集装箱码头泊位分配外文文献翻译.docx
《集装箱码头泊位分配外文文献翻译.docx》由会员分享,可在线阅读,更多相关《集装箱码头泊位分配外文文献翻译.docx(5页珍藏版)》请在冰豆网上搜索。
集装箱码头泊位分配外文文献翻译
文献出处:
Bierwirth,Christian,andFrankMeisel."Afollow-upsurveyofberthallocationandquaycraneschedulingproblemsincontainerterminals." EuropeanJournalofOperationalResearch (2014).
原文
Afollow-upsurveyofberthallocationandquaycraneschedulingproblemsincontainerterminals
Bierwirth;Christian;Meisel.
Berthallocationproblems
Scopeandclassificationscheme
Inberthallocationproblems,wearegivenaberthlayouttogetherwithasetofvesselsthathavetobeservedwithinaplanninghorizon.Thevesselsmustbemooredwithintheboundariesofthequayandcannotoccupythesamequayspaceatatime.Inthebasicoptimizationproblem,berthingpositionsandberthingtimeshavetobeassignedtoallvessels,suchthatagivenobjectivefunctionisoptimized.Avarietyofoptimizationmodelsforberthallocationhavebeenproposedintheliteraturetocapturerealfeaturesofpracticalproblems.In BierwirthandMeisel(2010),wehaveproposedaschemeforclassifyingsuchmodelsaccordingtofourattributes,namelya spatialattribute,a temporalattribute,a handlingtimeattribute,andthe performancemeasure addressedintheoptimization.Thevalueseachattributecantakearelistedin Fig.1
2.1.1.Spatialattribute
Thisattributeconcernstheberthlayout,whichiseitheradiscretelayout(disc),acontinuouslayout(cont),orahybridlayout(hybr).Incaseof disc,thequayispartitionedintoberthsandonlyonevesselcanbeservedateachsingleberthatatime.Incaseof cont,vesselscanberthatarbitrarypositionswithintheboundariesofthequay.Finally,incase hybr,thequayispartitionedintoberths,butvesselsmayshareaberthoronevesselmayoccupymorethanoneberth.Aparticularformofahybridberthisanindentedberthwherelargevesselscanbeservedfromtwooppositelylocatedberths.Thespatialattributeisextendedbyitem draft,iftheBAP-approachadditionallyconsidersavessel’sdraftwhendecidingonitsberthingposition.
2.1.2.Temporalattribute
Thisattributedescribesthearrivalprocessofvessels.Theattributereflectsstaticarrivals(stat),dynamicarrivals(dyn),cyclicarrivals(cycl),andstochasticarrivaltimes(stoch).Incaseof stat,weassumethatallvesselshavearrivedattheportandwaitforbeingserved.Incontrast,incaseof dyn,thevesselsarriveatindividualbutdeterministicarrivaltimesimposingaconstraintfortheberthallocation.Incase cycl,thevesselscallatterminalsrepeatedlyinfixedtimeintervalsaccordingtotheirlinerschedules.Incase stoch,thearrivaltimesofvesselsarestochasticparameterseitherdefinedbycontinuousrandomdistributionsorbyscenarioswithdiscreteprobabilityofoccurrence.Cyclicandstochasticarrivaltimesareconsideredinanumberofrecentpublicationsand,therefore,wehaveextendedtheoriginalclassificationschemewithregardtothesecases.Thetemporalattributeiscompletedbyvalue due,ifaduedateispresetforthedepartureofavesselorifamaximumwaitingtimeispresetforavesselbeforetheservicehastostart.
Temporalattribute
Thisattributedescribesthearrivalprocessofvessels.Theattributereflectsstaticarrivals(stat),dynamicarrivals(dyn),cyclicarrivals(cycl),andstochasticarrivaltimes(stoch).Incaseof stat,weassumethatallvesselshavearrivedattheportandwaitforbeingserved.Incontrast,incaseof dyn,thevesselsarriveatindividualbutdeterministicarrivaltimesimposingaconstraintfortheberthallocation.Incase cycl,thevesselscallatterminalsrepeatedlyinfixedtimeintervalsaccordingtotheirlinerschedules.Incase stoch,thearrivaltimesofvesselsarestochasticparameterseitherdefinedbycontinuousrandomdistributionsorbyscenarioswithdiscreteprobabilityofoccurrence.Cyclicandstochasticarrivaltimesareconsideredinanumberofrecentpublicationsand,therefore,wehaveextendedtheoriginalclassificationschemewithregardtothesecases.Thetemporalattributeiscompletedbyvalue due,ifaduedateispresetforthedepartureofavesselorifamaximumwaitingtimeispresetforavesselbeforetheservicehastostart.
Handlingtimeattribute
Thisattributedescribesthewayhowhandlingtimesofvesselsaregivenasaninputtotheproblem.Ittakesvalue fix,ifthehandlingtimesofvesselsareknownandconsideredunchangeable.Value pos indicatesthathandlingtimesdependontheberthingpositionsofvesselsandvalue QCAP indicatesthathandlingtimesaredeterminedbyincludingQCassignmentdecisionsintotheBAP.Incaseofvalue QCSP,thehandlingtimesaredeterminedbyincorporatingtheQCschedulingwithintheBAP.Inordertoclassifytherecentliteratureproperly,wehaveinsertedcase stoch asanewattributeforthescheme.Again,handlingtimescanbesubjecttoeitherdiscreteorcontinuousrandomdistributions.Asimilarextensionofourschemeisalsosuggestedby Carloet al.(2013),whoalsoopenittofurthersourcesofinfluenceonvesselhandlingtimes,likeoperationsoftransfervehiclesandyardcranes.However,aswehardlyfindinstantiationsofthesecasesintheliterature,werefrainfromextendingtheschemeinfurtherdirections.
