Hydrogen Induced Stress Cracking of Duplex Stainless Steel Subsea Components
Hydrogen Induced Stress Cracking of Duplex Stainless Steel Subsea Components
Hydrogen Induced Stress Cracking of Duplex Stainless Steel Subsea Components
Hydrogeninducedstresscrackingofduplexstainlesssteelsubsea
components
AmirBahramiandPaulWoollin
TWILtd
Cambridge,UK
Paperpresentedat29thInternationalConferenceonOffshoreMechanicsandArcticEngineering(OMAE2010),Shanghai,China,6
11June2010.
Abstract
Asmallnumberofduplexandsuperduplexstainlesssteelcomponentshavefailedinsubseaserviceduetohydrogeninducedstress
cracking(HISC).ThesignificanceofthesefailureshasledtoresearchtodefinecriticalloadingconditionsforHISC,toallow
confidentdesignofcomponentsinfuture.DatarelatingtotheFoinavensuperduplexhubfailureswerepublishedatOTCin1999and
NACECorrosionconferencein2001anddatafromTWIGroupSponsoredProjectswerepublishedatOMAEin2004.DNVRPF112
hasbeenbasedontheseandotherdata,toprovideaconservativeapproachtodesign.
ThereareanumberofgapsinthepublishedliteratureandinthedataavailablewhenDNVRPF112wasprepared,relatedto
differencesbetweensmallscalespecimenstestedattypicalseabedtemperatureandactualcomponentsinoperation,iethe
operatingtemperatureandpressure,anddatafromfullscaletestsonpipematerialwithfineaustenitespacingandhencegood
resistancetoHISC.
Thepaperpresentsnewdataontheseissuesandindicateswheretechnologygapsremain.
Introduction
Asmallnumberofduplexandsuperduplexstainlesssteelcomponentshavefailedinsubseaserviceduetohydrogeninducedstress
cracking(HISC).ThesignificanceofthesefailureshasledtoresearchaimedatdefiningcriticalconditionsforHISC,toallow
confidentdesignofcomponentsinfuture.DatarelatingtotheFoinavensuperduplexhubfailureswerepublishedatOTCin1999[1]
andtheNACEcorrosionconferencein2001[2]anddatafromothertwiinvestigationswerepublishedatOMAEin2004.[3]
ThesusceptibilitytoHISCshowsastrongcorrelationwithaustenitespacingandtestsperformedonsmoothsampleshaveshown
thatcoarsegrainedmicrostructuresaremoresusceptibletoHISC.[3]Smallscaleconstantloadtestsinseawaterwithcathodic
protection(CP)aregenerallyusedtocharacterisetheHISCsusceptibilityofduplexandsuperduplexmaterials.[2]Constantload,
tensileHISCtestsonhubmaterialsfromtheFoinavenfield,at1100mVSCEshowedthatifthesuperduplexmaterialwasloadedto
aninitialstrainof0.5%,creepandcrackinitiationwouldensueveryquicklyandfailurewouldeventuallyoccuratasubstantially
highertotalstrain,asaconsequenceoflowtemperaturecreep.[1]Testsonfullscalehubshavebeenreported,[1,2]whichindicated
thatoncecrackshadinitiated,theycouldpropagatethroughthicknessinabout10dayswithoutfurtherincreaseinappliedload.
HenceitwasconcludedthatinitiationofHISCmustbeavoidediffailureistobeprevented,andtheassessmentcriterionshouldbe
relatedtothethresholdinitialstressorstrainforinitiationofHISC.[1]
DNVRPF112[4,5]hasbeenbasedontheseandotherdata,toprovideaconservativeapproachtodesign.Howeverdataare
availabletoshowthatsomeproductformsincludingpipesaresubstantiallymoreresistanttoHISCthancoarsegrainedforgings,
duetofineraustenitespacing.RPF112doesnotallowadvantageofthisfinerspacingtobetakenreliably,asitisbasedona
measurementofaustenitespacing,whichisnotthesubjectofarecognisedstandardwithanecdotalevidencefromindustrythatitis
notreproducibleandhenceopentomisinterpretation.Therewerealsoanumberofothergapsintheavailabledataavailablewhen
DNVRPF112wasprepared,relatedto(i)differencesbetweensmallscalespecimenstestedattypicalseabedtemperatureand
ambientpressureandactualcomponentsinoperation,ieatelevatedoperatingtemperatureandpressure,and(ii)fullscaletestsfor
finegrainedpipematerial.Thispaperpresentsdataontheseissuesandindicateswheretechnologygapsremain.
