ebook img

Importance ranking of parameter uncertainties in geo-hazard assessments PDF

191 Pages·2016·8.67 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Importance ranking of parameter uncertainties in geo-hazard assessments

Importance ranking of parameter uncertainties in geo-hazard assessments Jeremy Rohmer To cite this version: Jeremy Rohmer. Importance ranking of parameter uncertainties in geo-hazard assessments. Modeling and Simulation. Universit´e de Lorraine, 2015. English. <tel-01319716> HAL Id: tel-01319716 https://hal-brgm.archives-ouvertes.fr/tel-01319716 Submitted on 22 May 2016 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destin´ee au d´epˆot et `a la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publi´es ou non, lished or not. The documents may come from ´emanant des ´etablissements d’enseignement et de teaching and research institutions in France or recherche fran¸cais ou ´etrangers, des laboratoires abroad, or from public or private research centers. publics ou priv´es. AVERTISSEMENT Ce document est le fruit d'un long travail approuvé par le jury de soutenance et mis à disposition de l'ensemble de la communauté universitaire élargie. Il est soumis à la propriété intellectuelle de l'auteur. Ceci implique une obligation de citation et de référencement lors de l’utilisation de ce document. D'autre part, toute contrefaçon, plagiat, reproduction illicite encourt une poursuite pénale. Contact : [email protected] LIENS Code de la Propriété Intellectuelle. articles L 122. 4 Code de la Propriété Intellectuelle. articles L 335.2- L 335.10 http://www.cfcopies.com/V2/leg/leg_droi.php http://www.culture.gouv.fr/culture/infos-pratiques/droits/protection.htm UNIVERSITE DE LORRAINE Ecole Nationale Supérieure des Mines de Nancy Laboratoire GeoRessources Ecole Doctorale RP2E THESE Présentée en vue du grade de DOCTEUR DE L’UNIVERSITE DE LORRAINE en Génie Civil-Hydrosystèmes-Géotechnique Par Jérémy ROHMER Importance ranking of parameter uncertainties in geo-hazard assessments Analyse de sensibilité des incertitudes paramétriques dans les évaluations d’aléas géotechniques le 16 Novembre 2015 Devant le jury composé de Michael OBERGUGGENBERGER– University of Innsbruck – Austria Rapporteur Michael BEER – University of Liverpool – UK Rapporteur Anne GEGOUT-PETIT – Université de Lorraine – France Examinateur Gilles GRANDJEAN – BRGM – France Examinateur Thierry VERDEL – Université de Lorraine – France Directeur de thèse Jack-Pierre PIGUET – Université de Lorraine – France Co-directeur de thèse Acknowledgements IthankT.VerdelandJ.-P.Piguet(Uni.Lorraine)foracceptingtosupervisethisPhDthesis.I amverygratefultomyco-authors,C.Baudrit(INRA),E.Foerster(CEA),A.Nachbaur(BRGM) andT.Verdel(Uni.Lorraine),fortheconstructiveworkandfruitfuldiscussionswhichledto thepublicationssupportingthepresentmanuscript. IalsothanktheBRGMdirectorateof"RisksandPrevention"division(J.L.Foucher,H.Fabriol andG.Grandjean)forsupportingmy"personalandprofessional"project. IamalsogratefultoD.Guyonnet(BRGM/ENAG)andG.Grandjean(BRGM)forintroducing metoFuzzysetsandnewuncertaintytheoriesandO.Sedan(BRGM)forhisveryvaluable adviceonpractical/operationalaspectsofnaturalriskassessments. Lastbutnotleast,Iamverygratefultomyfamily(Julia,myparents,myparentsinlawand mysisterinlaw)forsupportingme.SpecialthankstotheFrenchtraincompanySNCFforits repetitivedelays,whichletmeenoughtimeforwritingthismanuscript. Orleans,16November2015 J.Rohmer. i