYa fy FO) ® here Pip) is defined except for an additive constant, For the incompressible case, p= const, and Ap) = p/p while in the isothermal case for a so-called per- fect gas with pv = RT as equation of state, we draw the conclusion that L wo RT Pg where the denominstor is a constant. ‘Therefore, p= <2 p and Po R Pe) = 2 206 ». 0 Now £t follows from (2) that grad Pz) = E ered p. 2) 11 show this, we differentiate (2) with respect to x and find 2.2 df ae Le mR! pew! ay ee The sane ie true for the derivatives with reepect to y and 2 Using the above given expression for @ and the now found expression for Vo erad p, we can write Buler's equation (1) in the form $B = -eraa(en) - ered P= -eraa(enF) = -6 wraan s Z). e) E is carted the pressure head. rt ie equal to By for incompressible lvids and equal to a constant tines log p for the Seathermal case of a ecnprese thle fluid, Bquation (3) thue gives the acceleration at P equal to thenegative of the acceleration of grevity mitiplied by the dimensionless quantity, gradient of heteht plus pressure head. ‘According to Equation (3) the vector field of acceleration is the gradient of a potential tela or ens 9 ~ -potentiaa or B. Tn the case of equilitriu vaere Y= 0 or at mort Z= constant, we have gh + A= constant. Te is to be noticed that Baler's equations are tree So number with four dependent variables u,v, ¥ and p (at p is @ given function of p). Thue we need @ fourth equation fn onder that these four quantities can be detersined. We roceed to do is by using the feet thet So Hewton's seoond Tew the mass remaine constant for any deeinite Doly, ‘Thus, 4¢ xe consider # portion of the flats arbi trarily cut out fron the whole, the mass of which fe J p aV, we have to state that this value does not change throughout the motion et a be an clanent of the Youndery surface of the portion under cone sideration, tet Z be the velocity vector at a point of ak and ® the normal to Gh, After tine ot nts etenent AML be aiaplaced to 4 ney porttion ssa the voli arent out ty ake siven by ah q coa(nya)st. since the mass is given ty fp aV, our statement reads fefoaveo. Carrying out the differentiation we have to coneider the change of pin time and the shove mentionedshifting of the boundary. Thus we get 2 SF Bavs J oq con(nyayaa = 0 &) whore the firet integral is a volume integral accounting for the change in p at each point, and the second 1s a surface integral taking into account the change of ‘the boundary. ow using Gauss’ formule relating volume end surface integrals, J Kava fe cosoyx}ah, and coidering that a cor(e) = u coals) + ww v cos(n,y) + w cos(n,2), ve got t [xed Men. ona, fy OOM” flor te expression 2G), 30), 120 i xnom as the atvorgence of i and written div(e@). Thue (ls) takes the form J e+ aiv pfJav = 0, and this must hold for any volsme. In particular, for volimes closing dom on a point P we eee that finally the expression under the sign of integration must vanieh. Hence ab every point P we have aiv(e@) ae 2. ©) This is usually called the continuity equation. By calculating div(ed) we get another form of this equation! powien Bev Bev Be® Ie The last four terms of thie equation form the "material derivative of p. For if £ de any Amotion of the particle, then by the so-called Euler rule for differ. entiation we get the material derivative aoeee Ly tere ge Te RT aa Applying this to the preceding we get the continuity equation in the form patv d+ 2-0, or atv d= - Slog p. (5a) We note that Af the flutd 1s incompressible then p = constant, and the equation becames div T= 0 2, Bernoulli Equation. let P bea point of the fluid and Z the velocity vector, The direction of thie veetor is independent of the coordinate eystem; we will call de the element of Length in thie direction. The x-component of Biler's equation is au 3 2) B.