JOURNALOFSPACECRAFTANDROCKETS Vol.50,No.1,January–February2013 Assessment of Laminar, Convective Aeroheating Prediction Uncertainties for Mars-Entry Vehicles BrianR.Hollis∗ NASALangleyResearchCenter,Hampton,Virginia23681 and DineshK.Prabhu† ERCCorporation,MountainView,California94035 DOI:10.2514/1.A32257 An assessment of computational uncertainties is presented for numerical methods used by NASA to predict 1 laminar, convective aeroheating environments for Mars-entry vehicles. A survey was conducted of existing 218 experimentalheattransferandshock-shapedataforhigh-enthalpyreacting-gasCO2flows,andfiverelevanttest 1.6 serieswereselectedforcomparisonwithpredictions.Solutionsweregeneratedattheexperimentaltestconditions 14/ usingNASAstate-of-the-artcomputationaltoolsandcomparedwiththesedata.Thecomparisonswereevaluatedto 5 0.2 establishpredictiveuncertaintiesasafunctionoftotalenthalpyandtoprovideguidanceforfutureexperimental OI: 1 testingrequirementstohelplowertheseuncertainties. D g | or aa. Nomenclature suchnewmissions.Thus,anEntry,Descent,andLandingSystems http://arc.ai LHH=w0D === lwtiofatta-lltloef-rndetrehasagtlrperyaa,tmiJo=ekngthalpy,J=kg emAnnatrsaysly-vcsieashpiaScbtliueldiatyyrc.h(TEithDeecLstue-SrehAsig)thhwa-mtacsaoscusol,dnmdpuarorcsvt-eieddnet[rt1yh]estryhesaqttuemidreedfi(HnineMcdrMetahEsreSede) 15, 2013 | Kkbc == cebqmauc6ki=lwigbamrriudomlr2e(cid:2)acscotniostnanrat,te—cooerffigmcieonl=t,ccmm33=gmol(cid:2)sor vsaurecphheiicrtlseeocntuwicrietishnflaraaetar1bi)gleiadameraioeddrroyasnnhgaemelli;lcift2d-)etoc-eadlerlaraagrtgorera,tsioyhsy(teLpme=rDs(oH(cid:3)nIiAc0D:4aS–n0da:/n6od)r H CENTRE on March qRkRMfweC11;D ===== ffhcf(cid:2)orroe1eerareewnUtsseattt1rrrrraeedDrnaaammrsd=efi(cid:3)eauMRrc1ster,iaayomctnnheoranladtutstemhncebuoewmerfabfilelc,riWebna=ts,mecdm2o3n=gdmiaomle(cid:2)tser, CSsctvhuoaIorAprmenieoeDcbruesiSmnsop)antautcifiicanooolmnrrvseibhelotliihrungfoisahtcphtt-lrrieeaaoosltnpteiciosutleuanllsdsestihsemoeenfserao.eneroatrMsrfoetwbhiaxresearehrekdeominwasailghdnsr;ooc-Lwhaacisn=ontdeDnacsart3icusda)hreeeiarhrtsereoycdcspht.thheuIietarnrelsetlcFo,stnicugoii1rcn.ne–1flss3a,ai,nstth4tadae–b/nsoo8ldeefr. EARC RRmNax == mheamxiimspuhmerircaadlinuos,semradius,m asieornodsyynstaemmicardeescheolewrantoinr,Faignsd.2h–y4p.ersonic/supersonic retropropul- Y RES TU11 == ffrreeeessttrreeaammtveemlopceitrya,tumre=,sK uncTeortaiinnstuierseinthtehesmafoedtyelinagndandsupcrceedsisctioonf osfutchheamerisostihoenrms,odthye- LE z = Cartesiandistancefromnosealongsymmetryplane,m namic environment, which includes the surface heat transfer, NG (cid:4) = angleofattack,deg pressure and shear, and the integrated aerodynamic forces and LA (cid:5) = coneangle,deg momentsthattheentryvehiclewillexperience,mustbedefined.A ASA (cid:2)1 = freestreamdensity,kg=m3 wstehpichtowwaarsdpefurflofirlmlinegdtahsisparretqoufireamlaerngterisacptrievsiteynt[e2d]isnpotnhsisorsetdudbyy, N y NASA’sFundamentalAeronauticsProgram,HypersonicsProjectto b ed I. Introduction define,andultimatelyreduce,aerothermodynamicuncertaintiesfor d oa TWO important goals for NASA’s future Mars exploration severalhypersonicmissiontypes.Avarietyofphysicalphenomena wnl programsaretoperformaroboticsamplereturnmissionand thatinfluencetheHMMES aerothermodynamic environmenthave Do toenablehumanexplorationmissions.Bothgoalsrequirethesafe beenidentified.Theseinclude,butarenotlimitedtothefollowing: landing of much greater masses (in excess of 10 t) than those boundary-layer transition; TPS blowing, ablation, and roughness; of previous Mars missions. However, the heritage technology shock-layer radiation; flexible structure/flowfield interactions; employed in past NASA missions from Viking to Mars Science retropropulsion thruster and reaction control system jet/flowfield Laboratory (MSL), that of a rigid, 70 deg sphere-cone, ablative and jet/surface interactions; noncontinuum effects at high altitude thermalprotectionsystem(TPS),willnotbesufficienttoaccomplish andinseparatedwakeflows;surfacecatalysis;nonequilibriumgas kinetics, etc. A discussion of many of these phenomena and of computationalmethodsusedformodelingthemispresentedin[3]. Presented as Paper 2011-3144 at the 43rd AIAA Thermophysics Conference, Honolulu, HI, June 27–30, 2011; received 19 October 2011; Thesensitivitiesofthesecomputationalmethodstoinputparameters