PhD Candidate: Khue Nguyen
Start: Fall 2024
Serving as the backbone of every railway system, train timetables influence not only passenger and freight travelling plan and time management, but also subsequent planning stages in railway operation such as platforming, rolling stock scheduling, and crew scheduling. In countries whose rail networks employ multiple dependent routes, operators typically use cyclic timetables, that is, a schedule that repeats itself every cycle time. The Dutch railway functions on an hourly timetable, or the Basic Hourly Pattern (BHP), during which around 300 train services are operated. On a daily basis, they facilitate over 1.1 million passengers with more than 5200 trains.
This intensive schedule raises the need for maintenance activities such that malfunctioning on railway infrastructure is prevented. During the time where maintenance occurs, part of the railway infrastructure becomes unavailable, meaning that the nominal timetable is also infeasible and needs to be adjust. At Netherlands Railway (NS), the current timetabling adjustment procedure to adapt a maintenance activity takes place 26 weeks prior to its scheduled date. The adjustment is manually done by one or two planners, and the decisions are typically made through their expertise without the help of any solvers or computational tools. This, in combination with the complexity of the Dutch rail system, results in a six- or seven-week procedure just to adjust the BHP.
The goal of our research is to study the train timetabling adjustment problems in the presence of maintenance activities. In particular, we aim to develop mathematical models and computational tools to support the decision-making process such that better schedules are obtained in shorter time. Our approaches are applied on the Dutch railway instances to assess its performance in real-world cases. Our most recent model combines re-timetabling with track assignment can find optimal solutions in reasonable times.