Technical Summary Thecentraltopicofthepresentthesisisthetreatmentofepistemicuncertaintyingeo-hazard assessments(likelandslide,earthquake,etc.).Contrarytoaleatoryuncertainty(akarandom- ness,variability),epistemicuncertaintycanbereducedthroughadditionalmeasurements (labtests,insiteexperiments,etc.)ormodelling(e.g.,throughnumericalsimulations)orextra researchefforts.Amongthedifferenttypesofepistemicuncertainties,wefocusedhereonthe parametricone:thiscorrespondstotheincompleteknowledgeofthecorrectsettingofthe inputparameters(likevaluesofsoilproperties)ofthemodelsupportingthegeo-hazardas- sessment.Apossibleoptiontomanagethistypeofuncertaintyisthroughsensitivityanalysis: 1.identifythecontributionofthedifferentinputparametersintheuncertaintyonthefinal hazardoutcome;2.rankthemintermsofimportance;3.decideaccordinglytheallocationof additionalcharacterisationstudies. Forthispurpose,variance-basedglobalsensitivityanalysis(VBSA)isapowerfulprocedure, whichallows:i.incorporatingtheeffectoftherangeoftheinputvariationandofthenatureof theprobabilitydistribution(normal,uniform,etc.);ii.exploringthesensitivityoverthewhole rangeofvariation(i.e.inaglobalmanner)oftheinputrandomvariables;iii.fullyaccounting forpossibleinteractionsamongthem;andiv.providingaquantitativemeasureofsensitivity withoutintroducingaprioriassumptionsonthemodel’smathematicalstructure(i.e.model- free).Themostimportantsourcesofparameteruncertaintycanthenbeidentified(usingthe maineffects)aswellastheparametersofnegligibleinfluence(usingthetotaleffects).Besides, somekeyattributesofthemodelbehaviourcanbeidentified(usingthesumofthemain effects).Yet,totheauthor’sbestknowledge,thiskindofanalysishasrarelybeenconducted inthedomainofgeo-hazardassessments. Thiscanbeexplainedbythespecificitiesofthe domainofgeo-hazardassessments,whichimposeconsideringseveralconstraints,whichare atthecoreofthepresentwork. Mostnumericalmodelssupportinggeo-hazardassessmentshavemoderate-to-highcompu- tationtime(typicallyseveralminutes,evenhours),eitherbecausetheyarelarge-scale(e.g., landslidesusceptibilityassessmentatthespatialscaleofavalley),orbecausetheunderlying processesaredifficulttobenumericallysolved(e.g.,complexelastoplasticrheologylawlike theHujeuxmodeldescribingthecomplexcoupledhydromechanicalbehaviourofaslipsur- iii TechnicalSummary face).Despitetheextensiveresearchworkontheoptimizationofthecomputationalgorithms, VBSAremainscomputationallyintensive,asitimposestorunalargenumberofsimulations(> 1,000).Inthiscontext,VBSAcanbemadepossibleviathecombinationwithmeta-modelling techniques. Thistechnique consistsinreplacingthe long-runningnumerical model by a mathematicalapproximationreferredtoas“meta-model”(alsonamed“responsesurface”, or“surrogatemodel”),whichcorrespondstoafunctionconstructedusingafewcomputer experiments(typically50-100,i.e.alimitednumberoftimeconsumingsimulations),andaims atreproducingthebehaviourofthe“true”modelinthedomainofmodelinputparameters andatpredictingthemodelresponseswithanegligiblecomputationtimecost. TheapplicabilityofthecombinationVBSAandmeta-modelswasdemonstratedusingthe model developed by Laloui and co-authors at EPFL (Lausanne) for studying the Swiss La Frasse landslide. We focused on the sensitivity of the surface displacements to the seven parametersoftheHujeuxlawassignedtotheslipsurface.Inthiscase,asinglesimulationtook 4daysofcalculation. Ontheotherhand,evaluatingthemaineffects(firstordersensitivity