Evaluating the flooding level impacts on urban metro networks and travel demand: behavioral analyses, agent-based simulation, and large-scale case study

With urban residents’ increasing reliance on metro systems for commuting and other daily activities, extreme weather events such as heavy rainfall and flooding impacting the metro system services are becoming increasingly of concern. Plans for such emergency interruptions require a thorough understanding of the potential outcomes on both the system and individual component scales. However, due to the complex dynamics, constraints, and interactions of the elements involved (e.g., disaster, infrastructure, service operation, and travel behavior), there is still no framework that comprehensively evaluates the system performance across different spatiotemporal scales and is flexible enough to handle increasingly detailed travel behavior, transit service, and disaster information data. Built on an agent-based model (ABM) framework, this study adopts a data-driven ABM simulation approach informed by actual metro operation and travel demand data to investigate the impact of flood-induced station closures on travelers as well as the overall system response. A before-after comparison is conducted where the traveler behaviors in disaster scenarios are obtained from a discrete choice model of alternative stations and routes. A case study of the Shanghai Metro is used to demonstrate the ability of the proposed approach in evaluating the impacts of flood-induced station closures on individual traveler behavior under normal operation and a series of water level rise scenarios of up to 5m. It was found that, when the flood-induced station closures only affect a few river-side stations in the city center, the travelers experience only minor disruptions to their trips due to the availability of unaffected stations nearby as a backup. However, as the water level increases and more stations (mainly in the suburban area) are affected, up to 25% of trips are no longer being fulfilled due to the loss of entrances, exits, or transfer links. The system experiences overall less crowdedness in terms of passenger volume and platform waiting time with a few exceptions of increased passenger load due to concentrations of passenger flows to alternative stations under flooding-induced station closures. The proposed approach can be adapted to other disaster scenarios to reveal the disaster impacts on both aggregated and disaggregated levels and guide the design of more spatio- and temporally-targeted emergency plans for metro systems.