水库泥沙淤积的不确定因素分析毕业设计外文翻译Word文件下载.docx
《水库泥沙淤积的不确定因素分析毕业设计外文翻译Word文件下载.docx》由会员分享,可在线阅读,更多相关《水库泥沙淤积的不确定因素分析毕业设计外文翻译Word文件下载.docx(7页珍藏版)》请在冰豆网上搜索。
Abstract:
Significantadvanceshavebeenmadeinunderstandingtheimportanceofthefactorsinvolvedinreservoirsedimentation.However,predictingtheaccumulationofsedimentinareservoirisstillacomplexproblem.Inestimatingreservoirsedimentationandaccumulation,anumberofuncertaintiesarise.Thesearerelatedtoquantityofstreamflow,sedimentload,sedimentparticlesize,andspecificweight,trapefficiency,andreservoiroperation。
Inthisstudy,MonteCarlosimulationandLatinhypercubesamplingareusedtoquantifytheuncertaintyofannualreservoirsedimentationandaccumulatedreservoirsedimentationthroughtime.Inaddition,sensitivityanalysiswasperformedtoexaminetheimportanceofvariousfactorsontheuncertaintyofannualreservoirsedimentation.TheproposedprocedureshavebeenappliedtotheKennyReservoirattheWhiteRiverBasininColorado.Theuncertaintyofannualreservoirsedimentationandtheeffectofeachuncertainfactor,takenindividuallyandincombinations,ontheuncertaintyofaccumulatedreservoirsedimentationthroughtimehavebeenexamined.Theresultsshowthatannualstreamflowandsedimentloadarethemostimportantfactorsdeterminingthevariabilityofannualreservoirsedimentationandaccumulatedreservoirsedimentation.InthecaseofKennyReservoir,theuncertaintyexpressedbythecoefficientofvariationcanbeontheorderof65%forannualreservoirsedimentationand39%foraccumulatedreservoirsedimentationvolume.
Introduction
Reservoirsedimentationvarieswithseveralfactorssuchassedimentproduction,sedimenttransportationrate,sedimenttype,modeofsedimentdeposition,reservoiroperation,reservoirgeometry,andstreamflowvariability.Sedimentistransportedassuspendedandbedloadsbystreamsandriverscomingintoareservoir.Duetoflowdecelerationwhenariverapproachesareservoir,thesedimenttransportcapacitydecreases,andsomeoftheincomingsedimentistrappedanddepositedinthereservoir.Inaddition,thedepositedsedimentsmayconsolidatebytheirweightandtheweightofoverlyingwaterthroughtime.Predictingthesedimentcomingintoareservoir,itsdeposition,anditsaccumulationthroughouttheyears,afterconstructionofthedam,havebeenimportantproblemsinhydraulicengineering.Despitetheadvancesmadeinunderstandingseveralofthefactorsinvolvedinreservoirsedimentation,predictingtheaccumulationofsedimentinareservoirisstillacomplexproblem.Empiricalmodels,basedonsurveysandfieldobservations,havebeendevelopedandappliedtoestimateannualreservoirsedimentationload(RSL),accumulatedreservoirsedimentationload,(ARSL),andaccumulatedreservoirsedimentationvolume(ARSV)afteragivennumberofyearsofreservoiroperation.Likewise,severalmathematicalmodelsforpredictingreservoirsedimentationhavebeendevelopedbasedontheequationsofmotionandcontinuityforwaterandsediment.However,empiricalmethodsarestillwidelyusedinactualengineeringpractice.
Inestimatingresevoirsedimentinflow,reservoirsedimentation,andreservoirsedimentaccumulation,eitherbyempiricaloranalyticalapproaches,anumberofuncertaintiesarises.Themainfactorsaffectingreservoirsedimentationare
(1)quantityofstreamflow;
(2)quantityofsedimentinflowintoareservoir;
(3)sedimentparticlesize;
(4)specificweightofthedeposits;
and(5)reservoirsizeandoperation.Dependingontheparticularcaseathand,somefactorsmaybemoreimportantthanothers.Allofthesefactorsareuncertaintosomedegreeand,asaconsequence,reservoirsedimentationwillbeanuncertainquantitytoo.Inaddition,whichmodel(orprocedure)isapplicabletoestimatesomeoftheforegoingquantitiesand,infact,whichmodelistobeusedtoestimatetheamountofsedimentthatwillbetrappedinareservoirarequestionsthatcannotbeansweredwithcertainty.Forinstance,Fan(1988)obtainedinformationon34stream-,18watershed-,and20reservoir-sedimentationmodelsandstatedthatdifferentmodelsmaygivesignificantlydifferentresultsevenwhenusingthesamesetofinputdata.Suchanadditionalfactor,knownas‘‘modeluncertainty,’’maybequitealargecomponentoftheoveralluncertainty.Inanycase,theplannerandmanagerofareservoirmaybeinterestedinquantifyinghowtheuncertaintyofsomeofthefactorsaffectingreservoirsedimentationtranslateintotheuncertaintyofannualsedimentdepositionandaccumulatedsedimentdepositionthroughtime.Inthispaper,weaddresstheissuequantifyingtheeffectofparameteruncertaintyonreservoirsedimentationbasedonasetofpredefinedmodelsaswillbedescribedbelow.Theeffectofmodeluncertaintyisnotconsideredinthisstudy.
