Xiangyong Li , Jieqi Li, Y.P. Aneja, Zhaoxia Guo and Peng Tian

Recommend reason

The emerging electronic retailing (e-tailing) trend has led to a boom in the number of e-orders that e-tailers need to process. Relative to traditional retailers, e-tailers are facing a severe challenge on efficiently and effectively accomplishing the task of order fulfillment. This paper shows us how to find the best way to fulfill each customer’s order to minimize the transportation cost. The author presents a mixed integer programming formulation firstly and then introduces an adaptive large neighborhood search-based approach for this problem. The approach can produce high-quality solutions in short computing times.

About the author

Xiangyong Li: School of Economics and Management，Tongji University, professor. The main research directions are: designation and optimization of service network, data driven operation optimization, revenue management and pricing.

Jieqi Li：School of Economics and Management, Tongji University

Y.P. Aneja：Odette School of Business, University of Windsor

Zhaoxia Guo：Business School, Sichuan University

Peng Tian：Antai College of Economics & Management, Shanghai Jiao Tong University

Keywords

e-tailing; order allocation; order routing; order fulfillment; adaptive large neighborhood search

Brief introduction

This paper studies the integrated order allocation and order routing problem for e-order fulfillment. Order fulfillment is defined as the steps involved in receiving, processing and delivering orders to the end customers. Different from the traditional store setting, in an online setting, there is always a delay between the time one order is placed and the time the order is delivered. This time delay is also called the window of decision opportunity. If it doesn’t rush to make immediate allocation of inventories to every order after receiving it, the e-tailer will collect a group of orders and take advantage of the window of decision opportunity to make a better partition of orders.

The explosive growth of e-tailing has generated significant volumes of personal deliveries. It has been the consensus of the public, industry and officials at all levels of the government that it is desirable to reduce the impact of freight transportation on the city living conditions, in particular to reduce congestion and pollution as well as to increase mobility, without penalizing activities in urban areas. As a result, large trucks are banned from entering the central areas of the cities, and e-tailers’ fulfillment centers (FCs) storing products, are located on the outskirts of urban zones. The direct distribution is not an economic fulfillment mode for e-tailers facing government regulations. The e-tailers in China choose to deliver products in orders of end customers through “relay-style” transportation. In particular, urban vehicles move products to distribution centers (DCs) in urban areas, possibly by using corridors (sets of streets) specially selected to facilitate access to DCs and reduce the impact on traffic and the environment, whereas, green and environment-friendly street vehicles travel from DCs to perform delivery of products to end customers. So the author decomposes the new order fulfillment problem into two interconnected subproblems: (1) determining allocation of FCs’ inventories to match DCs’ consolidated demands, and routing urban vehicles (the first-level subproblem), and (2) assigning orders to the DCs and constructing delivery routes of street vehicles (the second-level subproblem). The e-tailer is able to reduce the total operation cost and improve the fulfillment efficiency by reevaluating the allocation assignment of FCs to orders and routing decisions on these orders in the window of decision opportunity.

The author presents a straightforward flow-based integer programming formulation for this problem. The first-level subproblem is processed by a greedy method, and the second-level subproblem is handled by the operators of the adaptive large neighborhood search (ALNS). The results show that the proposed heuristic is very efficient, by benchmarking its performance against a commercial solver and a greedy heuristic.