Previous|Next|Contents

ESDEPWG12

FATIGUE

Lecture12.8:BasicFatigueDesign
ConceptsinEurocode3
OBJECTIVE/SCOPE:

ThislecturecontainsthebackgroundinformationofthebasisoftheEurocode3rulesconcerningthefatiguedesignofstructuralelements.

PREREQUISITES

None.

RELATEDLECTURES

Lecture12.1:BasicIntroductiontoFatigue

Lecture12.2:AdvancedIntroductiontoFatigue

SUMMARY

ThelecturediscussesthemainfatiguedesignrulescontainedinEurocode3[1].Thesefatiguedesignrulesarebasedonfatiguetestresults
obtainedmainlyunderconstantamplitudeloading.Theclassificationofagivendetail,eitherweldedorbolted,resultsfromastatistical
evaluationofthefatiguetestdatawitha95%probabilityofsurvivalfora75%confidenceinterval.Theevaluationiscomparedwithasetof
equallyspacedSNcurveswithaslopeconstantofm=3.

ExplanationisgivenonthechoiceofanormaliseddoubleslopesSNcurve.Thenseveralfactors,introducedinEurocode3[1],affecting
thefatiguestrengtharealsodiscussed.

1.INTRODUCTION
TheprincipalobjectiveofthislectureistoreviewthemainruleswhicharethebasisforChapter9ofEurocode3[1]concerningthefatigue
strengthassessmentofsteelstructuraldetails.

ThemainprovisionsofEurocode3[1]relyuponasetoffatigueresistancecurves,equallyspaced,uponwhichareclassifiedasetof
constructionaldetails.TheconceptforfatiguestrengthdesignfollowstheRecommendationsoftheEuropeanConventionfor
ConstructionalSteelwork(ECCS).TheRecommendations[2]defineasetofequallyspacedfatiguestrengthcurveswithaconstantslopeof
m=3(fornormalstress),orm=5(forshearstress,hollowsectionjoints,andsomeparticulardetails).

Inadditiontothisapproachanotherconceptsupportedmainlybyrecentdevelopmentsandresearchinthefieldoffatiguefor"offshore"
structuresisreferredtoinEurocode3asthegeometricalstressconcentrationconcept(alsocalledthe"hotspotstress"method).

TodeterminethefatiguestrengthprovisionsgiveninEurocode3,acompilationoffatiguedataofvarioussourceswascarriedout.This
workhasprovidedanopportunitytoreevaluateexistingfatiguetestdataandallowedforamoreconsistentapproachtotheclassificationof
detailcategories.

2.PRACTICALIMPLICATIONOFDESIGNCRITERIA
2.1MainFactorsAffectingtheFatigueStrength

Fatigueofsteelstructuralcomponents,especiallyweldedsteeldetails,isaparticularlycomplexproblem,andmanyfactorsmayexertan
influenceonthefatiguelife.Table1listsanonexhaustiveinventoryofthesevariousfactorsandthosewhicharetakenintoaccounteither
explicitlyorimplicitlyinChapter9ofEurocode3areindicated.

WhilstsomefactorsaredealtwithinChapter9ofEurocode3,otherfactors,particularlythoserelatedtofabricationareconsideredinan
implicitmannerthroughdefineddiscontinuitiesorwelddefectsacceptancecriteriaandqualitycontrolrequirements.Thesegeneral
requirementswillbedefinedinastandardconcerningthe"Executionofsteelstructures".

