ebook img

Computational and Experimental Flow Field Analyses of Separate Flow Chevron Nozzles and Pylon Interaction PDF

12 Pages·2003·7.7 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Computational and Experimental Flow Field Analyses of Separate Flow Chevron Nozzles and Pylon Interaction

AIAA 2003–3212 Computational and Experimental Flow Field Analyses of Separate Flow Chevron Nozzles and Pylon Interaction Steven J. Massey Eagle Aeronautics, Inc., Hampton, VA 23669 Russell H. Thomas Aeroacoustics Branch, MS 166, NASA Langley Research Center, Hampton, VA 23681 Khaled S. Abdol-Hamid Configuration Aerodynamics Branch, MS 499, NASA Langley Research Center, Hampton, VA 23681 Alaa A. Elmiligui Analytical Services & Materials, Inc., Hampton, VA 23666 9th AIAA/CEAS Aeroacoustics Conference and Exhibit May 12–14, 2003/Hilton Head, SC Forpermissiontocopyorrepublish,contacttheAmericanInstituteofAeronauticsandAstronautics 1801AlexanderBellDrive,Suite500,Reston,VA20191–4344 Computational and Experimental Flow Field Analyses of Separate Flow Chevron Nozzles and Pylon Interaction StevenJ.Massey∗ EagleAeronautics,Inc.,Hampton,VA23669 RussellH.Thomas† AeroacousticsBranch,MS166,NASALangleyResearchCenter,Hampton,VA23681 KhaledS.Abdol-Hamid‡ ConfigurationAerodynamicsBranch,MS499,NASALangleyResearchCenter,Hampton,VA23681 AlaaA.Elmiligui§ AnalyticalServices&Materials,Inc.,Hampton,VA23666 Acomputationalandexperimentalflowfieldanalysesofseparateflowchevronnozzlesispresented. The goalofthisstudyistoidentifyimportantflowphysicsandmodelingissuesrequiredtoprovidehighlyaccurate flowfielddatawhichwilllaterserveasinputtotheJet3Dacousticpredictioncode. Fourconfigurationsare considered:abaselineroundnozzlewithandwithoutapylon,andachevroncorenozzlewithandwithoutapy- lon.Theflowissimulatedbysolvingtheasymptoticallysteady,compressible,Reynolds-averagedNavier-Stokes equationsusinganimplicit,up-wind,flux-differencesplittingfinitevolumeschemeandstandardtwo-equation k−εturbulencemodelwithalinearstressrepresentationandtheadditionofaeddyviscositydependenceon totaltemperaturegradientnormalizedbylocalturbulencelengthscale.ThecurrentCFDresultsareseentobe inexcellentagreementwithJetNoiseLabdataandshowgreatimprovementoverpreviouscomputationswhich didnotcompensateforenhancedmixingduetohightemperaturegradients. Introduction from the isolated case, the aerodynamic and noise reduc- Typically for modern commercial aircraft, engine in- tion performance of the device or strategy. This focus on stallation involves the engine-under-the-wing or the tail boththeaerodynamicaswellasacousticinteractioneffects mounted configuration. However, even within those two of installation and propulsion airframe aeroacoustics, will basic categories there are many design variables includ- becomemoreimportantasnoisereductiontargetsbecome ing structural, aerodynamic, and operational factors. For moredifficulttoachieve. thesestandardengineinstallationsanddesignrequirements Theworkinthispaperisacontinuationofanon-going theindustryhashighlydevelopeddesignandanalysispro- efforttostudytheeffectofthepylononseparateflownoz- cesses.1 zles including those equipped with the chevron jet noise The impact on the net radiated noise of the aircraft is reduction device. Building on preliminary work,2 the pri- an additional challenge that is emerging as increasingly mary objective of this work is to improve the capability importantinpropulsionairframeintegrationandisthegen- to simulate installed jet configurations and to develop un- eral motivation for the work reported here. One example derstanding of the characteristics of the flow field of the of an aeroacoustic effect of propulsion airframe integra- selected installation configurations through computational tion could be the effect of the pylon on the development