1、本科毕业论文（设计）外文翻译原文MULTIECHELONINVENTORYMANAGEMENTINSUPPLYCHAINSHISTORICALLY,THEECHELONSOFTHESUPPLYCHAIN,WAREHOUSE,DISTRIBUTORS,RETAILERS,ETC,HAVEBEENMANAGEDINDEPENDENTLY,BUFFEREDBYLARGEINVENTORIESINCREASINGCOMPETITIVEPRESSURESANDMARKETGLOBALIZATIONAREFORCINGFIRMSTODEVELOPSUPPLYCHAINSTHATCANQUICKLYRESP
2、ONDTOCUSTOMERNEEDSTOREMAINCOMPETITIVEANDDECREASEINVENTORY,THESEFIRMSMUSTUSEMULTIECHELONINVENTORYMANAGEMENTINTERACTIVELY,WHILEREDUCINGOPERATINGCOSTSANDIMPROVINGCUSTOMERSERVICESUPPLYCHAINMANAGEMENTSCMISANINTEGRATIVEAPPROACHFORPLANNINGANDCONTROLOFMATERIALSANDINFORMATIONFLOWSWITHSUPPLIERSANDCUSTOMERS,AS
3、WELLASBETWEENDIFFERENTFUNCTIONSWITHINACOMPANYTHISAREAHASDRAWNCONSIDERABLEATTENTIONINRECENTYEARSANDISSEENASATOOLTHATPROVIDESCOMPETITIVEPOWERSCMISASETOFAPPROACHESTOINTEGRATESUPPLIERS,MANUFACTURERS,WAREHOUSES,ANDSTORESEFFICIENTLY,SOTHATMERCHANDISEISPRODUCEDANDDISTRIBUTEDATRIGHTQUANTITIES,TOTHERIGHTLOCA
4、TIONSANDATTHERIGHTTIME,INORDERTOMINIMIZESYSTEMWIDECOSTSWHILESATISFYINGSERVICELEVELREQUIREMENTSSOTHESUPPLYCHAINCONSISTSOFVARIOUSMEMBERSORSTAGESASUPPLYCHAINISADYNAMIC,STOCHASTIC,ANDCOMPLEXSYSTEMTHATMIGHTINVOLVEHUNDREDSOFPARTICIPANTSINVENTORYUSUALLYREPRESENTSFROM20TO60PERCENTOFTHETOTALASSETSOFMANUFACTU
5、RINGFIRMSTHEREFORE,INVENTORYMANAGEMENTPOLICIESPROVECRITICALINDETERMININGTHEPROFITOFSUCHFIRMSINVENTORYMANAGEMENTIS,TOAGREATEREXTENT,RELEVANTWHENAWHOLESUPPLYCHAINSC,NAMELYANETWORKOFPROCUREMENT,TRANSFORMATION,ANDDELIVERINGFIRMS,ISCONSIDEREDINVENTORYMANAGEMENTISINDEEDAMAJORISSUEINSCM,IEANAPPROACHTHATADD
6、RESSESSCISSUESUNDERANINTEGRATEDPERSPECTIVEINVENTORIESEXISTTHROUGHOUTTHESCINVARIOUSFORMSFORVARIOUSREASONSTHELACKOFACOORDINATEDINVENTORYMANAGEMENTTHROUGHOUTTHESCOFTENCAUSESTHEBULLWHIPEFFECT,NAMELYANAMPLIFICATIONOFDEMANDVARIABILITYMOVINGTOWARDSTHEUPSTREAMSTAGESTHISCAUSESEXCESSIVEINVENTORYINVESTMENTS,LO
