Maximizing the throughput of multimodal logistic platforms by simulation-optimization: The Duferco case study

In this work a multimodal transport platform designed by Duferco near a steel factory in Belgium is considered. Aim of this paper is to present a simulation-optimization decision system for evaluating the maximum throughput of the platform in presence of uncertainties over the durations of the operations. A greedy algorithm and a local search phase has been implemented in order to create a feasible schedule of the planned operations over different time horizons (from a month to a year). Once a schedule has been computed using approximated input data, it is simulated, introducing stochasticity in the system parameters. From the simulation results it is possible to better evaluate some of the scheduling data, modify them accordingly, and re-schedule and simulate the system. This procedure is iteratively applied until an asymptotic state is reached. Extensive computational results show how the new solutions improve the company best practices.