solutionscomparedwithflowfieldexperimentsperformed of the exhaust plume and on the resulting jet noise. An- at the NASA Langley Jet Noise Laboratory. In parallel otherarea ofinstallation effectonnoise is theimpact that research, Hunter3 is developing an installed jet noise pre- installationhasonnoisereductiondevicessuchaschevron diction method that uses the flow field information from nozzles. Many noise reduction devices are studied funda- thesecomputationalsolutionsandcompareswithacoustic mentally on isolated components. In general, installation experimentsthathavealsobeenperformedattheJetNoise effectsmustbeconsideredbecausetheseeffectsmayalter, Laboratory on these same configurations. Together, these ∗SeniorResearchScientist,AIAAMember. three tools are being used to develop an understanding of †AerospaceEngineer,SeniorMemberAIAA. theacousticinteractionbetweenpylonandchevronnozzles ‡AerospaceEngineer,AssociateFellowAIAA. andthemoregeneralinteractionofpylonandjetwiththe §SeniorScientist. aim of developing flow and acoustic prediction capability Copyright(cid:176)c 2003bytheAmericanInstituteofAeronauticsandAstronautics, Inc. NocopyrightisassertedintheUnitedStatesunderTitle17,U.S.Code. The andnoisereductiontechnologyforinstalledjetconfigura- U.S.Governmenthasaroyalty-freelicensetoexerciseallrightsunderthecopyright tions. claimedhereinforGovernmentalPurposes. Allotherrightsarereservedbythe copyrightowner. 1OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 and pylon design combined with no added design feature forthejunctionsimilartothejunctioninthepreviouscon- figuration4. ExperimentalFlowConditions Surveys of the flow fields of configurations 1, 3, and 4Fareusedtocomparewiththeresultsofthesimulations in subsequent sections. The surveys were performed in theNASALangleyLowSpeedAeroacousticWindTunnel a)Config.1 b)Config.6 with a traverse apparatus moving a single rake of probes (see Figure 2). Mean total pressure and total temperature were measured with a rake that was populated with three total pressure and three total temperature probes and tra- versed in cross-sectional planes at x/D = 2, 5, 10, and 17.Ateachcross-sectionagridwasmappedwithmeasure- mentstakentypicallyin0.64cmincrementsatx/D=2and 5andin1.27cmincrementsatx/D=10and17.Thegrids were sized at each axial station to cover most of the core andthefanplume. Asaresult,severalhundredincrements andsamplesweretakeninordertocompleteasurveyata c)Config.3 d)Config.4F single axial station. Maintaining the core and fan jet op- Fig.1 Surfaceandsymmetryplanegridlines. erating condition was critical to obtaining a good survey. Table1liststhemeanvaluesofthecoreandfanoperating Configurations conditions and the standard deviation of those conditions The baseline configuration is a separate flow nozzle of foratypicalsurveyatasingleaxialstation,indicatingthat bypassratiooffivewithanexternalplug. Thenozzleand theoperatingconditionswereinfactmaintainedwellover pylon designs are from a nozzle study performed by Mc- the more than an hour of run time required for each axial Donnell Douglas (now Boeing) in 1996 and represents a station. generic design. The chevrons were designed for the core nozzleusingguidelinessimilartothoseusedintheNASA Table1 Meanandstandarddeviationvaluesforthecoreand AdvancedSubsonicTransportprogram. Thechevronsare fanoperatingconditionsduringthesurveyofconfiguration3 atx/D=5. designed to penetrate into the core flow by approximately thethicknessoftheboundarylayer.Thetrailingedgeofthe Mean StandardDev. baselinenozzleischosentocorrespondtothemid-pointof NPRCore 1.557 0.004 thechevronaxiallength. Thefourconfigurationsanalyzed T Core 826.1K 3.3K t inthisstudyare: NPRFan 1.747 0.005 T Fan 358.7K 0.8K Configuration 1: Baseline round core and fan nozzle t withnopylon(Figure1a). ComputationalFlowConditions