《Balance》观后感

上午看了一部短片叫balance——平衡。

讲述的是一块悬空的板子上站着5个人，起初他们都有着自己合适的位置，让板子保持平衡。 这样的关系很好，彼此都心知肚明自身的重心点：板子平衡，才能活着。世事无常，总是会有人在不经意间，打破你的安全感。

钓出来的箱子成了不安分的隐性导火线，诱惑降临，人们心中的平衡也逐渐被打破。 最后结局大家可想而知，箱子在板子的一端，仅有的存活者却站在另一端远远的望着。 谁也没办法得到自己想得到的东西，眼下孤苦一人，远不如五个人起初时的那般和谐。 所谓安全感，就是现代人最缺的东西。

因为没有安全感，所以大家都想无头苍蝇一样到处乱飞，嗅到点什么味道就趴在上面疯狂啃食，啃到的是美味的蛋糕算你运气好，要是一不小心啃到的是大便，那恭喜你，你的人生也算是完整了。 前些日子，闺蜜和男朋友分手了。说也好笑，原因竟然是男方极度实诚的一种念头：青春是用来挥霍的。他是个浪子，没办法停留在任何女人跟前，吃着碗里瞧着锅里。他是个傻子，用5年的时间和一个女人相处之后，却不知道其实所有的女人都是一样的。他遇到诱惑之后，没办法控制住自己，终是打破了宁静的日子。闺蜜的心被他伤死了，仅有的安全感都被他亲手捏碎了。

接下来就是各自过活，两条平行线，不会有交集。

安全感的建立，其实很难。特别是在这样的一个环境下，更是难上加难。

我们每天抱怨，抱怨那些打乱我们规矩秩序的人。

经常会听到一个熟悉的电视剧桥段：“我求求你了，请不要来打扰我的生活...”

显而易见，安全感受到威胁的时候，人都处于高度紧张神经紧绷的状态。所以安全感被

打破，是一件很令人讨厌的事情。

但有些人却专以打破安全感为乐，这种人倒也屡见不鲜了。 每个人的局限性在于没办法阻止对方做什么。就像你明明知道这个人会对你构成伤害，可因为时间环境或者其他的什么客观因素，他就是一直在你的世界里走来走，徘徊不定。

缺乏安定的因素，只会让我们变得越来越焦躁。

安全感的获取确是无处不在的。

简易言之就是从你的生活重心中索取安全感。

当你失恋的时候，你就把重心转移到工作中，用忙碌的工作时间冲淡独处时胡思乱想的成分。

当你职场失意时，就在亲情中获取安全感，想到家的温馨就觉得一切归零，终究告一段落。

有句俗话说，安全感是自己给的。

倒不如说，安全感是由自己控制发现获取的。

当你重心转移的同时，你更应该重新审视自己，定位自己。

这个阶段，我应该在哪个位置？应该就哪的重？避哪的轻？

这些都是你在获取安全感的同时需要思考的问题。

我们不能阻止此刻的安全感会被谁打破。

但我们可以控制下一秒的安全感从哪开始建立。

安全感一直在跟我们玩捉迷藏。

现在看见的。不一定是以后所藏匿的地方。

第二篇：balance

Phil.Trans.R.Soc.A(2007)365,823–839

doi:10.1098/rsta.2006.1944

Publishedonline16January2007

Aphysicalmodeloftheturbulentboundary

layerconsonantwithmeanmomentum

balancestructure

BYJOEKLEWICKI1,*,PAULFIFE2,TIEWEI3

1ANDPATMCMURTRY4DepartmentofMechanicalEngineering,UniversityofNewHampshire,

Durham,NH03824,USA2DepartmentofMathematics,and4DepartmentofMechanicalEngineering,

UniversityofUtah,SaltLakeCity,UT84112,USA3DepartmentofMechanicalandNuclearEngineering,PennsylvaniaState

University,StateCollege,PA16802,USA

RecentstudiesbythepresentauthorshaveempiricallyandanalyticallyexploredthepropertiesandscalingbehavioursoftheReynoldsaveragedmomentumequationasappliedtowall-bounded?ows.Theresultsfromtheseeffortshaveyieldednewperspectivesregardingmean?owstructureanddynamics,andthusprovideacontextfordescribing?owphysics.Aphysicalmodeloftheturbulentboundarylayerisconstructedsuchthatitisconsonantwiththedynamicalstructureofthemeanmomentumbalance,whileembracingindependentexperimentalresultsrelating,forexample,tothestatisticalpropertiesofthevorticity?eldandthecoherentmotionsknowntoexist.Forcomparison,theprevalent,well-established,physicalmodeloftheboundarylayerisbrie?yreviewed.Thedifferencesandsimilaritiesbetweenthepresentandtheestablishedmodelsareclari?edandtheirimplicationsdiscussed.

Keywords:wall-turbulence;scaling;?owphysics;meanmomentumbalance

1.Introduction

Fluiddynamicboundarylayersformin?owstangentialtoano-slipwall.Afoundationalnotionpertainingtotheboundarylayeristhatthereisalwaysaregionnearano-slipsurfacewithinwhichthedirecteffectsofviscosityaredynamicallysigni?cant(Prandtl1904;Schlichting1979).WithincreasingReynoldsnumber,thisregionnecessarilybecomesadecreasingfractionoftheoverall?owdomain.Inthismanner,thesolutiontotheNavier–StokesequationapproachesthatoftheEulerequationastheReynoldsnumberbecomeslarge(e.g.alimitingmodelfortheboundarylayerisavortexsheetpositionedin?nitesimallyabovethewall).Giventhis,acentralobjectiveofboundarylayertheoryistodeterminethisrateatwhichtheeffectsofviscositybecomespatially

localized.

*Authorforcorrespondence(joe.klewicki@unh.edu).

Onecontributionof14toaThemeIssue‘ScalingandstructureinhighReynoldsnumberwall-bounded?ows’.

823Thisjournalisq2007TheRoyalSociety

824J.Klewickietal.

Inconnectiontodynamics,attainingthisobjectivealsoservestorevealthescalingbehaviourofthemomentum?eld,sincedoingsoeffectivelydeterminesthenormalizationrequiredtowritethemomentumbalanceequationinaformthatremainsinvariantforvaryingReynoldsnumber.Perhaps,themostfamousexampleofthispertainstothe?atplatelaminarboundarylayer,asthewell-?p?????established1=Rxscalingbehaviourdirectlyre?ectstherateatwhichviscouseffectsdiminish(spatiallylocalize)withincreasingReynoldsnumber.Asmightbeexpected,identifyingthedominanttermsinthemomentumbalancealongwiththosenormalizationsthatretainthisbalanceindependentofReynoldsnumber(i.e.theReynoldsnumberp??????scalings)alsoprovidesabasisforeducing?owphysics.Forexample,the1=Rxbehaviourisalsodirectlyassociatedwiththedynamicsrealizedviathecompetitionbetweentheadvectionandthewall-normaldiffusionofaxialmomentum.

Ofcourse,determiningthescalingbehavioursassociatedwiththemeanmomentum?elddevelopmentinsmooth-wallturbulentboundarylayersisconsiderablymorechallenging.Primaryreasonsforthisarethat(i)themeanmomentumequationisindeterminateowingtotheappearanceoftheReynoldsstressesand(ii)insometurbulentwall-?ows,therearesubregionswithinwhichviscouseffectsarenegligibleinthemean.Thesechallengesnotwithstanding,signi?cantprogresshasrecentlybeenmadetowardsidentifyingtheappropriateapproximateformsofthemeanmomentumbalance(MMB)asthelayeristraversed,aswellasinidentifyingthosenormalizationsthatrenderthesesimpli?edformsoftheequationinvariantwithvariationsinReynoldsnumber(Fifeetal.2005a,b;Weietal.2005a).

TheprimarypurposesofthecurrenteffortaretopresentaphysicalmodeloftheturbulentboundarylayerthatisconsistentwiththestructureoftheMMBandtocompareitwiththeestablishedprevalentmodel.

2.Theprevalentphysicalmodel

Forthepurposeofcomparison,thissectionprovidesarelativelybriefdiscussionofthepredominant,well-accepted,modeloftheturbulentboundarylayer.

(a)Meanpro?le-basedlayerstructureandscalingbehaviours

Theprevalentphysicalmodelofthemeanstructureofthesmooth-wallturbulentboundarylayerhasratherdirectconnectiontothepropertiesofthemeanvelocitypro?le(e.g.Tennekes&Lumley1972;Pope2000;Davidson2004).(Herein,xistheaxialcoordinate,yisthewallnormalcoordinate,UandVarethevelocitycomponentsinthex-andy-directions,respectively,uppercaselettersrepresentmeanquantities,lowercaselettersdenote?uctuating

~ZUCu),anoverbarquantities,tildedenotesinstantaneousquantities(i.e.u

denotestimeaveragingandvorticitycomponentsareidenti?edbytheirsubscript.)Figure1ashowsrepresentativeturbulentboundary-layermeanvelocitypro?les.Thesepro?lesphave??????????beenmadenon-dimensionalusinginner

??w=risthefrictionvelocity,twisthewallshearvariables,utandn,whereutZt

stressandnisthekinematicviscosity,asdenotedbyasuperscript‘C’.Thus,aReynoldsnumberbasedonthefrictionvelocityandtheboundarylayerthickness,d,isgivenbydCZdut/n.Asisconventional,thedatain?gure1arePhil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel

(a)(b)1.0

AB0.8

0.6

0.4

C

110100

y+D10000.20.00100200300y+400500dU+/dy+<UV>+82530252015U+1050600

Figure1.Inner-normalizedmeanvelocityandstresspro?les.(a)Turbulentboundarylayermeanpro?lesandtheirassociatedlayerstructure:(A)viscoussublayer,(B)bufferlayer,(C)logarithmiclayerand(D)wakelayer(dataarefromKlewicki&Falco(1990)).(b)ViscousandReynoldsshearstresspro?lesinturbulentchannel?ow,dCZ590,whereinthiscasedisthehalfchannelheight.ThesedataarefromMoseretal.(1999).

plottedonsemi-logarithmicaxes.Themainfeaturesofthisgraphare(i)alayerimmediatelyadjacenttothesurfacewherethenormalizedpro?leislinear(A),(ii)alayerwherethedependenceofUConyCtransitionsfromlineartoapproximatelylogarithmic(B),1(iii)aregionofapproximatelylogarithmicvariation(C),followedby(iv)anextensiveouterregionwithinwhichUCvariesaccordingtoafunctionconsistingofalogarithmicpartplusawakepart,w(y/d).Dynamically,theviscoussublayer(A),0%yC%4,isidenti?edasaregionwhereviscosityhasamajoreffect.Inthebufferlayer(B),4%yC%30,theviscousandReynoldsstressesarebothdynamicallysigni?cant.Thelogarithmiclayer(C)extendsfromnearyCZ30toy/dx0.2andisseentobedominatedbytheeffectsofturbulentinertia.Withinthewakelayer(D),0.2%y/d%1,meanandturbulentinertiaarepredominant.

Attachingthesetime-averageddynamicalattributestolayersA–Distypicallyjusti?edbyexaminingtherelativemagnitudesofthemeanviscousstress,

(e.g.Tennekes&vUC/vyC,andtheReynoldsshearstress,TCeyCTZLumley1972;Pope2000;Davidson2004).Representativepro?lesofthesequantitiesareshownin?gure1b.ThesedatarevealthatthemagnitudeofTCiszeroatthewall,butrapidlyrisestoavaluethatisO(1)byyCx30.Conversely,vUC/vyCZ1.0atyCZ0,butdiminishestoaquantitythatismuchlessthanO(1)byyCx30.(Herein,theordersymbol,O($),willbeusedwithrespecttoe/0,whereeisasmallparameter,e.g.e2Z1/dC.Forexample,aZO(b)forpositivea(e)andb(e)istakentomeanthatbotha/bandb/aareboundedase/0.)Theseobservationsarecommonlyusedtosupportthenotionthat,inthemean,thedynamicaleffectsofinertiabecomegreaterthanthoseofviscositybeginninginlayerBandpredominantlysowiththeoutwarddistancefromlayerB.

ExaminationofthelayerthicknessesandthevelocityincrementsacrosstheselayersrevealstheReynoldsnumberdependenciesinherenttothisphysicalmodel.Thesedependenciesarelistedintable1.Asisapparentfromtheentries1Notethatapproximatelylogarithmicdoesnotexcludecertainpower-lawforms.Also,recentexperimentaldataprovideevidencethatthetraditionallyde?nedlogarithmiclayeriscomposedoftworegionshavinglogarithmic-likebehaviour,butnotnecessarilythesameslope(Osterlundetal.2000;McKeonetal.2004).

Phil.Trans.R.Soc.A(2007)

826J.Klewickietal.

Table1.Scalingbehavioursofthelayerthicknessesandlayervelocityincrementsassociatedwiththepredominantlyacceptedphysicallayerstructureoftheturbulentboundarylayer(C,DandEareconstants).

physicallayer

A

B

C

D(viscoussublayer)(bufferlayer)(logarithmiclayer)(wakelayer)DyincrementO(n/ut)(x4)O(n/ut)(x26)O(d)(/0.2d)O(d)(x0.8d)DUincrementO(ut)(x4)O(ut)(x9)O(UN)(x(ut/k)log(d/C))

O(UN)(x(ut/k)log(d/D)CE)

intable1,thelayersassociatedwiththedirecteffectsofviscosity(AandB)haveathicknessthatremainsa?xednumberofviscouslengthsindependentofReynoldsnumber.Correspondingly,withincreasingdC,theselayerthicknessesbecomeadiminishinglysmallfractionofdataratedirectlyproportionalto1/dC.Similarly,thevelocityincrementsacrosstheselayersaremeasuredbya?xednumberofut,andthusasdC/N,theybecomediminishinglysmallrelativetoUNatarateproportionaltout/UN.Conversely,boththelogarithmiclayerandthewakelayergrowatarateproportionaltodasdC/N.Ofcourse,whenmeasuredinviscouslengths,theselayerthicknessesareunboundedasdC/N.Inasimilarmanner,thevelocityincrementacrosseitherlayerCorDapproachesa?xedfractionofUNasdC/N.

(b)Dynamicalconsiderations

Importantelementsassociatedwiththepredominantmodelrelatetothederivationofthelogarithmicmeanvelocitypro?le.Whileotherderivationsexist,theconstructioncommonlydeemedmostrigorouspostulatestheexistenceofatwo-layermathematicalstructure(innerandouter),andassumesthattheinnerandouterscalingshaveacommonregionofoverlapwithinwhichbotharesimultaneouslyvalid(e.g.Millikan1939;Tennekes&Lumley1972;Pope2000;Panton2005).Underfurtherassumptionthatthepro?lestrictlyincreaseswithy,themeanvelocitygradient(simultaneouslyexpressedinitsinnerandouterforms)ismatchedintheoverlaplayerasyC/Nandh/0.Theclassicalformsofthelogarithmiclawofthewallanddefectlawsubsequentlyfollow.Accordingtothisdescription,theinnerlayerextendsfromthewalltotheouteredgeofthelogarithmiclayer,andtheouterlayerextendsfromtheinneredgeofthelogarithmiclayertod.Animportantphysicalimplicationofthismathematicaldescriptionisthecorrespondencebetweentheoverlaplayerandaninertialsublayer.Thus,considerableresearchhasbeendevotedtowardsunderstandinglogarithmiclayerturbulenceanditssimilaritiestothespectralversionofinertialrangeturbulence(e.g.Townsend1976;Perry&Abell1977;Perry&Marusic1995;Morrisonetal.2004;Davidsonetal.2006;McKeon&Morrison2007;Metzger2006).

