High Strength Steel Fatigue

Effect of TIGdressing on fatigue strength and weld toe geometry of butt
weldedconnectionsinhighstrengthsteel
Student:
S.H.J.vanEs
sjorsvanes@hotmail.com
Studentnumber:1311255
March2012
Graduationcommittee:
Prof.ir.F.S.K.Bijlaard StructuralEngineering,DelftUniversityofTechnology
Dr.M.H.Kolstein StructuralEngineering,DelftUniversityofTechnology
Dr.ir.R.J.M.Pijpers StructuralEngineering,DelftUniversityofTechnology
Dr.ir.M.A.N.Hendriks StructuralMechanics,DelftUniversityofTechnology
Ir.L.J.M.Houben RoadandRailwayEngineering,DelftUniversityofTechnology
3
Preface
DuringmybachelorthesisandmyjobasastudentassistantIhaveexperiencedworkingintheStevinIIlaboratory.When
I needed to choose a subject for my master thesis, I quickly knew that I wanted experimental testing to be part of my
project.Afterall,thismaybethelastchancetotakepartinexperimentalresearchwithsuchdepthandfreedomofsubject.
My supervisor during my first two years as a student assistantwas Richard Pijpers, who introduced me to his research of
fatigue strength of welded connections in high strength steel. When there was an opportunity to perform a small
experimentalprogrammebasedonhislargerdatabase,thiswasaperfectchanceformetostartmygraduationthesis.
An experimental programme cannot be performed alone, so I would like to thank the laboratory staff: Arjen van Rhijn
andJohnHermsenwhohelpedmepreparatingallspecimens;KeesvanBeekwhoprogrammedthesoftwareforthelaser
measurementsandfatiguestrengthtestsandfinallyMichelevanAggelenandFredSchilperoortwhohavefixedmanysmall
problemsIencounteredinthelaboratory.
OfcourseIalsothankmyexamcommitteefortheguidanceduringmythesis:Prof.ir.FransBijlaard,Dr.HenkKolstein,
Dr. ir. Richard Pijpers and Dr. ir. Max Hendriks. The frequent meetings and the possibility to ask a quick question without
appointmenthavebeenofgreatvalue.
FinallyIwouldliketothankmygirlfriendGretaandmyparentsforthesupportduringthisproject.
SjorsvanEs
March2012
5
Listofsymbolsandabbreviations
Latinsymbols
a cracklengthparameter [mm]
a* materialconstantinnotchstressapproach [mm]
a
i
initialcracksize [mm]
a
f
finalcracksize [mm]
c crackwidthparameters [mm]
C
0
materialconstantincrackpropagationcalculation [Nmm
3/2
]
d
0,9
depthofV
0,9
[mm]
F appliedforce [kN]
f
lm;Ni
loadingmodefactorapplicabletocrackinitiationlife []
f
lm;Np
loadingmodefactorapplicabletocrackpropagationlife []
f
lm;Nf
loadingmodefactorapplicabletototalfatiguelife []
f
m
meanstressandresidualstressfactor []
f
mat
materialfactortodeterminefatiguelimitofparentmaterial []
f
Ni
N
i
/N
f
[]
f
Np
N
p
/N
f
[]
f
t;Ni
thicknessfactorapplicabletocrackinitiationlife []
f
t;Np
thicknessfactorapplicabletocrackpropagationlife []
f
t;Nf
thicknessfactorapplicabletototalfatiguelife []
f
uc
influencefactorforvariationofundercut []
f
wh
influencefactorforvariationofweldheight []
f
influencefactorforvariationofweldtoeangle []
FATvalue see
C
h weldheight [mm]
K rangeofstressintensityfactor [Nmm
3/2
]
K
th
thresholdvalueofKbelowwhichnocrackpropagationoccurs [Nmm
3/2
]
K
f
fatiguenotchfactor []
K
f;adj
increasedvalueoffatiguenotchfactorafteradjustmentofthe
weldtoeparametersinunfavourabledirection []
K
t
elasticstressconcentrationfactor []
K
t;adj
increasedvalueofelasticstressconcentrationfactorafteradjustmentof
weldtoeparametersinunfavourabledirection []
k
hs
stressconcentrationfactorathotspot []
m slopeofSNcurveormaterialconstantincrackpropagationcalculation []
N numberofcycles []
N
C
numberofcyclesatFATvalue []
N
D
numberofcyclesatconstantamplitudefatiguelimit []
N
i
numberofcyclestocrackinitiation []
N
knee
numberofcyclesatfatiguelimit []
N
L
numberofcyclesatcutofflimit []
N
p
numberofcyclesduringcrackpropagation []
N
f
totalnumberofcyclesuntilfailure []
N
up
numberofcyclesatwhichtheBasquinrelationintersectswiththeyieldline []
P
s
probabilityofsurvival []
R stressratio []
R
eh
specifiedminimumyieldstrength [N/mm
2
]
R
0,2
specifiedoffsetyieldstrengthat0,2%strainafterunloading [N/mm
2
]
R
m
ultimatetensilestrength [N/mm
2
]
s multiaxialitycoefficienttodeterminefictitiousnotchradius []
V
0,9
highlystressedvolume [mm
3
]
w widthofhighlystressedvolume [mm]
Y compliancefunctionincrackpropagationcalculation []
Greeksymbols
strain []
notchradius [mm]
6
f
fictitiousnotchradius [mm]
* materialconstanttodeterminefictitiousnotchradius [mm]
weldtoeangle;allanglesaregivenindegrees []
stressrange [N/mm
2
]
C
stressrangeatFATvalue [N/mm
2
]
D
stressrangeatconstantamplitudefatiguelimit [N/mm
2
]
L
stressrangeatcutofflimit [N/mm
2
]
mean
meanstressrange [N/mm
2
]
95%
stressrangewith95%survivalpropability [N/mm
2
]
0,2%
offsetyieldstressat0,2%strainafterunloading [N/mm
2
]
a
stressamplitude [N/mm
2
]
a;E
endurablestressamplitudeinplainmaterial [N/mm
2
]
a;E;0
endurablestressamplitudeinplainmaterialatalternatingload [N/mm
2
]
E;specimen
endurablestressrangeinweldedspecimens [N/mm
2
]
E
endurablestressrangeinplainmaterial [N/mm
2
]
f
stressbelowwhichinfinitelifeisachieved [N/mm
2
]
hs
stressathotspot [N/mm
2
]
kaE
endurablestressatthenotch [N/mm
2
]
m
meanstress [N/mm
2
]
nom
nominalstress [N/mm
2
]
notch
stressatnotchroot [N/mm
2
]
r
residualstress [N/mm
2
]
y
yieldstress [N/mm
2
]
safetyfactor []
Abbreviations
AW Indicationthatthevalueconcernsasweldedspecimens
BM Basematerial
FZ FlusionzoneorFluidzoneincaseofTIGdressing
HAZ Heataffectedzone
HSS Highstrengthsteel
SG Straingauge
TIG IndicationthatthevalueconcernsTIGdressedspecimens(also:TungstenInertGas)
UC Undercut
VHSS Veryhighstrengthsteel
WT Weldtoe
WM Weldmaterial
7
Content
PREFACE............................................................................................................................................................. 3
LISTOFSYMBOLSANDABBREVIATIONS...................................................................................................................... 5
CONTENT............................................................................................................................................................ 7
1INTRODUCTION,PROBLEMANALYSISANDOBJECTIVES............................................................................................. 11
1.1 INTRODUCTIONANDPROBLEMANALYSIS ...................................................................................................... 11
1.2 OBJECTIVES ............................................................................................................................................ 11
2INTRODUCTIONINHIGHSTRENGTHSTEEL,FATIGUEANDTIGDRESSING................................................................... 13
2.1 CHAPTEROUTLINE ................................................................................................................................... 13
2.2 INTRODUCTIONINHIGHSTRENGTHSTEEL ..................................................................................................... 13
2.2.1 Material .....................................................................................................................................................13
2.2.2 Possibilitiesandlimitations .......................................................................................................................15
2.3 INTRODUCTIONINFATIGUE........................................................................................................................ 16
2.3.1 Definition...................................................................................................................................................16
2.3.2 Parametersthatinfluencethefatiguelife.................................................................................................16
2.3.3 SNcurve....................................................................................................................................................19
2.3.4 Highstrengthsteelandfatigue .................................................................................................................20
2.4 INTRODUCTIONINTIGDRESSING ............................................................................................................... 21
2.4.1 Weldimprovementtechniques..................................................................................................................21
2.4.2 TIGdressingprocessandinfluenceonfatiguestrength............................................................................22
3LITERATUREREVIEW:THEORY ................................................................................................................................... 23
3.1 INTRODUCTIONANDCHAPTEROUTLINE........................................................................................................ 23
3.2 NOMINALSTRESSAPPROACH ..................................................................................................................... 23
3.2.1 Principles ...................................................................................................................................................23
3.2.2 Calculationprocedure................................................................................................................................23
3.2.3 Benefits,drawbacksandapplication.........................................................................................................24
3.3 STRUCTURALSTRESSAPPROACH ................................................................................................................. 25
3.3.1 Principles ...................................................................................................................................................25
3.4 CRACKPROPAGATIONAPPROACH................................................................................................................ 27
3.4.1 Principles ...................................................................................................................................................27
3.5 NOTCHSTRESSAPPROACH......................................................................................................................... 29
3.5.1 Principles ...................................................................................................................................................29
4LITERATUREREVIEW:PRACTICE................................................................................................................................. 37
4.1 CHAPTEROUTLINE ................................................................................................................................... 37
4.2 LITERATUREREGARDINGFATIGUEANDHIGHSTRENGTHSTEEL........................................................................... 37
4.2.1 Strengthaccordingtocurrentdesigncodesandrecommendations .........................................................37
4.2.2 Behaviourofplainmaterial .......................................................................................................................37
4.2.3 Behaviourofnonplainmaterial ................................................................................................................38
4.3 LITERATUREREGARDINGTIGDRESSINGANDHIGHSTRENGTHSTEEL .................................................................. 41
4.3.1 TIGdressingprocess ..................................................................................................................................41
4.3.2 Strengthimprovementaccordingtocurrentdesigncodesandrecommendations ...................................41
4.3.3 InfluencesofTIGdressingonmaterialandgeometry...............................................................................43
4.3.4 InfluencesofTIGdressingonfatiguestrength ..........................................................................................46
8
5TESTSETUP................................................................................................................................................................ 49
5.1 CHAPTEROUTLINE ...................................................................................................................................49
5.2 TESTINGPROGRAMME..............................................................................................................................49
5.2.1 Identificationoftestspecimens ................................................................................................................ 49
5.2.2 Preparationofspecimens ......................................................................................................................... 49
5.3 TESTSETUP.............................................................................................................................................50
5.3.1 Measurementofweldgeometry............................................................................................................... 50
5.3.2 Measurementoffatiguelife ..................................................................................................................... 51
5.3.3 Measurementofmaterialhardness.......................................................................................................... 52
6PROCESSINGANDRESULTSOFLASERMEASUREMENTS ............................................................................................ 53
6.1 CHAPTEROUTLINE ...................................................................................................................................53
6.2 TESTOUTPUTANDPROCESSINGLASERMEASUREMENTS...................................................................................53
6.2.1 Testoutput................................................................................................................................................ 53
6.2.2 Determiningtheweldradius..................................................................................................................... 55
6.2.3 Determiningtheweldtoeangle,weldheightandundercut..................................................................... 59
6.3 RESULTS ................................................................................................................................................59
6.3.1 Observedgeometries ................................................................................................................................ 59
6.3.2 Comparisonbetweenrolledandcaststeel ............................................................................................... 60
6.3.3 Comparisonbetweendifferentsteelgrades ............................................................................................. 67
6.3.4 Distributionofweldgeometryparameters............................................................................................... 67
6.3.5 EvaluationofinfluenceofTIGdressing..................................................................................................... 72
7PROCESSINGANDRESULTSOFFATIGUETESTSANDHARDNESSMEASUREMENTS..................................................... 73
7.1 CHAPTEROUTLINE ...................................................................................................................................73
7.2 TESTOUTPUTANDPROCESSING ..................................................................................................................73
7.2.1 Testoutput................................................................................................................................................ 73
7.2.2 DeterminingN
i
,nominalstressandstressratio........................................................................................ 74
7.3 RESULTSOFFATIGUETESTS........................................................................................................................74
7.4 RESULTSOFHARDNESSMEASUREMENTS.......................................................................................................76
7.5 RESULTSOFCRACKMONITORING ................................................................................................................76
8ANALYTICALDETERMINATIONOFFATIGUESTRENGTH.............................................................................................. 77
8.1 INTRODUCTIONANDCHAPTEROUTLINE........................................................................................................77
8.2 FACTORSDETERMININGFATIGUESTRENGTH..................................................................................................77
8.3 DETERMINATIONOFSTRESSCONCENTRATIONFACTORANDFATIGUENOTCHFACTOR ............................................77
8.3.1 FEManalysisofweldtoe........................................................................................................................... 78
8.3.2 DeterminationofK
t
................................................................................................................................... 84
f
................................................................................................................................... 88
8.4 DETERMINATIONOFMEANSTRESSFACTOR ...................................................................................................91
8.5 DETERMINATIONOFTHICKNESSFACTOR.......................................................................................................92
8.6 DETERMINATIONOFLOADINGMODEFACTOR ................................................................................................92
8.7 PREDICTIONOFFATIGUESTRENGTHCURVE ...................................................................................................92
9ANALYSISOFFATIGUETESTRESULTS......................................................................................................................... 93
9.1 CHAPTEROUTLINE ...................................................................................................................................93
9.2 ANALYSISOFRAWDATA............................................................................................................................93
9.2.1 Exclusionofdata....................................................................................................................................... 93
9.2.2 Adjustmentoftestdata ............................................................................................................................ 95
9.2.3 Statisticalanalysis..................................................................................................................................... 97
9.3 FITTEDSNCURVES .................................................................................................................................99
9.3.1 Rawdata................................................................................................................................................... 99
9.3.2 Adjusteddata.......................................................................................................................................... 100
9.3.3 Discussion ............................................................................................................................................... 102
9.4 COMPARISON .......................................................................................................................................102
9.4.1 Comparisonofdatawithdesigncodes ................................................................................................... 102
9.4.2 Comparisonofdatawithasweldedfatiguetests................................................................................... 106
9.4.3 Comparisonofdatawithanalyticaldeterminationoffatiguestrength.................................................. 115
9.5 EVALUATIONOFTIGDRESSINGINFLUENCE.................................................................................................124
9
10CONCLUSIONSANDRECOMMENDATIONS............................................................................................................. 125
10.1 CONCLUSIONS.................................................................................................................................. 125
10.1.1 InfluenceofTIGdressingoffatiguestrengthofbuttweldedspecimens.................................................125
10.1.2 InfluenceofTIGdressingonweldgeometryofbuttweldedspecimens..................................................125
10.1.3 TheoreticalinfluenceofchangedweldgeometryonbehaviourofTIGdressedspecimens.....................126
10.2 EVALUATIONANDRECOMMENDATIONS................................................................................................. 126
10.2.1 Assumptionsandapproximations ...........................................................................................................126
10.2.2 Recommendationsforfurtherresearch...................................................................................................126
ANNEXA:REFERENCES .............................................................................................................................................. 129
ANNEXB:TESTSPECIMENS........................................................................................................................................ 131
ANNEXC:FATIGUETESTDATA................................................................................................................................... 133
ANNEXD:MATERIALCERTIFICATES ........................................................................................................................... 135
ANNEXE:PRODUCTIONDATASHEETSTIGDRESSING................................................................................................ 167
ANNEXF:MATLABSCRIPTSTOPROCESSWELDGEOMETRYDATA............................................................................. 177
ANNEXG:COLLABORATIONWITHTNO:ACOUSTICEMISSION................................................................................... 183
ANNEXH:HARDNESSMEASUREMENTS..................................................................................................................... 185
ANNEXI:CRACKMONITORINGRESULTS.................................................................................................................... 205
ANNEXJ:DIANAMODEL............................................................................................................................................ 211
10
11
1 Introduction,problemanalysisandobjectives
1.1 Introductionandproblemanalysis
Inmostdesigncodesandincommonengineeringpractice,thefatiguestrengthofweldedhighstrengthsteelstructuresis
assumedtobecomparablewiththefatiguestrengthofmildsteelstructures.
It is shown in literature (Maddox, 1991; Gurney, 1979) that perfect smooth specimens of high strength steel perform
better in fatigue tests than their mild steel counterparts. However, as soon as discontinuities, notches, surface roughness
and all other imperfections that are unavoidable in practice are taken into account, the advantage of high strength steel
quicklydiminishes.
To recover some of the advantages of high strength steel, several options can be used. The combination of welds and
other geometrical stress raisers can be omitted by design (for example by using cast steel nodes in trusses). Another
possibility,whichcanalsobecombinedwiththeaforementionedsolution,isaweldimprovement.Weldimprovementsare
proceduresexecutedafterweldingtoimprovethefatiguebehaviouroftheweldareabyreducingtensileresidualstresses,
improvinggeometry,removingweldflawsandinclusionsoracombinationoftheseimprovements.Thisresearchwillfocus
onaweldimprovementmethodcalledTIGdressing,inwhichtheweldtoeisremeltedtoprovideasmootherweldprofile.
TheprocedureofTIGdressinghasprovenonfilletweldstobebeneficialforbothlowstrengthandhighstrengthsteelsin
earlierresearch,butthiseffectisnotalwaystakenintoaccountincurrentdesigncodesandrecommendations.
ThisresearchwillcompareTIGdressedspecimenswithsimilarasweldedspecimensfromthesamematerialbatchesand
willattempttoexplaindifferencesinfatiguestrength,basedmainlyongeometrychanges.
1.2 Objectives
Theobjectivesofthisstudyare:
DeterminethefatiguestrengthofTIGdressingonvarioushighstrengthsteelbuttweldedspecimensinrelationto
thefatiguestrengthofsimilar,asweldedspecimens
Describe the change of weld toe geometry due to the TIG dressing process ondifferent high strength steel butt
weldedspecimens
RelatetheallegedimprovedfatiguestrengthbyTIGdressingtothechangedgeometrybymeansofatheoretical
analysis
To accomplish these objectives, a method needs to be found to accurately measure and describe the weld profile in a
consistent way. Furthermore fatigue tests will be executed on material from the same production batches as used in an
earlier research by Pijpers (2011). From this research, fatigue data and adjustment factors for various geometrical and
loadingparameterswillbeused.Togaininsightintothecurrentstateofknowledgeandatheoreticalbackgroundtocouple
the weld geometry to fatigue strength, this study will start with a literature research into fatigue, high strength steel and
TIGdressing.
12
13
2 Introductioninhighstrengthsteel,fatigueandTIGdressing
2.1 Chapteroutline
This chapter gives a general introduction in high strength steel, its production processes, possibilities and limitations.
Then a general introduction in fatigue is given, where the influencing factors of the fatigue process are discussed. The
materialandtheprocessoffatiguearethencombined.FinallyashortexplanationoftheTIGdressingprocessisgivenand
itseffectsonthematerialandfatiguebehaviour.
2.2 Introductioninhighstrengthsteel
2.2.1 Material
High strength steels are steels with a higher yield and tensile strength than the most commonly used steels. In current
practice in Europe, the steel grades S235 and S355 are most commonly used for mild steel structures. These steel grades
have a yield strength of at least 235 and 355 N/mm
2
, respectively. High strength steels have a specified minimum yield
strength(R
eh
)higherthan355N/mm
2
.Commonhighstrengthsteels(HSS)rangefromS355toS690.Higherstrengthsteels
are referred to as very high strength steels (VHSS). Again the steel grade refers to a specified minimum strength, but
because higher strength steels do not always show a clear yield point, the specified yield strength is the stress at which
afterunloadingapermanentdeformationof0.2%remains(R
0.2
)(seefigure2.1).
figure2.1Stressstrainrelationshipforahighstrengthsteel;Aistheproportionalitylimit,Btheelasticlimit,ythe
yieldpoint.LineCisusedtodeterminethepointwhereapermanentdeformationof0.2%remains:thespecified
yieldpoint.(source:EngineeringArchives,2008)
Highstrengthsteelscanbemanufacturedindifferentways.Themostcommonhighstrengthsteelsarenormalizedsteel,
thermomechanically rolled steel and quenched and tempered steel. All these manufacturing processes are focussed on
grainsizereduction,whichhasabeneficialinfluenceonthestrength.Normalizedsteelandthermomechanicallyrolledsteel
are available in moderate high strengths (up to S460). Quenched and tempered steel is available in higher strengths (VM
publication125,2008).
Thedifferenttreatmentshavetheirinfluenceonthemicrostructureofthematerial.Themicrostructureofsteeldepends,
amongotherthings,onthecarboncontent,temperatureandcoolingrate.Infigure2.2theironcarbondiagramisshown.
The three important phases are liquid (L), ferrite () and austenite (). Other forms are cementite (Fe
3
C) and pearlite
(+Fe
3
C).Austeniteisformedabovethetransitiontemperature(formostcarboncontents723C)andwilltransferbackto
ferritewhencooleddown.Thepossibleexcessofcarbon,whichisalmostalwaysthecasewhencarbonispresent,causes
cementite to be formed. However, when the steel is cooled down fast, the austenite will change to ferrite oversaturated
with carbon. This structure, called martensite, has a hard, crystalline and brittle structure with limited ductility. For this
reason,steelconsistingofmartensiteshouldbecheckedforsufficientductility.Asmartensiteisnotanequilibriumphase,it
isnotvisibleintheironcarbondiagram.Anotherpossiblecrystallinestructureisbainite.Bainiteiscreatedwhenaustenite
14
is cooled quickly, but not so quickly that martensite forms. Bainite consists of ferrite with a lot of dislocations combined
withcementite.Thedislocationsmaketheferriteharderthanordinaryferrite.
figure2.2Theironcarbondiagram(source:KEYtoMETALS.com)
Normalizedsteelmainlygainsitsstrengthbyalloyedelements.Thetoughnessofthesteeliskeptwithinboundariesby
the normalizing treatment, but the treatment also raises the strength to a certain extent, especially the ultimate tensile
strength.Tonormalizethesteelitisheatedtoabout920andthencooledbyair.
Thermomechanically rolled steel has a low content of alloying elements and mainly gains its strength by grain size
reduction. This is achieved by controlling and limiting the temperature during the end stages of the rolling. The relatively
low temperature during the final deformation of the components requires rather strong rolling equipment. When the
materialisprocessedfurther,heatingbeyondthetransitiontemperatureisnotallowedtopreventlossofstrength.
Quenched and tempered steel is quickly cooled down to achieve high strength. This produces a very strong, very hard
and very brittle material. To reduce hardness and restore ductility the material is reheated to a temperature below the
transition temperature. The energy for reheating can also be supplied by the core of thicker materials when only the
perimetersarequenched.Thisiscalledquenchedselftemperedsteel.
Thethreeproductionmethodseachleaveacharacteristicmicrostructure.Normalizedsteelconsistoffinegrainedferrite
and pearlite. Thermomechanically rolled steel starts with long austenite grains caused by the rolling processes. When
cooled down a very fine grained ferrite structure is created. The microstructure of quenched and tempered steels mainly
showsbainiteandmartensiteinacrystallinestructure.Thedifferentmicrostructuresaredepictedinfigure2.3.
15
figure2.3Fromlefttoright;microstructuresofnormalizedS460,thermomechanicallyrolledS460andquenchedand
temperedS1100(source:VMpublication125,2008)
2.2.2 Possibilitiesandlimitations
Possibilities
High strength steel offers designers certain advantages, but it also has some drawbacks. The first advantage of high
strengthsteelisdirectlyrelatedtoitshighstrength:lessmaterialisneededtoresistacertainforce.Structuresmadeofhigh
strengthsteelcanthereforebemadelighterthantheirconventionalcounterparts.Thisisespeciallybeneficialinmovable
structures, for example a movable bridge or mobile crane. Also most offshore structures benefit from this, because the
transport to the building site on barges is an important part of the design. The second advantage is related to welded
connections. In general, high strength steel structures have smaller plate thicknesses. Because the volume of added weld
material increases quadratic with increasing thickness of the plates to be connected, a significant cost reduction could be
madehere,especiallyinwesterncountrieswherewagesarehigh.
figure2.4Schematicviewoftheeconomicadvantagesofhighstrengthsteel(source:VMpublication125,2008)
Limitations
Alargedisadvantageofhighstrengthsteelisthematerialcostandavailability.Conventionalmildsteelsaremorereadily
available and more easily produced and are therefore cheaper. High strength steels that are available, are mostly only
availableasplatematerialandnotasprofiles.Thisdisadvantagemaybeoutweighedbythereductionofweldingcostsand
transportation costs. Another aspect of high strength steel is the fact that the Youngs modulus does not increase with
increasingstrength.Forsimplebeamstructures,highstrengthsteelwillthereforemoreeasilyreachthedeformationlimits
set by the design code or dictated by secondary structures, such as internal walls and windows, than conventional steel
structures. Structures that are stiff by their nature, such as truss structures, can overcome this problem. For the same
reason,stabilityofhighstrengthsteelstructuresandcomponentsalwaysneedsattention.Forcolumns,highstrengthsteel
therefore is only beneficial with highly loaded columns with relatively short buckling lengths, for example in high rise
buildings. In beam structures, a high strength steel girder will probably not be a class 4 cross section. In tensile elements
stabilityproblemsobviouslycannotoccurandthereforetheycanbeveryslender,withtheexceptionoftheareainwhich
connectionsaremade.
Aboltedconnectioncancauseproblems,becausetheyieldstrength(or0.2%proofstress)ismuchclosertothetensile
strength then it is for ordinary steels. The reduction of this ratio with increasing strength is visible in figure 2.5. The
allowable cross section reduction by holes is therefore limited. In general, all high strength steel structures and details
should be checked for sufficient deformation capacity because of the lower ultimate strain of higher strength steels (see
figure2.5).Weldedconnectionsarepossibleinhighstrengthsteel,butcurrentweldingmaterialsarelimitedtoanultimate
strengthof900N/mm
2
.ForsteelgradeshigherthenS890thiswillhaveconsequencesfortheweldedconnections.
Finally, the fatigue behaviour of high strength steel structures is commonly regarded as the same as for standard steel
structures. While a plain, polished specimen does show increasing fatigue strength with increasing yield strength, the
additionofsurfaceroughness,imperfections,notchesandresidualsstresses,allcausedbyproductionorwelding,severely
16
reducethefatiguestrengthofrealstructuresandlimitsittoalevelcomparablewiththefatiguestrengthofmildsteels.This
willbeelaboratedoninparagraph2.3.4.
figure2.5Overviewofmaterialbehaviourofsteelwithincreasingstrength(source:VMpublication125,2008)
2.3 Introductioninfatigue
2.3.1 Definition
Fatigue can be defined as the mechanism whereby cracks grow in a structure (ESDEP). These cracks grow under
fluctuatingstresses,generatedbyfluctuatingloads.Failureofafatigueloadedstructureoccurswhenthecrackhasreduced
thecrosssectionbysuchanamount,thattheremainingcrosssectioncannotcarrytheappliedtensileloads.
Fatigue can occur after a relatively low amount of cycles to very large numbers of cycles. In general, fatigue can be
dividedintolowcyclefatigue,mediumcyclefatigueandhighcyclefatigue.Exactboundariesbetweenthesethreeregimes
are not apparent. Eurocode limits its use for applications with more than 10
4
cycles, which could be seen as a suitable
boundarybetweenlowcycleandmediumcyclefatigue.
2.3.2 Parametersthatinfluencethefatiguelife
Infatigue,anumberofparametersareimportant,primarilyregardingthestressesinthematerial(seefigure2.6):
N =thenumberofcycles
=thestressrange.Thestressrangeisdefinedasthemaximumstressminustheminimumstress
m
=themeanstress
a
=thestressamplitude(halfofthestressrange)
R =thestressratio:
min
max
R
o
o
=
17
figure2.6Descriptionofafluctuatingstress(source:ESDEP)
The mean stress influences the fatigue strength of the material. When the mean tensile stress increases, the fatigue
capacity decreases. This has been derived by different authors, all showing roughly the same behaviour, see figure 2.7.
When the static strength of steel increases, a higher mean stress in fatigue conditions can be endured. However, the
sensitivitytothemeanstressincreaseswithincreasingstaticstrength(Haibach,2006),resultinginasteeperlineinfigure
2.7.
figure2.7MeanstresseffectshowninaHaighdiagram(source:ESDEP)
Inprinciple,compressivestressespreventcracksfromopening,andthereforegrowing.Therefore,compressivestresses
in theory increase the fatigue life of components. However, in most structures residual tensile stresses are present as a
result of production and welding. This means that cracks in areas of the structure that are nominally under constant
compression can still show crack growth. Therefore the mean stress effect is not always present in actual structures. For
weldedstructurestheresidualstressescanbeintheorderoftheyieldlimit, whichseverelyreducesthefatiguestrength.
Thisiscloselyrelatedtothemeanstresseffect,becauseacrackcannotdistinguisharesidualstressfromameanstress,and
iscalledresidualstresseffect.
Besidesthestressalsothegeometryofthematerialhasalargeinfluenceonthefatiguestrength.Fatiguecracksstartat
small defects in the material. These defects, called notches, can occur at the surface of the material due to roughness,
inclusions or surface defects, at large discontinuities such as bolt holes or at small discontinuities, for example near the
weld. At thesenotches thestress is concentrated, thereby increasing the chance ofa fatigue crack occurring in that area.
Thisisknownasthenotcheffect.
figure2.8Examplesofdiscontinuitieswherecrackscanoccur;ontheleftalargediscontinuity:acopehole'onthe
right,onsmallerscale,asmalldiscontinuity:aweld(source:ESDEP)
Corrosion at these discontinuities can further decrease the fatigue strength. Also in plain specimens there is a clear
influenceofcorrosiononthefatiguestrength,seefigure2.9.
18
figure2.9Influenceofcorrosiveenvironmentsonthefatiguestrengthofmaterials(source:ESDEP)
Thesizeofthespecimenalsoinfluencesthefatiguestrength.Whenthesizeofaspecimenincreases,thefatiguestrength
drops. The total strength of the component may increase, but the allowable stress decreases. This is known as the size
effect,andiscausedby(ESDEP):
Astatisticaleffect.Whenthesizeofacomponentincreases,thechanceofaweaklink,inthe formofanotch,
smallinclusionorresidualstress,increases.Thereforealsothechanceofaninitiatingcrackincreases.
Atechnologicalsizeeffect.Theproductionprocessesandtheirassociatedsurfaceconditionsupondeliveryhave
aninfluenceonthefatiguestrengthofacomponent.
Ageometricalsizeeffect.Whenthethicknessofaplateincreasesthestressgradientatanotch(infigure2.10a
weld) decreases. When the inclusions or surface defects have the same size asthey have in a thinner plate, the
stressesatthetipofthedefectarehigherinthecaseofathickplate(seefigure2.10).
A stress increase effect. When the plate thickness increases, the notch size in general does not scale up to the
sameamount,ormaynotscaleupatall.
19
figure2.10Influenceoftheplatethicknessonthefatiguestrength(source:ESDEP)
Finally there is the effect of the material strength. When the material size increases the fatigue strength of the plain
materialalsoincreases.However,whenacrackoccurs,thecrackgrowthrateinallsteelsisroughlythesame.Thismeans
thatonceacrackisinitiated,allsteelshaveasimilarlifetimeuntilfailurewhenexposedtothesamestress.Thereforehigh
strength steels can only show a longer fatigue life if the material can longer resist crack initiation, this will be further
elaboratedonin3.4.Becauseinactualstructuresplainsteelalwaysneedsaconnection,therewillalwaysbenotches,stress
raisersand,inthecaseofwelding,inclusionspresent,whichseverelyreducethecrackinitiationlife.Therefore,incurrent
designcodes,highstrengthsteelstructuresmostlyareregardedtohaveafatiguestrengthcomparablewithstandardsteel
structures.Thissubjectisstudiedfurtherin2.3.4and4.2.
figure2.11Materialstrengtheffect.Plainmachinedspecimensshowaclearincreaseinfatiguestrengthwhenthe
materialstrengthincreases.Atthesametimeitisclearthatthisincreaseisnotentirelyvisiblefornotchedspecimens
(source:ESDEP)
2.3.3 SNcurve
TheSNcurveistherelationbetweenastressrange()orstressamplitude(
a
)andtheaccompanyingnumberofcycles
tofailure.SNcurvecanbedefinedforplainmaterial,simpledetailssuchasaweldedplateorentireconnections.
TheSNcurvesgenerallyfollowtheformofBasquinsrelation:
20
b
a
N constant o = (1.1)
Mostlyboththehorizontalaxisandtheverticalaxisareshowninlogscale.Nowadays,mostSNcurvesaredescribedby
theformula:
log log log N a m o = A (1.2)
The parameters a and m (equivalent to b in equation (1.1)) are determined based on of tests or calculations. The
parameters can for example depend on material, detail, postweld treatments and weld quality guaranteed by certain
inspection methods. As mentioned before, the material strength is regarded as unimportant in most design codes. A
commonexampleofanSNcurveisshowninfigure2.12.ForloadingwithaconstantamplitudetheslopeoftheSNcurve
(m) is 3 and below a certain stress level no damage occurs, this is shown in figure 2.12 as the constant amplitude fatigue
limit.Iftheamplitudeoftheloadingvaries,therecanbedamagebelowthislimit.TheslopeoftheSNcurvechangesand
again,atacertainstresslevelnodamageoccurs,eveninthecaseofvariableamplitudeloading.Thislimitisknownasthe
cutoff limit. These characteristic points in the SN curve are denoted with the symbols N
D
for the constant amplitude
fatiguelimitandN
L
forthecutofflimitwiththeiraccompanyingstresses
D
and
L
.Thethirdcharacteristicpointinthe
curve, denoted with N
C
and
C
is the point that marks the detail category. Detail categories will be further explained
chapter3.2.
figure2.12AnSNcurve(sourceESDEP)
2.3.4 Highstrengthsteelandfatigue
Higher strength steels generally also have higher fatigue resistance. However, according to Pijpers (2011) this mainly
affects the crack initiation period (N
i
). After a crack has initiated, the crack growth rate is the same as for ordinary steels
(see 3.4). Local notches in welded joints (see paragraph 2.3.2) effectively reduce the crack initiation time to a number of
cycles which is negligible (ESDEP). Therefore, in conventional welded structures, where initial imperfections are always
present, the material strength is of little influence on the fatigue strength. Thiseffect isshown in figure 2.11. At very low
(nonpractical)strengththereisaclearinfluenceofthematerialstrengthonnotchedspecimens,butatthestrengthlevelof
standard strength steels (R
m
>400 N/mm
2
) and high strength steels there is almost no influence of the material strength
anymore.
Becauseoftheeffectshowninfigure2.11,mostdesigncodesdonotdistinguishstandardsteelfromhighstrengthsteel
whenregardingfatigueandifanimprovementismade,thecodeonlyshowsimprovementsinthelowerstrengthsteelsup
toacertainsteelgrade.Abovethissteelgrade(e.g.S355inNPRCEN/TS1300131)thefatigueperformanceofallmaterials
isregardedtobethesame.
Structures that can benefit greatly from the advantages of high strength steel (less material and therefore light; less
welding)mayalsobeloadedinfatigue.Examplesofthesestructuresarecranes,offshoreplatformsandmovablebridges.
Because the cross sections of high strength steel are reduced in comparison with their standard steel counterparts, the
stressrange()ismuchlargerinthehighstrengthsteelstructure.Ifthefatiguestrengthofahighstrengthsteelstructure
isindeednotmuchdifferentthanfromastandardsteelstructure,fatigueispotentiallyleadinginthedesignofdynamically
loadedstructures.
Tomakefulluseofthehighstrengthsteelsseveraloptionsareavailable.Thedetailsofthestructurecanbeadjustedto
provide a smoother stress flow, thereby reducing stress concentrations. This geometrical improvement can be done on a
small scale, for example by using tapered plates instead of buttwelding two plates together with different thicknesses. A
geometricalimprovementcanalsobeusedonalargerscale:adjustthedesignofthestructureforfatigue.Agoodexample
ofthisapproachistheuseofcaststeelnodesintrusses,bywhichthefatiguesensitiveweldsareremovedfromthehighly
stressed connection area (see figure 2.13). Another way to reduce the effective stresses is reducing the tensile residual
stresses in a welded structure. This is not always possible, and the effect depends on the stress ratio and mean stress. A
similareffectisreachedifthemeanstresscausedbyloadingislesstensile,forexamplebyloweringtheselfweightofthe
21
structure. Finally, the specimens can be treated in such a way after fabrication that surface defects and microcracks are
removed.Thiswillincreasethecrackinitiationtime,andtherebythetotalfatiguelife.
figure2.13Atrusswithcaststeelnodes(source:Pijpersetal.,2010)
2.4 IntroductioninTIGdressing
2.4.1 Weldimprovementtechniques
In figure 2.11 it is clear that for plain specimens the fatigue strength increases with increasing material strength.
However,ifnotchesareintroducedastrengthplateaucanbeobserved,whichlimitsthefatiguestrengthforsteelswithan
ultimate strength higher than approximately 400 N/mm
2
. These notches can be introduced by holes or changes in cross
sections,butare,incivilengineeringstructures,mostlycausedbywelding.Weldshaveaveryroughsurface,causedbythe
natureoftheprocess.Anotherdrawbackoftheprocessisthepossibilityofsmalldefectsintheweld.Theseimperfections
mighthavenoorasmallinfluenceonstaticstrength,butinfatigueloadingtheymayformtheoneweaklinkthatisneeded
to initiate a crack. Also, the weld itself causes a cross section change due to local thickening or the steep angle of a fillet
weld.Inaddition,theseweldsareingeneralpositionedatlocationswhichsufferfromstressconcentrationsduetoglobal
geometryofthestructure.Thenotcheffectisthereforeveryprominentinweldedstructures.
To reduce the strength reducing effects of theweld a numberof weld improvement techniques can be used. The most
importanttechniquesare(Haagensenetal.,2001):
Burrgrinding
Hammerpeening
Needlepeening
TIGdressing
Burrgrinding
Theaimofburrgrindingistoremovepossibleweldflawsattheweldtoewherefatiguecrackscaninitiate,byremoving
material with a high speed grinder. The stress concentration at the weld toe, caused by the sharp geometrical transition
fromparentmaterialtoweldmaterialisreducedbysmoothingtheweldprofile.
Hammerpeening
The aim of hammer peening is to introduce compressive stresses in the weld toe region by repeatedly hammering this
areawithapneumaticbluntnosedchisel.Theeffectofthehammerpeeningprocessreliesonthemeanstresseffect(see
2.3.2).Anotherbeneficialeffectmaybethesmoothingoftheweldtoeprofile.
Needlepeening
Theaimofneedlepeeningisalsotheintroductionofcompressivestressesintheweldtoeregion.Inthiscasethesingle
chisel is replaced by multiple, smaller chisels. This makes the process more suitable for larges areas to be treated. The
effectsofhammerpeeningandneedlepeening,andtheiraimsinfatiguestrengthimprovementarecomparable.
TIGdressing
TheaimofTIGdressingistoremovepossibleweldflawsbyremeltingthematerialattheweldtoe.Theremeltingshould
alsohaveabeneficialeffectonstressconcentrationsbecausetheweldgeometryismadesmoother.Althoughtheprocess
is carried out with welding equipment, no extra material is added. The melting of steel of course causes changes in the
stressstateofthematerial.Ifhighresidualstressesexistatthesurface,theywillbereducedtoacertainextent(see4.3.3).
Alsotheheataffectedzonewillbeenlarged.TheeffectofTIGdressingwillbediscussedfurtherin2.4.2.
22
More information on all these processes can be found in the IIW recommendations on weld improvement techniques
(Haagensenetal.,2001).BecauseinthisstudythefocusliesonTIGdressing,theinfluenceofthisprocessonthematerial
andgeometryisexplainedmorethoroughlyinthenextparagraph.
figure2.14Effectofdifferentweldimprovementtechniques:aswelded,burrgrinding,ultrasonicimpacttreatment
(comparabletotheeffectsofhammerandneedlepeening)andTIGdressed(sourcePedersenetal.,2010)
2.4.2 TIGdressingprocessandinfluenceonfatiguestrength
TIGdressinginvolvesremeltingofthematerial,whichmeansthattheoriginalgeometryoftheweldtoeisalteredintoa
newgeometry.Italsomeansthatanydefectsintheweldtoemayberemoved.Topreventthenewweldtoefromhaving
thesameimperfectionsandsharpgeometry,someprecautionsmustbetaken.
TheweldthatistobeTIGdressedneedstobepreparedbyremovinganymillscale,rust,oil,paintoranyotherpossible
weldcontaminant.Thiscanbedonebywirebrushing,butlightgrindingmightalsobenecessary.Ifthecleaningprocessis
notsufficient,gasinclusionsintheweldcanbetheresult,whichseverelylowerthefatigueperformanceoftheweld.
Toguaranteethenewgeometrytobebetterthantheoriginal,anumberofconditionsmustbemet.Theheatinputmust
notbetoohightopreventundercuts.TooptimizetheoverallshapeofthenewweldtoetheTIGtorchmustbepositioned
carefully.InIIWrecommendations,thetorchdistancetotheweldtoe,angleofthetorchintwodirectionsandtravelspeed
incombinationwiththeweldingcurrentarespecified(Haagensenetal.,2001).
TIGdressingimprovestheweldtoeinprinciplebyimprovingthegeometryandremovingweldtoeflaws.Thesmoothing
ofthegeometryreducesthestresspeakneartheweldtoe.Thislowerstresspeakalsohasfewerflawsatwhichtocausea
fatigue crack. A secondary benefit of TIG dressing may be the release of high tensile residual stresses caused by welding.
These influences all mostly influence crack initiation time. Because crack initiation time is very important when high
strengthsteelisconsidered(see2.3.4)thebeneficialeffectofTIGdressingmaybeexpectedtobelargerforhighstrength
steelweldedconnections.Thiseffectclearlyshowswhensomepreviousresearchisstudiedwhereimprovementsoffatigue
strengthvaryingfrom18%to85%arefound,evenwithinthesameresearchprogramme.Thiswillbeelaboratedonin4.3.4.
23
3 Literaturereview:Theory
3.1 Introductionandchapteroutline
Inallfatiguetestsawidescatterrangeisfoundbecauseoftheweakestlinkprocess.Acrackinitiatesatalocationwhere
globalgeometry,localgeometry,surfacedefects,materialdefectsandstressallcombinetoaworstcasescenario.Allthese
influencescannotexactlybemodeled,becauseoftherandomnatureofwelding.
A number of different calculation models have been developed to calculate the fatigue strength of a component. The
most common theories, nominal stress approach, structural stress approach and crack propagation approach, will be
coveredfirst.Subsequently,moreindepthanalyseswillbetreated,whichisspecificallyusedinthisresearch.
3.2 Nominalstressapproach
3.2.1 Principles
Thenominalstressapproachclassifiesawiderangeofwidelyuseddetailsandspecifiestheirfatiguestrengthatacertain
numberofcycles.InEurocode3thefatiguestrengthofdetailsisdeterminedat210
6
cycles.AllSNcurvesareparallelto
eachother,thereforeonlythefatiguestrengthat210
6
cyclesisneededtodeterminetheSNcurve(seefigure3.1),which
defines the allowable stress range at any number of cycles. The lines given in the code are design lines, which result in a
sufficientlysafestructure.
figure3.1AnumberofSNcurvesbelongingtodifferentdetailcategories(source:ESDEP)
3.2.2 Calculationprocedure
To design a structure, all stresses need to be determined and all details, welds and other discontinuities have to be
classifiedinacertaindetailcategory.Detailcategoriescandependforexampleonlocalgeometry,weldtype,weldquality,
post weld treatments and define the maximum allowable stress at 210
6
cycles (see figure 3.2). Then for each detail the
allowable number of cycles at the calculated stress level can be determined, by means of the standardized SN curves as
showninfigure3.1.Possiblemisalignmentsandthicknesseffectshavetobetakenintoaccountseparately.Thisallowable
numberofcyclescanthenbecomparedwiththeneedednumberofcyclesinthestructureslifetime.Afewexamplesofthe
24
detailcategoriesareshowninfigure3.2.Theexplainedprocedureisvalidfornormalstresses.Shearstressescanbetaken
intoaccountwithasimilarprocedure.
figure3.2Detailcategoriesfortransversebuttwelds(source:NENEN199319)
3.2.3 Benefits,drawbacksandapplication
The nominal stress approach can easily be applied on a wide range of designs because most common details are
incorporatedinthecodes.Calculationsarerelativelyeasyandquicktoperform.Whenadetailisnotclassified,thenominal
stressapproachcannotbeused.
In the widely used Eurocode 319 (2006) no distinction is made between different materials for the nominal stress
design,andthecalculationmethodislimiteduptoS700sincetheissueofEurocode3112(2007).Thismeansthatthefew
high strength steels that can be designed according to Eurocode (e.g. S460 and S690) are assumed to show no better
fatigue behaviour than standard steels. Also the use of post fabrication weld improvement techniques, other than stress
relief,isnotcoveredbyEurocode319.Foramorecompleteoverviewofthecurrentdesigncodesregardingfatigue,see
4.2.1.Therewillbeexplainedthatsomecodesdorewardpostweldtreatmentsorhighstrengthsteelswithhigherfatigue
strengthtoacertainextent.However,thenominalstressapproachremainsmostusefulforstandardapplications.
25
3.3 Structuralstressapproach
3.3.1 Principles
Inthiscasethemaximumstressatasocalledhotspotisdetermined,wherethestressreachesapeakatanotch.This
structural stress at the hot spot (
hs
) includes all stress raising effects at the detail, except the stress raising effect of the
weldgeometry.Thiseffectisleftoutoftheanalysisbecausetheexactweldgeometrydiffersgreatlyfromweldtoweld,and
isthereforeincorporatedinthescatterofthefatiguestrengthcurve.Themethodisthusverysimilartothenominalstress
approach,butapplicabletoallkindsofdetails,andnotjustthedetailslistedinthedesigncode.
figure3.3Someexamplesofstressdistributionatstructuraldetails(source:Hobbacher,2007)
ThestructuralstresscanbedeterminedbymeansofFEManalysisorbydirectmeasurementonthecomponent.Froma
certain number of measuring points the structural stress is then determined by extrapolation both in the case of FEM as
with direct measurements. An alternative method is a parametric calculation where the structural stress is previously
determined for a certain detail. A stress concentration factor (k
hs
) can then be determined directly from a parametric
formula.Thestructuralstresscanbecalculated:
hs hs nom
k o o = (3.1)
Once the structural hot spot stress is known the allowable number of cycles can be determined with the design SN
curves. Again these curves are dependent on the kind of detail but, as other stress raisers are already considered, only
applytosimpleweldingdetailssuchasdepictedinfigure3.4.
26
figure3.4Detailcategoriesforthestructuralhotspotstressmethod(source:NENEN199319)
Whenachosendetaildoesnotexactlycomplywiththedetailsgiveninthedetailcategoriesorwherenoclearlydefined
nominal stress exists, the nominal stress approach cannot be applied. Then the structural hot spot stress can be an
adequate tool to analyze the fatigue strength of a component. A large drawback of this method is the extensive FEM
research or actual tests that need to be executed when no parametric formulae are available to determine the hot spot
stress.
Thestructuralstressapproachcanonlybeappliedtodetailtypeswherethecrackgrowsfromtheweldtoe,becausethe
stress needs to be determined along a certain number of extrapolation points at the surface of a plate. In figure 3.5 this
rulesoutdetailsftoj.Therefore,tousethismethod,thedesignermustbesurethecrackwillinitiateattheweldtoe.
27
figure3.5Variouslocationswherecracksmayoccurinweldedjoints(source:Hobbacher,2007)
3.4 Crackpropagationapproach
3.4.1 Principles
Thecrackpropagationapproachtreatsthepartofthefatiguelifeofacomponentfromtheinitiationofacracktofailure.
Both boundaries need to be determined in advance. For the initiation of a crack a commonly used crack length is 0.15
millimeter.Thepointatwhichacomponentisconsideredfailedcanbedeterminedasthepointwherethecrackistrough
thickness, an actual failure or when the crack growth rate reaches a certain value after which relatively few remaining
cyclesareexpected.
OnceacrackhasinitiatedthecrackgrowthratecanbecalculatedwiththeParislaw:
0
m
da
C K
dN
= A (3.2)
Inwhich:
a cracklengthparameter[mm]
N numberofcycles[]
K rangeofstressintensityfactor[Nmm
3/2
]
C
0
,m materialconstants[Nmm
3/2
],[]
This relation holds if the crack is not too small (no crack propagation; K<K
th
) or not too large. The Paris law and its
limits for too small or too large cracks is shown in figure 3.6. From this figure it can clearly be seen that the number of
cyclesinregion1(ifK>K
th
)andregion3isrelativelysmall.Therefore,thetotalnumberofcyclesinthecrackpropagation
stage(N
p
)canbeapproximatedwithonlythecracksinregion2accordingtotheParisLaw.
28
figure3.6Crackgrowthratevs.stressintensityfactor(Parislaw)(source:ESDEP)
ThematerialconstantsC
0
andm
arematerialdependent,evenbetweendifferenttypesofsteel,butdonotdiffermuch
overtherangeofsteels,seefigure3.7.Thismeansthatonceacrackisinitiated,itwillpropagateatapproximatelythesame
ratewhetherhighstrengthorlowstrengthsteelisused.
figure3.7Crackgrowthratefordifferentsteels(source:ESDEP)
29
Thestressintensityfactorisusedtodescribethestressfieldaroundthecracktip,anddependsonthegeometryofthe
crack and the surrounding specimen. The stress intensity factor can be determined by means of FEM analysis, but for a
widerangeofjointsthestressintensityfactorscanbedirectlycalculatedwithparametricformulae.Thethresholdvalueof
K,belowwhichnocrackpropagationoccurs,dependsonthemeanstressandenvironmentalconditions.
Tostartthecalculationadescriptionofthestressfieldaroundthecracktipisneeded.Thisisdonebymeansofthestress
intensityfactorK,whichgenerallyhasthefollowingform:
o t = A K Y a (3.3)
InwhichY,calledthecompliancefunction,takesthecrackshapeandoverallgeometryofthesurroundingmaterialinto
account.Ifonlypropagatingcracksareconsidered(K>K
th
)theParislaw(equation(3.2))canberewrittento:
( )
1 1
f f
i i
a a
m m
m m
a a
da da
N
C K C
Y a o t
= =
A
A
} }
(3.4)
Because the crack growth rate changes when the crack grows this integral cannot be solved directly, but muste be
approximatedinsmallsteps.Ineachofthesestepsthestressintensityfactorisassumedconstant.
1 1
2 2 2
1 1 1
m m m
m m
i f
N
C Y a a o o t

