【astm標準】e 739 - 10 (2015)

1、Designation:E739?10(Reapproved2015)StardPracticefStatisticalAnalysisofLinearLinearizedStressLife(SN)StrainLife(εN)FatigueData1ThisstardisissuedunderthefixeddesignationE739thenumberimmediatelyfollowingthedesignationindica

2、testheyearofiginaladoptioninthecaseofrevisiontheyearoflastrevision.Anumberinparenthesesindicatestheyearoflastreapproval.Asuperepsilon()indicatesaneditialchangesincethelastrevisionreapproval.1.Scope1.1Thispracticecoverson

3、lySNεNrelationshipsthatmaybereasonablyapproximatedbyastraightline(onappropriatecodinates)faspecificintervalofstressstrain.Itpresentselementaryproceduresthatpresentlyreflectgoodpracticeinmodelinganalysis.Howeverbecausethe

4、actualSNεNrelationshipisapproximatedbyastraightlineonlywithinaspecificintervalofstressstrainbecausetheactualfatiguelifedistributionisunknownitisnotrecommendedthat(a)theSNεNcurvebeextrapolatedoutsidetheintervaloftesting(b

5、)thefatiguelifeataspecificstressstrainamplitudebeestimatedbelowapproximatelythefifthpercentile(P.0.05).Asalternativefatiguemodelsstatisticalanalysesarecontinuallybeingdevelopedlaterrevisionsofthispracticemaysubsequentlyp

6、resentanalysesthatpermitmecompleteinterpretationofSNεNdata.2.ReferencedDocuments2.1ASTMStards:2E206DefinitionsofTermsRelatingtoFatigueTestingtheStatisticalAnalysisofFatigueDataReplacedbyE1150(Withdrawn1988)3E468Practicef

7、PresentationofConstantAmplitudeFatigueTestResultsfMetallicMaterialsE513DefinitionsofTermsRelatingtoConstantAmplitudeLowCycleFatigueTestingReplacedbyE1150(Withdrawn1988)3E606E606MTestMethodfStrainControlledFatigueTesting3

8、.Terminology3.1ThetermsusedinthispracticeshallbeusedasdefinedinDefinitionsE206E513.Inadditionthefollowingterminologyisused:3.1.1dependentvariable—thefatiguelifeN(thelogarithmofthefatiguelife).3.1.1.1Discussion—Log(N)isde

9、notedYinthispractice.3.1.2independentvariable—theedcontrolledvariable(namelystressstrain).ItisdenotedXinthispracticewhenplottedonappropriatecodinates.3.1.3lognmaldistribution—thedistributionofNwhenlog(N)isnmallydistribut

10、ed.(Accdinglyitisconvenienttoanalyzelog(N)usingmethodsbasedonthenmaldistribution.)3.1.4replicate(repeat)tests—nominallyidenticaltestsondifferentromlyedtestspecimensconductedatthesamenominalvalueoftheindependentvariableX.

11、Suchreplicaterepeattestsshouldbeconductedindependentlyfexampleeachreplicatetestshouldinvolveaseparatesetofthetestmachineitssettings.3.1.5runout—nofailureataspecifiednumberofloadcycles(PracticeE468).3.1.5.1Discussion—Thea

12、nalysesillustratedinthispracticedonotapplywhenthedataincludeeitherrunouts(suspendedtests).MeoverthestraightlineapproximationoftheSNεNrelationshipmaynotbeappropriateatlongliveswhenrunoutsarelikely.3.1.5.2Discussion—Fpurpo

13、sesofstatisticalanalysisarunoutmaybeviewedasatestspecimenthathaseitherbeenremovedfromthetestisstillrunningatthetimeofthedataanalysis.4.SignificanceUse4.1Materialsscientistsengineersaremakingincreaseduseofstatisticalanaly

14、sesininterpretingSNεNfatiguedata.Statisticalanalysisapplieswhenthegivendatacanbereasonablyassumedtobearomsampleof(representationof)somespecificdefinedpopulationuniverseofmaterialofinterest(underspecifictestconditions)iti

15、sdesiredeithertoacterizethematerialtopredicttheperfmanceoffutureromsamplesofthematerial(undersimilartestconditions)both.1ThispracticeisunderthejurisdictionofASTMCommitteeE08onFatigueFractureisthedirectresponsibilityofSub

16、committeeE08.04onStructuralApplications.CurrenteditionapprovedOct.12015.PublishedNovember2015.iginallyapprovedin1980.Lastpreviouseditionapprovedin2010asE739–10.DOI:10.1520E073910R15.2FreferencedASTMstardsvisittheASTMwebs

17、itewww.astm.gcontactASTMCustomerServiceat.FAnnualBookofASTMStardsvolumeinfmationrefertothestard’sDocumentSummarypageontheASTMwebsite.3Thelastapprovedversionofthishisticalstardisreferencedonwww.astm.g.Copyright?ASTMIntern

18、ational100BarrHarbDrivePOBoxC700WestConshohockenPA194282959.UnitedStates1NisassumedtobethesameatlowSεlevelsasathighlevelsofSε).AccdinglylogNisusedasthedependent(rom)variableinanalysis.ItisdenotedY.Theindependentvariablei

