The order picking problem under a scattered storage policy

When warehouses are operated according to a scattered storage policy, each Stock Keeping Unit(SKU) is stored at multiple locations inside the warehouse. Such a configuration allows for improved picking efficiency, as now an SKU can be picked from the location that is most compatible with the other SKU’s in the picking batch. Seizing these benefits, however, comes at the cost of additional decisions to be made while planning the picking operations. Next to determining the sequence in which SKU’s will be retrieved from the warehouse, the location at which each SKU needs to be extracted has to be chosen by the planner. In this paper, we model the order picking problem under a scattered storage policy as a Generalized Travelling Salesperson Problem (GTSP). In this problem, the vertices of the underlying graph are partitioned into clusters from which exactly one vertex should be visited in each cluster. In our order picking application, each cluster contains all product locations of a single SKU on the order list. The aim is to design a pick tour that visits all product locations of the SKU’s on the pick list (i.e., visit each cluster exactly once) and minimizes the total travel distance. We present an ILP formulation of the problem and a variable neighbourhood heuristic, embedded in a guided local search framework. The performance of both methods is tested extensively by means of computational experiments on benchmark instances from the literature.