Otherconsiderationspertaintohowanygivenmodelsetstheconceptualframeworkforinterpretingmeasurementsandobservations.Notableamongtheserelatetodescribingthecharacteristicsanddynamicalbehavioursofboundary-layercoherentmotions.Acomprehensivereviewofboundary-layercoherentmotionsiswellbeyondthescopeofthepresenteffort.ItisusefulhowevertoidentifyPhil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel827

theimplicationsofthepredominantmodelrelativetotheinterpretationofdynamicalprocesses.Tothisend,issuesrelatingtotheself-sustainingmechanismsandtheso-calledinner/outerinteractionarenowbrie?ydiscussed.

Assupportedbythestresspro?lesof?gure1bandothernear-wallstatisticaldata(e.g.theturbulencekineticenergypro?le),theprevalentmodelplaceshighimportanceonthenear-wallregion(bufferlayerandbelow;e.g.Robinson1990).Numerouscoherentmotionshavebeenidenti?edasdynamicallysigni?cantinthisregion—includingstreaks,pockets,streamwisevortices,internalshearlayersandhairpinvortices.Indeed,thisregionisassertedbymanytobewheretheself-sustainingmechanismsofboundarylayerturbulenceprimarilyreside(e.g.seeanumberofcontributionstoPanton1997).Itisalsogenerallyaccuratetoattributethisfocusonthenear-wallregionwiththeidenti?cationofthebufferlayeraswherethedynamicstransitionfrombeingstronglyin?uencedbyviscositytowhereturbulentinertiadominates(Pope2000;Davidson2004).Thislastnotionalsohasbearingonhowtoconstructadescriptionoftheso-calledinner/outerinteraction.Properlycharacterizingthenatureoftheinteractionsbetweentheinnerandtheouterlayersisassertedbymanytobecentraltounderstandingturbulentboundarylayerdynamics(e.g.Kline1978;Falco1983;Thomas&Bull1983;Klewicki1989;Sreenivasan1989;Wark&Nagib1991).Thisassertion?ndssupportfromatleasttwocompellingarguments.Atperhapsthemostfundamentallevel,theturbulentboundarylayermaybeconsidereda?uiddynamicalmachinethat,onanaverage,convertsfree-streammomentumintotangentialforceactingatthe?uid/solidinterface.Giventhis,oneisthenfacedwithdescribinghowthenetmomentumtransferacrossthelayeroccurs,andthustheinner/outerinteraction.Similarly,aprimarycharacteristicoftheboundarylayeristhatwithincreasingdCtheratiooftheoutertotheinnerlength-scaleincreases.Theinner/outerinteractionisunavoidablyconfrontedifthedynamicalaccommodationtothisscaleseparationistobedescribed.Apparently,owingtoitsinherentcomplexities,theinner/outerinteractionremainsresistivetoathoroughcharacterization.Forexample,iftheviscous/inertialinteractionisseenascentraltotheinner/outerinteraction,thentheprevalentmodelplacesitinthebufferlayerforalldC.Ontheotherhand,byde?nition,theoverlapregioniswhereinnerandouterscalingsaresimultaneouslyvalid,andthusjusti?cationtoprimarilyassociateinner/outerinteractionswiththelogarithmiclayer.Doingso,however,requiresexplainingwhyinner/outerinteractionsshouldoccurinaninertialsublayer,2andnotintheregionwheretheviscousandReynoldsstressesareofthesameorderofmagnitude.

Lastly,asigni?cantandgrowingbodyofresults(Wark&Nagib1991;Meinhart&Adrian1995;Adrianetal.2000;Ganapathisubramanietal.2003,2005;Tomkins&Adrian2003;Morrisetal.inpress;Priyadarshanaetal.2007)supporttheperspectivethatthelogarithmiclayerisinstantaneouslycomposedofahierarchyofmotions,andthatthesemotionsarenominallyarrangedasuniformmomentumzonessegregatedbyrelativelynarrowvortical?ssures.Atpresent,itisnotreadilyapparenthowsuchaninstantaneousstructuremightgiverisetoamathematicaldescriptionbaseduponanoverlaplayer.ThistypeofEmpiricalobservationsrelatingtothebehaviourofthelogarithmicmeanpro?lehaverecentlypromptedproposalsregardingtheexistenceofameso-layer(Wosniketal.2000)intheregion30%yC%300andanextended‘bufferregion’(Osterlundetal.2000)outtoaboutyCZ200.Phil.Trans.R.Soc.A(2007)2

828J.Klewickietal.

hierarchicalstructurehoweverwouldseemtohavearathernaturalconnectiontotheearlier,phenomenological,logpro?lederivationbaseduponpostulatingthedistancefromthewallastheappropriatelength-scale(Prandtl1925).

3.Analternativephysicalmodel

Recentempiricalobservationsandmultiscaleanalysisprovidethebasisforanalternativephysicalmodeloftheturbulentboundarylayer(Fifeetal.2005a,b;Weietal.2005a,b).Animportantpremiseunderlyingthelayerstructuretobedescribed,andinturnthealternativephysicalmodel,isthattheMMBinitsunintegratedform(andinthiscaseasappliedtoboundarylayer?owoveraplanarsurfacelocatedatyZ0),

CCv2UCvTCCvUCvUUCVZC;e3:1Tvxvyvyvy

providesthetime-averageddescriptionofthedynamics.Theleftsideofequation(3.1)representsadvectionbythemean?ow,whiletheright-sidetermsrepresenttheviscousandReynoldsstressgradients,respectively.Forthe?atplate?ow,thereareonlythesethreedistinctdynamicaleffects,andthustheratioofanytwodeterminesthenaturebywhichtheequationisbalanced.

(a)MMB-basedlayerstructureandscalingbehaviours

Weietal.(2005a)exploredthestructureofboundarylayer,pipeandchannel?owsbyexaminingtheratioofthelasttwotermsinequation(3.1).Thedynamicsre?ectedbyequation(3.1)mustarisefromabalanceofatleasttwonon-negligibleterms,andthusinterpretationofthisratioisasfollows.(i)Ifjev2UC=vyC2T=evTC=vyCTj[1,thentheReynoldsstressgradienttermisnegligibleandequation(3.1)sumstozeroessentiallythroughabalance

ofthemeanadvectionandviscousstressgradientterms.

(ii)Ifjev2UC=vyC2T=evTC=vyCTj/1,thenthemeanviscousstressgradienttermisnegligibleandequation(3.1)sumstozeroessentiallythrougha

balanceofthemeanadvectionandReynoldsstressgradientterms.

(iii)Ifjev2UC=vyC2T=evTC=vyCTjx1,thentheReynoldsstressandtheviscousstressgradientsbalanceandareeithergreaterthanorofthesameorderof

magnitudeasthemeanadvectionterm.

Availablehigh-qualityexperimentalandDNSdata(Zagarola&Smits1997;Moseretal.1999;DeGraaff&Eaton2000)weredifferentiatedandtheindicatedratiowasexaminedasafunctionofyfordifferingdC.Thesketchof?gure2depictsthebehaviourofthestressgradientratioatany?xeddC.Asindicated,thedynamicalbalanceisdescribedbyafour-layerstructure.LayerIessentiallyretainsthecharacteroftheviscoussublayer,andintheboundarylayerisaregionwheretheviscousstressgradientnominallybalancesmeanadvection.InlayerII,themagnitudeoftheratioisveryclosetounity,andthusthislayeriscalledthestressgradientbalancelayer.Acrossthemesolayer(layerIII),theReynoldsstressgradientchangessignandthetermsinequation(3.1)undergoabalancebreakingandexchange(Fifeetal.2005b;Weietal.2005a).WithinlayerIII,allthethreePhil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel

ratio of stress

gradients

peak Reynolds

stress location829

?1y+

IIIIIIIV

Figure2.SketchoftheratiooftheviscousstressgradienttotheReynoldsstressgradientinboundarylayer,pipeandchannel?owsatanygivenReynoldsnumber.ThedottedlineinlayerIisforaboundarylayer,andthesolidlineisforapipeorchannel.