| |
|
A =
|
|
A
\ .
(3.5)
Inwhich:
a
i
theinitialcracksize[mm]
a
f
thefinalcracksize[mm]
If a limit is set to the crack size at which the specimen is considered failed, or the critical crack length is reached after
whichunstablegrowthoccurs(region3infigure3.6),thetotalamountofcyclescanbeobtainedbysummation.Tomake
the calculation, the material dependent parameters C and m need to be obtained. This can either be done directly from
testsorfromliterature.Alsothestressintensityfactorcanbeobtainedfromliterature,butaFEManalysiscanalsobeused.
However,thisFEManalysismustbemadefordifferentcrackdepthsofthesamecracktodeterminethechangeofKwhen
thecrackdimensionsincrease.
The crack propagation method can only calculate the number of cycles after crack initiation. Therefore, in design, it is
only useful when the crack propagation phase is dominant. In normal welded connections, this is generally the case
(ESDEP),inhighstrengthsteelweldedconnectionsthisapproximationcanbetooconservative.
If the compliance function is available from textbooks the analysis can be made rather quickly, especially when
specialized software is used which already incorporates the smallstepmethod depicted in equation (3.5). If the
compliance function is not available, in most cases it must derived from FEM analysis. Because the compliance function
amongotherthingsdependsonthecrackdepth,thisanalysiscanbeverytimeconsuming.
Thecrackpropagationapproachcanbeusedtodeterminethelivesofalreadydamagedstructures.Itcanalsobeusedto
determineserviceintervalsofstructures.Forthisthetimebetweenavisiblecrackandfailureiscalculated.
3.5 Notchstressapproach
3.5.1 Principles
As explained in 2.3.2, there is a clear notch effect in steel subjected to dynamic loading. At notches the stress is
concentratedwhichfacilitatestheinitiationof cracks.Theprincipleofthenotchstressapproachistocomparethisnotch
stress to the maximum stress a plain specimen can withstand. Therefore, at the root of a notch a small plain specimen is
imaginedwhichissubjectedtothesamestressesasthetipofthenotch.Inasimpleassessmentofthesenotchesonlythe
infinite fatigue life is considered and therefore only the fatigue limit is determined. If the plain machined specimens can
endureacertainstresslevelwithoutcracking,thenthisstresslevelcanalsobeenduredatthenotchroot.Inthismethod
oneofthemainassumptionsiselasticity.Thisisnotsurprising,because(large)plasticdeformationswilleventuallyleadto
30
cracksandfailure,whichcontradictswiththeinfinitelifethatwasassumed.Themethodcoversonlytheprocessofcrack
initiation.
figure3.8Atanotchthelocalstressesaredeterminedandappliedtoalocal'plainspecimen'
This does not take into account the fact that cracks may be initiated but do not propagate (dormant cracks) and that
minorplasticdeformationmaytakeplacewithouteffectonthefatiguelife(Radajetal.,2006).
Thestressconcentrationatthenotchiscalculatedwiththeelasticstressconcentrationfactor:
notch t nom
K o o = (3.6)
Experiments have shown that this elastic notch stress does not determine the fatigue behaviour of the notched
specimen.Instead,asomewhatlowerstresscanbelinkedtothefatiguebehaviourofthespecimen.Thiseffectiscalledthe
microstructuralsupporteffect(Radajetal.,2006).Thestressthatgovernsthefatiguebehaviourisastressaveragedovera
small length or volume, characteristic for the considered material. This microstructural support does not only occur near
very sharp notches (as shown in figure 3.8) but also at milder notches, provided that they are sufficiently small. The
microstructural support effect therefore depends not only on the material but also the geometry of the specimen,
specificallytheradiusoftheconsiderednotch(seefigure3.9)
figure3.9Onehalfofabuttweldandthecorrespondingnotchradius()atthenotchroot
Thismicrostructuralsupporteffecthasbeenrepresentedindifferentforms:
Criticaldistanceapproach(Peterson,1974)
Stressaveragingapproach(Neuber1937,1946and1968)
Stressgradientapproach(Siebeletal.,1993)
Highlystressedvolumeapproach(Kuguel,1961)
Thestressatthenotchthatactuallydefinesthefatiguebehaviourofthecomponentisexpressedas:
notch
f nom
K o o = (3.7)
WherethefatiguenotchfactorK
f
isdeterminedwithoneofthetheoriesdepictedabove.ThedifferencebetweenK
f
and
K
t
givessomeinformationaboutthesensitivityofthematerialtonotches.Thisisexpressedinthenotchsensitivity:
31
1
1
f
t
K
q
K
(3.8)
Anotchfactorof1representsamaterialthatisfullysensitivetonotches,becausethefatiguenotchfactorK
f
isequalto
theelasticstressconcentrationfactor.Ifq=0thematerialisinsensitivetonotchesbecauseK
f
=K
t
=1.
Different calculation approaches have been developed, each based on one of the earlier mentioned representations of
themicrostructuralsupporteffects.
Criticaldistanceapproach
Thismethod,developedbyLawrencefromtheoriginalconceptofPeterson,usesnotchstressanalysistodeterminethe
fatigue notch factor (Radaj et al., 2006). From there the method continues with the notch strain approach, which is not
coveredbythisresearch.OnlythedeterminationofK
f
willbediscussedhere.
The first step in the analysis is to determine where the crack will arise. For these locations the elastic notch stress
concentration factor needs to be determined. This can be done by FEM analysis or by using engineering formulae, if
available for the considered joint. The fatigue notch factor is derived from K
t
by using the critical distance approach
developed by Peterson. This approach states that the ratio between K
t
and K
f
depends on the ratio between a material
constanta*andthenotchradius.ThematerialconstantisapproximatedbyLawrence,andhasbeenappliedonlowand
highstrengthsteels:
1.8
2068
* 0.025
M
a
R
| |
=
|
\ .
(3.9)
Petersonalsofoundarelationbetweentheultimatetensilestrengthanda*.Thetworelationsaredepictedintable3.1.
Peterson also used some values based on hardness: a*=0.254 millimeter for softannealed steel (170HB) and a*=0.0635
millimeterforquenchedandtemperedsteel(360HB)(Radajetal.,2006).Thevaluesintable3.1differgreatlyinthelow
strengthrange,butareverysimilarinthehighstrengthrange.
R
m
[N/mm
2
] 345 518 690 863 1035 1380 1725
Lawrence 0.628 0.302 0.180 0.121 0.087 0.052 0.035
a*[mm]
Peterson 0.380 0.250 0.180 0.130 0.089 0.051 0.033
table3.1Materialconstanta*fromLawrenceandPeterson(Petersonderivedthedatafrombarsloadinginbending)
(Radajetal.,2006)
TherelationbetweenK
f
andK
t
isstatedas:
1
1
*
1
t
f
K
K
a
= +
+
(3.10)
Theshapeofthisfunctionisshowninfigure3.10.Inthisfigureitcanbeseenthatatacertaina*/ratioofabout1the
fatigue notch factor reaches its maximum. Because the notch radius scatters greatly at welds this worst case approach is
applicablefordesignanalysis.Ifthenotchgeometryanditsscatterisaccuratelyknown,arealisticapproachcanbemade.
32
figure3.10Elasticstressconcentrationfactor(K
t
)forabuttjointandfatiguenotchfactor(K
f
)fordifferentmaterials
(source:Radajetal.,2006)
Because the method proceeds in the notch strain domain, no clear adjustments for mean and residual stresses are
describedforadirectnotchstressanalysis.
Fictitiousnotchroundingapproach
ThismethodisdevelopedbyRadajandisbasedontheNeubermicrostructuralsupporthypothesis(Radajetal.,2006).It
hasmainlybeenusedforlowstrengthsteels,butisnotrestrictedtothesesteels.Totakeaccountofthereductionofthe
elastic stress to an effective stress the notch is imagined less sharp. This fictitious rounded notch leads to a lower elastic
stressconcentrationwhichisconsideredtoapproximatethefatiguenotchfactoroftheoriginalgeometry.
Thefictitiousnotchisgivenby:
*
f
s = + (3.11)
Inwhich:
f
thefictitiousnotchradius[mm]
theoriginalnotchradius[mm]
s amultiaxialitycoefficient[]
* amaterialconstant[mm]
It has been shown that if in the critical distance approach a*=0.25 millimeter is chosen, this corresponds to a
f
of 1
millimeter,becauseitcananalyticallybederivedthat
f
=4a*(Radajetal.,2006).Thematerialconstant*isconsideredto
depend on the yield limit of the material, see figure 3.11. This figure only covers low strength steels, except for ferritic
steels.
33
figure3.11Materialconstant*fordifferentsteeltypesandstrengths(source:Radajetal.,2006)
Radajadvisestheuseofa*of0.4millimeterincombinationwiths=2.5when(lowstrength)steelsareconsidered.This
is based on the assumption of cast steel for the weld deposit (for *) and plane strain combined with the Von Mises
multiaxial strength criterion (for s). If a worst case scenario of =0 is considered, this leads to a
f
of 1 millimeter. This
would,asdescribedabove,beequivalenttoavalueofa*of0.25millimeterinthecriticaldistanceapproach.
Static mean stresses can be taken into account via a Haigh diagram (see figure 2.7), but the Neuber microstructural
supporthypothesis,onwhichthemethodisbased,hasnotbeenprovenwiththeinclusionofthesemeanstresses.
Theintroductionofanenlargedradiuscancauseanundercutneartheweldtoe(seefigure3.12).Thismaycauseextra
stressconcentrationifthissignificantlyreducestheloadcarryingcrosssection,especiallywhenhighstressesareinvolved.If
theundercutoccurs,correctiontermsaregivenbyRadajetal.(2006).

figure3.12Undercutcausedbyfictitiousnotchrounding(source:Radajetal.,2006)
Highlystressedvolumeapproach
Sonsino(1993)hasdevelopedmethodsthattrytodeterminethestatisticalsizeeffect(see2.3.2)andtheeffectofmulti
axiallocalstresseswithinphaseandoutofphasestressamplitudeswithacalculationmethodbasedonthehighlystressed
volumeapproach.Inthisapproachthestatisticalsizeeffectiscombinedwiththemicrostructuralsupporthypothesis.Itis
assumedthatthecrackinitiationtimecanbedeterminedbasedonthestressesinalocalvolumeofmaterial.Thisvolume
hasbeendeterminedbySonsinoashavingadepthbelowthenotchandasurfaceareawherethenotchstresshasdropped
to90%ofitsmaximumatthenotch.SonsinoproposedthefollowingrelationforK
f
:
aE
f t
kaE
K K
o
o
= (3.12)
Inwhich:
aE
o theendurablestressamplitudeinplainmaterial[N/mm
2
]
kaE
o theendurablestressatthenotch[N/mm
2
]
Equation (3.12) expresses the fact that the endurable notch stress seems higher than the endurable stress for plain
material, instead of assuming that the stress at the notch is lower than elastically calculated, which was assumed in the
previous approaches. Thespecimens for whichthe strength ofplain material is determined, must of course not show the
highlystressedvolumeeffectthemselves,thereforetheymustbeofsufficientsize.Theendurablenotchstressdependson
thehighlystressedvolume(V
0.9
):
34
( )
0,9 kaE
f V o = (3.13)
Thehighlystressedvolumeisdefinedastheareawhere90%ofthemaximumnotchstressisexceeded.Thedepthofthe
regionisdeterminedbythenormalizedstressgradient(equations(3.15)and(3.16))
0,9 0,9
8
V d w
t
= (3.14)
0,9
0,1
d
_
= (3.15)
1
notch
notch
d
dn
o
_
o
= (3.16)
The notch stress gradient depends on the notch radius, cross section dimensions and loading type and can be found in
literatureordeterminedwitha FEManalysis.Therelationbetweenhighlystressedvolumeandallowablestresshasbeen
derivedbySonsino.Theresultsareshowninfigure3.13.
figure3.13Endurablenotchstressamplitudeatweldtoesinstructuralsteelasafunctionofthehighlystressed
volume;basedondifferenttestsincludingbending(B)andtension(T)loading(source:Radajetal.,2006)
Nowmaximumstressforinfinitefatiguelifeofthenotchedcomponentcandirectlybederivedfromtheinfinitefatigue
lifeofaplainmachinedspecimen.
Extensionintofinitelife
Although the notch stress approach was derived for infinite lives, there also have been attempts to extend the
application into the finite life regime. Schijve proposes to construct an SN curve on the basis of two asymptotes and the
intermediateBasquinrelation(Schijve,2001),seefigure3.14.Theupperasymptoteisdeterminedbytheultimatestrength
ofthespecimenandthemeanstress.Thelowerasymptotecanbecalculatedwiththenotchstresstheoryforinfinitelife,as
has been treated in the previous section. To complete the SN curve only the slope of the Basquin relation needs to be
known. Schijve circumvents this and proposes to define a fixed number of relations for N
up
and N
knee
as shown in figure
3.14.
35
logN
log
a
a
=R
m
a
=
f
N
up
N
knee
Basquinrelation
figure3.14EstimateofanSNcurve.Notetheuseofthestressamplitude(
a
)insteadofthestressrange()
The values proposed for N
up
and N
knee
are 10
2
and 10
6
respectively, based on tests on notched specimens. A remark is
madethatchoosing10
2
introducesaslightconservatismandthat10
3
wouldcorrespondbetterwithtestresults.Itmustbe
notedthatthemethodproposedisderivedfornotchedspecimens,whichtheauthordistinguishesfromweldedjoints.In
the chapter on welded joints a remark is made that the knee point (N
knee
): is found at a significantly higher fatigue life,
about 210
7
. For application for welded joints this most probably will be the more suitable value to use for N
knee
when
constructing the SN curve. The normative values for the fatigue limit specified by Eurocode (510
6
) and IIW
recommendations
1
(10
7
) (Hobbacher, 2007) lie in between the proposed values of Schijve for constant amplitude fatigue
loading.
AnothermethodwasproposedbyHcketal.(1981),whichwassummarizedbyGudehusetal.(1999).Herethefatigue
limit of a component is also determined by a notch stress analysis, but the slope of the Basquin relation also depends on
thedeterminedfatiguenotchfactoraccordingtoequation(3.17)forrolledsteelandequation(3.18)forcaststeel.
2
12
3
f
m
K
= + (3.17)
2
5.5
6
f
m
K
= + (3.18)
Thekneepointofthefatiguestrengthcurveisdependingontheslopeofthecurveaccordingtothismethod.Theknee
pointofthecurvecanbedeterminedwithequation(3.19)forrolledsteelandequation(3.20)forcaststeel.
2.5
6.4
10
m
knee
N

= (3.19)
3.6
6.8
10
m
knee
N

= (3.20)
Withtheuseofthesetwoequations,theslopeandkneepointofthefatiguestrengthcurvearedetermined.Whenthe
fatiguelimithasbeendeterminedwiththenotchstressanalysisandtheupperkneepoint(N
up
,seefigure3.14)isignored,
the fatigue strength curve is determined. The neglect of the upper plateau by not using N
up
results in a fatigue strength
curveonlyusableinmediumandhighcycleregimes.
1
TheIIWrecommendationsonlyshowaconstantamplitudefatiguelimitforstandardapplications.Forveryhighcycleapplications,also
beyond10
7
cyclestheSNcurveshowsaslope.
36
First the elastic notch stress concentration factor needs to be determined. This can be done with FEManalysis,
measuring or functional analysis methods based on theory of elasticity (Radaj et al., 2006). The elastic notch stress
concentrationfactoristhendefinedas(rewrittenfrom(3.6)):
notch
t
nom
K
o
o
= (3.21)
FromthiselasticnotchstressconcentrationfactorthefatiguenotchfactorK
f
needstobedetermined.Forthis,oneofthe
theoriesexplainedabovecanbeused.Themaximumstressatwhichnocrackinitiation(
notch;per
)occurscannowdirectlybe
obtained:
notch
n
f
K
o
o = (3.22)
; notch notch per

o o s (3.23)
Inwhich:
; notch per
o thepermissiblenotchstress[N/mm
2
]
notch
o theactualnotchstress,basedonthenominalstressandanalyseswithrespecttoK
f
[N/mm
2
]
The permissible notch stress can be derived from tests on plain specimens, or by mathematical relations between the
staticstrengthandfatiguestrength.
Ifanymisalignmentsarepresenttheymustbetakenintoaccount.Formulastotakemisalignmentsintoaccountaregiven
byHobbacher(2007).Oncethestresslimitforinfinitelifeisknown,anestimateofthemaximumstressforacertainfinite
lifecanbemadebymeansofthemethodofSchijve,ortherelationsfoundbyHck.
Thismethodrequiresmoreeffortthanthestandardmethods.Besidesthat,someinformationonthematerialandlocal
geometry to determine the correct value of K
t
and K
f
is necessary. When these can be determined accurately enough, a
reliableestimatecanbemadeofthefatiguelimitoftheconsideredcomponent.
Themethodwasdevelopedtodeterminethestresslevelbelowwhichnocrackinitiationoccurs.Thisisofcourseavery
important point in the SN curve, but limits design possibilities to a failfree design. With the additions of Schijve and
Haibach the method includes crack initiation over both the medium and high cycle domain. This means that for civil
engineeringstructuresthefullfatiguelifecanbeobtainediftheanalysisiscombinedwithacrackpropagationanalysis.
Thismethodcurrentlydoesnotfindalotofapplicationindesign,becauseoftherequiredeffort.
37
4 Literaturereview:Practice
4.1 Chapteroutline
In paragraph 2.3.4 it was mentioned that high strength steel does not always show a significantly better fatigue
behaviour than standard steel (see figure 2.11). In this chapter, this will be further investigated. First the current design
codes and recommendations are discussed with regard to fatigue and high strength steel. Then the actual behaviour
according to experiments will be discussed. First for plain material and then for non plain material, which contains both
notchedandweldedspecimens.
The effect of TIGdressing was discussed in general in 2.4.2. In this chapter the current state of design codes and
recommendationswithregardtoTIGdressinganditsbeneficialeffectsisdiscussed.ThentheinfluenceofTIGdressingon
thespecimenisdiscussed,followedbytheactualbehaviourofTIGdressedspecimens,bothforstandardandhighstrength
steel.
4.2 Literatureregardingfatigueandhighstrengthsteel
4.2.1 Strengthaccordingtocurrentdesigncodesandrecommendations
Three design codes will be discussed: Eurocode, IIW recommendations and NPRCEN/TS 1300131. Eurocode is taken
intoaccountbecauseitiswidelyusedinEuropeansteelstructures.TheIIWrecommendationsandtheNPRdesigncodeare
takenintoaccountbecausetheyalsocoververyhighstrengthsteels.
Eurocode199319(2006)
EurocodeisapplicableforsteelsuptoS700sincetheadditionalcodeNENEN1993112(2007)wasintroduced.Thecode
is mainly focused on the nominal stress approach, but also includes an annex addressing the structural stress approach.
Both the detail categories specified for the nominal stress approach and the structural stress approach do not distinguish
differentdetailcategoriesfordifferentsteelgrades.Thismeansthatanybeneficialeffectoftheparentmaterialstrengthis
notreferredtointhecode.Therefore,twoidenticalstructuresmadefromstandardsteelandhighstrengthsteelwillhave
thesamefatigueresistanceaccordingtoEurocode.
IIWrecommendations
The design recommendations by the International Institute of Welding (Hobbacher, 2007) cover design of structures
madeofstructuralsteelsuptoayieldstrengthof960N/mm
2
.Thedocumentcoversvariousfatigueapproaches:nominal
stress design, structural stress design, notch stress approach and crack propagation approach. If physical tests are
necessary,thedocumentgivessomeinformationontestproceduresandstatisticalanalysis.
The detail categories that are specified for the nominal stress approach and the structural stress approach do not
distinguishdifferentFATclassesforhighstrengthsteel,andthereforedonotdifferfromEurocodeinthisaspect.
NPRCEN/TS1300131
The CEN code is mainly applied to steel crane structures, but is interesting for this case because it, contrary to the
Eurocode and IIW recommendations, adjusts the fatigue strength of a component to a certain extent, depending on the
ultimatetensilestrengthofthematerial.Thecodeisapplicabletostructuresmadeofstructuralsteelsuptoayieldstrength
of960N/mm
2
andcoversthenominalstressdesign.
Thecodeisdifferentfromthecodesmentionedbefore,becauseitspecifiestheslopeoftheSNcurvedependingonthe
detail that is used, and specifies the FATvalue depending on the ultimate tensile strength of the base material. This
differentiation according to steel grade is only applied for plain material such as flat bars, plates and rolled profiles. The
code distinguishes steels with a yield strength lower than 275 N/mm
2
, between 275 N/mm
2
and 355 N/mm
2
and higher
than 355 N/mm
2
. The fact that this differentiation only takes place in the standard grades and classifies all high strength
steels together may not be optimal, but is certainly less conservative than the approach of Eurocode and the IIW. An
exampleofthedesignlineofNPRCEN/TS1300131isshowninfigure4.2.
4.2.2 Behaviourofplainmaterial
Differentauthorshavefoundthefatiguestrengthofaplainspecimenatacertainnumberofcyclestobeincreasingwith
increasingstaticstrength(Maddox,1991;Gurney,1979).Gurneypublishesasetofdatafrompreviousresearch,showingan
approximatelinearrelationbetweenfatiguestrengthandultimatetensilestrength(seefigure4.1).
38
figure4.1Relationbetweenultimatetensilestrengthandfatiguestrengthofplainmachinedspecimens.Blackdata
pointsareoriginallybendingtests,butanalyticallyconvertedtotensileloading(source:Gurney,1979)
Pijpersconcludednotonlythatthefatiguestrengthoftheplainspecimensincreaseswithincreasingtensilestrength,but
alsofoundaslopeoftheSNcurvelesssteepthanspecifiedincertaincodes(Pijpersetal.,2009).Thecomparisonbetween
Eurocode, NPRCEN/TS 1300131 and the results are shown in figure 4.2. A comparison is not made with the IIW
recommendations,thislinewillbesimilartotheonespecifiedbyEurocodebutmarginallymoreconservative.
figure4.2SNcurveofbasematerialsbasedonresearchfromPijpers(source:Pijpersetal.,2009)
InPijpersresearchtheS1100specimensdonotseemtoperformbetterthantheS690specimens.Infact,thecalculated
FATvalue for S690 was higher (391 N/mm
2
) than for S1100 (317 N/mm
2
). It was observed that in the S1100 specimens
cracksoccurredoutsidethetaperedcrosssection,whichhadbeengroundedtoplacestraingauges.Thecracksoccurredin
thewidercrosssectionwheregrindinghadnottakenplace.Apparentlytheseveryhighstrengthsteelsareverysensitiveto
surface roughness. Unfortunately, no data exists where cracks occurred in a smooth section of the S1100 specimens, but
thefactthatthecracksoccurredoutsidethetaperedsections,despitethehigherstressinthetaperedsections,showsthat
forsmoothmaterialhigherfatiguestrengthscanbereachedthandepictedinfigure4.2.
Different data sets show what was already stated in 2.3.2: if the static strength of steel increases, the fatigue strength
alsoincreasesforsmoothspecimens.
4.2.3 Behaviourofnonplainmaterial
Inpractice,moststructureswillnotmadeofplainmaterial.Structuresarewelded,withtheassociatedmicrocracksand
inclusions, have a certain surface roughness or other forms of notches. Gurney has collected data from previous
experiments where specimens with machined surfaces are compared with specimens which are tested as produced. The
effect of the present millscale and possible small defects is clear (see figure 4.3). Not only is the ratio between ultimate
tensile strength and fatigue strength lowered from 0.90.625 to 0.6250.4, but also above an ultimate strength of
approximately700N/mm
2
thefatiguestrengthdoesnolongerseemtoincreasewithincreasingultimatestrength.
39
figure4.3Comparisonofplainmachinedspecimenswithasproducedspecimens.Thescatterbandforspecimens
withmachinedsurfacesisderivedfromthedatainfigure4.1(source:Gurney,1979)
ThesensitivitytosurfaceroughnessthatwasassumedbyPijpers(see4.2.2)isclearlyshowninfigure4.3.Gurneyfound
thatthiseffectbecomesvisibleaboveacertainultimatetensilestrength,whichalsomatcheswithPijpersresearch,where
the S1100 specimens showed the very sensitive behaviour, while the S690 specimens cracked at the machined, reduced
crosssection.
ResultsofweldedconnectionsarealsoavailablefromGurney.Thesespecimenshavemoreseverenotchesthantheplain
machined specimens and asproduced specimens. Also,some residual stressesmay be introduced and the microstructure
of theheat affected zone maydiffer from theparent material. Theeffect of welding on the specimens is clearly visible in
figure4.4.Thestrengthoftheweldedconnectionsdoesnotshowthesameincreaseoffatiguestrengthwhenthetensile
strengthoftheparentmaterialincreases.Infact,theultimatetensilestrengthofthematerialdoesnotseemtohaveany
influenceatall.
figure4.4Fatiguestrengthofweldedspecimens(source:Gurney,1979)
Pijpershascollecteddatafromdifferentauthors,whichisveryinsightfulinthismatter,asshowninfigure4.5.Theplain
machined specimens show the linear relation as was earlier described. When some roughness, notch or weld defects are
includedthebeneficialeffectofhighstrengthsteelquicklydiminishes.
40
figure4.5Fatiguestrengthat210
6
cyclesfordifferentspecimens,bothbasematerial(BM)andtransversebutt
welded(TBW);thenumbersbetweenbracketsrefertothesourcesofthedatapointsintheoriginalarticle(source:
Pijpersetal.,2009)
Another large research project was executed by Anami, which focused only on welded joints. For different joints he
testedlocaljointspecimensandgirderspecimenswhichincludedthesametypeofjoint.Fromthelargeamountofdatathe
moststrikingexamplesaredepictedinfigure4.6andfigure4.7,inwhichitisclearthatthehighstrengthspecimensmostly
completely blend in with the lower strength specimens. He found that welded high strength steel structures might even
behaveworsethanstandardsteel.Thiseffectwasmostidentifiablewiththeoutofplanegussetgirderspecimens(aloaded
platewithagussetplateattached;nonloadcarryingfilletwelds),seetherightdiagraminfigure4.7.Anamisuggeststhat
this might be the case because in higher strength steel, higher residual stresses are possible. The fact that the effect is
mostly visible in the girder specimens is because in these specimens the residual stresses could arise due to constraints.
Other parameters, such as heat input and weld beat effects, where kept as constant as possible between the joints
specimensandgirderspecimens.Unfortunately,theappliedstressratioisnotstated.
figure4.6Fatiguetestdataforlongitudinalweldedjoints.Left:jointspecimens,right:girderspecimens.The
mentionedstrengthsaretensilestrengths(source:Anamietal.,2001)
41
figure4.7Fatiguetestdataforoutofplanegussetweldedjoints.Left:jointspecimens,right:girderspecimens.The
mentionedstrengthsaretensilestrengths(source:Anamietal.,2001)
Fromtheinformationgatheredtwoconclusionscanbedrawn.First,thereisindeedaneffectofthematerialstrengthon
the fatigue strength of a component or joint. It is an approximately linear relation between fatigue strength and material
tensile strength. However, in real constructions it will be very complicated to actually achieve this improvement, because
high strength steels are more sensitive to roughness, notches and defects, caused by production or welding, which could
easilyleadtoafatiguestrengthcomparabletothefatiguestrengthofstandardsteel.
4.3 LiteratureregardingTIGdressingandhighstrengthsteel
4.3.1 TIGdressingprocess
AccordingtoIIWrecommendationstheaimofTIGdressingistoremovetheweldtoeflawsbyremeltingthematerialat
theweldtoe.Italsoaimstoreducethelocalstressconcentrationeffectofthelocalweldtoeprofilebyprovidingasmooth
transitionbetweentheplateandtheweldface(Haagensenetal.,2001).
The remelting of the weld toe is done with a standard TIG machine. The process is relatively sensitive to weld
contaminantssuchasmillscale,rust,oilandpaint.ThereforetheareatobeTIGdressedmustbepreparedbycleaning,wire
brushingandlightgrinding.Alotofparametersareofinfluenceontheendresult,suchastheshieldinggas,travelspeed,
weldingcurrent,positionoftorch,etc.ForexactspecificationsofthecorrectTIGdressingprocedure,referenceismadeto
theIIWrecommendations(Haagensenetal.,2001).AnexampleofaTIGdressedspecimenisshowninfigure4.8,wherethe
rightsideoftheweldhasalreadybeendressedandtheleftsideisstillintheasweldedcondition.Thechangeingeometry
regardingtheweldtoeisclear.
figure4.8Twopolishedspecimens.Ontheleftaswelded,ontherightTIGdressed.
4.3.2 Strengthimprovementaccordingtocurrentdesigncodesandrecommendations
Eurocode
Eurocode NENEN 199319 does not specify any benefit from weld improvement techniques, such as TIGdressing.
However,forcertaindetailcategoriesgrindingisspecified.Forexampleintable8.3ofNENEN199319,thedetailcategory
112 is applicable for butt welds which, along with other quality specifications, have to be ground smooth. If this
42
specificationisnotmet,thedetailcategoryislowered.Unfortunately,burrgrinding,TIGdressingorhammerpeeningisnot
appreciatedinthismanner.
IIWrecommendations
The IIW recommendations acknowledge the possible beneficial effect of weld improvement techniques which are
classifiedinthreecategories:
Methodsforimprovementoftheprofile(machiningtheentireweldorweldtoeandTIG/plasma/laserdressing)
Methodsforimprovementofresidualstressconditions(peening,overstressing,thermalstressrelief)
Methodsforimprovementofenvironmentalconditions(painting,resincoating)
For the methods that improve the weld profile, recommendations are given. Only the possible improvements of TIG
dressingwillbefurtherexplainedhere.
First the applicability is discussed. The application of the recommendations is limited to joints that are suitable for
improvement,whicharetheoneswherethecrackstartsfromtheweldtoe.Furthermore,therecommendationsareonly
applicableforsteelswithayieldstrengthlowerthan900N/mm
2
,jointswithplatethicknesseswithinspecifiedboundaries
andhighcyclefatigue.SomeotherspecificationscanbefoundintheIIWdocument(Hobbacher,2007).Allimprovements
are given as an increase in fatigue strength with the as welded conditions as starting points and can only be used in the
nominalstressapproachorstructuralstressapproach.
The improvement in nominal stress design by TIG dressing is specified as a factor 1.3 in fatigue class for fatigue classes
with FAT90, with a maximum possible fatigue class of 112 after improvement. In structural stress design table 4.1 is
applicable. Note that no difference in improvement is specified between mild steel and higher strength steel, while for
other improvement techniques, such as needle peening and hammer peening, high strength steel welded joints are
assumedtohavehigherfatiguestrengthimprovement.Withbothmethodsitmustbenoticedthatthefatigueresistanceof
theimprovedjointisalwayslimitedbythefatigueresistanceofthebasematerial.
Material Loadcarryingfilletwelds Nonloadcarryingfilletweldsandbuttwelds

Mildsteel(f
y
<350N/mm
2
) 112 125
Higherstrengthsteel(f
y
>350N/mm
2
) 112 125
table4.1FATclassesforusewithstructuralstressapproachatjointsimprovedbyTIGdressing.Foraswelded
connectionstheFATclassesare90or100,dependingonthespecificdetail(source:Hobbacher,2007)
The beneficial effect of TIGdressing is not taken into account in the notch stress approach as prescribed by the IIW
because the actual weld geometry, which improves due to the dressing, is replaced in this code by an artificial geometry
witharadiusof1millimeter,becauseofthewidescatterandirregulargeometryofarealweldprofile.
NPRCEN/TS1300131
In this design code the different weld classes specified in NENEN 25817 are used. To the different detail categories,
different FAT values are given, depending on the weld quality. CEN/TS 1300131 specifies another extrahigh quality
categoryforwelds:B*.B*weldshavetocomplywith,inadditiontotheBqualityrequirements:
Fullpenetrationwithoutinitialpoints
Bothsurfacesmachinedflushgrounddowntoplatesurface;grindinginstressdirection
Theweldtoeposttreatedbygrinding,remeltingbyTIG,plasmaweldingorbyneedlepeening
Eccentricityofthejoiningplateslessthan5%ofthicknessofgreaterplate
Sumoflengthsofconcavitiesofweldlessthan5%oftotallengthoftheweld
100%NDT(inspectionalongwholeweldlength)
These B* quality welds have a higher FATvalue, just as the B quality welds have a higher FATvalue than the C quality
welds.Furthermore,thecodeallowsCqualityweldsofanytypetobeupgradedtoBqualityweldsforfatiguedesignifTIG
dressingisappliedatthepotentialcrackinitiationzoneinordertoincreasethefatiguestrength.
To illustrate the benefits of this quality improvement, an asymmetric butt joint with normal stress across the weld is
takenasanexample,seetable4.2.The11%benefitoftheupgradefromCqualitytoBqualityislessthanthe30%increase
the IIW prescribes, but the application of the upgrade is not limited to a certain FATvalue, as is the case in the IIW
recommendations.TheextensiveweldtreatmenttoupgradeaweldtoB*quality,includinggrindingandTIGdressingalso
grantsaboutthesameamountofextrafatiguestrength.
FAT Weldquality Relativeimprovement

