着火油罐辐射热场的数值模拟与实验研究

PAGEPAGE699NumericalSimulationandExperimentalStudyonThermalRadiationofBurningOilTanksYANGJuntao1,WEIDong2,WANGHelan1&WUJun1(1ScientificResearchCenter,ShanghaiFireResearchInstituteoftheMinistryofPublicSecurity,Shanghai,200438,China;2DepartmentofFireFightingcommanding,ArmedPoliceAcademy,Langfang,Hebei,065000,China)Abstract:Oiltankfiresaredisastrousanddifficulttoextinguishbecauseoftheirlargeburningrate,highflametemperatureandwidefirearea.Andtheadjacentoiltanksmaybeignitedduetoheatradiationordirectconnectiontotheburningtank.Inordertoexploretheinherentpricipleoffiredevelopmentandpredictthevariationofburningcharacteristics,therealfiretestsongasolineanddieseloiltankwithdiametersof1.0m,1.5mand2.7mwerecarriedout.AndthenFDS(FireDynamicsSimulator),whichisbasedonlargeeddysimulation,waschosentosimulateoiltankfires.Byanalyzingthecalculateddata,thehorizontalandverticaldistributionregularityofthethermalradiationwerediscussed.Comparedwiththeexperiments,thepredictedresults,especiallythepredictionforthermalradiation,ofFDSaccordedwiththeexperimentaldataverywell.Theconclusioninthispapercanbehelpfulforthedeterminationoffirefightingtacticsandthefirepreventionplanfortankfarm.Keywords:oiltankfires;thermalradiation;numericalsimulation;FDS(FireDynamicsSimulator)model;firetests1IntroductionThethermalradiationofoiltankfiresisthedirectorindirectreasonwhichresultsinthefirespreadbetweentanks,soitisnecessarytopredictthecharacteristicsofthermalradiationandtheirvariationtrendintheburningprocessforpreparingthefirepreventingmeasuresinoiltankfarm,indecidingthereasonabledistancebetweentanks,andensuringthesafetyoffirefighting.Largenumbersofexperimentalstudieshavebeencarriedoutathomeandabroad.Akita&Kashio[1]foundthattheintensityofthermalradiationreducesalongwiththeincreaseofdiameterofoilbasininthelocationswithsimilargeometricalcondition.Andtheyalsofound“smokeblockageeffect”.TheprincipleofthermalradiationchangingwiththediameterofoilbasinwassummedupbyHiroshiKoseki[2]throughthestudyofthermalradiationfluxofdifferentfuel.AndsomerelatedstudieshavebeenalsocarriedoutbysomeorganizationsinChina,suchasTianjinFireResearchInstitute[3],ArmedPoliceAcademy[4,5]andSateKeyLaboratoryofFireScienceetc.,andsomevaluableexperimentdataandresultshavebeenobtained.Thetruevariationtrendofthermalradiationinoiltankfirescanbeobtainedthroughexperimentstudy.Sotherealfiretestsongasolineanddieseloiltankwithdiametersof1.0,1.5and2.7mwerecarriedout.Inthetests,thehorizontalandverticaldistributionofthermalradiationwasmeasured.Inaddition,theexperimentsofferedtheboundaryconditiontothenumericalsimulation.Butaconsistentresultisdifficulttobeobtainedduetothebadrepeatabilityofoiltankfireexperimentswhicharestronglyinfluencedbyweatherandsurrounding.Sothewayofusingcomputertosimulatethedevelopingprocessandtheinfluenceofoiltankfirestotheenvironmentisamorepracticablemethod.Throughthecomparisonandanalysisof17fieldmodels[6],theburningprocessesofoiltankfiresweresimulatedbyusingFDS(FireDynamicsSimulator)andthevariationtrendofradiationfieldwereanalyzed.Atthesametimethesimulationresultswerevalidatedbyusingthetestdata.2OilTankFireExperiment2.1IntroductionofExperimentSystemTheexperimentsofoiltankfireswithdiametersof1.0,1.5and2.7mwerecarriedouttoobtainthebasicdataofthermalradiation.Thefiretestsinsituwereshowninfigure1.Inordertoknowtherelationshipofthermalradiationwithburningcharacterofflame,theradiantheatflux,aswellastheburningrate,thetemperatureofflame,tankwallandinneroil,thedirectionandvelocityofwindweremeasured.Theexperimentsystemincludedthreeparts:oiltank,datameasureandcollectionsystemandcoolingsystem.Anddatameasuringsystemmainlymeasuredtemperature,thermalradiationandburningrate.Inordertomeasuretheinfluenceofheatconvectionaroundtheburningoiltanktotheadjacenttanksandcomparetothemeasurementofthermalradiation,thewholeheatflux(thesumofradiantheatfluxandconvectiveheatflux)wasalsomeasured.Amongthese,theflametemperatureandwalltemperatureweremeasuredby1.5thermocoupleoftypeK,whichwerecalibratedwithanaccuracyof0.5℃,andtheoiltemperaturewasmeasuredby2.0thermocoupleoftypeKwithastainlesssteelprotectortube.Theambienttemperatureandhumidityweremeasuredbythewetanddrybulbthermometer,andtheaerovaneofPR-550typemeasuredtheinstantaneouswindvelocityandwinddirection.Then,theanalogsignalsoftemperature,heatflux,windvelocity,winddirectionwereconvertedintothedigitalsignalthroughYanhuaADAM5000seriesmodule,finallywereimportedintothecomputer.Theradiantheatfluxmeterandthewholeheatfluxmeterneedtobecooledbywaterinnormalworkingcondition,soacoolingwatersystem,drivenbythecentrifugalpump,withthewatertankof1000Lwasmadespeciallytoprotectthesedevicesandthetemperaturebar,soastomakethemendurehightemperatureflame.Inaddition,2groundmonitorand3mobilefirenozzlewereinstalled,whichcouldprotectoperators’safety.2.2MeasurementofThermalRadiationInordertomaketheexperimenttobeclosewiththerealsite,anadjacentandastimulanttankwerelocatedin1Dpointaroundtheburningtank.Theadjacenttankwassealedandcontainedsomeoil,andthestimulanttankwasopenandheatflowmeterscouldbeinstalledindifferentplacesofthesurfaceoftank.Onlytheradiationplaneswhichwerenormaltogroundwereconsideredandthedistributionprincipleofthermalradiationwasfocusedonintheseexperiments.Sotherewasnoheatflowmetermountedintheadjacentandstimulanttank,onlythechangeoftemperatureandstressofthemwereexamined.Thelocationofmeasuringpointswasshowninfigure2.TheverticalpointswerelocatedintheL=1.1D,1.5Dand2.0Dfromtwosidesoftank(Lwasthehorizontaldistancefromthecenteroftank).OnesidelocatedthewholeheatfluxmetersRL7,RL2andRL4,andtheothersidelocatedtheradiantheatfluxmetersRE4,RE1andRE2,andthewholeheatfluxmetersRL5,RL1andRL3,andallthemetersmeasuredatthesametime(Figure1).Inordertoexploretheverticaldistributionprincipleofradiationheat,heatfluxmetersRL8,RL6andRL4werelocatedinL=2.0Dpositionindifferentheightandtheverticaldistancebetweenanytwomeasurepointswas0.8m.FFig.1FiretestsinsituAdjacentTankBurningTankStimulanttank52L/D=1.5L/D=2.01.0D1.0DRL1,RE1RL3,RE2RL2RL4L/D=2.0L/D=1.5Fig.2MeasuresystemofthermalradiationL/D=1.1RL5,RE4L/D=1.1RL7RL6RL83NumericalSimulationResultsandDiscussionFDS(FireDynamicsSimulator)[7]isacomputationalfluiddynamics(CFD)modeloffire-drivenfluidflow.Thesoftware,whichwasdevelopedonthebasisofLES(LargeEddySimulation)byNIST,solvesnumericallyaformoftheNavier-Stokesequationsappropriateforlow-speed,thermally-drivenflowwithanemphasisonsmokeandheattransportfromfires.FDSusesamixturefractioncombustionmodel.Radiativeheattransferisincludedinthemodelviathesolutionoftheradiationtransportequationforanon-scatteringgraygas.TosimulateoiltankfiresbyusingFDS,thefirstthingrequiredtodoistoprogramappropriateinputdocument.ThefollowinginformationshouldbeincludedintheinputdocumentofFDS:geometricconfigurationofbuildingsincontrolarea,dimensionofcalculationunit,predefinedfiresource,fueltype,heatreleaserate,thermophysicalpropertyofbarriers,boundaryconditionsetc.Accordingtotheactualconditionofexperiment,inputdocumentwasprogrammed,andtheoiltankfireswithdiametersof1.0,1.5and2.7mweresimulated.Fig.3VariationtrendofradiationalongwithtimeforafixedpointRadiantflux/(kW·m-2Fig.3VariationtrendofradiationalongwithtimeforafixedpointRadiantflux/(kW·m-2)Time/sThevariationtrendofthermalradiationalongwithtimeforafixedpointinoiltankfireisshowninfigure3.Itshowsthatthermalradiationincreaserapidlyatthebeginningandbecomestableafter5s,withasmallfluctuationof1.5kW/m2orso.Itisalittledifferentwiththeactualtrendbecauseitisanidealsituationwithfixedheatreleaserateandtheflamespreadvelocitywasnotconsidered,sothetimefromignitiontostableburningwasveryshort.Whileinactualexperiment,becauseofflamespreadandmeasurelag,ittookabout60-70sfortheburningof1mdiameteroiltanktobecomestable.3.2HorizontalDistributionofThermalRadiationIntensityThehorizontaldistributionofaveragethermalradiationintensitywithnowindisshowninfigure4.Timeaveragethermalradiationintensityistheaverageintensityofthermalradiationinaperiodoftime.hwastheheightfromgroundandHgwastheheightoftank(thefollowingwerethesameifnospecialexplanationbegiven).Theconditionforcalculationwas:gasoil,ambienttemperatureTa=20C,calculationpointswerelocatedinh/Hg=0.4,0.6and0.8.Itshowsfromthefigurethatthehorizontaldistributiontrendofradiationintensityof1mand2.7mdiameterburningoiltankswereverysimilar.TheradiationintensitydecreasesalonewiththeincreaseofdimensionlessdistanceL/Dwithnowind.WhenL/D=3,thethermalradiationintensitywasmuchlowertoabout1kW/m2.Theradiationintensitywasgettinghighrapidlyalongwiththeincreaseofh/Hg.Apeakvalueofhorizontalradiationintensityexistsneartankwallwhenh/Hg<0.4for1moiltankandh/Hg<0.6for2.7moiltank.Thisheightwherepeakvalueoccurswasgettinghighalongwiththeincreaseoftheheightoftank.Andthepositionwherethepeakvalueoccurswasgettingfarfromtankalongwiththeincreaseofthediameteroftanks.Thiswasresultedfromthedecreaseofradiativeanglefactoroftargetsurfacesafterthecoveroftankwhentargetsurfacesneartankwall.