Table2.Scalingbehavioursofthelayerthicknessesandvelocityincrementsoftheproposedalternativemodel.(NotethatthelayerIVpropertiesareasymptoticallyattainedasdC/N.)physicallayer

I

II

III

IVDyincrement3)O(n/ut)(x??p??????????????Ond=ut(x1.6)??p??????????????Ond=ut(x1.0)O(d)(/d)DUincrementO(ut)(x3)O(UN)(xUN/2)O(ut)(x1)O(UN)(/UN/2)termsinequation(3.1)arenominallyofthesameorderofmagnitude,3andfromtheouteredgeoflayerIIItoyZd(i.e.layerIV),equation(3.1)ischaracterizedbyabalancebetweenthemeanadvectionandtheReynoldsstressgradient.

Thefeaturesof?gure2,depictedfor?xeddC,persistforthedCrangecurrentlyaccessibletoinquiry(e.g.spatialresolutioninthePrincetonSuperpipelimitedtheReynoldsnumbertoRC%41235).Inthisregard,itisalsorelevanttonotethatboundarylayerandpipe/channel?owsexhibitthesamebehaviourstowithinthedifferencesbetweenthemeanadvectionandpressuregradientpro?les.Thus,forexample,inlayerIItheirstructureisexpectedtobehighlysimilarsinceinthislayerthesetermsaremuchsmallerthanthedominantstressgradientterms.Quantitatively,thelayerthicknessesandthevelocityincrementsacrosstheselayershavebeenshownbothempiricallyandanalyticallytoexhibitdistinctReynoldsnumberdependencies.Table2presentsthesescaling3Ofcourse,rightatthepeakintheReynoldsstresstheReynoldsstressgradientpassesthroughzero,andthusinanarrowzonearoundthispointwithinlayerIIItheReynoldsstressgradientissmallerthantheothertwoterms.

Phil.Trans.R.Soc.A(2007)

830

(a)

80

60

40

×10?3200

?20

?40

?60

?80IIIIIIJ.Klewickietal.(b)III～(d +)1/2III～(d +)1/2IVlineU +?U = 1utIV?U = U∞/2

110100log(y+)

Figure3.AttributesoflayersIIandIII:(a)innernormalizedReynoldsstressgradient(solidline)andviscousstressgradient(dashedline)pro?les(dataarefromtheRqZ1410DNSofSpalart(1988))and(b)schematicdepictionoftheconnectionbetweenthegrowthratesoflayersIIandIIIandtheirassociatedvelocityincrements.

behaviours.Asisevident,layersIandIVadheretoinnerandouterscaling,respectively.Ontheotherhand,layersIIandIIIexhibitmixedscalingproperties.TheinnernormalizedthicknessoflayerIIgrowslikethegeometricp?????CmeanoftheReynoldsnumber(i.e.wd),whileitsvelocityincrementp?????remainsaCC?xedfractionofUN,independentofd.Similarly,DIIIywdC,whileitsvelocityincrementisonlyabout1.0ut,independentofdC.Thesescalingbehavioursdifferconsiderablyfromtheprevalentmodelandareassociatedwith????????????ptheexistenceofathirdfundamentallength-scale,nd=ut.Thislengthisintermediateton/utandd,requiredtoscaletheMMBinlayerIIIandbecomesincreasinglydistinctasdC/N.

ThemeandynamicsandscalingbehavioursassociatedwithlayersIIandIIIarecentral(andapparentlyunique)totheproposedphysicalmodel,andthuswarrantfurtherdiscussion.LayerIIiscalledthestressgradientbalancelayersincethedominantdynamicalmechanismsarethetermsontherightofequation(3.1);theirratiobeingK1in?gure2.Contrarytotheprevalentnotionthat,onanaverage,boundarylayerdynamicsaredominatedbyturbulentinertiaoutsidethebufferlayer(independentofdC),momentumbalancedatarevealthatanequalcompetitionbetweentheviscousforceandtheturbulentforcepersiststoay-locationnearthepeakintheReynoldsstress,Tmax.ConsistentwiththemathematicalhierarchyofscalinglayersrevealedbyFifeetal.(2005a),inthemodelposedbelowthiscompetitionisassociated(inatime-meansense)withthevorticalmotionsformingandevolvingfromthenear-wallvorticity?eld.Itissigni?canttonote,however,thatthebalanceinlayerIIcomesaboutviatwooppositesign,nearlyequal,butdecreasingmagnitudefunctions.ThesefunctionslosedominanceovermeanadvectionaslayerIItransitionsintolayerIII(?gure3a).

Thescalingsoftable2revealthatthelayerIIandIIIthicknessesarecoupled,suchthattheirvelocityincrementsfollowouterandinnerscaling,respectively.Thesepropertiesunderlienewinterpretationsrelatingto,forexample,thenatureoftheinner/outerinteractioninboundarylayers.ItisrelevanttonotethatthemajorportionsoflayersIIandIVandalloflayerIIIresidewithintheboundsofthetraditionallyde?nedlogarithmiclayer.TheloweredgeoflayerIIis?xedneartheedgeoftheviscoussublayer(independentofdC),whilethepositionofitsPhil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel

detached eddies

(?T/?y < 0)

831fully three-dimensionalvorticity field

IVIIIII

IFigure4.Schematicofsomeofthedynamicalattributesoftheproposedmodelfortheturbulentboundarylayer.Layernumbersp?????arethesameasthoseidenti?edin?gure2.Notethattheposition

CofTmax,ymax,varieslikedC.

p?????outeredgeextendstoincreasingyvalueslikedC,suchthatDIIUZUN/2.

(NotethatunderouterpnormalizationthepositionoftheouteredgeoflayerII?????Cmoves‘inward’like1=d.)Owingtothispositioningbehaviourp?????forlayerII,CboththeendpointsoflayerIIIvarywithd,particularly,likedC.Thus,whilethelayerIIIthicknessexhibitsthesameReynoldsnumberscalingbehaviouraslayerII,itsvelocityincrementisonlyx1ut.ThisarisesowingtothefactthatwithincreasingdC,layerIIIispositionedatincreasingyC-locationsinaregionwhereUCxlog(yC)(?gure3b).C

(b)Physicalinterpretation

Elementsofanewphysicalmodelarerepresentedintheschematicof?gure4.Unlikethephysicalpicturepromotedbytheprevalentview,thismodelisspeci?callyconstructedtobeconsistentwiththeMMBproperties.Thisleadstoanumberofnewinterpretationsrelatingtothedescriptionof?owphysics.Ofcourse,itisrecognizedthatanydefensiblephysicalmodelshouldalsoembracethenumerousindependentempiricalobservationsrelating,forexample,totheboundary-layercoherentmotionsdiscussedpreviously.Existingexperimentalobservationsarenowinterpretedinthecontextofthisnewmodel.

