A heuristic procedure for the crew rostering problem

This paper deals with the problem of planning work schedules in a given time horizon so as to evenly distribute the workload among the drivers in a mass transit system. An integer programming formulation of this problem is given. An iterative heuristic algorithm is described which makes use of a lower bound derived from the mathematical formulation. Furthermore, the algorithm at each iteration solves a multilevel bottleneck assignment problem for which a new procedure that gives asymptotically optimal solutions is proposed. Computational results for both the rostering problem and the multilevel bottleneck assignment problem are given.