revisionreceived27April2012;acceptedforpublication13May2012.This for numerical models of physical phenomena (as distinct from material is declared a work of the U.S. Government and is not subject to uncertainties determined through comparison with experimental copyrightprotectionintheUnitedStates.Copiesofthispapermaybemadefor data,whichisthesubjectofthisreport)havebeenexploredin[4]. personalorinternaluse,onconditionthatthecopierpaythe$10.00per-copy The current study is limited to uncertainties in the prediction feetotheCopyrightClearanceCenter,Inc.,222RosewoodDrive,Danvers, of laminar, attached high-enthalpy flow of CO over smooth, MA01923;includethecode0022-4650/13and$10.00incorrespondencewith nonablating surfaces. These restrictions define 2the most basic the∗CACerCo.space Engineer, Aerothermodynamics Branch. Associate Fellow validationcase(shortofperfect-gasflow)relevanttoMarsmissions. Withoutathoroughunderstandingofthecomputationaluncertainties AIAA. †Senior Research Scientist, Aerothermodynamics Branch; currently inthemodelingofthesephenomena,uncertaintiesinmodelingother NASAAmesResearchCenter,MoffettField,California94035.Associate phenomena, such as turbulence, roughness, and radiation, cannot FellowAIAA. properlybeaddressed. 56 HOLLISANDPRABHU 57 1 8 1 2 6 1. 4/ 1 5 2 0. 1 OI: D g | or a. a ai c. ar p:// htt 3 | Fig.1 HMMESarchitectureoptions. 1 0 2 5, 1 h arc The approach followed to define the uncertainties for laminar, M on attachedhigh-enthalpyCO2flowswasasfollows: E 1) Identify relevant sources of experimental data (i.e., high- TR enthalpysurfaceandflowfieldtestdataobtainedinCO flows).For N 2 E simplicity,only0degangle-of-attackdatawereconsidered. C H 2) Generate flowfield predictions at the selected test conditions C R usingthestate-of-the-artcomputationaltoolscurrentlyemployedin A E thedevelopmentofNASA’sMarsmissions.Althoughcomputations S RE may have been previously performed for some data sets using Y older software packages, itisimportant to emphasize that, for the E GL purposesofthecurrentstudy,newsolutionsweregeneratedforall AN cases.Thenewsolutionsweregeneratedtoensurethataconsistent A L methodology was applied using the latest software versions. AS Similarly, although other similar software tools exist outside of y N NASAandhavebeenemployedtomodelsuchflows,theywerenot d b consideredinthisstudy. e d 3)Performandassesscomparisonsbetweentheexperimentaldata a nlo Fig.2 Mid-L=Drigidaeroshell. and numerical results to determine the current state of the art in w o D Fig.3 Inflatableaerodynamicdecelerator. Fig.4 Supersonicretropropulsionsystem. 58 HOLLISANDPRABHU datathatdidnotmeetthecurrentcriteriawerealsonotedforfuture consideration(e.g.,turbulentdata,high-enthalpyairorN data). 2 A. ExperimentalDataSetsforHMMESLaminar,Convective UncertaintyAssessment Five sources of laminar high-enthalpy, convective heat transfer data in a CO environment were identified for inclusion in this 2 study. These data were obtained in the NASA Ames Research Center(ARC)42-InchShockTunnel,theGeneralAppliedSciences Laboratories(GASL)HYPULSEExpansionTube,theCaltechT5 Reflected Shock Tunnel, the University of Illinois Hypervelocity Expansion Tube, and the Calspan University of Buffalo Research Center(CUBRC)LENSIReflectedShockTunnel.Alldataarefrom 70degsphere-conemodelgeometries(Fig.5)similartothatofMSL, althoughthereissomevariationinnoseandcornerradiusdimensions (relativetothemaximumradius)ofthemodelstested. Test conditions from each of these studies are listed in 81 chronological order in Table 1 and a summary of thewind-tunnel 1 62 modelparametersforeachtestisprovidedinTable2.Notethatthe 1. 4/ conditionslisteddonotnecessarilyrepresentallthedataobtainedin 1 25 eachfacility;forsimplicity,thedataconsideredhereinwerelimited 10. to0degangleofattackatlaminarconditions.Additionaldatafor OI: Fig.5 Seventydegreesphere-conegeometry. nonzeroangleofattackandfortransitionalandturbulentboundary D g | layersarealsoavailableinmanyofthesedatasets.Itisalsoimportant or to note that these test conditions were actually determined from a aa. computational predictions for laminar high-enthalpy, convective combination of experimental flowfield measurements (e.g., shock c.ai aerothermodynamicenvironmentsofMarsmissions. speedandpitotpressure)andnumericalsimulationtools(toobtain http://ar