Table1Themainfactorsaffectingfatiguestrength

Designation of the factors affecting the fatigue Taken into

strength account in
Eurocode 3
 Stress
Stress or strain range *
Stress sequence
Frequency (no significant effect when < 40 Hz in
a non corrosive environment)
Mean stress (no effect in heat affected zone due *
to residual stresses) *
Residual stresses
Geometry
Nominal or geometrical stress
Local stress concentration *
Small discontinues *
scratches *(implicit)
grinding marks
surface pittings
weld defects or misalignments
Size effect (or scale effect)
*
Material Properties and Fabrication
Stressstrain behaviour of materials
Hardness
Chemical composition of steels
Metallurgical homogeneity
Electrical potential
Micro structural discontinuities (grain size, grain
boundaries)
Welding process
Weld heat treatment
Weld surface treatment
Environment
Corrosive atmosphere *(implicit)
Temperature *(implicit)
Humidity (hydrogen embrittlement)
Irradiation

2.2FatigueFailureCriteria

InthepreparationofEurocode3,classificationintodetailcategorieswasestablishedfromastatisticalanalysisoffatiguetestdataobtained
fromvariouslaboratorysources.Toobtainmorehomogeneoussamplesofthetestresults,particularattentionwaspaidtofailurecriteria
consideredinthesetests.

Severalfailurecriteriamaybeadoptedtocharacterizetheexperimentalfailureconditionattheendofafatiguetestinthelaboratory.Three
criteriaaregenerallyconsidered:

Firstappearanceofacrackeitherdetectedvisuallyordetectedbymeansofaphysicalmeasure,e.g.bytherecordofachangeinthe
localstraincondition.
Throughthicknesscrack:thefatiguecrackstartsfromthefrontsurfaceandgrowsthroughthethicknessofthetestpieceandreaches
thebacksurface.
Completefractureofthetestedspecimenorlargedisplacementofthetestedstructuralelementsuchthatthedisplacementbecomesso
importantthattheapplied"jackload"cannotbemaintained.Whenperformingafatiguetestonabeam,thefailuremaybe
conventionallydefinedasthepointwhenthemidspandeflectionreachesacertainlimit.

Generallyforsmallscalespecimens,thedifferencebetweenthefatiguelifeatcompletefractureandatamorerealistictolerablefatigue
cracksizeisnegligible.However,inalargescalestructuralelementtestedinfatiguethedifferencemaybehighlysignificant.

InEurocode3,thefatiguestrengthreferstothecompletefailureofthestructuralelement.Thisconditioncorresponds,usually,tothe
criteriongenerallyadoptedbystructurallaboratoriesorreportedinliterature.

2.3DesignStressesforFatigueAssessment
Differentstressesmayaffectthefatiguestrengthclassificationofastructuraldetail.Foraparticulardetail,thevariousoriginsofstresses
havetobeidentifiedinordertodefinemorepreciselythedesignstressesforthefatigueassessmentconceptsinvolvedinChapter9of
Eurocode3.

a.NominalStress
Considerauniformstructuralmembersubjectedtoasimpleaxialforceortoabendingmoment.Thenominalstressisthestressresultant
calculatedaccordingtothebasicstrengthofmaterial(Figure1).

Thenominalstressofamemberunderuniaxialstressis:

N= (2.1)

whereNisthenormalforceandAthegrosssectionarea.

Foraprismaticmembersectionunderabendingmoment,thestressresultantis:

M= (2.2)

where:

Mistheappliedbendingmoment

Iisthemomentofinertiaofthesection

visthedistancefromtheneutralaxistotheoutmostfibre.

b.Stressconcentrationeffectduetogeometricaldiscontinuities

Therearethreemainsourceswhichcancreateastateofstressconcentrationinastructuraldetail:

Theglobalgeometryofthestructuralelementwhichcontainsthestructuraldetail,e.g.attachmentsonabeamweborgusset
platesonabeamflange.

Thelocalstressconcentrationduetolocaldisturbanceoftheweldgeometry,boltholes,localvariationinstiffness,etc...For
example,ifaholeisdrilledinaplate,thestressdistributionacrossthesectioncontainingtheholewillbedifferentfromthe
nominalstressdistributionexistingintheplainplatecrosssection.Animportantstressgradientwilloccurinthevicinityofthe
hole.Thisgeometricalstressconcentrationisduetoboththedecreasefromthegrosssectiontothenetsectionandtothestress
"raiser"(concentrator)causedbythepresenceofthehole(Figure1).