7、STREVENUES,MISGUIDEDCAPACITYPLANS,INEFFECTIVETRANSPORTATION,MISSEDPRODUCTIONSCHEDULES,ANDPOORCUSTOMERSERVICEMANYSCHOLARSHAVESTUDIEDTHESEPROBLEMS,ASWELLASEMPHASIZEDTHENEEDOFINTEGRATIONAMONGSCSTAGES,TOMAKETHECHAINEFFECTIVELYANDEFFICIENTLYSATISFYCUSTOMERREQUESTSEGREFERENCEBESIDETHEINTEGRATIONISSUE,UNCE
8、RTAINTYHASTOBEDEALTWITHINORDERTODEFINEANEFFECTIVESCINVENTORYPOLICYINADDITIONTOTHEUNCERTAINTYONSUPPLYEGLEADTIMESANDDEMAND,INFORMATIONDELAYSASSOCIATEDWITHTHEMANUFACTURINGANDDISTRIBUTIONPROCESSESCHARACTERIZESCSINVENTORYMANAGEMENTINMULTIECHELONSCSISANIMPORTANTISSUE,BECAUSETHEREAREMANYELEMENTSTHATHAVETOC
9、OORDINATEWITHEACHOTHERTHEYMUSTALSOARRANGETHEIRINVENTORIESTOCOORDINATETHEREAREMANYFACTORSTHATCOMPLICATESUCCESSFULINVENTORYMANAGEMENT,EGUNCERTAINDEMANDS,LEADTIMES,PRODUCTIONTIMES,PRODUCTPRICES,COSTS,ETC,ESPECIALLYTHEUNCERTAINTYINDEMANDANDLEADTIMESWHERETHEINVENTORYCANNOTBEMANAGEDBETWEENECHELONSOPTIMALL
10、YMOSTMANUFACTURINGENTERPRISESAREORGANIZEDINTONETWORKSOFMANUFACTURINGANDDISTRIBUTIONSITESTHATPROCURERAWMATERIAL,PROCESSTHEMINTOFINISHEDGOODS,ANDDISTRIBUTETHEFINISHGOODSTOCUSTOMERSTHETERMSMULTIECHELONORMULTILEVELPRODUCTION/DISTRIBUTIONNETWORKSAREALSOSYNONYMOUSWITHSUCHNETWORKSORSC,WHENANITEMMOVESTHROUG
11、HMORETHANONESTEPBEFOREREACHINGTHEFINALCUSTOMERINVENTORIESEXISTTHROUGHOUTTHESCINVARIOUSFORMSFORVARIOUSREASONSATANYMANUFACTURINGPOINT,THEYMAYEXISTASRAWMATERIALS,WORKINPROGRESS,ORFINISHEDGOODSTHEYEXISTATTHEDISTRIBUTIONWAREHOUSES,ANDTHEYEXISTINTRANSIT,ORINTHEPIPELINE,ONEACHPATHLINKINGTHESEFACILITIESMANU
12、FACTURERSPROCURERAWMATERIALFROMSUPPLIERSANDPROCESSTHEMINTOFINISHEDGOODS,SELLTHEFINISHEDGOODSTODISTRIBUTORS,ANDTHENTORETAILAND/ORCUSTOMERSWHENANITEMMOVESTHROUGHMORETHANONESTAGEBEFOREREACHINGTHEFINALCUSTOMER,ITFORMSAMULTIECHELONINVENTORYSYSTEMTHEECHELONSTOCKOFASTOCKPOINTEQUALSALLSTOCKATTHISSTOCKPOINT,
