From vehicle-based punctuality to passenger-based reliable rapid transit

Abstract

Literature describes vehicle punctuality or line regularity, while a fair share of the passengers use multiple vehicles or lines in their journey. This neglects the transfer time that is needed for waiting from one vehicle to the next. Furthermore, the trade-off between holding time and the different components of a full passenger journey are often unclear. These shortcomings in assessing transit timetables are the research gap this study aims to close. To assess the influence of including transfers in evaluating timetable quality from a passenger perspective, this study developed a passenger-centric methodology that explicitly models transfer time, as well as waiting time and in-vehicle time. The demand for transfers is derived by a combined dataset of within-line demand and transfer demand, both originating from smartcard data. AVL data is used for simulating vehicle trips, which are stored in a network-time graph. Dynamic programming is used to efficiently calculate the journey time for all passengers. Across simulation runs, for every passenger, the 80th percentile of travel time is used to cover both the experienced speed and reliability. Summing this Rider-based Rapidness and Reliability (RRR), the performance of multiple holding strategies was compared for a case study in Groningen, The Netherlands.

Stakeholder perspectives on a timetable quality indicator that includes transfers were gathered through a focus group session. Findings suggest that holding strategies with more slack time perform worse when transfers are included, compared to when they are excluded. Hence, the main takeaway of this study is that transfer times should not be neglected in timetable-related research, as they influence results. Including all transfers that connect a study area with a cordon can yield different outcomes for the performance of holding control strategies. The root cause of these differences lies in the assumed passenger arrival patterns at transfer stops, which are typically clustered around scheduled times rather than uniformly distributed. Stakeholders acknowledged the importance of transfer time and agreed that a demand-oriented indicator is needed to better capture full journey experiences. Since every network differs in terms of transfer dependency, the generalizability of these findings is limited. Future research should focus on integrating transfers into agent-based models to account for capacity constraints, bunching, and elasticity effects, enabling a more dynamic analysis of transit systems over time.