Fluid mechanics : a geometrical point of view s. g. rajeev - The ebook is available for online readi

Page 1


https://ebookmass.com/product/fluid-mechanics-a-geometrical-

Instant digital products (PDF, ePub, MOBI) ready for you

Download now and discover formats that fit your needs...

Fluid Mechanics: A Very Short Introduction Eric Lauga

https://ebookmass.com/product/fluid-mechanics-a-very-shortintroduction-eric-lauga/ ebookmass.com

Fluid Mechanics, 8 Ed 8th Edition White

https://ebookmass.com/product/fluid-mechanics-8-ed-8th-edition-white/

ebookmass.com

Free-surface flow: environmental fluid mechanics Katopodes

https://ebookmass.com/product/free-surface-flow-environmental-fluidmechanics-katopodes/ ebookmass.com

Silent Hill 2 Mike Drucker

https://ebookmass.com/product/silent-hill-2-mike-drucker/

ebookmass.com

Transformations : women, gender and psychology Third Edition. Edition Crawford

https://ebookmass.com/product/transformations-women-gender-andpsychology-third-edition-edition-crawford/

ebookmass.com

Le carrousel infernal Joe Hill

https://ebookmass.com/product/le-carrousel-infernal-joe-hill/

ebookmass.com

The Summoning Smith

https://ebookmass.com/product/the-summoning-smith/

ebookmass.com

The Blind Duke's Ward : A Steamy Guardian/Ward Historical Regency Romance Novel (Dukes Ever After Book 1) Claire Devon

https://ebookmass.com/product/the-blind-dukes-ward-a-steamy-guardianward-historical-regency-romance-novel-dukes-ever-after-book-1-clairedevon/

ebookmass.com

Overcoming Secondary Stress in Medical and Nursing Practice Robert J. Wicks [Robert J. Wicks]

https://ebookmass.com/product/overcoming-secondary-stress-in-medicaland-nursing-practice-robert-j-wicks-robert-j-wicks/

ebookmass.com

Hitler: A Biography Peter Longerich

https://ebookmass.com/product/hitler-a-biography-peter-longerich/

ebookmass.com

FLUIDMECHANICS

FluidMechanics

AGeometricalPointofView

DepartmentofPhysicsandAstronomy

DepartmentofMathematics

UniversityofRochester

Rochester,NY

GreatClarendonStreet,Oxford,OX26DP, UnitedKingdom

OxfordUniversityPressisadepartmentoftheUniversityofOxford. ItfurtherstheUniversity’sobjectiveofexcellenceinresearch,scholarship, andeducationbypublishingworldwide.Oxfordisaregisteredtrademarkof OxfordUniversityPressintheUKandincertainothercountries

©S.G.Rajeev2018

Themoralrightsoftheauthorhavebeenasserted

FirstEditionpublishedin2018

Impression:1

Allrightsreserved.Nopartofthispublicationmaybereproduced,storedin aretrievalsystem,ortransmitted,inanyformorbyanymeans,withoutthe priorpermissioninwritingofOxfordUniversityPress,orasexpresslypermitted bylaw,bylicenceorundertermsagreedwiththeappropriatereprographics rightsorganization.Enquiriesconcerningreproductionoutsidethescopeofthe aboveshouldbesenttotheRightsDepartment,OxfordUniversityPress,atthe addressabove

Youmustnotcirculatethisworkinanyotherform andyoumustimposethissameconditiononanyacquirer

PublishedintheUnitedStatesofAmericabyOxfordUniversityPress 198MadisonAvenue,NewYork,NY10016,UnitedStatesofAmerica

BritishLibraryCataloguinginPublicationData Dataavailable

LibraryofCongressControlNumber:2018932827

ISBN978-0-19-880502-1(hbk.)

ISBN978-0-19-880503-8(pbk.)

DOI:10.1093/oso/9780198805021.001.0001

Printedandboundby CPIGroup(UK)Ltd,Croydon,CR04YY

LinkstothirdpartywebsitesareprovidedbyOxfordingoodfaithand forinformationonly.Oxforddisclaimsanyresponsibilityforthematerials containedinanythirdpartywebsitereferencedinthiswork.

Thisbookisdedicatedtomyteacher,A.P.Balachandran.

