Mixed-integer linear programming heuristics for the prepack optimization problem

In this paper we consider a packing problem arising in inventory allocation applications, where the operational cost for packing the bins is comparable, or even higher, than the cost of the bins (and of the items) themselves. This is the case, for example, of warehouses that have to manage a large number of different customers (e.g., stores), each requiring a given set of items. For this problem, we present Mixed-Integer Linear Programming heuristics based on problem substructures that lead to easy-to-solve and meaningful subproblems, and exploit them within an overall meta-heuristic framework. The new heuristics are evaluated on a standard set of instances, and benchmarked against known heuristics from the literature. Computational experiments show that very good (often proven optimal) solutions can consistently be computed in short computing times.