A study on the sealing performance of bolted flange joints with gaskets

内容需要下载文档才能查看 内容需要下载文档才能查看 内容需要下载文档才能查看

InternationalJournalofPressureVesselsandPiping84(2007)349–357

http://wendang.chazidian.com/locate/ijpvp

Astudyonthesealingperformanceofbolted?angejoints

withgasketsusing?niteelementanalysis

M.MuraliKrishna,M.S.Shunmugam,N.SivaPrasadÃ

DepartmentofMechanicalEngineering,IndianInstituteofTechnologyMadras,Chennai600036,IndiaReceived19January2006;receivedinrevisedform21January2007;accepted1February2007

Abstract

Gasketsplayanimportantroleinthesealingperformanceofbolted?angejoints,andtheirbehaviouriscomplexduetononlinearmaterialpropertiescombinedwithpermanentdeformation.Thevariationofcontactstressesduetotherotationofthe?angeandthematerialpropertiesofthegasketplayimportantrolesinachievingaleakproofjoint.Inthispaper,athree-dimensional?niteelementanalysis(FEA)ofbolted?angejointshasbeencarriedoutbytakingexperimentallyobtainedloadingandunloadingcharacteristicsofthegaskets.Analysisshowsthatthedistributionofcontactstresshasamoredominanteffectonsealingperformancethanthelimiton?angerotationspeci?edbyASME.

r2007ElsevierLtd.Allrightsreserved.

Keywords:Bolted?angejoints;Gasketcharacterization;Gasketcontactstress;Flangerotation;Axialboltforce

1.Introduction

Flangedjointswithgasketsareverycommoninpressurevesselandpipingsystems,andaredesignedmainlyforinternalpressure.Thesejointsarealsousedinspecialapplicationssuchasinnuclearreactorsandspacevehicles.Theconnectionofafuelducttoarocketengineisatypicalapplicationofthesejointsinspacevehicles.Preventionof?uidleakageistheprimerequirementof?angedjoints.Manydesignvariablesaffectjointperformanceanditisdif?culttopredictthebehaviourofjointsinservice.Anumberofdesigncodesandstandards,whichareprincipallybasedontheTaylor–Forgemethod[1],provideproceduresforthedesignof?angedjoints.EvenjointsdesignedwithcodessuchasASME,DIN,JISandBSexperienceleakageandthisproblemiscontinuouslyfacedbyindustry.Allthesecodesarebasedonmanysimpli?ca-tionsandassumptionsandhencemaynotpredicttherealbehaviourof?angedjointswithgaskets.

Thecomplexitiesassociatedwiththeanalysisofbolted?angejointswithgasketsareduetothenonlinear

Correspondingauthor.Tel.:+914422574679;fax:+914422574652.

E-mailaddress:siva@iitm.ac.in(N.SivaPrasad).

0308-0161/$-seefrontmatterr2007ElsevierLtd.Allrightsreserved.doi:10.1016/j.ijpvp.2007.02.001

behaviourofthegasketmaterialcombinedwithpermanentdeformation.Thematerialundergoespermanentdeforma-tionunderexcessivestresses.Thedegreeofelasticity(stiffness)isafunctionofthecompressivestresses,whichactonthegasketduringassemblyandafteritisputintoservice.Itiscommonlyrecognizedthatgasketstiffnesshasapredominanteffectonthebehaviourofthejointbecauseofitsrelativelylowstiffness.

Anotherinherentproblemwithboltedjointsis?angerotationandcontactstresses.Thesearecausedbytheboltpre-loadandincreasewhenthejointissubjectedtointernalpressure.TheASMEcodehasmadeanattempttocorrectthisproblembyaddingarigidityconstraint‘J’basedonthe?xedrotation.Thismaynotbeadequate,astherotationofthe?angeisnotauniquevalue.Flangerotationcausesvariablecompressionacrossthegasketfromtheinnerradiustotheouterradius.Duetothevariationincompression,thecontactstressesalsovaryalongtheradius.

