Integration of First Train Timetabling and Passenger Transfer Routing Problems in Urban Rail Transit Networks

Hao Li,Liujiang Kang,Huijun Sun,Zi-You Gao,Nsabimana Buhigiro
DOI: https://doi.org/10.2139/ssrn.4278829
2022-01-01
SSRN Electronic Journal
Abstract:Urban rail transit (URT) dominates metropolitan mobility operations. Large-scale URT networks have various timetabling-related problems during the first train operating periods, such as the long transfer waiting time and multiple transfer problems that result in greatly extended passenger travel time. This paper incorporates the shortest passenger travel paths into the first train timetabling problem in URT networks. We first propose a path-based first train timetabling (PB-FTT) problem, which is formulated as a mixed-integer nonlinear programming model to reduce the total travel time for passengers. To solve this NP-hard problem in a small-scale case with 94 decision variables, we propose a tailored branch-and-bound algorithm combined with the timetable-based Dijkstra algorithm (TD-TB&B), which solves the problem to optimality by 13.6 min, while exhausting 11.2 GB CPU memory. To further improve the efficiency of the exact algorithm, we design a rolling time heuristic (RTH) algorithm to replace the TB&B algorithm (namely TD-RTH algorithm). The TD-TB&B and TD-RTH algorithms are compared and tested in numerical experiments, showing that the TD-RTH algorithm reduces the CPU time by 4.55 min with an optimality gap of 0.12%. Finally, we apply the model and the TD-RTH algorithm to the Beijing URT Network. Results indicate that the PB-FTT model and TD-RTH algorithms can effectively improve the travel efficiency of passengers. Specifically, the total travel time is reduced by 40618 min and the transfer waiting time is decreased by 26.8% compared with the existing timetables.
What problem does this paper attempt to address?