Preface

Nexttocelestialmechanics,fluidmechanicsistheoldestpartoftheoreticalphysics. Eulerderivedthefundamentalequationsmorethan200yearsago.Yetitisfarfrom complete.WedonotyetknowwithmathematicalcertaintyifEuler’sequationshave regularsolutionsgivensmoothinitialdata.Moreimportanttophysics,thephenomenon ofturbulenceisstillmysterious.

Numericalmethodshavemademuchprogress(especiallyinapplicationstoengineering)inrecentyears.Theexponentialgrowthofcomputingpowerhasmadeitpossible todesignairplanes,submarinesandhouseholdapplianceswithoutcumbersometesting ofprototypes.Localweathercanbepredictedforaboutafortnight,afterwhicheventhe bestcomputersfail.Toolsfromstatisticalmechanicsandquantumfieldtheory(theareas thatroutinelydealwithaninfinitenumberofrandomvariables)oughttobeuseful.

Manyideasoftheoreticalphysics(e.g.,conformalinvariance)originatedinthestudy offluids.Abstractideas(suchasLiealgebras)appearhereinaconcreteandeasily visualizedsetting.InthisbookIwanttopresentfluidmechanicsasanonlinearclassical fieldtheory,anessentialpartoftheeducationofaphysicist.

Thecentralobjectoffluidmechanicsisavectorfield,thefluidvelocity.So,ageometric pointofviewisquitenatural.Atadeeperlevel,onecanunderstandEuler’sequationas ahamiltoniansystem,whosePoissonbracketsaredualtothecommutationrelationsof vectorfieldsandthehamiltonianisthekineticenergy(L 2 -norm).

ThisallowsonetothinkofEuler’sequationsforfluidsandhisequationforarigid bodyonthesamefooting:theybothdescribegeodesicsonaLiegroup.Ofcourse,the fluidismuchmorecomplicated:thegroupisinfinitedimensionalandthecurvature isnegative.Arnoldshowedthatthenotoriousinstabilitiesoffluidmechanicscanbe understoodintermsofthenegativecurvatureofitsgeometry.Althoughverynatural, thisneedssomemathematicalmachinery.Ihavetriedtopresentitinawayaccessibleto theoreticalphysicists.Clebsch’soldideaofusingcanonicalvariablesinfluidmechanics becomesusefulhere.

Moretraditionalsubjectslikeboundarylayertheory,vortexdynamics,andsurface wavesarealsoilluminatedbygeometry.ExpressingEulerandNavier–Stokesequations oncurvilinearcoordinatesisessential:mostboundariesofinterestarenotplanes.A littleRiemanniangeometrygoesalongwayhere.Thelineartheoryofinstabilities isunderstoodintermsofthespectrumofnon-normaloperators.HermannWeyl(of allpeople)foundanexcellentanalyticalapproximationfortheboundarycondition–Blasiustheoryofboundarylayers.Thevortexfilamentequationscanberelatedtothe Heisenbergmodelofmagnets;avortexinfluidmechanicsismappedtoasolitonofspin waves.

Throughout,Iwillpointtonumericalandanalyticalcalculations(mostlyusing Mathematica,butyoucanuseyourfavoritelanguage)togainunderstanding.Thisis notabookoncomputationalmethods.Youdon’tneedasupercomputeranymoretodo interestingwork,soeventhemosttheoreticalofuscanbenefitfromtheiruse.Switching fromananalytical/geometricalpointofviewtothemorepracticaldiscrete/numericalone andback,weunderstandbothbetter.

Studyingchaoticadvectionisagoodwayofdevelopingintuitionforfluidflowsand fordynamicalsystems.Aref(1984)givesasimpleexample,whichhasturnedouttobe usefulinengineeringdevicesthatmixfluidsefficiently.TheSmalehorseshoe,although notusuallythoughtofaspartoffluidmechanics,givesasolidlyestablishedmathematical modelofchaos.Manychaoticsystems(includingAref’s)haveaSmalehorseshoe embeddedinsidethem.SoIhaveincludedadiscussionofthesetopicsinanappendix. Therearemanyreasonstobelievethatrenormalizationisusefultounderstand turbulence.Althoughadiscussionofthosetheoriesisbeyondthescopeofthisbook, Ihaveincludedanappendixonmoreelementaryapplicationsofrenormalization:the IsingmodelandFeigenbaum’sapproachtodynamicalsystemsinonedimension.

