Three-level Logistic Network Location-allocation Model And Its Optimization With Facility Disruptions

Classic facility location model doesn’t consider facility disruptions, assuming thatfacilities are completely reliable. In dynamic environment, facilities disruptions sometimesoccur, and classic facility location is faced with many challenges: How to ensure the validityof the decision and enhance the logistics network anti-risk ability with facilities disruptions isan important problem. The facility disruption researches are single-resource and two-levellogistic networks at present. This paper researches a more common multi-resource, three-levellogistic network location-allocation model and optimization in practice.For facility disruptions, most of the researchers based on the idea of "R leveldistribution"("backup distribution"), but the model is complex and difficult to solve. A statevariable introduced reflects whether facilities are invalid or not, which simplifies the structureof model. Location decision belonging to the enterprise long-term decision, we pay moreattention to the long-term benefits decision, so the optimization objective function of themodel is to minimize the expect total cost of logistic.This model of the paper is nonlinear, which has been transformed into linear one by theMonte Carlo, and then the paper adopts the improved bacteria foraging algorithm to optimizethe first phrase location decision, and employs the fast assignment algorithm to optimize thesecond phrase allocation decision. To solve the problem of local optimum and fastconvergence, the two improved strategies are proposed: adaptive step length adjustmentstrategy and the indicativetumble strategy.Finally, based on the seven simulation examples, the validity of the model and algorithmis verified. The simulation experiment shows that the two-phrase heuristic algorithm can givesatisfactory solution in a short time; Compared with the traditional facility location model,location model with facility disruptions may increase the daily operation costs lightly, but canreduce economic loss significantly if failure occurs. For an actual logistics enterprise, we givea location-allocation decision with this model and the algorithm, and do a sensitivity analysisof the model parameters.This study has a certain theoretical and practical value to expand facilities locationtheory, to improve the heuristic algorithm, to guide the practice of decision-makers, and topromote scientific location decisions.