Detecting Braess paradox links with a mixed integer linear programme

Abstract

Braess’s paradox is a counterintuitive fact that adding new links to a network can increase travel costs due to routing competition. Real networks may have unreasonably constructed roads that cause Braess paradoxes. Therefore, to improve the performance of the transportation networks, it is a necessary task to identify Braess paradox affected locations and close them. In this paper, we study the Braess paradox detection problem. Given a transportation network, we seek for a set of Braess-tainted roads whose closure will reduce travel cost. The problem is formulated by a mixed integer linear programme that can be solved by employing commercial computing package. A numerical example is demonstrated to show the performance of the programme.