Advanced passenger information system : modelling and optimization

Abstract

In Hong Kong, over 90% of people use transit facilities for daily travel. The dense concentration of people in urban areas puts high pressure on existing bus networks and railways facilities. Congestion is common place particularly during peak periods. Alleviation could be brought about through a reassessment of the transit system performance and the implementation of the Advanced Passenger Information System (APIS) for provision of updated transit information to help passenger optimize their route choices before and/or during their travel. In this study, the interaction of transit network performance, such as transit service reliability by transit modes, and passenger route choice behavior is considered explicitly in a new transit network assignment model. The new model for resolving transit passenger flow patterns takes into account the attributes of APIS and in addition incorporates passenger response to APIS implementation when choosing travel routes. This research is, to the author's knowledge, the first devoted exclusively to evaluating the effects of public transport information on passenger route choice behavior and system performance, allied with the integration of passenger response to the transit service reliability. In this study, an approach is proposed to estimate a passenger origin-destination (O-D) matrix, using a partial set of observed passenger link flows and previous passenger O-D matrix under congested transit conditions. This study extends the literature on transit passenger O-D estimation problems. The effects of Pre-trip Transit Information System (PTIS) on passenger pre-trip planning and route choice behavior are considered. In other words, the study extends the investigation of the passenger O-D estimation problem to include a multi-class dimension. A bi-level programming model is proposed to obtain the updated passenger O-D matrix and simultaneous passenger flows by link. In addition, the En-route Transit Information System (ETIS) has been evaluated. It is assumed that there are three parties involved in ETIS, the users, the bus operators and the Government, each with a different objective. The passengers aim to maximize travel time saving by buying transit information, the bus operator aims to increase total revenue and the government aims to improve transit network performance. This study has attempted to investigate the effects of ETIS on each party, in congested transit networks, with elastic market penetration. A new bi-level programming model is proposed to investigate these effects. In the model the total transit passenger demand is fixed but the potential market penetration of ETIS has been formulated elastically. This study also has proposed a new probit-type reliability-based transit assignment model in a congested network which has unreliable transit services. A new disutility function has been developed to model the passenger route choice behavior under unreliable conditions, in which passengers make a trade-off between the transit service reliability and the expected route choice travel time. In the model, the total transit passenger demand is fixed but the in-vehicle travel times by transit line segments are formulated stochastically owning to unreliable traffic conditions on the road networks. The effects of congestion on transit network performance and passenger route choice behavior are explicitly considered.