Solution algorithms for the generalized train unit shunting problem

Abstract

This paper proposes different algorithms to tackle the Generalized Train Unit Shunting Problem (G-TUSP). This is the pre-operational problem of managing rolling stock in a station, between arrivals and departures. It includes four sub-problems: the Train Matching Problem, the Track Assignment Problem, the Shunting Routing Problem, and the Shunting Maintenance Problem. In our algorithms, we consider different combinations for the integrated or sequential solutions of these sub-problems, typically considered independently in the literature. We assess the performance of the algorithms proposed in real-life and fictive instances representing traffic in Metz-Ville station, which includes four shunting yards. It is a main junction between two dense traffic lines in the east of France. In a thorough experimental analysis, we study the contribution of each sub-problem to the difficulty of the G-TUSP, and we identify the best algorithms. The outcomes of our algorithms are superior to solutions manually designed by experienced railway practitioners.