定量火灾风险评估.ppt

Chapter5QuantitativeFireRiskAssessment,5.1OverviewInthischapter,quantitativefireriskassessmentisintroduced.Thetermquantitativefireriskassessmentreferstoanassessmentinvolvingnumericalquantificationsnotonlyoftheprobabilityafirehazard,orfirescenariooccurring,butalsotheconsequencesofthatfirehazardorfirescenario.,Bymultiplyingthenumericalvaluesofprobabilityandconsequenceeachfirescenarioisgivenanumericalfireriskvalue.Byaccumulatingthesumoftheriskvaluesfromallprobablefirescenarioswecanobtainanoverallfireriskvalue.Theoverallfireriskvaluecanbeusedforcomparisonswiththoseofalternativeorcode-compliantfiresafetydesigns.,Ingeneraltherearetwowaystoperformsystematicquantitativefireriskassessmentsasfollows:1.byusingachecklisttogothroughalistofpotentialfirehazardsandthequantitativeassessmentoftheirfirerisks;2.byusinganeventtreetogothroughasetofpotentialfirescenariosandthequantitativeassessmentoftheirfirerisks.,Inboththesemethods,thevaluesfortheprobabilityandconsequenceparametersareobtainedfromstatisticaldata,iftheyareavailable,orfromsubjectivejudgment,ifsuchdataarenotavailable.(仅在我们本课程的讲解中，采用了统计或主观估计的方式，火灾的结果还可以通过确定性的分析计算获得),5.2ChecklistMethodAswasdiscussedinChapter4,thechecklistmethodemploysthecreationofachecklistofpotentialfirehazardsandtheconsiderationoffireprotectionmeasures,eitherinplaceortobeadded,toarriveatanassessmentofthefirerisks.Thecreationofachecklistofpotentialfirehazardsallowsasystematiccheckofpotentialfirehazardsthatareinplace.,Thelistingoffireprotectionmeasuresalongsidewiththepotentialfirehazardsallowsaquickcheckofanysafetydeficienciesandanyneedtoprovideadditionalfireprotectionmeasurestominimizetherisk.Thechecklistmethod,therefore,isanenumerationofpotentialfirehazards,fireprotectionmeasures,eitherinplaceortobeadded,andtheassessmentoftheresidualfirerisks.,Itisusedtoidentifyanydeficienciesandanycorrectivemeasuresneededtominimizethefirerisks.Itdoesnotinclude,however,theconsiderationofthelogicaldevelopmentoffireevents,whichwillbediscussedinSection5.3usinganeventtree.,AnexampleofachecklistmethodemployingquantitativefireriskassessmentisshowninTable5.2.ThisisthesameexamplethatwasusedinChapter4,exceptthatquantitativeassessmentisemployedhereratherthanqualitativeassessment.,Thisexamplelooksatapotentialfirehazardinthelivingroomofahouseandtheconsiderationofanumberofadditionalfireprotectionmeasurestominimizetherisk.Obviously,therecouldbemanypotentialfirehazardsinahouse.Acompletefireriskassessmentwouldinvolvetheidentificationofallpotentialfirehazardsandtheconsiderationofvariousfireprotectionmeasurestominimizetherisk.,Atypicalhouseusuallyhassomefireprotectionmeasures,suchassmokealarms.Additionalfireprotectionmeasureswouldlowertheriskfurther.SimilartotheexampleinChapter4,thisexampleconsiderssixdifferentcombinationsofthreeadditionalfireprotectionmeasures.,Thethreeadditionalfireprotectionmeasuresare:(1)nosmokingmaterial(suchascigarettes)inthelivingroom,(2)sprinklers(3)regularevacuationdrills.Eachofthethreefireprotectionmeasureshasanimpactoneithertheprobabilityoffireoccurrenceortheconsequenceofafireoccurrence.,Forexample,themeasureofnosmokingmaterialinthelivingroomwouldhaveanimpactonloweringtheprobabilityoffireoccurrence;whereasthemeasuresofsprinklersandregularevacuationdrillswouldhaveanimpactonloweringtheconsequenceofafireoccurrencebysuppressingorcontrollingthefireorbyallowingtheoccupantstoevacuatemorequickly.,Itshouldbeemphasizedthatthisisjustanexampletoshowhowquantitativefireriskassessmentcanbecarriedoutusingachecklistmethod.Therearenostandardchecklistmethodsinfireriskassessment.,InTable5.2,theinherentfireriskvalues(withoutthehelpofanyfireprotectionmeasures)wereobtainedpreviouslyinChapter3.Table3.2inChapter3showsthattheprobabilityoffireoccurrenceinCanadianhouseswas1.75103fires/house/yearin1996andthepercentageofthesehousefiresthatoccurredinthemainlivingareawas8.5%.,Usingthesefigures,theprobabilityoffireoccurrenceinthemainlivingareainCanadianhousesin1996was,therefore,1.751038.5%or1.49104fires/house/year.,Table3.2inChapter3alsoshowsthattheconsequenceoffiresoriginatinginthemainlivingareain1996was43.2103deaths/fire,andtheresultantrisktolifefromthesefireswas6.43106deaths/house/year.ThesepreviouslyobtainedinherentriskvaluesareusedinthepresentexampleandareshowninTable5.2.,TheinherentriskvaluesinTable5.2werebasedonfirestatisticswhichincludedsomefireprotectionmeasures,suchassmokealarms,thatwererequiredbyregulations.Ifadditionalfireprotectionmeasuresareputinplace,theinherentfireriskswouldbefurtherreduced.,InTable5.2,theimpactofeachofthesixfireprotectioncombinationsisassessedusingaresidualmultiplicationfactoroftheinherentvaluesoftheprobabilityortheconsequence.Thisallowsthefireprotectionengineersandtheregulatorstoassesstheimpactofthesefireprotectionmeasuresbasedontheirassessmentsofthereductionoftheinherentvalues.,Onewaytoassesstheimpactoffireprotectionmeasuresisthroughtheuseofstatisticalinformation,iftheyareavailable.Unfortunately,suchinformationisnotalwaysavailable.Theinformationmaybeinthedatabasesofcollectionagencies,butnotnecessarilyintheirpublishedreportswhichusuallyshowgeneralinformationandnotthespecificinformationthatisrequiredforfireriskassessment.,Ifnosuchinformationisavailable,subjectivejudgmentmayberequired.Otherwise,theuseoffundamentalandrationalapproachtoquantificationisrequired,includingtheuseofmathematicalmodelingoffiredevelopmentandoccupantevacuation,whichwillbediscussedinlaterchapters.,Forexample,thereissomestatisticalinformationonthebenefitsofrestrictingsmokingmaterialandofinstallingsprinklers,butnotmuchinformationonthebenefitsofimplementingregularevacuationdrills.,NFPAstatisticsshowthatapproximately7%offiresinhomesarecausedbysmokingmaterials(NFPAFireStatistics,2006)andapproximately14%ofthesefiresoccurinthemainlivingarea(Hall,2006).Therefore,714%orapproximately1.0%offiresinhomesarefiresthatbothoriginateinthemainlivingareaandarecausedbysmokingmaterial.,IfthesefirestatisticscanalsoapplytoCanadianhomes,thenrestrictingsmokingmaterialinthemainlivingareawouldreducethenumberoffireoccurrenceinthemainlivingareafrom8.5%(seeTable3.2inChapter3)to7.5%ofhousefires.,Thereductionoffireoccurrencefrom8.5to7.5%is12%.Thecorrespondingresidualmultiplicationfactoroftheinherentprobabilityvaluebyrestrictingsmokingmaterialistherefore0.88,whichisshowninTable5.2.,NFPAstatisticsalsoshowthat,basedon19891998data,thereductionindeathsinoneandtwofamilydwellingswithsprinklersis51%whencomparedwithsimilardwellingswithoutsprinklers(KimberlyandHall,2005).Thecorrespondingresidualmultiplicationfactoroftheinherentconsequencevaluebyinstallingsprinklersistherefore0.49,whichisshowninTable5.2.,Withregardtothebenefitsofimplementingregularevacuationdrills,thereisnoinformationonthereductionofdeathratesthatiseasilyavailable.Forthisexample,wehavetomakeanassumption.Weknowthatifregularevacuationdrillsarecarriedout,therewillbefasterevacuationsandthereforelowerdeathrates.Forthisexample,letusassumeareductionofthedeathrateby60%.,Theresidualmultiplicationfactoroftheinherentconsequencevaluebyimplementingregularevacuationdrillsistherefore0.40,whichisshowninTable5.2.Inactualfireriskassessments,thisvalueneedstobejudgedandagreeduponbetweenthefireprotectionengineersandregulators.,Table5.2alsoshowsthattheimpactsontheconsequenceofinstallingsprinklersandofimplementingregularevacuationdrillsaremultipliedtogether.Thatis,thebenefitsofsprinklersandofregularevacuationdrillshaveacombinedresidualconsequencefactorof0.490.40or0.20.,Themultiplicationoftheresidualfactorsisbasedontheargumentthateachfireprotectionmeasurereducestheresidualdeathratebyacertainpercentageinsuccession.Thedeathrateisfirstreducedbythesprinklerssuppressingtheseverityofthefires,andthenfurtherreducedbyfasterevacuationoftheoccupants.,ThereductionoftheriskvaluesofthesixcombinationsofadditionalfireprotectionmeasuresisshowninTable5.2.Theresidualriskmultiplicationfactorsrangefrom0.88to0.20.Thequantificationoftheriskvaluesallowsnumericalcomparisonsofthevariousfireprotectionoptions.Thisisnotthecaseinqualitativefireriskassessment(Table4.4inChapter4).,ItshouldbeemphasizedagainthatthevaluesinTable5.2areselectedasanexampletoshowhowsuchachecklistmethodcanbecarriedout.Thesevalueswereselectedfromavailablestatisticalinformationwithoutmuchin-depthsearch.Asmorestatisticalinformationbecomesavailable,moreextensivesearchanddetailedanalysisareneededtofindthecorrectvalues.Inactualfireriskassessments,thesevaluesneedtobecarefullyselectedandagreeduponbystakeholders.,Inactualfireriskassessments,thesevaluesneedtobecarefullyselectedandagreeduponbystakeholders.Subjectivejudgmentoftheprobabilitiesandconsequencesprovidesaquickassessmentofthepotentialfirerisks.Morefundamentalandrationalapproachestoquantification,includingtheuseofmathematicalmodelingoffiredevelopmentandoccupantevacuation,willbediscussedinlaterchapters.,5.4Event-TreeMethodAneventtreeisanotherwaytoidentifypotentialfirehazards,assesstheirprobabilitiesandconsequences,andarriveatriskvalues.Differentfromthechecklistmethod,aneventtreeshowsmorethanalistofpotentialfirehazardsandfireprotectionmeasuresfortheassessmentoftheprobabilities,consequencesandeventuallytheriskvalues.,Theevent-treemethodinvolvestheconstructionofaneventtreeofvariousfirescenariossubsequenttotheinitiationofafirehazard,asdescribedinChapter2.Thefirescenariosprovidemorelogicalinformationforthejudgmentofprobability,consequenceandriskvalues,.Anexampleofanevent-treemethodemployingquantitativefireriskassessmentisshowninFigure5.1.ThisisthesameexamplethatwasusedinChapter4,exceptthatquantitativeassessmentisemployedratherthanqualitativeassessment.,InFigure5.1,thebranchingtodifferenteventsdependsonthesuccessorfailureofthefireprotectionmeasuresinplace.Thisexamplelooksatonefirehazardinanassumedapartmentbuildingandtheconsiderationofanumberofadditionalfireprotectionmeasurestominimizetherisk.,Thesameeventtreecanbeconstructedformorehazardsandmorefireprotectionmeasures.Acompletefireriskassessmentwouldinvolvetheidentificationofallpotentialfirehazardsandtheconsiderationofvariousfireprotectionmeasurestominimizetherisk.,Atypicalapartmentbuildingusuallyhassomefireprotectionmeasures,suchasfireresistantconstructionandfirealarms.Additionalfireprotectionmeasureswouldlowertheriskfurther.ThisexampleconsidersthesamesixdifferentcombinationsofthreeadditionalfireprotectionmeasureswhichwereconsideredinthechecklistmethodinSection5.3.,Thethreeadditionalfireprotectionmeasuresare:(1)nosmokingmaterial(suchascigarettes)intheapartments,(2)sprinklersand(3)regularevacuationdrills.Eachofthethreefireprotectionmeasureshasanimpactoneithertheprobabilityoffireoccurrenceortheconsequenceofafireoccurrence.,Forexample,themeasureofnosmokingmaterialintheapartmentwouldhaveanimpactonloweringtheprobabilityoffireoccurrence;whereasthemeasuresofsprinklersandregularevacuationdrillswouldhaveanimpactonloweringtheconsequenceofafireoccurrencebysuppressingorcontrollingthefireorbyallowingtheoccupantstoevacuatemorequickly.,Aswasinthediscussionofthechecklistmethod,theeventtreeinFigure5.1isonlyanexampletoshowhowaneventtreecanbeusedforquantitativefireriskassessment.Thedescriptionofeacheventisthejudgmentforthisexampleonly.Otherapplicationsmayhavedifferentdescriptions.Thedescriptionsallowmoretransparentdiscussionsandagreementsamongstakeholders.,Inanevent-treemethod,theprobabilityofeachfirescenarioiscalculatedusingtheprobabilityvaluesofsuccessorfailureofimplementingthefireprotectionmeasuresthatareassociatedwiththescenario.Someoftheseprobabilityvaluescanbeobtainedfromstatistics,iftheyareavailable.,Forexample,NFPAstatisticsshowthat,basedonthe19992002data,sprinklersinapartmentbuildingshaveareliabilityof96%ofactivatingandcontrollinglargefiresthatshouldactivatesprinklers(KimberlyandHall,2005).Notethatsmoulderingfiresandsmallfiresmaynotactivatesprinklers.,Ifnosuchinformationisavailable,thensubjectivejudgmentmayberequired.Forexample,thereisnostatisticalinformationthatcanbeeasilyfoundontheprobabilityofsuccessorfailureofimplementinganosmokingmaterialplansothattherewillbealowerrateoffireoccurrence.Withoutsuchstatisticalinformation,wehavetomakeanassumption.Letusassumeforthisexamplethattheprobabilityofsuccessofimplementinganosmokingmaterialplanis70%.,Similarly,thereisnostatisticalinformationthatcanbeeasilyfoundontheprobabilityofsuccessorfailureofimplementingaregularevacuationdrillsplaninapartmentbuildingssothattheoccupantswouldknowwhattodoincaseofafirealarmandwouldthereforeevacuatemorequicklythanwithoutsuchdrills.Letusassumeforthisexamplethattheprobabilityofsuccessofimplementingaregularevacuationdrillsplanis80%.,Inrealriskassessments,thesevaluesneedtobecarefullyanalysedandagreeduponbyfiresafetyengineersandregulators.Successisdefinedasthatthefireprotectionplanactuallyworks.,ThescenarioprobabilitiesareshowninFigure5.1.Forexample,ScenarioAhasaprobabilityof2.401003,whichistheproductof0.30(failureprobabilityofimplementinganosmokingmaterialplan)0.04(failureprobabilityofsprinklersystem)0.20(failureprobabilityofimplementingaregularevacuationdrillsplan).,Intheevent-treemethod,theprobabilityoffireoccurrenceforeachfirescenarioisassessedbasedontheinherentrateoffireoccurrenceandtheimpactofvariousfirepreventionmeasurestominimizethisinherentrateoffireoccurrence.InFigure5.1,theimpactofeachofthefireprotectionmeasuresontheinherentrateoffireoccurrenceisassessedusingaresidualprobabilitymultiplier.,在事件树方法中，每一个火灾场景出现的可能性基于火灾发生的固有频率和各种火灾安全措施对降低火灾发生固有频率的影响来确定。如图5.1，每一种火灾保护措施对火灾发生固有频率的影响通过一个残余可能性因子来体现（residualprobabilitymultiplier）。,Thisallowsthefireprotectionengineersandregulatorstoassesstheimpactofthesefireprotectionmeasuresbasedontheirassessmentsofthereductionoftheprobabilityoffireoccurrence.Someoftheseresidualprobabilitymultiplierscanbeobtainedfromstatistics,iftheyareavailable.Ifnosuchinformationisavailable,thensubjectivejudgmentmayberequired.,火灾安全工程师基于火灾安全措施对火灾发生可能性的减缩情况来确定火灾安全措施的影响。某些残余可能性因子可由统计数据获得，如果没有可用的统计数据，则需要进行主观判断。,比如没有“在公寓中使用阻燃材料”对减少火灾发生影响的统计数据。我们只能对此进行假设，在本例中假设“在公寓中使用阻燃材料”与“在建筑中禁烟”对火灾发生可能性的影响相同，其残余可能性因子为0.88(见Table5.2)。即在“在公寓中使用阻燃材料”成功，其结果是将火灾发生的频率下降到其固有频率的0.88。,Forexample,thereisnostatisticalinformationthatcanbeeasilyfoundonthereductionoffireoccurrenceofimplementinganosmokingmaterialplanforapartmentbuildings.Withoutsuchstatisticalinformation,wehavetomakeanassumption.Letusassumeforthisexamplethattheresidualprobabilitymultiplierofanosmokingmaterialplaninapartmentbuildingis0.88,thesameasthatforhousefires(seeTable5.2).,Thatis,theconsequenceofasuccessfulnosmokingmaterialplanisthereductionoftherateoffireoccurrenceto0.88ofitsinherentvalue.,ThisresidualprobabilitymultiplierforeachfirescenarioisshowninFigure5.1.Forexample,ScenariosA,B,C,D,allwithafailureofimplementingthenosmokingmaterialplan,havearesidualprobabilitymultiplierof1(noreduction);whereasScenarioE,F,G,H,allwithasuccessofimplementingthenosmokingmaterialplan,havearesidualprobabilitymultiplierof0.88.,每一个火灾场景的这个残余可能性因子在图5.1中显示出来。