An-investigation-of-fatigue-failures-of-turbine-bl

AninvestigationoffatiguefailuresofturbinebladesinagasturbineenginebymechanicalanalysisJianfuHou*,BryonJ.Wicks,RossA.AntoniouAirframesandEnginesDivision,DSTOAeronauticalandMaritimeResearchLaboratory,506LorimerStreet,FishermensBend3207,Melbourne,Victoria3001,AustraliaReceived28November2000;accepted25December2000AbstractBladefailuresingasturbineenginesoftenleadtolossofalldownstreamstagesandcanhaveadramaticeffectontheavailabilityoftheturbineengines.Thoroughfailureinvestigationisessentialfortheeffectivemanagementofengineairworthiness.Inthispaperbladefatiguefailuresareinvestigatedbymechanicalanalysesandbyexaminationoffailedblades.Aseriesofmechanicalanalyseswereperformedtoidentifythepossiblecausesofthefailuresbyexamininganomaliesinthemechanicalbehaviouroftheturbineblade.Anon-linearfiniteelementmethodwasutilisedtodeter-minethesteady-statestressesanddynamiccharacteristicsoftheturbineblade.Thesteady-statestressesanddynamiccharacteristicsofthebladewereevaluatedandsynthesisedinordertoidentifythecauseofbladefailures.#2002ElsevierScienceLtd.Allrightsreserved.Keywords:Turbinebladefailures;Finiteelementanalysis;Bladevibration;Stressanalysis;Fatiguefailure1.IntroductionBladefailuresingasturbineenginesmayhaveasevereimpactontheavailabilityofenginesandonthemanagementofengineairworthiness.Bladefailurescanbecausedbyanumberofmechanismsundertheturbineoperatingconditionsofhighrotationalspeedatelevatedtemperature.Ingeneral,bladefailurescanbegroupedintotwocategories:(a)fatigue,includingbothhigh(HCF)andlowcyclefatigue(LCF)[1–5]and(b)creeprupture[6–7].Bladefatiguefailuresareoftenrelatedtoanomaliesinmechanicalbehaviourandmanufacturingdefects.Toidentifythecausesofthebladefatiguefailures,acompleteinvestigationhastobecarriedout,integratingboththemechanicalanalysesandmetallurgicalexamination.Metallurgicalexaminationcanbeveryeffectiveindeterminingwhetherthefailureisrelatedtomaterialdefects,machiningmarks,poorsurfacefinish,initialflawsorheattreatment.However,theexaminationdoesnottakeaccountofpossiblevariations,fromthedesign,inthemechanicalbehaviourandcharacter-isticsoftheblade,andthesevariationsmaybedirectlylinkedtothemechanismoffailure.1350-6307/02/$-seefrontmatter#2002ElsevierScienceLtd.Allrightsreserved.PII:S1350-6307(01)00005-XEngineeringFailureAnalysis9(2002)201–211*Correspondingauthor.Tel.:[phone];fax:[phone].E-mailaddress:[e-mail](J.Hou).Fractographymaybeusedsuccessfullytodeterminewhetherfailureisduetofatigueorstressrupturebutitcanbequiteinadequateinidentifyingthecausesoffatigueduetovariationsinmechanicalbehaviour.Therefore,acompletefailureinvestigationshouldincludeamechanicalanalysisinordertodefinethesensitivityofthebladebehaviourtogasturbineengineperformancevariables.Thispaperfocusesonmechanicalanalysesofbladefatiguefailureswhichwerefoundinamilitaryengine.Aseriesofanalyseswereperformedandtheresultsevaluatedtoidentifythepossiblecausesoffatigue.Anon-linearfiniteelementmethodwasutilisedtodeterminethesteady-statestressesandthedynamiccharacteristicsoftheturbinebladeunderbothlaboratoryandserviceconditions.Thesteady-statestressesanddynamiccharacteristicsofthebladewerethenanalysedandevaluatedtoidentifythecausesoffatigue.2.ExaminationsoffailedbladesThebladegeometryandthebladefailurelocationareshowninFig.1.Examinationofthebladesfromafailedengineindicatedthatthe‘‘primary’’failure,Fig.2(a),wasduetotheseparationofabladeatthetopfirtree.Thisbladesubsequentlycausedthefracture,byimpact,ofalltheotherbladeaerofoilsofthefirstandsubsequentstages.This‘‘primary’’fractureoccurredaftertheformationofafatiguecrackapproxi-mately6–7mmlong.Furthermore,asignificantproportion(about25%)ofthebalanceofthebladesfromthefirststagehadsmallfatiguecracks(lessthan0.2mm)inthetopfir-treeroot.Fig.2(b)and(c)showatypicalcrackinausedbladeandfatigueprogressionmarksrespectively.Thefailuremechanismhasbeenassessedtobelowcyclefatiguecracksinitiatedearlyinservicewiththeirgrowthproceedingbyinter-mittenthighcyclefatigueathighmeanstresses.Fromanexaminationoftheretiredbladesitwasalsofoundthatbladetipcontactwiththeturbinecasingrubstraphadoccurredinallretiredblades,irrespectiveofthehoursofservice.Thiswasrevealedbytheremoval,byspallationandwear,ofthenickel-aluminidecoatingattheleadingedgeandatthesidesofthefins.3.Finiteelement(FE)modelling3.1.FEmodellingA3-Dfiniteelementmodelofabladeandasectorofthediscwerecreated,consistingof80,000solidelements,asshowninFig.3.Ten-nodedtetrahedralelementswereusedtomeshboththebladeanddiscduetothecomplexityofthegeometry.Mostofthesignificantgeometricfeaturesweremodelledandarelativelyfinermeshwasusedfortheregionoffailure,asshowninFig.3(b).Theglobalmeshdensitywaschosentominimisediscretisationerrorsinthefailureregion.Thetie-boltsandthespacerbetweenthefirststageandsecondstagediscswerenotphysicallymodelledandwererepresentedinsteadbyequivalentloadsandboundaryconditions.Cyclicsymmetrywasmain-tainedforthediscbyimposinganappropriatedisplacementconstraint.Interactionsbetweenthebladefirtreesanddiscfirtreeswererepresentedusinggeneralsurface-to-surfacecontact[