indices)shouldrequire>1,000differentsimulations,whichisherehardlyfeasibleusingthe numerical simulator. Thiscomputationburden was alleviated using akriging-type meta- modelconstructedusing30differentsimulations.Furthermore,theimpactofthemeta-model error(i.e.theadditionaluncertaintyintroducedbecausethetruesimulatorwasreplacedby anapproximation)wasdiscussedbytreatingtheproblemundertheBayesianformalism.This allowedassigningconfidenceintervalstothederivedsensitivitymeasures:theimportance rankingcouldthenbedoneaccountingforthelimitedknowledgeonthe“true”simulator(i.e. throughonly30differentlong-runningsimulations),henceincreasingtheconfidenceinthe analysis.Totheauthor’sbestknowledge,theapplicationofsuchkindsoftechniqueisoriginal inthedomainoflandslideriskassessment. ThesecondlimitationofVBSAisrelatedtothenatureoftheparameters(inputoroutput): theyarescalar. Yet,inthedomainofgeo-hazard,parametersareoftenfunctional,i.e. they arecomplexfunctionsoftimeorspace(orboth).Thismeansthatparameterscanbevectors withpossiblehighdimension(typically100-1,000).Forinstance,theoutputsoftheLaFrasse modelcorrespondtotemporalcurvesofthedisplacements(discreditedin300steps)atany nodesofthemesh,i.e. theoutputsarevectorsofsize300atanyspatiallocation. Another exampleisthespatialdistributionofhydraulicconductivitiesofasoilformation.Focusing firstonthefunctionaloutputcase,amethodologytocarryoutdynamic(global)sensitivity analysisoflandslidemodelswasdescribedcombining:1.basissetexpansiontoreducethedi- mensionalityofthefunctionalmodeloutput;2.extractionofthedominantmodesofvariation intheoverallstructureofthetemporalevolution;3.meta-modellingtechniquestoachievethe computation,usingalimitednumberofsimulations,ofthesensitivityindicesassociatedto eachofthemodesofvariation.Thesewereinterpretedbyadoptingtheperspectiveoftherisk iv TechnicalSummary practitionerinthefollowingfashion:“identifyingtheproperties,whichinfluencethemostthe possibleoccurrenceofadestabilizationphase(acceleration)overthewholetimeduration oronaparticulartimeinterval”.However,alimitationwasunderlined,namelythephysical interpretationofthedominantmodesofvariation,especiallycomparedtothetraditional time-varyingVBSA(moreeasilyinterpretable,butalsointractableforverylongtimeseries). Basedonthestudyonthefunctionaloutput,theapplicabilityoftheproposedmethodology wasalsoinvestigatedforthecaseoffunctionalinputsusingasanexample,arandomfield assignedtotheheterogeneousporosityofasoilformation. Finally, a third limitation of VBSA is related to the representation of uncertainty. By con- struction,VBSAreliesontools/proceduresoftheprobabilisticsetting.Yet,inthedomainof geo-hazardassessments,thevalidityofthisapproachcanbedebatable,becausedataareoften scarce,incompleteorimprecise.Inthiscontext,themajorcriticismsavailableintheliterature againstthesystematicuseofprobabilityinsuchsituationswerereviewed.Onthisbasis,the useofaflexibleuncertaintyrepresentationtoolwasinvestigated,namelyFuzzySetstohandle differentsituationsofepistemicuncertainty.Foreachsituation,examplesofrealcasesinthe contextofgeo-hazardassessmentswereused: • Vagueness due to the use of qualitative statements. The application of Fuzzy Sets wasillustratedinthecontextofsusceptibilityassessmentofabandonedunderground structures.Inparticular,itisshownhowtheso-called“thresholdeffect”canbehandled whentheexpertdefinesclassesofhazard/susceptibility; • Reasoningwithvagueconcepts.ThisishandledusingFuzzyLogic.Thisisillustrated withthetreatmentofimprecisionassociatedtotheinventoryofassetsatriskinthe