ebook img

Data Driven Ambulance Optimization PDF

49 Pages·2014·11.47 MB·English
by  
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 Data Driven Ambulance Optimization

Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Time-dependent ambulance allocation considering data-driven empirically required coverage Lara Wiesche, Dirk Degel, Brigitte Werners Management,OperationsResearch–Ruhr-UniversityBochum,Germany 1stInternationalWorkshoponPlanningofEmergencyServices: TheoryandPractice CWI, Amsterdam, 25–27 June 2014 LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 1/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Outline 1. Motivation 2. Basic idea and models 3. Data driven optimization model for tactical ambulance planning 4. A real world EMS planning problem in the city of Bochum 5. Conclusions and Outlook LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 2/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Motivation (cid:73) Aging population (cid:73) Age class of 70+ causes 50% of ambulance operations (cid:73) EMS demand increases source:StatistischesBundesamt2013 LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 3/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Motivation (cid:73) Aging population (cid:73) Age class of 70+ causes 50% of ambulance operations (cid:73) EMS demand increases source:StatistischesBundesamt2013 LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 3/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Motivation 1. Access to emergency medical services (EMS) is crucial 2. Ensuring optimal supply quality (cid:73) Short arrival time (cid:73) Coverage of entire demand area (cid:73) High degree of achievement y nc =⇒ Dependent on location and number of Efficie ambulances 3. Efficiency Q.uality (cid:73) Reduction of fixed costs (avoid overcapacity) (cid:73) Utilization of ambulances LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 4/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Quality criteria and objectives Evaluation of emergency medical services (objective of EMS provider): (cid:73) Degree of achievement (ex post): # operations within a time standard T # total number of operations . Objectives in literature (ex ante): ZeitabhängigerRadiusbei25(cid:20)khm(cid:21) A B C D E F G H J K L M N O P Q R (cid:73) single coverage models 1 129 1 2 55 67 92 105118130140148 2 (cid:73) double coverage models 3 44 56 68 80 93 106119131141 3 4 12 22 33 45 57 69 81 94 107120132142149 4 (cid:73) busy fraction models 56 1 5 1134 2234 3345 4467 5589 7701 W88236 9956 11008911221211333411W44343115501115556160 56 7 2 6 15 25 36 48 W602 72 84 97 110123135145152157161 7 (cid:73) queue models 8 3 7 16 W261 37 49 61 73 85 98 111124136146153158162 8 9 4 8 17 27 38 50 62 74 86 99 112125137147154159163 9 (cid:73) hypercube models 1101 190 1189 2289 3490 5512 W66344 7756 W88785 110001111134112267113389 1101 12 11 20 30 41 53 65 77 89 102115128 12 13 21 31 42 54 66 78 90 103116 13 14 32 43 79 91 104117 14 A B C D E F G H J K L M N O P Q R LaraWiesche (RUB) DataDrivenAmbulanceOptimization Wache Ra2die5n–in2Ab7hänJgiugkeniteder2Ge0sc1hw4indigkeitv(t)5/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Optimal quality in emergency medical services Research project: 2 years (work in progress) Focus: (cid:73) Analysis of required resources for EMS (cid:73) Analysis of time-dependent demand and speed fluctuations for EMS (cid:73) Tactical and strategic planning horizon (cid:73) Stochastic influences, uncertain parameters (demand, speed) Goal: (cid:73) Dynamic (and robust) optimization model =⇒ IT-based decision support tool for local EMS providers LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 6/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Decision support tool Decision support Optimization Simulation Forecast location, degree of trends, allocation and achievement, development resource planning resources, utilization empirical demand, evaluations population, required/adequate experts coverage LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 7/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Decision support tool Decision support Optimization Simulation Forecast location, degree of trends, allocation and achievement, development resource planning resources, utilization empirical demand, evaluations population, required/adequate experts coverage LaraWiesche (RUB) DataDrivenAmbulanceOptimization 25–27June2014 7/28 Motivation Basicmodels Optimizationmodel Arealworldproblem Outlook Basic covering location models Idea: Demand nodes i have to be covered within a time standard T M.H.FazelZarandietal./ScientiaIranica,TransactionsE:IndustrialEngineering18(2011)1564–1570 1565 2. Literaturereview The large number of publications in the area of MCLP preventsacomprehensivereviewinthispaper.Thus,some of the main contributions to the field are discussed in this section.TwovaluableandrecentreviewsbyReVelleetal.