artItwisilelxpbeectuesdedthaatsthgeuriedseullitnseosftfhoirsefufftourrtetoedffeofirtnsetthoersetdatuecoeftthhee mphaysssicfaralcmtioodnesl)s.aTshdeosethneuflmoewrificealldtcooodlsesewmhpilcohyasroem,ientohfeothrye,stoambee 3 | computationaluncertainties.Reductionofthemodelinguncertain- validated by comparison with the experimental data. Thus, a 201 ties will require the acquisition of new high-fidelity experimental completevalidationexercise,whichisoutsidethescopeofthisstudy, ch 15, gnraomuincds-praonpderfltiiegsh(ts-tuersftacdeathaeaotningboatnhdmpraecsrsousrceo,pfliocwafieerlodthsetrrumcotudrye-, wmooudledl-rienq-fluoirweisteimrautilvaetiorenfis.nementsoftestconditionpredictionsand Mar aerodynamicsforcesandmoments,boundary-layertransition,etc.) Theenthalpy/Reynoldsnumberrangeofthetestdataisshownin on and fundamental physical properties (chemical, vibrational, and Fig.6.Totalenthalpies(H (cid:5)H )rangedfrom1.9(neartoperfect RE electronicexcitationandrelaxationrates,transport properties, and gas)to12:3 MJ=kg(noneq0uilibrwiaullmchemistry).Reynoldsnumbers NT radiation emission and absorption rates). Concurrent development (basedondiameter)foralldataconsideredwerelessthan106,which H CE awnidllvaalsloidbateiorenqoufiraeddv.ancedcomputationalmethodsandalgorithms resulted in laminar conditions for all cases except one partly C transitionalrunintheCUBRCdataset. R A For comparison with flight conditions, reference values for E ES II. ReviewofExperimentalDataApplicable HMMES mid-L=D and HIADS configurations are shown on this R Y toHMMESMissions plot,aswellasreferencevaluesfortheMarsScienceLaboratoryand GLE Athoroughliteraturesearchwasperformedtoidentifysourcesof MarsVikingmissions.ItisnotablethattheMSLreferencecondition N isatthehighendoftherangeofReynoldsnumberandenthalpytest A experimentaldatarelevanttotheHMMESuncertaintyassessment. d by NASA L Dtcirnoaanentxadsfciteseireostsnsmsowefwa1ehsrueMerreeJsm=eplkoeegnsct)tts.sehAdowlcftehkororecuhcgpeohemnrstfihiocdearemlarrcaeettdauiocantilin,oconipnmsuwrpweoehsrCieictOihgoe2nsnuoeterrffasattthceeegdaM(hsi.eaeaar.-tt, chnmoiuginmshsdebiiotreinroi.nniIssne,wncwteohlhnaletlwrrpaeyisatths,aintnthhdetehapeHnerMaoaknrMdgheeeEraoStoiffnctgoemsncatdogcintnoidinotidnutiisdoteinaorgnfeorser.aattfheaercMtionraRrosefVy2nikotoilndg3s wnloade tthiaenaktimneotsicpsheorfeiCsO(cid:4)295a:r3e%eCxOpe2c,t2e:d7%toNb2,eanthde2.g0r%eatAesrt(buynvcoerlutaminet)y, 1. NASAARC42-InchShockTunnelData o componentbecauseofitspredominance,andsodatasetswithN (or Data were obtained [5] at laminar test conditions on a 70 deg D 2 otherminorcomponents)werelefttolaterconsideration.Additional sphere-cone model in CO , air, and CO -Ar environments in the 2 2 Table1 Testconditions Facility Facilitytype (cid:2)1,kg=m3 T1,K U1,m=s Re1;D H0(cid:5)Hw, Freestreammassfractions MJ=kg [CO ] [CO] [O ] [O] 2 2 NASAAmes42-InchShock Reflectedshocktunnel 3:10(cid:6)10(cid:5)4 200 4150 0:25(cid:6)106 8.5 1.000 0.000 0.000 0.000 Tunnel GASLHYPULSE(runs747,749) Shockexpansiontube 5:79(cid:6)10(cid:5)3 1088 4772 0:93(cid:6)106 12.3 1.000 0.000 0.000 0.000 CaltechT5(run2256) Reflectedshocktunnel 5:70(cid:6)10(cid:5)2 1407 2732 1:67(cid:6)106 6.1 0.831 0.108 0.061 0.000 CaltechT5(run2254) Reflectedshocktunnel 3:12(cid:6)10(cid:5)2 1828 3367 1:25(cid:6)106 10.6 0.550 0.287 0.151 0.012 CaltechT5(run2255) Reflectedshocktunnel 7:83(cid:6)10(cid:5)2 2188 3514 2:26(cid:6)106 11.3 0.592 0.259 0.139 0.009 CUBRCLENSI(series1,run12) Reflectedshocktunnel 5:26(cid:6)10(cid:5)4 691 2761 0:29(cid:6)106 4.9 0.901 0.063 0.036 0.00 CUBRCLENSI(series2,run16) Reflectedshocktunnel 1:48(cid:6)10(cid:5)2 361 1907 0:94(cid:6)106 1.9 1.000 0.000 0.000 0.000 CUBRCLENSI(series2,run12) Reflectedshocktunnel 8:70(cid:6)10(cid:5)3 931 2939 0:77(cid:6)106 6.0 0.843 0.100 0.057 0.000 CUBRCLENSI(series2,run08) Reflectedshocktunnel 8:96(cid:6)10(cid:5)3 892 2870 0:21(cid:6)106 5.6 0.863 0.087 0.05 0.000 CUBRCLENSI(series2,run13) Reflectedshocktunnel 6:06(cid:6)10(cid:5)3 1116 3373 0:16(cid:6)106 8.6 0.687 0.199 0.110 0.003 UniversityofIllinoisHET Shockexpansiontube 1:44(cid:6)10(cid:5)2 1172 3058 0:43(cid:6)106 5.7 1.000 0.000 0.000 0.000 HOLLISANDPRABHU 59 Table2 Wind-tunnelmodelinformation Facility Max Nose Corner Cone Modelinstrumentation Model Refs. radius,in. radius,in. radius,in. angle,deg material NASAAmes42- A(cid:3)2:74 A(cid:3)1:5 A(cid:3)0:039 70 Surfacejunctionthermocouples