ExperimentalanalysisofacompositeautomotivesuspensionarmM.PINFOLDandG.CALVERT(UniversityofWarwick/RoverGroupGaydon,UK)Received11November1992；revised26March1993Inapplicationswhereweightsavingandpartsintegrationcanbeachieved,theRoverGrouphasbeeninvestigatingthedesignandmanufactureofcomponentsfromcompositematerials.Themethodsusedinthedifferentstepsinthedesign-to-manufacturecycleinthehighvolumeautomotiveindustryarerelativelywellknownforasteelcomponent,butarenotsowellestablishedforacompositecomponent.Adesignmethodologyforcompositeshasbeenemerginginwhichaprincipalprocedureisdesignanalysis.Oneofthemostestablishedmethodsofanalysisisthatusingthefiniteelementtechnique,andthisisbeingsupplementedwithexperimentaltestsonprototypesusingphotoelasticanalysisandstresspat-ternanalysisbythermalemission,coupledwithconventionalstraingaugemoni-toring.Littleworkhasbeenundertakentocorrelatetheresultsobtainedfromthesedifferenttestmethodsandtocomparetheresultswithmeasurementsmadeonanactualcomponent.Thispaperpresentssomeoftheworkundertakenconcerningtheanalysisandtestingofacompositeautomotivesuspensionarm.Theresultsobtainedfromthethreedifferentanalysistechniquesarecomparedwithexperi-mentaltestresults,andtheiraccuracyisdiscussed.Keywords:autmotivesuspensionarm；stressanalysis；finiteelementmethod；photoelasticanalysis；SPATE；straingauges；sheetmouldingcompoundSolanddeWildestatethatcompositematerialshavebeenusedincreasinglyasstructuralmaterials.Areasonforthis.,isthatcompositematerialshavehighstrengthtoweightandhighstiffnesstoweightratioswhichcansignificantlyreducetheweightofastructure.Perhapsthemostimportantfeatureofcompositematerialsisthattheirmechanicalp:opertiescanbetailoredtomeetaspecificcriterion.However,Johnsonetal?suggestthatcompositedesign,analysisandfabricationtechnologymustundergomajordevelopmentsandsuccessfuldemonstrationsbeforesignificantstructuralcomponentswillbeincorporatedinproductionautomobilesandtrucks.Compositematerialshavetocompetewithsteelwithintheengineeringenvironment.WithintheautomotiveindustrythisrequiresacertainamountoftechnologytransferfromplacessuchastheAdvancedTechnologyCentreattheUniversityofWarwick,whichworkwithmaterialmanufacturersandautomotiveengineerstoenableunderstandingaboutthesematerialsasanalter-nativetothetraditionalmaterialssuchassteel.Ifcom-positesaretocompetewithtraditionalmaterialsinarealsense,thenautomotivedesignersneedtobefullyaware0010-4361/94/010059-05oftheirstrengthsandlimitationssothattheycanbeoneofperhapsmanyoptionsconsideredattheconceptstageofthedesign.Forthistohappenautomotiveengineersneedtocatchuponthetechniquesofdesigning,testingandmanufacturingcomponentsfromcomposites.Thiswillincludeunderstandinghowvariousmethodssuchasfiniteelement(FE)analysis,stresspatternanalysisbythermalemission(SPATE)andphotoelasticanalysiscanbeappliedtocompositecomponentsintheirdesignanddevelopment.Thusfarlittleworkappearstohavebeenundertakentostudywhethertheresultsobtainedfromthesedifferentanalysismethodscorrelatewithoneanotherorwithactualexperimentalresultsobtainedfromtestingarealcomponent.Inordertostudytheapplicationandcorre-lationofthedifferentanalysismethodstocompositematerials,acompositecomponent-anautomotivelowersuspensionarm-wasmanufactured.Thiscom-positecomponentwasanalysedbythethreemethodsdescribedaboveandalsotestedunderrealisticloadingconditions,withexperimentalresultsbeingobtainedfromstraingauges.1994Butterworth-HeinemannktdCOMPOSITES.VOLUME25.NUMBER1.199459,BallJointHousingFig.1ThecompositesuspensionarmDESIGNTheexistingsteellowersuspensionarmconsistsofninepiecesweldedtogetherwhilstthere-designedcompositecomponent-whichcanbeseeninFig.1-isasinglemouldedpart.Thematerialusedtomanufacturethesuspensionarmwasasheetmouldingcompound(SMC),comprisingapolyesterresinbondingagentwitha30%contentofrandomlyarrangedshortglassfibresandcal-ciumcarbonatefiIler.Theweightofthesteelsuspensionarmis2.53kgwhilstthere-designedSMCsuspensionarmcompletewithbushesandballjointweighs1.5kg.Thematerialpropertiesusedforthecompositesuspensionarmintheseanalyses,obtainedfromtestscarriedoutatRoversmaterialslaboratory,wereYoungsmodulus=10.5GPa,Poissonsratio=0.26anddensity=1.8x10-6kgmm-3.EXPERIMENTALTECHNIQUESPriortoundertakingexperimentalanalysisofanactualengineeringcomponent,someinitialvalidationworkwasrequiredtogainconfidenceinthetechniqueswhenappliedtosheetmouldingcompound.Therefore,fiatplates,beamsanddiscsconstructedfromSMCwereana-lysedundervariousloadingconditionsbeforeprogress-ingontothedesignedcomponent.Mostvalidationtestswerecarriedoutusingstrain-gaugedspecimenstocorrelatewiththefiniteelementanalysisresults.AlthoughitisrecognizedthatSMCisnotanisotropiematerialduetosomefibreorientationduringprocessing,forthepurposesofanalysisthemater-ialwasassumedtobeisotropic.Also,whentheactualSMCsuspensionarmwascutupandexamined,signifi-cantfibredistributionwasobservedintheribs.Itisfeltthatthecorrelationbetweentheexperimentalandanaly-sisresultsvalidatedthisassumptioninthecaseofthisparticularcomponent.StraingaugetestsBeforeundertakingtheexperimentaltestwork,thecom-positecomponentwasmountedviaitsrubbermountingbushesontoarelativelyinfinitelystiffstructure.Itisverydifficulttocoveralloftheloadingconditionswhencon-ductingexperimentaltestsandthusaworst-casescenarioisusuallyassumed.Theworst-caseloadingconditiononsuspensioncomponentsisknownaspot-holebrake.Thisattemptstosimulatethevehiclefallingintoadeeppot-holeat30mphwiththebrakesfullyappliedatthepointofimpact.Theresultantfore/aftandlateralloadsarethencalculatedbasedontheweightandvelocityofthevehicle.Duetothelimitationsofthetestrigthefullpot-holeloadscouldnotbeappliedtothecomponent,andthusreducedloadswiththesameresultantdirectionasthepot-holeloadswereappliedandtheresultsscaled.Theloadsappliedforthefullpot-holebrakecasewere24.2kNinXand8.2kNinY,andforthereducedloadcasewere5.9kNinXand2.02kNinY-seeFig.1.Thestraingaugesusedconsistedofsixthree-axisrosettegaugesand13single-gridgauges,with2.5mmgridlengths,chosentofitintotheradiiofthecomponentinanattempttomeasurethemaximumstrain,Gaugesweresituatedneartheballjointhousing,wheretheloadswereapplied,andaroundtheradiiofthebodymountingbushes,wherethecomponentwouldbemountedtothecarsubframe.Additionalstraingaugesweresituatedonsomeofthestrengtheningribsandclosetotheanti-rollbarmountingposition.SPATEanalysisStresspatternanalysisbythermalemission(SPATE)canbeusedtodeterminethesurfacestressesofcomponentsbystudyingthesmallchangesintemperatureduetocyclicloadingconditions.SPATEequipmentcomprisesadetectorunitwithscanninghead,ananaloguesignalprocessingunitandadigitalelectronicdataunit.Thesystemworksbydetectingtheminutetemperaturechangeswhichoccurwhenastructureiscyclicallyloaded.Theinfra-reddetectorscansthestructureandcorrelatesthemeasuredoutputwithareferencesignalfromtheloadingsystem.Anelectronicdataprocessingsystemcorrelatesthedetectedstress-inducedthermalfluctuationswiththeloadingreferencesignal.Acolourcontourmapofthesumoftheprincipalstresses(cr+4)isthenplotted,togetherwithabarchartgivingactualvalues.Thiscorrelationofsignalseffectivelyeliminatesallsignalfrequenciesotherthanthosecausedbytheloadingsystem,i.e.,allambienttemperaturefluctua-tions.TheSPATEsystemhasatemperatureresolutionof0.001C,andaspatialresolutionoflessthanImm.ThistypeofanalysishasbeenshownbyanumberofauthorsTMtoalsobeapplicabletonon-isotropicmater-ialssuchascomposites,andthesmallerrors(6%)demonstratedfromsuchstudieswhencomparedwiththeoreticalorFEresultsarefelttobeduetoinaccuraciesinthematerialdataused4.Itisapparentfromthestudiesundertakenthattheuseofthermoelasticstressanalysistoevaluatestressesandstrainsinanisotropiccompositematerialsismorecomplexthanforisotropicmaterials.However,ithasbeenshownthatthetechniquecanprovidevaluablequalitativeinformationonstressdistri-bution,effectsofsurfacedefectsandcrackgrowthpredictions.Ithasalsobeendemonstratedthat,givenaccuratedetailsofmaterialpropertiesincludingexpan-sioncoefficients,quantitativeresultscanbeobtaineddependinguponthedegreeofanisotropyofthematerial.PriortoundertakingafullSPATEanalysisofthesuspen-sionarmitwasnecessarytodetermineacalibrationfactorforthematerialused.Thiscanbeachievedintwoways,eitherbyloadingadiscofthematerialincompres-sionandcomparingtheSPATEoutputwiththetheoreti-60COMPOSITES.NUMBER1.1994calsolution,orbystraingaugingdirectlyontothecomponentinanareaofevenstressdistribution,therebyobtainingadirectcomparisonwiththeSPATEoutput.Bothmethodswereusedinthiscase,butdirectcalib-rationwithstraingaugescanovercomealotoftheproblems,thusallowingsignificantinformationtobeobtainedfromtheSPATEoutput.PhotoelasticanalysisThemajorityofphotoelasticworkinvestigatingthemac-romechanicalbehaviourofcompositematerialshasbeenundertakenusingphotoelasticcoatingtechniques.Thisisdonetoavoidthecomplexitiesofconstructingaphoto-elasticmodelwithanisotropicpropertiesandthuscon-structingacompositeliketheoriginalwhichwouldloseitstransparencyandcouldnotbeanalysed.However,forcomplexfibrelay-upsthiswouldbetheonlymethodofconductingphotoelasticanalysis,andthussomeresearchhasbeenundertakeninvestigatingtheuseoftheactualcompositesj7-30.Reasonableresultshavebeenobtainedfromsuchanalyses,butwithlimitationsduetotheneces-sityfortransparencywithinthecomposite.However,thecompositecomponentconsideredinthisstudywasmanufacturedfromSMCandthematerialwasassumedtobeisotropic,thussimplifyingthecreationofaphoto-elasticmodel.Athree-dimensionalepoxyresinmodelofthesuspensionarmwasconstructedforthephotoelasticanalysis.Themodelwasthenloadedinarepresentativemanner,withscaled-downloads,andsubjectedtoastressfreezingcycle.Thisinvolvesheatingthemodeluptothemater-ialsglasstransitiontemperature,atwhichpointtheYoungsmoduluschanges,andthemodeldeformsundertheappliedloads.Themodelisthenslowlycooled,avoidinganyuneventemperaturedistributionwhichcouldresultinunwantedthermalstresses.Duringthecoolingcyclethedeformationsandstressesarelockedintothemodel.Whenviewedunderpolarizedlightthethree-dimensionalmodelisajumbleofinterferencefringes.Inordertodeterminebothmagnitudeanddirec-tionoftheprincipalstressesatanypoint,asliceisremovedandobservedunderpolarizedlight.Bycount-ingthefringesthestressesinthemodelcanbecalculatedandconvertedintoactualstressinthecomponent.Thisisdonebymeansofproportionality,betweenthemodelandcomponentmaterials,andtheloadinganddimensio-nalparameters.Thelowersuspensionarmismountedtotherestofthecarviarubbermountingbushes.Investigationswerecarriedoutastothepossibilityofmodellingthesemountingbushes.However,experimentswithsiliconandfoamrubbersshowedthattherequiredscaled-downstiffnessofthebushesduringstressfreezingatelevatedtemperaturescouldnotbemaintained.Thephotoelasticanalysisthusassumedthatthesuspensionarmwassolidlymounted.FINITEELEMENTANALYSISThecompositesuspensionarmwasmodelledusingapproximately1300oftheSTIF45ANSYSsolidele-ments.Thesuspensionarmismountedtothesubframeviarubbermountingbushes；theseweremodelledwithspringelementstorepresentthestiffnessofthebushesandtocreatearealisticloaddistributionthroughoutthecomponent.LoadswereappliedtotheFEmodelviabeamelementsattheballjoint.ThreeloadcaseswereanalysedusingtheANSYSFEsoftware.Thefirstloadcasesimulatedthefullpot-holebrakeloads.Thesecondsimulatedthereducedloadusedinthetestsduetothelimitationsofthetestrig,toenablecomparisonswiththeresultsfromtheexperimentalstraingaugeanalysis.Thesetwoloadcasesusedspringelementstosimulatethestiffnessoftherubbermountingbushes.Thethirdloadcaseagainusedthereducedloadsbutthistimeomittedthespringelements；i.e.,thesuspen-sionarmwasmodelledasbeingsolidlymounted.ThisthirdloadcasewasrequiredtocorrelatewiththeSPATEandphotoelasticanalyses.RESULTSFiniteelementanalysisAnalysisofthesuspensionarmshowedthatthemaxi-mumequivalentstressinthecomponentfortheloadcaseconsideredisveryclosetotheultimatetensilestrengthoftheproposedmaterialforthepot-holeloadingcondition,whichistheworstloadingcondition.Thismeansthatthecomponentmayneedtobemanufacturedfromadiffer-entmaterial,orthatothermaterialsneedtobeposit-ionedinareasofhighstresstostrengthenthecomponentlocally.Duetoconstraintsupontheamountofcomputerdiscspaceavailable,thenumberofelementsusedwithintheFEmodelwasrelativelylowandthusthesizeoftheelementswithintheareaoftheradiiaroundthebodymountingbusheswastoolargetodetectanylargestressconcentrations.Also,thetypesofelementusedaroundtheseareas,duetothegeometryofthecomponent,wereamixtureofbrick,wedgeandtetrahedral.Thelattershapetendstobetoostifftogivegoodresultsandisnotrecommended.Ifmoredetailedresultswererequiredintheseareas,thentheseradiiwouldhavetobemodelledingreaterdetailwithmoreandsmallerelementsintheareasofhighstressgradient.PhotoelasticanalysisTheanalysisofthephotoelasticmodelofthesuspensionarmwasundertakenassumingthatthedirectionsofthemaximumprincipalstresseslayinahorizontalplanethroughthemodelinthedirectionofthefore/aftload.Whilstthisisnotstrictlytrueinpracticeduetolocalgeometryeffectsincertainareas,theassumptiongavesufficientlyaccurateresults.Ifobviousdiscrepancieswerefoundinparticularareasthenitwaspossibletotakeslicesfromdifferentplanes.Maximumstresseswereseentooccurinthevicinityoftheballjointhousingandthebodymounts.Duetotheabilityofphotoelasticanalysistopinpointverysmallareasofhighstress,themaximumstressvaluesgivenbyphotoelasticitytendedtobehigherthanthestraingaugeresults.Forexample,maximumstresslevelsintheinternalradiusoftheleadingbodymountwerefoundtobe43MPacomparedwithaSPATEvalueof26MPa.Thisdifferencecanbeexplainedbyexamin-ingtheslicetakenthroughthephotoelasticmodelwhichshowsthatthemaximumstressonlyoccursatapositionCOMPOSITES.NUMBER1.199461Table1.Stressresults(MPa)forfullloadcon-ditionsPositionStraingaugesFEPhotoelasticBalljointhousing176165176spanning3mmandthatthestressvalueseithersideofthemaximumarearound25MPa.SPATEanalysisTheinitialSPATEscanshowedlargebandsofstressrunningacrossthemountingareasandsomeconfusionastowhethertheseareaswereintensionorcompression.Theproblemwasidentifiedasexcessivemovementinthesuspensionarmbodymountingpositionsduetodistor-tionoftherubberbushesasexperiencedinthestraingaugetests.SPATEisequippedwithamotioncompen-satordeviceifrequired,whichdeflectsthescanningmirrorsinsidethedetectorintimewiththeoscillationsofthetest-piece,therebyeliminatingthemovement.How-ever,inthisparticularcase,thegeometryanddirectionofmovementcouldnotbeeliminatedovertheentireareaatthesametime,andthusitwasnecessarytoremovetherubberbushesandtoreplacethemwithaluminiumones.TheSPATEanalysiswasrepeatedwiththesolidbushesandshowedareasofhightensilestress(26MPa)alongtheleadingedgeandaroundtheinnerradiusoftheleadingbodymountingposition.Unfortunately,noSPATEanalysiscouldbeundertakenattheballjointendofthecomponentasitwasobscuredbythelargeloadingadaptorrequiredtofitthehydraulicactuatorsupplyingthecyclicloading.COMPARISONOFRESULTSItshouldbeclarifiedthatthestressvaluesquotedinthetablesfromthestraingaugeresultswerecalculatedfromtherosettegaugestogiveavalueofmaximumprincipalstress.Thephotoelasticanalysisalsogivesmaximumprincipalstressesunlessthevaluesaretakeninboardofafreeedgeinwhichcasetheyaredifferencesinprincipalstresses(o.-o-,).SPATEanalysisgivesanoutputintheformofthesummationoftheprincipalstresses(or.+a2)whereastheFEoutputcanbeinanyformrequired(inthiscaseyonMises).Duetothegeometryofthecompo-nentandthewayinwhichtheloadswereapplied,thevaluesofor2andcr3werealwayssmall,andthusdirectcomparisonscouldbemadebetweenthedifferentanaly-sismethodswithoutfurtherconversion.Tablelcomparestheresultsobtainedforthemaximumpot-holeloadconditions.Themaximumstressvaluesalloccurattheballjointareaandcorrelateverywell.Theseresultantstressesforthestraingaugesandphotoelasti-citywerecalculatedfromtheresultsobtainedforthereducedload.Themodelstresswasmultipliedbyaload-ingfactorastheratiobetweenthefore/aftandlateralloadingremainedconstantandinthesameproportionasthefullpot-holebrakeloadappliedtothesuspensionarlTI.TheresultsoftheanalysesundertakenwithreducedTable2.Stressresults(MPa)forredTJcedloadswithmountingbushesPositionStraingaugesFEInnerradiusofbody2520mountBalljointhousing4940Table3.Stressresults