[3] andBermanetal.[4]maybeconsultedforfurtherinformation regardingthesubject. MCLP was first presented by Church and ReVelle [1] in the mid 70s. Since then, numerous papers about various types of the problem have appeared. Some examples are theproposalofhierarchicalMCLPbyMooreandReVelle[5], modelingtheMCLPincompetitiveenvironmentsbyPlastria and Vanhaverbeke [6], gradual covering models by Church andRoberts[7],andBermanetal.[8,9],usingGISinMCLP byAlexandrisandGiannikos[10],andcoveringnodes,using parallelogramsbyYouniesandWesolowski[11],tonameafew. Another stream of research in the field of MCLP deals with employing heuristics and metaheuristics to tackle the computationalhurdleofsolvingMCLPmodels.Table1presents Figure1: Asamplesolutiontothecoveringlocationproblem. some of the main non-exact solution algorithms that have been used in order to solve variants of MCLP from the thesetofnodesandthesetofpotentialfacilitiesareidentical. source:Zarandietal.2011–Thelargescalemaximalcoveringlocationproblem;page1565 beginningofthe21stcentury.Thisreviewapparentlyshows Also,itisassumedthateachnodehasaspecificamountofde- LaraWiesche (RUB) mandtobeDmateat.DBreivseidneAs,mpbruolxainmceitOypotfimthizeatpioonpulationtothese25–2t7haJtumnee2ta0h14euristi8cs/h2a8vebeenanattractiveareaofresearchto facilitiesisdesirable.Thismeansthatfacilitiessuchaslandfills solveMCLPmodels. areexcludedfromthescopeofthispaper.Inaddition,anexoge- Asfarasweareconcerned,theliteratureonreal-worldcase nousversionofMCLPisconsidered.Hereby,themathematical studiesofMCLPiscomparativelylimited.Acasetoameliorate programofMCLPisgiven.First,wedefineproblemparameters theaccessibilityofdemandnodestofixedfacilitieshasbeen andvariablesasdiscussedin[2](forfurtherinformationabout considered by Murawski and Church [25], which is called theformulation,interestedreaderscanreferto[2]): the‘‘maximalcoveringnetworkimprovementproblem’’,and i,I Theindexandsetofdemandnodes, whichhasbeenemployedinGhanaasareal-worldcase.In another paper, Curtin et al. [26] studies the distribution of j,J Theindexandsetofeligiblefacilitysites, policepatrolareasinDallas,USA.Inaddition,Raticketal.[27] ai Thepopulationordemandatnodei can be considered as another real-world case study in the dij Theshortestdistance(ortime)fromdemandnodei Kohat district of Pakistan, by applying the model of Moore tofacilityatnodej, andReVelle[5].Recently,Basaretal.[24]surveyedthemulti- S Thedistance(ortime)standardwithinwhich periodversionofMCLPtolocateemergencyservicesinIstanbul, coverageisexpected, Turkey. Ni o{jf|Sdtioj≤noSd}e=i,thenodesjthatarewithinadistance ofpAarraemceentterisn,tseurcehstaisndMemCLaPndhsaosraerviseennthabeolouctatthioenuonfcdeertmaianntdy nodes.Batanovicetal.[28]suggestedaMCLPinnetworkswith p Thenumberoffacilitiestobeestablished, uncertainty,andstudiedtheproblemwiththreedifferenttypes xj Abinaryvariablethatequalsonewhenafacilityis ofdemandnodeweight: sitedatthejthnodeandzerootherwise, yi Abinaryvariablewhichequalsoneifnodeiis (a)Equalweights. coveredbyoneormorefacilitiesstationedwithinS (b)Relativedeterministicweights. andzerootherwise, (c)Linguistictermsasthedemandweights. Moreover,Corrêaetal.[21]isanotherrecentpublicationwhich Maximizez=aiyi, (1) addresses the problem of MCLP, considering congestion in i∈I queuelines.Arazetal.[29]presentedafuzzygoalprogramming Subjectto:yi≤xj, i∈I, (2) appappreoracbhyfBoerrmloacnatinetgaelm. e[3rg0e]nsctyudsieersvictheefagcrialidtiueasl. Acovreecreinngt j∈Ni problemwithuncertaindemands.Inanotherpublicationby xj=p, (3) BermanandWang[31],theysurveyedacaseinwhichthe j∈J demands are random variables with unknown probability 0≤yi≤1 i∈I, (4) dinistmriboudteiloinngs.FcaozveelrZinarganpdrioebtleaml.[s32in]istahneotfhuezrzyrecseenntsaet.teTmhepyt xj∈{0,1} j∈J. (5) modeledaclosed-loopsupplychainnetworkwithcoverage Therestofthepaperisorganizedasfollows:aconciseliterature considerations,andusedfuzzygoalprogrammingtosolveit. reviewofcoveringproblemsandrelatedissuesispresented Tothebestoftheauthors’knowledge,theonlyattempts inSection2.Section3isdedicatedtotheproposedsolution tomodelandsolveMCLPoflargersizesareReVelleetal.[2] algorithm.NumericalexamplesappearinSection4.Finally, andParkandRyu[15].Althoughthesetwopapersarevaluable conclusionsandoutlooksforpotentialfutureresearcharegiven contributionstotheliterature,theproblemsizeswhichwere inSection5. targetedarebelow900nodes,inbothpapers.Moreover,itis

Description:
Basic models. Optimization model. A real world problem. Outlook. Time-dependent ambulance allocation considering data-driven empirically required
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.