Robust predictive control for the management of multi-echelon distribution chains

The problem of robust inventory control for multiechelon distribution chains is addressed by using predictive control. Since the future demand of goods is assumed to be uncertain, we focus on a worst-case planning strategy that is consistent with the demand predictions and minimizes the maximum of a performance objective function. The resulting optimal decisions concern the delivery of goods in such a way to reduce the overall costs as to holding, transportation, and backlogs. As compared with previous works, the proposed approach allows one to deal with distribution chains of much larger dimension because of its intrinsic scalability. Of course, such an appreciable feature is paid in terms of loss of optimality. However, a convenient tradeoff can be achieved, as shown by means of simulations.