A simulation-based optimization method for production-distribution network design

Production-distribution network design is a critical decision that has significant impacts on a supply chain´s long-term performance. Supply chain dynamics, such as demand fluctuation, and transportation instability, are omitted in most mathematical models due to tractability. We present a simulation-based optimization method in this paper for multi-criteria production-distribution network design. The method consists of a multiobjective optimizer and a simulation module. The optimizer, based on a multiobjective genetic algorithm, is used to direct the search for compromised solutions regarding to various criteria. Candidate solutions are evaluated by the discrete-event simulation module, developed in a flexible manner that enables automatic simulation of various supply chain structures without human intervention. The method is applied to a case study from automotive industry. A set of Pareto-optimal solutions are obtained, including decisions on the open/close decisions on facilities, order splitting ratios and inventory control policies.