-L a+, or using Euler's rule for differentiation bu BraB- ch O+ BD, ead of the x-direction. Then ve Wow we get up the equation for the s-direction i have, since is the component of Zin the sqairection 2 By ghee aZ Hage Har D, awhich becomes ©) Tt is used mostly in so-called "steady" or This 8 know as the Bernoulli equation “pemanent” aotion. We eay that a notion is steady (permanent, stationary) if all the partial derivatives with respect to tine vanieh. Hence for steady motion da $= 0 ana (6) becomes (6a) For steady motion we have definite streamlines, t.e., curves which have two proper- tes! they are the pathways of the particles and they are curves such that the tangent at any point S2 in the sane direction as the velocity vector at that point. In the case of unsteady motion, hovever, the curves which are the paths of the articles do not coincide with the curves which have at any soment the property that ‘the tangent at any point is in the sane direction as the velocity vector at that point. It follows from (6a) that in stesdy motion we have Bene Z Hrn+Z const, o alongs streamline, We cali $5 the velocity head, hen this Last equation states that for steady motion: the sum of the velocity head, the gemetrical hetght, and pressure head is @ constant along a streamline. This equation must not be confused 2 vith the energy equation. While it is true that, £ is a kinetic energy and h is a measure of potential energy, there existe no third kind of "preseure energy".3. Clroutation, et AB be a curve in the fluid, P a point on the curve, Z the velocity vector at P, and 42 be an elonent of are along the curve, Then ve can take the oN , Line integral of the . velocity vector along 4B A e ) O, {acos(a,t)ar= fe - af «@ w Jo particular, if we have a closed curve or @ cireuit ve may consider the integral, from A to A, L.e., over the whole cfreuit, Then ve write it in the fom: 4 We are going to prove the following: Theorem. The value of the Line integral of q for a closed curve doce not change during the motion if the curve suuaye compored of the sane particles: &§t-d-o ® vere as vetere 4 te the anteriak derivative, thts statenent te equivalent 0 ‘the Buler equstion. Proof! ‘the change inf Z- df is due first to the change in Z for each yoint ‘and then to the change of the curve involved. We have Ba = mf § 8. a. g7- Be AL (a+ Satnat lot Here di' 4s the new element of arc into which d? ie transformed. LT asty From the figure it ie apparent that = Paced ods @s Bane. Hence dit. of = Boar at, sunetatuting above: fF. a? ats 2. Bary, ana appiying mater'e equation, fiz J+ Bas, and apptying quation, $ t-eras(emP) «d+ BG Pyar. Perey ‘The scalar product in the firet term on the right hand eide above equale the product of af by the soomponent of the gradient, Thus we find age. ge 2 aae af wet B= $F HM od gee = of E- 2 Da. a @) This integral, if it vere taken fron A to B wouldbe Gn Zl ) Ae A and B coincide in the case of a closed circuit, and since $e -h- 2 ie clearly single-valued, the integral around a closed curve ie zero, Q.E-D. Tt 4s underetood that this theorem holde true within a region where @ te ‘a continuous function of x, y, 2. In the case of eo-called discontinuous surfaces ‘we have to examine each particular situation. uwThis theorem of Lord Kelvin has @ converse which states that if the in- egret re ft d otion, then we bave what essentiaLiyBaler'e equation. 70 show this, eupoce ve bare a given vector field Hxjy2y8) and asome that &E= 0. for enh cloeed curve in the field. Mow we found that sere the seco tera under the tntgrat stn, ¢- Afar = 3 Ghar, te a eoten austereatial 19 f. Hence the intesral ef tne ancond tema ta sero, For the Se. a tegral of the first tema to vanish also, the component of $4 along the direction ta (8 Be eee eee ats (B),ae. moe the etretion ot aT te arsttrary, so ste component ot the acceleration in any direction 7 mist be a differential quotient with respect to 1, thts te tive only Af the acceleration te the gradient of a function V ot = eres v, mis se cosontianty Boler's equation for al ve have to do te to faentity -v vith ahs F For an ineompressible fluid the continuity equation ie simply div 7 = 0, Tats equation does not depend on p, Thus a velocity field satisfying the to eon astions $0 tor cach curve, aiv T= 0 © gives in any case a physically possible perfect flutd action, The corresponding »pressure distribution can be deduced from the Bernoulli equation (6). We can say, thet in the case of incompresstbility, (9) follows from (1) and (5) by eliminating P, Able f1uld we have the function e of p in the In the case of compre continuity equation. The cfreulation theorem states the extstence of a Ainction -¥, vinich we have to ddentity with gh + @ in this way we find the value of p and, therefore, of at any point and we have to check whether the velocity tunetion @ ‘then satisfies the continuity equation 4, Helmhotts' Vortex Theory. Given any vector field ¥ we can derive a related vector field by taking ‘88 components of the new vector such a vector Ss called the curt of ¥, We uit the proof thet this Ss, sn fact, © vector, i.e., that after a transformation of axes the new componente are given in ‘the seme form in terme of the new componente of #, For the vector field @ we set @= curl @ with components aon d,- ez po. (om, = BB (om D, - = 2 The vector @ ie called the vortex vector. The resson for thus calling @ = silee dz and dy, Aa menor fea A fn tbe x thrctte along m2, BE glee the rate of increase in the y component, ¥, of the velocity. As we move from A 6woble the second {nerease means a rotation of AD in the negative sense, Then ay ae 1 0, = 32 - $B ts the cum of the two rotations and 3 u, vill be a mean rotation z or the rotation of the bisector AP of the angle between AS and AD. into the bin rotation of the eleneat of volume. y However, a fluid notion vith the vatue of @ veing nonzero may Look quite akfferent than what we think of y ordinarily in speaking of "vortex". Consider for example —__1 s+ the flow in streamlines parallel |» au a, = - <0 gives a mean rotation eo Oy of the particles in the negative sense fas shown by the rotation of the bisector at PL into the bisector at 7. Thus thie straight Line flow is a vortex motion uscoarding to our definition The circulation T for any closed curve ¢ equals the sim of the cireula- tons for any network of meshes {nto which ¢ 4a divided up. For, §£ ve draw an ane joining to dis. A € tinct pointe Band E of Cas shown in the figure, ve have TP > as the ctreulation for the closed curve ABEA and Tas the etreulation for the closed curve DEB, and we find that T= 1, + Tp while f G.aP=-[ ak en) (ex) In the sone way Af we break up sry surface A of which C ie the boundary Snto a large mumber of small. elements we can write Tas the oum of the ofr culations for each element of the surface. Ic Finsliy, passing to the Limit as the number of such elements beccaes infinite and the dimensions of each elenent beccmes zero we have ref ar. @) BJo find af we consider an clement of area in the shape of a rectangle and take ite aides parallel to the x and y-axes. The sides then are ax and dy as shown in the figure. For this rectangle ar bt a vars (ve Banay- (oe Madar -vay y tere a, 1s the component of @ normal to the surface ele- ~ ment aA. For an arbitrary dy aa surface A stretched scross Cae the curve © this expression x for aP will then read ars o,ah vere @, te the component of @ normal to ah. If we let 7 te a unit veetor normal to A sn pointing in the direction in ubien « right banded screw would move when turning én the direction tn which we take the integral TP slong ¢, we have o, = +e, and using a= Tak we can write T-f @-f waa-f B- ax. Qo) ‘This equation is known as the theorem of Stokes. It gives a formal relation between the values J of a vector field and the deduced field of curl 7 = 3, At each point of the fluid we have the vector @ as well ae the vector Z. ‘A Line such that st each point in the fluid ite tangent Line coincides with the @ vector at thet point is called vortex Line, All vortex Lines through the pointe of a closed curve ¢, give us a vortex tube. the vortex Lines form the wall of agthe tube and, if ¢, Ss another eurve through all points of which the same Lines pass, any two surfaces A, and A; of which C) and Cy respectively are the voundaries may be considered as cross sections of the tube. Ce ey ‘Take a curve C which Lies completely on the wall of the vortex tube. For Uhte curve we have T= 0 according to (10) ae @, = 0 at each point of the wall ‘At sone Inter instant our vortex tube will be st @ new position, The particles which formed the curve C will now fom a new curve Lying on the new position of ‘the wall, By Kelvin's theorem (8) the cireulation has not changed, thus we have agein © for the curve into which C is transformed. Since © was any curve on the wall of the original tube, we have T= 0 for any curve on the nev surface into