Copper [5] InchShock B(cid:3)2:82 B(cid:3)1:5 B(cid:3)0:167 Tunnel C(cid:3)2:99 C(cid:3)1:5 C(cid:3)0:470 GASLHYPULSE 1.00 0.5 MP1(cid:3)0:05 70 Thin-filmtemperaturegauges Macorceramic [6–8] MP3(cid:3)0:10 MP4(cid:3)0:20 CaltechT5 3.50 1.75 0.35 70 Fast-responsecoaxialthermocouples SS304stainless [9,10] steel CUBRCLENSI, 12.0 6.0 0.60 70 Thin-filmtemperaturegaugesandcoaxial Stainlesssteel [11,12] series1 thermocouples CUBRCLENSI, 6.0 3.0 0.30 70 Coaxialthermocouples,thin-film Stainlesssteel [13] series2 temperaturegauges,andsilver calorimeters Universityof 1.00 0.5 0.05 70 Fast-responsecoaxialthermocouples AI2024andA2 [14] IllinoisHET toolsteel 1 8 1 2 6 1. 514/ NASAARC42-InchShockTunnel(alsosometimesreferredtoasthe Tunnel in CO2 and N2 test gases.The modelwas fabricated from 0.2 Ames3.5ftShockTunnel).Thefacility(sincedemolished)wasa stainless steel and instrumented with fast-response coaxial DOI: 1 Hcoematbtursatniosfne-rhdeaatteada,nrdeflfleocwtefideslhdoscckhtluienrneenliwmitahge1s0–w2e0remosbtteasitnteimdeosn. tahnedrm1o6coduepglesa.nTdesltaimnginwara,stpraenrfsoitrimonedal,ataanndgletusrobfulaetntatckdaotaf0w,1er1e, g | copper models instrumented with surface thermocouples. Three produced,althoughonlylaminar0degrunsareconsideredherein. a.or separateconfigurationswereactuallytested.Allwere70degsphere- Althoughaspecificvalueforexperimentaluncertaintywasnotcited a ai cone geometries with 1.5 in. nose radii; however, the model had for the CO data of [9], error bars shown on the plots were arc. interchangeableoutersectionswithvariouscornerradiithatresulted approximate2ly(cid:7)11%. p:// in2.74–2.99in.maximumradiusmodels.Besidesthe0degangle- 13 | htt o(cid:4)f(cid:3)-att1a0ckanrdun2s0cdoengs.iAdenreedxpheerriemine,natadlduitniocenratlairnutnysowfe(cid:7)re10p%erfworamsceidteadt 4. CUBRCLENSIReflectedShockTunnelData 15, 20 fortheheatingdatafromthisstudy. intMheeaCsUurBeRmCenLtsEoNnSMIRSLeflgeecotemdeStrhyomckoTdeulnsnwele.rTewpeortfeosrtmseerdieisnwCeOre2 ch performed:thefirstwithathermocoupleandthin-filminstrumented, Mar 2. GASLHYPULSEExpansionTubeData 12in.maximumradiusmodel[11,12]andthesecondwitha6in. on Testingwasperformed[6–8]intheGASL(nowATK)HYPULSE maximumradiusmodelinstrumentedwiththin-filmgauges,coaxial E R ExpansionTubeon70degsphere-conemodelsof1.00in.maximum thermocouples, and silver calorimeters [13]. Both models were T N radiusinairandCO testgases.Laminardatawereobtainedonthree fabricated from stainless steel. In addition to the 0 deg angle-of- E 2 C 70degsphere-conemodelsofvaryingcornerradiiandonablunted attackdataconsideredherein,runswerealsomadeat(cid:4)(cid:3)11,16,and CH hyperboloid with 70 degasymptotes. The models were fabricated 20 deg that produced laminar, transitional, and turbulent flows. R A from Macor ceramic and instrumented with thin-film heat-flux Although specific values for experimental uncertainties for each E ES gauges. In addition to the (cid:4)(cid:3)0 data considered herein, data are gaugetypewerenotgiven,errorbarsshownonplotsin[13]were Y R also available for 4 and 8 deg angles of attack. An experimental approximately(cid:7)5to(cid:7)10%. LE uncertainty of (cid:7)11% was cited for the heating data from this NG study. 5. UniversityofIllinoisHypervelocityExpansionTubeData A A L Themostrecentdataavailable[14]wereobtainedintheUniversity AS 3. CaltechT5ReflectedShockTunnelData ofIllinoisHypervelocityExpansionTube(HET).Thetestgeometry y N Heattransfertestswereperformed[9,10]ona3.50in.maximum wasa1.00in.maximumradius,70degsphere-conestainlesssteel d b radius,70degsphere-conemodelintheCaltechT5ReflectedShock model instrumented with fast-response coaxial thermocouples. e d Laminar data were obtained at angles of attack of 0, 11, and a nlo 16 deg. An uncertainty of (cid:7)8% was cited for the coaxial gauge w o measurementsbutanoverallexperimentaluncertaintywasnotgiven. D B. OtherRelevantSourcesofHigh-EnthalpyConvective AeroheatingData Other sources for high-enthalpy aeroheating data sets exist that may be of use in defining uncertainties in nonequilibrium kinetic modelseventhoughthesedatawerenotobtainedinpureCO flows. 2 These include, as previously noted, CO -Ar and air data from the 2 ARC42-InchShockTunnel[5],airdatafromGASL-HYPULSE[6– 8],andN datafromCaltechT5[10]. 