Title:
Generating clusters for urban logistics in hyperconnected networks
Generating clusters for urban logistics in hyperconnected networks
Author(s)
Hettle, Cyrus
Faugere, Louis
Kwon, Simon
Gupta, Swati
Montreuil, Benoit
Faugere, Louis
Kwon, Simon
Gupta, Swati
Montreuil, Benoit
Advisor(s)
Editor(s)
Collections
Supplementary to
Abstract
In the hyperconnected logistics model, a city is represented as a continuous mesh of small regions called unit zones. The clustering problem is to partition the set of unit zones into larger local cells and urban areas, and is critical in defining network operations. We give a mixed integer programming-based method for solving the clustering problem, which combines aspects of graph partitioning and min-cost flow problems. Our model aims to minimize expected operating cost, accounting for s expenses throughout the network, while incentivizing clusters that are resilient, geographically compact, and have balanced demand. To generate meaningful warm-starts for our MIP and achieve computational speedups, we adapt a graph partitioning method called striping. Solutions for the clustering problem can be integrated with methods for other problems in hyperconnected network design, significantly improving their tractability. Our techniques work effectively in tandem with methods for choosing hub candidate locations and routing flow. We show the effectiveness of our methods in redesigning SF Express’s hyperconnected network in Shenzhen.
Sponsor
Date Issued
2021-06
Extent
Resource Type
Text
Resource Subtype
Paper