The dynamic stocking location problem – Dispersing inventory in fulfillment warehouses with explosive storage

Online order fulfillment warehouses (F-Warehouses) are a critical component of the physical internet. F-Warehouses convert online customer orders into physical delivery packages with an operational goal of fast fulfillment. Structurally F-Warehouses are unlike traditional warehouses and key differentiators are, a very large number of small bin locations, an explosive storage policy, and commingled bin storage. In explosive storage, an incoming bulk is separated into small lots which can be stocked in any bin throughout the warehouse. As orders flow in, this inventory dispersion increases the probability that a picker can locate multiple ordered items within a short distance. The dynamic stocking decision then is, where to stock the arriving inventory lots, such that picking algorithms can generate more efficient picklists and achieve the fast fulfillment goal. The paper presents first, an objective function for the dynamic stocking location problem and then formulates it as a mixed-integer program. The objective is a function of the weighted order fillability of each bin and models’ item-to-item order correlations. Two heuristics that solve the problem in a time-efficient manner are developed. Both heuristics performed well, with one providing a solution within 10% of the MIP solution 82% of the time. Using a database driven simulator, experimental runs were conducted on a problem with 22,000 orders. The results show the proposed stocking objective and associated heuristic reduced fulfillment time by 6.69%–11.19% relative to a segmented random rule. Performance improved by as much as 3.4% when the stocking bandwidth was increased from 20 to 40, while fulfillment time declined consistently with an increasing explosion ratio.