Integrated Microsimulation Modelling of Crowd and Subway Network Dynamics For Disruption Management Support

In many large cities around the world, public transit networks have been carrying an ever-increasing burden of commuters. This has resulted in large movements of crowds in major transit hubs, and high levels of congestion in systems running near or at capacity. In such situations, service disruptions can negatively impact service and transit users well after they are resolved.;This dissertation presents Nexus, an innovative crowd dynamics and transit network simulation platform, that enables full simulation of all actors in the transit system and integrated dynamic transit assignment, while being both flexible and scalable to handle large-scale networks. Instead of developing new simulators for surface transit, trains and stations, Nexus enables interfacing of existing simulators together to form a network. An agent-based framework was overlaid to allow for responsive agents, while a virtual communication system permitted on-the-fly modifications to transit service operation.;With crowd behaviour in stations key to network performance, new models were constructed to better explain passenger behaviour at two critical locations. First, discrete choice models were developed of passenger choice of stairs versus escalators. Second, a simulation-based model of passenger behaviour on train platforms was developed using a diffusion-inspired approach and accounting for preferred waiting locations. Field data collected across several Toronto subway stations informed both models.;Finally, a proof-of-concept case study was conducted on the Toronto transit network. The impact of disruptions of various lengths at a key platform in the network was examined, and an illustrative example was conducted to show how the system could be used to test strategies to reduce impact on transit passengers.;Currently, transit agencies handle these disruption situations in an ad-hoc fashion. This is due to the lack of a tool capable of analyzing the network-level impacts of response strategies, especially with performance sufficient to handle large-scale systems. Existing tools are limited in their ability to accurately simulate all dynamic facets of the transit network, or model the behaviour of passengers following disruption events. As a result, they do not adequately capture the two-way interaction between crowds and mass transit service, or the complexities of train operation.