Computing Optimized Stock (Re-)Placements in Last-In, First-Out Warehouses

Within this work, we focus on the optimization of storage location assignments arising in warehouses with storage locations applying a last-in, first-out throughput policy. The sequence of goods to be stored is, however, not entirely known such that for each item the currently best storage location has to be identified almost immediately. Caused by this imperfect data and by stock removals concurrently performed it is necessary to apply relocation operations from time to time, which might range from a few operations to relocations lasting a working day depending on the workload of the warehousemen. For this purpose we propose an ad hoc stocking strategy as well as a storage relocation strategy based on variable neighborhood descent. Supported by experimental tests we compare variants of our approaches with each other and with formerly used stocking strategies showing that the number of conflicts could be significantly reduced by the proposed approach. Furthermore, an application of the relocation strategy can significantly improve warehouse states obtained due to imperfect stocking strategies, concurrently performed stock removals and insufficient information on production sequences.