112 B*quality 1.12(1.24inrelationtoFAT140)
100 Bquality 1.11
90 Cquality 1.00
table4.2FATvaluesspecifiedfordifferentweldqualities(source:NPRCEN/TS1300131)
43
4.3.3 InfluencesofTIGdressingonmaterialandgeometry
Geometry
IntheIIWrecommendationsonweldimprovementtechniques(Haagensenetal.,2001)detailedinformationisgivenon
theTIGdressingprogressandsomeglobalinformationisgivenonthegeometricalresultoftheprocess.Thetorchposition
resultingintheoptimumshapeintheoryandanactualspecimenaredepictedinfigure4.9.
figure4.9EffectofTIGdressingaccordingtoIIWrecommendations(source:Haagensenetal.,2001)
The geometrical changes of TIG dressing can be quantified if the weld toe radius and weld toe angle are defined (see
figure 4.10). The main effect of TIG dressing is the increase of the weld toe radius which results in a lower stress
concentrationattheweldtoe.Theweldtoeangleremainsapproximatelythesameforfilletweldsorbuttweldswithacap
thatishighenough.Forratherflatbuttweldstheanglewillbereduced.Thereductionofwelddefectssuchasmicrocracks
andsmallinclusionsismoredifficulttoquantify,andiscoveredinthescatteroftheresultsofmostresearches.Infact,the
TIGdressingshouldreducescatterbyreducingthedefects.
figure4.10Definitionofweldtoeradiusandweldtoeangle
DifferentauthorshavedoneresearchonthechangeofweldtoeradiusasaresultofTIGdressing.Pedersenhascollected
a number of studies and has compared the weld toe radius of three weld improvement techniques with the as welded
condition(Pedersenetal.,2010).Hisresultsareshowninfigure4.11.Itisclearthatasignificantimprovementinweldtoe
radius can be achieved by TIG dressing from an average of 11.5 millimeter to an average of 6 millimeter. However, the
variationoftheradiusalsoincreasesmuchcomparedwiththeasweldedcondition.Unfortunatelyitisnotclearonwhich
weld type these analyses are made, but context suggest the considered studies are mostly focused on fillet welds. This
wouldalsoclarifytheabsenceofmeasuredweldtoeangles.
figure4.11ComparisonoftoeradiiinasweldedandTIGdressedcondition;thenumbersbetweenbracketsreferto
thesourcesofthegraphsintheoriginalarticle(source:Pedersenetal.,2010)
Residualstressesandmicrostructure
Theadditionalweldprocesswillhaveitsinfluenceontheresidualstressdistributioninthespecimen,themicrostructure
and the associated hardness of the material in the (new) fusionzone and HAZ.A concern when TIG dressing isapplied to
highstrengthsteelconnectionsisthepossibilityofexcessivesofteningofmaterial(Pedersenetal.,2009).Hisinvestigation
showedahardnessdropof15%to20%intheTIGdressedarea.
44
AnumericalanalysisoftheinfluenceofTIGdressingonS690basematerialhasbeenmadebyHildebrand(Hildebrandet
al.,2006).HisresultsshowastrongincreaseinmartensiticstructureintheTIGdressedregionwherethepreviousstatewas
mainly bainite. He also found a reduction in residual stress at the surface of the joint, while internally the area with high
residualstressedincreased.Theresultsofthesimulationforabuttjointareshowninfigure4.12.
figure4.12Microstructureandresidualstressinsimulated,TIGdressedbuttjoint(source:Hildebrandetal.,2006)
TheseresultsagreewellwiththeresearchofLopezMartinez (1997), whereresidualstresseswheremeasuredinactual
specimens, before and after TIGdressing. The material that was used has a yield strength of 590 N/mm
2
and a tensile
strength of 757 N/mm
2
. He also found evidence that TIGdressing does not only influence the fatigue strength by
geometricalmodificationofthejoint,butalsobyloweringtheresidualstressesatthesurfaceoftheplates.Thetestswere
executedontestspecimensasshowninfigure4.13.Infigure4.15tofigure4.18hisfindingsareshown.
figure4.13GeometryofspecimenstestedbyMartinez(source:LopezMartinezetal.,1997)
45
figure4.14Upperviewoftestspecimenwithlocationsforresidualstressmeasurements(source:LopezMartinezet
al.,1997)
figure4.15LongitudinalresidualstressdistributioninAsectionofasweldedspecimen(source:LopezMartinezet
al.,1997)
figure4.16LongitudinalresidualstressdistributioninBsectionofasweldedspecimen(source:LopezMartinezet
al.,1997)
figure4.17LongitudinalresidualstressdistributioninAsectionofTIGdressedspecimen(source:LopezMartinezet
al.,1997)
figure4.18LongitudinalresidualstressdistributioninBsectionofTIGdressedspecimen(source:LopezMartinezet
al.,1997)
ItisclearthatthemaximumresidualstressincreasesduetoTIGdressing.Forexample,themaximumstressinthestress
state of figure 4.15 is 556 N/mm
2
while in the stress state of figure 4.17 this is increased to 699 N/mm
2
. At the surface
however, the residual tensile stresses are reduced from 360 N/mm
2
to 256 N/mm
2
. As explained in 2.3.2, tensile residual
stresses have a detrimental influence on the fatigue strength of a component. Therefore, a reduction of tensile residual
stresses can increase the fatigue strength of the joint. At some distance from the weld toe, section B in figure 4.14, the
tensilestressesatthesurfaceincreaseduetotheTIGdressing,comparefigure4.16andfigure4.18.Thishasofcoursethe
effect that the compressive residual stresses near the edge of the specimens increase. For welds, the most likely location
46
for crack initiation is at the weld toe at the surface. This is exactly the location where the residual stresses are reduced,
whichexplainsthebeneficialeffectofthechangeoftheresidualstressstate.
4.3.4 InfluencesofTIGdressingonfatiguestrength
InthepreviousparagraphsthemaininfluencesofTIGdressingarediscussed:
Increaseofweldtoeradius
Changeofresidualstressstate
Changeofmicrostructure
Reductionofweldflawsandinclusionsattheweldtoe
Especially the reduction of weld flaws and inclusions in combination with the increase in weld toe radius isassumedto
createthebeneficialbehaviourofTIGdressedspecimenswhencomparedwithasweldedspecimens.
In general, two results are reported in research concerning the increase in fatigue strength: the fatigue strength at a
certain number of cycles (usually 210
6
is taken as an example) increases and the fact that this strength increase is most
notable at a high number of cycles. This second result means a higher mvalue, or flatter SN curves. Some examples of
theseeffectsareshownbelow,fromastudybyDahleonfilletwelds.TheflatterappearanceoftheSNcurveiscausedby
theincreaseincrackinitiationtime,whichmostlyoccursatlowerstressesandhighercycles.Becausethehighcycleregime
has more benefit from the treatment, the SN curve rotates and the mvalue is increased. The fact that crack initiation
becomes important means that the governing factors for fatigue are stress concentration factor, material strength, mean
stress, residual stress, stress range, etc. (Dahle, 1998). For high strength steel the crack initiation time is more important
with respect to the total fatigue life than it is for standard steel, therefore the high strength steel data show a clearer
flatteningoftheSNcurveduetoTIGdressing(seefigure4.19)andincreaseoffatiguestrengththanstandardsteel.
figure4.19SNcurvesofTIGdressedfilletweldspecimenscomparedtoanasweldedreferencescatterband.Left:
S590steel,right:S900steel(source:Dahle,1998)
Dahle has coupled the change in mvalue, log(C) value and FATvalue to the material strength. The results of this
regression are shown in figure 4.20. It is clear that both the mvalue and the value of intersection with the vertical axis
increase when the material strength increases. This will also result higher FAT values. The algebraic formula for the
improvementoftheFATvalueis:
0,2
0,001056 0,65982 FAT R A = + (4.1)
ItmustbenotedthatthisFATvaluedoesnotcorresponddirectlytothestandarddetailcategories,butistobeusedwith
thealsoadjustedslopemandintersectionlogCasshowninfigure4.20.
Thisformulaleadstoarelativelylowincreaseinfatiguestrengthfornormalsteelstrengths,butmatcheswiththedata.A
remarkismadethatthewidescatterofthedataresultedinaveryconservativedesignline.
47
figure4.20InfluenceofmaterialstrengthonSNcurveparameters.Left:mandC.Right:FATvalue.(source:Dahle,
1998)
Dahle also calculated the increase in fatigue strength of the individual data points with respect to the reference
scatteringband.Intable4.3theaverageoutcomesofthiscalculationareshownpersteelgrade.Itisclearthatthematerial
strength has an influence on the beneficial effect from TIG dressing. Both data from constant amplitude loading and
variableamplitudeloadingareincorporatedinthecalculation.
SteelGrade Yieldstrength[N/mm
2
] Averageimprovementoffatiguestrength
350 398 1.185
590 615 1.424
700 780 1.725
900 900 1.894
table4.3Increaseinfatiguestrength(source:Dahle,1998)
SimilarresultsarisefromtheresearchofPedersenetal.(2009),wherethefreeslopeofthefatiguecurveofTIGdressed
S700 grade specimens increases to 5.9 compared to 4.2 for as welded specimens of mixed steel grade. The reported
beneficialeffectat10
5
cyclesisaround38%andaround70%at210
6
.Thisresearchalsoconsidersfilletweldsonly
More research on the beneficial effect of TIG dressing on fillet welds has been done by Huo (2005). This research does
not focus on high strength steel but distinguishes results for constant amplitude loading and variable amplitude loading.
Theresults,comparingtwopostweldtreatmentstotheasweldedcondition,areshowninfigure4.21.Thetestdatashow
much less scatter than the low strength data of Dahle, which leads to a greater increase of the fatigue strength:
approximately37%forsteelwith
y
=390N/mm
2
.
figure4.21Comparisonoftwoweldimprovementtechniqueswiththeasweldedsituationforasteelwithayield
limitof390N/mm
2
andatensilestrengthof590N/mm
2
.Testedonfilletweldspecimens(source:Huoetal.,2005)
TheeffectsdiscussedbyDahlearepartiallyvisibleinthedatapublishedbyHuo.Ifonlytheconstantamplitude(CA)data
publishedbyHuoareconsidereditisclearthatthefatiguestrengthoftheconnectionincreases,butthehighermvalueis
barelyvisible.Thismightbeduetothefactthatthesteelinquestionisofstandardquality,whiletheflatteningeffectinthe
studyofDahlewasmostnotableforhigherstrengthsteels.
48
OtherresultsarepublishedbyHaagensen,whofoundanaverageincreaseof44%inthefatiguestrengthofafilletweld
joint,butnochangeinslopeoftheSNcurve(Haagensenetal.,1987).Thisresultwasalsofoundforstandardsteelwitha
yieldstrengthof365N/mm
2
,whichmeansitdoesnotexcludetheflatteningeffectforhigherstrengthsteels.
AllresearchshowsanincreaseinfatiguestrengthofTIGdressedspecimens.Theincreaseinfatiguestrengthdiffersfrom
20% to 40% for standard steels and can be as high as 90% for very high strength steels. The research data by Dahle and
PedersenalsoshowsarotationoftheSNcurve,especiallyforhighstrengthsteelsinthecaseofDahle.Otherresearches
donotshowthiseffect,butalsodonotincorporatehighstrengthsteels,sotheydonotconfirmnorrefutethepresenceof
thiseffect.Alldiscussedresearchesconsiderfilletweldsonly.Whethertheeffectwillbesimilaronbuttweldsshallhaveto
showfromthistestingprogramme.
49
5 Testsetup
5.1 Chapteroutline
Inthischapterthetestingprogrammeandtestsetuparediscussed.Inparagraph5.2,theidentification,dimensionsand
fabricationofthespecimensisdiscussed.Thetestsetupsthathavebeenusedforthedifferentexperimentsareshownand
explainedinparagraph5.3.
5.2 Testingprogramme
5.2.1 Identificationoftestspecimens
In total there were 11 plates available for testing, each with different specifications. The specimens indicated by a V
consistoftworolledplatesconnectedbyaVbuttweld.ThespecimensindicatedbyaCconsistofarolledplateandacast
plateofsimilarstaticstrength,connectedbyaVbuttweld.Thecaststeelplateisingeneralslightlythicker(about5%)than
therolledsteelplate.Thenumberinthespecimenidentificationcodereferstothefirsttwonumbersofitssteelgrade.The
availableplatesare:
PlatesV89,V46,V69,C69andC46withapproximatedimensions330x25x950mm
3
.
PlatesC11andV11withapproximatedimensions330x20x950mm
3
.
PlateC89withapproximatedimensions320x25x950mm
3
,withalongitudinalcutalongthefulllengthexcept
fortheweldarea.
PlateC89awithapproximatedimensions160x25x950mm
3
.
With the available material 24 strips with a width of approximately 100 millimeter were produced for fatigue testing.
Each plate, except C89 and C89a, is therefore divided in three pieces of approximately 100 millimeter and one 25
millimeterresidualstripwhichcanbeusedforhardnessmeasurements.TheC89plateisdividedintwospecimensandthe
third C89 specimen is produced from plate C89a. The produced specimens are shown in Annex B in tabulated form. The
materialspecificationscanbefoundinthematerialcertificatesprovidedinAnnexD.
5.2.2 Preparationofspecimens
ThespecimenswerepreparedaccordingtoIIWrecommendationsonTIGdressing(Haagensenetal.,2001).Boththecap
and root sides of the Vbutt weld were TIG dressed. To provide a suitable geometry on which to apply TIG dressing, the
weldrootwasgroundoffuntilalimitedheightremained.TheweldrootsoftwospecimenscouldnotbeTIGdressed:
C69.Aftergrindingtheweldroot,smallporositieswerevisible.Topreventthesegasinclusionsswellingupduring
the TIG dressing process, resulting in an unsatisfactory end result, this root was not TIG dressed. It would be
interesting to investigate whether these gas inclusions would indeed have spoiled the TIG dressing process,
whethertheseinclusionswouldshowuponanXraytestandwhethertheywouldpassorfailthistest.Depending
ontheanswersonthesequestionstheseporositiescouldhaveseriousconsequencesforthepracticalapplication
oftheTIGdressingprocess.
C89. This plate consists of 2 butt welded specimens joined together longitudinally. One half of the plate had
insufficient weld height to prevent the TIG dressing process causing undercut. Therefore, this half was not TIG
dressedontheweldroot.
InboththesecasesonlytheweldtoesattheweldcapwereTIGdressed.Becausethemisalignmentduetoweldingwill
cause extra tensile stresses at the weld cap when the fatigue tests commence, the fatigue cracks are expected to initiate
there.ThereforethedecisionnottoTIGdresstheweldrootswillprobablynothaveanyinfluenceontheoutcomeofthe
tests.
Allplateswereflamecutintostripsofapproximately100millimeterinwidth.Subsequentlyareductionincrosssection
wasmadeneartheweldtopreventfatiguecracksintheparentmaterialnearorintheclampsofthetestingrig.Thefinal
dimensionsofallspecimensaredepictedinAnnexB.
50
5.3 Testsetup
5.3.1 Measurementofweldgeometry
TomakeanaccuratemeasurementoftheweldgeometryasensorhasbeenmountedtoanXYmotor.Thislasersensor
measures the distance from its own position to the object that is to be scanned. An overview of the measuring setup is
showninfigure5.1.
figure5.1Setuptomeasurethegeometryofwelds
The XY motor has a precision of 1/160 millimeter in both directions. The laser sensor has a resolution of 1/1000
millimeter overits full bandwidth which can be increased by sample averagingto 1/10000 millimeter. An overview of the
typeoflasersensorthathasbeenusedisshowninfigure5.2.
figure5.2Typeoflasersensorusedinmeasuringtheweldgeometry
Siliconrubbercasts
To speed up the total process of measuring and preparation of the specimens and to make the changing of specimens
moremanageable,siliconrubbercastsweremadeofthewelds.Thecastsareanexactnegativereproductionoftheweld.
The specifications of the used silicon rubber can be found in the manufacturers website. The used rubber is of the type
PS81020 from Polyservice in The Netherlands. In figure 5.3, laboratory staff is casting the silicon rubber onto the steel
plates.Theresultofthecastingisshowninfigure5.4.
51
figure5.3Castingofthesiliconrubberontothesteelplates
figure5.4SiliconrubbercastofC89aplate
5.3.2 Measurementoffatiguelife
ThepreparedspecimensaretestedunderaxialloadingintheStevinLaboratoryoftheDelftUniversityofTechnology.The
usedtestrigisdepictedinfigure5.5.Thetestingrighasamaximumloadof600kN.
figure5.5Schematicoftestingrig(imagesource:Pijpers,2011)
52
Thestartingpointofeachmeasurementistheunclamped,unloadedspecimen.Thismeansthatthedeformationarising
fromthestaticclampingoftheimperfectspecimensalreadyresultsinnonzeromeasuredtensilestrainsattheweldcap.
For all specimens the intention is to test at R=0.1, but due to the aforementioned tensile strains and the resulting
stresses,thisrequirestheminimumloadtobeacompressionloadinmanycases.Thisisdeemedunpractical,andpossibly
unsafe, therefore a minimal tensile load of 10 kN is taken as a lower bound for the minimum load. The actual Rvalue is
documented,basedonboththestraingaugesandthemeasuredforce.
Twelve strain gauges were placed on each specimen. The location of the strain gauges and global dimensions of the
specimensaredepictedinfigure5.6.ThenumberingofthestraingaugesisdepictedinAnnexB.
figure5.6Globaldimensionsofspecimenandlocationofstraingauges
Crackmonitoring
When one of the strain gauges starts showing a strain range which deflects significantly from its average, the system
shuts down the test rig so that, when restarted, crack monitoring can take place. When the crack occurs near one of the
straingaugesitcanbedetectedatacracklengthofabout10millimeter.Ifthecrackoccursfurtheraway,thecracklength
atdetectioncanbemuchlonger,upto40millimeter.
Cracksaremonitoredintwo ways.Tovisualizethecrack,some petroleumisbrushedontothesurfaceofthespecimen
andsubsequentlywipedaway.Thelengthofthecrackatthesurfacecanthendirectlybemeasuredwithaverniercaliper.
Thismeasurementcanbeexecutedrelativelyfrequent.
Anothermethodtomonitorcracksiscrackmarking.Thisinvolvesraisingthelowerstressinthespecimentoabout90%
oftheupperstressinthespecimen,therebygreatlyreducingthestressrange,for2%to10%oftheexpectedtotalnumber
ofcyclestofailure.Thisresultsinvisiblebeachmarksonthecracksurfaceafterfailureofthespecimen.Becausethestress
rangeismuchlowerduringthisprocedure,thenumberofstresscyclesmadeinthisprocedurecanbeneglectedinthetotal
amount of stress cycles before failure. A good example of the result of crack marking is shown in figure 5.7, but not all
specimens give the good results as shown in the image. Some specimens fail rather quick after crack initiation, thus
allowingonlytwoorthreecrackmarkings.Inotherspecimens,whilethecrackmarkingwasexecutedmoreoften,noneor
notallmarksarevisible.Ingeneral,crackmarkingcannotbeexecutedasfrequentlyasthesurfacemeasurement,because
themarkingsmusthavesomedistanceinbetweeneachother,preferableafewmillimeters.Furthermore,theassumption
thatthenumberofcyclesduringmarkingcanbeneglectedwillbecomedubiousifcrackmarkingisexecutedtofrequently.
figure5.7Beachmarksvisibleinaspecimen
5.3.3 Measurementofmaterialhardness
Ofallplatesasmallpieceofmaterialhasbeenpolishedandetchedtodistinguishtheweldmaterial,heataffectedzone
andparentmaterial.Onthedifferentzoneshardnessmeasurementshavebeencarriedout.Thisisastandardizedtestand
isexecutedaccordingtotheapplyingcodes.Formoredetailsonthetesting,referenceismadetoAnnexH.Theresultsare
discussedin7.4.
53
6 Processingandresultsoflasermeasurements
6.1 Chapteroutline
In this chapter the results of the laser measurements are discussed. In paragraph 6.2 is discussed how the raw data is
processedtousableresults.Anyissuesthatwereencounteredarediscussedandasolutionispresented.Thedistributionof
the weld toe parameters and any differences between cast and rolled steels or different steel strengths are discussed in
paragraph6.3.
6.2 Testoutputandprocessinglasermeasurements
6.2.1 Testoutput
Thedirectoutputfromthelaserscanisatablecontainingthepositionofthelaserandthemeasureddistance.Forpost
processingthelaserpositionandmeasureddistanceareseparatedandthemeasureddataareleveledoutbysubtractinga
plane from the data in such a way that the parent plate has value 0. The matlab scripts to perform these actions, and all
coming processing actions are shown in Annex F. Now the first results can be plotted in 3D, as shown in figure 6.1 and
figure6.2.
figure6.13DoverviewofatheweldofplateV46beforeTIGdressing,nogridlinesareshowntokeeptheimageclear
54
figure6.2Zoomedinonfigure6.1,withgridlinesata0.05x0.5mm
2
grid
From the same datasets weld cross sections can be plotted (see figure 6.3), which can be used to determine the
geometryoftheweldtoe.
figure6.3CrosssectionoftheweldofplateV46.Axesareinmillimeters.
Fourcharacteristicvalueshavebeenchosentodescribetheweldtoe:
1. Weldtoeradius
2. Weldtoeangle
3. Weldheight
4. Undercut
Infigure6.4thesecharacteristicvaluesareshownonafictive,exaggeratedweldtoecrosssection.
55
figure6.4Possiblemeasuredweldprofile(exaggerated)anditsaccompanyingschematization
6.2.2 Determiningtheweldradius
Theradiusismeasuredbycreatingafinegridaroundanestimatedmidpointforthecirclewhichcoincideswiththenotch
root.Foreachgridpoint thedistancetoeachdatapointofthe weldgeometryiscalculatedoverauserspecifiedsection.
Thegridpointthatshowstheleastvariationinthesedistancesisthebestsuitablemidpointandtheaccompanyingweldtoe
radiusisdeterminedbyaveragingalldistances.ThescriptofthisprocessisshowninAnnexF.Somestepsoftheprocessare
showninfigure6.3andfigure6.5.
figure6.5Optimizedcirclefortheleftweldtoe(red)ofthesameweldcrosssectionasshowninfigure6.3.The
optimalmidpointismarked.Axesareinmillimeters
Unfortunately,theprocessisnotcompletelyautomatic.Theusermustspecifytheboundariesofthegridbyspecifyingan
expected circle midpoint and the grid width and height. If this is not done properly the optimized midpoint will lie at the
edgeofthegrid,becausetheactualoptimalmidpointliesoutsidethegird.Inthiscasetheusercanspecifyalargergridor
relocate the expected midpoint until the optimized midpoint lies at a satisfactory distance from the edge of the grid. The
usermustalsospecifyboundariesbetweenwhichthecirclemustbeoptimized.Infigure6.5theseboundariesarechosenas
5.25 millimeter and 5.75 millimeter. Of course these boundaries are estimates and could also be estimated at slightly
differentvalues.Infigure6.6aplotisshownwheretheoptimizationboundarieswherevariedtoacertainextent.
56
figure6.6Differentradiifordifferentoptimisationboundaries.Axesareinmillimeters.
In figure 6.6 different circles are shown as the result of different optimization boundaries. Unfortunately, there is no
absolutewayofdeterminingwhichcircleisthebestoptimization.Anobviousindicationofthebestcircleisthevariationin
thedistancesfromitsmidpointtothedatapoints:thecircleforwhichthetotalvariationistheleastfitstheweldgeometry
the best. However, because the number of data points is reduced when the optimization boundaries are narrowed, it is
easierfortheprogramtomakeabetteroptimization.Therefore,narroweroptimizationboundarieswillalwaysresultina
lower variation. From figure 6.6 can be concluded that rather large differences in weld toe radius can be found if the
optimizationboundariesarevaried.Theboundariesandtheiraccompanyingradiiasdrawninfigure6.6areshownintable
6.1.
Optimisationboundaries[mm] Radius[mm]
5.155.85 1.61
5.205.80 1.40
5.255.75 1.20
5.305.70 1.24
5.355.65 1.04
5.405.60 0.74
table6.1Differentdeterminedradiifordifferentoptimisationboundaries
From figure 6.6 cannot clearly be concluded which of the circles fits best. If a zoomed in version of figure 6.6 is
considered(figure6.7)itisclearthatthecirclewiththesmallestradius(0.74millimeter)istoopessimistic.Thecirclewith
thelargestradiusistoooptimistic,especiallyontherightsideofthefigure.Totheeye,thecircleswithradii1.20,1.24and
1.40millimeterseemtomatchbestwiththeweld.
57
figure6.7Zoomedinonfigure6.6.Axesareinmillimeters.
Anotherremarkmustbemadeonthesmalldimpleintheweldgeometrythatcanbeseenaround5.55onthehorizontal
axisoffigure6.7.Thiscouldalsoberegardedasaverysmallnotch,withaverysmallradius.Thesetypesofdimplesorother
irregularitiesthatcouldbeinterpretedasasmallnotchoccurinalargenumberofweldcrosssections,bothasweldedand
TIGdressed.Achoicemustbemadetoeitherusethelargeradiusorusethesmallradiiascharacteristicvalue.
Thefollowingsectionswillfirstaddressthepossiblesubjectivityofthedeterminationoftheradiiandwaystomakeitas
objectiveaspossible.Thentheissuesconcerningtheadditionalsmallerradiiwillbediscussed.
Choosingthecorrectradiusfortheweldtoe
Asdescribedabove,thedeterminingoftheweldtoeradiuscannotbedoneentirelyautomatic.Theuseroftheprogram
influences the outcome to a certain extent. To determine whether the found weld toe radius is acceptable, a number of
checkscanbemade:
1. Wherethefoundradiusintersectsormeetsthemeasuredweldprofile,thetangentofthecirclemustcoincide
withthetangentofthemeasuredweldprofile.
2. When the boundaries are varied a small amount, the found radii must not greatly differ from each other. For
example, the three radii in figure 6.7 which are deemed to fit rather well, are all determined with different
boundariesbutdonotdiffergreatly.However,iftheboundariesofthesmallestcircleareslightlyenlargedthe
radiusvariesgreatly.Inthiscase,thesmallestcircleisinfluencedtoomuchbythesmalldimpleinthemeasured
weldprofile.
3. Theoverallimagemustbesatisfactory.Thisisofcourseextremelysubjective,butisoneofthemostimportant
tools to judge a radius. For example: for the program it is very difficult to filter out the small dimple in the
middleoffigure6.7,butapersoncaneasilyseethat,ifthelargeradiushastobedetermined,thesmalldimple
canbeignored.
Togetanestimateoftheamountofsubjectivityinvolvedintheprocess,theanalysisprocesshasbeencarriedouttwice
foroneweldcast.Theresultsofthisareshownintable6.2.Itisclearthatthedifferencesinmeanandstandarddeviation
areverysmall.Thedifferencesinmeanvaluecanevenbeneglectedwhenthesizeofthestandarddeviationistakeninto
account.
Radius Angle
Measurement 1 2 1 2
Mean[mm] 1.26 1.28 25.64 24.98
StandardDeviation[mm] 1.12 1.02 5.40 5.07
table6.2Comparisonofmeasuredweldradiiandanglefortwodifferentmeasurementsofthesameweldprofile
Thistestdoesnotprovethattheweldprofilemeasurementsarefullyobjective,butcanbeusedasanindicationofthe
amountofvariationtobeexpectedintheweldtoeparametersbytheuserinputintheprogram.
58
Choosingbetweenlargeandsmallradius
Infigure6.7tworadiicanbeseen,alargeone,forwhichthecirclesaredrawn,andasmallerdimpleinthemiddleofthe
figure.Boththeseradiicouldbeusedtodeterminethecharacteristicradiusforthisweldtoeprofile.Thisproblemisnot
uniquefortheweldprofileshowninthefigure,butoccursregularly.
To solve the problem, experts on welding and fatigue at the faculty of mechanical engineering of Delft University of
Technologywereconsulted.Asuggestionwasmadethatthesmallradiusmightbetoosmalltohaveinfluenceonthecrack
initiationofthespecimen.InfracturemechanicsthisproblemisaddressedwiththeKitagawaTakahashidiagram(Janssen
et al., 2002). This diagram describes the influence of a surface imperfection on the fatigue limit of the specimen. On the
horizontalaxisthesizeoftheimperfection(a)isshown,whichinthiscaseisthedepthofthesmallradiusinrelationtoa
smoothlargerradius.Ontheverticalaxisthethresholdstressrangetoinitiateacrackinplainmaterialisshown(
th
).The
diagramisbuildofoutoftwolines,onedescribingthefatiguelimit,andonelinedependingonthethresholdvalueofK,as
used in fracture mechanics (see 3.4). In the example of figure 6.8 the threshold stress to initiate a crack lies around 350
N/mm
2
for small defects. When defects increase to a size greater than a certain value, the threshold stress will start to
decline.Asaruleofthumbwecanstatethatbelowtheintersectionpointofthetwodottedlines,l
0
,thethresholdstress
doesnotdeclinebyalargeamount.Thismeansthatifasmalldefect(thesmallradius)issmallerthanl
0
itwillnotinfluence
theminimalstresstoinitiateacrack,andwillthereforenotinfluencethefatiguelife.
figure6.8ExampleofaKitagawaTakahashidiagramforanodularcastiron(
e
=320MPaK
th
=6.2MPam)
(source:Janssenetal.,2002)
To determine whether the small radii have to be taken into account, first l
0
has to be determined for the different
materials.Whenboth
e
andK
th
areknown,thevalueofl
0
canbedetermined.Toquicklyassesstheinfluenceofthesmall
radii not all materials are considered, but a selection of two materials is made:S690 and S1100. It is especially important
that S1100 is considered because previous research (Pijpers, 2011) has shown that these specimens are very sensitive for
surfaceroughness.
For the endurable stress range in plain material (
th
=
E
) reference is made to experiments carried out on this material
(Pijpers, 2011). On plain specimens, with mill skin in place, tests were performed. For both materials a runout was found
belowacertainstress.Thisstresswillherebeusedas
E
.Thevaluesoftheusedfatiguelimitsaregivenintable6.3.
Material
E
[N/mm
2
]
S690 270
S1100 378
table6.3Endurablestressesonplainmaterial
Different authors have reported on the value of K
th
. The different recommendations are shown in table 6.4. A remark
mustbemadeontherecommendationoftheIIW(Hobbacher, 2007),inwhich forverysmallsurfacecracksit isreported
that the K
th
may be lower than specified in table 6.4. However, based on the experiences of all other authors, where no
suchremarkismade,avalueof63Nmm
3/2
isassumedtobesufficientlyconservative.
59
K
th
[Nmm
3/2
] author
101upto120 Wanhill,1991
102.2upto208.7 Romeijn,2006
240173R Gruney,1979
170214R63 Hobbacher,2007
table6.4DifferentestimatesofK
th
Withthesedatathevalueofl
0
canbedetermined.Foralimitednumberofweldcrosssectionsananalysisismadeofthe
depthofthesmallradiusifoneispresent.Theresultsofthis,andthecalculatedl
0
areshownintable6.5.Itisclearthat
mostradiiliearoundthevalueofl
0
(notethattheKitagawaTakahashidiagramuseslogscalesonbothaxes).Thesmallradii
maythereforehaveaninfluenceonthefatiguestrengthofthatlocalmaterial,butasthisinfluenceisrelativelysmallitcan
probablybeignored,especiallywhenitistakenintoaccountthatforK
th
valuescanoccurwhichsignificantlyincreasethe
valueofl
0
.
V11 V69
x[mm] toeradius[mm] depthof'smallradius'[mm] x[mm] toeradius[mm] depthof'smallradius'[mm]
30 4.5732 0.0255 30 0.9837 0.0086
60 0.1614 N/A 60 0.8090 0.0149
90 1.7606 0.0189 90 1.7725 0.0287
120 2.7126 0.0171 120 0.3471 N/A
150 0.9171 0.0086 150 0.5420 0.0056
180 1.3947 N/A 180 0.2387 N/A
210 2.2831 0.0088 210 0.5109 N/A
240 1.7988 0.0047 240 1.1260 N/A
270 1.0579 0.0066 270 1.7184 0.0227
300 0.6709 N/A 300 2.0000 0.0187
330 3.6710 0.0381 330 0.4925 N/A
Mean 1.9092 0.0160 Mean 0.9583 0.0165
Stdev 1.3260 0.0114 Stdev 0.6212 0.0086
l
0
0.0088 l
0
0.0173
table6.5Analysisofoccurringsmallradiiandtheirdepthinrelationtoa'smooth'curve
Furthermore, these smaller radii occur mostly at some distance into the weld, where the maximum stress peak is not
located,andthereforesuffersfromalowerstressrange.
As a final argument to choose the larger radius as characteristic, is the fact that the smaller radii occur both in the as
welded specimens and TIGdressed specimens, with comparable regularity. Therefore, the large radius is the changing
factorandthusinterestingtorecord.
6.2.3 Determiningtheweldtoeangle,weldheightandundercut
Forthedeterminationofthesetreevaluesaplatformofstraightparentmaterialisusedasbasevalue.Fromthisbasethe
angle,heightorundercutattheweldtoecanbedetermined.Tosavetime,themeasurementsonlyincorporatealimited
distance of parent material (a few millimeters). This sometimes leads to difficulties because undercuts cannot always be
distinguishedfromthestraightparentmaterial.Ifadditionalresearchisdoneonthistopic,awidermeasurementscopeis
highlyrecommended.About10millimeterofparentmaterial,measuredfromtheweldtoe,willprobablysuffice.Itmight
bepossibletoincreasetheacrosstheweldgridsizeatdistancesgreaterthanafewmillimetersfromtheweldtoe,tokeep
themeasuringtimewithinlimits.
6.3 Results
6.3.1 Observedgeometries
Theanalysisofweldgeometrieshasresultedinaround750radii,anglesandundercutsandabout375weldheightsfor
boththeasweldedconditionandtheTIGdressedcondition.Theasweldedgeometriesalllookedverysimilar,buttheTIG
dressed geometries showed four characteristic shapes, which all appeared in different regularity. The global shape of the
observedweldtoeshapesforboththeasweldedandTIGdressedconditionareshowninfigure6.9.
60
figure6.9Observedweldtoeshapes(fictivegeometries,forcommunicationpurposes)
TypicalAswelded.Theasweldedspecimensgenerallyhave,totheeye,thesamegeometry.Theparentmaterial
shows no great undercuts. Then the weld material starts after a relatively small radius. The angle of the weld
materialisrathersteep.
TypicalTIGdressed,Type1.ThisisthedesiredeffectofTIGdressing(see4.3.3,especiallyfigure4.9).Theoverall
weldgeometryismuchsmoother.Theweldradiusismuchlargerthanintheasweldedstate,andtheweldtoe
angleisreduced.
Typical TIGdressed, Type 2. A significant undercut is visible, even to the naked eye on the weld itself. The
undercuthasdepthsuptoabout0.5millimeter.TheweldtoeradiusisgenerallylargerthanobservedinType1,
andtheweldtoeanglesomewhatsteeperthanforType1.Type2doesoccur,butnotveryfrequently,andalso
notineveryweld.
TypicalTIGdressed,Type3.TotheeyethislooksverysimilartoType1.However,asmallridgecanbefeltwitha
sharpobjectorthefingernail.Itseemstheweldhassolidifiedbeforeanentirelysmoothgeometrywasformed.In
the measurements, the radius and angle of the first, and in all cases smallest, radius was measured. The weld
angleofthistypeissimilartotheweldangleofType1.Type3occursratherfrequently,andcanbefoundinmost,
ifnotallstudiedwelds.
TypicalTIGdressed,Type4.Asmallweldtoeradiusfollowedbyarathersteepangle(similartoaswelded)fora
very short distance. After this, the angle declines back to similar values as found in Type 1 and Type 3. Type 4
occursfarlessfrequentthanType1andType3.
Itwouldbeveryinterestingtomakeaninventoryoftheexactfrequencyofthedifferenttypes,whichcouldbematched
to fabricating conditions (experience of the welder, heat input, etc.) and to the point of crack initiation. Furthermore, it
would be very interesting to investigate the existence and frequency of these typical geometries in TIGdressed welds in
practice. The welds studied in this research were all TIGdressed by the same welder, under the same conditions on the
same day. Furthermore, all considered weld cross sections are taken from a very limited number of welds, which could
influencetheresults.
6.3.2 Comparisonbetweenrolledandcaststeel
AlldatahavebeengatheredandcanbesplitinTIGdressedandasweldedorrolledandcaststeels.Itisvisibletotheeye
that TIGdressing has had great influence on the weld geometry. Whether there is a clear difference between rolled and
caststeelshastoresultfromthedatafromtheweldgeometrymeasurements.Onlytheparametersweldtoeradius,weld
toe angle and undercut will be compared, because the weld height is deemed independent from the parent material,
especiallysincecastplatesarealwaysweldedtoarolledplateinthisresearch.
61
Weldtoeradii
Bardiagramsofthedensityofmeasuredradiiforbothrolledandcaststeelareshowninfigure6.10.Anoverlayofthese
twodiagramsisshowninfigure6.11.Itisclearthatthecaststeelradiishowasharperpeak,incontrastwiththesmoother
diagram shown by the rolled steel data. The location of the peaks of both diagrams shows no large difference, especially
whenthespreadofthedataistakenintoaccount.
figure6.10Comparisonofasweldedradiibetweenrolledsteelandcaststeel[mm]
figure6.11Overlayofasweldedradii;cast()androlled()steel[mm]
62
DensitybardiagramsforTIGdressedradiiaredepictedinfigure6.12withanoverlayinfigure6.13.Theradiioftherolled
steelshowadensitypeakroughlyatthesamelocationasthecaststeelandanadditionalpeakofsmallerradii.Thiscanbe
asignificantdifference,butcanalsoarisefromthefactthatalldataresultfromalimitednumberofwelds.Oneweldbead
thathasalowerquality,caneasilyinfluencetheresultstoanextentthatthedifferencesbetweenthetwofigurescanbe
explained.Asidefromtheadditionalpeaktheglobalshapeofthediagramsappearsroughlythesame.
figure6.12ComparisonofTIGdressedradiibetweenrolledsteelandcaststeel[mm]
figure6.13OverlayofTIGdressedradiiofcast()androlled()steel[mm]
63
Weldtoeangles
In figure 6.14 density bar diagrams are shown for the weld toe angles of as welded specimens. An overlay of both
diagramsisshowninfigure6.15.Itisclearthatthediagramsareverysimilar.
figure6.14Comparisonofasweldedanglesbetweenrolledsteelandcaststeel
figure6.15Overlayofasweldedanglesofcast()androlled()steel
64
Infigure6.16densitybardiagramsareshownforweldtoeanglesofTIGdressedspecimens.Anoverlayisshowninfigure
6.17. The peak of both data samples differs, but when the spread of the data is taken into account, it is questionable
whetherthisisasignificantdifference.
figure6.16ComparisonofTIGdressedanglesbetweenrolledsteelandcaststeel
figure6.17OverlayofTIGdressedanglesofcast()androlled()steel
65
Undercuts
Density bar diagrams of the undercut of as welded specimens are shown in figure 6.18. An overlay of both diagrams is
showninfigure6.19.Itisclearthatthediagramslookverysimilar,withmostundercutsbeingzero.Itmustbenotedthat
both bar diagrams are cropped at 0.22 millimeter, while both the rolled steel and cast steel specimens shown a sporadic
datapointatdeeperundercuts.Forclarityofthecomparisonthediagramsarecropped.
figure6.18Comparisonofasweldedundercutsbetweenrolledsteelandcaststeel[mm]
figure6.19Overlayofasweldedundercutsofcast()androlled()steel[mm]
66
The density bar diagrams of the undercuts of TIGdressed specimens are shown in figure 6.20. An overlay of both
diagramsisshowninfigure6.21.Forsmallundercutsbothdiagramsshowsimilarbehaviour,butfordeeperundercuts,the
cast steel specimens show much more undercuts. This can attributed to the fact that the large majority of typical TIG
dressedtype2weldgeometries(seefigure6.9)occurredinoneweldintheC69TIGplateonthecastside.
figure6.20ComparisonofTIGdressedundercutsbetweenrolledsteelandcaststeel[mm]
figure6.21OverlayofTIGdressedundercutsofcast()androlled()steel[mm]
Conclusion
The overlay diagrams of the weld toe radii, weld toe angles and undercuts show different results. In some cases the
resultsmatchalmostexactly,andinsomecasesthecastandrolledsteelsshowratherdifferentresults.Beforeaconclusion
canbedrawn,afewfactorsmustbeconsidered.
Firstofall,thedatasetforcaststeelsismuchsmallerthanfortherolledsteels,sinceallcaststeelplatesareweldedtoa
rolledsteelplate.Thisresultsinaroughlythreetimeslargerdatasetfortherolledspecimens.Thiscouldexplainsomeof
thesharppeaksthecaststeeldatashow(e.g.seefigure6.11).Furthermoreitmustbenotedthatalldataresultfromonly
nine welds in total, which in turn may be built up from multiple weld beads. This means that in total only eighteen weld
beadshavebeenmeasured(oneforeachweldtoe).Ifoneofthoseweldbeadshasbeenfabricatedinalessthanperfect
way, this can very easily translate into large differences in the above analysis. For example, the large majority of large
undercutsthatarevisibleincaststeelappearintheC69TIGplate.
Based on these considerations and the fact that the differences between peaks generally do not exceed the standard
deviation,itisassumedthattheweldsincaststeelandrolledsteelhavesimilardistributionsofgeometry.
67
6.3.3 Comparisonbetweendifferentsteelgrades
The data sets of the individual measured plates consist of relatively few data points. This will result in very rough bar
diagrams.Thiseffectwasalreadynotedin6.3.2wheretherolledandcastmeasurementswerecompared.Inthiscasethere
is not a difference in data size, but the small size of the data set severely distorts the overlay of two bar diagrams.
Therefore, in this paragraph only mean values and standard deviations will be compared. The results for weld height are
nottakenintoaccount,becausethisdiffersgreatlybetweenplates,andmainlydependsontheskillofthewelderandthe
amountofaddedweldmaterial.Intable6.6totable6.8thecomparisonbetweenthedifferentmaterialscanbemade,both
for the as welded as TIGdressed situation. For the weld toe radius and weld toe angle hardly any differences can be
observed, and any differences are small when compared with the standard deviation. For the undercut in the as welded
situation there seems to be a material dependent influence, but as this influence is not apparent in the TIGdressed
situationitisconcludedthatthesteelgradehasnoinfluenceontheweldtoeparameters.
Aswelded TIGdressed
mean[mm] stdev[mm] mean[mm] stdev[mm]
S1100 1.4188 1.0184 4.5360 2.9817
S890 1.5691 0.9458 4.9847 4.4375
S690 1.5210 1.1274 4.5058 3.1078
S460 1.8761 1.6850 3.6718 3.2851
table6.6Comparisonofmeanvalueandstandarddeviationofweldtoeradii.Thegivenvaluesarealsoapplicableto
thecaststeelcounterpartofthementionedrolledsteels.
Aswelded TIGdressed
mean[] stdev[] mean[] stdev[]
S1100 29.0235 5.9754 14.5088 4.2531
S890 28.3556 5.5604 14.1985 4.8029
S690 27.9500 6.2554 17.0907 6.0301
S460 26.0104 7.9736 15.7443 3.7348
table6.7Comparisonofmeanvalueandstandarddeviationofweldtoeangles.Thegivenvaluesarealsoapplicable
tothecaststeelcounterpartofthementionedrolledsteels.
Aswelded TIGdressed
mean[mm] stdev[mm] mean[mm] stdev[mm]
S1100 0.0217 0.0534 0.0231 0.0369
S890 0.0176 0.0352 0.0161 0.0314
S690 0.0077 0.0120 0.0368 0.0850
S460 0.0067 0.0102 0.0142 0.0151
table6.8Comparisonofmeanvalueandstandarddeviationofundercuts.Thegivenvaluesarealsoapplicabletothe
caststeelcounterpartofthementionedrolledsteels.
6.3.4 Distributionofweldgeometryparameters
Based on the previous paragraphs, the data for cast and rolled steel or different steel grades will not be analyzed
separately.Inthisparagraphcollecteddataofallweldgeometryparameterswillbeplotted.
Weldtoeradius
Infigure6.22theresultsofthelasermeasurementsareshownforasweldedandTIGdressedspecimens.Anoverlayof
both diagrams is shown in figure 6.23. The overall trend is what is to be expected from TIGdressing: the weld toe radius
increases.However,verysmallradiistilloccurandtheirnumbersarestillsignificant.Thiswillbeduetotheoccurrenceof
typicalTIGdressedgeometriestype3and4(seefigure6.9).ThespreadoftheTIGdressedradiiismuchlargerthanforthe
asweldedradiiwiththeoccurrenceoftheaforementionedsmallradii,butalsoverylargeradiioccur.Atabulatedsummary
oftheresultsisgivenintable6.9.Notethatthevaluesoftheweldtoeradiiarenotnormallydistributed.Thevaluesintable
6.9shouldthereforenotbeusedassuch.
68
figure6.22ComparisonofasweldedandTIGdressedradii[mm]
figure6.23Overlayradiiofaswelded()andTIGdressed()specimens[mm]
Mean[mm] Standarddev.[mm]
Aswelded 1.5940 1.2256
TIGdressed 4.6807 3.5462
table6.9Meanandstandarddeviationofmeasuredradii
69
Weldtoeangle
both diagrams is shown in figure 6.25. The trend is exactly as expected: the average angle decreases. Of course a few
steeperanglesoccurintheTIGdressedsituation,butnotwithafrequencythatisdisproportionatewiththeoverallshape
ofthedistribution.Thespreadofbothdataisinthesameorderofmagnitude.Atabulatedsummaryoftheresultsisgiven
intable6.10.
figure6.24ComparisonofasweldedandTIGdressedangles[mm]
figure6.25Overlayanglesofaswelded()andTIGdressed()specimens[mm]
Aswelded 27.8480 6.5577
table6.10Meanandstandarddeviationofmeasuredangles
70
Undercut
both diagrams is shown in figure 6.27. The diagrams show a slight increase in non zero undercuts for TIGdressed
specimens,butbothTIGdressedandasweldedspecimenshavehardlyanyundercutslargerthan0.050.1millimeter.The
TIGdressedspecimenscontainafewdeeperundercutsasdescribedabove.Atabulatedsummaryoftheresultsisgivenin
table6.11.
figure6.26ComparisonofasweldedandTIGdressedundercuts[mm]
figure6.27anglesofaswelded()andTIGdressed()specimens[mm]
Aswelded 0.0136 0.0335
table6.11Meanandstandarddeviationofmeasuredundercuts
71
Weldheight
Because TIGdressing takes place at the weld toe, the weld height is assumed not be influenced to a great extent.
Therefore,onlyasweldeddatawillbeshownindiagrams.Acomparisonofmeanandstandarddeviationoftheweldheight
data between as welded and TIGdressed specimens is shown in table 6.12. The results in this table strongly support the
assumptionthattheweldheightisnotinfluencedbyTIGdressing.
Mean[mm] StandardDeviation[mm]
Aswelded TIGdressed Aswelded TIGdressed
C11 3.9384 3.9188 0.4697 0.4557
C89 4.1151 4.1485 0.2928 0.2747
C89a 4.3455 4.3803 0.2212 0.1790
C69 2.3018 2.2875 0.1935 0.1943
C46 1.7768 1.7068 0.1662 0.1107
V11 3.6734 3.7067 0.2278 0.2193
V89 3.0152 3.0438 0.2569 0.2421
V69 2.4594 2.4965 0.1445 0.1342
V46 2.1462 2.1428 0.1875 0.2023
table6.12ComparisonofmeanandstandarddeviationofweldheightsforasweldedandTIGdressedspecimens
Fromthegivenvaluesforthemeanandstandarddeviationintable6.12canbedeductedthattheweldheightcanvary
verysignificantlybetweenwelds,butshowsverylimitedvariationalongoneweld.Therefore,infigure6.28andfigure6.29
thedistributionofweldheightsisspecifiedforeachplateseparately.
figure6.28Distributionofweldheights.Castrolledplates[mm]
72
figure6.29Distributionofweldheights;Rolledrolledplates[mm]
6.3.5 EvaluationofinfluenceofTIGdressing
TheintentionandexpectedresultofTIGdressingistheincreaseoftheweldtoeradiusand,forbuttwelds,thereduction
oftheweldtoeangle.Boththeseeffectsareclearlyvisibleinthemeasuredweldgeometries.
The mean value and standard deviation that are found for the weld toe radius before and after weld improvement
comply reasonably with earlier research (see figure 4.11) with the remark that the measured radius for TIGdressed
specimens issomewhat lower than is shown inearlier research. However, if the actual distribution of as welded and TIG
dressedspecimensisobserveditisclearthat,althoughtheaverageradiusincreasessignificantly,alargeamountofsmall
radiistilloccursintheTIGdressedspecimens.Thefittingofastatisticaldistributionhasbeenomitted.Whiletheaswelded
datacomplyreasonablywellwithalognormaldistribution,theTIGdresseddatashowsanadditionalpeaknearthesmaller
radiiwhichmakesaccuratemathematicaldescriptiondifficult.
TheaverageweldtoeanglehasdecreasedsignificantlyandshowssmallerspreadafterTIGdressing.Thedatasetsbefore
and after TIGdressing show a distribution shape similar to a normal distribution. The distribution of undercuts after TIG
dressingshowssimilarbehaviourasbeforethetreatment,withtheexceptionofanincreaseofverydeepundercuts(typical
TIGdressedgeometrytype2,seefigure6.9).Thelargemajorityofundercutsliesbelow0.05millimetersdepth.Theweld
heightofeachseparatewelddoesnotseemtobeinfluencedbyTIGdressing.Thespreadofweldheightsovereachweldis
verysmall,butweldheightsofdifferentweldscanshowsignificantspread.
Whenfatigueisconsideredtobeaweakestlinkprocess,itisquestionablewhethertheseresultsjustifyafatiguestrength
increase due to changed geometry. The average weld geometry shows significant improvement with a larger weld toe
radiusandsmallerweldtoeangle,butintheworstcasescenariothismightnotbethecase.Especiallytheweldtoeradii
showasignificantpeaknearthesmallerradii.
A few remarks must be made concerning this conclusion. First of all, although the worst case radius does not seem to
show improvement, it is very likely combined with a smaller weld toe angle, which reduces the stress concentration near
the weld toe anyhow. Secondly, the measured smaller radii in the TIGdressed specimens mainly occur in typical TIG
dressedgeometriestype3and4(seefigure6.9).Intype3thestressconcentrationisexpectedtobelowerduetotheweld
plateaudirectlyaftertheweldtoe.Type4isratherrare,andaworstcasescenariowithacombinationofgeometrytype4
andotherparameterssuchasalargeinclusionandhighresidualstresses,whichmakesitcomparablewiththeaswelded
weld,isnotverylikely.
Ifadatabasewouldbemadeofthedifferentoccurringgeometries(seefigure6.9),thedependenciesbetweentheweld
toe parameters, which are very likely present, can be investigated and documented. This would make an analysis of the
worstcasescenariopossible.Thiswillbeelaboratedin8.3.2.
73
7 Processingandresultsoffatiguetestsandhardnessmeasurements
7.1 Chapteroutline
Inthischaptertheresultsofthefatiguetestsarediscussed.Theoutputofthetestingriganddeterminationofthevalues
ofN
i
,Rand
n
isdiscussedthefirsttwoparagraphs.Inparagraphs7.3,7.4and7.5theresultsofthefatiguetests,hardness
measurementsandcrackmonitoringarepresented.
7.2 Testoutputandprocessing
7.2.1 Testoutput
The fatigue tests result in data from the strain gauges, coupled to the number of cycles and notes made during the
experimentwhichcontaincracklengthatvariousstagesduringthecrackpropagationlifeandnumberofcyclesatfailure.
Whenthespecimenhasfaileditcanbeexaminedforbeachmarks,createdbythecrackmarkingprocedureasdescribedin
5.3.2. An example of the acquired strain gauge data is shown in figure 7.1, and an example of the acquired crack
propagationdataisshowninfigure7.2.
0
500
1000
1500
2000
0.E+00 1.E+05 2.E+05 3.E+05 4.E+05 5.E+05 6.E+05 7.E+05 8.E+05 9.E+05 1.E+06
N
[10
6
]
figure7.1StraingaugedatafromspecimenC692.Thecrackmarkingsareclearlyvisibleintheformofalowered
strainrange.
0
10
20
30
40
50
5.4E+05 5.6E+05 5.8E+05 6.0E+05 6.2E+05 6.4E+05 6.6E+05 6.8E+05
N
Crackdim.
[mm]
Width
[CM]Depth
[CM]Width
figure7.2CrackpropagationdatafromspecimenC692.Mostdatapointsaredirectlymeasuredduringthetest.The
crackmarkingdataarerarerduetothecomplexityoftheprocedure,butfitinwellwiththeotherdata.
74
7.2.2 DeterminingN
i
,nominalstressandstressratio
The test rig is programmed to shut down when the measured strain range deviates more than 5% from the average
measuredvalue.Whenthetestrigshutsdowntostartcrackmonitoring(see5.3.2),thecracksizeisfairlysubstantialmost
ofthetime.Forexample,thecracksizeatthefirstmeasurementinfigure7.2isalready14millimeter.BecauseTIGdressing
is expected to especially increase the crack initiation life of the specimens, the number of cycles at crack initiation (N
i
)
needs to be determined. This point is determined with the aid of the strain gauge data. The first point where one of the
straingaugelinesstartstodeflectfromitsaveragestraightlineismarkedasthecrackinitiationpoint.Incollaborationwith
TNOanalternatemethodtodetectthecrackinitiationsiteanddeterminethenumberofcyclestocrackinitiationhasbeen
experimented with. Instead of the strain gauge data, acoustic emission of the crack was used. The procedure and
experiencesareaddedinAnnexG.
Afterfailureofthespecimenthecracksurfacecanbeinspectedmorethoroughlytofindthecrackinitiationsite.Thisis
usuallydarkerduetothelongerexposuretopetroleum(see5.3.2,crackmonitoring).Alsothepresenceofbeachmarksdue
tocrackmarkingcanhelpidentifythecrackinitiationsite.Whenthelocationofcrackinitiationisdetermined,thenominal
stress is determined on the basis of measurements of the nearest strain gauge(s). In the tabulated fatigue test results in
AnnexCtheusedstraingaugesarespecified.AnnexBshowsthelocationofthenumberedstraingauges.
In the tabulated fatigue test results, two different values of the stress ratio R are specified. One is based on the strain
gauge(s)whichwerealsousedtodeterminethenominalstress.Theotherstressratioisbasedonthemeasuredvaluesof
theforce.
7.3 Resultsoffatiguetests
Inthisparagraphonlytheresultsofthefatiguetestswillbeshownwithoutanyadjustments(see9.2.2).Theresultswill
be split according to the static strength of the specimens. Cast and rolled results will be plotted in one graph. Based on
certainconsiderations,someresultswillbeexcludedfromtheanalysislater.Inthefiguresbelowallresultsareplotted.The
exclusionofdatawillbeelaboratedonin9.2.1.
Threedifferentfailuremechanismshavebeenobserved:
Basematerialfailure(indicatedwithBMinthegraphsandAnnexC)
Weldtoefailure(indicatedwithWTinAnnexC)
Weldmaterialfailure(indicatedwithWMinAnnexC)
The first two mechanisms were also observed in the as welded tests by Pijpers (2011). The weld material failure
mechanism was only very rarely observed in his quite substantial data set, while in the much smaller data set of this
research seven of such failures are observed, which accounts for roughly 30% of all results. A total of five failures were
foundinthebasematerialandonlyintheS890andS1100specimens.Severalspecimenshavenotfailedandwerestopped
atacertainnumberofcycles.Theserunoutsaremarkedinallgraphs.
The total failure of specimens is marked as defined as the number of cycles to failure: N
f
. The point where the first
deviationofastrainrangefromitsaveragecanbenoticedisdefinedasthenumberofcyclestocrackinitiation:N
i
.
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46
C46
runout
*
BMfailure
figure7.3ResultsoftestsonS460specimens.
75
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V69
C69
runout
*
BMfailure
figure7.4ResultsoftestsonS690specimens
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89
C89
runout
*
BMfailure
*
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11
C11
runout
*
BMfailure
76
7.4 Resultsofhardnessmeasurements
The results of the hardness measurements are shown in Annex H, in this paragraph the results will be discussed. The
general trend is that TIGdressing increases the hardness of the material, both where the TIGdressing has melted the
material and where the material has been influenced by the heat of TIGdressing. Exceptions to this trend are the
specimensC89,C11andV11,allofwhichareintheveryhighstrengthrange.Inthesecaseslittletonohardnessincrease
hasbeenobserved.Insomeexceptionalcases,thehardnesshasbeenloweredbyTIGdressingonthesespecimens.
The extensive study of these results lies out of the scope of this study, but the results can lead to some interesting
possibleconclusions.Thematerialhardnesscanbelinkedtotheultimatestrengthofthesteelaccordingtoequation(7.1)
(VanWortel,2006).
3.02 10
m
R HV = (7.1)
Ifthenotchstressanalysisisused,theultimatestrengthisusedindifferentstepsofthecalculationanddirectlylinkedto
the fatigue strength. The material constant a*, proposed by Lawrence (Radaj et al., 2006) (see 3.5.1 and equation (3.9))
depends on the ultimate strength. If the ultimate strength increases, a* decreases, which results in a higher notch
sensitivityofthematerial(seefigure3.10).
Differentauthorshavereportedontherelationbetweenthestaticstrengthandfatiguestrength(seefigure2.11).These
relationsgenerallytaketheformofalinearequation,whereincreasingstaticstrengthleadstoincreasingfatiguestrength.
Thiseffectwillbestrongerthantheeffectofa*(seefigure2.10,whereafactor2.5inR
m
leadstoafactorofabout1.3in
K
f
).
Ifboththeserelationscanbeproventobevalidintheconditionsthatareconsidered,theincreasedhardnesscausedby
TIGdressingmayincreasethefatiguestrengthoftheweld,evenwhenthechangedweldgeometry,residualstressesand
reduceddefectsareabsent.
EarlierresearchbyPedersenetal.(2010)hasshowntheoppositeoftheresultsfromthisresearch:thereahardnessdrop
of15to20%wasshown(see4.3.3).TheexacteffectofTIGdressingonthehardnessonthematerialandtheconsequences
oftheseeffectsthereforedeservesamorethoroughresearch.
7.5 Resultsofcrackmonitoring
Onallspecimenscrackmonitoringhastakenplace.However,insomecasesfewerdatapointsareavailablebecausethe
crack had been detected in later stadium, or the beach marks which should be visible due to crack marking could not be
found.AllusableresultsarecollectedinAnnexI.Theresultscanbeusedtoverifythenumberofcyclestocrackinitiation,
ortodeterminematerialparametersneededforacrackpropagationanalysis.Itcouldbeinvestigatedifthecracksgrowina
differentrateafterTIGdressing,andtowhatextentthiscanbeexplainedbythechangedgeometryonly.Thisresearchwill
omitsuchcalculations,becausethefocusliesontheextensionofthecrackinitiationlife.
77
8 Analyticaldeterminationoffatiguestrength
8.1 Introductionandchapteroutline
Aspreviouslydiscussed,TIGdressinghasanumberofpossibleinfluencesontheweldtoe:
Geometrychange
Reductionofimperfections
Changeinresidualstressstate
Changeofmetallurgicalstate
Thelasermeasurementsoftheweldgeometry(see5.3.1)resultinextensivedataconcerningthefirstparameter.Surface
imperfections can possibly also be traced to a certain extent with the laser measurements, but this has not been done in
thisresearchbecausedeeperimperfections(inclusions)cannotbetracedinthismanner.Nondestructivetestingcouldgive
insightintothechangeinnumberandsizeoftheseimperfections,buttolimitthedurationoftheresearchthesetestshave
notbeencarriedout.Thehardnessmeasurements(see7.4)giveinsightintothechangeofmetallurgicalstateoftheweld
toe.Thechangeinresidualstressshallbeestimated,basedonpreviousresearch.
In this chapter the fatigue strength of TIGdressed specimens will be predicted based on the measured or estimated
parametersstatedabove.Becausemoststandardfatigueassessmentsincorporatetheweldgeometryandimperfectionsin
the scatter, these methods for prediction cannot be used. The notch stress method (see 3.5) uses the actual geometry,
stress range, residual stress state and metallurgical state at the crack initiation site. This method is therefore suitable to
comparethestateofthespecimensbeforeandafterTIGdressingandpredictthefatiguestrength.
This chapter will first discuss the determination of the stress concentration factor and fatigue notch factor. Then other
influencefactorswillfollow,leadingtoapredictionofthefatiguestrengthoftheconsideredspecimens.
8.2 Factorsdeterminingfatiguestrength
Accordingtothenotchstressapproach,thecrackinitiationlifeofweldedjointscanbecomparedwiththecrackinitiation
lifeofplainmaterial,ifthecorrectionsaremadeforthestressconcentrationeffectofnotches.However,asmentionedin
2.3.2 there are other factors influencing fatigue life. All these factors will be considered to determine the expected crack
initiationlife:
Fatiguenotchfactor:K
f
(dependsontheelasticstressconcentrationfactor:K
t
)
Meanstressfactor:f
m
Thicknessfactor:f
t;w;Ni
(onlyapplicabletocrackinitiationlife)
Loadingmodefactor:f
lm;w;Ni
(onlyapplicabletocrackinitiationlife)
In the coming paragraphs each influence factor will be determined. Furthermore, the fatigue strength of the base
materialneedstobedetermined,towhichisreferredinthenotchstressanalysis.Differentrelationsbetweenthefatigue
endurancelimitandthestaticmaterialstrengthhavebeendeveloped.Pijpers(2011)hascomparedseveralformulas,and
found equation (8.1), as proposed by Haibach 2006, to be most applicable because of the distinction between rolled and
caststeel.
;0 aE mat u
f o o = (8.1)
Inwhich:
f
mat
materialfactorforeitherrolledsteel(0.45)orcaststeel(0.34)[]
Inequation(8.1), theusedvaluefortheultimatestrengthisthevaluespecifiedinthematerialcertificatesandhasnot
been linked to the hardness measurements. When the fatigue limit of the base material is corrected for the influences
describedabove,thefatiguestrengthcurvecanbedeterminedusingtheapproachbySchijveorHckasdescribedin3.5.1.
8.3 Determinationofstressconcentrationfactorandfatiguenotchfactor
Forbuttweldswithoutundercut,theelasticstressconcentrationfactorshavebeenstudiedbefore(Anthesetal.,1993)
andhavelettoaformulatodeterminetheelasticstressconcentrationfactorattheweldtoe.
78
3
2 1 2
2 3 sin
180
1 0 1 2 3
1 1 sin sin sin
180 180 180
l
b l l
t
t t
K b a a a a
r r
u
t
u u u
t t t
+ | |
+
|
\ .
(
( | |
| | | | | | | | | |
(
= + + + + + | ( | | | | |
|
(
\ . \ . \ . \ . \ . (
\ .

