WestinghouseAP1000internalsheatingratedistributioncalculationusinga3DdeterministictransportmethodOriginalResearchArticle

Costreductionandreliabilityincreasearesystematicallypursuedsystemsandcomponents;thisrequires,amongstother,theavailabilityofsophisticatedcomputerprogramsanddetailedanalysismodels. Asanexample,thecoreshroud,thestructurehavingthefunctiontomaintainthecorecenteredonitsaxis,isbeingdesignedintheWestinghouseAP1000,differentlyfrompreviousplants,asahighlyheterogeneousstructure.Itsthermal-mechanicalsizingmusttakeintodueaccountaccuaratelydeterminedinternalheatgenerationrates.Thelatter,ifdeterminedbycombining2Dand1Dneutronandγ-raycalculationswhichimplytheseparationofspatialvariablesandaremainlyapplicableforfluxesinthereactorbeltlineregion,mayincludeoverlyconservativemargins.Ontheotherhand,MonteCarlomethodsdonotallowaneasyquantificationoftheuncertaintiesrelatedtooverallcalculation. Three-dimensionaldeterministicmodels,basedonthediscreteordinatetransporttheory,havethepotentialtoprovideaccuaratedesigndata;theycanbealsoeffectiveprovidedthatthewell-knowndifficultytocreateandtuneacomplexgeometricalmodelinareasonabletimeisovercomeandadequatecomputerresourcesareavailabletoperformthecalculation(untilfewyearsago[Bottaetal.,1996.Three-DimensionalReactorPressureVesselFastNeutronFluenceCalculationsfortheAP600UsingTORT,3-DDeterministicRadiationTransportComputerPrograms:Features,ApplicationsandPerspectives,NEA/NSC/DOC,OECD/NEA.Paris,France],massiveparallelcomputers(i.e.CrayComputers)availableonlytolargenationallaboratoriesandselectedindustrieshadtobeused). ANSALDOisactingasWestinghousesubcontractoranditsupportedWestinghouseinallAP1000LicensingProcesstoNRCfrom1999tonowdaysfortheinternalheatingrategenerationrateandRPVfluencecalculations.AscomputerpowergrowthsupANSALDOrefineditscalculationmethodologyinordertoimprovethedesignitself.Therecentdevelopment,byENEABologna,ofapre-postprocessorbasedonthecombinatorialgeometry(thecomputerprogramBOT3P,[ORSI,2002.BOT3P:BolognaTransportAnalysisPre-Post-processorsVersion1.0,NuclearScienceandEngineering,no.142.AmericanNuclearSociety,USA,pp.349–354;ORSI,2004.BOT3P:BolognaTransportAnalysisPre-Post-processorsVersion3.0,NuclearScienceandEngineering,no.146.AmericanNuclearSociety,USA.pp.248–255;ORSI,2005.BOT3P:BolognaRadiationTransportAnalysisPre-Post-ProcessorsVersion4.0,NuclearScienceandEngineering,no.150.AmericanNuclearSociety,USA,pp.368–373;ORSI,2003.BOT3P:AMeshGenerationSoftwarePackagefortheTransportAnalysisCodesDORT,TORT,TWODANT,THREEDANTandMCNP,InternationalConferenceonSupercomputinginNuclearApplicationsSNA’2003.Paris,France;ORSI,2004.NEA-1678/05BOT3P4.1.BOT3P4.13DMeshGeneratorandGraphicalDisplayofGeometryforRadiationTransportCodes,DisplayofResults”,OECD/NEADataBank.Issy-les-Moulineaux,France])andtheongoingincrementofcomputerpowermakethethree-dimensionaldeterministicprogramTORT(partoftheDOORS3.2Package,[RSICCComputerCodeCollectionDOORS-3.2,1998,OakRidgeNationalLaboratory,OakRidge,Tennessee,USA])anattractivechoicetoperformthedeterminationofthecoreshroudheatingratedistributionasa3Dcontinuefunctions. ThispaperillustratesthecomputationalapproachfollowedbyANSALDONuclearetodeterminetheheatingratedistributionintotheWAP1000CoreShroud(seeFig.1).Thecalculationhasbeenperformedat67energygroups(47neutronsplus20gammas)inP3orinP5Legendreapproximationand2,000,000mesheshavebeenused.Thecomputationalpowerrequiredismassive:2GBofRAMandupto50GBofswapfilespeakstorageforeachrunwithanS8full-symmetricangularquadratureand120GBofswapfilespeakstoragewithanS16full-symmetricangularquadrature. ByusingalatestgenerationworkstationbasedonanINTELP-IV3.2GHzprocessorunderLINUXoperatingsystem(bothSUSE9.2orRedHat9.1andbetteraresuitable),eachrunrequireslessthanaweekoffullcomputertimewiththeP3LegendreapproximationwithanS8full-symmetricangularquadrature. ThecodeBOT3PhasbeenusedbothforthepreparationofthegeometricalmodelandaspostprocessoroftheTORTprogramoutputs.BOT3PpermittedtocutthemodeldevelopmenttimebyasignificantfactorandtoretrieveefficientlylargeamountofdatafromtheTORTbinaryfiles,allowingagraphicalreviewoftheinputandoutputdataand,finally,resultinginauniquetoolforQualityControlforcomplex3Dmodels. Thiscomputationalexperiencecanbeconsideredagoodworkingexampleforthe3Ddeterministictransportmethodrevivalbyconsideringthatthismethodiswidelyacceptedbylicensingauthorities.