(a)D=1(a)D=1m(b)D=2.7mFig.4HorizontaldistributionofaveragethermalradiationintensitywithnowindRadiantflux/(kW·m-2)Radiantflux/(kW·m-2)Itisshowninfigure5thattheverticaldistributionofaveragethermalradiationintensitywithnowindfor1mand2.7mdiameteroiltanks.Theconditionforcalculationwas:gasoil,ambienttemperatureTa=20C,calculationpointswerelocatedinL/D=1.5,2,2.5and3.Asisshowninfigure,thattheradiationintensityincreasesatfirstandthendecreasesalongwiththeincreaseofheight,andreachestothemaximumat0.4,andthiswasinaccordancewithexperiment.AccordingtothecomparisonoftheamountofthermalradiationintensityindifferentL/D,itisfoundthatthistrendofhighinmiddleandlowintwoendswasdecreasinggraduallywiththeincreaseofdistancefromtanksandfinallytrendslevel.Thedistributionofradiationwasahalfballsymmetricaldistributionalongthecenterlineofoiltanks,andtheprimarytrendsfordifferentdiametertanksweresimilar.(a)D(a)D=1m(b)D=2.7mFig.5VerticaldistributionofaveragethermalradiationintensitywithnowindRadiantflux/(kW·m-2)Radiantflux/(kW·m-2)ThevariationtrendofthermalradiationintensityofburningoiltanksobtainedbynumericalsimulationusingFDSwasstatedandanalyzedabove.Buttheseresultsandprinciplesareneededtobecomparedwithexperimentdatatoverifyitsaccuracy.4.1HorizontalDistributionofThermalRadiationIntensityFigure6showsthecomparisonofhorizontaldistributionofthermalradiationintensitybetweencalculationandexperimentdata.Thechosendataofexperimentandcalculationweretheaveragevaluesofaperiodoftimeinstableburningstageinordertoavoidtheinfluenceofturbulentfluctuation.Thevariationtrendsofexperimentandcalculationdatawerethesamenomatterforgasordieseloil,andthesetwokindsofvaluewereveryclosewithadeviationofabout10%.Radiantflux/(kW·m-2)Radiantflux/(kW·m-2)(a)D=1m,gas(b)DRadiantflux/(kW·m-2)Radiantflux/(kW·m-2)Radiantflux/(kW·m-2)Radiantflux/(kW·m-2)(c)D=2.7m,gas(d)DRadiantflux/(kW·m-2)Radiantflux/(kW·m-2)Fig.6Comparisonofhorizontaldistributionofthermalradiationintensity4.2VerticalDistributionofThermalRadiationIntensityRadiantflux/(kW·m-2)Radiantflux/(kW·m-2)Figure7showsthecomparisonofverticaldistributionofthermalradiationintensity.Thevariationtrendsofexperimentdataandcalculationarethesameforgasanddieseloil.Bothhaveextremums,andthepositionswhereoccurextremumswasgettinghigheralongwiththeincreaseofdiameteroftanks.Theexperimentdatadecreasedrapidlyafterextremum,whilethecalculationdatahasahighradiationareawherekeepshighradiationvalue.Figure7ashowsthatprecisionishighinlowerheightandbecomeloweralongwiththeincreaseofheight.AndwhenH=1.9m,thecalculationvaluewasthedoubleofexperimentvalue.Thisvariationmayberesultsfrom“smokeblockageeffect”.Intheexperiment,thesmokewasgettingthickerandthickeralongwiththeincreaseofexperimentscale,andmainlydistributesintheupsideofflameandwrapstheflamejustshowninfigure8.Theradiationfractionisdecreasedbecausesomeoftheradiationofsmokewasabsorbedbysmokeparticles.Butincalculationthesmokeblockageeffectwasignored.Andtheradiationfractionissetasaconstant,soalowerprecisioniRadiantflux/(kW·m-2)Radiantflux/(kW·m-2)(a)D=1m,gas(b)D=2.7m,gasRadiantflux/(kW·m-2)(c)D=1mRadiantflux/(kW·m-2)Fig.7ComparisonofverticaldistributionofthermalradiationintensityFig.8Fig.8Oiltankfireexperimentinsitu5Conclusions(1)AnumericalsimulationwascarriedoutontheradiationfieldofoiltankfiresbyusingFDS.Theresearchresultsarehelpfulfortheprotectionoffirefightersinfireandfiredesign.(2)Thehorizontaldistributionofthermalradiationofburningoiltankwereintypeofexponentialdecay,peakvaluesoccurredneartankwall,andtheheightwherepeakvalueoccu