(8.2)
Inwhich:
t theplatethickness[mm]
r theweldtoeradius[mm]
theweldtoeangle[];degrees
Otherparametersarespecifiedinthetablebelow:
Factor a
0
a
1
a
2
a
3
b
1
b
2
l
1
l
2
l
3
Tension 0.169 1.503 1.968 0.713 0.138 0.2131 0.2491 0.3556 6.1937
Bending 0.181 1.207 1.737 0.689 0.156 0.207 0.2919 0.3419 3.283
table8.1Influencefactorsforequation(8.2)
Unfortunately,thisformuladoesnotcoverweldswithundercuts.Whiletheinfluenceofsmallundercuts,suchasfound
in most geometries, is unknown, the influence of the large undercut of Typical TIGdressed type 2 (see figure 6.9) will
certainly be significant. Furthermore, the weld height is not incorporated in the formula, as it is assumed by Anthes that
weld height and weld angle are directly related. For the observed weld toe geometries after TIGdressing (see figure 6.9)
this is not the case, because the TIGdressing process has changed the angle but not the weld height. To obtain stress
concentrationfactorsforthespecificgeometrieswhichwerefound,aFEMmodelofthelocalweldgeometryismade.
8.3.1 FEManalysisofweldtoe
Aparametricmodelhasbeensetupinwhichthefollowingparameterscanbeadjusted:
Length
Thickness
Weldtoeradius
Weldtoeangle
Undercut
Weldheight
Because the weld is relatively long compared to the dimensions of the weld toe notch, a 2D model consisting of plane
strain elements is chosen. An example of a possible weld geometry and the accompanying mesh are shown in figure 8.1.
Onlyonesideoftheweldismodeled,andtheassumptionismadethatthiswillnotinfluencetheresult.Thiswillbeshown
tobeplausiblelateron.
figure8.1ExampleofFEMmodel
Themainconcernwhenmodelingtheweldtoenotchisthesizeoftheelementsneartheexpectedlocationofthestress
peak. IIW recommendations (Hobbacher, 2007) state that for linear elements the size of the elements should not exceed
1/6oftheweldtoeradius,or1/4oftheweldtoeradiusforhigherorderelements.Todetermineanacceptablemeshsize,
four geometries were analyzed three times, with radius to element size ratios of approximately 5, 10 and 15. In the
79
diagrams below the results of this investigation are shown. Each diagram shows three lines for the different radius to
element size ratios. The lines show the value of the largest principal stress along the arc of the weld toe. In the
accompanyinglegendsthenumberofelementsinthecurveoftheweldtoeisalsostated.
100
110
120
130
140
150
160
170
180
190
0.4 0.1 0.2 0.5 0.8 1.1 1.4
Arclengthfromstartofradius[mm]
[N/mm
2
]
R/E=14.32;5el.incurve
figure8.2Principalstressalongnotchradius.=2mm,=20,noundercut
160
180
200
220
240
260
280
0.1 0 0.1 0.2 0.3 0.4
[N/mm
2
]
figure8.3Principalstressalongnotchradius.=0.5mm,=30,noundercut
80
100
120
140
160
180
200
0.4 0.1 0.2 0.5 0.8 1.1 1.4 1.7
[N/mm
2
]
figure8.4Principalstressalongnotchradius.=2mm,=20,Undercut=R/25
80
0
50
100
150
200
250
300
350
0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
[N/mm
2
]
figure8.5Principalstressalongnotchradius.=0.5mm,=30,Undercut=R/5
Fromfigure8.2tillfigure8.5canbeconcludedthattheratiobetweenelementsizeandweldtoeradius(R/E)aloneisnot
enough to judge the element size, because, although the three depicted R/E ratios are the same for each diagram,
significantdifferencesinsmoothnessofthestressdiagramcanbeobserved.Theotherparameterwhichisspecifiedinthe
diagrams,thenumberofelementsinthecurve,providesabetterguidelinetothecorrectnessoftheelementsize.Asarule
ofthumbthemaximumelementsizewillbesetatR/E>10andaminimumof8elementsinthenotchradius.Aremarkmust
bemadethattheheightofthestresspeakdiffersnomorethan1.2%inthecasesconsidered.TheIIWguidelinetherefore
seems to be applicable, but a smoother stress diagram gives more faith in the results, especially when in the coming
analyses no mesh refinement is made. The choice for a possibly too fine mesh can be easily made, because the total
calculationtimeofthemodeldoesnotexceedoneminute.Giventhefactthatfurthermeshrefinementdoesnotresultin
significantly different values and that the stress pattern is smooth, it can be concluded that the FEM model performs
satisfactory.
Inanoverviewoftheresultsofthemodel(seefigure8.6)itisclearthatforallapplicablegeometriestheassumptionthat
the other side of the weld does not have to be incorporated is reasonable, because after a very short distance the stress
patternrecoversfromthestresspeakneartheweldtoenotchcomparedwiththegenerallylongwidth(>35millimeter)of
theweld.
figure8.6Stresspatterninhorizontaldirectionpresentedascontourlevels.=2, =20, no undercut

The developed model can be compared with the parametric formula by Anthes to see whether they give comparable
resultsforasetofrealisticparameters.Acomparisonforaweldprofilewithaheightof3millimeter,aweldangleof28,a
platethicknessof25millimeterandvaryingweldtoeradiusismadeinfigure8.7.Obviously,noundercutispresent,asthe
formula does not cover undercuts. Based on the comparison it can be assumed that the model and the formula give
reasonably similar outcomes. The stress concentrations differ by about 5%, and a little more for radii larger than
approximately4millimeter.This5%differencebetweentheparametricformulaandactualresultswasalreadyspecifiedby
Antheswhenpublishinghisparametricformula.SomeminordifferencescouldalsobeattributedtoadifferenceinPoisson
factor,whichisnotspecifiedbyAnthes,orthedifferenceinanalysis(BoundaryElementMethodbyAnthesvs.planestrain
FEM in this research). Most importantly it must be remarked that Anthes directly couples the weld toe angle and the
weldheight,whereasintheFEMmodeltheycanbegivenvaluesindependently.Themaximumdifferenceliesaround5%,
whichisinlinewiththeexpectedpossibledeviationspecifiedbyAnthes.However,ifanonregularweldismodeled,such
asaTIGdressedweldwithasmalleranglebutthesameweldheightastheasweldedspecimen,Anthesformulapredicts
thestressconcentrationfactorevenbetter.Thisisremarkablebecausethefixedweldangleweldheightrelationusedby
Anthes implies that for one of the analyzed situations the weld height used in the formula and the model should differ
significantly. If the weld height has any influence on the elastic stress concentration factor, this should become clear by
81
means of a great difference between the formula by Anthes and the FEM results for one of both analyzed geometries. A
morethoroughinvestigationontheinfluenceoftheweldheightisprovidedbelow.
0
1
2
3
4
5
6
0 1 2 3 4 5 6 7
Radius[mm]
Kt[]
Anthes(AsWelded)
FEM(AsWelded)
Anthes(TIGdressed)
FEM(TIGdressed)
figure8.7Comparisonofparametricformula(Anthesetal.,1993)andFEMmodelofweldtoe.Parametersof
analyzedgeometries(weldtoeangle,weldheight,platethickness,undercut):AsWelded(28,3,25,0);TIGdressed
(15,3,25,0)
Theinfluenceoftheweldtoeradiusandweldtoeangleisclearlyverylarge.Notonlycanthiseasilybeseeninfigure8.7,
but the presence of these two parameters in the formula of Anthes confirms that the value of K
t
will be very sensitive to
these parameters. The influence of the other parameters: weld height, plate thickness and undercut, is not apparent. To
gain more insight in the behaviour of K
t
under influence of changing parameters a short parameter study will follow. The
influence of the plate thickness will not be considered, because these values are very accurately known, and show little
variation,especiallywhencomparedwiththevariationofotherparameters.
Influenceofweldheight
To study the influence of the weld height on the elastic stress concentration factor, a number of analyses has been
executed on a typical as welded geometry and a typical TIGdressed geometry. The results are shown in figure 8.8 and
figure8.9.
The typical as welded geometry (figure 8.8) shows limited sensitivity to the variation of weld height. As expected, a
higher weld increases the stress concentration factor, and lowering the weld height decreases the stress concentration
factor.Fromacertainweldheight,inthiscasearound3millimeter,increasingtheweldheightdoesnotinfluencethestress
concentrationfactoranymore.Alsoaheightof2millimetershowslimitedvariationfromthebasevalueof3millimeter,up
to 6% for very small radii and no more than 4% for common radii (>0.5 millimeter). If the weld height is lowered to an
extreme value of 0.25 millimeter, the maximum variation with respect to the base value of 3 millimeter is 35% which
graduallydecreasesto27%forverylargeradii.Thisisaconsiderabledifference,butrelativelylimitedconsideredthatthe
weldheighthasreducedbymorethanafactor10.ItmustbenotedthatanydifferencesbetweentheK
t
curveswillbecome
smallerwhenthetransformationtoK
f
ismade.Thisalsoholdsfordifferencesduetoweldtoeangle,weldtoeradiusand
undercut.Thiseffectismostpronouncednearthesmallerradii.
82
1
2
3
4
0 1 2 3 4 5 6 7 8
Radius[mm]
Kt[]
H=4
H=3
H=2
H=1
H=0.5
H=0.25
figure8.8ComparisonofK
t
fordifferentweldheightsforatypicalasweldedgeometry.Parametersofanalyzed
geometries(weldtoeangle,weldheight,platethickness,undercut):(28,?,25,0)
ThetypicalTIGdressedgeometry(figure8.9)showssimilarbehaviourasthetypicalasweldedgeometry.Fromacertain
weld height, in this case 2 millimeter, further increase of the height hardly influences the stress concentration factor. A
heightvalueof1millimetershowsrelativelysmallvariationstothebasevalueof3millimeterofabout5%overthewhole
consideredrangeofradii.Theextremecaseofaweldheightof0.25millimeterleadstoavariationofmaximum18%with
respecttothebasevalueof3millimeter.
1
2
3
4
0 1 2 3 4 5 6 7 8
Radius[mm]
Kt[]
H=3
H=2
H=1
H=0.5
H=0.25
t
fordifferentweldheightsforatypicalTIGdressedgeometry.Parametersofanalyzed
geometries(weldtoeangle,weldheight,platethickness,undercut):(15,?,25,0)
Based on the parameter study it can be concluded that the stress concentration factor is relatively insensitive to weld
height variations around commonly observed values. Because all weld heights are recorded they can relatively easily be
incorporatedinthemodel,butthisconclusionisofgreatvaluewhenthemodelwillbecomparedtorealitylateron.When
weld heights as low as the weld plateau in typical TIGdressed geometry number 3 (see figure 6.9) are considered, a
significantinfluenceisobserved.Thiswillbeaddressedlateron.
It must be noted that this insensitivity of the model to weld height also explains why the model and Anthes formula
complysowellforvaryingweldtoeanglewhiletheweldheightinthemodeldidnotchange.
Influenceofundercut
Tostudytheinfluenceoftheundercutontheelasticstressconcentrationfactor,anumberofanalyseshasbeenexecuted
onatypicalasweldedgeometryandatypicalTIGdressedgeometry,theresultsareshowninfigure8.10andfigure8.11.
The stress concentration of the typical as welded geometry (see figure 8.10) increases as expected when an undercut
arises, and increases further when the depth of this undercut increases. For all common measured undercuts (<0.06
millimeter) this variation with respect to the base value of no undercut is not larger than 20%. For common radii (>0.5
millimeter)thevariationdecreasesto10%orless.
83
1
2
3
4
5
0 1 2 3 4 5 6 7 8
Radius[mm]
Kt[]
UC=0
UC=0,1
UC=0,02
UC=0,04
UC=0,06
t
fordifferentweldheightsforatypicalasweldedgeometry.Parametersofanalyzed
geometries(weldtoeangle,weldheight,platethickness,undercut):(28,3,25,?)
ThetypicalTIGdressedgeometry(seefigure8.11)showssimilarbehaviourtotheasweldedgeometry,butisabitmore
sensitive. For common undercuts the variation has a maximum value of 27%, which decreases to 12% or less when only
commonradiiareconsidered.
1
2
3
4
0 1 2 3 4 5 6 7 8
Radius[mm]
Kt [-]
UC=0
UC=0,02
UC=0,04
UC=0,06
UC=0,1
t
fordifferentweldheightsforatypicalTIGdressedgeometry.Parametersofanalyzed
geometries(weldtoeangle,weldheight,platethickness,undercut):(15,3,25,?)
Basedonthisanalysisitcanbeconcludedthattheinfluenceoftheundercutsthatweremeasuredisrelativelylimitedbut
hastobetakenintoaccountintheanalysis,withthepossibleexceptionoftypicalTIGdressedgeometrytype2(seefigure
6.9) where the influence may be larger. While the influence may be limited, the undercut data is available and easily
translated into the parametric model. Any influences of the undercut will therefore be taken into account in the fatigue
strengthprediction.
The larger undercuts, such as are present in typical TIGdressed geometry type 2, are not present in this analysis.
However,basedonthebehaviourforsmallerundercutsitisverylikelythattheincreaseofK
t
duetothelargerundercutwill
veryprobablybesmallerthanthedecreaseofK
t
duetotheverylargeradius(>10millimeter)whichiscommonlyassociated
withthisgeometry.
Relationbetweenmodelandreality
The model is of course a simplification of reality. In figure 8.12 a rough overlay of a possible weld geometry and its
accompanying model is shown. In two areas the model shows a clear difference with reality. First in the transition from
basematerialintotheundercut(markedwithline1inthefigure).Secondlyinthetransitionfromtheinclinedweldmaterial
tothetopoftheweldmaterial(markedwithline2inthefigure).Thefirstdifferenceisrathersmall,anddoesnotlieinthe
areaofthelargeststressconcentration.Theseconddifferencecanbeverylarge.However,thepreviousparameterstudy
84
has shown a relative insensitivity of the model to the weld height. The difference of between the two lines in figure 8.12
willthereforenothavealargeimpactonthestressconcentrationfactorattheweldtoe.Furthermore,thetopofthereal
weldprofileconsistofanumberofsidebysideweldbeads,andthereforeisnotcompletelyflat(e.g.seefigure6.3).With
the same reasoning as stated above, it can be argued that this will also not influence the stress concentration factor
significantly.
weldtoeradius
undercut
weldtoeangle
weldheight
schematisationofweldtoe
exampleofmeasuredweldprofile
1 2
figure8.12Overlayofanexaggeratedpossiblerealweldgeometryandthemodelusedtorepresentthisgeometry
Thepossibleweldgeometryasshowninfigure8.12canrepresentmostmeasuredgeometries,excepttypicalTIGdressed
type3(seefigure6.9).Inthisweldshapeaclearplateauisvisibleaftertheweldtoeradius.Thiscanbemodeledwiththe
availablemodelintwoways.Theplateaucanbeignored,resultingintoomuchweldmaterial,ortheplateaucouldbeseen
as the maximum height of the weld which results in ignoring most of the weld material. The two proposed models and a
fictiverealweldaredepictedinfigure8.13.
figure8.13TwopossiblerepresentationsoftypicalTIGdressedtype3andafictiverealweldprofile(exaggerated
drawingsoftheweld)
To analyze the consequences of choosing for one of those two models, one typical TIGdressed geometry has been
modeledtocalculatehowthestressconcentrationfactorofthisgeometrywillrelatetothestressconcentrationfactorsof
both proposed models. The results are presented in figure 8.14. It is clear that the real model lies in between the two
proposedmodels.Becauseoftimeconstraints,inthisresearchthechoicewillbemadeforoneofbothmodels.Thiswillbe
themodelontherightinfigure8.14becauseherethedifferencewiththerealmodelisthesmallest.Itmustbenotedthat
thisapproximationonlyhastobemadeforthespecificgeometrieswhichshowaplateauintheweld.
figure8.14Outcomeofthreedifferentmodelsforthesameweld(exaggerateddrawingsoftheweld)
t
Fatigue is a weakest link process. This property makes it very hard to predict the combination of geometry parameters
whichhavetobetakenintoaccounttocreatethecharacteristicweldtoeinamodel.ThefactthattheTIGdressedweld
toe radii showed significant spread and a small additional peak near the main peak of the as welded radii, indicating a
possiblelackofimprovementofthischaracteristicweldtoe,makestheanalysisevenmorecomplicated.
The choice is made to execute two types of analyses, one where for all parameters the main peak value is considered
bothforasweldedandTIGdressedspecimensandonewheretheradiusoftheTIGdressedspecimenshasasimilarvalue
as the as welded radii while all other parameters are assumed to be at their peak value. Subsequently, influencing
parameterswillbevaried,resultinginanexpectedbandwidthofK
t
.
85
Someparameterswillbecalculated,suchasthemeanofthe,nearlynormaldistributed,weldtoeangles.Otherswillbe
estimated with the aid of the diagrams provided in 6.3.4. This introduces some extra uncertainty, but calculation would
require the fitting of a probability density function, which also has an error margin, especially when the results do not fit
the distribution very well. If the results of the laser measurements of the weld toe would have been categorized and any
possible dependencies would have been documented, a statistic analysis could be performed which can, based on the
occurrenceofdifferentgeometries,moreaccuratelypredictthemeanandstandarddeviationofK
t
.Becausesuchelaborate
documentation of the weld toe parameters has been omitted in this research, the variation of K
t
and K
f
will be based on
estimates.
Firstanalysis,peakvalues
The base values of this analysis are depicted in table 8.2. It is indicated in the table for each variable how it has been
determined.
Aswelded TIGdressed determinedby

Weldtoeradius[mm] 1.0 3.8 estimate
Weldtoeangle[] 27.8 15.3 calculatedmean
Weldheight[mm] C11 3.9 3.9 calculatedmean
C89 4.1 4.1 calculatedmean
C89a 4.3 4.3 calculatedmean
V11 3.7 3.7 calculatedmean
Undercut[mm] 0 0 estimate
Platethickness[mm] S1100 20 20 measured
other 25 25 measured
table8.2Peakvaluesofweldtoes
Theweldtoeasdescribedintable8.2willbeconsideredasthebasevalue.Ashortsensitivityanalysiswillfollow,where
all values are related to this base value. In figure 8.15 to figure 8.18 the results are presented for a variation of a certain
amounttobothsidesofthechosenbasevalue.ThediagramshowsthevariationofK
t
inrelationtoitsbasevalueasstated
intable8.2,whenthisbasevalueisvaried(seealsoequation(8.4)).Fortheweldtoeangleandradiusthevaluesarevaried
for one standard deviation to both sides of the base value. For the undercut a value of 0.06 was chosen, based on the
diagramsin6.3.4.Theweldheightisvariedoverawiderbandtocoverthemeanandstandarddeviationofallconsidered
plates and has a base value (valued 1 in the graph) of 3. The assumption is that the influence of one parameter is not
influenced by the change of another parameter. If for example the radius is determined at a slightly different value, it is
assumedthattheinfluenceoftheweldtoeangle(seefigure8.16)issimilar.Basedonobservationsonthediagramsbelow,
thisseemsareasonableassumptionbecausethediagramsalllooksimilarwhiletheyaredeterminedfortwoverydifferent
geometries.
0.85
0.9
0.95
1
1.05
1 1.5 2 2.5 3 3.5 4 4.5 5
Weldheight[mm]
f
wh
[]
0.85
0.9
0.95
1
1.05
1 1.5 2 2.5 3 3.5 4 4.5 5
Weldheight[mm]
f
wh
[]
t=25
t=20
figure8.15VariationofK
t
withvariatingweldheightforaswelded(left)andTIGdressedjoints(right)
86
0.9
0.95
1
1.05
21.3 24.3 27.3 30.3 33.3
Weldtoeangle[]
f
[]
0.9
0.95
1
1.05
10.3 12.3 14.3 16.3 18.3 20.3
Weldtoeangle[]
f
[]
t=25
t=20
t
withvariatingweldheightforaswelded(left)andTIGdressedjoints(right)
1
1.02
1.04
1.06
1.08
0 0.02 0.04 0.06
Undercut[mm]
f
uc
[]
1
1.02
1.04
1.06
1.08
0 0.02 0.04 0.06
Undercut[mm]
f
uc
[]
t=25
t=20
t
withvariatingweldheightofaswelded(left)andTIGdressedjoints(right)
0.8
1
1.2
1.4
0 0.5 1 1.5 2
Weldtoeradius[mm]
f
[]
0.8
1
1.2
1.4
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
f
[]
t=25
t=20
figure8.18VariationofKtwithvariatingweldheightofaswelded(left)andTIGdressedjoints(right)
Withtheaidofthegraphsabove,alargerK
t
willbedetermined.Theassumptionisavariationofonestandarddeviation
ofallparameters,whichresultsinmultiplicationfactorsforeachweldtoevariable.BecauseK
f
alsodependsontheradius
of the weld toe (see 3.5), the influence of radius variation will be covered in the next section. As can be seen in the
diagramsabove,theinfluenceofthethicknessontheinfluenceoftheweldtoeradius,weldtoeangleandundercutisvery
limited. Therefore no difference in thickness for these parameters will be made. For the weld height the differences are
somewhatlarger,thereforeinthesefactorsthedifferencebetweenthe20and25millimeterplatesareincorporated.The
different correction factors that are derived can be found in table 8.3. The total influence of weld toe angle, weld height
andundercutvariationseemslimitedandcanbecalculatedwithequation(8.3).
87
Aswelded TIGdressed
Basevalue Variedto Factorf
i
i
Weldtoeangle[] 27.8 34.4 1.05 15.3 20.2 1.05
Weldheight[mm] C11 3.9 4.4 1 3.9 4.4 1
C89 4.1 4.4 1 4.1 4.4 1
C89a 4.3 4.5 1 4.3 4.5 1
C69 2.3 2.5 1.01 2.3 2.5 1
C46 1.8 2.0 1.01 1.8 2.0 1
V11 3.7 3.9 1 3.7 3.9 1
V89 3.0 3.3 1.01 3 3.3 1
V69 2.5 2.6 1 2.5 2.6 1
V46 2.1 2.3 1.01 2.1 2.3 1
Undercut[mm] 0 0.03 1.04 0 0.05 1.04
table8.3Influencefactorsforvariationofonestandarddeviationofweldtoeangle,weldheightandundercut
; t adj uc wh
K f f f
u
= (8.3)
Inwhich:
f
=influencefactorforvariationofweldtoeangle[]
f
uc
=influencefactorforvariationofundercut[]
f
wh
=influencefactorforweldheight[]
( )
( )
t
i
t
K i base value
f
K i base value
=
=
=
(8.4)
Secondanalysis,noimprovementinTIGdressedradius
Forthisanalysisallvaluesoftheasweldedweldtoearethesameasstatedintable8.2.ThevaluesoftheTIGdressed
specimenarealsoassumedthesame,withtheexceptionoftheweldtoeradius.Basedonobservationsonfigure6.22itis
very well possible that the weakest link radius of the TIGdressed specimens has a similar value as in the as welded
specimen,resultinginabasevalueradiusof1millimeter.AnadditionalsensitivityforthenewTIGdressedbasegeometry
ispresentedinfigure8.19andfigure8.20.
0.85
0.9
0.95
1
1.05
1 1.5 2 2.5 3 3.5 4 4.5 5
Weldheight[mm]
f
wh
[-]
0.85
0.95
1.05
10.3 12.3 14.3 16.3 18.3 20.3
Weldtoeangle[]
f
[]
t=25
t=20
t
withvariatingweldheight(left)andweldtoeangle(right)fortheTIGdressedweldtoe
1
1.02
1.04
1.06
1.08
1.1
0 0.02 0.04 0.06
Undercut[mm]
f
uc
[]
0.75
1
1.25
1.5
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
f
[]
t=25
t=20
t
withvariatingundercut(left)andweldtoeradius(right)fortheTIGdressedweldtoe
88
Aremarkmustbemadeonthevariationoftheweldtoeangleandundercutwhichseemtohavemoreinfluenceforthis
geometry than they have for the standard as welded geometry or the previously considered TIGdressed geometry.
Furthermore,theweldtoeradiusinfluenceseemsfarless,butthisisfullycausedbythechangeofbasevalue(=1instead
of=3.8)whichdistortsthediagram.ThecorrectionfactorsfortheadjustedTIGdressedweldtoearegivenintable8.4.
TIGdressed(=1)
i
Weldtoeangle[] 15.3 20.2 1.1

Weldheight[mm] C11 3.9 4.4 1
C89 4.1 4.4 1
C89a 4.3 4.5 1
C69 2.3 2.5 1
C46 1.8 2.0 1
V11 3.7 3.9 1
V89 3.0 3.3 1
V69 2.5 2.6 1
V46 2.1 2.3 1
Undercut[mm] 0.0 0.05 1.08
table8.4Influencefactorsforvariationofonestandarddeviationofweldtoeangle,weldheightandundercut
f
For the determination of K

t
the critical distance approach (see paragraph 3.5) will be used, because of the lack of
experience with high strength steel in the fictitious notch rounding approach and the highly stressed volume approach
would force the use of diagrams instead of formulas to determine the equivalent notch stress. These diagrams are not
availableforallsteelgrades.Ifequation(3.9)isusedtodeterminea*,thefollowingvaluescanbefoundforthedifferent
steeltypes:
Material R
m
[N/mm
2
] a*[mm]
G20Mn5(combinedwithS460) 622 0.217
599 0.233
611 0.224
G10MnMoV63(combinedwithS690) 799 0.138
785 0.143
841 0.126
G18NiMoCr36(combinedwithS890) 1042 0.086
1052 0.084
1070 0.082
G22NiMoCr56(combinedwithS1100) 1185 0.068
1163 0.070
1171 0.070
S460 590 0.239
S690 843 0.126
S890 1065 0.083
1051 0.085
S1100 1373 0.052
table8.5Calculatedvaluesofa*forthedifferentsteelplatesused
With the use of equation (3.10) the K
t
of all possible geometries can be transformed into a fatigue notch factor K
f
.
Although the values of R
m
anda* for the paired cast steeland rolled steel of the Cspecimensshow somedifference (see
table 8.5) the difference in fatigue notch factor (K
f
) is always less than 1% between the two halves of one Cplate in the
cases considered. Therefore, no distinction is made between cast steel and rolled steel cracks with regard to the fatigue
notchfactor.Infigure8.21tofigure8.23thebehaviourofK
t
andK
f
isshownforthefourdifferentconsideredsteelgrades,
both for the peak values, as well as the adjusted values (K
t;adj
and K
f;adj
) for deviating weld height, undercut and weld toe
angle (see table 8.4 for the influence factors). The figures are depicted for the as welded situation, TIGdressed situation
89
and TIGdressed situation without radius improvement. All plates have slightly different weld height (see table 8.2),
resulting in slightly different fatigue notch factors. These differences will be included in the analysis, but the differences
between plates of the same steel grade (e.g. V890 and C890) are not shown here, because the overall shape of the
diagramsisthesame,andanyquantitativedifferencesaresmall.
V460
1
1.5
2
2.5
3
3.5
4
0 0.5 1 1.5 2
Weldtoeradius[mm]
K[]
V690
1
1.5
2
2.5
3
3.5
4
0 0.5 1 1.5 2
Weldtoeradius[mm]
K[]
V890
1
1.5
2
2.5
3
3.5
4
0 0.5 1 1.5 2
Weldtoeradius[mm]
K[]
V1100
1
1.5
2
2.5
3
3.5
4
0 0.5 1 1.5 2
Weldtoeradius[mm]
K[]
figure8.21InfluenceofradiusvariationonK
t
()andK
f
()fordifferentsteelgradesintheasweldedsituation.
Peakvaluesareplottedinblack,adjustedvaluesingrey.
V460
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
V690
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
V890
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
V1100
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
t
()andK
f
()fordifferentsteelgradesintheTIGdressedsituation.
Peakvaluesareplottedinblack,adjustedvaluesingrey.
90
V460
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
V690
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
V890
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
V1100
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Weldtoeradius[mm]
K[]
t
()andK
f
()fordifferentsteelgradesintheTIGdressedsituation
withoutradiusimprovement.Peakvaluesareplottedinblack,adjustedvaluesingrey.
The fatigue strength curve of each material will be determined for different cases. One analysis uses the peak value of
the weld toe radius, combined with the peak value of all other weld toe parameters. In case of the TIGdressed situation
withoutradiusimprovementanexceptionismadefortheweldtoeradius,whichiskeptatthepeakvalueoftheaswelded
situation.Anotheranalysiswillbemadeusingtheweldtoeparameterswhicharecorrectedtosimulatethepossibilityofa
weakest link. All parameters except the weld toe radius were already varied over a value of one standard deviation in
8.3.2. A similar variation of the weld toe radius would lead to very small differences between the three considered
situations, because especially the TIGdressed radii show a large standard deviation. Because small radii in combination
withotherTIGdressedweldtoeparametersarealreadycoveredinaseparatesituation,such alargevariationisdeemed
too conservative and is therefore omitted. The chosen variation of the weld toe radius is determined at half of its peak
value,whichcoversthebulkoftheweldtoeradii(see6.3.4).Thisleadstoanadjustedweldtoeradiusof0.5millimeterfor
the as welded situation and an adjusted weld toe radius of 1.9 millimeter for the TIGdressed situation. The calculated
stress concentration factors and fatigue notch factors for both the peak values (K
t
and K
f
) and the corrected values after
variationoftheweldtoeparameters(K
t;adj
andK
f;adj
)canbefoundintable8.6andtable8.7.Someslightdifferencesmay
existbetweenthecastandrolledsideofthespecimens.Thisisnotdepictedintable8.6andtable8.7,butisincorporatedin
themodelandthecomparisonwiththefatiguetestsin9.4.3.
Aswelded TIGdressed TIGdressed(=1)

K
t
K
t;adj
K
t
K
t;adj
K
t
K
t;adj
C1100 2.16 2.79 1.47 1.78 1.79 2.35

C890 2.26 2.92 1.52 1.84 1.85 2.43
C890a 2.26 2.92 1.52 1.84 1.85 2.42
C690 2.17 2.83 1.50 1.82 1.83 2.40
C460 2.11 2.75 1.49 1.81 1.81 2.38
V1100 2.15 2.78 1.47 1.78 1.79 2.35
V890 2.22 2.89 1.52 1.84 1.84 2.42
V690 2.19 2.82 1.51 1.83 1.83 2.41
V460 2.15 2.80 1.50 1.82 1.82 2.39
table8.6Calculatedelasticstressconcentrationfactorsforthethreedistinguishedsituationsforpeakvaluesand
adjustedvalues
91
Aswelded TIGdressed TIGdressed(=1)

K
f
K
f;adj
K
f
K
f;adj
K
f
K
f;adj
C1100 2.10 2.61 1.47 1.76 1.75 2.22

C890 2.16 2.63 1.51 1.80 1.78 2.22
C890a 2.16 2.64 1.51 1.80 1.78 2.22
C690 2.04 2.45 1.49 1.77 1.74 2.12
C460 1.91 2.19 1.46 1.72 1.66 1.95
V1100 2.10 2.61 1.47 1.76 1.75 2.22
V890 2.12 2.61 1.50 1.80 1.77 2.21
V690 2.05 2.44 1.49 1.77 1.74 2.12
V460 1.93 2.20 1.47 1.72 1.66 1.94
table8.7Calculatedfatiguenotchfactorsforthethreedistinguishedsituationsforpeakvaluesandadjustedvalues
8.4 Determinationofmeanstressfactor
Thefatiguestrengthofspecimensisalsoinfluencedbyameanstress(seefigure2.7).Thismeanstressisdeterminedby
thestressratioR(see2.3.2)andbytheresidualstress
r
.Forweldedconnections,aresidualstressequalinmagnitudeto
theyieldstressisusuallyassumed.TheTIGdressingprocessisexpectedtolowerresidualstresses.Duetotimelimitations,
measurementsonresidualstressescouldnotbedone.Tomakeareasonableassumption,referenceismadetoliterature.
Asshownin4.3.3,LopezMartinezhasinvestigatedtheresidualstressbeforeandafterTIGdressing.BeforeTIGdressinghe
found a residual stress at the surface of 0.61
y
, which was reduced to 0.43
y
after TIGdressing. In deeper layers of the
materialtheresidualstressincreased,butbecausethecracksareassumedtoinitiateatthesurface,whichiswellsupported
byobservationsonthespecimens,asimilarresidualstressreductionisassumedtooccurinthespecimens.Whenfortheas
weldedspecimens
r
=
y
isassumed,theresidualstressafterTIGdressingcanbeassumedas:
0.43
0.7
0.61
r y y
o o o = = (8.5)
The mean stress of all specimens will be adjusted to a value of

m
=
y
. Therefore the analytical values will also be
adjustedtothisvalue.BasedonHaibach(2006),Pijpers(2011)proposedequation(8.6)todetermineameanstressfactor
includingtheeffectofresidualstresses:
( )
( )
;0
;0
;0
;
;0
;0
2
;0
1
when 1
1
1
when 1 1
1
1
3
when1 3
1 1
3
1
3
when3
1
m r
f aE
m r
m r
aE
f
aE
m Haibach
f
m r
aE f m r
f
aE
f
m r
aE
f
M
M
f
M
M
M
M
M
o o
o
o o
o o
o
o
o o
o
o o
o
o o
o
+
<
+
< <
+
+
+
< <
| |
+
+ +
|
|
\ .
+
<
(8.6)
with 0.00035 0.1
f m
M R =
92
Inwhich:
f
m;Haibach
;0 m
;0 m
for 0and 0
for 0and 0
aE r
aE r
o o o
o o o
= =
= =