Configuration 6: Baseline core and fan nozzle with Theflowconditionsforbothexperimentsandcomputa- pylon(Figure1b). tions are set at the take-off cycle point. The free stream Configuration3: Eightchevroncorenozzleandbase- Mach number is 0.28 with total pressure and tempera- linefannozzlewithnopylon(Figure1c). ture set to standard atmospheric conditions at sea level of 101353Paand 295Krespectively. Thefanpressureratio Configuration4F:Eightchevroncorenozzleandbase- is1.75withtotaltemperatureof359K.Thecorepressure line fan nozzle with pylon with the tip of a chevron ratiois1.56withtotaltemperatureof828K. alignedwiththepylon(Figure1d). NumericalMethod The configuration numbering scheme is consistent with previouslyusedconfigurationnumbers.2 Thecurrentcon- Thefluidflowissimulatedbysolvingtheasymptotically figurations 1 and 3 are identical in geometry to the same steady, compressible, Reynolds-averaged Navier-Stokes configurationssimulatedpreviouslyonlyinthisworkthey equationsusinganimplicit,up-wind,flux-differencesplit- are simulated with the results of the improved computa- ting finite volume scheme and standard two equation k-ε tionalformulationtobedescribedbelow. Thepyloninthe turbulence model4 with a linear stress representation and currentworkhasa divergingshelfangle of 1.5◦ (configu- theadditionofaneddyviscositydependenceontotaltem- rations6and4F)whilethepyloninthepreviousconfigu- peraturegradientthroughthevariationtheC closurecoef- µ rations2,4,and5hadacurvedshelfline. Thejunctionin ficient,seeAbdol-Hamidetal.5 Allcomputationsareper- configuration4Fissimplytheresultofaseparatechevron formedusingtheparallel,multiblock,structuredgridcode 2OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 Fig.2 Photograph,lookingupstream,oftheflowfieldsurveyexperimentintheNASALangleyLowSpeedAeroacousticWind Tunnel.Configuration1isinstalledonthejetenginesimulatorsurroundedbythewindtunnelnozzle.Rakeofprobes(centerof photo)isseenmountedonatraversemechanism. PAB3D.6 Viscous diffusion terms are modeled as uncou- chevron nozzle without pylon provide confidence that the pledintheflowdirectionandfullycoupledinthecrossflow azimuthaldirectionisalsofullyresolvedwith40cellsper directionanda3-factorschemeisusedfortheapproxima- chevron or 1.125◦ resolution. Grid sizes range from less tionofimplicitterms. than 300,000 cells for a 6◦- two cell wedge grid to over Grid sequencing is used to accelerate convergence by 16 million cells for the full 180◦ chevron/pylon grid, see solving 1/4 then 1/2 of the grid in each of the three com- Table 2 and Figure 1. All surfaces are gridded and run as putational directions. The general solution procedure fol- viscous with the exception of the lower bifurcator and a lowedistosolvetheregionnearthenozzleviatimemarch- verysmalltipregionoftheplug. Theaveragevalueofthe ing,thensolvetheplumeblocksviaspacemarching,then lawofthewallcoordinate,y+,ofthefirstcellcenterisap- finally run at least 200 time marching iterations on just proximately one with a minimum of 0.4 and a maximum theplumeblocksandadjacentupstreamblocksinorderto offive. Thesolutiondomainextendsoversixcorediame- smooth any inflections at the time/space marching block ters,(D = 12.8cm)upstreamofthefanexit,nearly32D boundary. Typical iteration counts for 3-D grids at each downstream and six D radially. In all plots, the origin of gridlevelare5,000, 4,000, 3,000and2,000. Axisymmet- the x-axis, x/D = 0.0, is set to the fannozzle exit, which ric cases are stiffer and require 10,000 iterations per grid putsthecoreexitatx/D=0.83andtheplugtipatx/D= level. A sample convergence history, Figure 3, in terms 1.87. of flow and turbulence residual is shown for four grid se- quence levels, where 0.5 denotes half of the spacing used Typicalruntimesforthefullyconvergedsolutionatthe fortheproductionruns. Convergedsolutionsforcenterline finegridlevelarelessthan40hoursfor16.2millioncells totaltemperature,showninFigure4,areessentiallyidenti- using up to 21 2-GHz Pentium 4 nodes on an unbalanced calatallgridlevelsanddemonstratenumericalverification grid. Balanced computations using 8 CPU’s for the eight for the axial and radial directions. Consistency between plumeblockscontaining600,000cellseach,yieldparallel theaxisymmetricsolutionsandtheperiodicsolutionsofthe efficienciesofupto97%oraspeedupfactorof7.8. 3OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 Modifiedk−εModel Thek−εmodelisthemostpopulartwo-equationturbu- 10›5 10›7 lencemodelandiswidelyusedacrossmanyflowregimes, qResidual eventhoughitwasdevelopedandtunedprimarilyfortwo- kResidual dimensional, incompressible flows. Thus, it is suspected 10›6 10›8 thatobservederrorsinresolvinghightemperaturejetmix- ing is a result of the model being applied outside of its al10›7 44 al envelope. The physical mechanism for the missing quan- Residu 22 11 10›9Residu toiftythoeftduernbsuilteynctomnitxriibnugtiiosnthtooutghhetKtoelbveind-uHeetlomthhoeltazbisnesntcae- q10›8 k bility. Experimentally7,8 ithasbeenshownthatthespatial .5.5 growthofatwo-dimensionalshearlayerisproportionalto 10›10 thesquarerootofthedensityratioofthetwomixingfluids. 10›9 InarecentstudybyTamandGanesan,9adensitybasedcor- rectionwasappliedontopoftheThiesandTam10modified 10›10 10›11 k−ε model, however it is strictly intended for free shear 0 20000 40000 60000 80000 flows and therefore not applicable to the current configu- Iterations rations. Withinstalledjetsitisparticularlyimportantthat themodelapplytobothfreeshearandwallboundedflows. Fig.3 Config. 1-axi,RoundNozzle:Residualhistoryatfour Forthepurposeofthepresentstudy, aminimallyinvasive levelsofgridrefinement. correctiononlytoturbulenteddyviscosityisusedtorectify themixingdeficiencyforhightemperaturejets. Toinsure thatthemodifiedk−εmodelreturnstotheoriginalmodel forcoldjetsandthatitremainsaccurateforwallbounded 1 flows,turbulencescaledtotaltemperature,T¯,ischosenas the variable for the additional eddy viscosity dependence, Coarse(44) where 0.9 Medium(22) Fine(11) k3/2(cid:107)∇T (cid:107) e) ExtraFine(.5.5) T¯ = t . (1) or0.8 JNLDATA ε Tt c ( TT The functional relationship was determined by the best T/T0.7 match to experimental data for the baseline round nozzle e rlin andisasfollows, e nt0.6 (cid:183) (cid:184) Ce C =C 1+ CTT¯3 (2) 0.5 µ µ0 1+CMMt2 wherethestandardk−εclosurecoefficientisC =0.09. 0.4 µ0 0 5 10 15 20 25 30 Thetemperaturegradientcoefficient,C ,issetto24.33by T X/D bestfittodataafterassigningtheturbulenceMachnumber coefficient, C to 4. Turbulence Mach number, M , is Fig.4 Config.1-axi,RoundNozzle:Centerlinetotaltemper- M t definedby atureatfourlevelsofgridrefinement. M2 =2k/a2 (3) t where a is the speed of sound. For this study no attempt Table2 Zonalgridcellcounts. ∗Config. 1-axiisaxisymmet- wasmadetoconstructageneralfunctionaldependenceon ricandfullytimemarched. turbulenceMachnumber,henceEq.2shouldonlybeused Time Space for subsonic jet flows. However, this has now been suc- Marched Marched cessfullyaddressedinAbdol-Hamidetal.5 Config. x/D ≤14 x/D >14 Total 1−axi∗ 2x148,464 2x148,464 Results 1 4,083,072 1,797,120 5,880,192 6 7,805,440 3,594,240 11,399,680 The three issues to be addressed in this section are as 3 10,125,312 3,594,240 13,719,552 follows;1)therectifiedCFDresultsusingthemodifiedk− 4F 12,620,096 3,594,240 16,214,336 ε model versus the original CFD results, 2) the validation ofCFDresultswithexperimentaldata,and3)explanation ofthephysicalcharacteristicsoftheflows. 4OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 1 1:BaselineRound(stdk›e) 1:BaselineRoundNozzle 6:RoundNozzlew/Pylon 0.9 3:ChevronNozzle Config.1,RoundNozzle. 