13、PLUSINTRANSITTOORONHANDATANYOFITSDOWNSTREAMSTOCKPOINTS,MINUSTHEBACKORDERSATITSDOWNSTREAMSTOCKPOINTSTHEANALYSISOFMULTIECHELONINVENTORYSYSTEMSTHATPERVADESTHEBUSINESSWORLDHASALONGHISTORYMULTIECHELONINVENTORYSYSTEMSAREWIDELYEMPLOYEDTODISTRIBUTEPRODUCTSTOCUSTOMERSOVEREXTENSIVEGEOGRAPHICALAREASGIVENTHEIMP
14、ORTANCEOFTHESESYSTEMS,MANYRESEARCHERSHAVESTUDIEDTHEIROPERATINGCHARACTERISTICSUNDERAVARIETYOFCONDITIONSANDASSUMPTIONSSINCETHEDEVELOPMENTOFTHEECONOMICORDERQUANTITYEOQFORMULABYHARRIS1913,RESEARCHERSANDPRACTITIONERSHAVEBEENACTIVELYCONCERNEDWITHTHEANALYSISANDMODELINGOFINVENTORYSYSTEMSUNDERDIFFERENTOPERAT
15、INGPARAMETERSANDMODELINGASSUMPTIONSRESEARCHONMULTIECHELONINVENTORYMODELSHASGAINEDIMPORTANCEOVERTHELASTDECADEMAINLYBECAUSEINTEGRATEDCONTROLOFSCSCONSISTINGOFSEVERALPROCESSINGANDDISTRIBUTIONSTAGESHASBECOMEFEASIBLETHROUGHMODERNINFORMATIONTECHNOLOGYCLARKANDSCARFWERETHEFIRSTTOSTUDYTHETWOECHELONINVENTORYMO
16、DELTHEYPROVEDTHEOPTIMALITYOFABASESTOCKPOLICYFORTHEPURESERIALINVENTORYSYSTEMANDDEVELOPEDANEFFICIENTDECOMPOSINGMETHODTOCOMPUTETHEOPTIMALBASESTOCKORDERINGPOLICYBESSLERANDVEINOTTEXTENDEDTHECLARKANDSCARFMODELTOINCLUDEGENERALARBORESCENTSTRUCTURESTHEDEPOTWAREHOUSEPROBLEMDESCRIBEDABOVEWASADDRESSEDBYEPPENAND
17、SCHRAGEWHOANALYZEDAMODELWITHASTOCKLESSCENTRALDEPOTTHEYDERIVEDACLOSEDFORMEXPRESSIONFORTHEORDERUPTOLEVELUNDERTHEEQUALFRACTILEALLOCATIONASSUMPTIONSEVERALAUTHORSHAVEALSOCONSIDEREDTHISPROBLEMINVARIOUSFORMSOWINGTOTHECOMPLEXITYANDINTRACTABILITYOFTHEMULTIECHELONPROBLEMHADLEYANDWHITINRECOMMENDTHEADOPTIONOFSI
18、NGLELOCATION,SINGLEECHELONMODELSFORTHEINVENTORYSYSTEMSSHERBROOKECONSIDEREDANORDERINGPOLICYOFATWOECHELONMODELFORWAREHOUSEANDRETAILERITISASSUMEDTHATSTOCKOUTSATTHERETAILERSARECOMPLETELYBACKLOGGEDALSO,SHERBROOKECONSTRUCTEDTHEMETRICMULTIECHELONTECHNIQUEFORCOVERABLEITEMCONTROLMODEL,WHICHIDENTIFIESTHESTOCK
19、LEVELSTHATMINIMIZETHEEXPECTEDNUMBEROFBACKORDERSATTHELOWERECHELONSUBJECTTOABUDGETCONSTRAINTTHISMODELISTHEFIRSTMULTIECHELONINVENTORYMODELFORMANAGINGTHEINVENTORYOFSERVICEPARTSTHEREAFTER,ALARGESETOFMODELSWHICHGENERALLYSEEKTOIDENTIFYOPTIMALLOTSIZESANDSAFETYSTOCKSINAMULTIECHELONFRAMEWORK,WEREPRODUCEDBYMAN
20、YRESEARCHERSINADDITIONTOANALYTICALMODELS,SIMULATIONMODELSHAVEALSOBEENDEVELOPEDTOCAPTURETHECOMPLEXINTERACTIONOFTHEMULTIECHELONINVENTORYPROBLEMSSOFARLITERATUREHASDEVOTEDMAJORATTENTIONTOTHEFORECASTINGOFLUMPYDEMAND,ANDTOTHEDEVELOPMENTOFSTOCKPOLICIESFORMULTIECHELONSCSINVENTORYCONTROLPOLICYFORMULTIECHELON
21、SYSTEMWITHSTOCHASTICDEMANDHASBEENAWIDELYRESEARCHEDAREAMORERECENTPAPERSHAVEBEENCOVEREDBYSILVERANDPYKETHEADVANTAGEOFCENTRALIZEDPLANNING,AVAILABLEINPERIODICREVIEWPOLICIES,CANBEOBTAINEDINCONTINUOUSREVIEWPOLICIES,BYDEFININGTHEREORDERLEVELSOFDIFFERENTSTAGES,INTERMSOFECHELONSTOCKRATHERTHANINSTALLATIONSTOCK
22、RAUETAL,DIKSANDDEKOK,DONGANDLEE,MITRAANDCHATTERJEE,HARIGA,CHEN,AXSATERANDZHANG,NOZICKANDTURNQUIST,ANDSOANDZHENGUSEAMATHEMATICMODELINGTECHNIQUEINTHEIRSTUDIESTOMANAGEMULTIECHELONINVENTORYINSCSDIKSANDDEKOKSSTUDYCONSIDERSADIVERGENTMULTIECHELONINVENTORYSYSTEM,SUCHASADISTRIBUTIONSYSTEMORAPRODUCTIONSYSTEM,
23、ANDASSUMESTHATTHEORDERARRIVESAFTERAFIXEDLEADTIMEHARIGA,PRESENTSASTOCHASTICMODELFORASINGLEPERIODPRODUCTIONSYSTEMCOMPOSEDOFSEVERALASSEMBLY/PROCESSINGANDSTORAGEFACILITIESINSERIESCHEN,AXSATERANDZHANG,ANDNOZICKANDTURNQUISTCONSIDERATWOSTAGEINVENTORYSYSTEMINTHEIRPAPERSAXSATERANDZHANGANDNOZICKANDTURNQUISTAS
24、SUMETHATTHERETAILERSFACESTATIONARYANDINDEPENDENTPOISSONDEMANDMITRAANDCHATTERJEEEXAMINEDEBODTANDGRAVESMODEL1985,WHICHTHEYDEVELOPEDINTHEIRPAPERCONTINUOUSREVIEWPOLICIESFORAMULTIECHELONINVENTORYPROBLEMWITHSTOCHASTICDEMAND,FORFASTMOVINGITEMSFROMTHEIMPLEMENTATIONPOINTOFVIEWTHEPROPOSEDMODIFICATIONOFTHEMODE
25、LCANBEEXTENDEDTOMULTISTAGESERIALANDTWOECHELONASSEMBLYSYSTEMSINRAUETALSMODEL,SHORTAGEISNOTALLOWED,LEADTIMEISASSUMEDTOBENEGLIGIBLE,ANDDEMANDRATEANDPRODUCTIONRATEISDETERMINISTICANDCONSTANTSOANDZHENGUSEDANANALYTICALMODELTOANALYZETWOIMPORTANTFACTORSTHATCANCONTRIBUTETOTHEHIGHDEGREEOFORDERQUANTITYVARIABILI