Thelocalstressconcentrationduetolocaldiscontinuitiesoccurringduringfabrication(misalignment,surfacescratch,pitting,
welddefect,etc).

Inmanycases,andbysimplification,thegeometricstressconcentrationisusuallycalculatedonthebasisofthenominalstressappliedto
thegrosssectionareaandthestressconcentrationfactorkG,as:

G=kG.nom(2.3)
Thisstructuralgeometricalstressconcentration,whichisdefinedasthemaximumprincipalstressexistinginthevicinityofthedetail,may
beevaluatedfromexperimentaltestsorfromfiniteelementmethods.

Thelocalstressconcentrationispresentinadditiontothestructuralgeometricstressconcentrationandmaybeduetolocaldisturbancesof
thelocalgeometryofthedetailsuchas:

localcrosssectionchange(geometryofweldsforexample).

localgeometricalimperfectionssuchasmisalignment.

smalllocaldiscontinuitiesinherenttotheactionoftheenvironmentorofthefabricationprocesssuchcorrosionpits,surface
scratches,draglinesduetoflamecutting,grindingmarks,weldingprocessdefectssuchasundercut,lackofpenetration,lackof
fusion,slaginclusions,porosities,hydrogeninducedcracking,etc.Theseverysmalldiscontinuitiesarepresentinevery
elementofengineeringstructures.Theirpresencedeterminesapotentiallocationforinitiationofafatiguecrack.

LocalstressconcentrationsaretakenintoaccountinanimplicitmannerinthederivationoftheSNcurvefromfatiguetestresults.Great
caremustbetakenwhenassessingfatiguestrengthfromtestsonsmallscalespecimensinsteadoflargescalespecimens.Thescaleeffect
duetoweldgeometrymayhaveagreaterinfluenceonthefatiguestrengthinsmalltestspecimensthaninlargetestspecimens.

Usually,fatiguespecimenshavebeentestedwithinherentdiscontinuities,andfatiguestrengthcurves,soderived,makeallowancefor
tolerabledefects.Theacceptancecriteriaforwelddiscontinuitieswhichwillbeproposedinthe"Executionofsteelstructures"standard
wouldguaranteethefitnessforpurposeofthefatiguestrengthdesignrulesofEurocode3.Inotherwords,thequalityassurancesystem
whichcoversthefabricationprocessshouldensurethatthefabricatedconstructionaldetailcomplieswiththerelevantqualityrequirement
specifiedinthestandardforthe"Executionofsteelstructures".

Whenassessingthefatiguestrengthbythesocalledgeometricstressrangemethod,accordingtoClause9.5.3ofEurocode3,thegeometric
stressconcentrationasdefinedbyEquation(2.3)mustbeproperlyevaluated.Thelocalgeometryoftheweldmustnotbetakenintoaccount
inthecalculationprocedureofthedesignstressrange,sincethelocaldiscontinuityeffectisalreadyintroducedinthederivationoftheSN
curves.However,whendeterminingthedesignstress,secondarystressesarisingfromjointeccentricityorduetojointstiffness,stress
redistributionduetobucklingorshearlag,andeffectssuchaspryingaction,shouldbetakenintoaccount.

3.DESIGNSTRESSSPECTRUM
3.1StressHistory

Afluctuatingstresstowhichastructuraldetailissubjectedmayhaveastresshistoryofconstantamplitudeorofvariableamplitude
(Figures2and3).
Forcumulativedamageanalysis,thestresshistoryissplitupintoindividualcyclesandrelatedstressrangeswhicharesummeduptoa
distributionofstressranges.Thisdistributionofstressrangesiscalledastressspectrum,seeLecture12.2.