4F:ChevronTipunderPylon 1:Data 3:Data 0.8 4F:Data e) or c Config.6,RoundNozzlew/Pylon. T(t0.7 / Tt 0.6 Config.3,ChevronNozzle. 0.5 0.4 0 5 10 15 20 25 30 Config.4F,ChevronTipunderPylon. X/D Fig.6 CFD:C factoronsymmetryplane. µ Fig.5 Totaltemperaturealongjetaxis. CFDRectification The effect of the temperature gradient modified k − ε model over the standard k − ε model for the round noz- Config.1,RoundNozzle:Standardk-εmodel. zle is to increase turbulent mixing and hence decrease the potential core length by 3.4 core diameters, D. The core lengthsaretabulatedinTable3andaredefinedforthex/D inwhichTt = 0.995Tt∞ alongtheaxisofthejet,seeFig- Config.1,RoundNozzle. ure 5. The action of the modified k −ε model in terms Table3 CFDPotentialCoreLength Config. Corelength,D Config.6,RoundNozzlew/Pylon. 1-stdk−ε 12.5 1-modk−ε 9.1 6 8.1 Config.3,ChevronNozzle. 3 4.6 4F 5.4 Config.4F,ChevronTipunderPylon. Fig.7 CFD:Totaltemperatureonsymmetryplane. oftheincreasededdyviscosityclosurecoefficientisshown in Figure 6, where the ratio of the actual to the standard value of 0.09 is plotted. In the code, the maximum ratio issetto5,however,mostoftheflowfieldsarewellbelow 3. Asaresultoftheundisturbedshearlayers,whichcreate strong temperature gradients, the round jet configuration Config.1,RoundNozzle:Standardk-εmodel. exhibitsthehighestamountofC amplification. CFDso- µ lutionsofthegeneralflowfieldaredepictedfortheround nozzle, Figures 7-10, with the standard k −ε model fol- lowedbyallconfigurationswiththemodifiedk−εmodel. Config.1,RoundNozzle. Note, for configuration 3., the symmetry plane was plot- ted for the solution rotated by 22.5◦ to be consistent with experimentalcrossflowplots. Config.6,RoundNozzlew/Pylon. CFDValidation As previously discussed, the present model was tuned on the axisymmetric grid to the round nozzle experimen- Config.3,ChevronNozzle. tal results. Particular attention was paid to matching the axial total temperature at x/D = 5 and 10. The model was then frozen for the other cases. Computed total tem- peraturealongtheaxisisplottedwithexperimentaldatain Config.4F,ChevronTipunderPylon. Fig.8 CFD:Totalpressureonsymmetryplane. Figure 5. Comparing the data points at x/D of 5 and 10, configuration 1 is on data, configuration 3 is less than 1 5OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 a result of the under prediction of mixing. For configura- tion1,T isinnearperfectagreementandp iswithinone t t contourlevelofagreement,seeFigures11and12.Forcon- Config.1,RoundNozzle:Standardk-εmodel. figuration3showninFigures14and15,themixinglobes total temperature and pressure differ by only two contour levels. With the addition of the pylon, configuration 4F Config.1,RoundNozzle. shows a slightly larger deviation from measured data, see Figures16and17. Focusingonthechevroninducedcross- flowplumesatx/D = 5,T istwotofourlevelsoff,with t thehighestdifferencebeinginasmallregionnearthelower Config.6,RoundNozzlew/Pylon. bifurcator. Thehigherthanexpectedtemperaturenearthe lower bifurcator is likely a result of not modeling it as a viscous surface, hence there is a lower level of turbulent Config.3,ChevronNozzle. mixing in this region. For total pressure, the contours are againseentobetwolevelshigherthanmeasured. Theper- cent error in terms of the contour range is summarized in Table 4. Calculated mass flow is also in good agreement Config.4F,ChevronTipunderPylon. Fig.9 CFD:Machnumberonsymmetryplane. withmeasureddata,seeTable5.Experimentally,therewas somediscrepancybetweentwomethodsofmeasurementof up to 0.36 kg/sec, however, in both sets of measurements the trend was the same, so the comparison of normalized valuesisvalidandinexcellentagreementwithCFD. Config.1,RoundNozzle:Standardk-εmodel. Table4 RelativeCFDCrossflowError Config. T Error p Error t t 1 ≈0% <5.5% Config.1,RoundNozzle. 