分销网络设计中的库存选址研究

华中科技大学博士学位论文分销网络设计中的库存-选址研究姓名：唐凯申请学位级别：博士专业：管理科学与工程指导教师：杨超20081025I————IIIIIAbstractDistributionreferstotheapproacheswhichtaketheproductsfromthesupplierstagetothecustomerstageinthesupplychain.InSupplyChainManagement(SCM),DistributionNetworkDesign(DND)hasalwaysbeenakeystrategicissueofanenterprisewithextensiveresearch.ThedistributionbusinessoftheproductsisoperatedandachievedbythedistributionnetworkthereforetheperformanceoftheenterpriseDNDhasadirectimpactonthetotallogisticcostandserviceleveloftheenterprise.Ingenerally,DNDincludesthreedifferentdecisionlevels:strategicdecisions(facilitieslocation),tacticaldecisions(inventorypolicy)andoperationaldecisions(transportation).TheDNDproblemistypicallysolvedfollowingasequentialapproach,inwhichtactical/operationaldecisionsaresimplifiedoromitted,andoncetheDNDissolved,thentheseomitteddecisionsmaybeincorporated.AwidefamilyofFacilityLocationProblems(FLP)solutionsisusedtosolvetheDND,whereparameters,demands,constraintsandtheobjectivefunctionareconsideredtobedeterministic.However,theFLPasstrategicentitiesarenotcapableofdealingwiththeinteractionsbetweenfacilitylocationandotherlevels’decisions,suchasinventorycontroldecisions.Thus,inordertomeetcustomerdemandswithlowesttotalcost,DNDshouldbefromtheperspectiveofoverallsituationoftheenterprises,tointegrateoneormoreofinventorycontrol,facilitieslocationandotherdecisionsintoasystemtooptimize,basingontheideaofsystematizationandintegration.Andfordesigningoptimaldistributionnetwork,DNDshouldalsoweightsyntheticallytransportation,production,inventory,facilitieslocationandotherrelatingcoststominimizethetotalsystemcost.Therefore,myresearchandapplicationoffacilitylocationproblemintegratedinventorymanagementand/orotherdecisionsarepresentedanddiscussedinthisthesisbasedonthetheoryofthisfield.Firstly,thethesisanalysesbackgroundmotivation,introducesthereasonsofchoosingthistopic,andproposestheresearchobjectionandresearchroute.Thenthethesisintroducesrelevantconceptsandcharacterofthedistributionnetwork,anddiscussesthefactorswhichwouldimpactDNDandthecomponentsofthecostofdistributionnetwork.IVMoreover,thethesisreviewstheorigination,evolutionandtheoriesoftraditionalFLP.Basedonthese,thethesissummarizesfurthertheresearchresultsofFLPappliedinDNDinthelasttenyear,andtheliteraturesinvolvingInventory-LocationModel(ILM),whichisthecoreissueinthethesis.Secondly,thethesisintroducesthebasicSingle-stageILM,andanalysesthepropertiesofthismodelwhicharedifferentfromthetraditionalFLP.Then,thethesisdevelopsajointinventory-locationmodel,whichconsidersthemarketselectiondecisions.ThismodelisformulatedasanonlinearintegerprogrammingmodelthatisNP-hard.Therefore,aheuristicsolutionapproachispresentedbasedonLagrangianRelaxationandGreedAlgorithm.Moreover,numericalexamplesaregiventoverifytheefficiencyofthisalgorithm.Thirdly,inordertodesignthedistributionnetworktocopewithcontinuallychangingmarketenvironmentinalongstrategyhorizon,thethesisaddressestwomultistageinventory-locationmodels,oneofwhichconsidersthecapitalbudget,andtheotherisnot.Inthesemodels,explicitlyhandlesparameteruncertaintybyallowingparameterstobedescribedbydiscretescenarios,eachwithaspecifiedprobabilityofoccurrence.Forsolvingthesemodels,aheuristicsolutionapproachisalsopresentedbasedonLagrangianRelaxationandGreedAlgorithm.Moreover,numericalexamplesaregiventoverifytheefficiencyofthisalgorithm.Fourthly,thethesisstudiesaninventory-locationmodelunderinflation.Thegrowinginflationratesathomeandabroadhavethetremendousimpactontheworldeconomyandthehuman’sdailylives,thusforsubsistenceanddevelopmentoftheenterprises,itiscriticaltodesignoptimaldistributionnetworkwhichshouldtacklehighinflation.Threegeneticalgorithmsarepresentedforsolvingourmodel,basedondifferentcodingforms,andtheeffectivenessofthesealgorithmsisanalyzedthroughdifferentsizesofinstances.Theresultsoftheanalysisshowthat,forminimizingthepresentvalueofthetotalcostunderthegrowinginflationrate,theoptimalpolicywouldneedtoincreasethenumberofthebuiltdistributioncenters.Fifthly,thethesisstudiesthecontentsandcharactersofthedistributionnetworkfordeterioratingitems.Thenfromtheperspectiveofinventorymodels,thethesisdevelopsanVinventory-locationmodelfordeterioratingitemswhichassumesthedeterioratingrateisaknownconstant.Twoalgorithmsarepresented:LagrangianRelaxationAlgorithmandGeneticAlgorithm.Moreover,thethesisdevelopsamorerealisticbutmorecomplexproduct-inventory-locationmodel,anddesignsanadaptionofGeneticAlgorithmtosolveit.Furthermore,numericalexamplesaregiventoillustratetheapplicationofthealgorithm,andshowthatGAisanefficientsolutionapproachforthisproblem.Finally,thethesisgivesconclusiontothecontentsandtheinnovativeachievementsoftheresearch,andpresentsthefuturescope,thepurposeandtheprospectofthistopicinfurtherstudies.Keywords:DistributionNetwork;Inventory;Location;DeterioratingItems;LagrangianRelaxation;GeneticAlgorithm111.11.1.1SupplyChainManagement,SCMBeamon1998SCMThomasGriffin199620“”“”20032006DistributionNetworkDesignDNDDND12AmbrosinoScutella20052DNDDNDDNDCrainicLaporte1997DNDDNDDNDShu