AeronauticalSciences InternationalCouncilofMe ofMe ... · PDF file0.007 0.006 0-005 0.004-0.002 0 0 006 0 004 0 002 0 0 006 0.004 Cf 0 002 0465, Rs -i5x 106 NM_ 3111 section 0 02 0.4

IC AS Paper No. 68-09

DETAILED EXPLORATION OF THE COMPRESSIBLE,VISCOUS FLOW OVER TWO-DIMENSIONAL AEROFOILSAT HIGH REYNOLDS NUMBERS

by

M. C. P. Firmin and T. A. CookAerodynamics DepartmentRoyal Aircraft EstablishmentFarnborough, U. K.

TheSixthCongressofMe

InternationalCouncilofMeAeronauticalSciences

DEUTSCHES MUSEUM, MUNCHEN, GERMANY/SEPTEMBER 9-13, 1968

Preis: DM 2.00

DETAILED EXPLORATION OF THE COMPRESSIBLE,VISCOUS FLOW OVER

TWO DIMENSIONAL AEROFOILS AT HIGH REYNOLDS NUMBERS

M. C. P. Firminand T. A. CookRoyalAircraftEstablishment,Farnborough,Hampshire,England

Abstract

Wind-tunnelinvestigationsintotheflowovertwo-dimensionalaerofoilsare discussedwithparti-cularemphasison the influenceof viscouseffectson aerodynamiccharacteristics.The experimentalworkwas undertakenusinglarge-chord-modelsspan-ning tunnelsboth withsolidand slottedwallsattheRoyalAircraftEstablishment.ThemeasurementsweremadewithReynoldsnumbersup to 15 x 106 andspeedsup to high subsonicwithmodelshavingaspan-to-chordratioof about3.

Measurementsof boundarylayerand wakedevelopmentare presentedwithmoredetailedmeasurementsin theregionof thetrailing-edgeofthe wings. The resultsare comparedwithmethodsof estimation,and wind tunnelwallcorrectionsarediscussed.

Introduction

The designof aircraftforflightat highsub-sonicspeedshas resultedin shapesthathaveasweptwinglayout. In general,thesweepanglesaresuchthattheflowoverthe wingsis sensiblysimi-lar to thatovertwo-dimensionalsections,andcurrenttrendswouldsuggestthatthissituationwillcontinuebut withsectionsthatmorecloselyapproachtheboundariesimposedby viscosityandcompressibility.As it is notyet possibleto makeexactcalculationsof theviscouscompressibleflowaboutaerofoils,predictionmethodsrelyon soundexperimentalbacking; yet as we explainin thispaper,evenexperimentson two-dimensionalsectionsare by no meansas simpleas thetheoreticalmodelmightimplysincetheyrequireparticularlycarefulplanningandmeticulousattentionto detail.

In thepast,it has notbeenpossibleat speedswherecompressibilitywas significant,to determinethe influenceof viscosityon thepressuredistri-butionoveran aerofoilas no generalsqlytionforpotential-flowexisted. RecentlySells0) hasobtainednumericalsolutionsfor theexactinviscidcompressibleflow overaerofoilsectionsat sub-criticalspeeds. Thisis an importantstepforwardand meansthatany differencebetweenmeasuredandtheoreticalpressuredistributions,suchas thatshownin Figure1 at highsubsonicspeeds,mustnowbe limitedto theviscouseffectswhichwe wishtoinvestigateandany undesirableenvironmentaleffectsdue to the conditionsunderwhichmeasure-mentsare made. Theseenvironmentalproblemsincludetheconstraininginfluenceof thewind-tunnelwallsand viscouseffects,bothof whichmaydetractfrom the two-dimensionalityof thetests.Theseproblemshaveto be overcomebeforetheinfluenceof theviscouseffectson two-dimensionalaerofoilscanbe determinedaccurately.An exampleof someexperimentalresultsis givenin Figure1and theyhavebeencorrectedfor theincreasedvelocityoverthe aerofoildue to thepresenceofthe wind-tunnelwallswhenthe aerofoilis at zerolift (i.e.theblockagecorrection)but notforotherwind-tunnelwallcorrections,whichwouldincreasethe differencein the quotedliftcoef-ficientsby about0.03assumingfirstorder