which the wall he changed. ‘This must hold for each infinitely small circuit too, therefore, ve have aP= 0 or @,=0 at any point of the new eurface. Tt follows that the new surface is agsin the wall of a vortex tube. In other vords, fluid particles which once form a vortex tube do so permanently. Any vortex Line fan be contidered as the Limit of a vortex tube of emall cross eection. After any interval of tine thie vortex Line will thas still be a vortex Line, 80 we have the firet theorem of Helholtz which states! Vortex Lines are material Lines; they a vortex tube remains consist permanently of the same particles. In other words, always a vortex tube. ext, consider the cireulation for any cross sectional curve of the tube, such ae or Cy in the leet figure, Since a, = 0 over the well of the tube 20connecting ¢, and C, we have by Stoke'e theorem that the circulation Ty for ©, equals the circulation 7, for Cy. In fact, we can choose for computing Ty, 1 according to (10), as our surface A either the cross section Ay or the wall of ‘the tube together with the crose section fy. In the eecond case the integral re- duces to the integral over A, since a, =O on the wall. Ae, therefore, J @,0k 1e the same value for each croge section of the tube we may call thie value the vorticity of the vortex tube under consideration, By the way, this expresses the fact that div @= div curl @= 0 in all cases. Wow we apply again Kelvin'e theorem (8) to the TP value of a cross section- al curve like ¢, or C,. Bach euch curve becomes a cross sectionsl curve of the now vortex tube. The theorem etates that the T value remaine unchanged and thus ve have the second theoren of Helmholtz which etstes: The vorticity of any vortex tube ts unchaged during the mottor Both of the theorems of Helmholtz which have been proven here hold for steady or unsteady motion and for compressible or incompressible fluide. ‘The only condition which was used in the proof as far ae compressible fluids are concerned is tht _p and p are connected by a single relation. In certain meteorological problems where this assumption dose not hold the Helmholtz theory has to be replaced by @ nore genersl one: Conversely, from the two theorems of Helmholta ve can deduce the theorem of Kelvin which states that the efrculation te constant during the motion, Tet ¢ be 8 closed curve. Through each point of © will pase s vortex Line, Tous this curve C can be considered in general os a ences sectional curve of a vortex tube, By Helmholts's Pret theorem, at a later ine stant thie vortex tube will be traneformed into a vortex tube aBy Helmholtz’ s second theorem, the vorticity, Ja, - aA (=P by Stokes theorea) 4is the same for the new tube as for the original one, Hence the circulation for ¢ 4s not changed. The only eae ve still have to examine is the one in which all vortex Lines through one point of C pass through another point of ¢. In this case the surface bounded by ¢ vill be part of a wall of a vortex tube and the cfroulation here will be zero. the surface will be transformed into a new surface which ie, by virtue of the firet Helmholtz theorem, part of the wall of # new vortex tube, and, by virtue of the second Helmholts theorem, the circulation will be zero here again. Thus in this case too Kelvin's theorem follows AS ve showed that Kelvin's theorem with the continuity equation te an equivalent of the Euler theory of perfect Fluide, we sce that Helmholts’s to theorens together with the continuity equation also form a complete bacle for the investigation of perfect fluid dymamice. If, in particular, we have an incompress- Ile fluld, then the continuity equation becoses div = 0, and a velocity fie1a satietying thie condition and the two theorems of Heluholte is physically porelble, A particular cage {2 the one where the velocity field te such that @= 0. for ell 4%, ¥, 2 ard t. In this case, for any closed curve witch can be made to be the voundary of @ surface of Fluid the circulation 1s zero, However, if ve have a sold cylinder of infinite length semersed in the fluid, (so as to leave a doubly connected space for the fluid), and a closed curve ¢ around this cylinder then the circulation must not vanish for ¢. If ve consider the boundary of a fluid surface to consist of two |G. c closed curves ¢ and ¢! around the cylinder, then 1¢ follows from a = 0 ‘that the circulation around the complete boundary 8 ero (where the Integration is understood to be taken in the sense shown in the figure), It meane thst in this case ali circuits surrounding the inmersed 22cylinder, and taken in the same sense, have the sane circulstion value, Prom the preceeding we draw the conclusion that a velocity distribution with div ¢+0 and eireuiation values equal for sil eiresite surrounding the same body 42 a physically possible motion of an incompressible fluid. 