M
f
meanstresssubfactor[N/mm
2
]
m
meanstress[N/mm
2
]
r
residualstress[N/mm
2
]
aE;0
fatigueendurancestrengthatalternatingload[N/mm
2
]
8.5 Determinationofthicknessfactor
The plate thickness of a specimen can influence the fatigue strength. Thicker plates generally show less favourable
fatigue behaviour. The calculation procedure as described in this chapter leads to the fatigue resistance at a reference
thickness of 25 millimeter. Because the fatigue test results are all adjusted to a thickness of 25 millimeter, the fatigue
strength prediction does not need to be adjusted. The determination and use of the thickness factor will be explained in
9.2.2.
8.6 Determinationofloadingmodefactor
Themethodthatisusedreferstotensileloadingofthespecimens.Inthefatiguetestinginthisresearchalsoonlytensile
loadingwillbeapplied.Duetomisalignmentstheloadwillbepartlyinbending,butonlytheresultsoftheactualtestswill
beadjustedforanyinfluenceofthebendingloadingmode.Thefatiguestrengthpredictionwillrefertopuretension,thus
eliminatingtheneedforanyloadingmodefactor.Thecorrectionofthefatiguetestresultswillbeexplainedin9.2.2.
8.7 Predictionoffatiguestrengthcurve
Thefatiguestrengthcurveforcrackinitiationisdeterminedinseveralsteps.Firstthefatiguelimitofthebasematerialis
determinedwithequation(8.1).Thisvalueisthencorrectedfornotches,meanandresidualstressanddisplayedasastress
rangeinsteadofastressamplitude,seeequation(8.7).
; ; ;0
2
E specimen m a E
f
f
K
o o A = (8.7)
Subsequently,thefatiguestrengthcurveisdeterminedwitheithertheapproachbySchijveorHcketal.(see3.5.1).The
kneepointforthefatiguestrengthcurveintheapproachbySchijveissetat210
7
accordingtotherecommendationinhis
publication(see3.5.1).
Thisresultsinseveraldesignlinesperplate.Firstofalltherearethethreeconsideredsituations:aswelded,TIGdressed,
andTIGdressedwithoutradiusimprovement.Furthermore,toinvestigatethesensitivityofthefatiguestrengthtovariation
oftheweldtoeparameters,twovaluesforK
f
havebeencalculatedin8.3.2and8.3.3.Oneofthesevalues,K
f;adj
istheresult
of an unfavourable variation of all weld toe parameters for a certain amount. This variation does not result in a specific
failurechance,andhasnomorevaluethanasensitivityanalysis.ThedifferencebetweenthelinesdeterminedwithK
f
and
K
f;adj
givesinsightinthesensitivityoftheresultsforslightvariationoftheweldtoegeometry.Thiswillbecommentedonin
9.4.3.
93
9 Analysisoffatiguetestresults
9.1 Chapteroutline
Inthischapterthefatiguedatawillbeanalyzed.First,aselectionandadjustmentofthetestdatawillbediscussed.Then
thestatisticalanalysisofthedatafollows,whichleadstomeanandcharacteristicSNcurves.Inparagraph9.3,thedataof
theTIGdressedspecimensiscomparedwithdifferentdesigncodes,asweldeddataandtheanalyticaldeterminationofthe
fatiguestrength.TheinfluenceofTIGdressingissummarizedinparagraph9.4.
9.2 Analysisofrawdata
9.2.1 Exclusionofdata
Notaldataisdeemedrepresentativeforanalysis.Insomecasesthefailuremechanismthatwasobservedcannotoccur
incommonpractice.Thesedatapointswillthereforebeexcludedandnotbeusedinthestatisticalanalysisofthefatigue
data,butcanstillbefoundinthetabulatedfatigueresultsinAnnexC.
Excludeddata
Thedatapointsofthefollowingspecimenshavebeenexcludedfromanalysis:
C462
C693
C892
V111
Duetoafailureofthetestingrig,specimenC462hasbeencompressedandhasbuckled(seefigure9.1).Thepermanent
deformation resulting from this buckling or the straightening needed for testing will very probably influence the result, if
the specimens is tested for fatigue strength. Therefore specimen C462 was not tested, and is thus absent in the fatigue
datatabulationinAnnexC.
figure9.1DamagedspecimenC462
ThebacksideofplateC69couldnotbeTIGdressed(see5.2.2).Thisresultedinaweldrootwithrathersharpnotches(see
figure9.2).WithspecimenC691thisstillresultedinweldcapfailure,butatspecimenC693thesenotchesinitiatedacrack
onthebackofthespecimen.Becausenotchesofthesedepthandsharpnesswillneverpassqualitycontrolinanypractical
structure,thisdatapointiseliminated.TopreventasimilarfailurewithspecimenC692,theweldrootwasgroundsmooth.
94
figure9.2BacksideofspecimenC691withsharpnotchesattheweldroot
C822 showed very large flaws in the cast plate, with depths of a few millimeters. To repair the plate these flaws were
removedbymillingandtheresultinggrooveswerefilledwithweldmaterial.Unfortunately,alsotherepairedspecimenwas
notfreeofflawsinthecastbaseplate.Furthermore,theweldrootcouldnotbeTIGdressed(seealso5.2.2).Thisresulted
insimultaneouscrackinitiationintheparentmaterialandtheweldroot.Becausebothcrackinitiationsiteswouldnothave
passedqualitycontrolinpractice,thisdatapointiseliminated.
Infigure7.6itcanclearlybeseenthatonedatapointdoesnotcoherewiththerestofthedata.Thisspecimen(V111)
hasfailedintheparentmaterial,ontherootsideoftheweld.Ontherootsideoftheweld,stressesaremuchlowerdueto
themisalignment.Althoughnocleardefectcouldbefoundwhichcanleadtoexclusionofthisdatapoint,thedatapointwill
not be included in the analysis. It is reasonable to assume that the defect that caused the initiation of this crack has not
beeninfluencedbytheTIGdressingbecausethecrackoccurredattherelativelylargedistanceofafewcentimetersfrom
theweldtoe.IfthisspecimenhadnotbeenTIGdressed,thecrackwouldlikelyariseatthesamelocation,becausethedata
pointalsolieslowerthanmostasweldeddatapointswhichweretestedbyPijpers(2011).Therefore,fortheinvestigation
oftheinfluenceofTIGdressing,thisspecimenprovidesnousefulinformation.
Rerunsofspecimens
Some specimens have shown crack initiation in the base material. Three of those base material cracks were outside of
thecrosssectionreduction.Thisconcernsthespecimens:
C112
C891
C892
Asdescribedabove,C892willbeexcludedfromtheanalysis.Theremainingtwospecimenshaveaclearcrackinitiation
siteataflawinthematerial,whichcouldhavebeenvisuallyassessedduringqualitycontrol.Becausethecrackoccurredat
asignificantdistancefromthecrosssectionreductionitwaspossibletocontinuetestingthespecimenafteritfailedinthe
basematerialatapproximatelythesamestresslevel.Forseveralreasonsthestresslevelofasecondtestwillneverbeat
exactly the same level. This is caused by the reduced specimen length, resulting in a changed misalignment, and the
uncertaintywherecrackswillinitiate.Thestressesinthesecondtestarematchedtostressesinthefirsttestfromastrain
gaugeclosetoalikely,highlystressed,crackinitiationpoint.Thisisnotnecessarilythecrackinitiationpointofthesecond
test.Tocalculatethecombinedstressforthefinalresult,equation(9.1)hasbeenused.
( )
( )
( )
1 2 1 2
1
, , , , ,
; ;
1
m m m m
i i S d L d j j S d
m
eq S d
i j
n n
D n n
o o o
o

A + A A
A =
+

(Hobbacher,2007) (9.1)
Inwhich:
D Minersumoftotaldamage[]
i,S,d
valueofcharacteristicequivalentstressrange[N/mm
2
]
m
1
slopeofSNcurveabovethekneepoint[]
m
2
slopeofSNcurvebelowthekneepoint[]
i,S,d
stressranges(loads)abovethekneepoint[N/mm
2
]
j,S,d
stressranges(loads)belowthekneepoint[N/mm
2
]
L,d
stressrange(resistance)atthekneepointoftheSNcurve[N/mm
2
]
n
i
numberofcyclesbelongingto
i
[]
n
j
numberofcyclesbelongingto
j
[]
InthecalculationtheMinersumoftotaldamageisassumedtobe1.Allstressrangesareassumedtolieabovetheknee
point of the SN curve, thereby omitting the need for values of m
2
and
L,d
. The slope of the SN curve above the knee
95
point(m
1
)isassumedtobe3.Alltheseassumptionsdonotinfluencetheresultmuchbecausethestressrangesofthetwo
tests are relatively close together. Other equivalent parameters are combined according to the number of cycles of each
test,seeequation(9.2).Thelimitedinfluence oftheassumptionsforequation(9.1)showwhenequation(9.2)isusedfor
thecalculationofthecombinedstress:practicallythesamecombinedstressfollowsfrombothformulas.
I I II II
eq
I II
X n X n
X
n n
+
=
+
(9.2)
Inwhich:
X
eq
equivalentparameter[]
X
I
parametervalueinfirsttest[]
N
I
numberofcyclesinfirsttest[]
X
II
parametervalueinsecondtest[]
n
II
numberofcyclesinsecondtest[]
All presented results in graphs for specimen C112 and C891 represent the results of the final failure. In the tabulated
testresultsinAnnexCthevaluesfromtheseparatedtestsareavailabletogetherwiththeircombinedvalue.
Failureoutsideweldregion
Finally,somespecimensmustbementionedwhichhaveshownfailureoutsidetheweldregion,butwillnotbeexcluded
fromthedata:
C111
C112(afterrerun)
V111
C891(afterrerun)
Allspecimensshowacrackintheparentmaterial,butinsidethereducedcrosssection.Becausethereisnoclear,large
defect at which the crack has initiated which should not pass quality control in practice, these data points will not be
excluded from the analysis. Their failure mode however, is different from all weld toe or weld material cracks. This could
either be the result of a defect and the higher sensitivity of very high strength steels (especially S1100) for these defects
(see 4.2.2) or an improvement of the fatigue strength in the weld region, caused by TIGdressing. Such an improvement
may have led to the base material being the weakest link. These data points will be marked in graphs to identify them if
theyshowsignificantdifferentbehaviourthantheotherspecimens.Anybasematerialfailuresfromtheasweldedtestswill
also be marked. If the base material failures after TIGdressing show better fatigue behaviour than the as welded base
material failures, this may be an indication that the base material failures in this research may have been caused by
improvingtheweldtoe,andshiftingtheweakestlinktothebasematerial.Failuresintheweldmaterialareconsideredto
necessarily be an improvement of the connection, because such failures did not occur in the research by Pijpers. For this
reason these results are judged as regular and will not be marked separately. In Annex C all different failure mechanisms
arelistedperspecimen.
9.2.2 Adjustmentoftestdata
To correctly compare the test data to the standardized cases in design codes, and to the previous experiments carried
outbyPijpers(2011),thenominalstresslevelswillbeadjusted.Forthedefinitionanddeterminationofthesefactorsthis
researchreliestoagreatextentontheworkbyPijpers(2011).Correctionfactorswillbeappliedtocorrectfor:
Meanstress;nominalstressesareadjustedtoameanstressvalueof
m
=0.5
y
Residualstress;nominalstressesareadjustedtoaloadingwithacertainrepresentativeresidualstress
Thickness;nominalstressesareadjustedtoareferencethicknessof25millimeter
Loadingmode;nominalstressesareadjustedtoapuretensionloadingmode
Theadjustednominalstresscanthenbecalculatedwiththeuseofequation(9.3)forbasematerialcracksandequation
(9.4)forweldtoecracks.
; ;
; ; ; ;
1 1
f f
n u Nf m n
t u N lm u N
f
f f
o o A = A (9.3)
Inwhich:
f
m
Meanstressfactor[]
f
t;u;Nf
Thicknessfactorforfatiguelifeofbasematerialcracks[]
f
lm;u;Nf
Loadingmodefactorforfatiguelifeofbasematerialcracks[]
96
; ;
; ; ; ;
1 1
f f
n w Nf m n
t w N lm w N
f
f f
o o A = A (9.4)
Inwhich:
f
m
Meanstressfactor[]
f
t;w;Nf
Thicknessfactorforfatiguelifeofweldtoecracks[]
f
lm;w;Nf
Loadingmodefactorforfatiguelifeofweldtoecracks[]
BecausethestudyofPijpersshowednocracksintheweldmaterial,noadjustmentfactorswerederivedforthisfailure
mode. Due to time limitations, the choice is made to treat weld material cracks as weld toe cracks in this research.
Furthermore,mostcorrectionfactorsderivedbyPijpersshowsomevariationforthicknessandweldtoeshape.Thesame
correctionsfactorshavebeenusedinthisresearch.Itisrecommendedtoinvestigatethechangeinthesecorrectionfactors
iftheTIGdressedgeometriesandweldmaterialcracksareaccountedfor.
Meanstressfactor
The as welded specimens tested by Pijpers (2011) were conservatively assumed to have a residual stress of 0 N/mm
2
.
Subsequentlytheresultswereadjustedtoavaluewhere
r
=
y
and
m
=
y
,whichisthestandardreferencevalueforSN
curves.AfterTIGdressing,theresidualstressesareexpectedtobelower,butcannotbecomecompressive,duetothestill
presentcoolingoftheweldtoe.Therefore,areasonableandconservativeassumptionstillisaresidualstressof0N/mm
2
in
thespecimen.TheTIGdresseddataarealsoadjustedtoareferencevalue.Forthis,theresidualstressreductiondeducted
fromLopezMartinez(see8.4)isused.Thetestresultsareadjustedto
r
=0.7
y
To adjust the test results, the assumed residual stress of 0 N/mm
2
and the measured value of
m
were used in
combinationwithequation(8.6)toobtaintheequivalentstressfor
m
=0and
r
=0.Fromthereadjustmentto
m
=
y
and
r
=0.7
y
can be determined using the same equation. The total result of these adjustments is displayed as f
m
in the
tabulatedfatiguetestresults.
Thicknessfactor
In general, if plate thicknesses increase, the resistance to fluctuating stresses at a notch decreases (see 2.3.1). The
thickness effect is described in Eurocode according to equation (9.5). Note that this factor applies to the total fatigue life
(N
f
)
0.2
; ;
25
f
t w N
f
t
| |
=
|
\ .
(9.5)
Pijpers (2011) usedequation (9.6) and equation (9.7) to calculate the thicknesseffect for the crack initiation life, based
onaparameterstudy.Pijpersdidnotfindaconfirmationofthethicknesseffectduringthecrackpropagationlife,resulting
inf
t;w;Np
=1.Equation(9.6)isvalidforweldtoeanglesof25,andmorepronouncedthicknesseffectshavebeenfoundfor
steeper angles (equation (9.7)). It therefore seems reasonable to expect a decreased thickness effect for TIGdressed
specimensduetothereducedweldtoeangle.
0.4
; ;
25
i
t w N
f
t
| |
=
|
\ .
for=25 (9.6)
0.7
; ;
25
i
t w N
f
t
| |
=
|
\ .
for=65 (9.7)
Although Pijpers uses a factor which is more conservative than Eurocode, and there is reason to assume a less
detrimental thickness effect because of the lower weld toe angle in TIGdressed specimens, equation (9.6) will be used
because of the differentiation between crack initiation life and crack propagation life. Equation (9.5) applies to the full
fatiguelife,whileonlythecrackinitiationlifeiscoveredbythenotchstressapproachandtheallegedimprovementofTIG
dressing applies mainly to the crack initiation life. Furthermore, the thickness effect proposed by Pijpers is more
pronouncedthanstatedinEurocodewhichmakestheuseofequation(9.6)conservative,atleastduringthecrackinitiation
life.
The adjustment for thickness is used to refer to a standard thickness of 25 millimeter. Most specimens approximate a
thickness of 25 millimeter, except for the S1100 specimens, which have a thickness of 20 millimeter. The thickness
adjustment factor consists of an adjustment factor which is applicable during the crack initiation life and an adjustment
factorwhichisapplicableduringthecrackpropagationlife.Thethicknessfactorforthetotalfatiguelifeforweldtoecracks
canthenbecalculatedwiththeuseofequation(9.8).
; ; ; ; ; ;
f i i p p
t w N t w N N t w N N
f f f f f = + (9.8)
97
Inwhich:
f
t;w;Ni
Thicknessfactorforcrackinitiationlife[]
f
Ni
Crackinitiationliferatio[]
f
t;w;Np
Thicknessfactorforcrackpropagationlife[]
f
Np
Crackpropagationliferatio[]
Pijpersdidnotfindathicknesseffectforbasematerialcracks,resultinginf
t;u;Nf
=f
t;u;Ni
=f
t;u;Np
=1.
Loadingmodefactor
Theadjustmentforloadingmodeisusedtorefertoastandardloadingmodeofpuretension.Becauseallspecimenshad
acertainmisalignment,partoftheloadwasbending.Thecorrectionfactorcanbecalculatedwiththeuseofequation(9.9)
forbasematerialcracksandequation(9.10)forweldtoecracks.
; ; ; ; ; ;
f i i p p
lm u N lm u N N lm u N N
f f f f f = + (9.9)
Inwhich:
f
lm;u;Ni
Loadingmodefactorforcrackinitiationlife;1.0(Pijpers,2011)[]
f
Ni
f
lm;u;Np
Loadingmodefactorforcrackpropagationlife;1.09(Pijpers,2011)[]
f
Np
; ; ; ; ; ;
f i i p p
lm w N lm w N N lm w N N
f f f f f = + (9.10)
Inwhich:
f
lm;w;Ni
Loadingmodefactorforcrackinitiationlife;1.2(Pijpers,2011)[]
f
Ni
f
lm;w;Np
Loadingmodefactorforcrackpropagationlife;1.37(Pijpers,2011)[]
f
Np
Theapplicablecorrectionfactor forthemisalignedaxiallyloadedspecimenscanbecalculated withtheuseof equation

(9.11).
; ; ;
; ;
1
1
f
m
lm w N ax
m
lm w Nf
K
f
K
f
=
| |
+ |
|
\ .
(9.11)
Inwhich:
K
m
Misalignment factor. [] Hobbacher (2007) proposed a formula based on geometry to calculate a stress increase
factor.InthisresearchK
m
isbasedontheratiobetweenF/Aandthemeasuredstressneartheweld[]
f
lm;w;Nf
Loadingmodefactorfortotalfatiguelifeforweldtoecracks.[]Ifbasematerialcracksareconsidered,f
lm;w;Nf
is
replacedbyf
lm;u;Nf
(cracksintheweldmaterialwillregardedasweldtoecracks)[]
9.2.3 Statisticalanalysis
The fatigue data are analyzed according to the method described by Brozzetti et al. (1989) which is also prescribed by
Eurocode.Foragivendatasetameanregressionlineisconstructedwithasurvivalprobabilityof50%(seeequation(9.12))
i N i
y a b x c = + + (9.12)
Inwhich:
y
i
LogN
f
[]
x
i
Log[]
a
N
intersectiononlogNaxis[]
b regressioncoefficient[]
sumofunknownrandomerrors[]
Theresidualrandomerrorsareminimizedwhena
N
andbareestimatedwiththeleastsquaresmethod:
98
( )
( )
2
2
i i i i
i i
n x y x y
b
n x x

=

(9.13)
( )
1
N i i
a y b x
n
=

(9.14)
Inwhich:
n numberofspecimens[]
TheslopeoftheSNcurve(bfromequation(9.13))caneitherbedeterminedbytheleastsquaresmethod,orfixedtoa
certainvalue.Especiallyforsmallbatchesoffatiguetests(smallerthan10)thefixedvalueoftheslopeisasensiblesolution.
A common value for the slope is 3 for as welded joints. As mentioned in 4.3.4, different authors have found the slope to
increase(i.e.aflatterSNcurve)whenTIGdressingisapplied.Therefore,althoughthenumberofspecimensislimited,an
SNcurvewithafreeslopeandafixedslopeof3willbothbedetermined.
ForcomparisonwithEurocodedetailcategoriesandasweldeddata,themeanstressrangewithaprobabilityofsurvival
of 50% and 95% at N = 210
6
cycles is calculated. Themean stress with a 50%probability of survival at 210
6
is calculated
withequations(9.15),(9.16)and(9.17).
50%
50%
10
x
o A = (9.15)
50%
50%
N
y a
x
b
= (9.16)
( )
6
50%
log 2 10 y = (9.17)
Tocalculatethestressrangewitha95%survivalprobability,thecoefficientofvariationfirstneedstobedetermined.This
canbedonewithequation(9.18).
( ) ( )
( ) ( )
2
2
2
whenbiscalculated
2
whenbisfixed
1
i N i
i N i
y a b x
n
s
y a b x
n

=

+
(9.18)
Inwhich:
s coefficientofvariation[]
Nowthecharacteristicstressrangeat210
6
cyclesiscalculatedfora75%confidencelevelof95%probabilityofsurvival.
Asstatisticaldistributionthestudenttdistributionisused.
95%
95%
10
x
o A = (9.19)
( )
50% 95%
95%
log y a
x
b

= (9.20)
95% 50% 0.95;n f

a a t s
= (9.21)
Inwhich:
t
0,95;nf
Thestudenttdistributionfactorfornfdegreesoffreedomanda95%probabilityofsurvival.nistheamountof
datapoints,fis1forafixedbvalueand2foracalculatedbvalue.[]
Brozettihasstatedsomeconditionsfortheseformulastogiveareliableresult.Runoutsshouldnotbeconsidered,where
allresultswith N
f
>510
5
shouldbeconsideredassuch.TheslopeoftheSNcurveshouldbefixedto3andtheminimum
samplesizeis12.Becauseliteraturehasshownthattheuseofhighstrengthsteel(Pijpers,2011)andTIGdressing(Dahle,
1998)canleadtohighermvalues,twoanalyseshavebecarriedoutforeachmaterial,onewiththeprescribedfixedslope
of3,andonewheretheslopeisoptimized.ThesamplesizesofallfittedSNcurveswillbemuchlowerthantheprescribed
valueof12.Becauseofthesedifferences,theresultsshouldnotbeseenasnormative,butonlyasindicative.
99
9.3 FittedSNcurves
9.3.1 Rawdata
In the following graphs, SN curves have been fitted for all non excluded data (see 9.2.1). This means that threefailure
mechanismsareconsidered:
Weldtoefailure
Weldmaterialfailure
Basematerialfailure
Allresultsareplottedtogetherfortworeasons.Firstofall,thenumberofdatapointsforeachfailuremechanismwould
be very small, sometimes only one or two if each failure mechanism would be considered separately. This would make a
decent analysis impossible. Secondly, the TIGdressing is expected to significantly improve the fatigue resistance of the
weld toe. This can lead to a change in the dominant failure mechanism, which still leads to the failure of the connection
overall.Thefatiguetestshaveshownthatnexttothetwofailuremechanismswhichwereobservedduringtheaswelded
tests (weld toe and base material) another failure mechanism has presented itself: weld material failure. This mechanism
hasnotbeenobservedintheasweldedtests,buthasbeenresponsibleforsevenofthetwentyfailedspecimensduringthe
TIGdressed tests. Apparently the weld toe fatigue strength has been improved, causing failure in the next weakest link.
Because all three failure mechanisms lead to failure of the component, and non practical failures have been excluded, all
dataisanalyzedtogether.
ThecharacteristicvaluesoftheSNcurvesthatresultfromthestatisticanalysisareshownintable9.1.TheSNcurvesare
plottedwiththeaccompanyingdatapointsinfigure9.3tofigure9.6.
calculatedslope fixedslope

mean
[N/mm
2
]
95%
[N/mm
2
] m[]
mean
[N/mm
2
]
95%
[N/mm
2
] m[]
S460 154 120 2.72 163 136 3
S690 212 165 4.45 171 126 3
S890 204 97 3.14 200 119 3
S1100 197 106 3.15 189 125 3
table9.1CharacteristicvaluesoffittedSNcurves.rawdata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46
C46
runout
*
BMfailure
m=2.72[50%]
m=2.72[95%]
m=3[50%]
m=3[95%]
figure9.3FittedSNcurvesforS460,rawdata
100
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V69
C69
runout
*
BMfailure
m=4.45[50%]
m=4.45[95%]
m=3[50%]
m=3[95%]
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89
C89
runout
*
BMfailure
m=3.14[50%]
m=3.14[95%]
m=3[50%]
m=3[95%]
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11
C11
runout
*
BMfailure
m=3.15[50%]
m=3.15[95%]
m=3[50%]
m=3[95%]
9.3.2 Adjusteddata
Inthefollowinggraphs,SNcurveshavebeenfittedforallnonexcludeddataafteradjustment.Forthesamereasonsas
explainedin9.3.1,alldatapointswillbeconsideredinonegraph.ThecharacteristicvaluesoftheSNcurvesthatresultfrom
101
thestatisticanalysisareshownintable9.2.TheSNcurvesareplottedwiththeaccompanyingdatapointsinfigure9.7to
figure9.6.

mean
[N/mm
2
]
95%
[N/mm
2
] m[]
mean
[N/mm
2
]
95%
[N/mm
2
] m[]
S460 136 102 2.53 152 123 3
S690 169 129 4.06 142 107 3
S890 145 61 2.73 153 91 3
S1100 141 76 3.05 140 92 3
table9.2CharacteristicvaluesoffittedSNcurves.adjusteddata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46
C46
runout
*
BMfailure
m=2.53[50%]
m=2.53[95%]
m=3[50%]
m=3[95%]
figure9.7FittedSNcurveforS460,adjusteddata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V69
C69
runout
*
BMfailure
m=4.06[50%]
m=4.06[95%]
m=3[50%]
m=3[95%]
102
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89
C89
runout
*
BMfailure
m=2.67[50%]
m=2.67[95%]
m=3[50%]
m=3[95%]
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11
C11
runout
*
BMfailure
m=2.70[50%]
m=2.70[95%]
m=3[50%]
m=3[95%]
9.3.3 Discussion
Someresultsshowsignificantdeviationfromthetotaldataset(forexample,theC89pointnearN=110
5
).Someofthese
pointscontainbasematerialfailure,butothers,suchastheexamplepointinfigure9.5,areothermechanisms.Becauseof
thelimitedsizeofthedatasetitishardtodeterminewhetherthesepointsareindeedarareexceptionorpartofawidely
scattereddataset.Exceptfortheexcludeddataasdescribedin9.2.1,nootherdatapointsareexcludedfromtheregression
inthisparagraphorcomingparagraphs.
The small data set leads to high values of t
0,95;nf
(see 9.2.3), resulting in large difference between the mean and
characteristiccurve.Thiseffectisevenstrongerbecauseofthelargeinfluencesingledatapointscanhaveonthestandard
deviationduetothesmallsamplesize.
Furthermore,itmustbenotedthatmostobservedrunoutsliegreatlyabovethefatiguelimitwhichisderivedfromthe
valueofthefittedSNcurveat510
6
cycles.Itisthereforequestionablewhetherthechosenkneepointiscorrect.Dueto
thelimitednumberofresults,especiallyforN
f
>110
6
onlyapresumptionwithrespecttothekneepointcanbeexpressed.
9.4 Comparison
9.4.1 Comparisonofdatawithdesigncodes
Theconnectionthatwastested,isjudgedbydifferentdesigncodesindifferentways.Intable9.3thecharacteristicdata
foragivenprobabilityofsurvival(P
s
)areshownforthedesignSNcurvesprescribedbydifferentcodes.Thedesigncurves
areplottedandcomparedwithoutanyfurthersafetyfactors.
103

c
[N/mm
2
] m[] P
s
[]
Eurocode* 71 3 95.0%
IIW** 92 3 95.0%
NPRCEN/TS1300131*** 100 3 97.7%
table9.3CharactericvaluesofdesignSNcurvesforthetestedspecimens
*WeldscheckedbyNDTfrombothsides,safetyfactornotconsidered.Design curvelimited tosteelsupto R
eh
=700
N/mm
2
** Welds checked by NDT and misalignment <10% of plate thickness. This criterion is not met in the specimens.
DesigncurvelimitedtosteelsuptoR
eh
=900N/mm
2
***Weld quality B. Detail category for asymmetric butt welds also incorporates asymmetric plate arrangement,
whichisnotthecasewiththetestedspecimens.DesigncurvelimitedtosteelsuptoR
eh
=960N/mm
2
When table 9.1, table 9.2 and table 9.3 are compared, it is clear that the very high strength steel specimens (S890 and
S1100)donotalwaysmatchthecodeswhenthecharacteristicstrengthsarecompared.However,thecharacteristiccurve
valueisinfluencedtoagreatextentbytheuncertaintyofthedatacausedbythesmallsamplesize.Below,alldatapoints
willbecomparedwiththedesignlinesspecifiedintable9.3.
Rawdata
In figure 9.11 to figure 9.14 the raw data points are compared with the design lines of the different codes and
recommendationsasspecifiedintable9.3.
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46
C46
runout
*
BMfailure
EN199319
IIWrecommendations
NPR_CEN/TS1300131
figure9.11S460datapointscomparedwithdesignlines,rawdata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[N/mm
2
]
V69
C69
runout
*
BMfailure
EN199319
IIWrecommendations
NPRCEN/TS1300131
104
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89
C89
runout
*
BMfailure
EN199319
IIWrecommendations
NPR_CEN/TS1300131
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11
C11
runout
*
BMfailure
EN199319
IIWrecommendations
NPRCEN/TS130010301
Adjustedresults
In figure 9.15 to figure 9.18 thedata points after adjustment are compared with the design lines of the different codes
andrecommendationsasspecifiedintable9.3.
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46
C46
runout
*
BMfailure
EN199319
IIWrecommendations
NPRCEN/TS1300131
figure9.15S460datapointscomparedwithdesignlines,adjusteddata
105
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V69
C69
runout
*
BMfailure
EN199319
IIWrecommendations
NPRCEN/TS1300131
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89
C89
runout
*
BMfailure
EN199319
IIWrecommendations
NPRCEN/TS1300131
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11
C11
runout
*
BMfailure
EN199319
IIWrecommendations
NPRCEN/TS1300131
Discussion
All specimens have performed according to the three considered design codes or better, both before and after
adjustmentofthedata.Itisnoteworthythatallfoundrunoutspecimensshowthisbehaviouratasignificantlyhigherstress
106
range(factor34forrawdataandafactor23foradjustedresults)thanspecifiedinthedesignlines.Aphenomenonwhich
wasalsonoticedwhenthedatapointswerecomparedwiththefittedSNcurves(see9.3).
9.4.2 Comparisonofdatawithasweldedfatiguetests
From the TIGdressed specimens all non excluded data are considered, for reasons explained in 9.3. In the study by
Pijpers(2011),dataweresplitaccordingtotheirfailuremechanism(basematerialorweldtoe).Theargumentstoinclude
alldataintheanalysiswhichwerevalidforTIGdressedspecimens,arenotfortheasweldeddata.Therefore,inprincipleit
isonlyinterestingtocomparetheresultsofthisresearchwithcapweldtoefailuresfromtheresearchbyPijpers.However,
allbasematerialfailuresthatwerefoundinthisresearchcanbetheresultofremovingtheweakestlinkfromthechainby
improving the weld toe, but could also already have been the weak link before TIGdressing. Therefore, the weld toe
failures from the cap side are plotted but also all base material failures are plotted. As was the case for the TIGdressed
datapoints,thebasematerialfailureswillbemarkedinalldiagrams.
Fullfatiguelife,rawdata
In table 9.4 the mean stress range and characteristic stress range at 210
6
cycles are depicted for the as welded

specimens and the TIGdressedspecimens. A graphic representation of the table can be found in figure 9.19. Although in
figure9.20tofigure9.23theTIGdresseddatapointsseemtolieinthemiddleandupperregionoftheasweldedscatter
band,themeanandcharacteristicvaluesintable9.4andfigure9.19donotshowastrengthincreaseduetoTIGdressingin
most materials. This strength decrease may be explained by the fact that most as welded specimens were tested under
bending,whichingeneralgivesbetterresultsinfatiguetests(see9.2.2).Theadjusteddatathereforemayshowadifferent
behaviour.ThedifferencebetweenasweldedandTIGdressedspecimensincreaseswhenthecharacteristicstressrangeis
considered for most materials. This can be caused by the fact that a small data set results in more uncertainty of the
variation, and therefore results in a lower characteristic stress range. Furthermore, the flattening of the SN curve due to
TIGdressing (see 4.3.4) cannot be observed in most results. Only the steel grade S690 behaves in a way that can be
expected, based on literature research: the mean and characteristic strength increase, and the mvalue of the SN curve
increases.

meanv
[N/mm
2
]
95%
[N/mm
2
] m[]
mean
[N/mm
2
]
95%
[N/mm
2
] m[]
S460TIG 154 120 2.72 163 136 3
S460AW 192 165 3.99 171 138 3
S690TIG 212 165 4.45 171 126 3
S690AW 193 124 3.68 183 108 3
S890TIG 204 97 3.14 200 119 3
S890AW 259 222 6.95 208 138 3
S1100TIG 197 106 3.15 189 125 3
S1100AW 169 114 2.56 180 130 3
table9.4Comparisonofmeanandcharacteristicvaluesat210
6
cyclesbeforeandafterTIGdressing,rawdata

0
50
100
150
200
250
S460TIG S460AW
0
50
100
150
200
250
S690TIG S690AW
0
50
100
150
200
250
S460TIG S460AW
0
50
100
150
200
250
S690TIG S690AW
0
50
100
150
200
250
S890TIG S890AW
0
50
100
150
200
250
S1100TIG S1100AW
0
50
100
150
200
250
S890TIG S890AW
0
50
100
150
200
250
S1100TIG S1100AW
figure9.19Graphicoverviewoftable9.4:meanandcharacteristicstressrangeat210
6
cycles,rawdata
107
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46TIG
C46TIG
V46AW
C46AW
runout
*
BMfailure
figure9.20ComparisonbetweenasweldeddataandTIGdresseddata,rawdata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[N/mm
2
]
V69TIG
C69TIG
V69AW
C69AW
runout
*
BMfailure
figure9.21ComparisonbetweenasweldeddataandTIGdresseddata,rawdata
*
*
* *
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89TIG
C89TIG
V89AW
C89AW
runout
*
BMfailure
figure9.22ComparisonofasweldeddataandTIGdresseddata,rawdata
108
*
* *
*
*
*
*
*
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11TIG
C11TIG
V11AW
C11AW
runout
*
BMmaterial
figure9.23ComparisonofasweldeddataandTIGdresseddata,rawdata
Fullfatiguelife,adjustedresults
In table 9.5 the mean stress range and characteristic stress range at 210
6
cycles are depicted for the as welded

specimensandtheTIGdressedspecimensafteradjustmentofthedatapoints.Agraphicrepresentationofthetablecanbe
found in figure 9.24. In the figure can be seen that still for most materials TIGdressing does not seem to provide a
significantincreaseinfatiguestrengthat210
6
cycles,exceptforsteelgradeS690.Itisremarkablethatthefatiguestrength
increase at 210
6
cycles of TIGdressed specimens seems to become larger when the SN curve with a fixed slope is
considered(seefigure9.24,righthandside),whileafreeslopeisexpectedtoperformbetterduetotheflatteningeffectof
theSNcurve(see4.3.4).Thisflatteningeffectisnotapparentintheresults.Whenlookingatthedatapointsinfigure9.25
tofigure9.28itisclearthattheTIGdresseddatapointsgenerallylieintheupperpartoftheasweldedscatterbandwith
the S1100 data points as possible exception. Because the mvalue of the curves decreases for all results and most data
points lie in the region below 110
6
cycles, the characteristic strength at 210
6
is strongly influenced. Based on the
observationsofthescatterplots,atpositiveeffectofTIGdressingislikely.

mean
[N/mm
2
]
95%
[N/mm
2
] m[]
mean
[N/mm
2
]
95%
[N/mm
2
] m[]
S460TIG 136 102 2.53 152 123 3
S460AW 150 127 4.72 127 96 3
S690TIG 169 129 4.06 142 107 3
S690AW 144 106 4.72 129 79 3
S890TIG 145 61 2.73 153 91 3
S890AW 168 122 3.77 155 106 3
S1100TIG 141 76 3.05 140 92 3
S1100AW 112 65 2.14 131 87 3
6
cyclesbeforeandafterTIGdressing,adjusteddata
109

0
50
100
150
200
250
S460TIG S460AW
0
50
100
150
200
250
S690TIG S690AW
0
50
100
150
200
250
S460TIG S460AW
0
50
100
150
200
250
S690TIG S690AW
0
50
100
150
200
250
S890TIG S890AW
0
50
100
150
200
250
S1100TIG S1100AW
0
50
100
150
200
250
S890TIG S890AW
0
50
100
150
200
250
S1100TIG S1100AW
6
cycles,adjusteddata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V46TIG
C46TIG
V46AW
C46AW
runout
*
BMfailure
figure9.25ComparisonofTIGdressedandasweldeddata,adjusteddata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V69TIG
C69TIG
V69AW
C69AW
runout
*
BMfailure
110
*
*
* *
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V89TIG
C89TIG
V89AW
C89AW
runout
*
BMfailure
*
*
*
*
*
*
* *
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
V11TIG
C11TIG
V11AW
C11AW
runout
*
BMfailure
Crackinitiationlife,rawdata
BecauseTIGdressingisexpectedtoinfluencethecrackinitiationlifetoagreaterextentthanthecrackpropagationlife,
thecrackinitiationlivesoftheasweldeddatawillbecomparedwiththecrackinitiationlivesoftheTIGdresseddata.NoS
Ncurvesarefittedtothedata,butonlytheoveralleffectofTIGdressingwillbediscussed.
The earlier discussed trend, where the TIGdressed data points mainly lie in the middle and upper region becomes
somewhat more distinct in figure 9.29 to figure 9.32. Another way to analyze the effect of TIGdressing on the crack
initiation time is to compare the average crack initiation life ratios (f
Ni
) of the TIGdressed data set and the as welded
dataset.Intable9.6thiscomparisonismade,andbasedonthistableitseemsclearthattheTIGdressinghasincreasedthe
crackinitiationlife.Animportantremarktothistablemustbemade.ThespecimenstestedbyPijperswereprovidedwith
threestraingaugesateachweldtoe,insteadoftwointhisresearch,butarealsowider.Ifacrackinitiatesclosertoastrain
gaugeitcanbedetectedearlier.TwostraingaugesinPijpersresearchwereplacedattheedgeofthespecimen,wherea
lot of cracks were found in that research. The results in table 9.6 may be distorted due to these differences, but it is not
suretowhatextentthesegeometricaldifferencesbetweenthespecimensinfluencetheresult.
f
Ni
TIG[] f
Ni
AW[]
S460 0.70 0.49
S690 0.64 0.40
S890 0.83 0.61
S1100 0.83 0.64
table9.6Comparisonoff
Ni
ofTIGdressedspecimensandasweldedspecimens.Runoutsarenotconsidered
111
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V46TIG
C46TIG
V46AW
C46AW
runout
*
BMfailure
figure9.29ComparisonofcrackinitiationlifeofTIGdressedandasweldeddata,rawdata
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V69TIG
C69TIG
V69AW
C69AW
runout
*
BMfailure
*
*
* *
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V89TIG
C89TIG
V89AW
C89AW
runout
*
BMfailure
112
*
**
*
*
*
*
*
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V11TIG
C11TIG
V11AW
C11AW
runout
*
BMfailure
Crackinitiationlife,adjustedresults
Whentheadjustedresultsareconsidered,thesameeffectasforrawdatacanbeobservedbecausethecrackinitiation
life ratios are the same. The TIGdressed specimens lie in the upper regions of the as welded scatter band. This effect is
somewhatclearerthanfortheadjustedfullfatigueliferesults.
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V46TIG
C46TIG
V46AW
C46AW
runout
*
BMfailure
figure9.33ComparisonofcrackinitiationlifeofTIGdressedandasweldeddata,adjusteddata
113
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V69TIG
C69TIG
V69AW
C69AW
runout
*
BMfailure
*
*
* *
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V89TIG
C89TIG
V89AW
C89AW
runout
*
BMfailure
*
*
*
*
*
*
* *
*
*
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V11TIG
C11TIG
V11AW
C11AW
runout
*
BMfailure
114
Discussion
ThescatterplotsoftherawdatashowlittleornobenefitduetoTIGdressing.Whentheresultsareadjustedaccordingto
9.2.2, the effect of TIGdressing seems positive, based on observations on the scatter plots. A remark must be made that
theusedcorrectionfactorfortheloadingmodehasbeendefinedasconservativebyPijpers.Thismeansthatitispossible
thatslightlyhigherfatiguestrengthshouldbeattributedtotheasweldedresultswhichwereloadedinpurebending.
When SN curves are fitted to the results, the TIGdressed results show a steeper SN curve most of the time. Because
most results are located below 110
6
cycles, this steeper curve results in lower mean and characteristic stress ranges at
210
6
cyclesforTIGdresseddatawhencomparedwithasweldedvalues.WhentheslopeoftheSNcurveisfixedtoavalue
of3,theresultsforTIGdresseddatabecomelessunfavourable,butaclearstrengthincreasecannotalwaysbeobserved.
To investigate the possible positive influence of TIGdressing in the medium cycle regime, the fitted SN curves of both
theasweldeddataandTIGdresseddataareplottedinfigure9.37.
S460
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
S690
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
S890
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
S1100
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
f
[]
[N/mm
2
]
figure9.37ComparisonoffittedSNcurves.Adjusteddata,calculatedslope.
Fromfigure9.37canbeconcludedthatespeciallyinthemediumcyclerangeabeneficialeffectduetoTIGdressingcan
beobservedformostmaterials.Therefore,themeanandcharacteristicvaluesofthefittedSNcurvesatN=410
5
(instead
ofN=210
6
)aredepictedintable9.7andfigure9.38.Thistableanddiagramclearlyshowameanfatiguestrengthincrease
duetoTIGdressingformaterialsS460andS690.Thetwosteelgradeswhichshowedalotoffailuresinthebasematerial,
S890andS1100stillperformaboutsimilartotheasweldedspecimens.Thecharacteristicvaluesofallsteelgradesdonot
alwaysshowtheresultsasdescribedabove,butthismaycoherewiththesmallerdatasetoftheTIGdressedspecimens,as
explainedearlier.

mean
[N/mm
2
]
95%
[N/mm
2
] m[]
mean
[N/mm
2
]
95%
[N/mm
2
] m[]
S460TIG 257 192 2.53 260 210 3
S460AW 211 178 4.72 217 164 3
S690TIG 251 191 4.06 243 183 3
S690AW 203 150 4.72 221 135 3
S890TIG 261 110 2.73 262 156 3
S890AW 257 187 3.77 265 181 3
S1100TIG 239 129 3.05 239 157 3
S1100AW 238 138 2.14 224 149 3
5
cyclesbeforeandafterTIGdressing,adjusteddata
115

