A hybrid model for crew scheduling in rail rapid transit networks

Human resources related problems are of vital importance for all the companies. Within the transportation world, they usually represent a very important part of the overall expenses. This paper presents a network-flow based crew scheduling model for rapid transit networks, where distances are not big but service frequency is very high. The crew scheduling problem has been traditionally modeled as a set covering/partitioning problem. Here, the presented approach faces the problem in a different way taking benefit from a linear payment policy and enabling an easier integration with other planning problems such as rolling stock circulations. For solving this new model an ad-hoc mathematical decomposition based on time personnel clustering is proposed. Then, a Fix and Relax matheuristic is employed to solve the resulting problem. Commercial Branch and Bound and heuristics are also tested so as to compare the performance of the presented solution approach. In order to demonstrate the quality of the results, experiments for real case studies drawn from the main Spanish train operator, RENFE, are presented showing a good balance between optimality and computational times. (C) 2019 Elsevier Ltd. All rights reserved.