26、TYEXPERIENCEDBYSEMICONDUCTORMANUFACTURERSSUPPLIERSLEADTIMEANDFORECASTDEMANDUPDATINGTHEYASSUMETHATTHEEXTERNALDEMANDSFACEDBYTHERETAILORARECORRELATEDBETWEENTWOSUCCESSIVETIMEPERIODSANDTHATTHERETAILERUSESTHELATESTDEMANDINFORMATIONTOUPDATEITSFUTUREDEMANDFORECASTSFURTHERMORE,THEYASSUMETHATTHESUPPLIERSDELIV
27、ERYLEADTIMESAREVARIABLEANDAREAFFECTEDBYTHERETAILERSORDERQUANTITIESDONGANDLEESPAPERREVISITSTHESERIALMULTIECHELONINVENTORYSYSTEMOFCLARKANDSCARFANDDEVELOPSTHREEKEYRESULTSFIRST,THEYPROVIDEASIMPLELOWERBOUNDAPPROXIMATIONTOTHEOPTIMALECHELONINVENTORYLEVELSANDANUPPERBOUNDTOTHETOTALSYSTEMCOSTFORTHEBASICMODELO
28、FCLARKANDSCARFSECOND,THEYSHOWTHATTHESTRUCTUREOFTHEOPTIMALSTOCKINGPOLICYOFCLARKANDSCARFHOLDSUNDERTIMECORRELATEDDEMANDPROCESSINGUSINGAMARTINGALEMODELOFFORECASTEVOLUTIONTHIRD,THEYEXTENDTHEAPPROXIMATIONTOTHETIMECORRELATEDDEMANDPROCESSANDSTUDY,INPARTICULARFORANAUTOREGRESSIVEDEMANDMODEL,THEIMPACTOFLEADTIM
29、ES,ANDAUTOCORRELATIONONTHEPERFORMANCEOFTHESERIALINVENTORYSYSTEMAFTERREVIEWINGTHELITERATUREABOUTMULTIECHELONINVENTORYMANAGEMENTINSCSUSINGMATHEMATICMODELINGTECHNIQUE,ITCANBESAIDTHAT,INSUMMARY,THESEPAPERSCONSIDERTWO,THREE,ORNECHELONSYSTEMSWITHSTOCHASTICORDETERMINISTICDEMANDTHEYASSUMELEADTIMESTOBEFIXED,
30、ZERO,CONSTANT,DETERMINISTIC,ORNEGLIGIBLETHEYGAINEXACTORAPPROXIMATESOLUTIONSDEKKERETALANALYSESTHEEFFECTOFTHEBREAKQUANTITYRULEONTHEINVENTORYCOSTSTHEBREAKQUANTITYRULEISTODELIVERLARGEORDERSFROMTHEWAREHOUSE,ANDSMALLORDERSFROMTHENEARESTRETAILER,WHEREASOCALLEDBREAKQUANTITYDETERMINESWHETHERANORDERISSMALLORL
31、ARGEINMOSTLWAREHOUSENRETAILERSDISTRIBUTIONSYSTEMS,ITISASSUMEDTHATALLCUSTOMERDEMANDTAKESPLACEATTHERETAILERSHOWEVER,ITWASSHOWNBYDEKKERETALTHATDELIVERINGLARGEORDERSFROMTHEWAREHOUSECANLEADTOACONSIDERABLEREDUCTIONINTHERETAILERSINVENTORYCOSTSINDEKKERETALTHERESULTSOFDEKKERETALWEREEXTENDEDBYALSOINCLUDINGTHE