5. Relation Between Vorticity and Bernoulli Function: Coneider the Bernoulli function (ay We showed thet for steady motion H= const, along a streamline. Now we wish to ‘examine the general behavior of H. For thie purpose we teke the gradient of i 2) 2 gered t= gran) + eraifn + Z Applying the equstion of motion (1), 2 ae gered x= rea) - 2 eared = eraa( ava FZ = eraal BeBe Mee BR Re ERR RR a UR ae a ay ds =~ 3° E> et Se 23Te expression va, - vai, ie well known as the x-component of the vector which is ‘the vector product of Y and @. For vector (or cross) product we will write (2x3). row repeating the cane procest for the y-couyonente end z-camponente ve get @xd. 03) grad H lay In the case of eteady motion the partial derivatives with respect to tine ere zero and (15) becomes gerd H= @ xa. (sy Yow 4¢ te well tnove that the vector obtained ax the vector protuct of two vectors te vormai to the plane formed by these two vectors, and hence (18) states that grad H at each point is normal to the plane formed by Z and @ at that point. Stnce tn general the gradient of a function te dizected eo as to be normal to the surface weere thls function Le constant, ve have that te constant slong the vortex Lines as veli as along the streanlinee, In fact, for eteady notion ve can fed tn ete vay a new proof of Helaholts's first Uheoren. If we consider the streenlines through vortex Line, gra 1s normal to the surface formed ty these streaslines, and Hin constant on this surface, Now Lf we consider the vorter Lines through the original stremline we see that they must Lie on the sane aurtece Hence the origiot) vortex Line noven along the streamlines on thle surface, tn steady notion, if J= 0 everyshere, then by (1K), ered N= 0. Tate neant i= const, everywhere throughout the fluld, thus in etesdy motion with ©, H hae the gone value along all streamlines, Now conversely, if H hee the tone valve along all etreeniines, then ve cannot eay definitely that Zs 0 every- vere. Hovever, we can any that elther dso or @ te peraiiel to Zot each point of the fluid. thle last case certatnly cannot happen tn two disenelons) flow, since here @ ie always normal to @ by definition. 2hvortex lines woays sau 6. Byuation for Vortex-Free Motion A motion is called yortexeree (ierotational) if B= 0 at alt pointe of the fluid. Ih this case a, = 0, 0 and we have yO ae = ° (as) ow if there exiets a function © euch that u-Z, (268) then (15) wilt autonaticaliy be satisfied. Also by considering 0, vey ely we get (6) aly Hore (és) and (16b) state that Z- grad @, wich may be changed to the statement = pot {mvs we sce thet the statement that « motion is yortex.sree Se equivalent to each of the following equations curl @=0; q= grade; @ = pot 7. ae) 25we assume a function @ of x, ¥, 2, snd taxe = grad 6, then we have a vortex.free velocity field, In onder to have a vortex-fhee velocity field satiafying all. plysieel consitions we only have to insure, in adaition, that @ ered 9 Sse he continuity equation, Here we have to desl separately with compressible and inconpressible fluids. Q) Ine eesible Fluid, In this case the continuity equation becomes ? may, be ? div T= 0. once B+ He Hao, and since Fx grado 7) fe, Se, Fo e a x ¥ Equation (17) is know as the Laplace equation. If ve take the integral of the Laplace equation which eatisfice the particular boundary conditions of a problem, we have the problem solved. Equation (17) may also be written in the form atv grad 0 We note that (17) does not involve the time tj the dependency upon t te only brought in by the boundary conditions. (2) Compressible Fluid. This case is more complicated and for the present we restrict