Performancemeasure
Thisattributeconsiderstheperformancemeasuresofaberthallocationmodel.Mostmodelsconsidertominimizetheportstaytimeofvessels.Thisisreachedbydifferentobjectivefunctions,e.g.whenminimizingwaitingtimesbeforeberthing(wait),minimizinghandlingtimesofvessels(hand),minimizingservicecompletiontimes(compl),orminimizingtardyvesseldepartures(tard).Ifsoftarrivaltimesaregiven,alsoapossiblespeedupofvessels(speed)istakenintoconsiderationattheexpenseofadditionalbunkercost.Othermodelsaimatreducingthevariableoperationcostofaterminalbyoptimizingtheutilizationofresources(res)likecranes,vehicles,berthspace,andmanpower.Anoftenconsideredfeatureistosavehorizontaltransportcapacitybyfindingberthingpositionsforvesselsclosetotheyard,whichiswhyweincludethisgoalbyitsownvalue pos.Rarelymetperformancemeasuresaresummarizedbyvalue misc(miscellaneous).Theintroducedmeasuresareeithersummedupforallvesselsintheobjectivefunction.Alternatively,iftheminimizationofthemeasurefortheworstperformingvesselispursued,i.e.amin–maxobjectiveisfaced.Vessel-specificprioritiesorcostratesareshownbyweights.Differentweights w1 to w4 addresscombinedperformancemeasures.
Literatureoverview
Intherelevantliterature,wehavefoundandclassified79newmodelsforberthallocation,mostofthempublishedafter2009. Fig.2 showstheBAPmodelsdevelopedbyresearcherssince1994byyearoftheirpublication,includingalsothoseapproachesreviewedin BierwirthandMeisel(2010).Thefigureshowsthattheinterestinberthallocationstartedwiththeearlypapersof Hoffarth(1994) and Imai,Nagaiwa,andTat(1997).However,thegrowthofpublicationsfollowedthepioneeringpaperof ParkandKim(2003),whocombinedberthallocationandQCassignmentforthefirsttime,andtheearlysurveyoncontainerterminaloperationsby Steenkenet al.(2004).Inparticular,journalpublicationsscaleduptotenandmoreperyearafter2010.Tothemidof2014,already13newjournalpapershavebeenpublishedoracceptedforpublication.Thecontinuouseffortspendonresearchinberthallocationconfirmsitasawell-establishedfieldtoday,whichstillshowspotentialforfutureresearch.
With Table1,wealsoprovideanoverviewofthemethodsthatareusedforsolvingtheBAPmodels.Notethatonlythemostsuccessfulmethodpresentedinapaperappearsinthetable.ItisnotsurprisingthatheuristicapproachesdominateastheBAPisknowntobe NP-hardinboth,thediscreteandthecontinuouscase,seee.g. Lim(1998) and HansenandOǧuz(2003).Exactmethodsareappliedinonlyonefourthoftheapproaches,rangingfromMILPformulationscombinedwithstandardsolverstohighlysophisticatedbranching-basedalgorithms.Amongtheheuristicapproaches,GeneticAlgorithmsandEvolutionaryAlgorithmstakethebyfarlargestsharewith40percent,see Fig.3(left).Therestofthemethodscompriseothermeta-heuristicslikeTabuSearchandSimulatedAnnealingaswellasproblemspecificheuristicslikelocalsearchtechniquesandgreedyrules.TherichnessofBAPmodelsfavorsmeta-heuristicapproachesastheyallowhandlingvariousproblemfeaturesflexibly.Ontheotherhand,asystematicevaluationofalgorithmsishinderedbythestrongheterogeneityofBAPmodels.Althoughcomparingmodelsisdefinitelynecessaryforassessingthesuitabilityofmethods,thecomparisonofalternativemodelsformulatedbydifferentresearchgroupsisjustemergingslowly,see Buhrkal,Zuglian,Ropke,Larsen,andLusby(2011),Umang,Bierlaire,andVacca(2013),and Imai,Nishimura,andPapadimitriou(2013).Tomakethisprocesssustainable,commonlyacceptedBAPbenchmarkinstancesareneededtoprovideauthorswiththeopportunitytoevaluatetheirwork.However,thecurrentbenchmarksareeithernotgeneralenoughtofulfillthisaimortheyaremerelyusedinsmallsubstreamsoftheentireresearchfield.Definingbenchmarkproblemsforgeneralberthallocationproblemsthatfulfilltheprinciplesofcomparability,unbiasedness,andreproducibilityremainsanopentopicforfutureresearch.
Inthefollowing,weabstainfromreviewingallpaperslistedin Table1 individually.Instead,thenextsubsectiondiscussesthosepapersinmoredetailthatcontainnovelfeaturesofwhichwethinktheymightreceiveparticularattentioninthefuture.
Newdevelopments
Newtechnologicaldevelopmentsthatimpacttheberthallocationprobleminc