Abstract: The actual demand and cost in logistics system arechanged with time. The facility location problem considering the time factor wasstudied in this paper. A dynamic logistics nodes location model considering capacitated,multi-source and multi-level logistics nodes was presented. We also consideredthe influence of the transition between hub and non-hub logistics nodes, andestablished a new node. Based on the previous multi-objective evolutionaryalgorithm based on external dominance clustering (ED-MOEA), a discrete ED-MOEA wasproposed to resolve this problem. Whensolving it, the objective was decomposed into a bi-objective problem. Thesimulation results show the effectiveness of the proposed model and thealgorithm.
Keywords:Dynamic Location, Multi-ObjectiveEvolutionary Algorithm, Logistics Nodes
1. Introduction
Facility locationand allocation is significant indesigning a distribution system ofsupply chain, which coversmost parts of the investment of initial cost. Itwill affectthe efficiency of vehicles in further operation performedin delivery,which affectedby the optimal degree of the facilities in companies. An efficient logistic network is very helpfulin reducing operation cost in supply chain. Facility location existed when a manufacturefactory, retail chain storeor 3PL (Third-PartyLogistics) company choosesits sub-factory,warehouseor logistics center,which is a practical problem. Facility location problem was firstly studied byCooper [1]in 1963.The problem evolved with the development of the sociaty.The formulations of the problem ranged from simple linear, single-stage,single-product, un-capacitated and deterministic models to non-linear models [2]-[5].Many researchers proposed their ideasto solve the problemsin thepast years [6]-[12]. Facility location problemswerestudied vastly for theirwidely applicationsin real-life.Most of them are the staticlocation problems, which rarely consider the time factor. As the demand and the cost structure change with time in actual logisticdistribution system in supply chain, the time should beconsidered in the procedure of facility location. Dynamic location problems aredifferent from staticlocation problemsin the timefactor of both demand and cost structure. Ballou analyzed the dynamic warehouselocation firstly in 1968. Butveryfew referencesof dynamic location problemscan be found [13-16].Evolutionaryalgorithms (EAs) in general (i.e. genetic algorithms, evolutionary programming,etc.)are heuristic search techniques inspiredby Darwin’sevolutionary theory. Evolutionary algorithms have become very popular asoptimizers in the last few years. As a consequence of this popularity, theiruse has spanned many areas. Inthis paper, we present an optimization model to solve the capacitated, multi-source, multi-levelmulti-period, logistics node location problem,based on the research of the node problem [17] and dynamic location problem [18-19]. Based on the algorithm ED-MOEA [20],a discretemulti-objective evolutionary algorithm based onexternal dominated clustering(dED-MOEA) is employedhereto solve the presented model.