(c)Characteristicvorticalmotions

~es0providesperhapsthemostusefulcriterionfordistinguishingtheSinceu

boundarylayerfromthefree-stream,theprocessofdescribingtheproposedmodelbeginsbyattributingthecharacteristicvorticalmotionstolayersI–IV.SincethepropertiesoflayerIIIarelargelyattributedtotheinteractionbetweenlayersIIandIV,thediscussionproceedsaccordingtoaI,II,IV,IIIorder.(i)LayerI

Existingphysicalandnumericalexperimentsrevealthatthevorticity?eldintheregion0%yC%3hasaperturbedsheet-likecharacterthatpredominantlymeandersinthex-andz-plane,buthaslittlewall-normalcomponent(e.g.Kim

~z(i.e.oppositesigntoetal.1987;Balintetal.1991;Klewicki1997).Positiveu

Phil.Trans.R.Soc.A(2007)

832J.Klewickietal.

themeanvorticity)isvirtuallynon-existentinthisregion(Klewickietal.1990;Rajagopalan&Antonia1993;Metzger&Klewicki2001).Perturbationsofthissheet-likedistributionofvorticityarebelievedbymanytoconstitutetheinitialconditionsthattriggertheevolutionofmotionsarisingoutoflayerI(Offen&Kline1974;Perry&Chong1982;Wallace1982;Smithetal.1991).

(ii)LayerII

AttheloweredgeoflayerII,thevorticity?eldishighlyalignedinthespanwisedirection.Acrossthislayer,thevorticity?eldthreedimensionalizes.TheassociatedcharacteristicsarearapidreductioninjUzj,theappearanceof000~zandatrendtowardsuxpositiveuZuyZuz,wheresuperscript‘0’denotesr.m.s.(e.g.Klewickietal.1990;Balintetal.1991).Asindicatedin?gure4,thislayerisprimarilyassociatedwiththeattachededdies;attachedinthesensethattheiretiologyisattributabletothethreedimensionalizationofthenear-wallvorticity?eld.Notethatforallofthedescriptionsgiven,theeddiesandthedynamicsattributedtoanygivenlayershouldbeviewedasprobabilisticwithachangingmixtureofcharacteristicpropertiesintraversingfromonelayertothenext.Thus,attachededdiesareviewedaspredominantlypopulatinglayerII,andtoadecreasingdegreewithincreasingyCacrosslayerIIIandintolayerIV.Asigni?cantbodyofresearchsupportsthehypothesisthathairpin-likevorticesconstituteabasicbuildingblockofwallturbulence(e.g.Theodorsen1952;Head&Bandyopadhyay1981;Wallace1982;Perryetal.1986;Smithetal.1991;Adrianetal.2000;Ganapathisubramanietal.2005).Broadlyspeaking,hairpinvorticesareseentoformviatheredistributionofnear-wallvorticity(mainly

~z),andsubsequentlyevolveoutward.Whileearlier?owcomposedofu

visualization-basedevidenceindicatedthatthesemotionsmightextendfromthesublayertotheedgeoftheboundarylayer(Head&Bandyopadhyay1981),anincreasingandpredominantlymorerecentbodyofresultsindicatethatatsomey-positionthevorticalmotionsloseconnectionwiththeirnear-wallorigin.Consistentwiththesestudies,theprimarycandidateeddiesforlayerIIarehairpin-likevorticesthatcollectivelyorganizeintohierarchicalpackets.Inthisregard,theevolutionofthevorticityinthesemotionsisgenerallyattributedtoacompetitionbetweenvortexstretchingandviscousdiffusion.

(iii)LayerIV

Propertiesofthevorticity?eldinlayerIVincludeessentiallyequalvorticitycomponentintensitiesandanegligiblemeanvorticity.Thevorticityprobabilitydistributionshoweverarerelativelylongtailed—re?ectingtheincreasinglyintermittentnatureofthevorticity?uctuations(e.g.Balintetal.1991).Giventhesecharacteristics,thevorticalmotionsassociatedwithlayerIVareidenti?edasdetachededdies;detachedinthesensethatthesespatiallycompactmotionsareuncorrelatedwiththenear-wallvorticity?eld,andnominallyadvectwiththelayerIVmean?ow.Thoughaspeci?cgeometricformforthedetachededdiesis

~zstructure,thenotcompletelyestablished,directmeasurementsofnear-wallu

increasinglythree-dimensionalnatureofthevorticity?eldwithincreasing~e$ueZ0(KlewickidistancefromthesurfaceanddataconsiderationsrelativetoV

etal.1990;Rajagopalan&Antonia1993)supporttheexpectationthatatsomey-positionthecharacteristicvorticalmotionsbecomespatiallylocalizedandPhil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel833

topologicallyformclosedloops.Detachededdiesarehypothesizedtopredomi-nantlypopulatelayerIV,andtoadecreasingdegreewithdecreasingyCacrosslayerIIIandintolayerII.Thesimplestformofsuchaneddyisavortexring-likemotion.Falco’searlier‘typicaleddy’observations(Falco1977,1983,1991)supporttheexistenceofintermediatescalering-likeeddiesinboththeinnerandouterregions.Regardlessoftheexactgeometricformofthedetachededdies,a

~z,knowntobehypothesizedcharacteristicfeatureisthattheycontainpositiveu

prevalentinlayerIV.

(iv)LayerIII

GiventheattributesoflayersIIandIV,layerIIIisviewedasazonewithinwhichthecharacteristiceddytransitionsfromattachedtodetachedwithincreasingy.Thus,withinlayerIIItheexpectationisto?ndanearlyequalmixtureofattachedanddetachededdies.Similarly,theprocessbywhichattachededdiesmightevolveintodetachededdiesisexpectedtobecharacteristicoflayerIII.Inthisregard,therecentsimulationsofBakeetal.(2002)providecompellingevidencefortheformationofvortexringsfromthepinch-offofthelegsofhairpin-likevorticesduringthelateststagesoftransition.ThisprocesswaspreviouslyproposedbyFalcoasamechanismforring-likemotionformationintheturbulentboundarylayer,andwasexplorednumericallybyMoinetal.(1986).Theattached–detachededdydecompositionofthevorticity?eld?ndssupportfromvisualstudies(Falco1983,1991;Klewicki1997),two-pointvorticitycorrelations(Klewicki&Falco1996;Metzger&Klewicki2001)andDNSandPIVstudies(Jimenez&delAlamo2004;Christensen&Wu2005;Ganapathisubramanietal.2005).Furthermore,theinclusionofthedetachededdyconcepthasbeenfoundtoimprovecoherentmotion-basedmodelperformance(Perry&Marusic1995).Overall,layerIIIisnominallyviewedastheregionwheretheattachedanddetachededdiesinteractwithhighestprobability.

(d)Characteristicdynamics

ThepropertiesoftheMMBsuggestthatspeci?cdynamicalattributesmaybeassociatedwiththeattached/detachededdystructureproposed.Forexample,undertheproposedmodelattachededdiesformandevolveacrosslayerII,andthustheirdynamicalsignatureisthattheyproduceinstantaneouscontributionstopositiveKv=vy.Similarly,thecharacteristiceddiesoflayerIVaredetached.Therefore,theirdynamicalsignatureisthattheyproducenegativeKv=vy.Thisidenti?cationofbothasourceandasinkcharacterwiththeReynoldsstresstermintheMMBisapparentlydistincttothepresentmodel.

Inthecontextofthesedynamicalsignatures,itisusefultoexaminetheequation??vv??KvCwKu:ZzKyCe3:2Tvyvx

Forturbulentchannel?ow,thelasttermisidenticallyzero,whileforboundarylayersthistermisgenerallysmall.4Toagoodapproximation,however,thegradientoftheReynoldsstressislargelyestablishedbythedifferenceofthe4Hinze(1975)associatedthevorticaltermswithTownsend’sactivemotionsandthestreamwisederivativetermswiththeinactivemotions.