Ihavenottriedtosurveyeverysub-fieldindetail:suchacomprehensivesurveywould beaboutasusefulasamapofacountryonascaleof1:1.Afewcasesareexaminedin detail.Theoreticalphysicsisbasedongeneralprinciples(conservationlaws,variational principles)andspecialcasesthatcanbeunderstoodanalyticallyorbysimplenumerical computations.Thetechniquesdevelopedthiswaycanbeadaptedtothosethatarisein applications.Thatsaid,mychoiceoftopicsisnecessarilysubjective.Theemphasison Lietheoryanddynamicalsystemsisunusualforabookatthislevel.

Muchremainsuncovered.Itwouldhavebeennicetotalkofquantumfluids.The greatmysteryofturbulenceandattemptstomodelitcouldhavebeenreviewed.Models ofweatherprediction,oceanography,andastrophysicaljetsareallthriving.Thewhole fieldofmagnetohydrodynamicsisgivenshortshrift.Eachofthesetopicswouldrequire abookofitsown.

Thisisasortofsequelto AdvancedMechanics Rajeev(2013).Aknowledgeof mechanics,linearalgebra(eigenvalueproblems)andpartialdifferentialcalculusisthe mainprerequisite.Thesectionswitha ∗ intheirnamecanbeskippedonafirstreading; theycontainmoreadvancedmaterial.

Thebookisaimedatphysics/mathematicsgraduatestudents;someengineering studentswillalsofindituseful.Thereissomeoverlapandsomerepetition,becausesome ideas(e.g.,dynamicalsystems,Liealgebras)aresorecurrentinphysics.Othersareonly outlined,andworkingthemoutyourselfisanessentialpartoflearningthesubject.

Acknowledgments

Icannotthankmywifeandchildrenenoughfortheirsupportovertheyears.A.P. Balachandrantaughtmuchmorethanphysics:theaudacitytogointonewareas.Thanks totheindulgenceofmycolleaguesinPhysicsandMathematicsDepartmentsforletting meexploredirectionsthatarenotprofitableintheshortterm.

IamgratefultoA.Kar,G.KrishnaswamiandM.Bhattacharyaforcommentingon partsofthemanuscript.ThanksfordiscussionswithA.Iosevichonfractals;D.Geba onregularityofNavier–Stokes;V.V.Sreedharonconformalinvariance;andV.P.Nair andT.Padmanabhanoncountlesstopicsofphysicsinourearlyyears.ThankstoSonke AdlungforencouragingmetowritethisbookandtoHarrietKonishiformanyhelpful suggestions.ThanksalsotoAlanSkullandLydiaShinojfortheexcellentediting.