2 Other blunt-cone aeroheating data were obtained in several facilitiesaspartoftheNATOAGARDWorkingGroup18activity [15].Althoughtheprimarypurposeofthisprogramwastoobtain rarefied wake flow data [16], high-enthalpy forebody aeroheating datawerealsogeneratedintwoofthetests[17]:thoseconductedat DLR-high enthalpy shock-tunnel Göttingen (DLR-HEG) and CUBRCLENS. The DLR-HEG tests was conducted on a 70 deg sphere-cone modelwithhigh-enthalpyrunsinair,CO ,andCO -N mixtures. 2 2 2 Fig.6 Comparisonofgroundtestandflightconditions. Unfortunately,explicitinformationregardingfreestreamspeciemass 60 HOLLISANDPRABHU Table3 Forwardreactionratecoefficients No. Reaction A B C Type Third-partymultiplier Refs. 1 CO(cid:10)M!C(cid:10)O(cid:10)M 2:3(cid:6)1020 (cid:5)1:00 1:29(cid:6)105 Dissociation [28] 2 CO(cid:10)O!O (cid:10)C 3:9(cid:6)1013 (cid:5)0:18 6:92(cid:6)104 Exchange [29] 2 3 CO (cid:10)M!CO(cid:10)O(cid:10)M 1:5(cid:6)1025 (cid:5)2:50 6:60(cid:6)104 Exchange (cid:6)2forM(cid:3)atoms [30] 2 4 CO (cid:10)O!O (cid:10)CO 2:1(cid:6)1013 0.00 2:78(cid:6)104 Dissociation [28] 2 2 5 O (cid:10)M!2O(cid:10)M 2:0(cid:6)1021 (cid:5)1:50 5:94(cid:6)104 Dissociation (cid:6)5forM(cid:3)atoms [29] 2 concentrations is incomplete in the documentation available IV. AssessmentofExperimental–Computational [18–20].Becauseitwasnotwithinthescopeofthisstudytoperform Comparisons computationstodeterminefacilityoperatingconditions,thesedata A. SurfaceHeatTransferComparisons werenotanalyzed.Ifsuchinformationcouldbeobtained,thesedata Comparisons were made between the LAURA and DPLR shouldbefactoredintofuturework. predictions and the experimental heat transfer data for each test TheCUBRCLENSAGARD-18testwasconductedona70deg condition. These comparisons are shown in Figs. 7–17 and are sphere-cone model in air and N at enthalpies between 5 and 181 10 MJ=kg. Data were obtained [221] on both the blunted-cone odradtae,rendoitnjutesrtmosnofainfaccreilaistyin-bgyt-oftaaclileitnythbalapsyis.ovInerethaechwfihoglueres,etthoef 2 forebodyandinthewakeonaninstrumentedsting. 1.6 experimentaldataareshownwitherrorbarsof(cid:7)15%.Thisvalueis 14/ purely a nominal figure used as a consistent reference and is not 5 0.2 III. ComputationalMethods derivedfrom the published reports; the actual stated experimental OI: 1 Flowfield solutions were generated at each test condition using uncertainties varied from approximately 5 to 15% as discussed aa.org | D NnhuyAmpSeerArsi’ocsnaLilcAsUimaReuArloaatthinoednrmDtoPodLoylRsnacemomdiecpslo.TycehodemspebuyctoadNtieoAsnSasrAetahfnoedrtwcooenxpttrieinnmusuaivrmye ptchoreenvdniiootiunoscnlayt.aoFlpyottirioctn,hsef.uclolLymAcpUautRtaaAltyitoinacn,aldarnedDsuPsluLtspR,eprrcreaedtsaiucllyttistoicnwsweaarrelelsbghoeounwnenrdaaflorlyyr c.ai documentation and background material are available (e.g., [22– ar p:// 25])foreachcode.Althoughtherearesomealgorithmicandphysical htt modeling differences between them, both are structured-grid 3 | Navier–Stokes solvers for flows with vibrational and chemical 1 20 nonequilibrium.Pastcode-to-codecomparisonshavedemonstrated 15, goodagreementbetweenthemforentryvehicleaerothermodynamic ch simulations relevant to HMMES, such as for the Mars Science ar M Laboratory[26]andFireII[27]missions. on In this study,the physicalmodel employed was that of laminar RE flow,withtwo-temperature(translationalandvibrational/electronic) NT representationandnonequilibriumchemicalkinetics.Afive-reaction E C chemistry model for the CO -CO-O -O-C system was used, with CH forward reactions as defined2by the2modified Arrhenius form of R A Eq.(1)andbackwardratesdeterminedfromtheequilibriumconstant E S definitionvia Eq. (2). The forward reaction coefficients are listed E Y R inTable3.Thecarbonatomproducingreactions(nos.1and2)were E included in this set for completeness; however, these reactions L NG were found to be negligible for the range of test conditions LA considered: A (cid:2) (cid:3) S A cm3 d by N kF(cid:3)A(cid:6)TB(cid:6)exp(cid:8)(cid:5)C=T(cid:9) units of gmol-s (1) F1:i9g.M7J=CkUgBcRasCe.LENS I Reflected Shock Tunnel: Series 2, run 16, e ad (cid:2) (cid:3) (cid:2) (cid:3) Downlo kB(cid:3)KkFC units of gmcmol62-s or gmcmol3-s (2) For the comparisons with experimental data, the freestream boundary conditions were taken from the published facility test conditions. Thewall temperature boundary condition was set to a constant300Kvalue;althoughheattransferratesforthesecasescan be very high, the facility run times were in the millisecond to microsecond ranges, over which time the increase in wall temperaturewasnegligiblewithrespecttotheboundary-layeredge temperatures. A surface catalysis boundary condition was also required.Severalsurfacecatalysismodeloptionswereemployed:a noncatalytic option, a fully catalytic option (forces atoms to recombinetohomogenousdiatomics)anda“supercatalytic”option, in which recombination to 100% CO is enforced at the surface. 