32、INVENTORYCOSTSATTHEWAREHOUSETHESTUDYBYMOHEBBIANDPOSNERSCONTAINSACOSTANALYSISINTHECONTEXTOFACONTINUOUSREVIEWINVENTORYSYSTEMWITHREPLENISHMENTORDERSANDLOSTSALESTHEPOLICYCONSIDEREDINTHEPAPERBYVANDERHEIJDENETALISANECHELONSTOCK,PERIODICREVIEW,ORDERUPTOPOLICY,UNDERBOTHSTOCHASTICDEMANDANDLEADTIMESTHEMAINPUR
33、POSEOFIIDASPAPERISTOSHOWTHATNEARMYOPICPOLICIESAREACCEPTABLEFORAMULTIECHELONINVENTORYPROBLEMITISASSUMEDTHATLEADTIMESATEACHECHELONARECONSTANTCHENANDSONGSOBJECTIVEISTOMINIMIZETHELONGRUNAVERAGECOSTSINTHESYSTEMINTHESYSTEMBYCHENETAL,EACHLOCATIONEMPLOYSAPERIODICREVIEW,ORLOTSIZEREORDERPOINTINVENTORYPOLICYTH
34、EYSHOWTHATEACHLOCATIONSINVENTORYPOSITIONSARESTATIONARYANDTHESTATIONARYDISTRIBUTIONISUNIFORMANDINDEPENDENTOFANYOTHERINTHESTUDYBYMINNERETAL,THEIMPACTOFMANUFACTURINGFLEXIBILITYONINVENTORYINVESTMENTSINADISTRIBUTIONNETWORKCONSISTINGOFACENTRALDEPOTANDANUMBEROFLOCALSTOCKPOINTSISINVESTIGATEDCHIANGANDMONAHAN
35、PRESENTATWOECHELONDUALCHANNELINVENTORYMODELINWHICHSTOCKSAREKEPTINBOTHAMANUFACTURERWAREHOUSEUPPERECHELONANDARETAILSTORELOWERECHELON,ANDTHEPRODUCTISAVAILABLEINTWOSUPPLYCHANNELSATRADITIONALRETAILSTOREANDANINTERNETENABLEDDIRECTCHANNELJOHANSENSSYSTEMISASSUMEDTOBECONTROLLEDBYABASESTOCKPOLICYTHEINDEPENDENT
36、ANDSTOCHASTICALLYDEPENDENTLEADTIMESARECOMPAREDTOSUMUP,THESEPAPERSCONSIDERTWOORNECHELONINVENTORYSYSTEMS,WITHGENERALLYSTOCHASTICDEMAND,EXCEPTFORONESTUDYTHATCONSIDERSMARKOVMODULATEDDEMANDTHEYGENERALLYASSUMECONSTANTLEADTIME,BUTTWOOFTHEMACCEPTITTOBESTOCHASTICTHEYGAINEXACTORAPPROXIMATESOLUTIONSINMULTIECHE
37、LONINVENTORYMANAGEMENTTHEREARESOMEOTHERRESEARCHTECHNIQUESUSEDINLITERATURE,SUCHASHEURISTICS,VARYMETRICMETHOD,FUZZYSETS,MODELPREDICTIVECONTROL,SCENARIOANALYSIS,STATISTICALANALYSIS,ANDGASTHESEMETHODSAREUSEDRARELYANDONLYBYAFEWAUTHORSAMULTIPRODUCT,MULTISTAGE,ANDMULTIPERIODSCHEDULINGMODELISPROPOSEDBYCHENA