Foravariableamplitudestresshistory,thereisaneedtodefinesuchastresscycleassociatedwithaparticularstressrange.Thereare
severalproceduresforcyclecountingmethods.Eurocode3referstothe"reservoirmethod"whichgivesasoundrepresentationofthestress
variationcharacteristicbyallowingapropercontributionofeachstressrangetothefatiguedamageprocess.Thisstressrangecounting
methodisthemostcommonlyaccepted.Thiscountingmethodissomewhatsimilartothewellknown"rainflowcountingmethod".The
"rainflow"andthe"reservoir"countingmethodsdonotleadtoexactlythesameresult.However,intermsoffatiguedamagebothcounting
proceduresgiveverycloseresults,andfor"long"stresshistoriestheygivenearlythesameresult.

3.2StressHistogram
Themostcommonwayofrepresentingirregularstresshistoriesforfatigueanalysisistosumupthestressrangesofequalamplitude,andto
obtainadistributionofstressrangeblockswhichiscalledastresshistogram(orastressspectrum)consistingofanumberofconstantstress
rangeblocks.Eachblockischaracterizedbyitsnumberofcyclesniandstressrangei(Figure4).Theorderingofthedifferentblocks
doesnotmakeanydifferencesincethedamagecalculationrulesspecifiedinEurocode3referstothelinearcumulativedamageruleof
PalmgrenMiner.Howeverforconveniencethestresshistogramiscommonlypresentedwithstressblocksrankedindecreasingorder
(Figure5)whichoftencanbeapproximatedbyatwoparametersWeibulldistributionsuchas:
=0 (2.4)

4.FATIGUEDESIGNCURVESCLASSIFICATIONCONCEPT
TheclassifiedfatiguedesigncurvesadoptedinEurocode3,arethesameasproposedinthe"EuropeanConventionforConstruction
SteelworkFatigueRecommendations"[2].TheECCSFatigueRecommendationswereoneofthefirstattemptstoprovideuniformitytothe
determinationofthefatiguestrengthdesigncurves.

TheECCSRecommendationsdefineasetofequallyspacedSNcurvesplottedonaloglogscale.Referencetothesecurvesallowsadetail
categorytobeclassified(representative)ofaparticularstructuraldetailwhichcorrespondstoanotcheffectoracharacteristicgeometrical
discontinuity).Thisclassificationhasbeendeterminedbyaseriesoffatiguetestresults,fromwhichastatisticalandaprobabilistic
evaluationisperformed,seeLecture12.7.

Eachindividualfatiguestrengthcurveisdefinedinaconventionalway(Figure6)byaslopeconstantofm=3(slope=1/3).Theconstant
amplitudelimitissetat5millioncycles.Theslopeconstantm=3wasabestfitforalargenumberofdifferentstructuraldetailstestedin
fatigue.Thefigureof5millioncyclesfortheconstantamplitudefatiguelimitisacompromisebetween2millioncyclesfor"good"details
and10millioncyclesfordetailswhichcreateaseverenotcheffect.Foranystressrangeofconstantamplitudebelowthislimit,nofatigue
damageisexpectedtooccur.

Whenadetailissubjectedtovariablestressranges,whichisgenerallythecaseinreality,severaloptionsmayoccur:

Ifnostressrangeofvariableamplitudeexceedsthefatiguelimit,nofatiguedamageassessmenthastobecarriedout.
 Ifatleastonestressrangeblockexceedsthisfatiguelimit,adamagecalculationhastobeperformedonthebasisofthelinear
cumulativedamagerule,referredasthePalmgrenMiner'srule.