Experimentalprocedure
Introduction
SmallscaleHISCtestswereperformedontensilespecimenstakenfromasfairlycoarsegrainedsuperduplexstainlesssteel.Tests
wereof30daydurationandperformedatapotentialof1100mVSCE.Comparativetestswereperformedat20and80C(1bara),
and1and100bara(20C).
LargescalefourpointbendHISCtestingwascarriedoutonfinegrainedgirthweldedandfilletweldedseamlesspipesfora
maximumdurationexceedingsixmonths.Residualstressmeasurementsweretakenpriortotesting.Testingwasperformedin
seawaterundercathodicprotectionat1100mVandstrainwasrecordedduringtestingatdifferentlocationsontheweldedpipe.
Visualinspection,dyepenetrantexamination,metallographicandfractographicstudieswereperformed.
Materials
ThemicrostructuralcharacteristicsofthefivematerialsaresummarisedinTable1.MaterialAhadafairlycoarsealignedaustenite
structure,withfairlyconsistentausteniteislandsize.MaterialBhada'primary'coarse,alignedaustenitestructureandfiner,random
equiaxed'secondary'austeniteislandsinbetweenthecoarseunits.Themeasurementsofaustenitespacingweremadeinthis
material(i)includingallaustenite,and(ii)toreflectonlythecoarseprimaryaustenitespacing,ieignoringthefinesecondary
austenite.
Table1Materialsexamined
Materialtype
Averageaustenitespacing(m)
Ferrite(%) Hardness(HV5) Thirdphases
20(transverse)
555
261
None
39(longitudinal)
Transverse:8(11ignoringfineaustenite)
B:22%Cr(UNSS32205)pipe
565
244
None
Longitudinal:11(43)
A:25%Cr(UNSS32760)bar
EffectoftemperatureandpressureonsmallscaleHISCtests
TwoseriesofconstantloadtensileHISCtestswereperformedonmaterialA(25%Crsuperduplexbar)withspecimensmachinedin
thelongitudinaldirectionprechargingandtestingwereat1100mVSCEand1barapressureat20and80C,innaturalseawater.A
saltbridgewasused,sothattheAg/AgClreferenceelectrodecouldbekeptcool.
AdditionalconstantloadHISCtestswereperformedat1100mVSCE,20CinnaturalseawateronmaterialA(specimensmachined
longitudinally)withprechargingandtestingat100bara.Theautoclavewaspressurisedwithnitrogen.AnAg/AgClreference
electrodewasused.Specimenswerestressedtoaroundthethresholdstressesforcrackpropagationandinitiationin30days,as
establishedatroomtemperatureandpressure(749and553MParespectively).Allowancefortheinternalautoclavepressurewas
madetothemeasuredappliedstressonthespecimens,asdescribedinEFC17.[6]
Aftertestthespecimenhydrogencontentsweremeasuredbyvacuumhotextractionandcracknumbersanddepthsweremeasured
onmetallographicsectionsthroughspecimensthathadnotfailedattheendoftest.
Fullscaleweldedpipetests
PipesamplesinmaterialB(22%Crduplex),whichwere4mlong,with15mmwallthicknessand168mmouterdiameterwereused
forfullscaletestinginapurposebuilt,fourpointbendloadframe.Twospecimenweldgeometrieswereexamined:
(i)apipewithagirthweldatmidlength.
(ii)apipewithafilletweldtoacircularpatchatmidlength,simulatingananodeattachmentpad.Thegirthweldedpipes,
designatedGW1andGW2,wereweldedbymechanisedTIG(withthepipehorizontalandrotated)employingaZeron100X
superduplexfillerwire.Thefilletweldedpipe,designatedFW1,wasweldedbymanualTIG,employingsimilarwire.
