A Novel Scattered Storage Policy Considering Commodity Classification and Correlation in Robotic Mobile Fulfillment Systems

The commodity storage assignment problem (CSAP), which assigns stock-keeping units (SKUs) to a suitable location for matching the customer demand patterns, is crucial for improving the order picking efficiency. In this study, we jointly consider the SKUs classification and correlation, and propose a new scattered storage policy named scattered-correlation storage policy based on the commodity classification (SCSPCC) for mitigating CSAP in the robotic mobile fulfillment systems (RMFS). We call the new problem CSAP-SCSPCC. To address this problem, we construct a mixed-integer programming model, and propose a novel variable neighborhood search with self-adaption and simulated annealing acceptance mechanisms (SA-VNSSA). Besides, a heuristic algorithm is proposed to select the minimum number of shelves to evaluate the optimization effect of SA-VNSSA and SCSPCC in terms of the number of shelf transports. Extensive numerical experiments are conducted on small-, medium-, and large-scale instances, respectively. The results reveal that the proposed model and algorithms are reasonable and effective in solving CSAP-SCSPCC compared with the state-of-the-art methods. Specifically, SA-VNSSA outperforms the three state-of-the-art comparison algorithms [i.e., SA-1 (Muppani and Adil, 2008), SA-Pop (Assadi and Bagheri, 2016), and SA-2 (Zhang et al., 2019)] by more than 4.19% and 3.23% on average in medium- and larger-scale instances, respectively. The comparisons between SCSPCC and CDSAP (Mirzaei et al., 2021) and DCP (Zhang et al., 2019) show that the order picking efficiency is improved by our SCSPCC more than 6.31%. It is can be concluded that SCSPCC is efficient and robust to match the SKU storage pattern and customer demand patterns in e-commerce retail.