Inthislastoption,twocaseshavetobeconsideredforthecumulativedamagecalculationwhensomestressrangesarebelowtheconstant
amplitudefatiguelimit:

EitherthedamagecalculationismadesimplyassumingthattheSNcurveofslopeconstantm=3isextendedbeyondtheconstant
amplitudefatiguelimit.
Orthedamagecalculationismadeassumingthatbeyondtheconstantamplitudefatiguelimit,theSNcurveofslopeconstantm=3
isextendedbyastraightlineofslopeconstantm=5.Theinterceptofthisstraightlinewiththeverticallineat10millioncycles
providesacutofflimit.ThereasonforusinganSNcurvewithtwoslopestocumulativedamagecalculationsisthatitisan
approximatewaytotakeintoaccounttheprogressivereductionoftheconstantamplitudefatiguelimitasaresultofthedamage
causedbythestressrangesabovethatlimit.Inthisway,eventuallyallstressrangesinthespectrumbecomedamaging.Fracture
mechanicsconfirmthisdecreaseoftheslopeoftheSNcurveinthelongfatigueliferange.

Inbothcases,allcyclesbelowcutofflimitcanbeignoredwhenevaluatingthefatiguedamage.ItshouldbenotedthatEurocode3leaves
thedesignengineerfreetouseeitherthesingleslopeSNcurveorthedoubleslopeSNcurve.

Experimentalresultshaveindicatedthatwithintherangeofhighnumbersofcycles,achangeintheslopeofthefatiguestrengthoccurdue
toadecreaseofthecrackgrowthrate.Theintroductionofadoubleslopeconceptandaconstantamplitudefatiguelimitat5millioncycles
isstillamatterofcontroversy.Despiteanumberofcriticisms,particularlyconcerningtheincreaseincomplexityoftheanalysis,Eurocode
3haskeptthedoubleslopecurvebecausethisrulemay,forsomedetailcategories,improvetheaccuracyofthefatiguecheck.However,
thisimprovementcannotexpectedforalltypesofstructuraldetail,andallstressspectra.Insomecases,especiallyforthosedetailswitha
veryseverenotcheffect,thedoubleslopecurvemaynotleadtoaconservativeresult.

Somedetails,forexample,coverplatedbeams,haveshownaconstantamplitudefatiguelimitofalmost10millioncycles.Toavoidnon
conservativeconditions,somedetails(whichgenerallyhaveseverenotcheffect)havebeenclassifiedincategoriesslightlylowerthantheir
fatiguestrengthat2millioncycleswouldhaverequired.TheconceptofthespecifiedECCSfatiguedesigncurves,whichconsistsof14
equallyspacedcurves,anewdesignfatiguestrengthcurveisnotrequiredforeachnewstructuraldetail.

The"gridsystem"ofSNcurveshasbeenestablishedasfollows.Theverticaldistanceoftheordinatelogscalebetweeneachfatigue
strengthcurveshasbeenobtainedbydividingthedifferencebetweenoneorderofmagnitudeinto20equalspaces(Figure7).Forexample,
takingtworeferencevaluesasc=100MPaandc=1000MPaat2millioncycles,thecalculationofthespacingisdeterminedfromthe
following:

ThegeneralSNcurveequationmaybewrittenas:

logN=loga3log(4.1)

sowithc=100MPa(log2000000=6,30103)

loga=6,30103+3log100=12,301(4.2)

andforc=1000MPa

loga=6,30103+3log1000=15,301(4.3)

Thespacingbetweentwocontiguouscurvesrepresents

loga=(15,30112,301)/20=0,15(4.4)
Sostartingfromthereferencevaluesofc=100MPa,withloga=12,301,thesubsequentvaluesofcmaybeobtainedfromEquation
(4.1)asgiveninTable2.

Table2Characteristicfatiguestrengthat2millioncycles

loga c(roundedvalue)

... ...

12,601 125

12,451 112

12,301 100

12,151 90

12,001 80

... ...

Table2showsthatthenumberdefiningthecharacteristicfatiguestrengthat2millioncycles,usedasadetailcategoryidentification,isa
roundedvalue.

5.FATIGUETESTRESULTS
Generallyfatiguestrengthcurvesareevaluatedfromseriesoffatiguetestsperformedonspecimenswhichtypicallyreproducethedetailto
bestudied.Thefatiguestrengthcurves(SNcurves)canbemostaccuratelydeterminedwhenagroupoffatiguespecimensaretestedat
differentstressrangelevels.However,thereisnorecognizedstandardmethodforfatiguetestinganddesignexperiments.Asaresult,the
fatiguetestdatafoundintheliteraturearesomewhatnonhomogeneous.