Residualstressesweremeasuredpriortotestingusingthecentreholedrillingtechniqueintheweldtoe/HAZareas.Thelocations
werechosentominimiseimpactonsubsequenttest.Theweldedpipesweretestedwithacellmountedaroundtheweldarea,
containingnatural,flowingseawateratatemperatureof10Cwithapotentialof1100mVSCEappliedbypotentiostat.Thepipes
wereprecharged,withoutloadapplied,forsevendays,priortotest.TheloadappliedtopipeGW1andFW1wasthenincreased
incrementallytoidentifythethresholdloadformacroscopiccrackdevelopment.Eachloadingstepwasmaintainedforsevendays
andthepipewasexaminedfortheonsetofcrackingemployingabinocularmicroscope.Aftersevendaysexposuretheappliedload
wasincreasedandtheprocedurerepeateduntilcrackinginitiated.Inordertorecordthestrainduringtesting,thepipeswerestrain
gauged.Followingdeterminationofanapproximatethresholdloadfromthefirsttwotests,thegirthweldedpipeGW2wastested
withloadingtogive0.5%totalstrain,asmeasuredonstraingaugesawayfromtheweldandheldforaperiodof6months.The
weldedpipeswereexaminedbydyepenetrantinspection(DPI)attheendofthetesttoidentifyanyfinecrackingattheweldand
sectionsweretakenthroughrelevantareasidentifiedbyDPI.
Stressconcentrationfactors(SCFs)fortheweldtoeswereestimatedbaseduponcomparisonoftheweldgeometriesandprevious
finiteelementanalysisperformedatTWI.[7]ThetwoweldgeometriesexaminedhadthefollowingestimatedSCFs:
(i)girthweldedpipes.TheSCFwasabout2.60.2.Theerrorsquotedallowforthevariabilityofgeometryinrealweldscompared
withFEmodels.ThemainvariablesdeterminingtheSCFatthetoeofabuttweldaretheangleattheweldtoeandtheoverall
profileoftheweld.Forthegirthwelds,thetoeanglewasmeasuredtobeabout45degreesandtheweldoverfillwasconsideredto
beintheformofacirculararc.
(ii)filletweldedpipe.TheSCFwasestimatedas2.80.2.Themainvariablesherearethetoeangleandtheratiobetweentheweld
leglengthandplate(pipe)thickness.Thetoeanglewas2025andtheleglengthandplate(pipe)thicknesseswere8mmand
15mm,respectively.
Results
Effectsoftemperatureandpressure
HISCtestingat20and80CFigure1presentstheresultsoftensileHISCtestingof25%CrsuperduplexmaterialAat20and
80C,1barapressureand1100mVSCE.Intermsofappliedstress,thethresholdforspecimenfailurein30dayswaslowerat80C
thanat20Cbyaround3%butnoreductioninthestressforcrackinitiation(ieformationofsmallcracksthatdidnotpropagate
throughthickness)in30dayswasfound,Figure1a.TheequivalentdataareplottedintermsofstraininFigure1b,whichshowsa
similarsmallreductioninstrainforcrackingat80C.However,the0.2%proofstressofmaterialEwasaround13%lowerat80C
than20C(530and608MParespectively),indicatingthattheHISCcrackinitiationandpropagationthresholdsat20Cwerelower
thanat80C,whenconsideredintermsofnormalisedstress,ieappliedstressdividedby0.2%proofstress.
Fig.1.ResultsofsmallscaletensileHISCtests
at20and80C(allat1barapressure)Fig.1a)
Plottedintermsofappliedstress
Fig.1b)Plottedintermsofmeasuredstrain
HISCtestingat1and100baraFigure2presenttheresultsofsmallscaletensileHISCtestingof25%CrsuperduplexmaterialA
at1and100bara,20C,and1100mVSCEinseawater.Areductioninthestressthresholdforcrackpropagation,byabout4%,was
notedat100barabutnoreductioninthresholdstressforcrackinitiationwasnoted.However,itshouldbenotedthatfewertests
wereperformedat100bara,suchthatthethresholdvalueswerelesspreciselydeterminedthanat1bara.Hence,itispossiblethata
smallshiftincrackinitiationthresholdexistsbutitcouldbenomorethan5%.
