A two-step optimization model for the distribution of perishable products

Optimal planning for distribution networks of perishable products is addressed by means of a two-step approach pursuing both strategic and tactical goals. The distribution network is represented as a directed graph, and discrete-time dynamic equations are devised to model the storage and transportation of products. The first decision step consists of an optimization problem to account for a strategic viewpoint. In particular, this problem allows to select optimal values for replenishment cycles of products, safety stocks, and amounts of products to be transferred among the nodes of the network by considering uncertainty on long-term demand predictions. The second decision step requires the solution of rolling-horizon optimization problems at the various time buckets that exploit accurate, short-term predictions of customers' demands according to a tactical perspective. In this case, the strategic decisions on the amounts of products to be transferred are tuned according to the available short-term predictions of demands. The effectiveness of the proposed approach is showcased by simulations in a case study, in comparison with a classical lot-for-lot strategy.