Itisclearthat,undersuchcircumstances,areviewofexistingfatiguedataandtheirstatisticalevaluation,evenwhenlimitedtothesame
detailcategory,mayleadtolargediscrepanciesintheresults.Suchdifferencesmaybeattributed,notonlytothefatiguetestingpracticein
eachlaboratory,butalsotothedetailedfabricationprocedureandqualityachievedinthepreparationofthespecimens.Discontinuitiesplay
amajorroleinfatiguestrength,particularlyforweldeddetailsandcarefulconsiderationmustbegiventotheweldqualitywhichmay
considerablyaffectthevariationinfatiguestrength.

Fatiguespecimensarefabricatedwithcertaininherentdiscontinuitieswhicharenotfullyknownormaynotbeproperlyevaluatedin
laboratoryreports.Insuchcases,itisgenerallyratherdifficulttoappreciateifthefabricationqualityofspecimensisrepresentativeof
currentworkshoppractice.Moreover,whenperformingastatisticalanalysisonfatiguetestdatafromdifferentorigins,aratherlarge
variationoffatiguestrengthmayresult.Carefulattentionmustbepaidtothehomogeneityofthefatigueresistance.

TheseconsiderationswereborneinmindduringthepreparationofEurocode3.Thefatiguetestresultswhichwerestatisticallyanalyzed
andthenclassifiedaccordingtotheproceduredescribedfulfilcertainrequirements:

Prioritywasgiventotestresultsfromfullsizespecimenscomparedtosmallscalespecimenssimulatingthesamestructuraldetail.
Foracomparablequalityofweldments,smallerweldedtestspecimensexhibitahigherfatiguestrength(andahigherconstantslope)
thanfullsizetestspecimens.Thisdifferenceinfatiguebehaviourismainlyduetothefactthatfullsizespecimenslockinmore
residualweldingstressesthansmallsizespecimensdo.Thisdifferenceisresidualstressmagnitudeistheresultofvariationsin
mechanicalconstraintsduringwelding.
Inweldedspecimensthestressrange()andthenumberofcyclestofailure(N)wereconsideredasthemainparametercontrolling
thefatiguestrengthcurve.
Aminimumof12fatiguetestresultswererequiredtoreachacertainsignificancelevelandtoleadstatisticallytoaconfident
interpretationofthetestresults.

6.CUMULATIVEDAMAGERULE,EQUIVALENTSTRESSRANGECONCEPT
6.1PalmgrenMinerSummation
Inreallife,structuralelementsaresubjectedtovaryingfatigueloads,andnottoconstantamplitudefatigueloadings.Eurocode3refersto
thePalmgrenMinersummationtoevaluatethecumulativedamage(Figure8).Thisruleisbasedontheassumptionthatthetotaldamage
accumulatedbyastructuralelementundervaryingstressranges,isobtainedbythelinearsummationofthedamageofeachindividual
stressrange,i.e:
D= (6.1)

where:

niisthenumberofcyclesofconstantamplitudestressrangesi

Niisthetotalnumberofcyclestofailureunderconstantamplitudestressrangei.

Thestructuralelementisdesignedsafelyagainstfatigueif:

D1(6.2)

Noaccountofthedamageistakenforanyvaryingstressrangesfallingbelowthecutofflimit.

6.2EquivalentStressRange
TheconceptofequivalentstressrangehasbeenintroducedintheECCSRecommendations[2]andisalsoreferredtoinEurocode3.The
definitionoftheequivalentstressrangeisconventional.Itcanbesaidthattheequivalentstressrangeconceptissimplerthanadirect
PalmgrenMinersummationwhentheSNcurveisofuniqueslope(1/m).Theexpressionis,inthiscase,quitesimpleandtherecalculation
ofthedamageforeachSNcurveisthereforeavoided:

equ= (6.3)

withm=3orm=5asappropriate.