theoreticalwall correctionsapply. The inviscidflowcalculationsfor thisfigurehaveinvolvedanextrapolationfrom justbelowthe criticalspeedwhichcannotbe justifiedon purelytheoreticalgrounds,but sincethemeasurementsthemselvesarewhollysubcriticalthisis consideredreasonable.

iI

iei

s.

......,

c.;(64.1.0)

.

Cp

0

'

C0040

f

CL0 441

4

.2

I I

0

0.1

0.4 011" 'la IC

. a

04

a

la 04wrimental 40044

04

Inviacid f km (Sane)

14 0445Re if X se

Mal. 101 Section

04

rt

Fig I Typical pressure distribution

Viscouseffectsof course,occurdue to flowinregionsof high shearsuchas in theboundarylayerand in the wake. At highReynoldsnumbers,theflowin theseviscousregionsis usuallyturbulentandconsequentlynot amenableto theoreticaltreatmentunlesssimplifyingassumptionsaremade. Theviscouslayershave thusto be determinedexperi-mentally. It shouldbe possiblehoweverto treattheselayersas normalboundarylayersandwakeswithconstantstaticpressurenormalto thesurface,exceptnearthe trailingedgeof an aerofoi;wheretheflow curvaturemaybe large. nchemannk2)evensuggestsit is possiblefor theinfluenceof vorti-cityin the curvedflownearthetrailing-edgetocauseearlierseparationthanwouldotherwisebeexpected.Consequently,detailedmeasurementsof theflowin sucha regionareparticularlydesirable.

Whenconsideringtheoverallforceson an aero-foil,the contributionto theliftfrom thefric-tionalforcescan usuallybe neglectedand theliftdeterminedfromsurfacepressuresalone,but fordragthefrictionalforcesareimportant.The aero-foildragis made up in twoparts; theformdrag,obtainedfrom the surfacepressureson thewinganddue to the displacementeffectof theboundarylayer,and thefrictionaldragwhichhas to be measuredbysomeothermeans. Waketraversemethods,in whichconvenientassumptionsaremadeabouttheflow,canbe usedto determineoveralldrag,but doubtsstillremainabouttheaccuracyof suchmethodsso directmeasurementsof the dragarealsodesirable.

There was, therefore,a need to design anexperiment in which as much information as possibleis obtained about the detailedflow characteristics.

1

The workmustobviouslybe at highReynoldsnumbers(sothatboundarylayertransitionproblemsarereducedas muchas possible)andat thehighsub-sonicspeedsrelevantto presentsweptwingdesigns.Alsoit is essentialthattheconditionsunderwhichthe wingsare testedbe knownaccurately,inparticulartheeffectsof windtunnelwallinter-ference,whichare knownto increaserapidlyat thehighersubsonicspeeds,mustbe determinedcarefully.

In thispaperwe willgivedetailsof aselectionof themeasurementsmadeso far at R.A.E.,indicatewheredifficultiesin measuringtechniqueshaveoccurredand makecomparisonswithexistingtheorieswherethisis useful. A completeanalysisof all thesemeasurementshasnotyetbeenmadeandfurtherexperimentalworkstillneedsto be done.Someof the theoriesused,particularlyfor theboundary-layerdevelopment,willtendtobe some-whatparochialsincecomputerprogrammesareusuallyinvolvedand onlythosereadilyavailableto us havethusbeenused.