比如ScenariosA,B,C,D，“在公寓中使用阻燃材料”全部失败，残余可能性因子为1，意味着没有减少火灾的发生；而ScenarioE,F,G,H,“在公寓中使用阻燃材料”成功，具有的残余可能性因子为0.88，即火灾发生的可能性，降低到固有值的0.88。,Thatis,theconsequenceofanosmokingmaterialplanisthereductionoftherateoffireoccurrenceto0.88ofitsinherentvalue.,Alsoinanevent-treemethod,theconsequenceofeachfirescenarioisassessedbasedontheinherentconsequenceofthefireandtheimpactofthevariousfireprotectionmeasurestominimizetheconsequence.InFigure5.1,theimpactofeachofthefireprotectionmeasuresontheconsequenceisassessedusingaresidualconsequencemultiplier.,Thisallowsthefireprotectionengineersandregulatorstoassesstheimpactofthesefireprotectionmeasuresbasedontheirassessmentsofthereductionoftheconsequence.Someoftheseresidualconsequencemultiplierscanbeobtainedfromstatistics,iftheyareavailable.,Forexample,NFPAstatisticsshowthat,basedonthe19891998data,thereductionindeathsinapartmentbuildingswithsprinklersis81%whencomparedwithsimilarbuildingswithoutsprinklers(KimberlyandHall,2005).Theresidualconsequencemultiplierofasprinklersystemthereforeis0.19.Thatis,theconsequenceofinstallingasprinklersystemisthereductionofthedeathrateperfireto0.19ofitsinherentvalue.,Ifnosuchinformationisavailable,thensubjectivejudgmentmayberequired.Forexample,thereisnostatisticalinformationthatcanbeeasilyfoundonthedeathreductionbenefitofimplementingaregularevacuationdrillsplan.Withoutsuchstatisticalinformation,wehavetomakeanassumptionagain.,Letusassumeforthisexamplethattheresidualconsequencemultiplierofaregularevacuationdrillsplanis0.40.Thatis,theconsequenceofaregularevacuationdrillsplanisthereductionofthedeathrateperfireto0.40ofitsinherentvalue.,ThisresidualconsequencemultiplierforeachscenarioisshowninFigure5.1.Forexample,ScenarioBhasaresidualconsequencemultiplierof0.40,whichistheproductof1.00(residualconsequencemultiplierofafailedsprinklersystem)0.40(residualconsequencemultiplierofasuccessfulregularevacuationdrillsplan).,Figure5.1showstheresidualriskvaluesofallthefirescenarioswhicharebasedonthesuccessorfailureofthreefireprotectionmeasures.Theprobabilityvalueofeachfirescenarioistheproductoftheindividualprobabilityvaluesofallthebranchesthatareassociatedwiththatscenario.,Theresidualprobabilitymultiplierofeachscenarioistheproductoftheindividualprobabilitymultipliersofallthefireprotectionmeasuresthatareassociatedwiththatscenario.Theresidualconsequencemultiplierofeachscenarioistheproductoftheindividualconsequencemultipliersofallthefireprotectionmeasuresthatareassociatedwiththatscenario.,Finally,theresidualriskmultiplierforeachscenarioistheproductof(scenarioprobability)(residualprobabilitymultiplier)(residualconsequencemultiplier).Forexample,ScenarioEhasascenarioresidualriskmultiplierof4.931003,whichistheproductof5.601003(scenarioprobability)0.88(residualprobabilitymultiplier)1.00(residualconsequencemultiplier).,InFigure5.1,themultiplicationoftheresidualmultipliersisbasedontheargument,asdiscussedinSection5.3,thateachfireprotectionmeasurereducestherateoffireoccurrence,ortheseverityofthefire,orthedeathrateperfire,insuccessionbyacertainpercentage.,Itshouldbeemphasizedagainthatthevaluesintheexampleareselectedbytheauthorasanexampletoshowhowsuchevent-treemethodcanbecarriedout.Inrealriskassessments