Fire-Dynamics-Simulator-(Version-4.0)-Simulation-f

FireDynamicsSimulator(Version4.0)SimulationforTunnelFireScenarioswithForced,Transient,LongitudinalVentilationFlowsEstherKimandJohnP.WoycheeseandNicholasA.Dembsey*,DepartmentofFireProtectionEngineering,WorcesterPolytechnicInstitute,100InstituteRoad,Worcester,MA01609–2280,USAAbstract.Inthisstudy,aseriesofsensitivityanalyseswereconductedtoevaluateacomputationalfluiddynamic(CFD)model,FireDynamicsSimulator(FDS)version4.0,fortunnelfiresimulations.Atunnelfiretestwithafiresizeontheorderofa100MWwithforced,time-varyinglongitudinalventilationwaschosenfromtheMemorialTunnelVentilationTestProgram(MTVTP)afterconsideringrecenttunnelfireaccidentsandtheuseofCFDmodelsinpractice.AcarefulstudyofgridsizeandparametersusedintheLargeEddySimulation(LES)turbulencemodel—turbulentPrandtlnumber,turbulentSchmidtnumber,andSmagorinskyconstant—wascon-ducted.MoredetailedanalyseswereperformedtorefinethesmokelayerpredictionofFDS,especiallyonbackflow(i.e.,areversedsmokeflowneartheceiling).Also,energyconservationwascheckedforthisscenarioinFDS.AsimpleguidelineisgivenforsmokelayersimulationsusingFDSforsimilartunnelfirescenarios.Keywords:tunnelfire,CFD,FDS,modelevaluation,transientflow,longitudinalventilation,backlayering,backflowNomenclatureD*characteristicfirediameter_QheatreleaserateDgridlengthscalelviscositycpspecificheatcapacityCsSmagorinskyconstantDmassdiffusivitykconductivityPrturbturbulentPrandtlnumberQ*characteristicfiresizeScturbturbulentSchmidtnumberTtemperature*Correspondenceshouldbeaddressedto:NicholasA.Dembsey,E-mail:[e-mail],44,137–166,2008
2007SpringerScience+BusinessMedia,LLC.ManufacturedinTheUnitedStatesDOI:10.1007/s10694-007-0028-212u,v,wflowvelocitiesinx,yandzdirectionqdensitySubscripts¥ambientLESforLargeEddySimulation1.IntroductionOverthepast10years,severalaccidentsintunnelshavedemonstratedtheseriousconsequencesofuncontrolledfiresinconfinedspaces[1].TheMontBlanc(France/Italy)incidentin1999involvedmultipleheavygoodsvehicles(HGVs)andpassengercarsandhadanestimatedpeakheatreleaserate(HRR)of300–380MW,whichissignificantlyhigherthantypicaldesignfiresfortunnelanalyses[2,3].Overfortypeoplewerekilledandthetunnelwassignificantlydamaged.Twoyearslater,asimilarfireintheSt.Gotthard(Switzerland)tunnel[4]resultedin11fatalities.Inadditiontothehumancost,tunnelclosuresforstructuralrepairs,asexperiencedfromthesepastaccidents,havesignificanteconomicimpactoverawiderregion.Theincreasingfrequencyofthesefireincidentshasledtogreaterefforttoidentifyandreducethepotentialfirerisksintunnels.Oneofthetoolscommonlyusedbyengineerstostudyanddesignfire-safetun-nelsiscomputationalfluiddynamics(CFD)simulation,whichismorecost-effectivethanconductinglarge-scalefireexperiments.Themodelparametersmustbecare-fullytunedforeachscenariotoensurethatappropriateresultsaredeveloped.Simu-lationswithoutthiscriticalstepmaygiveafalsebasisforengineeringdecisions.Aswithanyothertypesofmodels,CFDsimulationshavelimitationsinpre-dictingthedynamicsofthefluid,asdeterminedbyassumptionsinherentinthemodel.Asalgorithmsbecomemorecomplex,itishardertoidentifythelimita-tionsofthemodel.Hence,testingthemodelunderknownconditionsandcom-paringthesimulatedresultstodatafromaparticularfiretestisimportantbeforeusingthemodeltopredicttheresultsofasimilar,untestedscenario.Thismethod-ologyidentifiescomputationallimitationsoftheCFDmodels.Inthisstudy,FireDynamicsSimulator(FDS)Version4.0[5,6],whichisawell-knownCFDtoolforthefireprotectionengineeringfield,wasused.ThismodelwasdevelopedbytheNationalInstituteofStandardsandTechnology(NIST).FDSwasevaluatedviasensitivityanalysesofgridsizesandturbulentparameterstodetermineitsabilitytoresolvelargefires,ontheorderofa100MW,intunnelswithforced,transient,longitudinalsmokeventilation.OtherresearchthatpresentsasimilarmodelvalidationworkusingFDSfortunnelfirescenarioshasbeenconducted[7–9].Thesewillbebrieflydiscussedintheback-groundsection.Thisstudyisdifferentfromothersimilarworksintermsofusingarelativelylargefiresizeandtime-varyingventilationvelocity(2.5–4m/s)tosim-ulatetheforced,transient,longitudinalventilationsystem.TheresultsfromthisworkidentifylimitationswhenusingFDSforsimilartunnelfiresandprovideguidelinesforsimulatingthesetypesoffirescenarios.138FireTechnology20082.BackgroundInthissection,backgroundresearchonFDSVersion4.0[5,6],theMemorialtun-nelfiretest[10]621A,andotherFDSmodelevaluationwork[7–9]ontunnelfiresareintroduced.TheresearchresultsforFDSandmodelevaluationareincludedforgeneralinformationaboutthemodelandpaststudiesusingFDSontunnelfires.TheMemorialtunnelfiretestisdiscussedtoprovidegeneralinformationabouttheexperimentandtoexplainwhyTest621Awasofspecialinteresttothisstudy.2.1.FireDynamicsSimulatorVersion4.0FDS[5,6]simulatesthermallydrivenflowswithrelativelylowspeedcomparedtothespeedofsound.Themodelassumesthattheflowisweaklycompressible,whichallowslargertimestepsbecauseacousticpressurewavesarenotconsideredinthecalculations.Inturbulentflows,therearewiderangesoflengthandtimescalesinvolvedinthedynamicsoftheflow.FortheturbulencecalculationsinthegasphasetheusercanapplyeithertheDirectNumericalSimulation(DN