1.1Thevelocityfield1

1.2Space-timeapproach2

1.3EulerianvsLagrangianpicture3

1.4Integralcurves3

1.5Themethodofcharacteristics4

1.6Conservationlaw5

1.7Densitiesvsscalars6

1.8Steadyflows8

1.9Incompressibleflow8

1.10Irrotationalflow8

1.11Irrotationalandincompressibleflow9

1.12Jacobimatrix10

1.13Flownearafixedpoint11 2Euler’sEquations 14

2.1Conservationofmomentum14

2.2Thestresstensor15

2.3Incompressibleflow16

2.4Conservationofenergy17

2.5Helmholtzequation18

2.6Steadyflows18

2.7Equationsofstate19

2.8TransportformofEuler’sequation21

2.9LinearizationofEuler’sequations:sound24

2.10InviscidBurgersequation25

2.11Scaleinvariance26

2.12d’Alembert’sparadox:limitationstheidealfluidmodel28

3TheNavier–StokesEquations 30

3.1Viscosity30

3.2Viscousincompressibleflow33

3.3Dissipationofenergyatconstantdensity33

3.4Dissipationofenergyforcompressibleflows34

xii Contents

3.5Scaleinvariance:Reynoldsnumber35

3.6Navier–Stokesincurvilinearcoordinates36

3.7Diffusionandadvection37

3.8Thediffusionkernel38

3.9Growthofentropyindiffusion41

3.10Theadvection–diffusionkernel42

4IdealFluidFlows 45

4.1Statics45

4.2SolutionsofLaplace’sequation49

4.3Complexanalyticmethods51

4.4Fluidwithastirrer52

4.5Flowpastacylinder56

4.6Thed’Alembertparadox58

4.7Joukowskiairfoil59 4.8Surfacewaves60

5ViscousFlows 64

5.1PipePoiseuilleflow64

5.2CircularCouetteflow65

5.3Stokesflow66

5.4Stokesflowpastasphere66

5.5Vortexwithdissipatingcore71

6Shocks 72

6.1TheBurgersequations72

6.2TheCole–Hopftransformation74

6.3Thelimitofsmallviscosity78

6.4Maxwell–Lax–Oleneikminimumprinciple78

6.5TheRiemannproblem80

7BoundaryLayers 83

7.1Prandtl’stheory83

7.2TheBlasiusreduction84

7.3Weyl’smethod88

7.4Dragonaflatplate89

7.5Limitationsofboundarylayertheory90

8Instabilities 91

8.1TheRayleigh–Taylorinstability91

8.2LinearizationofNavier–Stokesequations94

8.3Orr–Sommerfeldequation95

8.4Transientsolutionsoflinearequations99

8.5Normaloperators100

8.6Anon-normaloperator101

8.7Anonlinearmodelwithtransients102

8.8Stabilityregained104

8.9Rapidlychangingexternalforce105

8.10TheKapitzapendulum107

9IntegrableModels 109

9.1KdV109

9.2Thesolitonsolution110

9.3Multi-solitonsolutions111

9.4Laxpair111

9.5HamiltonianformalismofFadeevandZakharov113

9.6ThehamiltonianformalismofMagri114

9.7Thevortexfilament114

9.8Geometryofcurves115

9.9Velocityfromvorticity:theBiot–Savartlaw116

9.10Thevelocityfieldofavortexfilament117

9.11Regularizationandrenormalization118

9.12RelationtotheHeisenbergmodel121

10HamiltonianSystemsBasedonaLieAlgebra 123

10.1Rigidbodymechanics123

10.2Vectorspaces124

10.3Liealgebra126

10.4TheVirasoroalgebra133

10.5Hamiltonianforthetwo-dimensionalEulerequations135

10.6Spectraldiscretizationoftwo-dimensionalEuler138

10.7Clebschvariables139

10.8Hamiltonianformof3DEulerequations140

10.9Poissonbracketsofvelocity142

10.10Analogywithangularmomentum143

11CurvatureandInstability 144

11.1Riemanniangeometry144

11.2Covariantderivative145

11.3Geodesicdeviationandcurvature146

11.4Curvatureasabi-quadratic148

11.5Addingdissipation149

11.6Liegroups151

11.7InfinitedimensionalLiegroups152

11.8Geometryofleft-invariantmetrics152

11.9Geodesicsonagroupmanifold153

xiv Contents

11.10Covariantderivativeonagroup155

11.11Curvatureofaleft-invariantmetric156

11.12Geodesicson SO(3)157

11.13Thediffeomorphismgroup158 11.14Liealgebraofvectorfields160 11.15Diffeomorphismsofthecircle160

11.16The L2 -metricforvectorfieldson R3 161

11.17Incompressiblediffeomorphisms162 11.18Curvatureofthediffeomorphismgroup164

12Singularities 166

12.1Norms: L 2 , L p ,Sobolev167 12.2Thedissipationofenergy169 12.3SolutionofNavier–Stokesbyperturbationtheory170 12.4Leray:finitetimeregularity171 12.5Scaleinvariantsolutions173

13SpectralMethods 175

13.1TheChebychevbasis175 13.2Spectraldiscretization176 13.3Sampling177 13.4Interpolation178 13.5Differentiation178 13.6Integration179 13.7ThebasicODE180 13.8Downsampling181 13.9SpectralsolutionoftheOrr–Sommerfeldequation183 13.10Higherdimensions185

14FiniteDifferenceMethods 186

14.1Differentialanddifferenceoperatorsinonedimension186 14.2Padéapproximant188 14.3Boundaryvalueproblems189 14.4Explicitschemeforthediffusionequation190 14.5Numericalstability192 14.6Implicitschemes192 14.7Physicalexplanation194 14.8ThePoissonequation195 14.9DiscreteversionoftheClebschformulation197 14.10Radialbasisfunctions199