Theequivalentstressrangeequdependsonlyonthefatigueloadspectrumandtheslopeconstantm.Insuchacase,knowingequ
evaluatedaccordingtoEquation(6.3),itiseasytochoosedirectlyadetailcategorywhichwillhaveanadequatefatigueresistance.

6.3EquivalentStressRangeforanSNCurvewithadoubleSlopesConstant

WhenthebasicSNcurveisofdoubleslope,theexpressionoftheequivalentstressrangebecomesmoreunwieldy.Thepracticabilityofits
applicationisquestionable,exceptifusingthelimitstatefunctionasdefinedbythefollowingequation:

f.equRd/f(6.3)

ThederivationofequwhentheSNcurvehasadoubleslopeisgivenbelow:

a.DamagecalculationforadoubleslopeSNcurvewhenthestressrangeisbelowandabove D

SupposetherearesomestressrangeblockswheretherangeisbelowthevalueofDandsomeaboveD(Figure9)itisassumedthat
theproperpartialsafetycoefficientshaveintroducediniandj.
blockiwheni>D

blockjwhenj>D

Fromthedefinitionthedamageisgivenby:

D= (6.5)

takingintoaccounttheSNcurveslopeforeachsetofstressrangeblocks:

D= (6.6)

Equation(6.6)maybewrittenas:

D= (6.7)

FromFigure9:

ND=aD3=bD5

NDcorrespondstothefatiguelimitoftheSNcurveat5millioncycles.

a/b=1/D2(6.8)

Hence:

D= (6.9)

where:

Q=nii3+njj3(j/D)2

ThedamagemaybecalculatedusingeitherEquation(6.5)orEquation(6.9)directly.

b.Calculationoftheequivalentstressrange equforadoubleslopeSNcurve

Inthisparticularcase,adecisionmustbemadeastowhichslopethedefinitionofequrefers.Thechoiceofaslopeconstantof3or5
makesabsolutelynodifferencetothefinalresultofthecalculationofequwhentheloadspectrumstraddlesbothpartsofthedoubleslope
SNcurve.Thecalculationoftheequivalentstressrangeequisderivedbelowfromaslopeconstantofm=3ofthedoubleslopeSN
curve(notedasequ.3).Thesamedemonstrationholdsforaslopeconstantofm=5.Bydefinition:
D= (6.10)

where:

Nequistheequivalentnumberofcyclesatfailureundertheequivalentstressrangeequ

Nisequaltoni+nj

EvaluatingNequonthebasisoftheSNcurveofslopeconstantm=3:

D= (6.11)

byequatingEquations(6.6)and(6.11),thedamageis:

D= (6.12)

thenEquations(6.11)and(6.12)give:

equ3= (6.13)

therefore:

equ.3= (6.14)

Rd.3isdefinedasthefatigueresistancecorrespondingtoequ.3ontheSNcurveofconstantslopem=3.

Rd.3=D(ND/N)1/3(6.15)

FromEquations(6.14)and(6.15):

= = (6.16)

ThisexpressionisequaltothedamageasgivenbyEquation(6.9):

= (6.17)

Remarks:

1.Bothfatigueassessmentformats,thePalmgrenMinersummation,andtheequivalentstressrangeconcept,arerigorouslyequivalent
intermsofdamage.
2.ReferenceintheabovedemonstrationismadetoDandNDcorrespondingtothe"knee"pointofthedoubleslopeSNcurve.Since
theSNcurveiswrittenas:

N(Rd)m=a=constant

anotherreferencevaluemaybetaken,forexample:

D3ND=C3NC=constant

C,beingthestressrangeatNC=2millioncycles.