3 12.0% 11.0% 4F 12.0-24.0% 11.0% Config.6,RoundNozzlew/Pylon. Table 5 Mass flow Comparison. ∗Config. 6 was estimated basedonarearatio. Config. CFD Exp. Error Config.3,ChevronNozzle. 7.835kg/s 7.902kg/s 1 1.000 1.000 0% 6 0.940 0.94∗ 0% 3 0.993 0.994 -0.1% Config.4F,ChevronTipunderPylon. 4F 0.939 0.936 0.3% Fig.10 CFD:Turbulentkineticenergyonsymmetryplane. D under mixed, and configuration 4F is between 1 and 2 D undermixed. Relativeerrorscomparedtothepredicted FlowPhysics roundcorelengtharelessthan11%forconfiguration3and Thepurpose of creating a chevroncore nozzle is to en- 11%to22%forconfiguration4F. hancemixingandhencereducenoisewithverylittlethrust Incomparisonswithcrossectionaldatafortotaltemper- loss. Preliminarycalculationsforthecurrentchevronnoz- ature and pressure, see Figures 11 - 17, the thickness and zlesshowathrustlossoflessthanonepercent. Toclearly shapeoftheshearlayersareseentobeinexcellenttogood showthemixingeffectsofthechevroncoreandthepylon, agreementwithmeasureddata. Note,noexperimentaldata a non-dimensionalized, mass averaged turbulence kinetic was taken for configuration 6. Also, for configuration 3. energy and total temperature expressions are introduced, theCFDsolutionwasrotatedintheazimuthaldirectionby showninFigures18and19respectively. Theexpressions 22.5◦ to align it with the experimental data. For the total follow those published by Kenzakowski et al.11 and are temperature plots, the contour range extends from 0.39 to definedasfollows: 0.99withalevelspacing,∆Tt,of0.04. Fortotalpressure, (cid:82) (cid:113) therangeisfrom1.00to1.72withspacingof∆pt =0.04. kρudA In the experimental data plots, the data points were taken kˆ = A(cid:82) q (4) ρudA at ∆x and ∆y = 0.1D intervals and are marked with a (cid:82) A smallblackdot. Thistendstosmearoutthinshearlayers, φρudA Tˆ = (cid:82)A (5) butdoesnotaccountforthemismatchwithCFD,whichis t ρudA A 6OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 Experiment CFD Experiment CFD T /T p /p t tCORE t t∞ Fig. 11 Config. 1, Round Nozzle: Total temperature cross Fig.12 Config. 1, RoundNozzle: Totalpressurecrosssec- sectionsatx/Dof2,5,10and17. tionsatx/Dof2,5,10and17. where flow velocity vectors at x/D = 2, shown in Figure 20. T −T φ= t tFAN (6) Intheseplots,uniformlyspacedvectorswereplottedwith T −T tCORE tFAN the length proportional to the magnitude over contours of withintegrationonlyperformedoverareaswhere the axial velocity component, u. In both cases, as the py- lonandplugcloseout,thecrossflowvelocityvectorsshow 0.005≤φ≤1. (7) flow moving in the direction to fill in the wake. For the From the figures it is clear that the introduction of the roundnozzlewithoutapylon,anoverallinwardflowisob- chevron core nozzle greatly enhances turbulent mixing, servedduetothesecondaryflowsetupbytheplug. With while the pylon installation contributes to the mixing by theadditionofthepyloninconfiguration6,astrongerover- onlyasmallamountfortheroundnozzleandevenlessfor all vertical flow toward the pylon is setup. Also, because the chevron nozzle. Chevrons are seen to be an excellent the pylon is also tapered in the horizontal direction, an method to enhance mixing without introducing much tur- additional horizontal velocity is induced. This flow then bulenceinthenear-fieldcomparedtotheroundnozzlewith separatesatthesharpbottomedgeofthepylonandrollsup pylonandactuallyreducingitinthefar-field,seeFigure18. into a small vortex. For the chevron nozzle without a py- The mechanism for the enhanced mixing due to both lon, configuration3, anoctagonalshapedinfluenceonthe pylon and chevrons can be seen by examining the cross- plug flow at the jet center is seen in the crossflow as well 7OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 Tt/TtCORE,CFD pt/pt∞,CFD Experiment CFD T /T t tCORE Fig.13 Config.6,RoundNozzlew/Pylon:Totaltemperature andpressuresectionsatx/Dof2,5,10and17. Fig.14 Config. 