Phil.Trans.R.Soc.A(2007)

834J.Klewickietal.

indicatedvelocityvorticitycorrelations.Giventhis,theinterpretationisthatinlayerIItheattachededdiesinteractwiththevelocity?eldtogenerateanetpositivesum,andinlayerIV,thedetachededdiesgenerateanetnegativesum.Thedominanttermsinequation(3.1)indicatethatthedynamicsunderlyingtheevolutionofattachededdiesischaracterizedbyacompetitionbetweenviscousshearforcesandturbulentinertia.Duringthisevolution,thesemotionsactasasourcetothemeanmomentum,andateachReynoldsnumberthissource-likecharacteris,onanaverage,depletedatymax,thepositionofthepeakintheReynoldsstress,Tmax.Similarly,detachededdydynamicsinlayerIVischaracterizedbyacompetitionbetweenmeanadvectionandturbulentinertia.Onaverage,thissink-likecharacterextendsfromymaxtod.The?ow?eldinteractionsunderlyingequation(3.2)ineitherlayerIIorIVhavebeenshowntohavesigni?cantcontributionsfromintermediatescalemotions(Priyadarshana&Klewicki2003).Physically,itisrationaltoattributethistothefactthatasdC/Nvelocityspectrapeakatdecreasinglylowwavenumber,whilevorticityspectrapeakatincreasinglyhighwavenumber.Thus,accordingtoequation(3.2),thevelocityandvorticity?eldsmustcorrelateoversomeintermediatewavenumberrangeinorderfortheretobeanetmomentumtransportviaturbulentinertia.Thisargument,however,apparentlyonlypartiallyholds.MeasurementsoveraverybroadrangeofdCindicatethatwithincreasingdCaspectrallylocalscaleselectionoccurs.Thisresultsinsigni?cantcontributionstothelong-timecorrelationtoarisefromportionsofthecospectranearthepeaksintheparticipatingvelocityandvorticityspectra,respectively(Priyadarshanaetal.2007).

(e)Inner/outerinteractions

TheMMB-basedmodelsupportsaclearsetofperspectivesregardinginner/outerinteractions,someofwhicharenowbrie?ydiscussed.OneinterpretationofthelayerI–IVvelocityincrementsisthatthenetcirculationassociatedwiththeoutwardtransportofvorticityfromlayerII(supportingboundarylayergrowth)isasymptoticallybalancedbyanetinwardtransportofmomentumfromlayerIV(requiredforthegenerationofasurfacedragforce;?gure4andtable2).Theseattributesrathernaturallyalignwiththemomentumsource/sinkcharacterascribedtotheattachedanddetachededdies,respectively.Speci?cally,theattachedanddetachededdiesareviewedastheactivevorticalmechanismsbywhichlayerIIturbulenceissustained(andnear-wallvorticityistransportedoutwardtolayerIV),andbywhichfree-streammomentumisextractedandtransportedintolayerII,respectively.Thesephysicalattributes?ndconsistencywithperturbedboundarylayerstudies,indicatingthattheinnerregionrapidlyadjusts,andfollowingperturbation,rapidlyrecovers(e.g.Smits&Wood1985).Conversely,recoveryoftheouterlayerrequiresverylongredevelopmentlengthsdownstreamofaperturbation(e.g.Eaton&Johnston1981).

Anotherimportantfeatureisthatthestatisticalcentreoftheinner/outerinteractionisinlayerIII.ThiscentrestheinteractiononthezonewheretheMMBundergoesabalancebreakingandexchange;thenetresultbeingthatthedynamicschangefromabalancebetweentheviscousforceandtheturbulentinertiatoabalancebetweentheturbulentinertiaandthemeanadvection.Similarly,thepresentmodelalsoattachesdynamicalimportancetotheinteractionbetweenthemotionsthatproducepositiveKv=vyinlayerIIandthosethatproducenegativeKv=vyPhil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel835

inlayerIV.5Inconnectionwiththis,increasingscaleseparationrequirestheaforementioneddynamicalaccommodationwithincreasingdC,andthusthepresentmodelidenti?esanintermediaterangeofscalesthatcontinuallyadjustswithdCasplayinganessentialroleintheinner/outerinteraction.

(f)Thelogarithmiclayer

Asnotedin§2b,theprevalentmodelhasratherdirectconnectionstothemeanpro?leanditsoverlaplayer-basedderivationofthelogarithmicpro?le.Conversely,multiscaleanalysesoftheMMB(forchannelandCouette?ow)revealthatalogarithmicpro?leispossible,butnotbyassumingtheexistenceofanoverlaplayer.Speci?cally,throughtheuseofanadjustedReynoldsstressfunction,Fifeetal.(2005a,b)revealthattheMMBadmitsascalinglayerhierarchythatnaturallygivesrisetoalength-scaledistribution,andthatthesecharacteristiclengthsasymptoticallyscalewithy.Thus,theseanalysesprovideajusti?cation,rigorouslyfoundedintheMMB,forthedistancefromthewallscaling-basedderivationofthelogarithmicpro?le.TheyalsoidentifytheconditionsrequiredfortheMMBtoadmitanexactlogarithmicmeanpro?le.Brie?y,foreachmemberofthefamilyofadjustedReynoldsstresses(associatedwiththevalueofaparameterb),therecorrespondsalayerIII-likestructureacrosswhichtheMMBundergoesabalancebreakingandexchange.Thelength-scalesassociatedwitheachoftheselayerscanberigorouslyderivedaccordingtotheconditionthatwhentheyareusedtonormalizetheMMB,theMMBattainsaninvariantform.Itfollowsthatanexactlylogarithmicmeanpro?lewilloccurwhenthelocallynormalizedsecondderivativeoftheReynoldsstress(divergenceoftheLambvector)remainsinvariantoverarangeofy(i.e.forarangeofb).ThelayerhierarchyinitiatesnearyCZ30andterminatesaty/dx0.5(Fifeetal.2005a).Fromthisanalysis,itissurmisedthatpurelylogarithmicbehaviourisexpectedonlyoverarangeofyinteriortothesebounds,i.e.where‘endeffects’donotdisrupttheself-similarityofadjacentblayers.Ingeneral,however,thescalehierarchycoverspartoflayerII,alloflayerIIIandpartoflayerIV.Interestingly,asnotedinWeietal.(2005a),thiscoverageofdifferingmomentumbalancelayersprovidesanaturalexplanationfortheexistenceandtheobservedbreakpointsofthevaryinglogarithmic-likepro?lebehavioursnotedin§2.Itisalsoworthmentioningthesimilaritiesbetweenthismathematicalstructureandthe?ndingsregardingthehierarchicalmotionpropertiesofthelogarithmiclayer.Amongtheseisthepotentialcorrespondencebetweenagivenblayerandthestatisticalensembleofhairpinvortexpacketsthatrisetoagiveny-location.

4.Summary

AphysicalmodeloftheturbulentboundarylayerthatisbaseduponthepropertiesoftheMMBhasbeendescribedandcomparedwiththeprevalent,well-established,model.SomedistinctionsbetweentheperspectivesprovidedbythetwomodelsareInterestingly,supportforparticularlyintensevorticalmotioninteractionsatintermediatescaleisgivenbytheremarkableobservationthatasdCvariesthepeakinthedissipationspectraexhibitsthesameReynoldsnumberdependenceasdoesthewidthoflayerIII(Tsuji1999).

Phil.Trans.R.Soc.A(2007)5

836J.Klewickietal.

nowbrie?ynoted.Intheprevalentmodel,theregionwheretheeffectsofviscosityaredeemedsigni?cantshrinkslike1/dCwithincreasingdC.Inthepresentp?????model,theregionwheretheeffectsofviscosityaresigni?cantshrinkslike1=dC.Forthereasonsdiscussedin§1,theimplicationsofthisdifferencearefarreaching.TheprevalentmodelderivesperspectivesrelativetotheReynoldsstressasasourceofturbulentmomentumtransport.Thepresentmodelderivesperspectivesfromthesource/sinkpropertiesoftheReynoldsstressgradientrelativetoaffectingatimerateofchangeofmomentum.Distinctfromtheprevalentmodel,thepresentmodelcentrestheinner/outerinteractioninthetraditionallogarithmiclayerandintheregionwherethemeanviscousforcetransitionstohavingahigher-ordercontributiontothedynamics.Theprevalentmodelismostoftendiscussedinconnectionwiththeassumptionofanoverlaplayerandthecorrespondingderivationofthelogarithmicmeanpro?le.Thepresentmodelincorporatesthescalehierarchy-basedderivationadmittedbytheMMB.