ourselves to steady motion. Also we will neglect the influenc ot erovity, tees, th the Bernoutlt equation ve vill anit the grad h (tske = 0. or constent). Gur justification for this 1s thet in sero-dynanicel probleas the velght of the fold partioies in mort cases 18 of no tnportance. How, since we are considering vortex-tree notion we have = grad @, Or, voat is equivalent = 0, and this implicn = const. Hence 2 Fs oe et const, ro) 26Wow, since P= { 2B, and p is a function of p, (18) gives us s relation be- tween p and q. Taking the derivative in (18), we have ao, aaq+aP=0, ain+ B= 0, Bx -pa. as) Since our function @ must also satisfy the continuity equation, ve consider the Latter, which for compressible fluids and steady motion states atr(sa) = 0. By appealing to the definition of divergence it can easily be shown that this lest equation can be transformed to oaivds a8 ow for vortex-tree motion div P= af, hence ae = at (20) Mao since p isa function of p, and p a function of q as show shove, ve have, by the rule for implicit aifferentistion Bw op 2, 3s Gp aq Os If we introduce for brevity ae (of dimension 4 (21) P eTand substitute thie and (19) in the Last equation ve get. B- bm - 2. The expreesion (20) for Ad now becomes 2 0 = 53 5 (22) The quantity © defined by (21) is often celled the velocity of sound, since in a tube where $2 nas the constant value c* the propagation of sound waves has the velocity c, We remark that for the incompressible case ve have dp = 0 and hence © which was previously obtained. If ve now o= a, ‘Thus equation (22) gives ae consider (a) the icothermal case, then p const. qn this case (22) becomes %o 8 oe gak D- (25) Tous since Z= grad ® we have here a differential equation of second order and third degree in @. In the thermal case this equation replaces the simple A= 0 which ve had for the incompressible case. (b) Te adiabatic case is e particular case of the so-called polytr case for whiten For adiabatic processes, = 1.4, Here 28Ee ae Substituting thie expression for BF into é Bs D = gt= const. we have 2 web. If we let ¢, be the value of © ata point where the value of q is gp this equation becomes En = ate Jeo Subtracting the last two equations eliminates gH giving te gg o-Ee-H hich aabotttate task into (@2) yield the final form of the eitferentie equation for @ in the polytropte ease 2 rag oo ge EK) Fee - ay d= grado. nen the right hand side of (24) is amsll in magnitude we can find approximate eolu- tions. First take the right hand side zero and solve the resulting equstion, = 0. Call the solution, which satisfies the boundary conditions, @, end intro. duce this qj inte the right hand side of (24). then we have to solve an equation of the form 4p = f(x,y,2) where f ts @ know function. So we continue step ty 29step. Such a method of solution offers difficulty, hovever, vhen q approaches ¢ An magnitude. 7+ Momentum Theorems for Perfect Fluid Notion. We now proceed to derive two integrals of the Buler equation (1). Multi- plying Bquation (1) ty it becomes g- a6 +2 e ge = -ps ered (n+ 2). Yow #=-g grad h 12 a vector pointing vertically dowwards. Denoting by r the specitio wetgnt pg, we may write 7 for -pg grad h, Then the sbove equation can be written ot -F- cade 2) since grad P= i grad p as was show in Section 1. If we integrate thie equation over a volune V of the fluid and write W tor the vector J ay, the weight of V, ve have the firet Momentum equation @) by” Bav-i- Agee Tet 7 be the position vector of the elenent of volume aV ae shom in the figure, ana let an ett ZO of the whole =O 5 center of mase volume V, Take the crose product of ¥' with each term of Bquation (25) and integrate the 30at over V getting J 0? x av = xe - 7 Gx area p)av (en mo “ where FexW= J Fx Fav, his is the second nonentun equation or the equation « of monent of momentum. We want to traneform the mmentum equations (26) and (27) over into equa- tone containing surface integrals. Take (26) firet. In thie equation a AB ° Ag sof 7 and if 4V= de dS se chom in the figure, where de te the a elenent of length along @ velocity: Line and a8 de the area of the crocs section of the bundle of euch Lines shown, we have Nov pa d3. in the mass crossing the face 48 of the element of volume in unit