Capacitated facilities location problems with unreliable facilities.

摘要

Modern supply chains are subject to a wide range of risks such as demand uncertainty, natural disasters, terrorist attacks, etc. In this thesis, we consider large scale disruption risks in supply chain networks, We study the problem of locating facilities in a supply chain when the facilities are unreliable in providing service. Snyder and Daskin [1] presented deterministic formulations for the Uncapacitated Facilities Location Problem (UFLP) and the p-Median problem when the facilities are unreliable. We extend these models and propose a scenario-based formulation for the Capacitated Facilities Location Problem (CFLP). We observe that the scenario-based formulation can be represented as a two-stage stochastic program with relatively complete recourse. Due to the intractability of the scenario-based formulation to exact solution methods, we propose a sampling based procedure known as the Sample Average Approximation (SAA) to approximately solve this model. We investigate the performance of this sampling-based procedure. We show, from our experiments that the SAA procedure performs well for a wide range of parameters. We also propose as an extension a formulation for the two-echelon facility location problem when the facilities are unreliable and a decomposition method to solve the approximation problems.