An utility-based storage assignment strategy for e-commerce warehouse management

Abstract

Order picking is one of the most labor intensive and time consuming parts of e-commerce warehouse order management. An effective solution to the item storage assignment problem can significantly improve the efficiency of warehouse operations and improve customer service levels. The problem, however, is quite challenging due to distinct item characteristics (volume, velocity, variety), multiple item relationships (co-occurrence patterns, complementarity, compatibility and substitution effects), layout configurations (single vs multi-block, pure vs mixed-shelves, dedicated vs scattered storage), order-picking strategies (stand-alone, batched, zoned) and routing strategies (traversal, return, midpoint) and is NP hard. Several heuristic approaches have been proposed in the literature to address the storage assignment problem. More recent and state-of-the-art approaches to the problem have considered item relationships to improve the efficiency of order picking. We argue that such methods are largely myopic in nature and consider just pair-wise item correlations and also ignore quantities of items in an order. Moreover, current methods make several simplifying assumptions about layout configurations, order-picking and routing strategies. In this paper, we propose a novel utility-based storage assignment strategy for the item storage assignment problem. It uses a top-k high utility itemset mining method and a heuristic algorithm for effective storage assignment. We conduct rigorous experimental evaluation of the proposed idea and compare it against other related methods. The experimental results on wide variety of layout configurations, routing strategies and order distributions show very promising results for the proposed storage assignment strategy. The findings of this study are likely to be useful for warehouse managers in improving operational efficiencies and delivering better customer service levels.