Metro crew planning with day-off pattern, duty type, and rostering scheme considerations

The metro crew planning must consider various complex factors in real scenarios, such as day-off requirements, duty types, and rostering rules. The metro crew planning problem is often divided into crew scheduling and rostering problems, which are modeled separately and solved sequentially. However, the solution determined in a sequential planning process may not guarantee the optimality of the entire crew planning problem. This study is a first attempt at including rostering pattern consideration in the metro crew planning problem, where a rostering pattern is defined as the combination of a day-off pattern, a set of duty types and a rostering scheme. To solve this complicated problem, we first generate a multiple-layer time–space network where duty time windows with specific duty types are represented by different layers. We then model the considered crew planning problem using a path-based integer program on the time–space network and develop two column generation-based heuristics to solve the problem, where dual prices are particularly used in generating train paths. A computational study is conducted with real-life data derived from Hefei Metro to examine the effectiveness of the modeling and solution methods as well as observe the benefits of roster pattern designs. • Crew planning with day-off pattern, duty type, and rostering scheme is studied. • An integer linear program is formulated with path-selection variables. • Two column generation-based heuristics are developed. • Performance of the proposed heuristic is tested computationally.