Sawaetal.[2]presentedamathematicalmodelfordeterminingthecontactstressdistributioninapipe?angeconnectionbasedonthetheoryofelasticity,treatingitasanaxisymmetricproblem.Thesealingperformance,effectivegasketwidthandthemomentsactingatthejoint

350

M.MuraliKrishnaetal./InternationalJournalofPressureVesselsandPiping84(2007)349–357

werediscussed.Sawaetal.[3]extendedtheirearlierinvestigationbyanumericalmethodconsideringthestress–straincurveofthegasketaspiecewiselinear.BouzidandDerenne[4]developedananalyticalmethodconsider-ingtherotational?exibilityofthe?angefordeterminingthecontactstressesinordertopredictthejointtightness.Intheseinvestigations,neithertheactualnonlinearitynorhysteresisofthegasketmaterialwastakenintoconsideration.

Inthepresentwork,a?niteelement(FE)modelfor?ndingthecontactstressesinagaskethasbeendeveloped.Nonlinearityandhysteresisofthegasketundervariousloadingandoperatingconditionsaretakenintoaccount.

Experimentshavebeencarriedoutfor?ndingtheloadingandunloadingcharacteristicsofthegasketmaterials,whichareinturnusedintheFEA.Thein?uenceof?angerotationsonthesealingperformanceofdifferentgasketmodelswithvaryingloadingandoperatingconditionshasbeenstudied.Theincreaseintheaxialboltforcewhenthejointissubjectedtoaninternalpressurehasalsobeenanalyzed.Thedistributionofcontactstressonthegasketfordifferentloadingconditionshasbeenstudied.Inthepresentanalysisweld-neck(WN)raised-face(80mmNPS,Class-600,ASME/ANSIB16.5)?angeswithselectedspiral-woundringgasketsareconsideredwithmaterialpropertiesatroomtemperature.

2.Gasketedjointcon?gurationandmaterialproperties2.1.Geometryofthe?ange,gasketandbolt

Allstandard?angesareprovidedwithavarietyofmachinedfaces.Raised-face?angesarepreferableoverfull-face?angesformediumserviceconditions,andringgasketswithsmallcontactareareducetheboltpre-loadrequiredtocompressagasket.Figs.1(a)and(b)showthedimensionsofthepipe?ange(80mmNPS,weldneck,Class600,ASME/ANSIB16.5)[5]andthespiralwoundgasketusedinthe?niteelementanalysis(FEA).M20sizeboltsareconsideredintheanalysis.

2.2.Materialpropertiesofthe?angeandthebolt

The?angeandtheboltmaterialpropertiesareassumedtobehomogenous,isotropicandlinearlyelastic.Materialsfor?angesandboltsarechosenasforgedcarbonsteel(A105,Young’smodulus,E¼195GPa,Poisson’sratio,n¼0.3)andchromiumsteel(A193-B7,E¼203GPa,n¼0.3).

2.3.Gasketcharacterization

Gasketsareoftenmultilayeredmaterials,exhibitingnonlinearbehaviourinloadingandunloadingconditions.Themodulusofelasticityforthebolting-uporcompres-sion(loading)stageisdifferentfromthedecompression(unloading)stageofthegasketduetotheinternalpressure.

6.05

23 4.6 5.4Fig.1.Dimensionsofthe?angeandgasketusedintheFEA:(a)?angeand(b)spiral-woundgasket.

Whenthegasketisdecompressed,itshowsstronghysteresiswhichisnonlinearandleadstopermanentdeformationusuallycon?nedtothrough-thickness.Thecontributiontothestiffnessfrommembrane(inplane)andtransversesheararemuchsmallerandhenceneglected.AsinputdataforFEA,eachreloadingcurveisassumedtobeidenticalwiththeunloadingoneforsimplicity.

