A Hypergraph Model for the Rolling Stock Rotation Planning and Train Selection

This paper is about an integrated optimization approach for timetabling and rolling stock rotation planning in the context of passenger railway traffic. Given a set of possible passenger trips, service requirement constraints, and a fleet of multiple heterogeneous self-powered railcars, our method aims at producing a timetable and solving the rolling stock problem in such a way that the use of railcars and the operational costs are minimized. To solve this hard optimization problem, we design a mixed-integer linear programming model based on network-flow in an hypergraph. We use this models to handle effectively constraints related to coupling and decoupling railcars. To reduce the size of the model, we use an aggregation and disaggregation technique combined with reduced-cost filtering. We present computational experiments based on several French regional railway traffic case studies to show that our method scales successfully to real-life problems