3.Specialcaremustbetakenwhencalculatingequ.3andRd.3:bothexpressionsmustbeevaluatedwiththesameslopeconstant.
4.Thevaluesofequ.3andRd.3areclearlydifferentandmaynotbeusedindiscriminatelywhenplottingfatiguetestresultsonalog
versuslogNdiagram.Generallywhenfatiguetestshavebeenperformedundervariablestressrangeamplitude,theequivalent
stressrangeasgivenbyEquation(6.3)hasbeenusedtoplottheexperimentalresults.
7.RESIDUALSTRESSEFFECT
Weldedjointsinstructuraldetailscontaintensileresidualstressesinthevicinityoftheweldbead.Figure10showsthattheirmagnitude
maybeashighastheyieldstressoftheweldmentmetal.Figure10alsoshowshightensileresidualstressesneartheedgeswhichwere
flamecut.

Itiswellestablishedthatthepresenceofresidualstressesofsuchmagnitudemakesthefatiguestrengthofaweldedjointindependentofthe
appliedloadratio,anddependentonlyontheappliedstressrange.Thefullsignificanceofthetensileresidualstressesduetoweldingwas
notappreciatedoriginally,sincemanyfatiguetestresultswereobtainedfromweldedspecimenswhichweretoosmalltoretainthemajor
partoftheweldingresidualstressessuchaswouldoccurinlargestructuralcomponents.

Itisevidentthattensilestressesplayasignificantroleinthepropagationofacrack,sincetheytendtoactasaopeningmodeduetotensile
stressesappliedatthecracklips.Thecrackpropagationrateislikelytobereduced,whenthecrackgrowsintoazoneofcompression
residualstress.

ItisinrecognitionofthisphysicalcrackpropagationbehaviourthattheRratio(R=min/max)hasbeenconsideredinEurocode3Chapter
9fornonweldedorstressrelieveddetails.Figure11showsthecomparisonbetweenfatiguetestresultsandtwo"bonusfactor"ruleswhich
werestudiedwhendraftingChapter9.Therulewhichwasfinallyselectedtakesintoaccountoftheeffectofcompressivestressrangesby
multiplyingthepartofthestressrangeincompressionbyafactorof0,6.Thevalidityofthisrulehasbeencomparedwithfatiguetest
resultsperformedonnonloadcarryingweldcruciformjointsforvariousRratiosrangingfrom3,0to0,8.Thesefatiguetestswerecarried
outonsmallspecimens.
8.CONCLUDINGSUMMARY
Fatiguebehaviourofstructuraldetailsisgovernedbymanyfactorswhicharebynaturerandom.
Thepresentstateofknowledgeprovidessufficientinformationforreasonablycomprehensiveandsafefatiguedesignrules.Itis
recognizedthattheextremeliferegionofthefatiguestrengthcurveisnotwellestablished.Thecurrentavailabilitytestdatainthis
regionareverysparse.
Thequalityoffatiguedesigniscloselyrelatedtotheattentiongiventostructuraldetails,i.e.notonlytothegeometricalshapeandthe
dimensions,butalsotothequalityoffabricationandtheacceptabledefects,etc.
Incarryingoutthefatigueassessmentofstructures,thedesignermust,first,carryoutaproperfatigueloadanalysisinorderto
evaluatecorrectlythestressresultantsactingondetails.
Thedesignermustthenselecttheproperfatiguestrengthcurverelatedtoeachofthesedetails.
Boththeanalysisandcurveselectionrequireskillinrecognitionandinterpretationofthemaindesignfactorsaffectingfatigue
resistance.

9.REFERENCES
[1]Eurocode3:"DesignofSteelStructures":ENV199311:Part1.1,Generalrulesandrulesforbuildings,CEN,1993.

[2]EuropeanConventionforConstructionalSteelwork:RecommendationsfortheFatigueDesignofSteelStructures.ECCSPublication43,
1985.

[3]Eurocode1:"BasisofDesignandActionsonStructures",CEN(inpreparation).

Previous|Next|Contents