TheFEprogramANSYSoffersanumberofelementstomodelgaskets.Theseelementsconsidergeometricandmaterialnonlinearitiesandmembraneandtransverseshearareneglected.Thusthepressure–versus–closurebehaviourcanbedirectlyappliedtocharacterizethegasketmaterial.Aloadcompressivemechanicaltest(LCMT)[6,7]hasbeencarriedoutfor?ndingthemechanicalcharacteristicsofgasketmaterialwhichareinturnusedintheFEA.ANSYSsoftwarehasaprovisiontoinputtheLCMTdatapoints.Fig.2showsthematerialpropertiesfordifferentspiral-woundgaskets.Thenonlinearcompressionbehaviourwithtwodifferentunloadingcurvesforspiral-woundgasketsisconsideredtopredictthenonlinearbehaviourofthegasketintheanalysis.

3.Finiteelementmodelling3.1.Discretization

Athree-dimensionalFEmodelhasbeendevelopedforbolted?angeconnectionswithgasketsusingANSYS[8].Thesejointspossessgeometriccharacteristicswhicharesymmetricalaboutanaxis.Theycanbede?nedintermsof

内容需要下载文档才能查看

M.MuraliKrishnaetal./InternationalJournalofPressureVesselsandPiping84(2007)349–357

351

10080Pressure, MP

a

60402000.0

0.2

0.4

0.60.81.0Closure, mm10080Pressure, MPa

60402000.0

0.2

0.4

0.60.81.0Closure,mm

1.2

1.4

1.2

1.4

10080Pressure, MP

a

60402000.0

0.2

0.40.60.8Closure,mm

1.0

Fig.2.Characteristicsofdifferenttypesofspiral-woundgasketsobtainedexperimentally:(a)asbestos?lled(b)graphite?lledand(c)PTFE

?lled.

Fig.3.Finiteelementmeshofbolted?angejointwithspiral-woundgasketwitheightbolts.

gasketandeightbolts(1/4model).Solidelements(SOLID185)[9]areusedtomodelthegeometryofthe?ange.

3.2.Modelingofthegasket

Thegasketismodeledwithinterfaceelements(INTER195)[9].Theseelementsarebasedontherelativedeformationofthetopandbottomsurfacesandofferadirectmeanstoquantifythethrough-thicknessdeforma-tionofthegasketjoints.Anelementmidplaneiscreatedbyaveragingthecoordinatesofnodepairsfromthebottomandtopsurfacesoftheelements.Thestressnormaltothemidsurfaceofanelementinthegasketlayeristhesameasthegasketpressure.Asin-planedeformationandtransverseshearareneglected,thereisonlyonecomponentwhichisnormaltothegasket.Thuscompletegasketbehaviour(through-thicknessdeformationofthegasket)ischaracterizedbyapressure–versus–closure(relativedisplacementoftopandbottomgasketsurfaces)relationship.

3.3.Pretensioninthebolts

Pretensionelements(PRETS179)[9]showninFig.4areusedtomodeltheloadinaboltedjointduetotighteningatthetimeofassembly.Allpretensionelementswillhavea

aprimarysegmentwhichisrepeatedatequallyspacedintervalsabouttheaxisofsymmetry.Takingintoaccounttherotationalsymmetry,aquarter(901segment)modelofthejointisconsideredintheanalysisforaneightboltmodel.Similarly,one-sixth(601)andone-?fth(721)segmentsareconsideredforsixand10boltmodels,respectively.Fig.3showsthemeshdivisionusedintheFEanalysisforabolted?angejointwithaspiral-wound

352

M.MuraliKrishnaetal./InternationalJournalofPressureVesselsandPiping84(2007)349–357

K

Z

X

Fig.4.PRETS179element:(a)beforeadjustmentand(b)afteradjustment.

commonpretensionnode(K).ThisnodeisthethirdnodeforthepretensionelementwhereasnodesIandJareonthesectionedmid-surfaceofthebolt.SidesAandBonthepretensionsectionareconnectedbyoneormorepretensionelements,oneforeachcoincidentnodepair.Apretensionnode(K)isusedtocontrolandmonitorthetotaltensionloads.Inthe?rststage(bolting-up),loadwasappliedtothepretensionnodeasaforce.Theforce‘‘locks’’onthesecondstage(pressurized),allowingadditionalloads.Theeffectoftheinitialloadispreservedasadisplacementafteritislocked.Theboltandnutthreadsaremodeledaspartofanunthreadedshankwiththeminordiameterofthebolts,andtheboltheadandnutareassumedcylindricaltoavoidmeshingdif?culties.