Lastly,somesimilaritiesanddifferencesbetweenthescalehierarchyandtheoverlaplayerareworthnoting.Onesimilarityisthattheyconstituteaninternalintermediatelayer,insulatedfromboundaryconditioneffects,thatexhibitsself-similarbehavioursaccordingtolocaldynamics.Thenotionthatalogarithmiclayerisaninertialsublayer,however,requiresclari?cation.ThescalehierarchyexistsinlayersII,IIIandIV.AccordingtotheMMB,IIandIIIarelayerswherethemeanviscousandReynoldsstressgradientshavethesameorderofmagnitude.Thus,oftheselayers,onlyIVcanrationallybeconsideredaninertiallayer.Interestingly,recentempiricalresultsrelatingtowherethe‘true’logarithmiclayerstartsareconsistentwiththisnotion(seeWeietal.2005a).Lastly,althoughitisnotreadilyapparenthowtheoverlaplayerandthescalehierarchydescriptionsmightbeequivalent,thepossibilityremainsthatbotharevaliddescriptionsofthesamephysics.

ThisworkwassupportedbytheNationalScienceFoundationundergrantCTS-0555223andtheDepartmentofEnergyundergrantW-7405-ENG-48.

References

Adrian,R.J.,Meinhart,C.D.&Tomkins,C.D.2000Vortexorganizationintheouterregionoftheturbulentboundarylayer.J.FluidMech.422,1–54.(doi:10.1017/S0022112000001580)Bake,S.,Meyer,D.G.W.&Rist,U.2002TurbulencemechanisminKlebanofftransition:aquantitativecomparisonofexperimentanddirectnumericalsimulation.J.FluidMech.459,217–243.(doi:10.1017/S0022112002007954)

Balint,J.,Wallace,J.&Vukoslavcevic,P.1991Thevelocityandvorticityvector?eldsofaturbulentboundarylayer.Part2.Statisticalproperties.J.FluidMech.228,53–86.

Christensen,K.T.&Wu,Y.2005Characteristicsofvortexorganizationintheouterlayerofwallturbulence.InProc.FourthInt.Symp.onTurbulenceandShearFlowPhenomena,Williamsburg,VA,pp.1025–1030.

Davidson,P.2004Turbulence:anintroductionforscientistsandengineers.NewYork,NY:OxfordUniversityPress.

Davidson,P.A.,Nickels,T.B.&Krogstad,P.-A.2006Thelogarithmicstructurefunctionlawinwall-layerturbulence.J.FluidMech.550,51–60.(doi:10.1017/S0022112005008001)

DeGraaff,D.B.&Eaton,J.K.2000Reynoldsnumberscalingofthe?atplateturbulentboundarylayer.J.FluidMech.422,319–346.(doi:10.1017/S0022112000001713)

Eaton,J.K.&Johnston,J.P.1981Areviewofresearchonsubsonic?owreattachment.AIAAJ.19,1093–1102.

Phil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel837

Falco,R.E.1977Coherentmotionsintheouterregionofturbulentboundarylayers.Phys.Fluids20,S124–S132.(doi:10.1063/1.861721)

Falco,R.E.1983Newresults,areviewandsynthesisofthemechanismofturbulenceproductioninboundarylayersanditsmodi?cation.AIAApaperno.83-0377.

Falco,R.E.1991AcoherentstructuremodeloftheturbulentboundarylayeranditsabilitytopredictReynoldsnumberdependence.Phil.TransR.Soc.A336,103–129.

Fife,P.,Wei,T.,Klewicki,J.&McMurtry,P.2005aStressgradientbalancelayersandscalehierarchiesinwallboundedturbulent?ows.J.FluidMech.532,165–189.(doi:10.1017/S0022112005003988)

Fife,P.,Klewicki,J.,McMurtry,P.&Wei,T.2005bMultiscalinginthepresenceofindeterminacy:wall-inducedturbulence.MultiscaleModel.Simul.4,936–959.(doi:10.1137/040611173)

Ganapathisubramani,B.,Longmire,E.K.&Marusic,I.2003Characteristicsofvortexpacketsinturbulentboundarylayers.J.FluidMech.478,35–46.(doi:10.1017/S0022112002003270)

Ganapathisubramani,B.,Hutchins,N.,Hambleton,W.T.,Longmire,E.K.&Marusic,I.2005Investigationoflarge-scalecoherenceinaturbulentboundarylayerusingtwo-pointcorrelations.J.FluidMech.524,57–80.(doi:10.1017/S0022112004002277)

Head,M.R.&Bandyopadhyay,P.1981Newaspectsofturbulentboundarylayerstructure.J.FluidMech.107,297–338.(doi:10.1017/S0022112081001791)

Hinze,J.O.1975Turbulence.NewYork,NY:McGraw-Hill.

Jimenez,J.&delAlamo,J.C.2004ComputingturbulentchannelsatexperimentalReynoldsnumbers.InProc.15thAustralasianFluidMechanicsConference,TheUniversityofSydney,Australia,paperno.AFMC00038.

Kim,J.,Moin,P.&Moser,R.1987Turbulencestatisticsinfully-developedchannel?owatlowReynoldsnumber.J.FluidMech.177,133–166.(doi:10.1017/S0022112087000892)

Klewicki,J.C.1989Ontheinteractionsbetweentheinnerandouterregionmotionsinturbulentboundarylayers.Ph.D.dissertation,MichiganStateUniversity.

Klewicki,J.C.1997Self-sustainingtraitsofnear-wallmotionsunderlyingboundarylayerstresstransport.InSelf-sustainingmechanismsofwallturbulence(ed.R.L.Panton),pp.135–166.Southampton,UK:ComputationalMechanicsPublications.

Klewicki,J.C.&Falco,R.E.1990Onaccuratelymeasuringstatisticsassociatedwithsmall-scalestructureinturbulentboundarylayersusinghot-wireprobes.J.FluidMech.219,119–142.(doi:10.1017/S0022112090002889)

Klewicki,J.C.&Falco,R.E.1996Spanwisevorticitystructureinturbulentboundarylayers.Int.J.HeatFluidFlow17,363–376.(doi:10.1016/0142-727X(96)80001-W)

Klewicki,J.C.,Gendrich,C.P.,Foss,J.F.&Falco,R.E.1990Onthesignoftheinstantaneousspanwisevorticitycomponentinthenear-wallregionofturbulentboundarylayers.Phys.FluidsA2,1497–1500.(doi:10.1063/1.857599)

Kline,S.J.1978Theroleofvisualizationinthestudyofthestructureoftheturbulent.InCoherentstructureofturbulentboundarylayers,anAFOSR/LehighUniversityWorkshop(edsC.R.Smith&D.E.Abbott),p.1.Bethlehem,PA:LehighUniversity.

McKeon,B.&Morrison,J.2007Asymptoticscalinginturbulentpipe?ow.Proc.R.Soc.A365,771–787.(doi:10.1098/rsta.2006.1945)

McKeon,B.,Li,J.,Jiang,W.,Morrison,J.&Smits,A.2004Furtherobservationsonthevelocitydistributioninfully-developedpipe?ow.J.FluidMech.501,135–147.(doi:10.1017/S0022112003007304)

Meinhart,C.&Adrian,R.1995Ontheexistenceofuniformmomentumzonesinaturbulentboundarylayer.Phys.Fluids7,694–696.(doi:10.1063/1.868594)

Metzger,M.M.2006Lengthscalesandtimescalesofthenear-surfaceaxialvelocityinahighReynoldsnumberturbulentboundarylayer.Int.J.HeatFluidFlow27,534–541.