2 Thechoiceofwallcatalysismodelhasbeenshowntohavealarge effectonsensitivityofaeroheatingpredictions[4]but,unfortunately, asdiscussedin[3],therearelittledataavailableatlowtemperatures (such as generated in these tests) on catalysis effects for CO 2 reactions to help justify the selection of a particular catalysis Fig.8 CUBRC LENS I Reflected Shock Tunnel: Series 1, run 12, model. 5:1MJ=kgcase. HOLLISANDPRABHU 61 1 8 1 2 1.6 Fig.9 CUBRC LENS I Reflected Shock Tunnel: Series 2, run 8, Fig.12 Caltech T5 Reflected Shock Tunnel: Run 2256, 6:1MJ=kg 14/ 5:6MJ=kgcase. case. 5 2 0. 1 OI: D g | or a. a ai c. ar p:// htt 3 | 1 0 2 5, 1 h c ar M n o E R T N E C H C R A E S E Y R Fig.10 UniversityofIllinoisHETExpansionTube:5:7MJ=kgcase. Fig.13 ARC42-InchShockTunnel:8:5MJ=kgcase. E L G AN consistentwithdifferencesbetweenthembeingmuchsmallerthan With the exception of the GASL HYPULSE case (which is the A L thedifferencesproducedbyselectionofthewallcatalysisboundary highest enthalpy condition), the current comparisons between AS condition. Thus, to maintain clarity on the plots, only the fully predictions and measurements are similar to the comparisons y N catalyticDPLRresultswillbeshownalongwiththeLAURAresults previouslypublishedforeachstudy.FortheGASLHYPULSEcase d b foreachboundarycondition. (Fig.17),thereisalargeoverpredictionofthedatainthecurrentstudy e d a o nl w o D Fig.11 CUBRC LENS I Reflected Shock Tunnel: Series 2, run 12, Fig.14 CUBRC LENS I Reflected Shock Tunnel: Series 2, run 13, 6:0MJ=kgcase. 8:6MJ=kgcase. 62 HOLLISANDPRABHU 1 8 1 2 1.6 Fig.15 CaltechT5ReflectedShockTunnel:Run2254,10:6MJ=kg 14/ case. 5 2 0. Fig.18 Comparisonofnoncatalyticpredictionswithdata(averaged 1 OI: errormagnitude). D g | a.or were determined from the definition of the equilibrium constant a c.ai (e.g., [32]). When the chemistry models employed in [6–8] were 3 | http://ar sdouanbteasmtwitouadsteeodlboisrnettrhoveeLodAt.hTUehrRiissA“o,cbsosirmerreivclaatt,r”ioabnguritesmenmeoretentloyt wtionititedhnedtnheteidfyetoxthpiemercpiamluyestnehtoaaflt 01 adiscrepancybetweencurrentandpreviouscomparisons. 2 5, Thedifferencesbetweenpredictionandmeasurementshownfor h 1 each of the gauges in Figs. 7–17 were averaged for each case to c Mar determine an overall uncertainty for that case. Averages were n generatedforsupercatalyticandnoncatalyticcomparisonintermsof o E both the average magnitude (absolute value) of the differences R NT betweenpredictionandmeasurementandtheaverageofthesigned CE (positive/negative) differences. These averages are plotted in H Figs. 18–21 vs total enthalpy. From these figures, several general C AR observationscanbemade: SE 1) The average magnitude per case of the differences between E Y R Fig.16 CaltechT5ReflectedShockTunnel:Run2255,11:3MJ=kg noncatalytic predictions and experimental data varied between 15 LE case. and45%. G 2) The average magnitude per case of the differences between N A supercatalytic predictions and experimental data varied between A L thatwasnotevidentintheoriginalcomparisons.Thisdifferencewas approximately5and30%,withtheexceptionofa148%difference AS traced to the chemical–kinetic models employed. In the original fortheGASLHYPULSEcase.Uponreviewofadraftofthispaper y N study,explicitforwardandbackwardreactionrateswerespecifiedas bytheoperatorsofHYPULSE,‡itwasindicatedthatthepublished d b perEvansetal.[31],whereasinthecurrentstudy,thebackwardrates resultsfromthistestmaycontainadatareductionerrorthatproduced e ad erroneouslyhighheatingvalues.However,timelimitationspreclude o wnl reviewingandrevisingthesedatainthepresentpublication. Do 3)Thenoncatalyticpredictionsaveragedfromapproximately1to 35%lowerthanthedatapercase,expectfortheGASLHYPULSE casewherethepredictionswere45%higherthanthedata. 4)Thesupercatalyticpredictionsaveragedfrom5to35%higher thanthedatapercase,withtheexceptionofa148%overprediction fortheGASLHYPULSEcaseandtheunderpredictionforthelow- enthalpyCUBRCLENScase. 