A multi-criterion multi-class dynamic transit trip assignment model

Abstract

A dynamic transit trip assignment model developed for the operational planning of an existing metro system has the following salient features:(1) use a schedule-based network formulation to model the movement of trains; (2) use multi-interval assignment (15-minutes duration for each time interval) to predict train loading; (3) assume a multivariable linear path cost model that includes in-vehicle time, waiting time, walking time and line change penalty; (4) divide passengers into different classes based on their route choice preferences; (5) calibrate the model using passenger boarding and alighting flows at station platforms; (6) validate the model using cordon counts. As a case study, the model is applied to the Hong Kong Mass Transit Railway network for a 4-hour period from 6 a.m. to 10 a.m. The input data consists of full sample multi-interval station-to-station O-D trip matrices compiled from data retrieved from ticketing gates. Multi-class assignments for single period and multi-period are implemented and a calibration procedure estimates the proportion of different user classes of travelers to be assigned to the network.