0
50
100
150
200
250
S460TIG S460AW
0
50
100
150
200
250
S690TIG S690AW
0
50
100
150
200
250
S460TIG S460AW
0
50
100
150
200
250
S690TIG S690AW
0
50
100
150
200
250
S890TIG S890AW
0
50
100
150
200
250
S1100TIG S1100AW
0
50
100
150
200
250
S890TIG S890AW
0
50
100
150
200
250
S1100TIG S1100AW
5
cycles,adjusteddata
The exceptions to the trend described above are the data from S690 steel. Here, the data points in the scatter plot
behavebetterandthefittedSNcurvesalsoshowastrengthincreaseduetoTIGdressingathighercycles(210
6
).Thefitted
SNcurveoftheTIGdresseddatahasahighermvaluethanthefittedSNcurvetotheasweldeddataincaseoftheraw
data.Whenthetabulatedfatiguetestdataareconsulted(seeAnnexC)itcanbeseenthatallfailedS690specimensfailed
attheweldtoe.Nobasematerialorweldmaterialfailureshavebeenobservedinthesespecimens.Closerinspectionofthe
weld leads to no additional information: the welds in S690 look similar to the welds of S460, where almost only weld
material failures occurred. It may be interesting to analyze the geometry of the weld cap with the aid of the already
available laser measurement data in future research. Also any changes in residual stress in the weld material may have
occurred due to the TIGdressing procedure, which, if it concerns a residual stress increase, can lead to worse fatigue
behaviouroftheweldmaterialthanintheasweldedcase.
An opposite trend can be seen in the S1100 specimens. Only one S1100 specimen has failed at the weld toe, and all
othersfailedinthebasematerialandtheS1100TIGdresseddatapointsmoreorlessblendinwiththeasweldedresults,
especiallyathighcycles.
It may be possible to explain these observations in the following way. TIGdressing has improved the weld toe and
removed it as weakest link from the chain. For specimens where even after TIGdressing the weld toe was the, now
stronger, weakest link, an improvement in fatigue strength can be observed. Specimens which have weak links with
marginallyhigherfatiguestrengthelsewhereinthebasematerialorweldmaterialshowonlymarginalimprovement.Ifthe
weaklinkintheasweldedspecimenwasplacedoutsidetheweldregion,resultinginabasematerialfailure,thiswillstillbe
thecaseintheTIGdressedsituation.Inthatcase,notimprovementisobserved.Inthegraphsofthesteelsthatshowbase
material failures, it can be observed that the base material failures indeed blend in with the as welded results, while the
rareweldtoeorweldmaterialfailureslieslightlyabovetheasweldedscatterband.
The crack initiation life of the TIGdressed specimens seems to be improved with respect to the as welded specimens,
resulting in a higher N
i
/N
f
ratio. However, due to differences in geometry between the TIGdressed specimens and the as
weldedspecimens,theseresultsmayhavebeendistorted.
9.4.3 Comparisonofdatawithanalyticaldeterminationoffatiguestrength
Theusedanalysismethodisonlyvalidforthecrackinitiationlifeofweldtoecracks.Therefore,onlythecrackinitiation
lifeofspecimensfailedattheweldtoeandrunoutswillbecomparedwiththeanalyticalresults.Thisselectionresultsina
relatively limited number of data points which can be compared for the TIGdressed case, because several specimens
showedotherfailuremechanismsthanaweldtoefailure(see7.3).
Furthermore, the analysis method contains corrections for the means stress. The results of the analysis will therefore
onlybecomparedwiththeresultsofthetestscorrectedformeansstress,thicknessandloadingmode.
Comparisonofasweldeddatawithmodelbasedonpeakvaluesofasweldedgeometry
Infigure9.39tofigure9.49thecomparisonismadebetweenthemodelasdescribedandderivedinchapter8andthe
fatigue test results by Pijpers in the as welded situation. Almost all predicted results lie significantly higher than the test
results.EspeciallythemodelasderivedbySchijveshowsverydifferentbehaviourthanreality.Whenthekneepointinthe
modelbySchijvewouldbeadjustedto10
6
cycles(asisrecommendedfornotched,notweldedspecimens)itwouldshow
similarresultsasthemodelderivedbyHck.
116
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V46AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
figure9.39ComparisonofcrackinitiationlifeofV46specimenswithweldtoefailureandmodeledfatiguestrength
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C46AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
figure9.40ComparisonofcrackinitiationlifeofC46specimenswithweldtoefailureoncastsideandmodeled
fatiguestrength
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C46AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
figure9.41ComparisonofcrackinitiationlifeofC46specimenswithweldtoefailureonrolledsideandmodeled
fatiguestrength
117
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V69AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C69AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C69AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
118
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V89AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C89AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C89AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
119
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V11AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C11AW
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
ComparisonofTIGdresseddatawithmodelbasedonpeakvaluesofTIGdressedgeometry
fatigue test results in the TIGdressed situation. Similarly to the as welded situation, almost all predicted results lie
significantlyhigherthanthetestresults.Itseemsthatthediscrepancybetweenthemodelandtheresultsisevenlargerin
theTIGdressedsituation.
120
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C46TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V69TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C69TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
121
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V11TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C11TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
ComparisonofTIGdresseddatawithmodelbasedonpeakvaluesofTIGdressedgeometryandasweldedradius
fatiguetestresultsintheTIGdressedsituation.Theresultsshowasimilarbehaviourasfortheprevioustwocomparisons
withthemodel.BecauseahigherstressconcentrationfactorandfatiguenotchfactorisattributedtotheTIGdressedweld
toes, the model and actual tests approach each other better than in the previous analysis where only peak values were
considered.However,becausetheasweldedresultsalsodidnotmatchverywell,thisisnovalidreasontoassumethatthe
modelwithpeakvaluesoftheTIGdressedgeometrywithanasweldedradiusisthebestmodeloftheTIGdressedjoint.
122
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C46TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V69TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C69TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
123
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
V11TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
10
100
1000
1.0E+04 1.0E+05 1.0E+06 1.0E+07
N
i
[]
[N/mm
2
]
C11TIG
runout
Schijve(peak)
Schijve(adjusted)
Hck(peak)
Hck(adjusted)
fatiguestrength
Discussion
Almostalltestresults,bothasweldedandTIGdressedshowsignificantdifferenceswiththemodelledbehaviour.Theas
weldedmodelseemstoperformthebest.BothTIGdressedmodelsperformalittlebitworse,withthepeakvaluemodel
beingtheleastdependable.
Thesensitivityofthemodeltoslightdifferencesintheweldtoeparametersisrelativelylimitedascanbeseenfromthe
figures.Thegreylinesindicateavariedweldtoetoanextentthatthebulkofallweldtoesarecovered(see8.3),butthis
only results in a rather small fatigue strength variation. The variation that is found is smaller than a typical fatigue test
scatterband(seethefiguresabove).
The use of the TIGdressed model with the as welded radius instead of all peak values leads to a difference of 1320%,
depending on the steel grade. The model predicts a fatigue strength increase due to TIGdressing of 30% for S460, which
steadily increases to a 52% strength increase for S1100. This fatigue strength increase difference is partly related to the
differenceinK
f
andispartlycausedbytheassumedhigherabsolutereductionofresidualstressesinhigherstrengthsteels.
Duetotheinaccuracyofthemodel,itisveryhardtoratethesenumbersattheirtruevalue.However,whenthesevalues
arecomparedwithfigure9.38,itseemsthatthemodeloverestimatesthebeneficialeffectofTIGdressing.
TomakeagoodestimationofthevalidityofthemodelfortheTIGdressedspecimens,itisimportantthattheaswelded
modelgivesgoodpredictionsofthefatiguestrength.Becausethisisnotthecase,itisveryhardtoevaluatetheaccuracyof
theTIGdressedmodel.Whentheseissuesaresetasideandonlytherelativedifferencebetweentheasweldedmodeland
TIGdressedmodelisconsidered,itisnoticedthatthetestdataliesfurtherbeneaththepredictedlineinthe TIGdressed
situationthanintheasweldedsituation.Thisisexpectedtobetheotherwayaround,becauseonlythechangeingeometry
124
and residual stresses is incorporated in the model. The removal of weld flaws should also give a beneficial effect, which
shouldleadtoatestedfatigueresistancewhichishigherthanthemodel.Unfortunately,thisisnotthecase.
AremarkmustbemadeontheusedmodeltodeterminedK
f
.Asisexplainedin3.5.1,literaturerecommendstheuseof
a* in the determination of K
f
, because this leads to the maximum value of K
f
and thus the most conservative result.
Becausetheactualgeometryoftheweldwasmeasuredinthisstudy,thevalueoftheradiustodetermineK
f
wasbasedon
these measurements and not on the most conservative value possible. When the radius would be set at this worst case
value an increase in K
f
would be found, resulting in a lowering of the predicted SN curves, thereby increasing the model
accuracy.TheincreaseofK
f
canvaryfromafewpercentforS460formorethan30%forS1100.Theactualvaluecouldnot
bedeterminedbecausethea*valueofthismaterialissosmall,thattheaccompanyingradiushasnotbeenmodelled.The
indication of 30% increase in K
f
is based on the smallest radius that has been modelled. However, when this approach
would be used, the measurement of the actual weld toe radius is no longer used, which partly eliminates the beneficial
effectfortheTIGdressedspecimens.
To omit this question, another approach could also be used. It is possible to link the fatigue notch factor to the stress
gradient.Thismethodisrelatedtothehighlystressedvolumeapproachasdiscussedin3.5.1,buthasnotbeenconsidered
in this research. A more thorough analysis of the possibilities to determine K
f
, while incorporating the measured weld
profileshouldbesubjectoffurtherinvestigation.
9.5 EvaluationofTIGdressinginfluence
ThereareseveralindicationsthattheTIGdressingprocedurehashadapositiveeffectonthefatiguestrengthoftheweld
toe.Firstofall,anewfailuremechanismhasshown,whichwasnotobservedintheasweldedspecimenstestedbyPijpers.
Sevenofthetwentynonexcludedresultsshowedthisnewfailuremechanism:acrackintheweldmaterial.Whetherthisis
only caused by weld toe improvement, or TIGdressing also had influence on the weld material, possibly by increasing
residual stresses, may be the subject of further investigation. Other specimens still showed failure at the weld toe, but
these data points can mostly be found in the upper regions or above the scatter band of the as welded specimens, thus
suggestingafatiguestrengthimprovement.Finally,somespecimensfailedinthebasematerial.Whetherthisbasematerial
was already a weak link in the chain before TIGdressing or it became the weakest link after the weld toe was improved
cannot be proven. However, the scatter plots in 9.4.2 show that the basematerial failures more or less blend inwith the
basematerialresultsfromtheasweldedtests.Thisindicatesthattheobservedfailurelocation wouldalsohavebeenthe
weakestlinkinanasweldedsituation.Thismightbeapossibleexplanationfortherelativelypoorperformanceofsomethe
TIGdressed specimens of the steel grades S890 and S1100 when compared with the fatigue strength increase of lower
strength specimens. Apparently the, very likely improved, weld toe was not the weakest link, even in the as welded
situation.Onespecimenthatwasexcludedfromtheanalysis,specimenV111,mayalsofallintothiscategory.Herealsoa
basematerialfailurewasobserved,whichwouldbetheweakestlink,bothintheasweldedstateandtheTIGdressedstate.
TheoveralltrendseemstobeanimprovementattheweldtoeasaresultofTIGdressingespeciallyinthemediumcycle
regime. Clear information on the high cycle regime is not available, due to the scarcity of test data is this regime. Due to
otherfailuremechanisms,notnecessarilythewholespecimenshowsfatiguestrengthincrease.Iftheweldtoeisimproved,
butthenextweaklinkonlybehavesmarginallybetter,theTIGdressinghaslimitedinfluence.Itmustalsobenotedthatthe
runouttestresultsmostlylieintheupperregionoftheasweldedrunouts,butgreatlyabovethefittedfatiguelimitsorthe
fatiguelimitsspecifiedbycodes.ThisisnottranslatedintothefittedSNcurves,andthereforeworthanextraremark.
The analytical results do not comply with the test data in a sufficient way to use the developed model to predict the
fatiguestrengthofcomponents.Thetestresultsdeviatesignificantlyfromthepredictedvaluesandthemodeledresultsare
nonconservative.Fromtheanalyticalapproachitisfoundthatvariationsoftheweldtoeparameterswithinsuchlimitsthat
thebulkofallmeasuredweldtoesarecovered,haverelativelylowconsequencesonthefatiguestrength.Thedifferences
that are foundwhen such an analysis is madeare generally smaller than a typical fatigue test scatter band.To effectively
predictthefatiguestrengthofbuttweldedspecimens,extraresearchisnecessary.
Because of the limited number of tests per material and the limited accuracy of the model, this report refrains from
strengthincreasefactorsandonlytheoveralltrendhasbeendescribed.Forthesamereason,andbecauseS890andS1100
donotshowaclearstrengthimprovementasdiscussedearlier,nomaterialdependentimprovementisdescribed.
To define strength increase factors for S460 and S690, and to examine whether S890 and S1100 specimens as a whole
showastrengthincreasemoretestingisnecessary,especiallyinthehighcycleregime.
125
10 Conclusionsandrecommendations
10.1 Conclusions
10.1.1 InfluenceofTIGdressingoffatiguestrengthofbuttweldedspecimens
Design codes do not always incorporate a fatigue strength increase due to TIGdressing. The fact that all TIG
dressedspecimensshowafatiguestrengthabovethedesignlineoftheconsidereddesigncodesisthereforenot
surprising.
WhentheTIGdressedspecimensarecomparedwithsimilarasweldedspecimens,thescatterplotsshowatrend
towards a better fatigue behaviour of the TIGdressed specimens for all specimens which have shown failure in
theweldarea.ForsteelgradesS460andS690thisconcernsalmostallspecimens.Forthehigherstrengthsteelsa
lotofspecimenshaveshownbasematerialfailure.Quantitativeconclusionscannotbedrawnduetotherelatively
smallnumberoftests.
Thematerialsthathaveshownbasematerialfailuresintheasweldedcondition,S890andS1100,showlittleto
no improvement of the fatigue strength, because any beneficial effect of the weld toe improvement can only
haveaneffectonspecimensthatwouldnothaveshownbasematerialfailureintheasweldedcondition.Thefew
specimens that have shown weld material failure after TIGdressing in these steel grades generally show the
abovedescribedtrendtowardsfatiguestrengthimprovement.
Thefailuremechanismintheweldmaterialhasbecomemuchmoreprominent.Thisfailuremechanismhasonly
very rarely been observed in the as welded tests, but has been observed on a significant number of the TIG
dressedspecimens.Thisisaverygoodindicationthattheweldtoeindeedhasbeenremovedastheweakestlink
fromthechain.WhethertheweldmaterialissimplythenextweaklinkorhasbeeninfluencedbyTIGdressingis
notclear.
The main fatigue improvement of TIGdressed specimens is expected from improvement of the crack initiation
life. When scatter plots of TIGdressed and as welded specimens are compared, the difference between the as
weldedscatterbandandtheTIGdressedspecimensseemslargerwhenonlythecrackinitiationlifeisconsidered,
thus confirming this expectation. It must be noted that differences in geometry of the specimens and
instrumentation which were present, can distort this graph because they influence the accuracy with which the
crackinitiationtimecanbedetermined.
10.1.2 InfluenceofTIGdressingonweldgeometryofbuttweldedspecimens
Asweldedweldtoegeometriesoftheconsideredspecimensgenerallyshowonegeometryvariant.Whenthese
weldsareTIGdresseddifferentweldtoegeometriescanbeobserved(seefigure6.9).Differentvariantsoccurin
differentfrequenciesbutnosinglegeometryvariantdominatesthemeasurements.
The peak value of the distribution of weld toe radii increases from ~1.0 millimeter for as welded specimens to
~3.8 millimeter for TIGdressed specimens. The TIGdressed results also show a peak in the region of the as
weldedpeak.ThespreadofweldtoeradiiincreasesduetoTIGdressing.
Thepeakvalueofthedistributionofweldtoeanglesdecreasesfrom~27.8forasweldedspecimensto~15.3for
TIGdressedspecimens.Thespreadofthedatabeforeandafterweldtoeimprovementissimilar.
The peak value of the distribution of undercuts is similar for the as welded specimens as for the TIGdressed
specimens and lies at 0 millimeter. The number of nonzero undercuts is somewhat larger for the TIGdressed
specimens. One specific geometry variant shows significantly deeper undercuts than all other considered weld
toe geometries, both as welded as TIGdressed. This geometry is also consistently associated with a very large
weldtoeradiusandthereforewillverylikelynotbetheweakestlinkinfatigue.
Because the width of the weld is much larger than the influence of TIGdressing, the weld improvement has no
influence on the weld height. Within one plate specimen the weld height shows only a slight variation. Larger
differencesinweldheightbetweenspecimenswerefound.
Thediscussedinfluences,withexceptionoftheundercutinfluences,leadtoanaverageweldtoegeometrywith
alowerstressconcentrationfactor.However,duetotheoccurrenceofanumberofsmallerweldtoeradiiinthe
TIGdressed specimens, which is disproportionate with respect to the overall distribution, the theoretical
improvement of the fatigue strength is not necessarily as high as is to be expected based on the average
improvement.
126
10.1.3 Theoretical influence of changed weld geometry on behaviour of TIGdressed
specimens
Thefoundspreadintheweldgeometrydatais,basedonasensitivityanalysis,notlikelytocauseascatterband
withthewidththatisusuallyassociatedwithfatiguetesting.
The fatigue notch factor shows reduction due to TIGdressing with slight dependency on the material static
strength. Higher strength steels show a greater reduction in fatigue notch factor, due to the difference in the
parameter a*. Further differences in fatigue strength improvement due to TIGdressing can be sought in other
factors,suchasresidualstressreduction.Intheusedmodelinthisresearchtheresidualstresswasassumedtobe
a fixed fraction of the yield strength, thereby increasing the fatigue strength increase differences between
differentsteelgrades
The model shows a reduction of the fatigue notch factor due to the changed weld toe geometry, even if no
improvementoftheweldtoeradiusisachieved.Thisreductionisthenmostlycausedbyareductionoftheweld
toeangle.
ThemodelthatisusedshowspooraccuracyforbothasweldedandTIGdressedspecimens.Also,thediscrepancy
betweenthemodelandtestedspecimensisnotconstant.Anyquantitativeorqualitativepredictionsofafatigue
strengthincreaseduetoTIGdressingcanthereforenotbeexpressed.
10.2 Evaluationandrecommendations
10.2.1 Assumptionsandapproximations
In this paragraph the main assumptions andapproximations will be described so that the conclusions from 10.1 can be
ratedattheirtruevalue.
All weld toe geometries have been analyzed together, while a few different typical geometries have been
observed. One model has been used to represent all TIGdressed geometries, leading to an approximation for
typicalTIGdressedgeometrytype3(seefigure6.9).
Noprobabilitydistributionhasbeenfittotheresultsoftheweldtoeparameters.Thepeakvaluesnowhavebeen
estimatedbasedonthebardiagrams.Fittingadistributionwouldmakeanaccuratecalculationofthepeakvalue
ofthedistributionpossible,butnotforalldatasetsanaccuratefitcouldbemade.Theaccuratecalculationofa
peakvaluewithafitteddistributionisthereforeanillusion.
Whenanalyzingtheweldtoegeometry,discussionaroseastowhetheralargenotchorasmallernotchinsidethis
mainnotchshouldbedocumented.Basedonvariousarguments,onlythemainnotchwasdocumented.
ThisresearchonlyfocusesonthegeometricalimprovementofTIGdressing.Anyimprovementbyremovingweld
toeflawsisnottakenintoaccount.Thereductionofresidualstressesisapproximated,basedonliterature.
Thecriticaldistanceapproachhasnotbeenproveninthefiniteliferegion.Thisresearchhasextendedtheinfinite
lifethathasbeendeterminedwiththisapproachintothefinitelife.
Peterson(1974)hasrecommendedtheuseofK
f;max
whendeterminingthefatiguestrengthofnotchedspecimens.
Thisresearchhasusedtheactualmeasuredgeometry.
Correction factors for thickness, loading mode, mean stress and residual stress have been acquired from the
researchbyPijpers(2011).Theinfluenceofthesefactorscanalsobedependentontheweldtoegeometry.Inthis
researchthesamecorrectionfactorsasusedbyPijperswereused,althoughitisnotcertainthatthesefactorswill
haveexactlythesamevalueforTIGdressedjoints.
Variousspecimenshaveshownfailureinsidetheweldmaterial.Thisfailuremechanismwasnotobservedinthe
asweldedjointsbyPijpers(2011).Therefore,nocorrectionfactorsforthickness,loadingmode,meanstressand
residualstressareavailable.Thesecrackshavebeenregardedasweldtoecracks.
10.2.2 Recommendationsforfurtherresearch
RecommendationsregardingtheTIGdressingprocedure
At some locations the TIG dressing procedure could not be executed because after grinding gas inclusions
appeared. To prevent them from expanding rapidly when heated and possibly ruining the weld, these locations
were not TIGdressed. It should be investigated whether these gas inclusions would be visible on an Xray test,
and in what frequency these inclusions appear. Furthermore, it should be investigated whether welds where
theseinclusionshavebeenobservedcanbeTIGdressedwithoutanyproblems.
DifferentTIGdressedgeometrieshavebeenobserved(seefigure6.9).Onlyoneofthesegeometriesisexpected
and described by the IIW and other literature. Further research should determine whether these different
geometriesalwaysoccurforbuttweldedspecimens.Theweldsthatwerestudiedinthisresearchwererelatively
fewinnumberandallexecutedbyonepersonunderthesameconditions.Togainamorecompleteimageofthe
possiblegeometries,thesefactorsshouldbevaried.
127
Recommendationsregardingthelasermeasurements
AllweldgeometrieshavebeenconsideredtogetherafterthesplitbetweenasweldedandTIGdressedwasmade.
Wheneachweldtoeparametersandtheirmutualdependencyarerecordedandcoupledtooneoftheweldtoe
variantsspecifiedinfigure6.9,thevalueofthedatabasewillincreasegreatly.
In a lot of weld toe geometries, different notches can be observed. In this study the choice is made to consider
only the largest main notch, based on a relatively quick assessment of the influence of the smaller notches.
Further research should show to what extent these smaller notches influence the fatigue behaviour and should
try to match the crack initiation sites on the specimens on any possible local smaller notches which may be
present.
In this research a very limited amount of parent material has been incorporated in the research to limit the
measuringtime.Extendingtheamountofmeasuredparentmaterialtoabout10millimeterwillgreatlyincrease
theabilitytoaccuratelydeterminetheweldtoeangle,weldtoeheightenundercut,becausealltheseparameters
aredefinedwiththeparentmaterialasbaseline.
Anumberofspecimenshavefailedintheweld material.Althoughdataisavailableofthisareaoftheweld,the
exactgeometryofanypresentnotcheshasnotbeendetermined.Toaccuratelymodelthesefailures,thisareaof
the weld should be analyzed thoroughly. This analysis may also lead to an explanation of the fact that the S460
specimenshaveshownalotofweldmaterialfailuresandtheS690specimenshavenonewhiletheyhave,tothe
eye,asimilarweldmaterialgeometry.
Recommendationsregardingthefatiguetesting
To avoid runouts, most specimens were tested at relatively high stresses. This leads to an uncertain
determinationoftheslopeoftheSNcurve.Moretestsshouldbedoneatlowerstressrangesleadingtofailureat
highercycles
Ahighuncertaintyexistsinthevariationofthedatasetduetothelimitednumberofdatapointspersteelgrade.
AdditionaltestingshouldgivemorecertaintyoftheactualscatterbandoftheTIGdressedresults.
Recommendationsregardingthemodellingandadjustmentofthefatiguestrength
Inthisresearchanassumptionwasmadeconcerningthereductionofresidualstressesattheweldtoebeforeand
after TIGdressing. Additional research on residual stresses in TIGdressed butt welded joints is necessary to
accurately model a fatigue strength increase. The residual stresses in the weld material are also interesting to
measure, because an increase of residual stresses in the weld material may explain the occurrence of the extra
failuremechanismintheweldmaterial.
When the weld toe geometries have been analyzed more thoroughly as described above, it may be possible to
determineaworstcasescenariobycombiningdifferentweldtoeparametersaccordingtotheiractualoccurrence
anddependencies.ThiscanleadtoanaccurateworstcasevalueofK
f
.
All correction factors for thickness, mean stress and loading mode have been based on the research by Pijpers
(2011).Theinfluenceofsomefactorsdependsontheweldgeometryaswell.Furtherresearchshouldinvestigate
whethertheusedcorrectionfactorsarealsovalidforTIGdressedgeometriesordefinenewcorrectionfactors.
All correction factors for thickness, mean stress and loading mode have been based on the research by Pijpers
(2011). Because in the research by Pijpers no weld material failures were observed, no such correction factors
were available for the current research. Additional research should focus on correction factors when failure is
observedintheweldmaterial.
The developedmodel of the fatigue strength isinaccurate. Dueto assumptionsmade when applying the critical
distanceapproach,thedeterminationofK
f
maybelessprecise,leadingtoalessaccuratemodel.Omittingthese
assumptions may improve model accuracy but will partly neutralize the beneficial geometrical influence of TIG
dressing. Other methods to determine the fatigue notch factor, such as the stress gradient approach, may also
leadtobetterresults.
Recommendationsregardingunuseddataofcrackmonitoringandhardnessmeasurements
ThecrackmonitoringdatawhichispresentedinAnnexI,hasnotbeenanalyzed.Althoughlimited,thisdatabase
maygiveinsightintoanyinfluencesoftheTIGdressingprocedureonthecrackpropagationlifewhencombined
with similar measurements on as welded specimens from Pijpers (2011). This calculation could also be used to
verifythemeasurednumberofcyclestocrackinitiation.
The hardness measurements show rather different results than found in literature. Based on observations,
interesting conclusions can be drawn if a relation can be made between hardness and fatigue strength. A more
thoroughresearchofmoreTIGdressedspecimenscanleadtomoreconclusiveresults.
128
129
AnnexA:References
[1] Anami, K and Miki, C. Fatigue strength of welded joints made of highstrength steels, Progress in Structural
EngineeringandMaterials,Vol.3,pp.8694,2001
[2] Anthes,R.J.VolkerKottgen,V.B.Seeger,T.KerbformzahlenvonStumpfstenundDoppelTsten,Schweien
undschneiden,vol.45no.12pp.685688,1993
[3] Brozzetti, J., Hirt, M. A., Ryan, I. Sedlacek, G. Smith, I.F.C. BAckbround information on fatigue design rules
StatisticalevaluationChapter9,document9.01,Eurocode3Editorialgroup,1989
[4] Dahle, T. Design fatigue strength of TIGdressed welded joints in highstrength steels subjected to spectrum
loading,InternationalJournalofFatigue,Vol.20,No.9,pp.677681,1998
[5] EngineeringArchives.com,2008
[6] ESDEPLectureWG12:Fatigue
[7] FMECWMConstructiestaalsoortenmethogesterkte,VMpublication125,2008
[8] Gudehus,H.Zenner,H.LeitfadenfurBetriebsfestigkeitsrechnung,VerlagStahleisenGmbH,1999
[9] Gurney,T.R.Fatigueofweldedstructures,CambridgeUniversityPress,1979
[10] Haagensen, P.J. and Maddox, S.J. IIW Recommendations on Post Weld Improvement of Steel and Aluminium
Structures,IIWdocumentXIII181500,2001
[11] Haagensen, P.J. Drgen, A. Slind, T. and rjaster, O. Prediction of the improvement in fatigue lif of welded
jointsduetogrinding,TIGdressing,weldshapecontrolandshotpeening,Proceedingsofthe3rdInternational
ECSCOffshoreConferenceonSteelinMarineStructures(SIMS87),Delft,1987
[12] Haibach,E.BetriebsfestigkeitVerfahrenundDatenzurBauteilberechnung,SpringerVerlag,2006
[13] Hildebrand, J. Starcevic, I. Werner, F. Heinemann, H. and Khler, G. Numerical simulation of TIGdressing of
welded joints Joint International Conference on Computing and Decision Making in Civil and Building
Engineering,Montral,2006
[14] Hobbacher,A.Recommendationsforfatiguedesignofweldedjointsandcomponents,IIWdocumentXIII2151
07/XV125407,2007
[15] Hck,M.,Thrainer,L.,Schtz,W.BerechnungvonWhlerlinienfurBauteileausStahl,StahlgussundGrauguss
SynthetischeWhlerlinien,BerichtNr.ABF11(ZweiteberarbeiteteFassung),VerlagStahleisen,1981
[16] Huo,L.Wang,D.Zhang,Y.InvestigationofthefatiguebehaviouroftheweldedjointstretedbyTIGdressingand
ultrasonicpeeningundervariableamplitudeload,InternationalJournalofFatigue,Vol.27pp.95101,2005
[17] Janssen,M.Zuidema,J.Wanhill,R.J.H.FractureMechanics,DelftUniversityPress,2002
[18] Kuguel,R.Arelationbetweentheoreticalstressconcentrationfactorandfatiguenotchfactordeducedfromthe
conceptofhighlystressedvolume,ProceedingsofASTM61,732744,1961
[19] LopezMartinez,L.Lin,R.Wang,andBlom,A.F.InvestigationofresidualstressesinasweldedandTIGdressed
specimens to static/spectrum loading Proceedings of NESCO: Welded HighStrength Steel Structures,
Stockholm,1997
[20] Maddox,S.J.Fatiguestrengthofweldedstructures,AbingtonPublishing,1991
[21] NENEN199319,Eurocode3:DesignofsteelstructuresPart19:Fatigue,2006
[22] NENEN 1993112, Eurocode 3: Design of steel structures Part 112: Additional rules for the extension of EN
1993uptosteelgradesS700,2007
[23] NENENISO 5817, Welding Fusion welded joints in steel, nickel, titanium and their alloys (beam welding
excluded)Qualitylevelsforimperfections,2007
[24] NeuberH.Kerbspannungslehre,SpringerVerlag,1937
[25] NeuberH.TheoryofNotchStresses,AnnArborMich,1946
[26] Neuber H. ber die Bercksichtigung der Spannungskonzentration bei Festigkeitsberechnungen, Konstruktion,
Vol20,No.7,pp.245251,1968
[27] NPRCEN/TS 1300131, Cranes General Design Part 31: Limit states and proof of competence of steel
structures,2005
[28] Pedersen,M.M.Mouritsen,O..Hansen,M.R.andAndersenJ.G.ExperiencewiththeNotchStressApproachfor
FatigueAssessmentofWeldedJointsProceedingsoftheSwedishConferenceonLightWeightOptimizedWelded
Structures,Borlnge,2010
[29] Pedersen, M.M. Mouritsen, O.. Hansen, M.R. Andersen J.G and Wenderby, J. Comparison of Post Weld
TreatmentofHighStrengthSteelWeldedJointsinMediumCycleFatigue,IIWdocumentXIII227209,2009
[30] Peterson,R.E.Stressconcentrationfactors,NewYork,JohnWiley,1974
[31] Pijpers, R.J.M. Kolstein, M.H. Romeijn, A. and Bijlaard, F.S.K. Fatigue experiments on very high strength steel
basematerialandtransversebuttwelds,Advancedsteelconstruction,Vol,5,No.1,1432,2009
[32] Pijpers, R.J.M. Kolstein, M.H. Fatigue strength of truss girders made of Very High Strength Steel, as presented
onthe:13
th
InternationalSymposiumonTubularStructures,pp.499505,2010
130
[33] Pijpers, R.J.M. Fatigue strength of welded connections made of very high strength cast and rolled steels,
Dissertation,M2I,2011
[34] Radaj, D. Sonsino, C.M. and Fricke, W. Fatigue assessment of welded joints by local approaches second
edition,Woodheadpublishinglimited,2006
[35] Romeijn,A.LecturenotesCT5126Fatigue,DelftUniversityofTechnology,2006
[36] Siebel, E. and Stieler, M., Ungleichfrmiche Spannungsverteilung bei schwingender Beanspruchug, VDI
Zeitschrift,Vol.97,No.5,pp.121126,1993
[37] Schijve,J.FatigueofStructuresandMaterials,KluwerAcademicPublishers,2001,
[38] Sonsino, C.M., Zur Bewertung des Schwingfestigheitsverhaltens von Bauteilen mit Hilfe rtlicher
Beanspruchungen,Konstruktion,1993,Vol.45,pp.2533,1993
[39] Tricoteaux, A. Fardoun, F. Degallaix, S. Sauvage, F. Fatigue crack initiation life prediction in high strength
structuralsteelweldedjoints,FatigueandFractureofEngineeringMaterials,Vol.18,No.2pp.189200,1995
[40] Wanhill, R.J.H. Durability analysis using short and long fatigue crack growth data, Proceedings of the
InternationalConferenceonAircraftDamageAssessmentandandRepair,Melbourne,1991
[41] Wortel, H van. Integrity of High Strength Steel Structures; Final Report of Work Package: Fabrication, TNO
Report,2006
131
AnnexB:Testspecimens
Onthefollowingpagesthegeometryoftheproducedtestspecimenswillbedepicted.
The geometry of tested specimens is tabulated for each specimen separately. The figure below gives guidance to the
locationofthevariousmeasurements.
132
W
1
W
2
W
3
W
4
W
L
1
W
L
2
M
1
M
2
T
1
T
2
T
3
T
4
T
5
T
6
T
7
T
8
S
i
d
e
S
i
d
e
V
4
6
9
9
1
0
0
7
0
7
0
1
0
0
1
0
1
4
9
5
1
7
.
7
7
.
4
2
5
.
4
2
5
.
3
2
5
.
3
2
5
.
1
2
5
.
3
2
5
.
1
2
5
.
5
2
5
.
7
R
o
l
l
e
d
R
o
l
l
e
d
V
4
6
9
9
1
0
1
7
0
7
0
1
0
0
1
0
1
5
0
5
1
8
.
0
8
.
3
2
5
.
5
2
5
.
6
2
5
.
3
2
5
.
2
2
5
.
3
2
5
.
4
2
5
.
5
2
5
.
5
R
o
l
l
e
d
R
o
l
l
e
d
V
4
6
1
0
1
1
0
3
7
0
7
0
1
0
2
1
0
6
5
1
4
9
8
.
0
7
.
7
2
5
.
5
2
5
.
6
2
5
.
2
2
5
.
1
2
5
.
3
2
5
.
1
2
5
.
5
2
5
.
4
R
o
l
l
e
d
R
o
l
l
e
d
V
6
9
1
0
2
1
0
6
7
0
7
0
1
0
1
1
0
2
4
5
4
8
4
.
8
4
.
3
2
4
.
6
2
4
.
6
2
4
.
5
2
4
.
5
2
4
.
4
2
4
.
5
2
4
.
6
2
4
.
6
R
o
l
l
e
d
R
o
l
l
e
d
V
6
9
1
0
3
1
0
6
7
0
7
0
1
0
4
1
0
5
4
4
4
6
6
.
1
6
.
7
2
4
.
6
2
4
.
6
2
4
.
4
2
4
.
5
2
4
.
5
2
4
.
4
2
4
.
6
2
4
.
7
R
o
l
l
e
d
R
o
l
l
e
d
V
6
9
1
0
4
1
0
9
7
0
7
0
1
0
7
1
1
1
4
5
4
3
6
.
7
6
.
7
2
4
.
8
2
5
.
0
2
4
.
6
2
4
.
6
2
4
.
6
2
4
.
7
2
4
.
7
2
4
.
7
R
o
l
l
e
d
R
o
l
l
e
d
V
8
9
9
3
9
7
7
0
7
0
9
8
1
0
0
4
4
4
5
3
.
9
4
.
1
2
4
.
6
2
4
.
6
2
4
.
6
2
4
.
7
2
4
.
7
2
4
.
6
2
4
.
9
2
4
.
7
R
o
l
l
e
d
R
o
l
l
e
d
V
8
9
9
8
9
9
7
0
7
0
9
1
9
8
4
2
4
2
4
.
2
3
.
8
2
4
.
7
2
4
.
7
2
4
.
6
2
4
.
6
2
4
.
7
2
4
.
7
2
4
.
7
2
4
.
7
R
o
l
l
e
d
R
o
l
l
e
d
V
8
9
9
9
1
0
0
7
0
7
0
1
0
1
1
0
4
4
4
4
2
3
.
8
3
.
2
2
4
.
7
2
4
.
7
2
4
.
6
2
4
.
5
2
4
.
6
2
4
.
6
2
4
.
7
2
4
.
7
R
o
l
l
e
d
R
o
l
l
e
d
V
1
1
9
8
9
9
7
0
7
0
9
8
9
9
3
9
3
8
2
.
7
2
.
3
1
9
.
9
2
0
.
0
2
0
.
0
2
0
.
0
1
9
.
8
1
9
.
8
1
9
.
9
1
9
.
9
R
o
l
l
e
d
R
o
l
l
e
d
V
1
1
1
0
2
1
0
3
7
0
7
0
9
2
1
0
3
3
5
3
6
5
.
0
4
.
6
2
0
.
0
2
0
.
0
2
0
.
1
1
9
.
9
2
0
.
0
1
9
.
9
2
0
.
1
2
0
.
0
R
o
l
l
e
d
R
o
l
l
e
d
V
1
1
1
0
4
1
0
6
7
0
7
0
1
0
2
1
0
5
3
0
3
2
4
.
3
4
.
3
2
0
.
1
2
0
.
1
1
9
.
8
2
0
.
0
1
9
.
8
1
9
.
8
2
0
.
0
1
9
.
9
R
o
l
l
e
d
R
o
l
l
e
d
C
4
6
1
0
4
1
0
9
7
0
7
0
1
0
0
1
0
1
5
1
5
1
9
.
6
7
.
7
2
5
.
5
2
5
.
6
2
5
.
5
2
5
.
4
2
7
.
1
2
7
.
0
2
7
.
0
2
7
.
1
R
o
l
l
e
d
C
a
s
t
C
4
6
C
4
6
1
0
5
7
0
7
0
1
0
5
1
1
1
4
9
4
9
8
.
4
6
.
5
2
5
.
4
2
5
.
5
2
5
.
1
2
5
.
1
2
6
.
9
2
7
.
2
2
8
.
1
2
9
.
1
R
o
l
l
e
d
C
a
s
t
C
6
9
1
0
6
7
0
7
0
1
0
7
1
1
2
5
3
5
1
7
.
1
8
.
5
2
6
.
0
2
5
.
7
2
6
.
0
2
5
.
9
2
4
.
7
2
4
.
7
2
4
.
7
2
4
.
7
C
a
s
t
R
o
l
l
e
d
C
6
9
1
0
3
1
0
5
7
0
7
0
1
0
2
1
0
3
4
9
5
1
6
.
4
7
.
6
2
6
.
0
2
5
.
9
2
5
.
6
2
5
.
6
2
4
.
5
2
4
.
6
2
4
.
7
2
4
.
7
C
a
s
t
R
o
l
l
e
d
C
6
9
1
0
5
1
0
8
7
0
7
0
1
0
4
1
0
5
4
8
4
8
8
.
4
7
.
4
2
4
.
7
2
4
.
5
2
4
.
4
2
4
.
5
2
5
.
4
2
5
.
4
2
5
.
7
2
5
.
3
R
o
l
l
e
d
C
a
s
t
C
8
9
1
0
2
1
0
3
7
0
7
0
1
0
2
1
0
4
4
4
4
4
2
4
.
3
2
4
.
8
2
4
.
4
2
4
.
6
2
4
.
4
2
4
.
5
2
4
.
5
2
4
.
6
R
o
l
l
e
d
C
a
s
t
C
8
9
1
0
3
1
0
4
7
0
7
0
9
7
9
8
4
5
4
4
3
.
3
3
.
0
2
4
.
7
2
4
.
7
2
4
.
4
2
4
.
6
2
4
.
3
2
4
.
4
2
3
.
7
2
3
.
7
R
o
l
l
e
d
C
a
s
t
C
8
9
a
1
0
4
1
0
5
7
0
7
0
1
0
1
1
0
6
4
5
4
4
4
.
3
3
.
2
2
4
.
8
2
4
.
9
2
4
.
7
2
4
.
7
2
3
.
9
2
3
.
9
2
3
.
7
2
4
.
1
R
o
l
l
e
d
C
a
s
t
C
1
1
9
8
7
0
7
0
9
7
9
9
4
3
4
2
3
.
5
3
.
4
2
0
.
2
2
0
.
1
2
0
.
0
2
0
.
1
2
0
.
9
2
0
.
9
2
1
.
3
2
1
.
2
R
o
l
l
e
d
C
a
s
t
C
1
1
9
9
1
0
0
7
0
7
0
1
0
0
1
0
1
4
0
4
4
2
.
5
2
.
4
2
0
.
0
2
0
.
1
2
0
.
0
2
0
.
0
2
0
.
8
2
0
.
7
2
1
.
2
2
1
.
0
R
o
l
l
e
d
C
a
s
t
C
1
1
9
7
9
9
7
0
7
0
9
7
1
0
0
4
1
4
1
1
.
1
1
.
0
2
0
.
0
2
0
.
0
2
0
.
0
2
0
.
1
2
0
.
5
2
0
.
3
2
0
.
7
2
0
.
7
R
o
l
l
e
d
C
a
s
t
R
e
m
a
r
k
s
:

A
l
l
m
e
a
s
u
r
e
m
e
n
t
s
i
n
m
i
l
l
i
m
e
t
e
r
s

C
4
6
2
:
b
e
n
t
d
u
e
t
o
t
e
s
t
r
i
g
e
r
r
o
r

C
6
9
2
:
w
e
l
d
r
o
o
t
i
s
g
r
o
u
n
d
s
m
o
o
t
h
d
u
e
t
o
l
a
c
k
o
f
T
I
G
d
r
e
s
s
i
n
g

