Content uploaded by N.T. Switzner
Author content
All content in this area was uploaded by N.T. Switzner on Mar 13, 2021
Content may be subject to copyright.
Assessingsteelpipetoughness
usingchemicalcompositionandmicrostructure
Paper#57
NathanSwitzner,JoelAnderson,MichaelRosenfeld,
RSI‐PS
JonathanGibbs,PeterVeloo,RamonGonzalez
PacificGasandElectricCompany
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
Toughness
•Toughnessisthe
abilityofthematerial
towithstandcrack
growth.
•In‐servicemetal
productsexhibit
cracksorcrack‐like
defectsofavarietyof
shapesandsizes. https://www.rosen‐group.com/global/solutions/services/pipeline‐cracks
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
PipeToughness
•Toughnesscanbean
importantaspectofmaximum
allowableoperatingpressure
(MAOP)reconfirmationbythe
engineeringcritical
assessment(ECA)process.
•Toughnessisusedindefect
analysisandremaininglife
assessments.
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ToughnessfromaSteelmakingPerspective
•Tomeettoughness
targets,steel
manufacturers
optimizechemical
compositionand
microstructurealong
withotherprocessing
parameters.
https://www.steel‐grips.com/10‐news/286‐successful‐remote‐
commissioning‐of‐level‐2‐automation‐for‐continuous‐caster‐cc21
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
Motivation:ANondestructiveToughnessTest
•Thereisnowidelyacceptednon‐destructivemethodforestimating
toughness.
•Operatorsmaycutoutandtestpipeorhottapcoupons.
•Astatisticalmetallurgicalmodeltoestimatetoughnessbasedonin‐
situchemicalanalysisandmicrostructuredatawouldprobablyfall
inthe"othertechnology"category,whichrequiresnotification.
https://epcmholdings.com/ho
t‐tapping‐advice‐on‐
successful‐pipeline‐tapping/
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
DataSourcesforToughness,Composition,and
Microstructure
•Standarddestructivelaboratorymethodswereusedtodeterminethe
compositionforeachofthepipefeaturesinthisstudy.
•Calculationsofuppershelfenergyandtransitiontemperatureforafull‐size
CharpyV‐notchsample.
“Vendor1” ~1000pipesand
features
CharpyV‐notch
ToughnessData
Laboratory
CompositionData
“Vendor2” ~120pipesand
features
CharpyV‐notch
ToughnessData
Laboratory
CompositionData
PG&E ~41pipesand
features
CharpyV‐notch
Toughnessdata
Laboratory
CompositionData
Microstructural
Data
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
KeyMeasurementsfromtheCVNTest
•Sincefracturetoughnessisstrongly
controlledbytemperaturefor
steel,CVNtestsareperformedat
multipletemperatures.
•Recordeddatainclude:
•Test temperature
•Energyabsorbed
•Percentshear(ductile)
appearance
Rosenfeld, M., “Procedure improves line pipe Charpy test
interpretation,” Oil and Gas Jrnl, 1997, Apr, pp 40-46.
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
AlternativeCalculationMethodforUpper
ShelfEnergyandTransitionTemperature
•Based onwork byRosenfeld,API579Section9F.2.2.c
provides analternativemethod to calculate:
•UppershelfenergyandTransitiontemperature
•Thecalculationmethodrequires:
•Pipewallthickness
•Test temperature
•Samplesize(forsubsizesamples)
•Energyabsorbed
•Percentsheararea
Ref:Rosenfeld,M.,“ProcedureimproveslinepipeCharpytestinterpretation,”OilandGasJrnl,1997,Apr,pp40‐46.
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ElementEffectsonCVNUpperShelfEnergy
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ElementEffectsComparison
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ElementEffectsComparison
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ElementEffectsComparison
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ModelTypesandIterationstoPredictUpper
ShelfEnergyandTransitionTemperature
•PredictUpperShelfEnergy
•Basedon10compositional
elements(train/test):
•Linearmodel
•Randomforest
•Basedon4compositional
elements(train/test):
•Linearmodel
•Randomforest
•PredictTransitionTemp.
•Randomforestmodels:based
on9elementsandthen4
elements(train/test)
•PredictUpperShelfEnergyand
TransitionTemperature:
•Linearmodelusingboth
compositionand
microstructure(notrain/test)
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
RandomForestModelforUpperShelfEnergy
•Arandomforestmodelisadecisiontree‐basedmodel
thatisiteratedhundredsorthousandsoftimeson
randomsubsetsofthedata.
•Resultsofallthe“trees”areaveragedtoproducea
singlemodel.
•Minimizestheinfluenceofasmallnumberofoutliers.
•Thedataweresplitbetweentrainingandtestsets,with
80%and20%ofthedata,respectively.
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
R2value
rf_a =0.84
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
R2value
rf_a =0.84
rf_b =0.80
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
RandomForestModelforTransition
Temperature
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
R2value
rf_a =0.51
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
MicrostructureEffects
•Microstructurereferstothegrain
structureandfeaturesinapolished
andetchedsampleusuallyonly
visiblethroughamicroscope.
•Thegrainsizeandtheamountof
darkphase(pearlite=ferrite+iron
carbide)aretwoquantifiable
aspectsknowntoaffecttoughness.
light
phase
grain
dark
phase
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
R2=0.76
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
ConcludingRemarks
•Toughnessisanimportant
materialpropertyfor
inhibitingcrackgrowth.
•Therearenowidely
recognizedNDEmethodsto
estimatetoughness.
•Uppershelfenergy(USE)and
transitiontemperature(TT)
arekeypropertiesforpipeline
remaininglifecalculations.
•USEandTTcanbecalculated
usingempiricalequations
basedondestructiveCVN
data.
•DestructiveCVN,composition
andmicrostructuredatawere
usedtobuildlinearand
randomforestmodelsto
estimateUSEandTT.
NathanSwitzner
Assessingsteelpipetoughnessusingchemicalcompositionandmicrostructure
FutureWork
•Focusonlowertoughness
andhighertransition
temperaturesteels(atrisk
steels)
•Examinethecompositionand
microstructureofoutliers
•Expandthecompositionand
microstructuredatabases
•BeginvalidationusingNDE
compositionand
microstructuredata
•Sensitivityanalysis
FormoreinformationcontactNathan
Switzner:nswitzner@rsi‐ps.com