Thisisprimarilyduetothefactthattime-varyingloading(e.g.,WISPER)isassociatedwithnonconstantamplitude.Inotherwords,realtime-varyingwind-inducedloadingisirregularandstochastic,andtheassociatedloadhistoryaffectsthecompo-nentfatiguelifeincomplexways.Theprocedureusedinthisworktocorrelatethematerialfatiguestrengthwiththecomponentfatiguestrength/lifeisbasedontheuseofacycle-countingalgorithm(theso-calledRain owcycle-countinganalysis;Ref10),alinearizedGoodmandiagram(e.g.,Ref11)toaccountfortheeffectofmean-stress/strainonthematerialfatiguelife/strengthandtheMinerÕslinear-superpositionprinciple/rule(Ref12).Therain owanalysis,theGoodmandiagram,andtheMinerÕsrulearebrie yoverviewedintheremainderofthissection.
2.3.1Rain owAnalysis.Whenatime-varyingloadsignalisrecordedoverasamplingperiod,andneedstobedescribedintermsofathree-dimensionalhistogram(eachbinofwhichbeingcharacterizedbyarangeofthesignalamplitudeandarangeofthesignalmean-value),proceduresliketherain owcountingalgorithmareused.Withinthisprocedure,the rststepinvolvesconvertingtheoriginalloadsignalintoasequenceofloadpeaksandvalleys.Then,thecycle-countingalgorithmisinvoked.Tohelpexplaintherain owcycle-countingalgorithm,asimpleloadsignal(afterthepeak/valleyreconstruction)isdepictedinFig.6(a),withthetimeaxisrunningdownward.
Fig.6(a)Applicationoftherain owcycle-countingalgorithmtoasimpleloadsignalafterthepeak/valleyreconstruction.Leaseseetextforexplanation;and(b)theresultingthree-dimensionalhistogramshowingthenumberofcycles/half-cyclesineachmeanstress/strain-stress/strainamplitude

bin
794—Volume19(6)August2010Withintherain owcycle-countingalgorithm,separatecountingofloadhalf-cyclesiscarriedoutfortheonesstartingfromthepeaksandtheonesstartingfromthevalleys.InFig.6(a),onlythehalf-cyclesoriginatingfromthepeaksareanalyzed.Ahalf-cyclethenstartsfromeachpeakandendswhenoneofthefollowingthreecriteriaismet:
(a)whentheendofthesignalisreached(CaseAin
Fig.6a);
(b)whenthehalf-cycleinquestionrunsintoahalf-cycle
whichoriginatedearlierandwhichisassociatedwithahigherpeakvalue(CaseBinFig.6a);and
(c)whenthehalf-cycleinquestionrunsintoanotherhalf-cyclewhichoriginatedatalatertimeandwhichisassoci-atedwithahighervalueofthepeak(CaseCinFig.6a).Onceallthehalf-cyclesareidenti edtheyareplacedinbins,eachbinbeingcharacterizedbyarangeoftheloadamplitudeandtheloadmean-value.Anexampleoftheresultingthree-dimensionalhistogramshowingthenumberofcycles/half-cyclespresentintheloadsignalassociatedwithagivencombinationoftheloadamplitudeandtheloadmean-valueisdepictedinFig.6(b).
2.3.2GoodmanDiagram.BeforepresentingthebasicsoftheGoodmandiagram,itisimportanttorecognizethatfatiguelifeofamaterialisafunctionofboththestress/strainamplitudeandthestress/strainmean-value.Often,thestress/strainmean-valuesarequanti edintermsofanR-ratiowhichisaratioofthealgebraicallyminimumandthealgebraicallymaximumstress/strainvalues(associatedwiththeconstant-amplitudecyclic-loadingtests).Fromthede nitionofthemeanstress/strain,itcanbereadilyshownthatfatigue-loadingtestscarriedoutunderconstantR-ratioconditions,correspondtothetestsinwhichthemeanstress/strainscaleswiththecorre-spondingamplitude.ToconstructtheGoodmandiagram,constant-R/constant-amplitudefatigue-testresultsareplotted,inastress/strainamplitudeversusstress/strainmean-valuediagram.Asdepicted,inFig.7,constant-Rdatafallontoa

line
Fig.7AnexampleoftheGoodmandiagramshowingconstantfatigue-lifedata(dashedlines)andconstantR-ratiodata(thesolidlinesemanatingfromtheorigin)
JournalofMaterialsEngineeringandPerformance
热门文档