A modelling platform for optimizing time-dependent transit fares in large-scale multimodal networks

With the continuous growth of urban areas around the world, overcrowding in large transit networks has become a persistent problem, with far-reaching impacts similar to those caused by congestion in large road networks. Moreover, instead of serving as a relief for large transportation systems, congested transit networks have increased delay-related traffic and transit costs. In light of these problems, cities seek cost-effective and relatively fast-to-implement strategies to mitigate transit system congestion, one of which is time-based fare structures. By implementing time-based fares, the transit demand may shift out of the congested peak periods, easing transit travel conditions. Although time-based fares are already in use in some transportation systems, their implementation is usually based on simplified what-if analyses. Such analyses of fare structures in previous studies have lacked a comprehensive evaluation of people's responses to these fares and is usually applied to simple or sometimes hypothetical transportation networks. Therefore, this paper presents a platform for analyzing and optimizing time-based transit fares in large networks, taking into consideration the effects of these fares on people's choices of mode, departure time, and route in addition to the interactions between transit vehicles and general traffic. As a case study, the largest metropolitan area in Canada, the Greater Toronto Area, is tested. The results show that the optimal time-based fares help spread the transit demand to the shoulders of the peak. However, the savings in weighted average multimodal door-to-door travel time over the whole network are slightly small compared to the large increase in peak-hour fares.