time, since the velocity line in tangent to Z at eath point, no fluid crosses the walls of the ftlsment or bundle, then by the continuity equation the sane axount fa aS mist leave the opposite face in unit tine, Hence pq 45 is constant along a fLlanent. At the point where the filament leaves the volume V we see fron the figwe, if Tis a unit vector paraltel to the outward normal of the element of surface at, that oq 45 = oq,dA. The component of { in the direction of # ts awritten q,. The eecond integral on the right hand side of the foregoing equation can thus be written Ja as Bae = f on astd at point (2) - F at point (1)). Te positive value oq 48, as vas show sbove, equals the product pa,ah at the point (2). On the other hand, it equale the product -pa4A at the point (1), since there the component q, is negative. If ve introduce thie into the last ox: pression we see that On the other hand, it follove from the Gaues transformation { grad pav=f pit an, w @) Hence (26) becomes x ~ ae [ oSav+f odaaa-W-f pita. (28) ("Gy Se (a) Suppose we have sotia bodies imersed in the fluid as surface whieh we denote by C.9. Then in equation (28) the surface @ A consiete of the €.8, plue the Cs, surfaces of the bodies with the oN unit nomal veetore # in the Airections shom. Let F be 32i an. force exerted by the fui on the solid bodies, which te obviously ff (woty’ sure.) Then Algo on @ solid body 9, = 0 80 {yt feat Substituting these two expressions into (28) and solving for ¥ we have the first Momentum equation in the form which gives us the force exerted by the fluid upon any a0lid bodies imersed in it Fetes oBav-s Maaa-s var. 2 . wy ® " as. as. oe ext, we consider Equstion (27). We have to transform firet, the volune integral of # x ag ‘and second, the volime integral of 7x grad p. the £ z es cia ay component of Fx ea is xoq 2 - inien equals upon as oa 55 Pa SS a eat (errs) But = q cor(x,a), v= 4 cos(y,q) walle coe(x,a), % = cos(y,a) 3ence xq Sy og B= og S ov-y2) and Fx og Be og 2 Exh om Be oo 2 xd. Bx av po? x Bars foe x Aur x spav a Sav fy" ‘Da,aa while again by the Gauss transformation: 1G x grad p)av =f p(F xH)aa. ww) (a) Hence (27) becomes 2.3 a ee J ol x Bhan + fo Dagan « BT J px ar, (yn BI Gy Oo) For bodies imersed in the fluid the moment of the forces is We fo x Baa. (vody surfs.) Then in the last equation { e@ x tan = H+ fp x Baa, a es. which gives us the second momentum equation in the final general form 3hHBB pox Bhav fox Bayer - f @ PGF x ayaa. G0) ‘This equation gives the moment of the forces which the fluid exerts upon the solid bodies inside the controlling surface. ‘These equations tor F and H will be used primarily for cases of steady a notion when 22-9 and under conditions when W can be neglected. ‘Then 3 He ¥ be neglected. Th mu faa + an oy = of Fx (oda, + pan. LaF eo If we specialize further to incompressible potential flow, ve have 9 const. and J= 0, ten it foliows that P= 2 and eince H in the case of potential flow is a universal constant, 2 Ba Zs const. 2 or p= -S+ const. we can neglect the const. as it vill drop out of the integral of pil, Hence : Fe Ga, - ta. Lacs Fe But g,=qcor(n,s) and Z= a? where P is a unit vector along the velocity Girection. go ¥ 2G - 2 cos(n,s))aa BL Fas. and 35HAG EPH Ge eRconleyven 8 let =i 2B cos(nje), rt is then seen from the figure that ise unit veotor whieh is the sage of the vector # reflected in the plane nomal to the vector 2, mis plane ts tangent to the potential aurface ¢ = const, and passing through our point, Thus we have $= const. oan 3) #2 F-3y Fas BL oR xe oy where i is the unit vector symmetrical to with respect to the potential eur- face, These tvo equations give the force and moment exerted by @ perfect, incompressible, irrotational flutd upon material bodies inside the surface ¢.8. vhen ‘the flow is steady and the effect of gravity can be neglected. Problen, Given a steady motion with the velocity field (9-2) (ye? ynyz-nere) (22x) (Psy 2? yaya nets?) (oy) (Py? 4e? nyayenenre?), find the etreamlines, the vortex lines, the surfaces on which H is a constant and the pressure distribution. 