Combining robustness and recovery in rapid transit network design

Abstract

When designing a transport network, decisions are made according to an expected value for network state variables, such as infrastructure and vehicle conditions, which are uncertain at a planning horizon of up to decades. Because disruptions, such as infrastructure breakdowns, will arise and affect the network on the day of operations, actions must be taken from the network design. Robust network designs may be implemented but they are extremely expensive if disruptions do not realise. In this paper, we propose a novel approach to the network design problem where robustness and recovery are combined. We look for the trade-off between efficiency and robustness accounting for the possibility of recovering from disruptions: recoverable robust network design. Computational experiments drawn from fictitious and realistic networks show how the presented approach reduces the price of robustness and recovery costs as compared to traditional robust and non-robust rapid transit network design approaches.