Fig.2.ResultsofsmallscaleHISCtestsat1and
100bar,20CFig.2a)Plottedintermsof
normalisedappliedstress,ieappliedstress/0.2%
proofstress
Fig.2b)Plottedintermsofmeasuredstrain
PosttestcharacterisationSubstantiallyhigherhydrogenpickupwasobservedat80Ccomparedto20C,byafactorof5to8,
whilstmeasurementsindicatedanapproximatefactoroftwoincreaseinhydrogenpickupat100baracomparedto1bara(bothat
20C).Itwasnotedthatthereweremanymorecracksformedat80Cthanat20Cbutincreasingtemperatureto80Cseemedto
havereducedcrackdepth.Raisingpressureincreasedcrackdepthforagivenstrain.
FullscaleHISCtestsonweldedpipes
ResidualstressAmaximumtensileresidualstressof453MPawasmeasuredintheHAZofgirthweldedpipeGW1about2mmfrom
thefusionline.Maximumcircumferentialandaxialtensileresidualstressesof447MPaand309MPawereobservedintheHAZofpipe
GW1.Amaximumtensileresidualstressof392MPawasmeasuredintheHAZofpipeGW2andmaximumcircumferentialandaxial
tensileresidualstresseswere389MPaand279MPa.ForthefilletweldedFW1pipeamaximumtensileresidualstressof404MPawas
measuredintheHAZabout1.5mmfromthefusionline.
ThresholdstressandstrainforHISCFigure3showsthe'crack'and'nocrack'appliedstresslevelsfromgirthweldedpipeGW1,
ascalculatedfromthelongitudinalstrainmeasurementsawayfromtheweldandthestressstraincurveforthe22%Crparent
materialB,indicatingtheapproximatethresholdload.Theglobalstrain,ieawayfromstressconcentrations,duringtestingwas
constantandnostrainingduetolowtemperaturecreepwasobservedatthegauges.
Fig.3.'Crack'and'nocrack'stresslevelsinthe
fullscalegirthweldedpipeGW1
Nocrackingwasobservedaftersevendaysat0.63%strain,correspondingtoanominallongitudinalstressof584MPa.The
equivalentnormalisedstress(withrespecttothe0.2%proofstress)was1.05.Crackingoccurredafterlessthan12hoursatastrain
levelof0.85%(measuredfarfromtheweldtoe)correspondingtoanominalstressof601MPa(normalisedstress=1.08).Failurein
pipeGW1occurredattheweldtoeandpropagatedthroughtheHAZ(Figure4)intotheparentmaterial.
Fig.4.Crackattheweldtoeinthe
girthweldedpipeGW1
TheappliedloadforthepipeGW2,testedatoneloadforsixmonthswithoutcracking,isindicatedbythedashedline(measured
125mmfromweldcaptoegauge6).Themaximumtotalstrainmeasuredfarfromtheweldtoewasalittleover0.5%,
correspondingtoastressof571MPaandanormalisedstressof1.03.
Fig.5.Fracturesurfaceofthecrack
ingirthweldedpipeGW1
Discussion
Operatingconditions
Increasingtemperatureto80Candincreasingpressureto100barabothtendedtoreducethethresholdstressforspecimenfailure
in30daysbutlittleornoeffectonsmallcrackinitiationwasfound.Whenthereductionofproofstressat80Cistakeninto
account,crackingbehaviourisimprovedintermsofnormalisedstressbutitisconcludedthatsensitivitytoHISCisnotsubstantially
changedat80C.Themechanismoftheeffectofpressureonhydrogenpickupisnotimmediatelyapparent,althoughoneeffect
mightbetoinhibitrecombinationofHatoms.Itappearsthatconditionsthatacttoincreasehydrogenpickuptendtoenhance
specimenfailure,iecrackpropagation,buthavelittleeffectoncrackinitiation,atleastforthedurationsstudied.Thefactthat
increasedsurfacehydrogencontentdidnotsignificantlyaffectHISCinitiation,suggeststhatthesurfacehydrogenlevelisalready
adequatelyhighforeasycrackinitiationat20Cand1bara.However,agreaterbutstillquitesmall,effectofpressureoncrack
propagationwasnoted.Thisisconsistentwithanincreasedhydrogenlevelsubsurface,arisingfromahighersurfaceconcentration
(thehydrogendiffusioncoefficientbeingunchanged).