14.11ALagrangiandiscretization201

15GeometricIntegrators 203

15.1Liegroupmethods203 15.2Exponentialcoordinates204

AppendixADynamicalSystems 209

A.1Jacobimatrixatafixedpoint209

A.2Stableandunstablemanifolds210

A.3Arnold’scatmap211

A.4Thehomoclinictangle213

A.5TheSmalehorseshoe214

A.6Binarycode215

A.7Iteratedfunctionsystemsontheinterval217

A.8Normalformofthehorseshoe219

C.1TheIsingmodel228

C.2Transfermatrix230

C.3Renormalizationdynamics231

C.4Cayleytree234

C.5SpontaneousmagnetizationoftheIsingmodelontheCayleytree236

C.6TheIsingmodelonsquareandcubiclattices239

C.7Iterationsofafunction239

C.8Feigenbaum’srenormalizationdynamics241

C.9TheFeigenbaum–Cvitanonicequation241

Listoffigures

1.1Integralcurveofafoliation13

4.1Densityofaself-gravitatingfluid48

4.2Flowaroundastirrerinthehalf-plane53

4.3Flowaroundastirrerinsideacircularboundary55

4.4Flowaroundacylinder57

4.5FlowaroundaJoukowskiairfoil60

4.6Groupvelocityofsurfacewaves63

6.1Velocityatashock75

6.2Densityatthesameshock75

6.3SolutiontotheBurgersequation80

7.1TheBlasiussolution87

8.1Solutiontoanon-normaloperator102

8.2Transients103

9.1Vortexfilamentsoliton121

13.1Spectralmethodforanoscillator182

13.2Orr–Sommerfeldspectrum184

14.1TheAdams–Bashforthmethod189

14.2FDMforaboundaryvalueproblem191

14.3Thestableregioninimplicitschemes193

14.4Implicitvsexplicitschemes194

14.5ExactsolutionofaPoissonequationcomparedtothenumericalsolution196

14.6RBFinterpolation200

A.1ManifoldsofArnold’scatmap212

A.2Aglimpseofahomoclinictangle213

A.3Thehorseshoeofthecatmap214

A.4Thehorseshoemap216

A.5Periodicorbitofasequence223

B.1Chaoticadvection226

C.1RenormalizationoftheIsingmodel233

C.2Cayleytree235

C.3SpontaneousmagnetizationoftheCayley–Isingmodel237

VectorFields

Afluidiscomposedofalargenumberofmolecules.Thesemoleculesareinrapidmotion, eachinadifferentdirection.Theycollidewitheachother,whichtendstorandomizethe molecularvelocities.Inthefluidapproximation,wethinkofthesystemascomposedof “fluidelements”whicharelargeenoughtocontainamultitudeofmoleculesbutstill smallcomparedtothesizeofthevesselcontainingthem.

Theaveragevelocityofafluidelementwillbemuchsmallerthantheindividual molecularvelocities.Thedistancebetweenmoleculesissosmallthatwecanregardthe densityandvelocityofafluidascontinuousfunctionsofspaceandtime.Afunction inspace-timeiscalledafield.Thus,thefluidpressureisascalarfieldwhilefluid velocityisavectorfield.Otherexamplesofscalarfieldsaretemperature,entropy,and theconcentrationofsomechemicalpollutantcarriedbythefluid.

1.1Thevelocityfield

Let xi for i = 1,2,3bethecoordinatesofsomepointwithinthefluid.Tobeginwith, thinkofthemasCartesiancoordinates.Later,wewillseethatcurvilinearco-ordinates workjustaswell.Wewillfollowtheconventionofgeometryinwritingtheindexon coordinatesassuperscripts. vi (x, t ) arethecomponentsofthefluidvelocityattime t and position x (sometimeswewillomittheindexonthecoordinate).Thismeansthatafluid elementat xi attime t willmoveto xi + vi (x, t ) atthenextinstant t + ,where isan infinitesimallysmalltimeinterval.