5)Alinearfitcanbemadeforanyoftheaveragecomparisonsto showthatthedifferencesincreasewithtotalenthalpy.However,the qualityofsuchfitsisquestionablegiventhesmallnumberofdata points available and the skew resulting from the very large differencesfortheGASLHYPULSEcase. From these results, it is tempting to conclude that the supercatalyticboundaryconditionisappropriatebecauseitprovides theclosestagreementwiththedata,withtheexceptionoftheGASL HYPULSEcase,whichcouldbeeliminatedasanoutlierpoint(andis subject to possible revision due to data reduction error as noted previously).Infact,forNASA’sMSLmission,thisconclusionwas Fig.17 GASLHYPULSEExpansionTube:(cid:2)12:3MJ=kgcase. ‡FromprivatecommunicationwithChing-YiTsai,ATK(Aug.2011). HOLLISANDPRABHU 63 1 8 1 2 6 1. 4/ 1 5 2 10. Fig.19 Comparisonofnoncatalyticpredictionswithdata(averaged Fig.21 Comparisonofsupercatalyticpredictionswithdata(averaged OI: signederror). signederror). D org | appliedbecausethesupercatalyticoptionprovidesthemostdesign aa. conservatism[33]. differenceswerealsoobservedinthecomparisonsbetweenpredicted c.ai However, although possibly appropriate as a vehicle design andmeasuredshockshapes.Suchflowfielddifferencescannotbea http://ar puanrcaedritgaimnt,ysauncahlysais.cOonthcleurseiovnidecnacnenmotusbtealsaoppbleiecdontsoidaererdig.oFriorsuts, ffuacntcotrisontoocfosnusrifdaecre. catalysis, thus there must be other uncertainty 13 | thatthelimitedsetofdataonCO2surfacecatalysisdonotsupport[3] 20 thesupercatalyticmodelformetallicsurfacesatlowtemperatures. 5, Second,thattheHYPULSEdatawereobtainedonmodelsfabricated B. Shock-ShapeComparisons 1 ch fromMacorceramic, whichwouldbeexpected toshowevenless Comparisons were made between the predicted and measured Mar catalyticefficiencythanametallicmodel;thus,thistestmayprovide shockshapes(withtheexceptionofCaltechrun2256forwhichno on adistinctlydifferentphysicalenvironmentthantheothersandcannot image was available) as shown in Figs. 22–30. The differences RE beimmediatelydismissed.Third,thatthefreestreamenvironmentof betweenpredictedandmeasuredshockstandoffdistancesatthenose NT theGASLHYPULSEtest(andalsotheUniversityofIllinoisHET areshowninFig.31withrespecttototalenthalpyandinFig.32with CE test) was generated in a significantly different manner than the respect to freestream density. With the exception of the lowest CH CUBRCLENS,CaltechT5,andAmes42-InchShockTunneltests; enthalpyCUBRCLENScase(series2,run16)andperhapstheARC AR thefirsttwofacilitiesoperateasshock-expansiontunnels,whereas 42-Inchcase,therewerelargedifferencesbetweenthepredictedand SE theotherthreeoperateasreflectedshocktunnels.Asdiscussedin measuredshockshapesforallcases.Shockstandoffdistanceswere E Y R [34], there is considerable uncertainty as to the vibrational underpredictedfortheCUBRCLENSandGASLHYPULSEcases E nonequilibrium state of the freestream in reflected shock tunnels, andoverpredictedfortheCaltechT5andUniversityofIllinoisHET L G whichcouldproducesignificantuncertaintiesintheresultssurface cases(theavailableimagequalitywastoolowforassessmentofthe N A heattransfermeasurements.Andfinally,aswillbeshowninthenext ARC 42-Inch Shock Tunnel case). These comparisons suggest L A section, regardless of surface catalysis model selected, major fundamentaldifferencesbetweenactualandcomputedchemicaland S A N y b d e d a o nl w o D Fig.20 Comparisonofsupercatalyticpredictionswithdata(averaged Fig.22 CUBRC LENS I Reflected Shock Tunnel: Series 2, run 16, errormagnitude). 1:9MJ=kgcase. 64 HOLLISANDPRABHU 1 8 1 2 6 1. 4/ 1 5 2 0. 1 OI: D aa.org | F5:i6g.M23J=kCgUcBaRseC. LENS I Reflected Shock Tunnel: Series 2, run 8, F6:i0g.M25J=kCgUcBasReC. LENS I Reflected Shock Tunnel: Series 2, run 12, ai c. ar p:// vibrational relaxation rates, which would strongly influence the vehicle, which is a significant level for a mission that requires 13 | htt pwrheedtihceterdthseudrfifafceerehnecaetsiningmraetaessu.rTehduasn,ditprceadnincotetdbheeasatiindgfroartecsewrtaeirne pshreacpiessionforentthrey.cTahseesdiinffetrheinscesstudinypwreedriectmeducahndlamrgeearsuthreadnsthhoocske 0 h 15, 2 doureottohetrhefascutrofrasc.eAcnatda,lybseics,auchseemhiicgahl-eanntdhvalipbyratfiaocnilailtireast’esfr(eoersbtroetahm), guennceerratateindtiefsorcatnhebeMexSpLectceadsetoibnel[a3r4g]e,;hanodwesvoer,ththeeraeearoredynnoadmatica