Metzger,M.M.&Klewicki,J.C.2001AcomparativestudyofwallregionstructureinhighandlowReynoldsnumberturbulentboundarylayers.Phys.Fluids13,692–701.(doi:10.1063/

1.1344894)

Phil.Trans.R.Soc.A(2007)

838J.Klewickietal.

Millikan,C.1939Acriticaldiscussionofturbulent?owsinchannelsandcirculartubes.InProc.FifthInt.CongressofAppliedMechanics,pp.5772–5776.NewYork,NY:Wiley.

Moin,P.,Leonard,A.&Kim,J.1986Evolutionofacurvedvortex?lamentintoavortexring.Phys.Fluids29,955–963.(doi:10.1063/1.865690)

Morris,S.,Stolpa,S.,Slaboch,P.&Klewicki,J.Inpress.Nearsurfaceparticleimagevelocimetrymeasurementsinatransitionallyrough-wallatmosphericboundarylayer.J.FluidMech.

Morrison,J.,McKeon,B.,Jiang,W.&Smits,A.2004Scalingofthestreamwisevelocitycomponentinturbulentpipe?ow.J.FluidMech.508,99–131.(doi:10.1017/S0022112004008985)

Moser,R.D.,Kim,J.&Mansour,N.N.1999Directnumericalsimulationofturbulentchannel?owuptoRtZ590.Phys.Fluids11,943–945.(doi:10.1063/1.869966)

Offen,G.R.&Kline,S.J.1974CombineddyestreakandHydrogen-bubblevisualobservationsofaturbulentboundarylayer.J.FluidMech.62,223–239.(doi:10.1017/S0022112074000656)Osterlund,J.,Johansson,A.,Nagib,H.&Hites,M.2000Anoteontheoverlapregioninturbulentboundarylayers.Phys.Fluids12,1–4.(doi:10.1063/1.870250)

Panton,R.L.1997Self-sustainingmechanismsofwallturbulence.Southampton,UK:ComputationalMechanicsPublications.

Panton,R.L.2005Reviewofwallturbulenceasdescribedbycompositeexpansions.Appl.Mech.Rev.58,1–36.(doi:10.1115/1.1840903)

Perry,A.E.&Abell,C.J.1977Asymptoticsimilarityofturbulencestructuresinsmooth-andrough-wallpipes.J.FluidMech.79,785–799.(doi:10.1017/S0022112077000457)

Perry,A.E.&Chong,M.S.1982Onthemechanismofwallturbulence.J.FluidMech.119,173–217.(doi:10.1017/S0022112082001311)

Perry,A.E.&Marusic,I.1995Awall-wakemodelfortheturbulencestructureofboundarylayers.Part1.Extensionsoftheattachededdyhypothesis.J.FluidMech.298,361–388.(doi:10.1017/S0022112095003351)

Perry,A.E.,Henbest,S.&Chong,M.S.1986Atheoreticalandexperimentalstudyofwallturbulence.J.FluidMech.165,163–199.(doi:10.1017/S002211208600304X)

Pope,S.B.2000Turbulent?ow.Cambridge,UK:CambridgeUniversityPress.

Prandtl,L.1904Uber?ussigkeitsbewegungenbeisehrkleinerreibung.InVerhandlungendesDrittenInternationalenMathematiker-KongressesinHeidelberg1904(ed.A.Krazer),pp.484–491.Leipzig,Germany:Teubner.

Prandtl,L.1925BerichtuberdieEntstehungderTurbulenz.Z.Agnew.Math.Mech.5,136–139.Priyadarshana,P.&Klewicki,J.C.2003Reynoldsnumberscalingofwalllayervelocity–vorticityproducts.InReynoldsnumberscalinginturbulent?ow(ed.A.J.Smits),pp.117–122.Princeton,NJ:KluwerAcademicPublishers.

Priyadarshana,P.,Klewicki,J.,Treat,S.&Foss,J.2007Statisticalstructureofturbulentboundarylayervelocity–vorticityproductsathighandlowReynoldsnumbers.J.FluidMech.570,307–346.(doi:10.1017/S0022112006002771)

Rajagopalan,S.&Antonia,R.A.1993Structureofthevelocity?eldassociatedwiththespanwisevorticityinthewallregionofaturbulentboundarylayer.Phys.FluidsA5,2502–2510.(doi:10.1063/1.858763)

Robinson,S.K.1990Coherentmotionsintheturbulentboundarylayer.Annu.Rev.FluidMech.23,601–639.(doi:10.1146/annurev.?.23.010191.003125)

Schlichting,H.1979Boundarylayertheory.NewYork,NY:McGraw-Hill.

Smith,C.R.,Walker,J.D.A.,Haidari,A.H.&Sobrun,U.1991Onthedynamicsofnear-wallturbulence.Phil.Trans.R.Soc.A336,131–175.

Smits,A.J.&Wood,D.H.1985Theresponseofturbulentboundarylayerstosuddenperturbations.Annu.Rev.FluidMech.17,321–358.(doi:10.1146/annurev.?.17.010185.001541)Spalart,P.1988DirectsimulationofaturbulentboundarylayeruptoRqZ1410.J.FluidMech.187,61–98.(doi:10.1017/S0022112088000345)

Sreenivasan,K.R.1989Theturbulentboundarylayer.InFrontiersinexperimental?uidmechanics,vol.46(ed.M.Gad-el-Hak).SpringerLectureNotesinEngineering,pp.159–209.Berlin,Germany:Springer.

Phil.Trans.R.Soc.A(2007)

Turbulentboundarylayermodel839

Tennekes,H.&Lumley,J.1972A?rstcourseinturbulence.Cambridge,MA:MITPress.

Theodorsen,T.1952Mechanismofturbulence.InProc.2ndMidwesternConf.onFluidMechanics,pp.1–19.Columbus,OH:TheOhioStateUniversity.

Thomas,A.S.W.&Bull,M.K.1983Ontheroleofpressure?uctuationsindeterministicmotionsintheturbulentboundarylayer.J.FluidMech.128,283–322.(doi:10.1017/S002211208300049X)Tomkins,C.D.&Adrian,R.J.2003Spanwisestructureandscalegrowthinturbulentboundarylayers.J.FluidMech.490,37–74.(doi:10.1017/S0022112003005251)

Townsend,A.A.1976Thestructureofturbulentshear?ow.Cambridge,UK:CambridgeUniversityPress.

Tsuji,Y.1999Peakpositionofdissipationspectruminturbulentboundarylayers.Phys.Rev.E59,7235–7238.(doi:10.1103/PhysRevE.59.7235)

Wallace,J.M.1982Onthestructureofturbulentshear?ow.Apersonalview.InDevelopmentsintheoreticalandappliedmechanics(edsT.J.Chung&G.Karr),pp.1–12.Huntsville,AL:DepartmentofMechanicalEngineering,UniversityofAlabama.

Wark,C.E.&Nagib,H.M.1991Experimentalinvestigationofcoherentstructuresonturbulentboundarylayers.J.FluidMech.230,183–208.(doi:10.1017/S0022112091000757)

Wei,T.,Fife,P.,Klewicki,J.&McMurtry,P.2005aPropertiesofthemeanmomentumbalanceinturbulentboundarylayer,pipeandchannel?ows.J.FluidMech.522,303–327.(doi:10.1017/S0022112004001958)

Wei,T.,McMurtry,P.,Klewicki,J.&Fife,P.2005bMesoscalingoftheReynoldsshearstressinturbulentchannelandpipe?ows.AIAAJ.43,2350–2353.

Wosnik,M.,Castillo,L.&George,W.K.2000Atheoryforturbulentpipeandchannel?ows.J.FluidMech.421,115–145.(doi:10.1017/S0022112000001385)

Zagarola,M.V.&Smits,A.J.1997Scalingofthemeanvelocitypro?leforturbulentpipe?ow.Phys.Rev.Lett.78,239–242.(doi:10.1103/PhysRevLett.78.239)

Phil.Trans.R.Soc.A(2007)