Givenascalarfield f (x, t ),therearetwonotionsoftimederivativethatareimportant influidmechanics.Theobviousoneisthepartialderivative,

inwhichwelookatthechangeintimeatafixedlocationinthefluidelement.Theother isthetotalderivativeormaterialderivative

1.2Space-timeapproach

Inrelativisticphysics,wemustthinkofphysicalquantitiesasfunctionsofspace-time. Thisisafour-dimensionalmanifoldwhosecoordinates xμ arethethreecoordinatesof space xi , i = 1,2,3plustime t ,whichisusuallythoughtofasthezerothcoordinate ofspace-time x0 = t .Thispointofviewisalsoconvenientinsomenon-relativistic situations,1 includingfluidmechanics.Ifweset v0 = 1:

Thusthevelocityfieldcanbethoughtasafirstorderdifferentialoperator(i.e.,the materialderivative)inspace-time:

Evenifthefluidelementsareatrest, vi = 0,theyaremovingforwardintime.Evenif thefluidismoving,therateatwhichitmovesforwardintimeisunaffected.(Thisisthe non-relativisticapproximation).Thecoordinates (t , x) donotneedtobeCartesian;and theydonotneedtobemeasuredinaninertialframe.Wecanmakeany(possiblytimedependent)transformationofthespacecoordinates.Butthetimecoordinateshouldnot change,becauseinthenon-relativisticlimitallobserversagreeonthedefinitionoftime. Ofcourse,thereisatheoryofrelativisticfluids(PoissonandWill,2014).Thisbook happensnottobeaboutthat.

Applyingthechainruleofdifferentiation

1 RecallthetitleofFeynman’sclassicpaper,

Integralcurves 3 and

Thustime-dependenttransformationsofthetypeweareconsideringpreservethe conditionthatthetimecomponentofvelocityisone.

1.3EulerianvsLagrangianpicture

Innon-relativisticmechanicsthereissomethingspecialaboutaninertialreferenceframe: Newton’slawsasoriginallystatedholdinsuchframes.Influidmechanicsthisiscalled theEulerframeortheEulerianpicture.AtransformationfromoneEulerianframeto anotheristime-independent.Itcanstilltransformspacecoordinatesinany(possibly nonlinear)way.

Ifweallowtime-dependenttransformations,wecanevenchoose φ i suchthatthefluid velocityiszeroeverywhereinthenewsystem.Thenewsystemwouldbeco-movingwith thefluid:anon-inertialreferenceframe.ThisistheLagrangianpicture.TheEulerianand Lagrangianpicturesarerelatedbythetransformationsatisfyingthedifferentialequation obtainedbysetting ˜ vi = 0ineqn(1.1):

WewillmostlysticktotheEulerianpicture,asdynamicallawsoffluidmotion(suchas theEulerequations)areeasiesttounderstandfromthispointofview.Butsomepersistent structuresoffluids(vortices,jetstreams)areeasiertoseeintheLagrangianpicture. Sometimeswewillswitchtothispicture.

1.4Integralcurves

Imagineaspeckofdustcarriedalong(oradvected)bythefluid.Itsposition ξ i (t ) will changewithtimeaccordingtothedifferentialequation

Givenitsinitialposition, ξ i (0),thisdifferentialequationcanbesolvedtodetermine thepathfollowedbythedustparticle.Geometrically,thisisacurvestartingat ξ i (0) whosetangentateachpointisthevelocityvectoratthatpoint(andinstantoftime). Theprocessofsolvingadifferentialequationisakindofintegration.So ξ i (t ) iscalled

theintegralcurveofthevectorfield vi (x, t ).Exactlyonesuchcurvepassesthrough everypointinspace.Piecingtogetheralltheintegralcurves,wehaveafunction i (x, t ) satisfying

andtheinitialcondition

1.5Themethodofcharacteristics

Ascalarfieldissaidtobeadvectedbytheflow vi (x, t ) ifitsatisfies

Thatis,itisconstantalongthepathofaparticlecarriedalongbytheflow.Supposewe aregiventheinitialvalueofthisscalarfield f0 (x) = f (x,0) andwewanttopredictwhat willbeitsvalueatsomelatertime t .Wemustthenstartatthepoint x attime t andtrace backalongtheintegralcurveof v tofindthepoint η(x, t ) whereitwasattimezero.You canseethatthisistheinverseoftheproblemwediscussedinthesectiononintegral curves:

(η(x, t ), t ) = x

Soifweknowalltheintegralcurves (x, t ) andcanfindtheinversefunctionabove, wecansolvetheequationforadvectedscalarfields

f (x, t ) = f0 (η(x, t ))

Since (x, t ) isdeterminedby v(x, t ) ,soisitsinverse η(x, t ).Forexample,if v(x, t ) = v aconstant,thesolutionis f (x, t ) = f0 (x vt ) ascanbeeasilyverified.