36Probien 2, xf H=Z and do te the differential clement of Length tn the 2 a oe direction of 3, show that Gp= 9 SE te equivatent to the two theorens of Helanoltz ”(CHAPTER II MOTION IN TWO DIMENSIONS - AIRWING OF INPINITE SPAN If all quantities are independent of 2, and w=0 0 that the motion is psrallel to the xy-plane, ve say that it is a plane motion or two dimensional, In auch @ cate We may envisage a yolime of fluid of unit thickness moving parallel to ‘the xy-plane. 1. Steady Motion of an Incmpressible Pluta. Xf the motion is steady, all partial derivatives with respect to t will be zero and, Af the fluta is incompressible, we will have p= cont. We make these ‘two assumptions in this section. Tae two theorems of Helmholte and the equation of continuity are, as show in Chapter I, equivalent to Buler's hydro.mechanical set up as far as the velocity distribution is concerned, The continuity equation div = 0 becomes Boo xy This equation is identically satisfied if ve assme function y euch that a by o- 2 ve. a) ‘The function y(x,y) 18 called the stream function since v(x,y) = const. gives ve a streamline for the motion. This follove from Equation (1) aa we conclude from ‘there that the directional derivative of y in any direction gives the component of the velocity in a direction making an angle of -90° with the direction of the derivative, since J is tangent to a streamline, tt hi oe cmon sna to cin itentin; tance BY © tere + ie mrerared slong the stetnin, ich wowne that 4+ contt. long enh arene, the valve of § SDL ay fem streamline to streamline, If we define as dn the element in » direction normal 38to de and such that the angle from de to dn is 490°, ve have Beg, a n Baa, av= aan. ‘thus, the increment ay in y from one streamline to the next equals the amount of f1ula pasting between these two streamlines per unit time. (an is the erose section of the filament of width dn and of unit thickness.) It follows that the amount of fluid crossing any cruve AB, shown in the igure, in unit time is jg Yys Which fs the value of y at B minus the value of ¥ at A. In ordar to apply the theorens of Helmholtz we need to bring in the vortex e a ea vector @. since the motion is two dimensional w= 0, = 0 and 73 9 oe 0, 0 a a mH aya Oe my ‘Thus in two dimensional motion the vector @ is normal to the xy-plane. ‘The first theorem of Helmholtz then says that each vortex line, normal to ‘the xy-plane, must travel slong @ streamline. The second theorem states that as an ie a constant. Notice in the foregoing figure the shaded area will nove between the streamlines as show. By the equation of contimalty these two shaded fareas are equal eo de dn remains constant too. Hence @ met stay conetant along cach streamline, But from Rquation (1) fy Fv on ER Bey s0 the constancy of @ gives us as equation of motionay = PCY). (2) Bach function (sy) sxtiefying this equation can be the stress function of an ine compressible perfect fluid; e.g. ¥ = (x + y°)* with ay = 16(x° + y?) = 160y is a solution while ¥= x? + y? with ay = 6(x+y) contradicts (2). We have nov to determine the pressure distribution since thus far ve heve studied the equations from which p had been eliminated. We start with the re~ lation comseting oth A (Choice, Section 5), thin ts, ar 2 = 0, ExS= ered gt. Here Z ie nomal to @. thus the absolute value of 7 x @ ie equal to ao. Me direction of the vector Zx@ is normal to Z and to @, therefore, parallel to the xy-plane and nomal to the tangent of the streamline. According to the right hand rule the positive sense of @x i is opposite to the positive normal as de- fined above, if @, is poritive, Hence end since (9) 18 -a, we have 6. rw. In this vay the value of Hf for each stzeanline is found when we know #(¥)! H= 3s Fay @) ho(except for an additive constant, which ts of no importance). Then from the defini ton of we find the pressure 2 p= (t= b - Foe. Ga) In some cases it is convenient to have the expression for w or A} in intr coordinates. We let the positive Airection of the x-axie coincide with the direction of @ and the Airection of the poeitive y-axis with that of the normal 7 et the point considered. Then 2% a eh tod voile qq = 0, 5" 4 0. Mis Le ween from the figure In which the radive of as curvature R= 92. There dg, = qt and co a a-ak ‘The expression for a, becomes ae- 2. Here R is positive when the center of curvature is on the sane side of the a