Abstract
Recently, we have concerned strategic optimization on logistics network design and developed an efficient two-level solution method. To cope with extremely large-scale problems, in this paper, we propose an extended algorithm of the two-level algorithms that utilizes the analogy between the algorithm and master-worker configuration of PC cluster for parallel computing. To enhance efficiency both in speed and accuracy, we adopted a population-based algorithm like particle swarm optimization (PSO) and developed a new binary algorithm of PSO to deal with the binary decision variables standing for open or close of DC sites. Then, we have developed a parallel procedure that can reduce the inefficiency coming from idle time and overhead and finally bring about a high performance for the parallel computing suitable for the present goal and circumstance. Through numerical experiments, we confirmed effectiveness of the proposed method through the certain aspects to evaluate effectiveness of the parallelization.
