A two-phase heuristic algorithm for designing reliable capacitated logistics networks under disruptions

Abstract

This paper considers the reliable capacitated logistics network design problem (RCLNDP) with system disruptions, which is concerned with locating facilities, constructing transportation links, and allocating their capacities to customers in order to satisfy the demand with minimum expected total cost. Both the facilities and the transportation links are subject to random disruptions, and the expected total cost accounts for the costs of facility location, link construction, and flows in both normal and disrupted conditions. We model this problem as a two-stage stochastic program in which the decision maker establishes plans for facility location and link construction in the first stage (before disruptions are realised) and may choose link flows in the second stage. This is a large-scale mixed-integer optimisation problem and is therefore difficult to solve. Hence, we propose an efficient two-phase heuristic with three possible initial solution-generation methods. [Received 1 November 2015; Revised 3 March 2016; Revised 22 May 2016; Revised 8 June 2016; Accepted 10 June 2016]