C
8
9
2
:
l
a
r
g
e
d
e
f
e
c
t
s
a
r
e
r
e
m
o
v
e
d
a
n
d
f
i
l
l
e
d
i
n
w
i
t
h
w
e
l
d
m
a
t
e
r
i
a
l
.
S
p
e
c
i
m
e
n
s
i
s
,
e
v
e
n
a
f
t
e
r
t
r
e
a
t
m
e
n
t
,
n
o
t
e
n
t
i
r
e
l
y
d
e
f
e
c
t
f
r
e
e
,
b
a
s
e
d
o
n
v
i
s
u
a
l
i
n
s
p
e
c
t
i
o
n
133
C
r
a
c
k
l
o
c
a
t
i
o
n
N
R
(
S
G
)
R
(
F
)
N
/
R
b
a
s
e
d
o
n
S
G
N
i
N
p
N
f
N
i
/
N
f
(
F
/
A
)
K
m
f
N
i
f
N
p
f
l
m
;
N
f
f
l
m
;
N
i
f
l
m
;
N
p
f
t
;
N
f
f
t
;
N
i
f
t
;
N
p
f
m
V
4
6
1
W
M
_
c
a
p
_
m
i
d
d
l
e
3
1
9
0
.
5
3
0
.
0
2
6
3
2
7
2
7
9
6
1
1
5
2
5
6
3
8
8
0
5
2
0
.
7
0
2
3
0
1
.
3
9
0
.
7
0
0
.
3
0
1
.
0
6
1
.
0
5
1
.
0
8
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
V
4
6
2
W
M
_
c
a
p
_
m
i
d
d
l
e
2
4
8
0
.
3
9
0
.
0
3
5
2
,
3
,
4
,
5
3
7
9
6
0
2
1
1
5
3
0
6
4
9
4
9
0
8
0
.
7
7
1
7
3
1
.
4
3
0
.
7
7
0
.
2
3
1
.
0
6
1
.
0
5
1
.
0
9
1
.
0
0
1
.
0
0
1
.
0
0
0
.
9
7
V
4
6
3
W
M
_
c
a
p
_
e
d
g
e
2
1
2
0
.
4
3
0
.
0
3
5
3
,
5
6
0
3
8
1
9
1
3
9
0
1
1
7
4
2
8
3
0
0
.
8
1
1
4
6
1
.
4
6
0
.
8
1
0
.
1
9
1
.
0
6
1
.
0
6
1
.
0
9
1
.
0
0
1
.
0
0
1
.
0
0
0
.
9
6
V
6
9
1
N
/
A
2
0
6
0
.
3
3
0
.
0
4
2
4
,
5
7
0
2
3
3
1
7
0
7
0
2
3
3
1
7
1
.
0
0
1
4
7
1
.
4
0
1
.
0
0
0
.
0
0
1
.
0
5
1
.
0
5
1
.
0
8
1
.
0
1
1
.
0
1
1
.
0
0
0
.
8
5
V
6
9
2
W
T
_
r
o
l
l
e
d
_
c
a
p
_
m
i
d
d
l
e
2
7
7
0
.
3
1
0
.
0
2
8
4
,
5
3
4
0
7
6
5
9
8
7
9
5
4
3
9
5
6
0
0
.
7
8
2
0
6
1
.
3
4
0
.
7
8
0
.
2
2
1
.
0
5
1
.
0
4
1
.
0
7
1
.
0
1
1
.
0
1
1
.
0
0
0
.
8
7
V
6
9
3
W
T
_
r
o
l
l
e
d
_
c
a
p
_
m
i
d
d
l
e
3
8
0
0
.
2
8
0
.
0
2
3
3
4
7
4
9
4
8
3
7
9
0
1
3
1
2
8
4
0
.
3
6
2
6
4
1
.
4
4
0
.
3
6
0
.
6
4
1
.
0
8
1
.
0
5
1
.
0
9
1
.
0
0
1
.
0
1
1
.
0
0
0
.
9
0
V
8
9
1
W
M
_
c
a
p
_
m
i
d
d
l
e
3
6
4
0
.
2
0
0
.
0
2
2
2
,
3
,
5
5
9
4
1
7
2
6
0
2
4
2
6
5
4
4
1
4
0
.
9
1
2
9
0
1
.
2
6
0
.
9
1
0
.
0
9
1
.
0
4
1
.
0
4
1
.
0
6
1
.
0
0
1
.
0
0
1
.
0
0
0
.
8
1
V
8
9
2
N
/
A
2
6
6
0
.
2
2
0
.
0
3
2
2
,
3
,
4
,
5
5
5
5
1
3
5
1
0
5
5
5
1
3
5
1
1
.
0
0
2
1
5
1
.
2
4
1
.
0
0
0
.
0
0
1
.
0
3
1
.
0
3
1
.
0
5
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
9
V
8
9
3
W
M
_
c
a
p
_
m
i
d
d
l
e
3
9
4
0
.
1
3
0
.
0
2
3
4
,
5
2
5
0
3
8
9
1
0
2
6
7
3
3
5
3
0
6
2
0
.
7
1
3
2
5
1
.
2
1
0
.
7
1
0
.
2
9
1
.
0
4
1
.
0
3
1
.
0
5
1
.
0
0
1
.
0
1
1
.
0
0
0
.
8
1
V
1
1
1
B
M
_
r
o
l
l
e
d
_
r
o
o
t
_
e
d
g
e
1
8
1
0
.
2
3
0
.
1
1
8
1
2
1
9
0
2
5
8
4
8
6
5
7
2
3
8
9
1
5
0
.
8
0
2
2
5
0
.
8
0
0
.
8
0
0
.
2
0
1
.
0
0
1
.
0
0
0
.
9
8
1
.
0
0
1
.
0
0
1
.
0
0
0
.
6
8
V
1
1
2
W
T
_
r
o
l
l
e
d
_
c
a
p
_
m
i
d
d
l
e
4
5
2
0
.
4
6
0
.
0
2
6
2
,
3
9
5
0
8
3
4
7
4
2
0
1
4
2
5
0
3
0
.
6
7
3
2
5
1
.
3
9
0
.
6
7
0
.
3
3
1
.
0
6
1
.
0
5
1
.
0
8
1
.
0
6
1
.
0
9
1
.
0
0
0
.
8
6
V
1
1
3
N
/
A
2
4
0
0
.
1
9
0
.
0
4
3
2
,
3
,
4
,
5
9
2
7
0
0
6
3
0
9
2
7
0
0
6
3
1
.
0
0
1
7
6
1
.
3
6
1
.
0
0
0
.
0
0
1
.
0
5
1
.
0
5
1
.
0
8
1
.
1
0
1
.
1
0
1
.
0
0
0
.
6
9
C
4
6
1
W
M
_
c
a
p
_
m
i
d
d
l
e
3
0
9
0
.
4
9
0
.
0
2
7
2
2
9
1
1
4
8
6
5
6
7
2
3
5
6
8
2
0
0
.
8
2
2
1
3
1
.
4
5
0
.
8
2
0
.
1
8
1
.
0
6
1
.
0
5
1
.
0
9
0
.
9
9
0
.
9
9
1
.
0
0
0
.
9
8
C
4
6
2
C
4
6
3
W
T
_
r
o
l
l
e
d
_
c
a
p
_
m
i
d
d
l
e
4
0
8
0
.
5
5
0
.
0
2
3
2
,
3
4
1
8
2
7
6
4
2
0
9
1
0
6
0
3
6
0
.
3
9
2
7
3
1
.
4
9
0
.
3
9
0
.
6
1
1
.
0
8
1
.
0
6
1
.
1
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
2
C
6
9
1
W
T
_
r
o
l
l
e
d
_
c
a
p
_
m
i
d
d
l
e
3
8
0
0
.
3
2
0
.
0
2
6
4
9
4
5
0
9
5
8
2
7
8
1
5
2
7
8
7
0
.
6
2
2
8
0
1
.
3
6
0
.
6
2
0
.
3
8
1
.
0
6
1
.
0
5
1
.
0
8
1
.
0
0
1
.
0
0
1
.
0
0
0
.
9
1
C
6
9
2
W
T
_
r
o
l
l
e
d
_
c
a
p
_
m
i
d
d
l
e
2
9
8
0
.
3
2
0
.
0
3
2
4
5
2
3
6
7
0
1
4
1
8
0
6
6
6
5
4
7
6
0
.
7
9
1
9
8
1
.
5
1
0
.
7
9
0
.
2
1
1
.
0
7
1
.
0
6
1
.
1
0
1
.
0
1
1
.
0
1
1
.
0
0
0
.
8
8
C
6
9
3
W
T
_
r
o
l
l
e
d
_
r
o
o
t
_
m
i
d
d
l
e
1
2
4
8
.
6
7
0
.
0
3
6
8
,
9
,
1
0
,
1
1
3
0
7
3
5
3
9
1
1
8
4
3
4
3
1
9
1
9
7
3
0
.
9
6
1
7
1
0
.
7
3
0
.
9
6
0
.
0
4
0
.
9
4
0
.
9
4
0
.
9
1
1
.
0
1
1
.
0
1
1
.
0
0
0
.
7
5
C
8
9
1
B
M
_
c
a
s
t
_
s
i
d
e
2
5
4
2
,
8
7
9
5
9
0
2
3
0
7
1
4
8
2
6
6
1
6
0
.
9
6
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
3
C
8
9
2
B
M
_
c
a
s
t
_
c
a
p
_
m
i
d
d
l
e
2
2
8
0
.
2
5
0
.
0
2
5
6
(
*
7
0
/
9
8
)
5
4
4
1
9
5
1
9
5
2
1
0
6
3
7
1
0
.
5
1
2
3
6
0
.
9
7
0
.
5
1
0
.
4
9
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
8
C
8
9
a
1
W
M
_
c
a
p
_
m
i
d
d
l
e
4
1
1
0
.
1
6
0
.
0
1
7
2
,
3
,
4
,
5
7
3
6
5
9
2
5
6
5
1
9
9
3
1
0
0
.
7
4
3
4
2
1
.
2
0
0
.
7
4
0
.
2
6
1
.
0
3
1
.
0
3
1
.
0
5
1
.
0
1
1
.
0
1
1
.
0
0
0
.
8
2
C
1
1
1
B
M
_
c
a
s
t
_
c
a
p
_
e
d
g
e
2
5
4
0
.
1
3
0
.
0
3
2
3
3
7
5
3
4
4
5
5
4
7
2
4
3
0
8
1
6
0
.
8
7
2
1
8
1
.
1
6
0
.
8
7
0
.
1
3
1
.
0
0
1
.
0
0
1
.
0
1
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
4
C
1
1
2
B
M
_
c
a
s
t
_
c
a
p
_
e
d
g
e
2
3
8
6
1
8
5
4
7
9
3
6
5
3
0
2
1
9
2
0
0
9
5
0
.
9
7
1
.
0
0
1
.
0
0
1
.
0
1
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
4
C
1
1
3
W
T
_
c
a
s
t
_
c
a
p
_
e
d
g
e
/
(
B
M
_
c
a
s
t
_
c
a
p
_
e
d
g
e
)
4
1
2
0
.
2
7
0
.
1
3
0
5
1
5
4
8
9
6
3
9
4
9
0
1
9
4
3
8
6
0
.
8
0
3
4
4
1
.
2
0
0
.
8
0
0
.
2
0
1
.
0
3
1
.
0
3
1
.
0
5
1
.
0
7
1
.
0
8
1
.
0
0
0
.
8
1
R
e
r
u
n
s
p
e
c
i
m
e
n
s
C
r
a
c
k
l
o
c
a
t
i
o
n
N
R
(
S
G
)
R
(
F
)
N
/
R
b
a
s
e
d
o
n
S
G
N
i
N
p
N
f
N
i
/
N
f
(
F
/
A
)
K
m
f
N
i
f
N
p
f
l
m
;
N
f
f
l
m
;
N
i
f
l
m
;
N
p
f
t
;
N
f
f
t
;
N
i
f
t
;
N
p
f
m
a
t
S
G
o
f
f
i
n
a
l
f
a
i
l
u
r
e
C
1
1
2
I
B
M
_
c
a
s
t
_
c
a
p
_
e
d
g
e
1
7
7
0
.
0
9
0
.
0
4
8
6
(
*
7
0
/
1
0
0
)
7
2
4
7
5
0
1
0
7
1
4
7
8
3
1
8
9
7
0
.
8
7
1
4
7
1
.
2
0
0
.
8
7
0
.
1
3
1
.
0
0
1
.
0
0
1
.
0
1
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
1
2
5
3
C
1
1
2
I
I
B
M
_
c
a
s
t
_
c
a
p
_
e
d
g
e
2
2
5
0
.
2
8
7
0
.
0
3
4
6
1
0
2
2
8
9
6
6
5
3
0
2
1
0
8
8
1
9
8
0
.
9
4
2
2
8
0
.
9
9
0
.
9
4
0
.
0
6
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
5
2
2
5
C
1
1
2
e
q
B
M
_
c
a
s
t
_
c
a
p
_
e
d
g
e
2
3
8
6
1
8
5
4
7
9
3
6
5
3
0
2
1
9
2
0
0
9
5
0
.
9
7
1
.
0
0
1
.
0
0
1
.
0
1
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
4
C
8
9
-
1
I
B
M
_
c
a
s
t
_
r
o
o
t
_
m
i
d
d
l
e
1
5
7
0
.
2
4
0
.
0
2
5
1
2
(
*
7
0
/
1
0
3
)
6
3
3
1
6
7
1
1
3
2
7
1
7
4
6
4
3
8
0
.
8
5
1
5
7
1
.
0
0
0
.
8
5
0
.
1
5
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
3
2
4
8
C
8
9
-
1
I
I
B
M
_
c
a
s
t
_
s
i
d
e
_
e
d
g
e
2
9
9
0
.
0
3
8
0
.
0
2
5
2
,
8
4
9
4
6
4
3
0
7
1
4
8
0
1
7
8
0
.
6
2
2
6
8
1
.
1
2
0
.
6
2
0
.
3
8
1
.
0
0
1
.
0
0
1
.
0
1
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
7
2
9
9
C
8
9
1
e
q
B
M
_
c
a
s
t
_
s
i
d
e
_
e
d
g
e
2
5
4
2
,
8
7
9
5
9
0
2
3
0
7
1
4
8
2
6
6
1
6
0
.
9
6
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
1
.
0
0
0
.
7
3
AnnexC:Fatiguetestdata
134
135
AnnexD:Materialcertificates
E
r
l
a
u
t
e
r
u
n
q
e
n
s
i
e
h
e
R
u
c
k
s
e
l
t
e
/
E
x
p
t
i
c
a
t
l
o
n
s
v
o
i
r
a
u
v
e
r
s
o
/
S
e
e
r
e
v
e
r
s
e
f
o
r
e
x
c
l
a
n
a
t
l
o
n
s
I
w
w
w
.
d
i
l
l
i
n
o
e
r
.
d
e
/
c
e
r
t
i
f
i
c
a
t
e
)
D
I
L
L
I
N
G
E
R
H
U
T
T
E
A
0
2
I
N
S
P
E
C
T
I
O
N
C
E
R
T
I
F
I
C
A
T
E
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
2
0
0
4
I
N
S
P
E
C
T
I
O
N
R
E
P
O
R
T
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
1
9
9
1
+
A
l
:
1
9
9
5
+
A
S
P
E
R
I
S
O
1
0
4
7
4
:
1
9
9
1
A
1
0
A
d
v
i
c
e
o
f
d
i
s
p
a
t
c
h
N
o
.
1
D
a
t
e
o
f
d
i
s
p
a
t
c
h
2
7
6
2
5
3
-
1
7
.
1
0
.
0
7
A
0
8
1
M
a
n
u
f
a
c
t
u
r
e
r
's
o
r
d
e
r
l
A
0
3
C
e
r
t
i
f
i
c
a
t
e
N
o
.
3
2
5
6
5
1
-
0
0
1
S
h
e
e
t
1
1
.
.
.
A
0
5
E
s
t
a
b
l
i
s
h
e
d
I
n
s
p
e
c
t
i
n
g
b
o
d
y
L
R
A
0
6
P
u
r
c
h
a
s
e
r
F
i
n
a
l
r
e
c
e
i
v
e
r
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
0
7
.
1
N
o
.
A
0
7
.
2
N
o
.
3
0
2
0
9
4
7
B
0
1
P
r
o
d
u
c
t
H
O
T
R
O
L
L
E
D
P
L
A
T
E
S
B
0
2
1
S
t
e
e
l
d
e
s
i
g
n
.
B
0
3
A
n
y
s
u
p
p
l
.
r
e
q
u
i
r
e
m
e
n
t
s
D
I
L
L
l
M
A
X
6
9
0
T
D
I
L
L
I
N
G
-
E
0
4
:
0
3
B
0
1
B
9
9
D
e
s
c
r
i
p
t
i
o
n
o
f
t
h
e
p
r
o
d
u
c
t
B
1
4
B
0
8
B
0
9
B
1
0
B
1
1
B
1
2
B
0
4
B
0
7
.
2
B
0
7
.
1
A
0
9
I
t
e
m
N
u
m
b
e
r
T
h
i
c
k
n
e
s
s
W
i
d
t
h
L
e
n
g
t
h
T
h
e
o
r
e
t
i
c
a
l
P
r
o
d
u
c
t
d
e
l
i
v
e
r
y
H
e
a
t
N
o
.
R
o
l
l
e
d
p
l
a
t
e
N
o
.
1
P
u
r
c
h
a
s
e
r
a
r
t
i
c
l
e
n
u
m
b
e
r
N
o
.
o
f
p
i
e
c
e
s
m
a
s
s
c
o
n
d
i
t
i
o
n
T
e
s
t
N
o
.
M
M
K
G
0
1
1
1
2
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
3
6
7
Q
+
A
3
1
2
0
0
5
3
5
7
7
2
-
0
1
0
1
1
1
2
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
3
6
7
Q
+
A
3
1
2
0
0
5
3
5
7
7
2
-
0
2
0
1
1
1
2
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
3
6
7
Q
+
A
3
1
2
0
0
5
3
5
7
7
2
-
0
3
0
1
1
1
2
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
3
6
7
Q
+
A
3
1
2
0
0
5
3
5
7
7
2
-
0
4
0
1
1
1
2
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
3
6
7
Q
+
A
3
1
2
0
0
5
3
5
7
7
2
-
0
5
0
1
1
1
2
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
3
6
7
Q
+
A
3
1
2
0
0
5
3
5
7
7
2
-
0
6
*
*
6
1
4
2
0
2
0
2
1
1
2
,
0
0
X
2
5
0
0
X
1
2
0
0
0
2
8
8
0
Q
+
A
3
1
2
0
0
5
3
5
8
2
1
-
0
1
0
2
1
1
2
,
0
0
X
2
5
0
0
X
1
2
0
0
0
2
8
8
0
Q
+
A
3
1
2
0
0
5
3
5
8
2
1
-
0
2
0
2
1
1
2
,
0
0
X
2
5
0
0
X
1
2
0
0
0
2
8
8
0
Q
+
A
3
1
2
0
0
5
3
5
8
2
1
-
0
3
*
*
3
8
6
4
0
.
0
3
1
1
5
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
9
5
9
Q
+
A
3
1
1
9
1
3
2
6
8
2
9
-
0
1
0
3
1
1
5
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
9
5
9
Q
+
A
3
1
1
9
1
3
2
6
8
2
9
-
0
2
0
3
1
1
5
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
9
5
9
Q
+
A
3
1
1
9
1
3
2
6
8
2
9
-
0
3
0
3
1
1
5
,
0
0
X
2
0
5
5
X
1
2
0
0
0
2
9
5
9
Q
+
A
3
1
1
9
1
3
2
6
8
2
9
-
0
4
*
*
4
1
1
8
3
6
0
4
1
1
5
,
0
0
X
2
5
0
0
X
1
2
0
0
0
3
6
0
0
Q
+
A
3
1
1
9
1
1
2
6
8
3
9
-
0
2
0
4
1
1
5
,
0
0
X
2
5
0
0
.
-
X
1
2
0
0
0
-
e
3
6
0
0
Q
+
.
w
.
$
$
l
1
I
3
1
1
9
1
1
2
6
8
3
9
-
0
3
,
r
I
h
"
t
'.
*
*
2
7
2
0
0
0
5
1
2
0
,
0
0
X
2
0
5
5
X
1
2
0
0
0
3
9
4
6
Q
+
A
3
1
1
9
1
3
2
6
8
2
6
-
0
1
0
5
1
2
0
,
0
0
X
2
0
5
5
X
1
2
0
0
0
3
9
4
6
Q
+
A
3
1
1
9
1
3
2
6
8
2
6
-
0
2
0
5
1
2
0
,
0
0
X
2
0
5
5
X
1
2
0
0
0
3
9
4
6
Q
+
A
3
1
1
9
1
3
2
6
8
2
6
-
0
3
*
*
3
1
1
8
3
8
-
A
0
4
Z
0
1
1
Z
0
2
/
Z
0
3
W
e
h
e
r
e
b
y
c
e
r
t
i
f
y
.
t
h
a
t
t
h
e
m
a
t
e
r
i
a
l
s
h
a
v
e
b
e
e
n
d
e
l
i
v
e
r
e
d
i
n
a
c
c
o
r
d
a
n
c
e
A
G
d
e
r
D
i
l
l
i
n
g
e
r
H
u
t
t
e
n
w
e
r
k
e
A
0
1
o
r
o
.
2
.
O
k
l
.
S
T
U
l
'
r
/
!
'
-
A
B
a
D
X
H
P
o
s
t
f
a
c
h
1
5
8
0
,
D
6
6
7
4
8
D
i
l
l
i
n
g
e
n
/
S
a
a
r
S
t
u
t
t
g
a
r
t
O
f
f
i
c
e
a
.
I
n
s
p
e
c
t
i
o
n
d
e
p
a
r
t
m
e
n
t
L
l
o
y
d
'
s
E
M
E
A
1
A
.
t
'I
2
2
M
a
n
u
f
a
c
t
u
r
e
r
's
I
e
c
t
o
r
7
1
0
8
B
.
M
U
E
L
L
E
R
m
a
r
k
T
e
s
t
H
o
u
s
e
M
a
n
a
a
e
r
I
n
s
p
e
c
t
o
r
's
s
t
a
m
p
D
a
t
e
1
7
.
1
0
.
0
7
E
D
I
H
K
1
"
'
-
"
E
r
l
a
u
t
e
r
u
n
q
e
n
s
i
e
h
e
R
u
c
k
s
e
t
t
e
/
E
x
p
f
i
c
a
t
i
c
n
s
v
o
i
r
a
u
v
e
r
s
o
l
S
e
e
r
e
v
e
r
s
e
f
o
r
e
x
p
l
a
n
a
t
i
o
n
s
(
w
w
w
.
d
i
l
l
i
n
g
e
r
.
d
e
l
c
e
r
t
i
f
i
c
a
t
e
)
.
.
D
I
L
L
I
N
G
E
R
H
U
T
T
E
A
0
2
I
N
S
P
E
C
T
I
O
N
C
E
R
T
I
F
I
C
A
T
E
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
2
0
0
4
I
N
S
P
E
C
T
I
O
N
R
E
P
O
R
T
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
1
9
9
1
+
A
1
:
1
9
9
5
+
A
S
P
E
R
I
S
O
1
0
4
7
4
:
1
9
9
1
A
1
0
A
d
v
i
c
e
o
f
d
i
s
p
a
t
c
h
N
o
.
1
D
a
t
e
o
f
d
i
s
p
a
t
c
h
A
0
8
1
M
a
n
u
f
a
c
t
u
r
e
r
'
s
o
r
d
e
r
l
A
0
3
C
e
r
t
i
f
i
c
a
t
e
N
o
.
S
h
e
e
t
2
7
6
2
5
3
-
1
7
.
1
0
.
0
7
3
2
5
6
5
1
-
0
0
1
2
/
.
.
.
A
0
5
E
s
t
a
b
l
i
s
h
e
d
I
n
s
p
e
c
t
i
n
g
b
o
d
y
L
R
A
0
6
P
u
r
c
h
a
s
e
r
F
i
n
a
l
r
e
c
e
i
v
e
r
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
0
7
.
1
N
o
.
A
0
7
.
2
N
o
.
3
0
2
0
9
4
7
B
0
1
P
r
o
d
u
c
t
H
O
T
R
O
L
L
E
D
P
L
A
T
E
S
B
0
2
1
S
t
e
e
l
d
e
s
i
g
n
.
B
0
3
A
n
y
s
u
p
p
l
.
r
e
q
u
i
r
e
m
e
n
t
s
D
I
L
L
I
M
A
X
6
9
0
T
D
I
L
L
I
N
G
-
E
0
4
:
0
3
8
0
1
-
8
9
9
D
e
s
c
r
i
p
t
i
o
n
o
f
t
h
e
p
r
o
d
u
c
t
B
1
4
B
0
8
B
0
9
B
1
0
B
1
1
I
t
e
m
N
u
m
b
e
r
T
h
i
c
k
n
e
s
s
W
i
d
t
h
L
e
n
g
t
h
N
o
.
o
f
p
i
e
c
e
s
.
.I
i
l
I
,
-
'
A
0
9
P
u
r
c
h
a
s
e
r
a
r
t
i
c
l
e
n
u
m
b
e
r
B
0
7
.
1
R
o
l
l
e
d
p
l
a
t
e
N
o
.
1
T
e
s
t
N
o
.
2
6
7
9
8
-
0
2
2
6
7
9
8
-
0
3
1
8
5
1
4
-
0
1
1
8
5
1
4
-
0
2
1
2
1
2
4
-
0
2
1
2
1
2
4
-
0
3
2
6
7
6
4
-
0
2
2
6
7
9
7
-
0
1
2
6
7
9
7
-
0
2
2
6
7
8
7
-
0
2
2
6
7
8
7
-
0
3
4
0
9
9
1
-
0
1
4
1
3
2
1
-
0
1
4
1
3
2
1
-
0
2
4
1
3
2
0
-
0
1
'8
!
!
'
4
1
3
2
0
-
0
2
B
0
7
.
2
H
e
a
t
N
o
.
3
1
1
9
1
5
3
1
1
9
1
5
3
1
2
7
0
0
3
1
2
7
0
0
3
1
1
9
1
3
3
1
1
9
1
3
3
1
2
4
1
6
3
1
2
4
1
6
3
1
2
4
1
6
3
1
2
4
1
6
3
1
1
9
1
5
3
1
1
9
1
5
3
1
1
9
1
5
3
1
2
7
0
0
3
1
2
7
0
0
3
1
2
7
0
0
B
0
4
P
r
o
d
u
c
t
d
e
l
i
v
e
r
y
c
o
n
d
i
t
i
o
n
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
Q
+
A
8
1
2
T
h
e
o
r
e
t
i
c
a
l
m
a
s
s
K
G
4
8
0
0
4
8
0
0
9
6
0
0
4
9
3
2
4
9
3
2
9
8
6
4
6
0
0
0
7
2
0
0
7
2
0
0
1
4
4
0
0
5
9
1
8
5
9
1
8
1
1
8
3
6
7
2
0
0
7
2
0
0
1
4
4
0
0
X
1
2
0
0
0
8
8
7
8
X
1
2
0
0
0
,
8
8
7
8
1
7
7
5
6
1
0
8
0
0
9
8
6
4
9
8
6
4
1
9
7
2
8
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
X
1
2
0
0
0
x
1
2
0
0
0
x
1
2
0
0
0
M
M
X
2
5
0
0
X
2
5
0
0
X
2
5
0
0
X
2
5
0
0
X
2
0
5
5
X
2
0
5
5
.
.
X
2
0
5
5
X
2
0
5
5
X
2
0
5
5
X
2
0
5
5
X
2
5
0
0
X
2
0
5
5
X
2
0
5
5
X
2
5
0
0
X
3
0
0
0
X
3
0
0
0
3
0
,
0
0
3
0
,
0
0
3
0
,
0
0
3
0
,
0
0
2
5
,
0
0
2
5
,
0
0
2
5
,
0
0
2
0
,
0
0
2
0
,
0
0
2
5
,
0
0
2
5
,
0
0
4
5
,
0
0
4
5
,
0
0
4
5
,
0
0
5
0
,
0
0
5
0
,
0
0
11211211121121121121112
*
*
*
*
*
*
*
*
*
*
*
*
*
*
0
7
0
7
1
9
1
9
8
l
1
I
t
0
6
0
6
1
0
1
0
1
1
1
1
0
8
0
9
0
9
2
0
2
1
2
1
A
G
d
e
r
D
i
l
l
i
n
g
e
r
H
u
t
t
e
n
w
e
r
k
e
P
o
s
t
f
a
c
h
1
5
8
0
,
0
-
6
6
7
4
8
D
i
l
l
i
n
g
e
n
/
S
a
a
r
I
n
s
p
e
c
t
i
o
n
d
e
p
a
r
t
m
e
n
t
A
0
4
D
X
H
M
a
n
u
f
a
c
t
u
r
e
r
's
m
a
r
k
Z
0
1
I
Z
0
2
l
~
~
_
W
e
h
e
r
e
b
y
c
e
r
t
i
f
y
,
t
h
a
t
t
h
e
a
b
o
t
m
e
n
t
i
t
e
d
m
a
t
e
r
i
a
l
s
h
a
v
e
b
e
e
n
d
e
l
i
v
e
r
e
d
i
n
a
c
c
o
r
d
a
n
c
e
H
.
-
J
.
C
y
r
a
n
s
k
i
I
t
i
"
>
'
r
l
<
'
~
S
t
u
t
t
g
a
r
t
O
f
f
i
c
e
'
.
I
.
l
l
o
y
d
'
s
R
e
g
i
s
t
e
r
E
M
E
A
'
Y
t
.
.
.
1
\
s
p
e
c
t
o
r
l
'
r
l
f
-
B
.
M
U
E
L
L
E
R
T
e
s
t
H
o
u
s
e
M
a
n
a
g
e
r
I
n
s
p
e
c
t
o
r
'
s
s
t
a
m
p
D
a
t
e
1
7
.
1
0
.
0
7
E
D
T
H
K
1
A
0
1
0
0
D
I
L
L
I
N
G
E
R
H
U
T
T
E
E
r
l
a
u
t
e
r
u
n
a
e
n
s
i
e
h
e
R
Q
c
k
s
e
i
t
e
/
E
x
p
l
i
c
a
t
i
o
n
s
v
o
i
r
a
u
v
e
r
s
o
/
S
e
e
r
e
v
e
r
s
e
f
o
r
e
x
o
l
a
n
a
t
l
o
n
s
(
w
w
w
.
d
i
l
l
i
n
a
e
r
.
d
e
/
c
e
r
t
i
f
i
c
a
t
e
l
A
0
2
I
N
S
P
E
C
T
I
O
N
C
E
R
T
I
F
I
C
A
T
E
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
2
0
0
4
A
1
0
A
d
v
i
c
e
o
f
d
i
s
p
a
t
c
h
N
o
.
1
A
0
8
1
M
a
n
u
f
a
c
t
u
r
e
r
's
o
r
d
e
r
l
S
h
e
e
t
D
a
t
e
o
f
d
i
s
p
a
t
c
h
A
0
3
C
e
r
t
i
f
i
c
a
t
e
N
o
.
I
N
S
P
E
C
T
I
O
N
R
E
P
O
R
T
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
1
9
9
1
+
A
1
:
1
9
9
5
+
A
S
P
E
R
I
S
O
1
0
4
7
4
:
1
9
9
1
2
7
6
2
5
3
-
1
7
.
1
0
.
0
7
3
2
5
6
5
1
-
0
0
1
3
/
.
.
.
A
0
5
E
s
t
a
b
l
i
s
h
e
d
I
n
s
p
e
c
t
i
n
g
b
o
d
y
A
0
6
P
u
r
c
h
a
s
e
r
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
0
7
.
1
N
o
.
3
0
2
0
9
4
7
6
0
1
P
r
o
d
u
c
t
L
R
F
i
n
a
l
r
e
c
e
i
v
e
r
A
0
7
.
2
N
o
.
H
O
T
R
O
L
L
E
D
P
L
A
T
E
S
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
6
0
2
1
S
t
e
e
l
d
e
s
i
g
n
.
D
I
L
L
I
M
A
X
6
9
0
T
6
0
3
A
n
y
s
u
p
p
l
.
D
I
L
L
I
N
G
-
E
0
4
:
0
3
r
e
a
u
i
r
e
m
e
n
t
s
B
0
1
-
B
9
9
D
e
s
c
r
i
p
t
i
o
n
o
f
t
h
e
p
r
o
d
u
c
t
B
1
4
B
0
8
B
0
9
B
1
0
B
1
1
B
1
2
B
0
4
B
0
7
.
2
B
0
7
.
1
A
0
9
I
t
e
m
N
u
m
b
e
r
T
h
i
c
k
n
e
s
s
W
i
d
t
h
L
e
n
g
t
h
T
h
e
o
r
e
t
i
c
a
l
P
r
o
d
u
c
t
d
e
l
i
v
e
r
y
H
e
a
t
N
o
.
R
o
l
l
e
d
p
l
a
t
e
N
o
.
1
P
u
r
c
h
a
s
e
r
a
r
t
i
c
l
e
n
u
m
b
e
r
N
o
.
o
f
p
i
e
c
e
s
m
a
s
s
c
o
n
d
i
t
i
o
n
T
e
s
t
N
o
.
M
M
K
G
2
4
1
6
0
,
0
0
x
2
5
0
0
x
1
2
0
0
0
1
4
4
0
0
Q
+
A
3
1
2
7
0
0
4
0
8
9
8
-
0
1
*
*
*
3
5
1
8
2
5
0
0
B
0
6
M
a
r
k
i
n
g
o
f
t
h
e
p
r
o
d
u
c
t
I
T
E
M
N
O
.
:
0
1
-
1
1
,
1
3
,
1
9
-
2
4
S
T
E
E
L
D
E
S
I
G
N
A
T
I
O
N
D
I
L
L
I
M
A
X
6
9
0
T
H
E
A
T
N
O
.
/
T
R
A
D
E
M
A
R
K
/
R
O
L
L
E
D
P
L
A
T
E
N
O
.
-
T
E
S
T
N
O
.
/
I
N
S
P
E
C
T
O
R
'
S
S
T
A
M
P
C
1
0
-
C
2
9
T
e
n
s
i
l
e
t
e
s
t
B
1
4
S
0
7
.
2
B
0
7
.
1
B
0
5
C
0
1
C
0
2
1
C
0
3
C
1
0
C
1
1
C
1
2
C
1
3
C
1
4
C
1
5
I
t
e
m
H
e
a
t
N
o
.
R
o
l
.
p
l
a
t
e
l
R
e
f
e
r
e
n
c
e
(
h
e
a
t
)
t
r
e
a
t
m
e
n
t
C
0
1
T
e
m
p
.
A
N
o
.
T
e
s
t
N
o
.
G
R
.
C
M
P
A
%
R
E
H
R
P
0
2
R
M
L
O
=
5
D
0
2
3
1
2
0
0
5
3
5
7
9
9
*
K
4
Q
R
T
7
5
0
7
9
4
1
9
,
0
0
3
3
1
1
9
1
3
2
6
8
2
9
K
4
Q
R
T
8
4
6
8
4
8
1
8
,
1
0
4
3
1
1
9
1
1
2
6
8
3
9
K
4
Q
R
T
8
0
5
8
2
9
1
8
,
2
0
5
3
1
1
9
1
3
2
6
8
2
6
K
4
Q
R
T
7
8
8
8
3
1
1
8
,
4
0
6
3
1
1
9
1
5
2
6
7
8
7
K
4
Q
R
T
8
0
7
8
5
1
1
6
,
8
0
7
3
1
1
9
1
3
2
6
7
9
8
K
4
Q
R
T
7
9
8
8
4
2
1
7
,
1
0
8
3
1
1
9
1
5
2
6
7
6
4
4
n
I
'
.
K
4
Q
7
9
2
8
4
3

1
7
,
9
<
b
i
Z
;
'
I
I
.
,
.
.
"
.
,
l
i
l
t
.
1
0
3
1
2
4
1
6
1
8
5
1
4
K
4
Q
R
T
8
3
7
8
6
1
1
6
,
1
1
3
3
1
2
4
1
6
1
8
6
0
3
*
K
4
Q
R
T
8
0
3
8
4
8
1
6
,
4
2
0
3
1
2
7
0
0
4
0
9
9
1
K
4
Q
V
R
T
8
2
7
8
7
2
1
7
,
0
2
2
3
1
2
7
0
0
4
0
8
5
5
*
K
4
Q
V
R
T
7
8
7
8
4
4
1
9
,
4
2
3
3
1
2
7
0
0
4
0
8
9
5
*
K
4
Q
V
R
T
8
0
4
8
6
2
1
7
,
2
/
'
'
\
A
0
4
W
e
h
e
r
e
b
y
c
e
r
t
i
f
y
,
t
h
a
t
t
h
e
a
b
o
v
f
m
e
n
t
i
o
t
d
m
a
t
e
r
i
a
l
s
h
a
v
e
b
e
e
n
d
e
l
i
v
e
r
e
d
i
n
a
c
c
o
r
d
a
n
c
e
A
G
d
e
r
D
i
l
l
i
n
g
e
r
H
u
t
t
e
n
w
e
r
k
e
A
0
1
D
X
H
H
.
-
J
.
C
y
r
a
n
s
k
R
J
f
,
f
-
J
:
l
'
r
l
f
-
A
B
a
P
o
s
t
f
a
c
h
1
5
8
0
,
D
-
6
6
7
4
8
D
i
l
l
i
n
g
e
n
/
S
a
a
r
S
t
u
t
t
g
a
r
t
O
f
f
i
c
e
1
A
.
(
\
I
n
s
p
e
c
t
i
o
n
d
e
p
a
r
t
m
e
n
t
M
a
n
u
f
a
c
t
u
r
e
r
'
s
L
l
o
y
d
'
s
R
e
g
i
s
t
e
r
E
r
.
,
.
'F
.
A
S
e
r
'
B
.
M
U
E
L
L
E
R
m
a
r
k
-
s
p
e
c
t
o
r
T
e
s
t
H
o
u
s
e
M
a
n
a
a
e
r
I
n
s
p
e
c
t
o
r
's
s
t
a
m
p
D
a
t
e
1
7
.
1
0
.
0
7
E
D
I
H
K
1
E
r
l
a
u
t
e
r
u
n
o
e
n
s
i
e
h
e
R
O
c
k
s
e
i
t
e
/
E
x
o
l
i
c
a
t
i
o
n
s
v
o
i
r
a
u
v
e
r
s
o
/
S
e
e
r
e
v
e
r
s
e
f
o
r
e
x
o
l
a
n
a
t
i
o
n
s
(
w
w
w
.
d
i
l
l
i
n
o
e
r
.
d
e
/
c
e
r
t
i
f
i
c
a
t
e
l
D
I
L
L
I
N
G
E
R
H
U
T
T
E
A
0
2
I
N
S
P
E
C
T
I
O
N
C
E
R
T
I
F
I
C
A
T
E
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
2
0
0
4
I
N
S
P
E
C
T
I
O
N
R
E
P
O
R
T
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
1
9
9
1
+
A
1
:
1
9
9
5
+
A
S
P
E
R
I
S
O
1
0
4
7
4
:
1
9
9
1
A
1
0
A
d
v
i
c
e
o
f
d
i
s
p
a
t
c
h
N
o
.
1
D
a
t
e
o
f
d
i
s
p
a
t
c
h
2
7
6
2
5
3
-
1
7
.
1
0
.
0
7
A
0
8
1
M
a
n
u
f
a
c
t
u
r
e
r
'
s
o
r
d
e
r
l
A
0
3
C
e
r
t
i
f
i
c
a
t
e
N
o
.
3
2
5
6
5
1
-
0
0
1
S
h
e
e
t
4
/
.
.
.
A
0
5
E
s
t
a
b
l
i
s
h
e
d
I
n
s
p
e
c
t
i
n
g
b
o
d
y
L
R
A
0
6
P
u
r
c
h
a
s
e
r
F
i
n
a
l
r
e
c
e
i
v
e
r
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
0
7
.
1
N
o
.
A
0
7
.
2
N
o
.
3
0
2
0
9
4
7
6
0
1
P
r
o
d
u
c
t
H
O
T
R
O
L
L
E
D
P
L
A
T
E
S
6
0
2
1
S
t
e
e
l
d
e
s
i
g
n
.
6
0
3
A
n
y
s
u
p
p
l
.
r
e
q
u
i
r
e
m
e
n
t
s
D
I
L
L
I
M
A
X
6
9
0
T
D
I
L
L
I
N
G
-
E
0
4
:
0
3
C
1
0
-
C
2
9
T
e
n
s
i
l
e
t
e
s
t
6
1
4
6
0
7
.
2
6
0
7
.
1
6
0
5
C
0
1
C
0
2
1
C
0
3
C
1
0
C
1
1
C
1
2
C
1
3
C
1
4
-
C
1
5
I
t
e
m
H
e
a
t
N
o
.
R
o
l
.
p
l
a
t
e
l
R
e
f
e
r
e
n
c
e
(
h
e
a
t
)
t
r
e
a
t
m
e
n
t
C
0
1
T
e
m
p
.
A
N
o
.
T
e
s
t
N
o
.
G
R
.
C
M
P
A
%
R
E
H
R
P
0
2
R
M
L
O
=
5
D
2
4
3
1
2
7
0
0
4
1
1
3
9
*
K
4
Q
V
R
T
7
6
2
8
3
5
1
7
,
0
C
4
0
-
C
4
9
I
m
p
a
c
t
t
e
s
t
6
1
4
8
0
7
.
2
6
0
7
.
1
6
0
5
C
0
1
C
0
2
1
C
0
3
C
4
1
C
4
0
C
4
4
C
4
6
C
4
5
C
4
2
C
4
3
I
t
e
m
H
e
a
t
N
o
.
R
o
l
.
p
l
a
t
e
!
R
e
f
e
r
e
n
c
e
(
h
e
a
t
)
t
r
e
a
t
m
e
n
t
C
0
1
T
e
m
p
.
W
i
d
t
h
o
f
T
y
p
e
o
f
T
e
s
t
i
n
g
m
e
t
h
o
d
E
n
e
r
g
y
I
n
d
i
v
i
d
u
a
l
v
a
l
u
e
s
M
e
a
n
N
o
.
T
e
s
t
N
o
.
G
R
.
C
t
e
s
t
p
i
e
c
e
t
e
s
t
p
i
e
c
e
J
o
u
l
e
A
V
=
J
v
a
l
u
e
0
2
3
1
2
0
0
5
3
5
7
9
9
*
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
7
6
1
5
3
1
2
1
1
5
0
0
3
3
1
1
9
1
3
2
6
8
2
9
'
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
9
4
1
9
4
2
0
5
1
9
8
0
4
3
1
1
9
1
1
2
6
8
3
9
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
7
7
1
6
2
1
8
2
1
7
4
0
5
3
1
1
9
1
3
2
6
8
2
6
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
7
8
1
8
2
1
7
5
1
7
8
0
6
3
1
1
9
1
5
2
6
7
8
7
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
8
1
1
9
4
1
7
9
1
8
5
0
7
3
1
1
9
1
3
2
6
7
9
8
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
5
4
1
7
7
1
7
7
1
6
9
0
8
3
1
1
9
1
5
2
6
7
6
4
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
0
9
1
0
5
8
8
1
0
1
1
0
3
1
2
4
1
6
1
8
5
1
4
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
1
3
2
1
0
3
1
2
7
1
2
1
1
3
3
1
2
4
1
6
1
8
6
0
3
*
K
4
Q
O
-
4
0
C
H
P
-
V
6
0
0
A
V
9
1
8
3
8
9
8
8
2
0
3
1
2
7
0
0
4
0
9
9
1
K
4
Q
V
-
4
0
C
H
P
-
V
6
0
0
A
V
2
0
7
2
1
3
2
2
0
2
1
3
2
2
3
1
2
7
0
0
4
0
8
5
5
*
K
4
Q
V
-
4
0
C
H
P
-
V
6
0
0
A
V
2
0
8
2
1
6
2
1
8
2
1
4
2
3
3
1
2
7
0
0
4
0
8
9
5
*
K
4
Q
V
-
4
0
C
H
P
-
V
6
0
0
A
V
9
7
1
2
9
1
2
6
1
1
7
2
4
3
1
2
7
0
0
4
1
1
3
9
*
K
4
Q
V
-
4
0
.
C
H
P
-
V
6
0
0
A
V
5
0
.
-
9
4
.
.
.
.
.
1
0
1
8
2
"
l
!
!
'
I
!
!
l
I
I
I
I
l
.
.
-
-
!
I
I
F
I
A
G
d
e
r
D
i
l
l
i
n
g
e
r
H
i
i
t
t
e
n
w
e
r
k
e
P
o
s
t
f
a
c
h
1
5
8
0
,
0
-
6
6
7
4
8
D
i
l
l
i
n
g
e
n
/
S
a
a
r
I
n
s
p
e
c
t
i
o
n
d
e
p
a
r
t
m
e
n
t
A
0
4
D
X
H
M
a
n
u
f
a
c
t
u
r
e
r
'
s
m
a
r
k
W
e
h
e
r
e
b
y
c
e
r
t
i
f
y
,
t
h
a
t
t
h
e
a
b
o
v
e
:
r
e
n
t
i
o
n
e
\
m
a
t
e
r
i
a
l
s
h
a
v
e
b
e
e
n
d
e
l
i
v
e
r
e
d
i
n
a
c
c
o
r
d
a
n
c
e
I
w
n
n
t
n
e
v
,
v
,
,
,
,
,
,
.
'
f
.
x
I
H
.
-
J
.
C
y
r
a
n
s
k
,
S
t
u
t
t
g
a
r
t
O
f
f
i
c
e
S
J
1
A
.
i
\
/
U
O
Y
d
'
S
R
e
g
i
s
t
e
r
E
M
E
A
F
t
e
_
i
f
-
I
.7
T
n
e
c
t
o
r
l
'
r
/
!
-
B
.
M
U
E
L
L
E
R
T
e
s
t
H
o
u
s
e
M
a
n
a
o
e
r
I
n
s
o
e
c
t
o
r
'
s
s
l
a
m
o
D
a
l
e
1
7
.
1
0
.
0
7
E
D
I
H
K
1
A
0
1
E
r
l
a
u
t
e
r
u
n
q
e
n
s
i
e
h
e
R
G
c
k
s
e
i
t
e
/
E
x
p
l
i
c
a
t
i
o
n
s
v
o
i
r
a
u
v
e
r
s
o
/
S
e
e
r
e
v
e
r
s
e
f
o
r
e
x
p
l
a
n
a
t
i
o
n
s
(
w
w
w
.
d
i
l
l
i
n
g
e
r
.
d
e
/
c
e
r
t
i
f
i
c
a
t
e
l
D
I
L
L
I
N
G
E
R
H
U
T
T
E
A
0
2
I
N
S
P
E
C
T
I
O
N
C
E
R
T
I
F
I
C
A
T
E
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
2
0
0
4
I
N
S
P
E
C
T
I
O
N
R
E
P
O
R
T
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
1
9
9
1
+
A
1
:
1
9
9
5
+
A
S
P
E
R
I
S
O
1
0
4
7
4
:
1
9
9
1
A
1
0
A
d
v
i
c
e
o
f
d
i
s
p
a
t
c
h
N
o
.
1
D
a
t
e
o
f
d
i
s
p
a
t
c
h
2
7
6
2
5
3
-
1
7
.
1
0
.
0
7
A
o
a
l
M
a
n
u
f
a
c
t
u
r
e
r
'
s
o
r
d
e
r
l
A
0
3
C
e
r
t
i
f
i
c
a
t
e
N
o
.
3
2
5
6
5
1
-
0
0
1
S
h
e
e
t
5
/
.
.
.
A
0
5
E
s
t
a
b
l
i
s
h
e
d
I
n
s
p
e
c
t
i
n
g
b
o
d
y
L
R
A
0
6
P
u
r
c
h
a
s
e
r
F
i
n
a
l
r
e
c
e
i
v
e
r
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
0
7
.
1
N
o
.
A
0
7
.
2
N
o
.
3
0
2
0
9
4
7
6
0
1
P
r
o
d
u
c
t
H
O
T
R
O
L
L
E
D
P
L
A
T
E
S
B
0
2
1
S
t
e
e
l
d
e
s
i
g
n
.
B
0
3
A
n
y
s
u
p
p
l
.
r
e
q
u
i
r
e
m
e
n
t
s
D
I
L
L
l
M
A
X
6
9
0
T
D
I
L
L
I
N
G
-
E
0
4
:
0
3
C
7
0
-
C
9
9
C
h
e
m
i
c
a
l
c
o
m
p
o
s
i
t
i
o
n
%
-
H
e
a
t
a
n
a
l
y
s
i
s
6
0
7
.
2
C
7
0
H
e
a
t
C
5
1
M
N
P
S
N
C
U
M
O
N
I
C
R
V
N
B
T
I
B
3
1
1
9
1
1
Y
0
,
1
6
5
0
,
2
7
5
1
,
2
9
0
,
0
1
0
0
,
0
0
0
6
0
,
0
0
3
9
0
,
0
1
4
0
,
1
1
1
0
,
0
4
5
0
,
0
2
7
0
,
0
0
0
0
,
0
2
4
0
,
0
0
3
0
,
0
0
1
9
3
1
1
9
1
3
y
0
,
1
7
0
0
,
2
8
9
1
,
2
9
0
,
0
1
0
0
,
0
0
0
8
0
,
0
0
2
7
0
,
0
2
3
0
,
1
0
1
0
,
0
2
4
0
,
0
2
4
0
,
0
0
0
0
,
0
2
4
0
,
0
0
4
0
,
0
0
1
9
3
1
1
9
1
5
y
0
,
1
6
0
0
,
2
8
2
1
,
2
8
0
,
0
1
1
0
,
0
0
0
7
0
,
0
0
4
2
0
,
0
2
4
0
,
0
9
9
0
,
0
3
2
0
,
0
2
5
0
,
0
0
0
0
,
0
2
4
0
,
0
0
3
0
,
0
0
1
9
3
1
2
0
0
5
Y
0
,
1
3
4
0
,
2
6
5
1
,
4
3
0
,
0
1
1
0
,
0
0
1
1
0
,
0
0
3
5
0
,
0
1
6
0
,
0
0
7
0
,
0
2
3
0
,
3
0
1
0
,
0
0
1
0
,
0
2
0
0
,
0
0
2
0
,
0
0
1
7
3
1
2
4
1
6
y
0
,
1
6
8
0
,
2
7
5
1
,
2
9
0
,
0
1
1
0
,
0
0
0
5
0
,
0
0
3
1
0
,
0
1
7
0
,
2
0
7
0
,
0
3
2
0
,
3
0
4
0
,
0
0
0
0
,
0
2
9
0
,
0
0
2
0
,
0
0
1
8
3
1
2
7
0
0
Y
0
,
1
6
4
0
,
2
6
9
1
,
2
9
0
,
0
0
9
0
,
0
0
1
1
0
,
0
0
4
5
0
,
0
1
8
0
,
2
4
7
0
,
1
1
6
0
,
3
0
4
0
,
0
0
1
0
,
0
2
7
0
,
0
0
2
0
,
0
0
1
9
6
0
7
.
2
C
7
0
H
e
a
t
Z
R
A
L
-
T
3
1
1
9
1
1
Y
0
,
0
0
0
2
0
,
0
7
9
3
1
1
9
1
3
Y
0
,
0
0
0
2
0
,
0
7
7
3
1
1
9
1
5
Y
0
,
0
0
0
2
0
,
0
8
0
3
1
2
0
0
5
Y
0
,
0
0
0
2
0
,
0
5
4
3
1
2
4
1
6
y
0
,
0
0
0
2
0
,
0
8
1
3
1
2
7
0
0
Y
0
,
0
0
0
2
0
,
0
7
7
C
9
4
H
e
a
t
a
n
a
l
y
s
i
s
C
a
r
b
o
n
e
q
u
i
v
a
l
e
n
t
I
A
l
l
o
y
i
n
g
r
e
s
t
r
i
c
t
i
o
n
s
6
0
7
.
2
H
e
a
t
3
1
1
9
1
1
F
O
-
0
2
=
0
,
4
1
F
O
-
5
1
=
0
,
0
2
3
1
1
9
1
3
F
O
-
0
2
=
0
,
4
J
.
,
F
O
-
5
1
=
0
,
0
2
~
.
~
I
r
.
.
.
.
r
I
1
i
a
"
"
,
/
;
'
;
;
"
:tk
-.,
I
I
,
.
.
~
"
I
I
I
,
s
.
~
"
y
3
1
1
9
1
5
F
O
-
0
2
=
0
,
4
0
F
O
-
5
1
=
0
,
0
2
3
1
2
0
0
5
F
O
-
0
2
=
0
,
4
4
F
O
-
5
1
=
0
,
0
2
3
1
2
4
1
6
F
O
-
0
2
=
0
,
4
9
F
O
-
5
1
=
0
,
0
3
3
1
2
7
0
0
F
O
-
0
2
=
0
,
5
0
F
O
-
5
1
=
0
,
0
3
A
G
d
e
r
D
i
l
l
i
n
g
e
r
H
u
t
t
e
n
w
e
r
k
e
P
o
s
t
f
a
c
h
1
5
8
0
,
0
-
6
6
7
4
8
D
i
l
l
i
n
g
e
n
/
S
a
a
r
I
n
s
p
e
c
t
i
o
n
d
e
p
a
r
t
m
e
n
t
A
0
4
D
X
H
M
a
n
u
f
a
c
t
u
r
e
r
'
s
m
a
r
k
l
'
r
l
f
-
B
.
M
U
E
L
L
E
R
T
e
s
t
H
o
u
s
e
M
a
n
a
a
e
r
I
n
s
p
e
c
t
o
r
'
s
s
t
a
m
o
D
a
t
e
1
7
.
1
0
.
0
7
E
D
I
H
K
1
A
0
1
E
r
l
a
u
t
e
r
u
n
q
e
n
s
i
e
h
e
R
O
c
k
s
e
i
t
e
/
E
x
p
l
i
c
a
t
i
o
n
s
v
o
i
r
a
u
v
e
r
s
o
/
S
e
e
r
e
v
e
r
s
e
f
o
r
e
x
p
l
a
n
a
t
i
o
n
s
(
w
w
w
.
d
i
l
l
i
n
g
e
r
.
d
e
/
c
e
r
t
i
f
i
c
a
t
e
l
.
.
D
I
L
L
I
N
G
E
R
H
U
T
T
:
E
A
0
2
I
N
S
P
E
C
T
I
O
N
C
E
R
T
I
F
I
C
A
T
E
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
2
0
0
4
I
N
S
P
E
C
T
I
O
N
R
E
P
O
R
T
3
.
2
A
S
P
E
R
E
N
1
0
2
0
4
:
1
9
9
1
+
A
1
:
1
9
9
5
+
A
S
P
E
R
I
S
O
1
0
4
7
4
:
1
9
9
1
A
1
0
A
d
v
i
c
e
o
f
d
i
s
p
a
t
c
h
N
o
.
1
D
a
t
e
o
f
d
i
s
p
a
t
c
h
A
0
8
1
M
a
n
u
l
a
c
t
u
r
e
r
'
s
o
r
d
e
r
l
A
0
3
C
e
r
t
i
f
i
c
a
t
e
N
o
.
S
h
e
e
t
2
7
6
2
5
3
-
1
7
.
1
0
.
0
7
3
2
5
6
5
1
-
0
0
1
6
A
0
5
E
s
t
a
b
l
i
s
h
e
d
I
n
s
p
e
c
t
i
n
g
b
o
d
y
L
R
A
0
6
P
u
r
c
h
a
s
e
r
F
i
n
a
l
r
e
c
e
i
v
e
r
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
N
C
O
F
E
R
W
A
L
D
R
A
M
,
O
O
S
T
E
A
0
7
.
1
N
o
.
A
0
7
.
2
N
o
.
3
0
2
0
9
4
7
6
0
1
P
r
o
d
u
c
t
H
O
T
R
O
L
L
E
D
P
L
A
T
E
S
6
0
2
1
S
t
e
e
l
d
e
s
i
g
n
.
8
0
3
A
n
y
s
u
p
p
i
.
r
e
q
u
i
r
e
m
e
n
t
s
D
I
L
L
I
M
A
X
6
9
0
T
D
I
L
L
I
N
G
-
E
0
4
:
0
3
C
9
4
C
a
r
b
o
n
e
q
u
i
v
a
l
e
n
t
f
o
r
m
u
l
a
I
A
l
l
o
y
i
n
g
r
e
s
t
r
i
c
t
i
o
n
s
F
O
-
0
2
=
C
+
(
M
N
/
6
)
+
(
C
R
+
M
O
+
V
)
/
5
+
(
N
I
+
C
U
)
/
1
5
F
O
-
5
1
=
V
+
N
B
0
0
1
M
a
r
k
i
n
g
a
n
d
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
s
u
r
f
a
c
e
a
p
p
e
a
r
a
n
c
e
,
s
h
a
p
e
a
n
d
d
i
m
e
n
s
i
o
n
a
l
p
r
o
p
e
r
t
i
e
s
I
T
E
M
N
O
.
:
0
1
-
1
1
,
1
3
,
1
9
-
2
4
R
E
S
U
L
T
O
F
M
A
R
K
I
N
G
,
S
U
R
F
A
C
E
,
S
H
A
P
E
A
N
D
D
I
M
E
N
S
I
O
N
S
:
N
O
R
E
M
A
R
K
S
S
U
R
F
A
C
E
A
S
P
E
R
E
N
-
1
0
1
6
3
-
A
2
T
H
I
C
K
N
E
S
S
A
S
P
E
R
E
N
-
1
0
0
2
9
-
A
L
E
N
G
T
H
A
N
D
W
I
D
T
H
A
S
P
E
R
E
N
-
1
0
0
2
9
F
L
A
T
N
E
S
S
A
S
P
E
R
E
N
-
1
0
0
2
9
-
T
4
H
Z
0
1
-
Z
9
9
C
o
n
f
i
r
m
a
t
i
o
n
s
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
1
I
T
E
M
N
O
.
:
0
1
-
1
1
,
1
3
,
1
9
-
2
4
T
E
R
M
S
A
N
D
C
O
N
D
I
T
I
O
N
S
O
F
A
.
M
.
S
O
C
I
E
T
Y
(
A
0
5
)
A
P
P
L
Y
:
L
L
O
Y
D
S
R
E
G
I
S
T
E
R
E
M
E
A
;
]
t
I
.
.
.
_
_
_
_
.
-
.
_
!
!
!
!
I
I
!
.
',.t'!''s
l!!l.J
!l!!.
-
-
-
-
-
-
-
-
A
G
d
e
r
D
i
l
l
i
n
g
e
r
H
u
t
t
e
n
w
e
r
k
e
P
o
s
t
f
a
c
h
1
5
8
0
,
0
-
6
6
7
4
8
D
i
l
l
i
n
g
e
n
/
S
a
a
r
I
n
s
p
e
c
t
i
o
n
d
e
p
a
r
t
m
e
n
t
A
0
4
D
X
H
M
a
n
u
f
a
c
t
u
r
e
r
'
s
m
a
r
k
Z
0
1
/
Z
0
2
/
Z
0
3
W
e
h
e
r
e
b
y
c
e
r
t
i
f
y
,
t
h
a
t
t
h
e
m
a
t
e
r
i
a
l
s
h
a
v
e
b
e
e
n
a
e
l
i
v
e
r
e
d
i
n
a
c
c
o
r
d
a
n
c
e
w
i
t
h
t
h
e
t
e
r
m
s
o
f
o
r
d
e
r
.
/
\
.
H
.
-
J
.
C
y
r
a
n
s
k
i
l
l
b
v
d
I
I
"
x
S
t
u
t
t
g
a
r
t
O
f
f
i
c
e
R
i
S
L
l
o
y
d
'
s
!
:
M
E
A
I
Y
'
I
J
A
A

