Quantifying the heterogeneity of direct and indirect effects of a public transport disruption with agent-based simulation

Abstract

This paper studies and quantifies the direct and indirect effects of a disruption in the public transport network on passengers using agent-based simulation. In particular, we study the behavior of agents in the network to measure the spatial and temporal extent of the impacts (delays, disutility) of a disruption. Due to the dynamic nature of public transport systems, disruption’s impact on a particular part of the public transport network propagates through the network in both time and space dimensions. Besides, we attempt to evaluate passengers’ behavior in as realistic as possible scenarios, where information about the disruption is scarce, or the disruption is even completely unexpected, and overcome the difficulties of a real-life passenger’ behavior simulation. For such an aim, we add a new extension to the within-day replanning module in the agent-based simulation for public transport (MATSim). We apply our agent-based simulation to the case of the public transport system of Zürich, Switzerland. Our simulation approach quantifies precisely the number of directly and indirectly affected agents by the disruption, respectively those passengers who cannot carry their trip as planned because the services are disrupted and need to reroute their trip in the network; and those passengers whose services are not disrupted, but experience additional crowding effects due to the rerouted, directly affected, passengers. For both groups of travelers, we also study the delay that they experience, and the variation in their utility score of traveling. We prove how those effects relate to a large spatial and temporal heterogeneity, and moreover, they depend strongly on the information available, and replanning actions that the agents might undertake.