Thismethodofusingasystemofordinarydifferentialequations(ODEs)tosolve apartialdifferentialequation(PDE)isaparticularcaseofthemethodofcharacteristics.See(CourantandHilbert,1962),Chapter1foranexpositionofthe generalcase.

Whenthevectorfieldistime-independent(steady)wecanexpectasolutiontothe advectionequationthatisalsosteady.Thatis,

Example1.1

If v(x) = x2 , x1 theintegralcurvesarecircles.Asteadysolutiontotheadvectionequation isanyfunctionthatisrotationinvariant.

Thepathsofadvectedparticlescanbequitecomplicated;ingeneraltheycanbe determinedonlybynumericalintegration.

Example1.2 Thedoublegyre

Theparameter controlsthetimedependence;ifitiszerothevelocityisindependentoftime.

Exercise1.1 WriteaprograminMathematica(oryourownfavoritelanguage)which plotstheintegralcurvesofthedoublegyreforvariousvaluesof andtheinitial point.Asmallchangeintheinitialconditioncanleadtoabigchangeintheoutcome afterenoughtime.Ananimationwillbemoreinterestingthanastaticplot.The values α = 1.1, = 0.25aregoodchoices.

1.6Conservationlaw

Thetotalmasscontainedwithinsomeregion V ofthefluidis V ρ(x, t )dx.Sincemass isconservedinnon-relativisticphysics,thechangeofthisquantitymustbeequaltothe inflowofmassintotheregionthroughitsboundary.

Here dSi istheareaofaninfinitesimalelementontheboundary,whichisthoughtof asavectorpointingalongtheoutwardnormal.AppealingtoGauss’theorem

Thustheconservationofmassbecomes(sometimeswewillsuppressthearguments offunctionsforsimplicityofnotation)

Sincethisholdsforanyregion V wemusthave

Thesameargumentholdsforanyscalarquantitythatisconserved.Forexample,supposeourfluidisasolutionoftwokindofmoleculesthatdonotinteractchemicallywith eachother.Thentheirnumberdensities ρ1 (x, t ), ρ2 (x, t ) willbeseparatelyconserved.

Suppose

istherelativeconcentrationofonetypeofmolecule comparedtotheother.Fromtheconservationlawswecandeducethat

Thatis,therelativeconcentrationisanadvectedscalarfield.Thisisoneofthereasons whyadvectedscalarsareinterestinginfluidmechanics.

1.7Densitiesvsscalars

Notallphysicalquantitiesdescribedbyasinglerealnumberateachpointarescalars.For example,theycouldbedensities.Thedifferenceisinhowtheychangeunderacoordinate transformation.Ifwemakeachangeofcoordinates(assumedtobetime-independent forsimplicity)

ascalartransformsinasimpleway:

where φ 1 istheinversetransformation.

Ifthetransformationisinfinitesimal(i.e.,differsfromtheidentitybyaninfinitesimally smallquantitytimesavectorfield),

thechangeinthefunctionisalsosmall

Notethatthechangeis(uptoafactor ) thedirectionalderivativeofthefunction alongavectorfieldwhosecomponentsare vi .Vectorfieldsareinfinitesimaltransformationsofspace.Thechangeofanyquantityundertheactionofavectorfieldiscalledits Liederivative.Forascalaritisjustthedirectionalderivative:

Adensitydescribestheamountofsomephysicalquantity(mass,charge,etc.)in asmallvolume.Soitisthecombination ρ(x)d 3 x whichisinvariantundercoordinate transformations.

Recallthat

.Infinitesimally,

Thus,theinfinitesimalchangeofadensityunderavectorfieldisthecoefficientof in theabove

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.