Railway timetabling for the future? A pilot study on a single track line

Abstract

No country uses optimisation methods to establish a value maximising allocation of operators’ demand for scarce railway infrastructure. Using the same input as was used in the year 2020 (T20), when the official timetable for one specific line in Sweden was developed by the infrastructure manager, this paper presents the outcome of the combined use of dual optimisation and rapid branching to produce a solution to an actual optimisation problem. These methods generate a timetable with the same number of trains as in T20 and some qualities that are better than those of the official timetable. The pilot study therefore confirms that it is worthwhile to further develop formal optimisation mechanisms towards larger-scale applications. Being able to handle conflicts-of-interest on a single-track line means that the problem on a double track line has fewer restrictions.