38、NDLEETODEALWITHMULTIPLEINCOMMENSURABLEGOALSFORAMULTIECHELONSCNETWORKWITHUNCERTAINMARKETDEMANDSANDPRODUCTPRICESTHEUNCERTAINMARKETDEMANDSAREMODELEDASANUMBEROFDISCRETESCENARIOSWITHKNOWNPROBABILITIES,ANDTHEFUZZYSETSAREUSEDFORDESCRIBINGTHESELLERSANDBUYERSINCOMPATIBLEPREFERENCEONPRODUCTPRICESINTHECURRENTP
39、APER,ADETAILEDLITERATUREREVIEW,CONDUCTEDFROMANOPERATIONALRESEARCHPOINTOFVIEW,ISPRESENTED,ADDRESSINGMULTIECHELONINVENTORYMANAGEMENTINSUPPLYCHAINSFROM1996TO2005HERE,THEBEHAVIOROFTHEPAPERS,AGAINSTDEMANDANDLEADTIMEUNCERTAINTY,ISEMPHASIZEDTHESUMMARYOFLITERATUREREVIEWISGIVENASTHEMOSTUSEDRESEARCHTECHNIQUEI
40、SSIMULATIONALSO,ANALYTIC,MATHEMATIC,ANDSTOCHASTICMODELINGTECHNIQUESARECOMMONLYUSEDINLITERATURERECENTLY,HEURISTICSASFUZZYLOGICANDGASHAVEGRADUALLYSTARTEDTOBEUSEDSOURCEATASKINGUMUSANDAFUATGUNERITURKEY,2007“MULTIECHELONINVENTORYMANAGEMENTINSUPPLYCHAINSWITHUNCERTAINDEMANDANDLEADTIMESLITERATUREREVIEWFROMA
41、NOPERATIONALRESEARCHPERSPECTIVE”IMECHEVOL221PARTBJENGINEERINGMANUFACTUREJUNE,PP15531570译文供应链下的多级存货管理从历史上看，多级供应链、仓库、分销商、零售商等，已经通过大量的库存缓冲被独立管理。竞争压力的增加和市场的全球化迫使企业发展能够快速满足客户需要的供应链。为了保持竞争力，降低库存，这些企业必须交互使用多级库存管理，同时降低运营成本，改善客户服务。供应链管理是一种集成的管理思想和方法，它执行供应链中从供应商到最终用户以及公司内部不同部门之间的物流的计划和控制等职能。最近几年，供应链管理这个领域已经引起
42、了广泛关注，并且它被视作能够帮助企业获得竞争优势的一种工具。供应链管理是为了在满足服务水平的同时使系统成本达到最小，从而把供应商、制造商、仓库和商店的业务效率都整合起来，使商品以正确的品质、在正确的地点、在正确的时间进行生产和销售的一套方法。可见，供应链由各种各样的成员或者阶段组成。供应链是一个动态的、随机的和复杂的系统，因此，需要许许多多的人来参与。在制造企业中，存货通常占企业总资产的20到60。因此，对于那些追求利润的企业，存货管理政策是关键。当考虑到整个供应链，也就是一个由采购、转化和运输公司组成的网络时，存货管理在很大程度上就与这整个供应链有关。在供应链管理中，存货管理的确是一种主要方
43、法，它能从整体上解决供应链问题。因各种不同的原因，存货以不同形式存在在供应链中。在整个供应链中，存货管理失衡，经常会引起“牛鞭效应”，即需求逆流而上，逐级变异放大的一个阶段。这种效应引起企业过多的存货积压，使收入减少，运输效率降低，扰乱了库存计划和产品生产计划，同时降低了企业的服务水平。许多学者已经对这些问题进行了研究，并且强调了对有效地满足客户需求的供应链各阶段之间进行整合的必要性。除了整合问题，为了确定一个有效地供应链库存政策，还必须处理不确定性问题。除了对供应（比如交货时间）和需求的不确定性，与生产和销售过程相关的信息延迟也是供应链的一个特点。多级供应链中的库存管理是一项重要的内容，因为
44、有许多方面两者都必须相互配合，协调合作。它们还必须对它们的库存进行协调安排。有许多因素使成功的库存管理变得复杂，例如。需求的不确定、交货时间、投产日期、产品价格、成本等，尤其是在不确定性的需求和交货时间下，管理者不能够将多级供应链中的存货管理得最优。大多数制造企业被组织起来形成了一个制造和分销为一体的网络，这个网络包括了原材料的采购、加工和产品的销售。当一个产品经过多个阶段才到达最终用户时，多级或者多层次生产/分销网络这些代名词也和前面所述的这样的网络意思相同。因各种不用的原因，存货以不用的形式存在在整个供应链中。在任何一个制造过程中，它们可能作为原材料、在制品或者产成品存在。它们存在于配送仓
