Strategic Design of Robust Global Supply Chains under Uncertainty

We will present a model for the strategic design of the global supply chain of an individual company. The model presented is a multi-period, two-stage stochastic, multi-country, multi-product, multi-echelon formulation based on forecasted exchange rates and with bill of materials (BOM) flow conservation. The objective is maximization of the expected value of the time-discounted world-wide after-tax net cash flows NPV(NCF). The uncertainty of the data is explicitly considered through the inclusion of scenarios in the deterministic equivalent problem (DEP). However, the problem size of the DEP is so large that convergence of standard Benders decomposition requires an excessive amount of computer time. We have implemented several acceleration techniques for Benders decomposition. The following observations are based on the statistical analysis of the required computing times for an industrial case study. 1) Acceleration techniques significantly decrease the computation times. 2) Strengthening of the dual variables contributes the most, while a primal heuristic contributes the least. 3) The largest reductions in computer time are achieved when all acceleration techniques are used in concert. We are continuing the numerical experiment to further differentiate the effectiveness of the acceleration techniques.