Rescheduling Railway Traffic on Real Time Situations Using Time-Interval Variables

• Published in: Integration of AI and OR Techniques in Constraint Programming Vol. 10335; pp. 312 - 327
• Main Authors: Cappart, Quentin, Schaus, Pierre
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2017; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: First Come First Serve; Freight Train; Improvement Ratio; Route Activity; Track Segment
• ISBN: 9783319597751; 3319597752
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-59776-8_26

In the railway domain, the action of directing the traffic in accordance with an established timetable is managed by a software. However, in case of real time perturbations, the initial schedule may become infeasible or suboptimal. Subsequent decisions must then be taken manually by an operator in a very limited time in order to reschedule the traffic and reduce the consequence of the disturbances. They can for instance modify the departure time of a train or redirect it to another route. Unfortunately, this kind of hazardous decisions can have an unpredicted negative snowball effect on the delay of subsequent trains. In this paper, we propose a Constraint Programming model to help the operators to take more informed decisions in real time. We show that the recently introduced time-interval variables are instrumental to model this scheduling problem elegantly. We carried experiments on a large Belgian station with scenarios of different levels of complexity. Our results show that the CP model outperforms the decisions taken by current greedy strategies of operators.