Validation of a schedule-based capacity restraint transit assignment model for a large-scale network

Abstract

This paper describes the application of a capacity restraint trip assignment algorithm to a real, large-scale transit network and the validation of the results. Unlike the conventional frequency-based approach, the network formulation of the proposed model is dynamic and schedule-based. Transit vehicles are assumed to operate to a set of pre-determined schedules. Passengers are assumed to select paths based on a generalized cost function including in-vehicle and out-of-vehicle time and line change penalty. The time-varying passenger demand is loaded onto the network by a time increment simulation method, which ensures that the capacity restraint of each vehicle during passenger boarding is strictly observed. The optimal-path and path-loading algorithms are applied iteratively by the method of successive averages until the network converges to the predictive dynamic user equilibrium. The Hong Kong Mass Transit Railway network is used to validate the model results. The potential applications of the model are also discussed.