Periodic Timetabling with Integrated Routing: Toward Applicable Approaches

Abstract

Periodic timetabling is an important, yet computationally challenging, problem in public transportation planning. The usual objective when designing a timetable is to minimize passenger travel time. However, in most approaches, it is ignored that the routes of the passengers depend on the timetable, so handling their routing separately leads to timetables that are suboptimal for the passengers. This has recently been recognized, but integrating the passenger routing in the optimization is computationally even harder than solving the classic periodic timetabling problem. In our paper, we develop an exact preprocessing method for reducing the problem size and a heuristic reduction approach in which only a subset of the passengers is considered. It provides upper and lower bounds on the objective value, such that it can be adjusted with respect to quality and computation time. Together, we receive an approach that is applicable for real-world problems. We experimentally evaluate the performance of the approach on a benchmark example and on three close-to-real-world instances. Furthermore, we prove that the ratio between the classic problem without routing and the problem with integrated routing is bounded under weak and realistic assumptions.