A Heuristic Approach to the Design of Optimal Cross-Docking Boxes

Multinational companies frequently work with manufacturers that receive large orders for different products (or product varieties: size, shape, color, texture, material), to serve thousands of different final destinations (e.g., shops) requesting a combination of different quantities of each product. It is not the manufacturers' task to create the individual shipments for each final destination. But manufacturers can deliver part of their production in so-called cross-docking boxes (or other containers) of a few+ (say, three) types, each type containing a given assortment, i.e., different quantities of different products. At a logistics center, the shipments for each destination are then made of cross-docking boxes plus additional "picking" units. The expensive part is the picking, since cross-docking boxes require little or no manipulation. The problem we solve in this paper is, given a large set of orders for each destination, to design the cross-docking box types in order to minimize picking. We formally define a variant of this problem and develop a heuristic method to solve it. Finally, we present extensive experimental results on a large set of real-world benchmarks proving that our approach gives high-quality solutions (optimal or near optimal) in a very limited amount of time.