FINAL REPORT FATIGUE RELIABILITY OF STEEL HIGHWAY

FINALREPORTFATIGUERELIABILITYOFSTEELHIGHWAYBRIDGEDETAILSPeterJ.Massarelli,Ph.D.FacultyResearchAssociateThomasT.Baber,Ph.D.FacultyResearchScientistVirginiaTransportationResearchCouncil(ACooperativeOrganizationSponsoredJointlybytheVirginiaDepartmentofTransportationandtheUniversityofVirginia)InCooperationwiththeU.S.DepartmentofTransportationFederalHighwayAdministrationCharlottesville,VirginiaAugust2001VTRC02-R4iiDISCLAIMERThecontentsofthisreportreflecttheviewsoftheauthors,whoareresponsibleforthefactsandtheaccuracyofthedatapresentedherein.ThecontentsdonotnecessarilyreflecttheofficialviewsorpoliciesoftheVirginiaDepartmentofTransportation,theCommonwealthTransportationBoard,ortheFederalHighwayAdministration.Thisreportdoesnotconstituteastandard,specification,orregulation.Copyright2001bytheCommonwealthofVirginia.iiiABSTRACTTheexpectedlifeofasteelhighwaybridgesubjectedtorandom,variable-amplitudetrafficcyclesishighlydependentondamageaccumulationcausedbyvariousfatiguemechanisms.Thisstudyaddressedsomeoftheissuesassociatedwithdevelopingprobabilisticreliabilitymodelsforsteelbridgestructuresundervehiculartrafficloadings.Specifically,methodsforincorporatinginspectiondata(e.g.,thepresenceandsizeofcracks)intoclassicalandfracturemechanics–basedfatiguemodelstopredictfracture-criticalelementdamageaccumulationwereanalyzed.Theblockloadingmethodintroducedhereassignsadamagestatetothecrackeddetailaftereachvehiclepassage.Thismethodislimitedtoloadingblockswheretheplasticzoneatthecracktipdevelopedbythepseudo-staticresponseofthebridgemustbelargerthanthecrackgrowthcausedbythesubsequentdynamiccyclesforeachvehiclepassage.Itmustalsobeassumedthatthecracklengthisfixedduringaloadingblock.Thismethodaccountsforclosureeffectsbypredictingtheoccurrenceofdamageonlywhenthecrackisopened.Whenabridgeinspectionrevealsfatiguecracks,fielddataintheformofstrainmeasurementscanbecollectedinthevicinityofthefracturedetailtoidentifythecharacteristicstressblockandthefluctuationsinthecurveattributesattributabletorandomnessofthetrafficloading.Then,thedatacanbeanalyzedtoidentifythestatisticalpropertiesoftheattributeparameters,whichincludethemagnitudeofthepseudo-staticresponse,thenumberofdynamiccyclesineachblock,andthedurationofthevehiclepassage.Withthisinformation,thedistributionofstressblockscanbeestimated,andtheblockloadingmethodcanbeemployedtocalculatethefatiguelifetimeofeachcriticaldetail.FINALREPORTFATIGUERELIABILITYOFSTEELHIGHWAYBRIDGEDETAILSPeterJ.Massarelli,Ph.D.FacultyResearchAssociateThomasT.Baber,Ph.D.FacultyResearchScientistINTRODUCTIONFatigueandfracturebehaviorareimportantconsiderationsindeterminingtheconditionofmetalstructuressubjectedtocyclicloads.Specifically,theexpectedlifeofasteelhighwaybridgesubjectedtorandom,variable-amplitudetrafficcyclesishighlydependentupondamageaccumulationcausedbyvariousfatiguemechanisms.Theanalysisofstructuralfatiguedamageisgenerallydividedintotwoconcerns:resistanceandload.Resistancereferstotheload-carryingcapabilitiesofthestructure,andloadencompassestheexternalforcesappliedtothestructure.Structuralfailureisseldomattributedtoloadconsiderations;theoccurrenceofstressesexceedingthosepredictedbythedesignerisrare.Brittlefracturefailuresubsequenttofatiguecrackingisthemostcommoncauseofsteelbridgecomponentfailureandoccursinbridgesmainlybecausespecificdetailshavealowerfatigueresistancethanthedesignersoriginallythought(Fisher,1984).FisherandYuceoglu(1981)compiledqualitativedataconcerningthecrackingof142bridgesintheUnitedStatesandCanada.Theyidentifiedout-of-planedistortionandlargeinitialdefectsasthetwomostcommonspecificcausesoffatiguecrackdevelopment.BarsomandRolfe(1987)attributedcomponentdistortiontotheresidualstressesthatsometimesariseafterweldingmaterialshavecooled,andtheyreportedthatweldingmayalsoberesponsibleforanumberoftheinitialdefectssuchasgaspockets,undercutsandcracks.Anumberofspecificdetailsinsteelbridgeshavebeenidentifiedasbeingespeciallysusceptibletocracking.Xanthakos(1994)compiledalistofexpectedfracturelocationsthatissummarizedinTable1.Stress-lifefatigueanalysisiscurrentlybasedontheso-calledS-NdiagramsthatrelateN(loadingcyclestofailure)andtheamplitudeorrangeoftheappliedcyclicstressS.However,theempiricalnatureofstress-lifemethodsimposesseverelimitationsontheusefulnessofS-Ndata.TheS-Nmethoddoesnotaccountforthetruestress-strainresponseofthematerial;localplasticityeffectsareignoredandallstrainsareconsideredtobeelastic;andanyunusualdetailgeometrywouldprecludetheapplicationofpreexistingdata.Despitethelimitationsofclassicalfatiguemethods,theS-NapproachtofatigueanalysisisthebasisforthecurrentAmericanAssociationofStateHighwayandTransportationOfficials(AASHTO)designspecifications.2Thereisanongoingprocesstodevelopnewandmorereliablemodelstopredictfatiguedamage.Fatiguecrackdevelopmentiscommonlysubdividedintotwostages:crackinitiationandcrackpropagation(growth).Stressconcentrationsarisingfrommaterialdefectscanalsocausecrackinitiationtooccur.Theresultofthefatiguecrackinitiationprocessisamacrocrack.Macrocrackscanalsoformastheresultofthecoal