3.5.Loadingandboundaryconditions

FEanalysisconsistsofboltpre-loadingandpressureloadingconditions.

3.5.1.Boltpre-loadingcondition

Agasketed?angeinthebolting-upstageisanalysedtoobtaintheinitialstressesanddeformationsinthe?angesandgasketduetotheclampingforcesduringassembly.Duetorotationalsymmetry,thedisplacements(boundaryconditions)andloadsareappliedonasinglesegmentintheanalysis.Forthemodelcreatedinacylindricalcoordinatesystem(r,y,z),thecircumferentialdisplacements(Uy¼0)areconstrainedinthe01and901planesforaneight-boltmodel.Similarly,circumferentialdisplacementsinthe601and721planesarearrestedalongwiththe01planeforsixand10boltmodels,respectively.Inthebolting-upstage,onlytheboltpre-load(F)hasbeenapplieduniformlyonthepretensionelementsectionthroughthepretensionnode.

3.5.2.Pressureloadingcondition

Thegasketed?angeisanalysedbyconsideringinternalpressureinadditiontotheboltpre-load.Duetotheinternalpressure,ahydrostaticendforceandpressureforceareinducedonthejoint.Thehydrostaticendforceexertedontheclosedendofapipesystemiscalculatedbasedontheinnerdiameterofthepipeandthetotalhydrostaticendforceiscalculatedbasedontheinnerdiameterofthegasket.Thepressureforceisthedifferencebetweenthetotalhydrostaticendforceandthehydrostaticendforceontheareainsidethe?ange.Thehydrostaticendforcehasbeenapplieduniformlyintheaxialdirectionatoneendofthepipeandtheotherendhasbeen?xedintheaxialdirection(Uz¼0).Theforceduetointernalpressurehasbeenapplieduniformlyintheaxialdirectionontheinnerfacesofboththe?angesurfaces.Thesameboundaryconditionsofconstrainingthecircumferentialdisplace-ments(Uy¼0)asappliedinthebolting-upconditionaremaintainedinthisstagealso.

Twentyloadstepsforeachofthebolting-upandpressurizedstageshavebeenconsidered.Itisfoundthat

3.4.Contactinterfaces

Inthepresentmodel,theboltandnutaretreatedasasingleentityandthe?angeringisaseparateentity.Sincethebehaviourofthesetwoaredifferentintermsofload–deformationcharacteristics,surface-to-surfacecon-tactelementsareusedtomodelcontactinterfacestocarryoutananalysisasathree-dimensionalcontactproblem.Asthemembers(?anges,gasket)andboltsaredeformablebothcanbetreatedasinthe?exible-to-?exiblecategory.Contacthasbeenconsideredbetweentheboltheadand?angeringinterfaceaswellasthenutfaceand?ange

interface.Theoptionwithnoslippingfrictionisusedsimulateinteractionofthecontactsurfaces.Since?angeringisstifferthanthebolt,itismodeledasatargetsurface(TARGE170)[9]andthebolthead/nutfaceismodeledasthecontactsurface(CONTA173)[9].Thecontactelementsthemselvesoverlaythesolidelementsdescribingtheboundaryofadeformablebody.Theseelementsconnectthenodesattheinterfacesorgapwithahighstiffnesswhentheinterfaceisincontactandwithverysmallorzerostiffnesswhentheinterfaceseparates.

Aftercarryingoutconvergencestudies,themodelisdiscretizedwith24,396solid,192interface,160pretensionand612contactelementsandatotalof32,642nodes.