Severalmethodsofuncertaintyanalysishavebeendevelopedandappliedinwaterresourcesengineering.Themostwidelyusedmethodsarefirst-orderanalysis(FOA)andMonteCarlosimulation(MCS).FOAisbasedonlinearizingthefunctionalrelationshipthatrelatesadependentrandomvariableandasetofindependentrandomvariablesbyTaylorseriesexpansion.Thismethodhasbeenappliedinseveralwaterresourcesandenvironmentalengineeringproblemsinvolvinguncertainty.Examplesincludestormsewerdesign;
ground-water-flowestimation,predictionofdissolvedoxygen;
andsubsurface-flowandcontaminanttransportestimation.InMCS,stochasticinputsaregeneratedfromtheirprobabilitydistributionsandarethenenteredintoempiricaloranalyticalmodelsoftheunderlyingphysicalprocessinvolvedingeneratingstochasticoutputs.Then,thegeneratedoutputsareanalyzedstatisticallytoquantifytheuncertaintyoftheoutput.ManyexamplesofuncertaintyanalysisbyMCScanbefoundinwaterresourcesandenvironmentalengineering.Someexamplesincludesteady-stateground-water-flowestimationandwater-qualitymodeling.Scaviaetal.(1981)madeacomparisonofMCSandFOAfordetermininguncertaintiesassociatedwitheutrophicationmodeloutputssuchasphytoplankton,zooplankton,andnitrogenforms.TheyindicatedthatbothMCSandFOAagreewellinestimatingthemeanandvarianceofmodelestimates.However,MCShastheadvantageofprovidingbetterinformationabouttheoutputfrequencydistribution.
Latinhypercubesampling(LHS)isanalternativesimulationprocedurethathasbeendevelopedforuncertaintyanalysisofphysicalandengineeringsystems.ThebasicideabehindLHSistogeneraterandomstochasticinputsinastratifiedmannerfromtheprobabilitydistributions.InthiswaythenumberofgeneratedinputscanbereducedconsiderablyascomparedtoMCS.TheypointedoutthatthepointestimatemethodyieldsalargermeanandvariancethanthoseobtainedbytheFOAandLHSmethods.Furthermore,instudyingtheimportanceofstochasticinputsontheoutputbysensitivityanalysis,LHSyieldsmoreinformationthantheothertwomethods.
Inthisstudy,uncertaintyanalysisbasedonMCSandLHSmethodsareconductedtoestimatetheprobabilitydistributionofannualreservoirsedimentationvolume(RSV).Inaddition,sensitivityanalysisisperformedtoseetherelativeimportanceofstochasticinputsinestimatingthevariabilityofRSV.Furthermore,uncertaintyanalysisofARSVthroughouttimeisperformedusingMCS.Inthisprocedure,annualstreamflowsaregeneratedbyastochastictimeseriesmodel.Theeffectofparameteruncertaintyinthestochasticmodelontheoutput(i.e.,ARSV)isalsoconsidered.
EstimationOfAnnualAndAccumulated
ReservoirSedimentLoad(Mass)AndVolume
Reservoirsedimentationvolumedepends,amongotherfactors,onthequantityofsedimentinflow,thepercentageofsedimentinflowtrappedbythereservoir,andthespecificweightofthedepositedsedimentconsideringtheeffectofcompactionwithtime.Theincomingsedimentloadandthestreamflowdischargeareusuallymeasuredathydrometricgaugingstations,andasedimentratingcurveisconstructed.Thesedimentratingcurveexpressestherelationshipbetweentherateofsedimentdischargeandtherateofstreamflowdischargeandisusuallyrepresentedgraphicallyonlogarithmiccoordinates.Incomingsedimentisgenerallycomposedofsuspendedsedimentandbedload.Whenthebedloadcannotbeobtainedbymeasurements,itcanbeestimatedbyformulas.
Inestimatingannualsedimentload,ithasbeencommonpracticetouseannualsedimentratingcurvesforbothsuspendedsedimentandbedload.Theannualsedimentratingcurveistherelationbetweenannualsedimentloadandannualstreamflowdischarge.Twomethodscanbeconsideredfordeterminingannualsedimentratingcurves.Asimplemethodinvolvesthefollowingsteps:
(1)Foragivenyearcalculatedailysedimentloadsfromdailysedimentratingcurves;
(2)addalldailysedimentloadsanddividethesumbythenumberofdaysintheyear,thenthisvaluerepresentstheannualaveragesedimentloadintonsperday;
(3)repeatSteps1and2forallyearsofrecord;
and(4)plottheannualaveragesedimentloadversustheannualaveragestreamflowforeachyearintherecord.Analternativemethodisbasedonestimatingannualsedimentloadsusingflowdurationcurves.Inanycase,anannualsedimentratingcurvecanbeconstructedbysimpleregressionanalysisafterlogarithmictransformationofannualaveragestreamflowdischargesandannualaveragesedimentloads.Colby(1956)statedthatinactualpracticedailysedimentratingcurvescouldbeassumedtobeequivalenttoinstantaneoussedimentratingcurves.
Dailyratingcurvesofsuspendedsedimentandbedloadmayberepresentedas
(1)
(2)
whereQSD=dailysuspendedsedimentload(tons/day);
QBD=dailybedload(tons/day);
QWD=dailyaveragestreamflowd