contextofseismicriskanalysis; • Imprecision.Thisishandledbydefiningpossibilitydistributions,whichhaveastrong linkwithFuzzySets. Thisisillustratedwiththerepresentationofuncertaintyonthe amplificationfactoroflithologicalsiteeffectsinthecontextofseismicriskanalysis; • Imprecisionontheparametersofaprobabilisticmodel.Thisishandledinthesetting offuzzyrandomvariables.Thisisillustratedusingaprobabilisticdamageassessment modelinthecontextofseismicriskanalysis,namelytheRISK-UE,level1model. Onthisbasis,theissueofsensitivityanalysisconsideringamixtureofrandomnessandimpre- cisionwasaddressed.Basedonaliteraturereview,amajorlimitationwasoutlined,namely thecomputationtimecost:newtoolsforuncertaintyrepresentationbasicallyrelyoninterval- valuedtools,theuncertaintypropagationtheninvolvesoptimisationprocedure,whichcan behighlycomputationallyintensive. Inthiscontext,anadaptationofthecontributionto failureprobabilityplotof[LiandLu,2013],tobothhandleprobabilisticandpossibilisticin- formationwasproposed.Theanalysiscanbeconductedinapost-processingstep,i.e.using onlythesamplesofrandomintervalsandofrandomnumbersnecessaryforthepropagation v TechnicalSummary phase, henceatnoextracomputationalcost. Besides, itallowsplacingonthesamelevel randomandimpreciseparameters,i.e.itallowsthecomparisonoftheircontributioninthe probabilityoffailuresothatconcreteactionsfromariskmanagementperspectivecanbe decidedaccordingly.Theapplicabilityofthiseasy-to-usetoolwasdemonstratedusingreal cases,whereitisquestionabletouseprobabilitiestotreatuncertainty.Thefirstapplication casecorrespondstostabilityanalysisofsteepslopes.Themainimpreciseparametersinthis casearethetensioncrack’sheightlocatedintheupperpartofthecliffandthetoeheight.The secondonecorrespondstothestabilityanalysisofanabandonedundergroundquarry,where theextractionratiowasimprecisebecauseitcouldonlybeestimatedwithgreatdifficulties (duetotheparticulargeometryofthequarry).Finally,athirdexamplewasused,namelythe stabilityanalysisofaminepillarpresentingthinlayersofclay,whosepropertiesaredifficultto evaluateinpractice.Thislastexampleimposedtorelyonmeta-modellingtechniquestoease thejointuncertaintypropagationphaseusingthelong-runningmechanicalnumericalcode. Insummary, thepresentworkshouldbeseenasanefforttohandleepistemicparameter uncertaintiesingeo-hazardassessments.First,theachievementisofmethodologicalnature (methodologyforconductingVBSAusinglongrunningsimulators,methodologyforconduct- ingVBSAadaptedtofunctionaloutputs,methodologyforconductingsensitivityanalysiswhen bothimprecisionandrandomnessarepresent).Thismethodologicalworktakesadvantagesof therecentadvancesinthestatisticalcommunity(VBSA,basissetexpansion,FuzzySets,Fuzzy randomvariables,hybridpropagation,etc.)toanswerpracticalquestions(whatdrivestheun- certaintyontheresultsofthehazardassessment?Howtoconductmultiplesimulationswhen thesimulationcodetakesonehourtoberun?Howshouldtheuncertaintybetreatedwhen theonlypiecesofinformationavailablerestricttovaguestatementsandafewquantitative estimates?).Agreatattentionhasbeenpaidtoinvestigatetheapplicabilityofeachproposed technique/procedurei.e.byhighlightingtheprosandconsthroughtheconfrontationtoreal cases.Thisconstitutesthesecondachievementofthepresentwork. vi

Description:
Ce document est le fruit d'un long travail approuvé par le jury de Traditionally, rational decision-making under uncertainty is based on probabilities
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.