19、sdenotedX.ItmaybeeitherSεlogSlogεrespectivelydependingonwhichappearstoproduceastraightlineplotftheintervalofSεofinterest.ThusEq1Eq2maybereexpressedasY5A1BX(3)Eq3isusedinsubsequentanalysis.ItmaybestatedmepreciselyasYX5A1B

20、XwhereYXistheexpectedvalueofYgivenX.NOTE4—Ftestingtheadequacyofthelinearmodelsee8.2.NOTE5—TheexpectedvalueisthemeanoftheconceptualpopulationofallY’sgivenaspecificlevelofX.(Themedianmeanareidenticalfthesymmetricalnmaldist

21、ributionassumedinthispracticefY.)6.TestPlanning6.1TestplanningfSNεNtestprogramsisdiscussedinChapter3ofRef(1).4Plannedgrouping(blocking)romizationareessentialfeaturesofawellplannedtestprogram.Inparticulargoodtestmethodolo

22、gyinvolvesuseofplannedgroupingto(a)balancepotentiallyspuriouseffectsofnuisancevariables(fexamplelabatyhuity)(b)allowfpossibletestequipmentmalfunctionduringthetestprogram.7.Sampling7.1Itisvitalthatsamplingproceduresbeadop

23、tedthatassurearomsampleofthematerialbeingtested.Aromsampleisrequiredtostatethatthetestspecimensarerepresentativeoftheconceptualuniverseaboutwhichbothstatisticalengineeringinferencewillbemade.NOTE6—Aromsamplingprocedurepr

24、ovideseachspecimenthatconceivablycouldbeed(tested)anequal(known)opptunityofactuallybeingedateachstageofthesamplingprocess.Thusitispopracticetousespecimensfromasinglesource(plateheatsupplier)whenseekingaromsampleofthemate

25、rialbeingtestedunlessthatparticularsourceisofspecificinterest.NOTE7—Proceduresfusingromnumberstoobtainromsamplestoassignstressstrainamplitudestospecimens(toestablishthetimederoftesting)aregiveninChapter4ofRef(2).7.1.1Sam

26、pleSize—TheminimumnumberofspecimensrequiredinSN(εN)testingdependsonthetypeoftestprogramconducted.ThefollowingguidelinesgiveninChapter3ofRef(1)appearreasonable.TypeofTestMinimumNumberofSpecimensAPreliminaryexplaty(explaty

27、researchdevelopmenttests)6to12Researchdevelopmenttestingofcomponentsspecimens6to12Designallowablesdata12to24Reliabilitydata12to24AIfthevariabilityislargeawideconfidencebwillbeobtainedunlessalargenumberofspecimensareteste

28、d(See8.1.1).7.1.2Replication—ThereplicationguidelinesgiveninChapter3ofRef(1)arebasedonthefollowingdefinition:%replication=100[1?(totalnumberofdifferentstressstrainlevelsusedintestingtotalnumberofspecimenstested)]TypeofTe

29、stPercentReplicationAPreliminaryexplaty(researchdevelopmenttests)17to33minResearchdevelopmenttestingofcomponentsspecimens33to50minDesignallowablesdata50to75minReliabilitydata75to88minANotethatpercentreplicationindicatest

30、heptionofthetotalnumberofspecimenstestedthatmaybeusedfobtaininganestimateofthevariabilityofreplicatetests.7.1.2.1ReplicationExamples—Goodreplication:Supposethattenspecimensareusedinresearchdevelopmentfthetestingofacompon

31、ent.Iftwospecimensaretestedateachoffivestressstrainamplitudesthetestprograminvolves50%replications.Thispercentreplicationisconsideredadequatefmostresearchdevelopmentapplications.Poreplication:Supposeeightdifferentstresss

32、trainamplitudesareusedintestingwithtworeplicatesateachoftwostressstrainamplitudes(noreplicationattheothersixstressstrainamplitudes).Thistestprograminvolvesonly20%replicationwhichisnotgenerallyconsideredadequate.8.Statist

33、icalAnalysis(LinearModelY=ABXLogNmalFatigueLifeDistributionwithConstantVarianceAlongtheEntireIntervalofXUsedinTestingNoRunoutsSuspendedTestsBothCompletelyRomizedDesignTestProgram)8.1Fthecasewhere(a)thefatiguelifedatapert

34、aintoaromsample(allYiareindependent)(b)thereareneitherrunoutsnsuspendedtestswhereftheentireintervalofXusedintesting(c)theSNεNrelationshipisdescribedbythelinearmodelY=ABX(mepreciselybyYX5ABX)(d)the(twoparameter)lognmaldis

35、tributiondescribesthefatiguelifeN(e)thevarianceofthelognmaldistributionisconstantthemaximumlikelihoodestimatsofABareasfollows:A?5Y2B?X(4)B?5(i51k~Xi2X!~Yi2Y!(i51k~Xi2X!2(5)wherethesymbol“caret”(^)denotesestimate(estimat)