Regional express delivery network planning: A location-routing model and two-tier adaptive GA

Abstract

Modern express delivery system has the ability to deliver mail thousands of kilometers away to the rural doorstep within a few days. This ability depends largely on the robustness and efficiency of regional express delivery networks for distributing and collecting packages. This study aims to minimize the total cost of a regional express delivery network by developing a location-routing model with novel structures. Routing depots are considered non-hub nodes within a hub location network, while external interfaces are introduced as gateways for inter-regional mail exchange. To reduce the model scale, a decomposition approach is proposed, which divides the model into a master problem and a cluster of sub-problems. Furthermore, a two-tier adaptive GA is designed for the sub-problems. Numerical experiments simulate the SF Express operation in the main urban area of Beijing. Computational results show that: 1) The decomposition approach effectively addresses a real-scale problem by transforming it into some small-scale ones. 2) The two-tier adaptive GA achieves high effectiveness in tactic and temporal efficiency compared with the branch and bound method. 3) The proposed model is robust in terms of rents, fuel costs, discount factors and the quantity passing through external interfaces.