3,ChevronNozzle: Totaltemperaturecross sectionsatx/Dof2,5,10and17. as the axial flow. However, the strongest effect by far is carriesturbulenceenergy2fromboththeboundarylayerof onthecore/faninterface. Note,inthisfigure,configuration thepylonaswellasthecore/fanshearlayer,itisexpected 3 has not been rotated as in the previous plots, therefore tobeanadditionalsourceofnoise. its chevrons are oriented as shown in Figure 1c, and are ConcludingRemarks aligned with configuration 4F, where the chevron tip is at the z/D = 0. It is observed that the chevrons allow the The purpose of this study was to provide insight into core flow to stream out in a strong crossflow plume much the flow physics of pylon-jet interactions and to provide likeamushroomcloud,wherethehotcoreflowrushesup the input mean flow field to acoustic prediction methods whiledrawingthecoolerflowdownarounditsbase. When for installed jet noise. Important flow phenomenon were theeffectofthepylonisaddedinforconfiguration4F,two identifiedandthemodelingtechniquesrequiredtoprovide notableeffectsareobserved;1)becauseofthepylonblock- highlyaccuratemeanflowfielddataspecificallyforuseas age, no fan flow can be drawn down near the pylon, and inputtoacousticpredictionmodelswereestablished. Four 2)thevortexonthecrossflowplumeadjacenttothepylon configurations were considered; a baseline round nozzle constructively interferes with the separation vortex com- with and without a pylon, and a chevron core nozzle with ing from the bottom edge of the pylon and forms a much andwithoutapylon.Theflowwassimulatedbysolvingthe stronger vortex than in configuration 6. Since this vortex asymptotically steady, compressible, Reynolds-averaged 8OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212 Experiment CFD Experiment CFD p /p T /T t t∞ t tCORE Fig.15 Config. 3,ChevronNozzle: Totalpressurecrosssec- Fig.16 Config.4F,ChevronTipunderPylon:Totaltemper- tionsatx/Dof2,5,10and17. aturecrosssectionsatx/Dof2,5,10and17. Navier-Stokesequationswithatemperaturegradientmodi- less than 11% for configuration 3 and 11% to 22% for fiedtwo-equationk−εturbulencemodel.ThecurrentCFD configuration 4F. This study has also clearly shown that results are seen to be in good to excellent agreement with installation effects are significant and introduce local and JetNoiseLabdataandshowgreatimprovementoverprevi- global changes to the flow. Of critical importance is the ouscomputationswhichdidnotcompensateforenhanced interface between the chevron induced vortical flow and mixingduetohightemperaturegradients. Intermsofthe the bottom edge of the pylon which in the configuration potentialcorelength,whichistheclearestsingleindicator 4F interferes constructively to produce a potential source of error, the new round nozzle computations are in com- of noise. With a firm understanding of the flow physics plete agreement with experimental data. This is in sharp androbustnumericalmodel,futureresearchwillaimtode- contrasttoinitialcomputationsusingtheunmodifiedk−ε signmoreaffectivenoiseattenuatingnozzlemodifications, model, which over predicted potential core length by 3.4 whilefurtherrefiningCFDpredictioncapabilities. corediametersor37%.Currentresultsforthechevroncore Acknowledgments nozzlewithoutthepylonshowapotentialcorelengtherror oflessthanonecorediameter. Theinclusionofthepylon ThecontributionsoftheentireJetNoiseLabtothePy- increasestheerrortobetweenoneandtwocorediameters. lon Effects Experiment are gratefully acknowledged. The Relative to the round nozzle core length, these errors are authorsalsowishtothankS.PaulPaoandCraigHunterof 9OF11 AMERICANINSTITUTEOFAERONAUTICSANDASTRONAUTICSPAPER2003–3212

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.