Underground logistics system (ULS) is recognized as sustainable alleviator to road-dominated urban logistics infrastructure with various social and environmental benefits. The purpose of this study is to propose effective modeling and optimization method for planning a hub-and-spoke ULS network in urban region.
Underground freight tunnels and the last-mile ground delivery were organized as a hierarchical network. A mixed-integer programming model (MIP) with minimum system cost was developed. Then a two-phase optimization schema combining Genetic-based fuzzy C-means algorithm (GA-FCM), Depth-first-search FCM (DFS-FCM) algorithm and Dijkstra algorithm (DA), etc. was designed to optimize the location-allocation of ULS facilities and customer clusters. Finally, a real-world simulation was conducted for validation.
The multistage strategy and hybrid algorithms could efficiently yield hub-and-spoke network configurations at the lowest objective cost. GA-FCM performed better than K-means in customer-node clustering. The combination of DFS-FCM and DA achieved superior network configuration than that of combining K-means and minimum spanning tree technique. The results also provided some management insights: (1) greater scale economies effect in underground freight movement could reduce system budget, (2) changes in transportation cost would not have obvious impact on ULS network layout and (3) over 90% of transportation process in ULS network took place underground, giving remarkable alleviation to road freight traffic.
Demand pairs among customers were not considered due to lacking data. Heterogeneity of facilities capacity parameters was omitted.
This study has used an innovative hybrid optimization technique to address the two-phase network planning of urban ULS. The novel design and solution approaches offer insights for urban ULS development and management.
