Recent Progress in Multi-State Supernetwork Modeling for Travel Demand Analysis

This talk includes three model extensions developed with co-workers centered at multi-state supernetwork modeling for travel demand analysis. Motivated by the potentially large share of shared autonomous vehicles (SAVs) in the future, the first topic concerns an activity-based bi-level system optimal model inclusive of a hub-based relocation strategy to moderate the supply and demand of SAVs, in which the lower-level captures the travelers’ activity-travel scheduling behavior by a tolerance-based dynamic activity-travel assignment (DATA) model and the upper-level determines the fleet size, initial distribution, and hubs of SAVs. To address the difficulty of scalability in the family of DATA models, the second topic discusses an improved column generation algorithm for speeding-up DATA. The algorithm adopts a varied temporal resolution scheme, combining exploration and exploitation strategies, to assign flows to narrow time regions rather than to the whole time horizon. The third model extension is about a new framework of a dynamical activity-travel rational adjustment process in multi-state supernetworks based on a needs-based theory. This formalism couples dynamic activity generation, activity-travel scheduling, and traffic flow evolution in a strong sense, and offers appealing explanations for day-to-day traffic dynamics and equilibria.