A simulation-optimization approach for a service-constrained multi-echelon distribution network

Academic research on (s,S) inventory policies for multi-echelon distribution networks with deterministic lead times, backordering, and fill rate constraints is limited. Inspired by a real-life Dutch food retail case we develop a simulation-optimization approach to optimize (s,S) inventory policies in such a setting. We compare the performance of a Nested Bisection Search (NBS) and a novel Scatter Search (SS) metaheuristic using 1280 instances from literature and we derive managerial implications from a real-life case. Results show that the SS outperforms the NBS on solution quality. Additionally, supply chain costs can be saved by allowing lower fill rates at upstream echelons.