最小化拆单率的在线零售商多仓商品摆放优化策略研究 本期目录 >>
Title: Inventory Allocation Policy to Distribution Centers with Minimum Split Orders in Online Retailing
作者 李建斌;李乐乐;黄日环
Author(s): LI Jianbin; LI Lele; HUANG Rihuan
摘要: 拆单是指一张订单需要从多个仓库独立进行配送,或由一个仓库在不同时间分别配送,而这会额外增加商家的物流成本。本文对在线零售商多个仓库的商品摆放策略进行了研究,目的是降低由于商品摆放策略不同所引起的拆单。文中考虑了总体不缺货情况下,基于品类和数量的两类拆单情况,并建立数学模型,改进了热销品算法(Bestseller)得到了环形算法(Loop),对相应仿真问题进行求解,并与原热销品算法进行对比。结果表明环形算法能够在兼顾各配送中心负载的同时进一步降低品类拆单率,进而降低总体拆单率。备货量的分析表明,在线零售商可以通过增加备货量来降低甚至消除数量拆单的影响。对品类约束的研究表明,在单仓商品品类较少时,品类拆单对总的拆单率影响更大;单仓商品品类较多时,数量拆单影响更大。效率方面,由于环形算法不需要计算相关性矩阵,效率相比热销品算法要高。
Abstract: Order splitting refers to a retailing order is delivered in multiple shipments, which adds to fulfillment cost. The effect is much more significant while a retailer holds a large number of SKUs. In this paper, SKU allocation policy to multiple DCs in online retailing is explored to minimize amount of split orders. Split orders due to SKU and due to unit are both considered in our research. An integer formulation of split orders problem is given. A new algorithm Loop is designed and compared with the benchmark. The result shows that the algorithm performs well in reducing split orders, workload balance of DCs and efficiency. Further analysis shows that split orders due to unit can be minimized or even eliminated when online retailer keeps higher inventory level. And discussion on quantity of SKUs points out that more orders are split due to SKUs when distribution centers hold relatively less SKUs while order splitting due to unit becomes significant when more SKUs are held in each one distribution center. As for efficiency of computation, the Loop algorithm is more efficient without requirement for coappearance matrix Firstly, based on literature review, we transfer minimizing operation cost to minimizing split orders. And three main factors lead to split orders are given. The first is SKUs that a certain order demands are not stocked in one distribution center (DC), which is called split orders due to SKU. The second is inventory level of certain SKUs can not fulfill such an order, which is called split orders due to unit. The third is order will be split because of particular SKUs such as pre-sale products. Order splitting can be minimized by optimizing SKU allocation policy, which is to decide SKU allocation to DCs and maximum inventory level of each SKU in each DC that the SKU is allocated. An integer formulation considering both two factors is given and proved as NP-hard. Secondly, according to analysis of sales data from an online retailer in China, we found that a minority SKUs account for majority sales. Therefore SKUs can be defined as bestseller and normal SKUs. Allocating bestsellers to all the DCs can minimize split orders and balance the workload. With such an idea, Loop Algorithm is designed. Thirdly, to test the efficiency of Loop Algorithm, a numerical example is given. We use simulated orders based on analysis of sales data mentioned above to decide SKU allocation and calculate split orders. Three algorithms are compared: Loop Algorithm, Bestsellers designed by Catalan and Fisher and Sales as benchmark. The comparison includes two aspects: split rate and workload balance of DCs. Split rate refers to percentage of split orders. The result shows that Loop performs the best in both aspects. Fourthly, to distinguish two kinds of split orders, a further discussion is given. Based on sensitiveness analysis, feasible quantity of both SKUs and units is given considering reasonable holding cost. More qualitative conclusion is given according to numeric experience. In summary, split orders is a newly raised problem with rapid growth of online retailing and it will receive more concern in the future because how to minimize fulfillment cost has been one of the most important issue to online retailers.
关键词: 在线零售;运营管理;拆单;启发式算法
Keywords: online retailing; operation management; split orders; heuristics
基金项目: 国家自然科学基金资助项目;国家自然科学基金资助项目;中央高校基本科研业务专项基金;湖北省科技计划软科学类项目;教育部新世纪优秀人才基金
发表期数: 2017年 第3期
中图分类号: 文献标识码: 文章编号:
参考文献/References:

[1] Xu P J. Order fulfillment in online retailing: What goes where[D]. Massachusetts Institute of Technology, 2005.

[2] Amazon.com, Inc. SEC form 10-K for the fiscal year ended December 31, 2013. http://goo.gl/QF8cRQ.

[3] Amazon.com, Inc. SEC form 10-K for the fiscal year ended December 31, 2011. http://goo.gl/ifctT.

[4] Xu P J, Allgor R, Graves S C. Benefits of reevaluating real-time order fulfillment decisions[J]. Manufacturing and Service Operations Management, 2009,11(2):340-355.

[5] Jane C, Laih Y. A clustering algorithm for item assignment in a synchronized zone order picking system[J]. European Journal of Operational Research, 2005,166(2):489-496.

[6] Parikh P J, Meller R D. Selecting between batch and zone order picking strategies in a distribution center[J]. Transportation Research Part E: Logistics and Transportation Review, 2008,44(5):696-719.

[7] Küçükaydin H, Aras N, Kuban Altınel I. Competitive facility location problem with attractiveness adjustment of the follower: A bilevel programming model and its solution[J]. European Journal of Operational Research, 2011,208(3):206-220.

[8] Sun H, Gao Z, Wu J. A bi-level programming model and solution algorithm for the location of logistics distribution centers[J]. Applied Mathematical Modelling, 2008,32(4):610-616.

[9] Awasthi A, Chauhan S S, Goyal S K. A multi-criteria decision making approach for location planning for urban distribution centers under uncertainty[J]. Mathematical and Computer Modelling, 2011,53(1):98-109.

[10] Pan J C, Shih P. Evaluation of the throughput of a multiple-picker order picking system with congestion consideration[J]. Computers & Industrial Engineering, 2008,55(2):379-389.

[11] Parikh P J, Meller R D. Estimating picker blocking in wide-aisle order picking systems[J]. IIE Transactions, 2009,41(3):232-246.

[12] Catalán A, Fisher M L. Assortment allocation to distribution centers to minimize split customer orders. The Wharton School, University of Pennsylvania, Working Paper, 2012.

[13] Feige U, Krauthgamer R, Nissim K. Approximating the minimum bisection size (extended abstract): Proceedings of the thirty-second annual ACM symposium on Theory of computing[Z]. STOC. ACM, 2000: 530-536.

[14] Karp R M. Reducibility among combinatorial problems[M]. Springer, 1972. 85-103

[15] Andreev K, Racke H. Balanced graph partitioning[J]. Theory of Computing Systems, 2006,39(6):929-939.

[16] Chen L, Hsu F, Chen M, et al. Developing recommender systems with the consideration of product profitability for sellers[J]. Information Sciences, 2008,178(4):1032-1048.