Theeffectofincreasingtemperatureisnotasimpleone.Thesurfacehydrogenconcentrationat80Cisexpectedtobemuchhigher
thanat20C,ashigherchargingcurrentdensitiesarefoundathighertemperatures,[8]andthismightaccountforthegreater
numberofsurfacecracksat80Ccomparedto20C.Hydrogendiffusionisalsofasterat80C,leadingtohighersubsurface
hydrogencontents.However,thisdidnotleadtogreatercrackdepthsat80C.Onepotentialexplanationofthiscouldbethat
propagationiscontrolled,atleastinpart,bytheaustenitestructureratherthantherateofhydrogenpenetrationaheadofthecrack
alone.
Fullscalepipebehaviour
Theworkshowedthatfullscaleweldedduplexstainlesssteelpipes,withfineaustenitespacing,cathodicallyprotectedat
1100mVSCEcantolerateaglobalstressof1.03timesthe0.2%proofstress,equivalenttoatotalstrainof0.5%,bothmeasured
wellawayfromanystressconcentratorpriortotheonsetofHISCataweldwithandSCFof2.62.8.Onlysmalldifferenceswere
observedbetweenfilletweldedandgirthweldedpipes.Thepipetestfailureloadsaresubstantiallygreaterthanforcoarsegrained
superduplexforgings,whichfailedunderanappliedstrainofonly0.25%atastressconcentrator.[2]Theinitiationandsubsequent
propagationofHISCapparentlyoccurredatverysimilarstressinthelargescalesamples.Thestressandstrainattheweldtoeare
notknown,asthereisnosimplerelationshipbetweentheSCFandthelocalconditionsattheweldtoe,whichwillbeaffectedby
locallowtemperaturecreep.Whentheworkwasoriginallyundertaken,therewasnoopportunityorstrongneedtoanalysethe
resultsfurtherbutthereisnowanopportunitytomodelthestressandstrainattheweldtoeandtocomparewiththeRPF112
allowableloading.
Remainingissues:materialswithimprovedhiscresistanceandsubsurfaceflaws
RPF112representsabigstepforwardindesigntoavoidHISC.However,itmaybemoreconservativethannecessaryforsome
productforms,e.gfinegrainedwroughtpipe,partlybecauseitisprimarilyvalidatedagainstdataforverycoarsegrained
superduplexforgingmaterialandpartlybecauseitmakesconservativeassumptionsregardingtheeffectofresidualstress.Thishas
ledtoreportsfromindustrythatsomecomponentsarebeingoverdesigned,leadingtoproductionofcomponentsthatarethicker,
heavierandmoreexpensivethannecessary.However,someproductforms,includingpipes,aresubstantiallymoreresistantto
HISCthancoarsegrainedforgings,atleastpartlyduetofineraustenitespacing.WhilstRPF112allowsadvantageofthisenhanced
HISCresistancetobetakenformaterialwithaustenitespacing<30m,thequalificationprocessisnotrobust,asitisbasedona
microstructuralmeasurementofaustenitespacing,whichisnotthesubjectofarecognisedstandard,andotherfactorssuchasthe
alignmentandaspectratiooftheaustenitearenotconsidered.Thereisevidencethattheaustenitespacingmeasurementprocedure
describedinRPF112isnotreproducibleandhenceopentomisinterpretation,eginrelationtothetreatmentof'fineequiaxed
austenite'clusters.DevelopmentofarobustqualificationtestformaterialswithimprovedHISCresistancewouldbebeneficial.
RPF112recommendsthatweldsinnewdesigns,areasofuncertaintyand'critical'itemsmustbefreeofflaws.Unfortunatelymany
itemsfallintothesecategoriesandthenecessaryinspectionistypicallydifficulttoundertakeandtheneedtoeliminateflawsmay
forceunnecessaryrepairs.Theconcernisthatsuchcomponentscouldpasshydrotestandearlyserviceloading,egiftheyare
locatedoutsideoftheinitiallyshallowhydrogencharged,stressedzone,particularlyforcomponentswhicharemainlysubjectto
hydrogeningressfromtheexternalsurface.Howeverthereisariskoffailureinthemediumtolongtermifsuchflawsbecomefully
orpartiallyencapsulatedinahydrogencharged,stressed,zonenear.Ashydrogendiffusesfurtherintoamaterialwithtime,therisk
offailurefromsuchembeddedflawswillincrease.