M.MuraliKrishnaetal./InternationalJournalofPressureVesselsandPiping84(2007)349–357

353

20loadstepsadequatelyrepresentthenonlinearbehaviourandgiveconvergenceofthestressvaluesinbothstages.4.FlangerotationwiththeASMEcode

Angularrotationofa?angeunderthein?uenceofboltpre-loadandreactionforcesiscalled?angerotation.Thisismeasuredwithrespecttothecentreofthecross-sectionofthe?ange.TheASMEcodehasarigidityindex‘J’tocheckthe?angerotation.Therigidityfactorforintegral(weld-neck)isequivalent

toa?angerotationlimit0.31.Thismaynotbeadequate,astherotationofthe?angeisnotauniquevalueasitchangeswithinternalpressure.

Whenthegasketjointissubjectedtointernalpressure,thestressonthegasketisreducedasthepressuretriestomovethejointmembersapart.Asstressisreduced,thegasketexpandsduetoitselasticnature.However,undertheloadsusedinpracticalconditionsitdoesnotrecovertotheinitialgeometry.Experiencehasshownthatthereducedstressonthegasketunderpressurizedconditionsisasimportantastheinitialstress.Whenthegasketstressistoosmall,thejointmayleakorthegasketmaybeblownoutofthejoint.TheASMEcoderulesde?nethisreducedstress(referredtoasresidualstress)intermsofa‘‘maintenance’’(‘‘mollifying’’)gasketfactor‘‘m’’whichisdimensionless.Thedesiredminimumresidualstressissaidtobetheproductof‘‘m’’andthecontainedpressure(P).Thecodealsospeci?esa‘‘y’’factorwhichistheinitialgasketstressorsurfacepressurerequiredtopre-loadorsealthegaskettopreventleaksinthejointasthesystemispressurized[6,7].5.FEAresults5.1.Flangerotation

Whenboltsaretightenedtoachievethedesiredsurfacepressure,thesealingmaterial(gaskets)isdeformed.Duetotheeccentricityoftheboltpre-load,thehydrostaticendforceexertedonthe?anges,theseatingforce(gasketreactionload)andtheinternalpressureontheinnerfacesofthe?ange,abendingloadactsonthe?ange,therebycausingarotationofthe?ange.Thegasketissubjectedtoanon-uniformcontactstressduetothisrotationofthe?angeanddif?http://wendang.chazidian.comrgerotationmaycausebucklingoftheinnerwindingsorseparationofthesealingelementinthecaseofspiral-woundgasketsduetopossiblecontactofthe?angewiththeraisedfaceoutsidediameterwhichwillactasapivotpoint.

Fig.5showstherotationofthe?angeintermsoftheaxialdisplacementsatthebottomsurfaceoftheringportionalongtheradialdirectionforanasbestos?lledspiral-woundgasketmodelforaboltpre-loadF¼30kNwithdifferentinternalpressures(P¼0,3,5and10MPa).Itisobservedthataxialdisplacementsvarynon-linearlyintheradialdirection.Whilecalculatingthe?angerotations,

-1.04-1.06m

m ,-1.08tnemecal-1.1

spiD laix-1.12A32-1.14

-1.16

35

45

55

65758595105

Radial Distance, mm

P: 1 Fig.5.Flangerotationintermsofaxialdisplacementsalongtheradialdirection.

0.065

0.06

2

.

ged0.055

,noitatoR0.05

0.045

0.04

2

46810

Internal Pressure, MPa

Fig.6.Variationof?angerotationwithinternalpressure.

onlytheaxialdisplacementsattheinnerandouterradiusofthe?angeareconsideredforsimplicity.Fig.6showsthecalculated?angerotationwithanasbestos?lledspiralwoundgasketatdifferentboltpre-loadsandinternalpressures.Therotationislargewhenthejointissubjectedtointernalpressure.Table1showsthe?angerotationfordifferentspiral-woundgasketswitheightbolts.Itisobservedthatthecalculated?angerotationsarewellbelow0.31,asspeci?edbyASME[10].5.2.Variationofaxialboltforce

Whenthejointistightenedwithaboltpre-load,aninitialtensileforceintheboltandaninitialcompressive