ActualMeasurements

For measurementsof the tyTementionedabove,windtunnelswithlargeworking-sectionsarerequiredto enablereasonablylargechordmodelstobe usedwithoutinvokingserioustunnelinterferencedifficultiesas regardsboth chord-to-heighteffectsand end-wall(i.e.low aspect-ratio)effects. Withlargechordmodels,accurateanddetailedboundary-layerand wakesurveyscanbe madeat realistically-highReynoldsnumbersand increasedprecisionis possiblein modelmanufacture.Ourworkwas,in fact,done in the8 ft x 8 ft (2.4m x2.4m) closed-walltunnelat Bedfordand the 8 ft x6 ft (2.4m x 1.8m) slotted-walltunnelatFarnboroughusingmodelsof up to 76 cm chord. Thus,both themodelspan-to-chordand tunnelheight-to-chordratioswere3 (orgreater)whilethe trailing-edgeboundary-layerthicknessesweretypically1 to3 cm.

In the 8 ft x 8 ft (2.4m x 2.4m) tunnel,awingwasmountedas shownin Figure2. Boundarylayerand wakemeasurementsweremadeusingrear-mountedtraversingprobes. The wingwas splitintosevensimilarspanwisepartseachindividuallymounted,so thatit was possibleto makeoverallforcemeasurementson a centralsegmentof the wingusinga straingaugebalance. The smallgapsbetweenthe segmentsweresealedotherthanwhenoverallforcemeasurementswerebeingmade. In the 8 ft x6 ft (2.4m x 1.8m) tunnelthe wingwasmountedverticallyas shownin Figure3. Boundary-layerandwakeprobesweremountedfromwithinthe thicknessof wingand traversedby remotelycontrolleddrives.Enlargementsare shownof the probesprotrudingfromthe wing. Measurementsin thewakefurtherdown-streamthan10% of thechordbehindthe wingweremadeusingrear-mountedtraversingprobes. Thisarrangementmadesure thatthe influenceof theprobesand theirsupportson theflow averthe wingwaskept to an absoluteminimumand thattheloca-tionof the probeswasknownaccuratelyevenformeasurementsin the wakecloseto the trailing-edgeof the wing. The wallslotarrangementwas alteredfor thiswork,the open-arearatioof thewallsparallelto a spenwisegeneratorbeingvariedbyusingadditionalplatesbehindtheusualslots,andtheslotsin thewallsnormalto a spanwisegeneratorwereclosedexceptfor a regionwell

downstreamof thewing. Anotherwingwas testedwiththe samesectionbut havingabouthalfthechordso thattunnelinterferenceeffectsoouldbeinvestigated.The smallerwingdidnothavepro-visionfor makingboundarylayermeasurements,however.

Fig.2: Aerofoil mounted in 8 ft x 8 ft 12,4 m x 2,4 m I wind tunnel

Fig.3: Aerofoil mounted in 8 ft x 6 ft (2,4 m x 1,8 m) wind tunnel

showing remotely controlled boundary layer probes

In all theworkcomprehensivemeasurementsofsurfacepressuresweremadewhilesurfacetubetechniqueswereusedfor skinfrictionmeasurements.

On themodels,boundary-layertransitionprob-lemshaveas far as possiblebeenavoidedby fixingtransitionwitha sparsedistributionof smallspheres('ballotini')stuckontothe wingsurfacein a thinband at about5%chordbackfromtheleadingedge. At theReynoldsnumbersof the teststhe diameterof thesphereswasverysmallcomparedwiththe wingchord,a diameter-to-chordratioof0.00015beingtypical.Thisapproachavoidedthelocationof the transitionregionhavingto bedeterminedfor eachcombinationof free-streamMachnumberand wingincidence,and it alsoensuredtransitionwouldoccurneartheleading-edgeof thewingsat an almostconstantlocationon the chord,thushelpingto ensuretwo-dimensionalflow.

2

Although there is some uncertaintyabout thetransitionprocess itself, it is thoughtthatinterms of the chord the boundary-layerflow sqtlesdown to a truly turbulent layerquite quickly0)after transit