v
l
'
r
l
f
-
B
.
M
U
E
L
L
E
R
T
e
s
t
H
o
u
s
e
M
a
n
a
g
e
r
I
n
s
p
e
c
t
o
r
's
s
t
a
m
p
D
a
t
e
1
7
.
1
0
.
0
7
E
D
I
H
K
1
A
0
1
166
167
AnnexE:ProductiondatasheetsTIGDressing
On the following pages the production data sheets from the TIGdressing are displayed. Unfortunately some of the
commentsareinDutchlanguage.
PRODUCTI ON D A T A S H E E T
f or WELD TOE I MPROVEMENT by
TIG DRESSING
Jf l Pm; \ Internati onal Institute of
l Âpr - Wel di ng
^=
3
="' Commi ssi on XIII
WELDINp SPECIFICATION
Base material: 3 rPtj O
Filler material:
Welding procedure No: C C?
c
\
COMPONENT
Type:
Identification: <! Cvh ^
EQUIPMENT
Make and model: l A<duA pi Pv
Power capacity: QS'oo fikv^ f
Electrode diameter: ^
y
2
Gas cup diameter: I 0 A V .
Shieldina aas: AR<>tJ L.\ b
SKETCH OR PHOTO
TREATMENT DATA
Dressing position HG. 2G) etc: \ (z>
Preheat temperature: OFr-> 1^
Gas flow rate: 11>~ ^ /
Work angle, sideways: r*> O - o
Work angle, weld direction: ( C\
c
Travel speed (S): /' Q.2- / T
Vol t aaef VWp* - i% Current fl): 9 ^ ^ /
u
6 0 x S x /
Heat input: n=
1 000x1 /
(kJ/mm)
SKETCH OR PHOTO
/WSPEC7/OW
Visual jELPhoto Geometry
measurements
Equipment: Lc AN^j C
REMARKS
TOE GEOMETRY MEASUREMENTS
Spacing of measurement points (mm):
Max. Mi n. Mean
Toe radius (mm):
Groove depth (mm):
Measurements report:
REMARKS
OPERATOR
Experience: a). Hours: b). Length of weld treated: i ^ ^
Operator's name: CA \Aa^^^T\ W-e' L. rA
nota- S^(JÎn b 111
1 uai e. '
PRODUCTI ON DATA S H E E T
TIG DRESSING
International Insti tute of
Wel di ng
Commi ssi on XIII
WELDING SPECIFICA TION
Base material: d~. a
Filler material:
Welding procedure No:
COMPONENT
Type:
Identification: Z
EQUIPMENT
Make and model: -.^v^fiP^
SKETCH OR PHOTO
Power capacity: f -c y> JA-IAV, /"
Electrode diameter: _Ai 2.
Gas cup diameter: ^ i s / w ^
Shielding gas: ^jJX** (=\
1

TREATMENT DATA
Dressing position (1G, 2G) etc:
Preheat temperature: O C -
Gas flow rate: l , r^-
Work angle, sideways: for> - w
a ''
Work angle, weld direction: ( Q
Travel speed (S): l?-2 ^ w - y
Voltage (V): . Current (I): ^ T A^-yj
Heat input: H=
(kJ/mm)
1 0 0 0 x #
INSPECTION
Visual Photo Geometry
measurements
Equipment:
Spacing of measurement points (mm): ,
REMARKS
x
-a l o o v Aw^.
Max. Mi n.
l ean
Toe radius (mm):
Groove depth (mm):
leasurements report:
OPERATOR
?
,
b). Length of weld treated: OZL~L
l,VVv<A
Experience: a). Hours:
Operator's name: (^2i <j
Date
TIG DRESSING
Internati onal Insti tute of
Wel di ng
Commi ssi on XIII
WELDING SPECIFICATION
Base material: S 6
Filler material:
Welding procedure No: d " T ^ QJL<h) rV>
EQUIPMENT
Make and model: /^pt
Power capacity: '2 Cr ^j f ) fin p
Electrode diameter:
Gas cup diameter:
Shielding gas: /) Ac y w ^ ; /
TREATMENT DATA
Dressing position (1G, 2G) etc: < C
Preheat temperature: AÔ
Gas flow rate: C zAh
Work angle, sideways:
Work angle, weld direction: jfi
O
Travel speed (S): j 0 1 / ^. ^ / f/ua
Voltage (V): 0 - / ^ Current (I):' Artf
Heat input: H=
(kJ/mm)
6 0 x S x /
1 000x1 /
COMPONENT
Type:
Identification:
SKETCH OR PHOTO
INSPECTION
measurements
Equipment:
REMARKS
Max. Mi n. Mean
Toe radius (mm): .
Groove depth (mm):
Operator's name: /
OPERATOR
b). Length of weld treated: "b
Date: i l - 20 U
TIG DRESSING
Internati onal Institute of
Wel di ng
Commi ssi on XIII
Base material: _> //COCJ
Filler material:
Welding procedure No: i jf-; p i / ? / : > S " V ' ' ^
COMPONENT
Type: ,
Identification: ( 11
EQUIPMENT
Make and model: kxj*j.
Power capacity: lÔ* p
Electrode diameter:
Gas cup diameter: _
SKETCH OR PHOTO
y 2 <zi
Shielding gas: Afiy x-J 2. . . d
TREATMENT DATA
Preheat temperature:
Gas flow rate: IS I P/r>>
I A/
A Q_ Work angle, sideways:
Work angle, weld direction: io
Travel speed (S)
Voltage (V): i}~ f< Current (I):
<
j Aw
Heat input: H=
(kJ/mm)
6 0 x 5 x 7
1 000x1 /
INSPECTION
measurements
Equipment: .
REMARKS
5
?
Max. l i n. Mean
Toe radius (mm):
Groove depth (mm): _
OPERATOR
b). Length of weld treated: 3,3>2 mm Experience: a). Hours: ,
Operator's name: O - C$\Z/X*S 6t.ro
Date: ).M / , - 96 jt
I
TIG DRESSING
Internati onal Institute of
Wel di ng
Commi ssi on XIII
Base material: & X
c
f<^>
Filler material:
Welding procedure No: / ij d&M
EQUIPMENT
Make and model: /(kniPp><
Power capacity: _ 2. ^0
7.1 A? /7l Electrode diameter:
Gas cup diameter: _
Shielding gas: AiiQa-AJ u-
TREATMENT DATA
Dressing position (1G, 2G) etc: ' C~
Preheat temperature: - / 2 -6 *_
c
Gas flow rate: /r> L I
Work angle, sideways: / o 0/D
Work angle, weld direction: / d
Travel speed (S): th^Cj T
Voltage (V): / ?, <l Current (I): Q^ Ah f
(
Heat input: H=
(kJ/mm)
6 0 x S x /
-\000xV
INSPECTION
measurements
Equipment:
TO GEOMETRY MEASUREMENTS
Max. l i n. Mean
Toe radius (mm): _
Groove depth (mm): .
leasurements report:
COMPONENT
Type:
Identification:
SKETCH OR PHOTO
REMARKS
/ \ f 2 i/o I C
A W/ / d t o W4S c/ f
vô o -2, **T7 ^ <rs^ ^ A
OPERATOR
b). Length of weld treated: t j / j j
Operator's name: GJ jDg. f^so.s c?A
C
^
Date
TIG DRESSING
|\. International Insti tute of
Wel di ng
Commi ssi on XIII
Base material: > . E^fg
/
Filler material:
EQUIPMENT
Make and model: A/ f / r VPp'
Power capacity: 1 -S' o P
2 Electrode diameter: _ _
Gas cup diameter: /> /n
Shielding gas: $L0J ON
TREATMENT DATA
Dressing position (1G, 2G) etc: > &
Preheat temperature: / OO
c
Gas flow rate:
Work angle, sideways:
Work angle, weld direction: />
Travel speed (S):
m
Voltage (V): /Jh
/
7
~
/ S Current (I): OÔ
Heat input: H=
(kJ/mm)
6 0 x S x /
1 000xV
COMPONENT
Type:
Identification:
7Z
SKETCH OR PHOTO
INSPECTION
measurements
Equipment:
REMARKS .
/ OO
C
C
Max.
Iin.
Mean
Toe radius (mm):
Groove depth (mm):
OPERATOR
Experience: a). Hours: b). Length of weld treated: 3>3p tt^/f
Operator's name: (g^/t^l &-
-3-
Date : 3 2&LL
TIG DRESSING
M^M Wel di ng
" Commi ssi on XIII
Base material: Q ^ f
C J
Filler material:
Weldinq procedure No: " T^nt ao^sl v2
COMPONENT
Type:
Identification: V A*
EQUIPMENT
Make and model: idiS/npp'
c
i.Ôci W
i I
Power capacity: ' 2 ^ o j@sn^
Electrode diameter: r*> /*v
Gas cup diameter: JO t*
1
^
Shieldina aas: A/Z4,a>sS /
{
^
SKETCH OR PHOTO
TREATMENT DATA
Dressina position (1G. 2G) etc: ' d
Preheat temperature: DO
<
Gas flow rate: /vT Z f /r*.t,\>

Work anqle. sideways: />o -
Work angle, weld direction: '
Travel speed (S): f 'jH* I
VoI t aaeM: / 3 - / c ? Current (I): 2 b / V J ?
, . 6 0 x S x /
Heat input: n = =
1 000x1 /
(kJ/mm)
SKETCH OR PHOTO
INSPECTION
measurements
Equipment:
REMARKS ,
/QO / ? <r> p
1 3 - / / , V
Max. Mi n. Mean
Toe radius (mm):
Groove depth (mm):
REMARKS ,
/QO / ? <r> p
1 3 - / / , V
OPERATOR
Experience: a). Hours: b). Length of weld treated: 3QV'j /"/*
Date: 3 ^ I gg ' /
TIG DRESSING
/{jifjwlfc, International Insti tute of
Wel di ng
Commi ssi on XIII
Base material: - 8><a CJ
Filler material:
COMPONENT
Tvoe:
Identification: ^
/
EQUIPMENT
Make and model: /Q-mpdi "2A'oo \rJ
J
r
\
Power capacity: 9 < C> A /ft \
v
Electrode diameter: 3 . % /w m
Gas cup diameter: /O m PI.
Shieldinq aas: A iloîu -6
SKETCH OR PHOTO
TREATMENT DATA
Dressing position (1G. 2G) etc: /
Preheat temperature: /'IS ~c
Gas flow rate: / / */mx/\j
Work anale, sideways: &t> - Qo '
Work anqle, weld direction: id
Travel speed (S): j'70 n /* t
Voltaae (V): / } - /&, Current (I): 2<<o p
u
.. . JeoxSxi
Heat input: H= =
1 000x1 /
(kJ/mm)
SKETCH OR PHOTO
INSPECTION
Visual Photo . Geometry
measurements
Equipment:
REMARKS
2 A p
Max. Mi n. Mean
Toe radius (mm):
Groove depth (mm):
REMARKS
2 A p
OPERATOR
Experience: a). Hours: b). Length of weld treated: 3 2 3 <*
IV1
Operator's name: . QQC>JO&'V hA<H
Date: A O - ^ 2 ^
5
TIG DRESSING
Wel di ng
Commi ssi on XIII
Base material: S l/lQ>
Filler material:
1
4 r
EQUIPMENT
Make and model: /L/k'^vPpi 9 , ^ o d ^
Power capacity:
Electrode diameter:
Gas cup diameter: .
Shielding gas: Hgc^ ^ . d
/ O
/ V ) / O- .
J
TREATMENT DATA
1

Preheat temperature: / Qr5> _
Gas flow rate: /-"> / /, I r
Work angle, sideways: - 9s
Work angle, weld direction: /a
d
<3_
Travel speed (S):
Voltage (V): / ? - / | Current (I): 2 Sj
Heat input: /-/=
(kJ/mm)
6 0 x 5 x /
1 000xV
COMPONENT
Type:
Identification: 1 / / (
SKETCH OR PHOTO
INSPECTION
measurements
Equipment:
REMARKS
Max.
lin.
l ean
Toe radius (mm): _
Groove depth ( mm): .
Experience: a). Hours: .
OPERATOR
b). Length of weld treated: 323 ^
INR
>
Operator's name:
t
Date: ?G I I
177
AnnexF:Matlabscriptstoprocessweldgeometrydata
Splittingoflaserpositionandmeasureddataandsubtractingaplane.

%split A in X, Y en Z values
gr=size(A);
rijen_m=gr(1,1);
kolommen_n=gr(1,2);
Xs=A(1,:);
X=Xs(1,2:kolommen_n(1,1)); %create vector containing x-values
Ys=A(:,1);
Y=Ys(2:rijen_m(1,1),1); %create vector containing y-values
Zster=A(2:rijen_m(1,1),2:kolommen_n(1,1)); %extract Z values

%Construct a plane to level the weld cross sections at 0
x1=0; %first point to equal to zero
y1=7;
z1=1.4481;

x2=53; %second point to equal to zero
y2=7;
z2=0.1411;

x3=0; %third point to equal to zero
y3=140;
z3=0.7353;

gridsizex=0.01; %define grid size of data
gridsizey=7;

F=[x1 y1 1; x2 y2 1; x3 y3 1]; v=[z1; z2; z3];
abc=F\v;

VLAK=zeros(rijen_m(1,1)-1,kolommen_n(1,1)-1);
for n=1:(kolommen_n(1,1)-1)
for m=1:(rijen_m(1,1)-1)
xwaarde=X(1,n);
ywaarde=Y(m,1);
VLAK(m,n)=abc(1,1)*xwaarde+abc(2,1)*ywaarde+abc(3,1);
end;
end;

%Substract created plane from Z values
Z=(Zster-VLAK)*-1;

%Plot Weld
surf(X,Y,Z,'LineWidth',0.1,'Linestyle','-') %create 3D plot
title('Weld geometry'), xlabel('Dimension across the weld'),
ylabel('Dimension along the weld'), axis equal;

clear Xs Ys Zster x1 y1 x2 y2 x3 y3 z1 z2 z3 xwaarde ywaarde VLAK gr m n
abc F v;
178

Determiningtheweldtoeradius
%define optimalisation boundaries

ondergrens=46; %lower boundary, use steps of size: [gridsizex]
bovengrens=47; %upper boundary, use steps of size: [gridsizex]
apphmid=2; %estimate of height of circle midpoint6
hoogtegrid=2.5; %height of grid in mm
breedtegrid=2.5; %width of grid in mm
gridstap=0.025; %grid spacing

%building of the grid
lgrid=breedtegrid/gridstap+1;
hgrid=hoogtegrid/gridstap+1;
hcoor=zeros(1,lgrid);
vcoor=zeros(hgrid,1);
for k=1:lgrid
hcoor(1,k)=(bovengrens+ondergrens)/2-breedtegrid/2+(k-1)*gridstap;
end
for l=1:hgrid
vcoor(l,1)=apphmid+hoogtegrid/2-(l-1)*gridstap;
end

%build two vectors to show the grid
vectorgridplothor=hcoor;
m=1;
while m<hgrid;
vectorgridplothor=[vectorgridplothor hcoor];
m=m+1;
end
vectorgridplotver=vcoor(1,1)*ones(1,lgrid);
mm=2;
while mm<hgrid+1;
hulpvector=vcoor(mm,1)*ones(1,lgrid);
vectorgridplotver=[vectorgridplotver hulpvector];
mm=mm+1;
end
scatter(vectorgridplothor,vectorgridplotver,'.g')
axis equal
hold on;

%for all gridpoints, calculate distance to all data points within
%optimalisation boundaries, determine average distance and deviation
Mafw=zeros(hgrid,lgrid);
Mstr=zeros(hgrid,lgrid);
vanscancol=round(ondergrens/gridsizex+1);
totscancol=round(bovengrens/gridsizex+1);
scanpointsx=X(1,vanscancol:totscancol);
scanpointsy=Z(rij,vanscancol:totscancol);
grscptx=size(scanpointsx);
afst=zeros(grscptx(1,2),1);

for q=1:hgrid
for r=1:lgrid
for t=1:grscptx(1,2)
afst(t,1)=sqrt((hcoor(1,r)-scanpointsx(1,t))^2+(vcoor(q,1)-
scanpointsy(1,t))^2);
end
179
gem=mean(afst);
vgem=gem*ones(grscptx(1,2),1);
afw=afst-vgem;
Mstr(q,r)=gem;
Mafw(q,r)=sum(abs(afw));
end
end

%find minimum value in matrix Mr
[minpercol,rijentry]=min(Mafw);
[mintot,colentry]=min(minpercol);
entryrijcol=[rijentry(1,colentry) colentry];

%find calculated optimum radius and midpoint coordinates
ropt=Mstr(entryrijcol(1,1),entryrijcol(1,2));
afwijkingvangem=Mafw(entryrijcol(1,1),entryrijcol(1,2));
xopt=hcoor(1,entryrijcol(1,2));
yopt=vcoor(entryrijcol(1,1),1);

%draw optimum midpoint
scatter(xopt,yopt,'or');

%draw circle
for tel=1:1000
cirkelx(tel)=ropt*cos(tel/100)+xopt;
cirkely(tel)=ropt*sin(tel/100)+yopt;
end
plot(cirkelx, cirkely);
clear cirkelx cirkely grscptx afst afw vgem Mstr Mafw minpercol rijentry
colentry afwijkingvangem;
clear ondergrens bovengrens apphmid hoogtegrid breedtegrid gridstap lgrid
hgrid hcoor vcoor;
clear k l vectorgridplothor vectorgridplotver mintot entryrijcol
afwijkingvangem tel;

Determiningtheweldtoeangle
%determine weld toe angle

vanmoe=44; %start of straight parent material, use steps of size
[gridsizex]
totmoe=45; %end of straight parent material, use steps of size [gridsizex]
vanlas=50; %start of straight weld material, use steps of size [gridsizex]
totlas=52; %end of straight weld material, use steps of size [gridsizex]

vanhoekmoecol=round(vanmoe/gridsizex+1);
tothoekmoecol=round(totmoe/gridsizex+1);
vanhoeklascol=round(vanlas/gridsizex+1);
tothoeklascol=round(totlas/gridsizex+1);

Xmoepoints=X(1,vanhoekmoecol:tothoekmoecol);
Zmoepoints=Z(rij,vanhoekmoecol:tothoekmoecol);
Xlaspoints=X(1,vanhoeklascol:tothoeklascol);
Zlaspoints=Z(rij,vanhoeklascol:tothoeklascol);

moe=polyfit(Xmoepoints,Zmoepoints,1);
las=polyfit(Xlaspoints,Zlaspoints,1);

%plot of both fitted lines
180
xmoe=vanmoe:gridsizex:totmoe;
zmoe=moe(1)*xmoe+moe(2);
plot(xmoe,zmoe)
hold on;
xlas=vanlas:gridsizex:totlas;
zlas=las(1)*xlas+las(2);
plot(xlas,zlas);

hoekmoe=atand(moe(1)); %angle of parent material with horizontal
hoeklas=atand(las(1)); %angle of weld material with horizontal
hoekverschil=abs(hoeklas-hoekmoe); %angle between parent material and
weld
clear vanhoekmoecol tothoekmoecol vanhoeklascol tothoeklascol Xmoepoints
Xlaspoints Zlaspoints moe las xmoe zmoe xlas zlas hoekmoe hoeklas;

Determiningtheweldheight
%determine undercut of weld in relation to parent material

left=3.; %left location of parent material, use steps of size [gridsizex]
right=50; %right location parent material, use steps of size [gridsizex]
vanh=5; %start of area under investigation for weld height use septs of
size [gridsizex]
toth=52; %end of area under investigation for weld height use steps of size
[gridsizex]

%construct refenence line
leftcol=round(left/gridsizex+1);
rightcol=round(right/gridsizex+1);
leftz=Z(rij,leftcol);
rightz=Z(rij,rightcol);
xline=[left right];
zline=[leftz rightz];
parent=polyfit(xline,zline,1);
plot(xline,zline);

%determine distance of weld to reference line
vanhcol=round(vanh/gridsizex+1);
tothcol=round(toth/gridsizex+1);
scanpointsx=X(1,vanhcol:tothcol);
scanpointsz=Z(rij,vanhcol:tothcol);
Dist=zeros(1,grscptx(1,2));
for dtel=1:grscptx(1,2);
Dist(1,dtel)=scanpointsz(1,dtel)-
(parent(1,1)*scanpointsx(1,dtel)+parent(1,2));
end
weldheight=max(Dist);
clear left right vanh toth leftcol rightcol leftz rightz xline zline
parent;
clear scanpointsx scanpointsz vanhcol tothcol grscptx Dist;

Determiningtheundercut

%determine undercut of weld in relation to parent material
vanmoe=48; %start of straight parent material, use steps of size
[gridsizex]
totmoe=49; %end of straight parent material, use steps of size [gridsizex]
181
vancut=47; %start of area under investigation for undercut use steps of
size [gridsizex]
totcut=51; %end of area under investigation for undercut use steps of size
[gridsizex]

vancutmoecol=round(vanmoe/gridsizex+1);
totcutmoecol=round(totmoe/gridsizex+1);
vancutweldcol=round(vancut/gridsizex+1);
totcutweldcol=round(totcut/gridsizex+1);

Xmoepoints=X(1,vancutmoecol:totcutmoecol);
Zmoepoints=Z(rij,vancutmoecol:totcutmoecol);
moe=polyfit(Xmoepoints,Zmoepoints,1);

scanpointsx=X(1,vancutweldcol:totcutweldcol);
scanpointsz=Z(rij,vancutweldcol:totcutweldcol);

reflijnx=X(1,vancutweldcol:totcutweldcol);
for reftel=1:grscptx(1,2)
reflijnz(1,reftel)=moe(1)*scanpointsx(1,reftel)+moe(2);
end
plot(scanpointsx,reflijnz);

Dist=zeros(1,grscptx(1,2));
for dtel=1:grscptx(1,2)
Dist(1,dtel)=reflijnz(1,dtel)-scanpointsz(1,dtel);
end
undercut=max(Dist);
clear vanmoe tot moe vancut totcut vancutmoecol totcutmoecol vancutweldcol;
clear totcutweldcol Xmoepoints Zmoepoints moe scanpointsx grscptx;
clear scanpointsz reflijnx reftel reflijnz Dist dtel;
182

183
AnnexG:CollaborationwithTNO:Acousticemission
In this research the location of cracks and crack initiation time have been determined with the aid of strain gauges.
Anotherpossibilityistheuseoftheacousticemissionofthesteelwhenacrackisgrowing.Todetectthesesoundemissions
ofthesteel,sensorshavebeenappliedonanumberofspecimens.Toreducebackgroundnoisefromthehydraulicsofthe
testingrig,plasticinedampershavebeenapplied.TheresultingtestsetupcanbeseeninfigureG.1.
figureG.1Left:testsetupwithsensorsanddampersappliedtothespecimen.Right:closeupofthreesensors
withoutanycablesattachedandtwodampersmadeofplasticine,pretensionedwithbicycletires.
Due to the nature of the testing rig a very high background noise level was measured. This can be damped by the
dampers as depicted in figure G.1 or by placing insulating material between the clamps of the test rig. Because the
considered rig is an axial test rig which uses clamping to fix the specimens, the second option for background noise
reductionwasnotpossible.Thedampersareabletoreducethebackgroundnoisetoacertainextent,butnotalwaystothe
desiredlevel,dependingonthethicknessofthespecimenandthemagnitudeoftheappliedloadandfrequency.
The sensors areable to locate a crack via triangulation of the signal at differentsensors. The software then produces a
summationofallhitswhichresultsinanimageascanbeseeninfigureG.2.Thislocationofthecrackcanonlybedonein
a rather coarse manner because of the relatively large size of the sensors in relation to the weld. In practice, larger
structureswillbeobservedandsmallersensorsareavailable.
184
figureG.2Increaseofhitsduringthetestofaspecimenfromthefirsthit(topleft)tothefinalcrack(bottomright)
Thegreensymbolsrepresentthesensorsonbothsidesontheweld,thereddotsarehits.
Mostresultsoftheexperimentareoflimitedimportanceforthecurrentresearch.Fortheseresultsreferenceismadeto
TNO publications on the matter. It is however very interesting to compare the crack initiation time as defined in this
researchandthecrackinitiationtimeasfoundwiththeacousticemission.Inthisway,thecrackinitiationtimeasdefined
bythemethodofdeviatingstraingaugescanbeevaluated.Unfortunatelytheseresultsarenotavailableatthepublishing
dateofthisreport.
If additional research is done on the matter, it would be very insightful to obtain the reports of TNO which contain
severalpracticaltips.Furthertestingshould,whenpossible,bedoneinsuchatestingrig,thatdampingofthebackground
noiseispossiblebyplacingisolationmaterialbetweentherigandthespecimen(i.e.abendingtest).
185
AnnexH:HardnessMeasurements
Fromeachplatematerialonestripofresidualmaterialhasbeenpolished.Pijpers(2011)hasshownonsimilarspecimens
from the same material that the hardness does not vary significantly over the height of the specimen, but only shows
differences between the different material states (fusion zone FZ, heat affected zone HAZ and base material BM). From
eachspecimentheasweldedsituationisplotted.AllmeasurementsareexecutedafterTIGdressing,butonlyinthemiddle
ofthespecimen,whichisassumedtobeoutoftheinfluenceofTIGdressing.
Then, the comparison with the effect of TIG dressing is made. The different zones that occur after the initial welding
process(FZ,HAZandBM)caneachseparatelybeinfluencedbytheTIGdressing,bybeingremeltedorheataffected.Inthe
diagramsthedifferentzonesthatoccurafterinitialweldingarestatedastheasweldedsituation,andtheeffectsofTIG
dressing (melting of heat affecting) on these original zones are plotted for comparison. For example: figure H.4 shows
datapoints which were in the BM zone after welding. Some areas of the BM zone have been influenced by TIG dressing,
denotedbyTIGFluidZoneandTIGHeatAffectedZone.Thesepointsareplottedinwhiteandgrey.Becauseofthecomplex
geometryofthedifferentzonesafterTIGdressing,theXaxisinthesecomparisondiagramshasnomeaning,thediagramis
simplyacomparisonofdifferentmeasuredvalues.
AnexampleoftheasweldedstateofaspecimenisshowninfigureH.1.TheeffectofTIGdressingisshowninfigureH.2.
NotethatinfigureH.2thefluidzoneofTIGdressingatthecapsideoftheweldextendsintotheasweldedFZ,HAZandBM.
Thisisinrealitynotalwaysthecase.Similarremarkscanbemadeattheotherfluidzonesandheataffectedzones.
FZ HAZ HAZ BM BM
figureH.1Zonesintheasweldedspecimens
figureH.2EffectofTIGdressing
186
C11
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
Distancefrommiddle[mm]
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
figureH.3C11:Aswelded.Castmaterialontheleft,rolledmaterialontherightsideoftheweld
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.4C11:BasematerialbeforeandafterTIGdressing[Cast]
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.5C11:BasematerialbeforeandafterTIGdressing[Rolled]
187
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.6C11:HeataffectedzonebeforeandafterTIGdressing[Cast]
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.7C11:HeataffectedzonebeforeandafterTIGdressing[Rolled]
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.8C11:FluidzonebeforeandafterTIGdressing
188
C89
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
figureH.9C89:Aswelded.Castmaterialontheleft,rolledmaterialontherightsideoftheweld
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
As Welded
TIG Fluid Zone
TIG Heat Affected Zone
189
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
190
C89a
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
HV10
BaseMaterial
WeldMaterial
HeatAffectedZone
figureH.15C89a:Aswelded.Castmaterialontheright,rolledmaterialontheleftsideoftheweld
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.16C89a:BasematerialbeforeandafterTIGdressing[Cast]
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.17C89a:BasematerialbeforeandafterTIGdressing[Rolled]
191
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.18C89a:HeataffectedzonebeforeandafterTIGdressing[Cast]
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone[nodata]
TIGHeatAffectedZone
figureH.19C89a:HeataffectedzonebeforeandafterTIGdressing[Rolled]
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.20C89a:FluidzonebeforeandafterTIGdressing
192
C69
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
HV10
Base Material
Heat Affected Zone
Weld Material
figureH.21C69:Aswelded.Castmaterialontheright,rolledmaterialontheleftsideoftheweld
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
193
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
194
C46
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
figureH.27C46:AsweldedCastmaterialontheright,rolledmaterialontheleftsideoftheweld
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
195
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
196
V11
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
30 20 10 0 10 20 30 40
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
figureH.33V11:Aswelded
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone[nodata]
TIGHeatAffectedZone
figureH.34V11:BasematerialbeforeandafterTIGdressing
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.35V11:HeataffectedzonebeforeandafterTIGdressing
197
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.36V11:FluidzonebeforeandafterTIGdressing
198
V89
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
199
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH40V89:FluidzonebeforeandafterTIGdressing
200
V69
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
-40 -30 -20 -10 0 10 20 30 40
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
figureH.42V69:BasematerialbeforeandafterTIG0dressing
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
201
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
202
V46
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
40 30 20 10 0 10 20 30 40
HV10
BaseMaterial
HeatAffectedZone
WeldMaterial
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
203
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
550.0
HV10
AsWelded
TIGFluidZone
TIGHeatAffectedZone
204
205
AnnexI:Crackmonitoringresults
Inthefollowing diagramstheresultsofthecrackmonitoringareplotted.Allvaluesaregiveninthecrackdimensionsa
andc.Foredgecracksthisisequaltothelengthanddiameterofthecrack,forsurfacecrackstheactualcrackwidthis2c
Only the specimens which have shown usable results are plotted. Note that the crack depth a measured from surface
measurements can differ from the crack marking measurements because the depth of the crack at the side is not
necessarily the same as the maximum depth anywhere in the cross section. In case of a rerun specimen, only the second
testisconsidered.
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
1.00E+06 1.02E+06 1.04E+06 1.06E+06 1.08E+06 1.10E+06
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
figureI.1CrackpropagationdataofspecimenC112
0.0
2.0
4.0
6.0
8.0
10.0
6.00E+04 7.00E+04 8.00E+04 9.00E+04 1.00E+05 1.10E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
206
0.0
10.0
20.0
30.0
40.0
50.0
60.0
7.00E+04 8.00E+04 9.00E+04 1.00E+05 1.10E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
figureI.3CrackpropagationdataofspecimenC89a1
0.0
10.0
20.0
30.0
40.0
1.00E+05 1.10E+05 1.20E+05 1.30E+05 1.40E+05 1.50E+05 1.60E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
5.40E+05 5.80E+05 6.20E+05 6.60E+05 7.00E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
figureI5CrackpropagationdataofspecimenC692
207
0.0
5.0
10.0
15.0
20.0
25.0
30.0
3.10E+06 3.12E+06 3.14E+06 3.16E+06 3.18E+06 3.20E+06
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
3.20E+05 3.30E+05 3.40E+05 3.50E+05 3.60E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
6.00E+04 8.00E+04 1.00E+05 1.20E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
208
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
2.10E+05 2.20E+05 2.30E+05 2.40E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
figureI9CrackpropagationdataofspecimenV111
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
1.00E+05 1.10E+05 1.20E+05 1.30E+05 1.40E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
6.35E+05 6.40E+05 6.45E+05 6.50E+05 6.55E+05 6.60E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
209
0.0
5.0
10.0
15.0
20.0
25.0
30.0
2.80E+05 3.00E+05 3.20E+05 3.40E+05 3.60E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
3.60E+05 3.80E+05 4.00E+05 4.20E+05 4.40E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
7.00E+04 9.00E+04 1.10E+05 1.30E+05
N
a/c [mm]
Crackwidthc
Crackdeptha
Crackwidthc[CM]
Crackdeptha[CM]
210
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
3.00E+05 3.20E+05 3.40E+05 3.60E+05 3.80E+05
N
a/ c [ mm]
Crack width c
Crack depth a
Crack width c [CM]
Crack depth a [CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
4.00E+05 4.20E+05 4.40E+05 4.60E+05 4.80E+05 5.00E+05
N
a/ c [ mm]
Crack width c
Crack depth a
Crack width c [CM]
Crack depth a [CM]
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
6.50E+05 6.70E+05 6.90E+05 7.10E+05 7.30E+05
N
a/ c [ mm]
Crack width c
Crack depth a
Crack width c [CM]
Crack depth a [CM]
211
AnnexJ:DianaModel
Themodelthatwasusedneedstobeeasilyadjustabletocalculatethestressconcentrationfactorfordifferentweldtoe
geometries. To achieve this, an excel sheet was designed which generates a list of TNO Diana commands, based on the
giveninputparameters.Thesheetoffersthepossibilitytoalterthegeometry,butalsothemeshingdivisionofthedifferent
contourlines.Becauseofthelargechangesingeometry,themeshingdivisionneedstobechangedinsuchawaythatthe
automaticmeshinggeneratorinTNODianacanfindasuitablemesh.
Inthisannex,onlytwogeometrieswillbecovered,withthefollowingparameters:
Radius: 2mm
Weldtoeangle: 15.3degrees
Weldheight: 3mm
Length: 100mm
Thickness: 25mm
Undercut: 0mmor0.1mm
Any different geometries are simply an alteration of the described procedure. The difference between the two
consideredgeometriesistheoccurrenceofanundercut,whichslightlyaltersthemodelbuildup.Themodelisconstrained
inhorizontaldirectionandallrotationsontherightside.Ahorizontallineloadisappliedontheleftsideofthemodel.

figureJ.1Left:overviewofthecontourlinesandmeshingaidlinesofthemodel.Right:detailofweldtoewith
undercut(top)andwithoutundercut(bottom)
Inbetweenthemeshingaidlines(asseeninthedetailsinfigureJ.1)andintheleftrectangularsectiontowhichtheload
is applied a mapped mesh is used. This results in a mesh similar to the one depicted in figure 8.1. Finally, the Diana
commandstoconstructthemeshoffigureJ.1withundercutareshowbelow.
FEMGENAnnex_UC
PROPERTYFEPROGDIANASTRUCT_PE
yes;
UTILITYSETUPUNITSLENGTHMILLIMETER
UTILITYSETUPUNITSMASSKILOGRAM
UTILITYSETUPUNITSFORCENEWTON
UTILITYSETUPUNITSTIMESECOND
UTILITYSETUPUNITSTEMPERATURECELSIUS
CONSTRUCTSPACETOLERANCEABSOLUTE0.0005
MESHINGDIVISIONDEFAULT2
!definebasicweldshape
GEOMETRYPOINTP1000
GEOMETRYPOINTP210000
GEOMETRYLINESTRAIGHTL1P1P2
GEOMETRYPOINTP30250
GEOMETRYPOINTH100280
212
GEOMETRYLINESTRAIGHTL5P5H
GEOMETRYLINEANGLECL6L590HXY100
GEOMETRYLINEANGLECL7L315.3P4XY40
GEOMETRYPOINTINTERSECTP8CL6CL7
UTILITYDELETELINESCL6CL7
YES;
GEOMETRYLINESTRAIGHTL6HP8
!defineweldrootnotch
GEOMETRYLINEPARALLELCL1L7P62
GEOMETRYPOINTINTERSECTCP1CL1L3
GEOMETRYLINEANGLECL2CL10CP1xy7.20043217846358
GEOMETRYLINECIRCLEP112
GEOMETRYLINEPERPENDICCL3L7P11P1510
GEOMETRYPOINTINTERSECTP20CL3L11
UTILITYDELETELINESL7
YES;
GEOMETRYPOINTINTERSECTP21L3L10
GEOMETRYLINEARCL14P21P15P11
GEOMETRYLINEARCL15P15P20P11
UTILITYDELETELINESCL3CL1L9L8CL2L3L10L11
YES;
UTILITYDELETEPOINTSP16P13P9P10P6P7CP1P14P12P4
YES;
!renaminginorderlyfashion
UTILITYNAMEL13L3
UTILITYNAMEL4L9
UTILITYNAMEL5L8
UTILITYNAMEL6L7
UTILITYNAMEL12L6
UTILITYNAMEL15L5
UTILITYNAMEL14L4
UTILITYNAMEP2P9
UTILITYNAMEP3P2
UTILITYNAMEP21P3
UTILITYNAMEP15P4
UTILITYNAMEP8P6
UTILITYNAMEP5P8
UTILITYNAMEP20P5
UTILITYNAMEHP7
UTILITYNAMEP11PCENTER
!createmeshingaidlines
GEOMETRYLINECIRCLEPCENTER5
GEOMETRYLINEPERPENDICCL1L6PCENTERP2020
UTILITYNAMEP20P11
UTILITYDELETELINESL16L17L18L19
YES
GEOMETRYLINEARCL11P11P12PCENTER
GEOMETRYLINEANGLECL2L60P12XY40
UTILITYDELETELINESCL1CL2
YES
UTILITYDELETEPOINTSP17P18P19P21P22
YES
213
GEOMETRYLINEANGLEL10L1180P11XY10
GEOMETRYLINEPERPENDICCL1L10P23PCENTER10
GEOMETRYPOINTINTERSECTP10L3CL1
UTILITYDELETELINESCL1
YES
UTILITYDELETEPOINTSP24
YES
GEOMETRYLINESTRAIGHTl15P5P12
UTILITYDELETELINESL3L7
YES
YES
YES
!createsurface
GEOMETRYSURFACE4SIDESSURF1L1+L9L8+L7+L12+L11+L10+L13
L3L19
GEOMETRYSURFACE4SIDESSURF2L10L14+L4L16L13
GEOMETRYSURFACE4SIDESSURF3L11L15L5L14
!boundaryconditions
PROPERTYBOUNDARYCONSTRAINTL81456
PROPERTYBOUNDARYCONSTRAINTL91456
!loading
PROPERTYLOADSPRESSURELO11L2100X
!materials
PROPERTYMATERIALS1100ELASTICISOTROP2100000.3
PROPERTYATTACHALLMATERIALS1100
!properties
PROPERTYPHYSICALTH1GEOMETRYTHINFSHL1
PROPERTYATTACHALLTH1
!meshing
CONSTRUCTSETSURFSAPPENDSURFACESALL
MESHINGTYPESSURF1CQ16E
214
MESHINGDIVISIONLINEL190
MESHINGOPTIONSALGORITHMPAVINGSURF1
MESHINGOPTIONSALGORITHMMAPPEDSURF2
MESHINGGENERATEALL
EYEFRAME

High Strength Steel Fatigue

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

High Strength Steel Fatigue

Uploaded by

Copyright:

Available Formats

Effect of TIGdressing on fatigue strength and weld toe geometry of butt

log log log N a m o = A (1.2)

; notch notch per

Material Loadcarryingfilletwelds Nonloadcarryingfilletweldsandbuttwelds

FAT Weldquality Relativeimprovement

figure8.6Stresspatterninhorizontaldirectionpresentedascontourlevels.=2, =20, no undercut

Aswelded TIGdressed determinedby

Weldtoeangle[] 15.3 20.2 1.1

For the determination of K

Aswelded TIGdressed TIGdressed(=1)

C1100 2.16 2.79 1.47 1.78 1.79 2.35

Aswelded TIGdressed TIGdressed(=1)

C1100 2.10 2.61 1.47 1.76 1.75 2.22

The mean stress of all specimens will be adjusted to a value of

Theapplicablecorrectionfactor forthemisalignedaxiallyloadedspecimenscanbecalculated withtheuseof equation

95% 50% 0.95;n f

cycles are depicted for the as welded

cycles are depicted for the as welded

Gas flow rate: /vT Z f /r*.t,\>

%define optimalisation boundaries

%determine weld toe angle

%determine undercut of weld in relation to parent material

You might also like