arc conditions are typically determined through a combination of availableforcomparisons. M diagnosticmeasurementsandnumericalmethods,itisalsopossible n E o that the freestream conditions were not accurately characterized. TR This possibility is more likely for the reflected shock tunnels, as C. OverallAssessmentofComparisons EN discussedin[14,34]. Withoutattemptingtoassignthecausetoeitherexperimentalor C H Although the focus of this study is primarily on heat transfer computational methods, this survey reveals that very large RC uncertainties,thepredictionofshockshapesalsoaffectsaerodynamic differences exist between the two with respect to convective ESEA udnisctreirbtauitniotinesa.nTdhtehsuhsoocfktshheafpoercisesananinddmicotomreonftsthreessuulrtfiancgefproremssuthree afleorwohs.eDatiifnfgerreantecsesainndhfleoawttfiraenldsfsehroractkesshvaaprieesdfboerthwigeehn-e1n5thaanldpy14C8O%2 R Y integrationofthatpressuredistribution.Ithasbeenshown[35]that and were strongly influenced by the selection of surface catalysis LE theuseofdifferentmodels(Camac[36]andMillikan–White[37])for modelsforthecomputations.Shockstandoffdistancecomparisons G N vibrationalrelaxationratesofthepolyatomicCO moleculeproduces revealed differences ranging from 50% underprediction to 50% A 2 L anuncertaintyinthetrimangleofadegreeormorefortheMSL-entry overprediction. A S A N y b d e d a o nl w o D Fig.24 UniversityofIllinoisHETExpansionTube:5:7MJ=kgcase. Fig.26 ARC42-InchShockTunnel:8:5MJ=kgcase. HOLLISANDPRABHU 65 1 8 1 2 6 1. 4/ Fig.27 CUBRC LENS I Reflected Shock Tunnel: Series 2, run 13, 1 25 8:6MJ=kgcase. 10. Fig.29 CaltechT5ReflectedShockTunnel:Run2255,11:3MJ=kg OI: case org | D MaTrhseesxepcloomraptiaornismonisssaiorensstamrtulisntgbleypcoonosriadnedretdh.eFirirismt,pilnicahtiisotnosricfoarl aa. context, is the successful Viking mission. Viking peak heating thepredictedandmeasuredshockshapesdifferindicatesthatother c.ai conditionswereonthelowrangeofthesedata,wheretheuncertainty phenomena beside wall catalysis are not being modeled properly http://ar (cid:7)in2h0e%atibnagsemdiognhtthbeeseascsoigmnpeadriaso“nrse.aOsofnfacboluer”sev,anlueiethienrtthheerdaantageseotsf acnodursme,atyheaslesodifcfoernetnricbeusteintsohoacekroshheaaptiensgpruondcuecretaainntiaeesr.odAynnda,moicf 3 | reviewedhereinnorthecomputationaltechniquesemployedexisted uncertainty that cannot be assessed because there are no relevant 201 whenthatmissionwasdeveloped.Instead,analyticalmethodswere force-and-momentdataagainstwhichtocomparepredictions. 5, appliedwithhighlevelsofconservatism. It is a requirement of this study that some numerical values be 1 ch Incontrast,theMSLmission,forwhichmostofthesedatasets assignedtothecomputationalaeroheatinguncertaintyofthestate-of- Mar were developed, is at the high end of the test range. Noncatalytic the-arttoolsusedbyNASAforMars-entryproblems.Althoughsuch on predictions were up to (cid:4)35% lower than the experimental data, an assessment is recognized as being overly simplistic, it at least RE whereas supercatalytic predictions were up to (cid:4)35% higher providesastartingpointforconsiderationoftheissue.Thus,based NT (exceptingtheGASLHYPULSEcase).Itisfortunatethat,infact,the onthecomparisonspresentedinFigs.7–17,approximateuncertainty CE supercatalyticmodelwasemployedinthedesignprocessbecauseit estimates for laminar, convective aeroheating have been made in CH overpredicts all experimental data except a single low-enthalpy terms of total enthalpy. Because of the previously noted doubts AR condition. astotheappropriatenessofthesupercatalyticboundaryconditions, SE Looking to the future, HMMES-class missions will experience these estimates are based on the averaged error magnitude E EY R emnitshsaiolpnisesarweeolultbsiedyeonthdearnayngeexoisftitnhgetteesstt ddaattaa,.iBteccaanunsoetHsaMfeMlyEbSe reenstuhlatlipnyg(f<ro5mMtJh=ekgn)o,n(cid:7)c1a5ta%lytuicncberotuainndtayr;y2)cfoonrdmitoiodne:ra1te)efnotrhallopwy GL assumed that the supercatalytic assumption will providesufficient (5–10MJ=kg), (cid:7)30% uncertainty; and 3) for high enthalpy AN conservatismforaeroheatingpredictions.Additionally,thefactthat (10–20 MJ=kg),(cid:7)60%uncertainty. L A S A N y b d e d a o nl w o D Fig.28 CaltechT5ReflectedShockTunnel:Run2254,10:6MJ=kg Fig.30 GASL HYPULSE Expansion Tube: Run 749, 12:3MJ=kg case. case.