Urban transit network design using spanning tree: A case study of Canberra transit network

TL;DR

This paper introduces a heuristic approach using tabu search to optimize transit network design by minimizing passenger-kilometers, demonstrated through a case study of Canberra's bus network.

Contribution

It presents a novel tabu search heuristic for designing transit networks that efficiently approximates optimal solutions in large-scale scenarios.

Findings

The heuristic effectively finds promising spanning trees for the Canberra network.

Relaxing link constraints can further reduce total passenger-kilometers.

The approach offers policy implications for transit network improvements.

Abstract

Transit network design plays an important role in public transport. With the simplicity of spanning tree, this paper adopts the concept of spanning tree to help (re-)design a public transit network that addresses passenger utility by minimizing the total passenger-kilometers, which can be formulated as a mixed-integer optimization model. However, searching for an optimal minimum passenger-kilometer spanning tree is intractable for a large network. This paper proposes a tabu search based heuristic to quickly output a promising spanning tree. The efficacy of the proposed tabu search based heuristic is verified using the Canberra bus network data. With the flexibility of the proposed tabu search heuristic and the efficiency of the transit system, this paper proposes a greedy algorithm to relax the number of links constraint to add more links that can further minimize the total passenger-kilometers. With the case study of Canberra transit network, this paper provides implications to policy makers to further improve the design of a public transit network.