45、库，存在于运输途中，或者存在于管道里，它们存在于这些设备的每个链接处。制造商从供应商处采购原材料，将它们加工成产品并销售给分销商，然后由分销商销售给零售商或者用户。当一个产品经过多个阶段才到达最终用户，它就形成了一个多级库存系统。某一库存节点的级库存等于这个库存节点上的所有库存加上转移或者正在转移的任何一个后续节点的库存，减去后续节点的缺货。在商界有关多级库存系统的分析已经有着悠久的历史。在许多领域，多级库存管理系统被广泛运用于向客户分销产品。鉴于这些系统的重要性，许多研究人员通过各种各样的条件和假设开始研究他们的运行特点。自从哈里斯提出经济订货批量模型以来，研究人员和实际工作者更加积极地关注
46、在不同操作参数和模型假设条件下系统的分析和模型设计。在过去的十年里，对于多级库存管理模型的研究已经获得了重要成就，主要是因为通过利用现代信息技术，使各个过程和分销阶段的供应链的整体控制逐渐变成可能。克拉克和斯卡夫最早研究两阶段存货模型。他们证实了库存系统的基础存货政策的最优性，并提出了一种用于计算最佳订货批量的政策。贝斯勒和凡诺特进一步发展了两阶段模型，使其包含一般块茎结构。上面提到的车间仓库问题通过埃本和施拉格分析一个缺货的中央仓库模型解决了。他们在相等的订货点分配假设条件下，对订购批量做出了更近似的表述。一些作者也已经考虑到了在各种形式下的这个问题。由于多阶段问题的复杂和棘手，哈德利和怀廷
47、建议对库存系统采用单层次、单阶级模型。夏布鲁克把一个订购政策看做是一个仓储和零售商的两级模型，他假设零售商的缺货是完全积压的，而且，夏布鲁克还建立了矩阵（可收回项目控制的多级技术）模型，它明确了在有预算约束的一个低级阶段中使库存水平最小化，这个模型是管理服务部分库存的第一个多级模型，此后，很多研究者提出了一大套模型，他们一般都是在多级框架下寻求最佳批量和安全库存。除了分析性模型，仿真模型也被开发了出来用于研究多级库存问题中复杂的相互作用问题。到目前为止，相关的一些文献主要关注于对需求的预测，以及对多阶段供应链库存政策的发展。需求随机的多阶段系统的存货控制政策已经具有了一个广泛的研究领域。近年来
48、有许多论文都包含了斯尔福和派克的观点。用于定期评估标准的统一采购的优点是可以通过规定不同阶段的订购水平获得连续不断的评估标准，这是就所有库存而言，而不是单指设备。劳以及其他人，迪克斯和戴科克，唐格和里，密特拉和查特基，哈里加，陈，阿克斯特和章，诺齐克和特纳基斯特以及赛欧和郑都在他们的研究中利用了数学模型技术去管理供应链中的多级存货。迪克斯和戴科克的研究考虑到了不同的多级存货系统，比如配送系统或者生产系统，并且假设订单在一个固定的时间内到达。哈里加提出了由若干个装配或者整理和储存设备串联在一起组成的单个周期生产系统的随机模型。阿克斯特，诺齐克和特纳基斯特在他们的文章中都提到了两阶段库存系统。阿克
49、斯特和特纳基斯特认为零售商都面临不变的无偏好的泊松需要。麦彻和查特基研究了博特和格拉夫模型，并在他们关于实行快速配送商品的观点的论文“对随机需求下的多级存货问题策略不间断回顾”中进行了进一步的阐述。这个模型的提出和改进能够扩展多个阶段和两阶段配送系统的内容。在劳尔的模型中，假设订货时间忽略不计，需求率和生产率是确定的，而且保持固定不变的情况下，缺货是不允许的。赛欧和郑运用分析模型分析两个重要因素，这两个因素能够帮助半导体制造商根据经验推测订单数量变化的最高程度一个是供应商的订货时间，另一个是预测的需求更新情况。他们认为那儿的零售商面临的外部需求与两个连续的时间段相联系，并且零售商利用最新的需求信息来更新它们未来的需求预测。此外，他们还认为供应商的供货时间是变动的，而且受零售商的订货量的影响。顿和里在他们的论文中再次阐述了克拉克和斯卡夫的连续多级存货系统并得出了三个关键的结论。第一，他们提出了最佳多级存货水平的最小近似值以及克拉克和斯卡夫关于基本模型的整个系统成本的最大值。第二，他们利用马丁格尔预测理论模型说明了克拉克和斯卡夫的最优存货政策结构保留了在与时间线关联下的按需处理。第三，他们把近似值拓展到了与时间相关联的需求的过程和研究，特别是对于一个回归需求模型，订货时间的影响和一系列存货系统性能的相关性。通过对有关利用数学模型