RPF112takesaconservativeapproachtodealingwithflaws,asthereisanabsenceofreliabledataonthefracturetoughness
applicabletoburiedflawsinduplexstainlesssteelashydrogendiffusestothemoveralongperiodoftime.
Slowstrainratefracturetoughnesstestsonduplex/superduplexweldsexposedtocharginginseawaterwithCPgiveCTODvaluesas
lowas0.05mm.However,thefactthatnoHISCfailuresfromburiedflawshavebeenreportedindicatesthatthesevaluesarenot
appropriateandthatfurtherstudiesarerequiredtoquantifythefracturetoughnessrelevanttosuchflaws.Contributingfactors
includethehydrogenconcentrationataburiedflaw,lowtemperaturecreepandcracktipacuity.
Conclusionsandrecommendations
1. SusceptibilityofsuperduplexstainlesssteeltoHISCisbroadlysimilarat20and80C,despitesubstantiallyhigherhydrogen
pickupat80C,whichisabout5to8timeshigherthanat20C.
2. Increasingpressureto100baragaveamodestdecreaseinthresholdstressforHISCcrackpropagationbuthadnoeffecton
crackinitiation.Hydrogenpickupwasabouttwiceashighat100baracomparedto1bara,at20C.
3. ThethresholdstressforHISCoflargescalewelded22%Crpipes,withfineaustenitespacing,was1.03timesthe0.2%proof
stress,measuredawayfromtheweld,equivalenttoastrainof0.5%.Thereisanopportunitytoreanalysethesedatato
comparewithRPF112allowableloads.Comparisonwithpreviousfullscaletestsoncoarsegrainedsuperduplexforgings
wouldassistfurtherunderstandingoftheeffectsofmicrostructure.
4. ThereisaneedforareliableandreproduciblemethodofqualifyingamaterialwithimprovedHISCresistance,associated
withaustenitespacing<30m.
5. Ifaflawissufficientlyclosetotheoutsidesurfacesuchthathydrogencandiffusetooneofitstipsovertime,then
theoreticallythismayleadtoHISC.However,nofailuresfromsuchfeatureshavebeenreported,suggestingadiscrepancy
betweentestingandtheservicesituation.Thelikelihoodoffailureinthefuturefromsuchflawsthereforeneedstobe
quantifiedandassessed.
References
1. TaylorTS,PendlingtonTandBirdR:'Foinavensuperduplexmaterialscrackinginvestigation'.Procconf'Offshore
TechnologyConference(OTC)1999',Houston,1999,paperOTC10965,467480.
2. WoollinPandMurphyW,'Hydrogenembrittlementstresscrackingofsuperduplexstainlesssteel',ProcconfCorrosion2001,
NACEInternational,paper01018.
3. WoollinPandGregoriA,'Avoidinghydrogenembrittlementstresscrackingofferriticausteniticstainlesssteelsunder
cathodicprotection',ProcconfOMAE2004,paper51203.
4. RecommendedpracticeDNVRPF112,'Designofduplexstainlesssteelsubseaequipmentexposedtocathodicprotection',
2008.
5. VargasPM,WastbergSandWoollinP,'Stressbaseddesignguidelinesforhydrogeninducedstresscracking(HISC)
avoidanceinduplexmaterials',ProcconfOMAE200979504.
6. EFCpublication17,2ndEdition,'Corrosionresistantalloysforoilandgasproduction:guidanceongeneralrequirementsand
testmethodsforH2Sservice',ManeyPublishing,2002.
7. GurneyTR:'Finiteelementanalysesofsomejointswiththeweldstransversetothedirectionofstress'.TWIResearchreport
E/62/75,March1975.
8. TurnbullA,GriffithsAandReidT,'Hydrogenembrittlementofduplexstainlesssteelssimulatingserviceexperience',Proc
conf'Corrosion99',NACEInternational,paper148.
TWILtd,GrantaPark,GreatAbington,CambridgeCB216AL,UK.Tel:+44(0)1223899000