Seamless urban metro networks: A shared depot perspective on rolling stock scheduling

Publication Type

Journal Article

Publication Date

12-2025

Abstract

This study focuses on a novel integrated optimization problem that involves train timetables and rolling stock circulation plans in urban metro networks that features shared rolling stock depots for multiple interconnected metro lines. An integer programming model is developed to determine train dispatch times, the allocation of rolling stock resources at depots, rolling stock connection, and passenger assignment. The model considers objectives that encompass both passenger service levels and operational costs. To address the computational complexity of large-scale problems, two exact solution approaches are developed. (1) A Benders decomposition algorithm decomposes the proposed model into a rolling stock scheduling problem and a set of independent passenger assignment subproblems. Upon analyzing the theoretical properties of the integer subproblems, these subproblems are replaced by their linear relaxation problems to enhance computational efficiency. (2) An improved Benders decomposition algorithm, which incorporates novel optimality cuts, is designed to accelerate the solution process. Numerical experiments using real-world data from the Xi’an Metro validate the effectiveness of the proposed approaches. Computational results demonstrate that the improved Benders decomposition algorithm consistently yields a high-quality solution for the whole-day case and outperforms the traditional Benders decomposition method and CPLEX solver. Compared with independent depots for each metro line, the shared depot strategy yields a noteworthy 6.8 % reduction in the number of carriages that put into operation. This reduction highlights key improvements in resource utilization, and overall operational effectiveness.

Discipline

Artificial Intelligence and Robotics | Transportation

Research Areas

Intelligent Systems and Optimization

Publication

Transportation Research Part B: Methodological

Volume

204

First Page

1

Last Page

31

ISSN

0191-2615

Identifier

10.1016/j.trb.2025.103379

Publisher

Elsevier

Additional URL

https://doi.org/10.1016/j.trb.2025.103379

This document is currently not available here.

Share

COinS