@misc{BreugemBorndoerferSchlechteetal., author = {Breugem, Thomas and Bornd{\"o}rfer, Ralf and Schlechte, Thomas and Schulz, Christof}, title = {A Three-Phase Heuristic for Cyclic Crew Rostering with Fairness Requirements}, issn = {1438-0064}, url = {http://nbn-resolving.de/urn:nbn:de:0297-zib-74297}, abstract = {In this paper, we consider the Cyclic Crew Rostering Problem with Fairness Requirements (CCRP-FR). In this problem, attractive cyclic rosters have to be constructed for groups of employees, considering multiple, a priori determined, fairness levels. The attractiveness follows from the structure of the rosters (e.g., sufficient rest times and variation in work), whereas fairness is based on the work allocation among the different roster groups. We propose a three-phase heuristic for the CCRP-FR, which combines the strength of column generation techniques with a large-scale neighborhood search algorithm. The design of the heuristic assures that good solutions for all fairness levels are obtained quickly, and can still be further improved if additional running time is available. We evaluate the performance of the algorithm using real-world data from Netherlands Railways, and show that the heuristic finds close to optimal solutions for many of the considered instances. In particular, we show that the heuristic is able to quickly find major improvements upon the current sequential practice: For most instances, the heuristic is able to increase the attractiveness by at least 20\% in just a few minutes.}, language = {en} } @article{BreugemSchlechteSchulzetal., author = {Breugem, Thomas and Schlechte, Thomas and Schulz, Christof and Bornd{\"o}rfer, Ralf}, title = {A three-phase heuristic for the Fairness-Oriented Crew Rostering Problem}, series = {Computers \& Operations Research}, volume = {154}, journal = {Computers \& Operations Research}, doi = {https://doi.org/10.1016/j.cor.2023.106186}, abstract = {The Fairness-Oriented Crew Rostering Problem (FCRP) considers the joint optimization of attractiveness and fairness in cyclic crew rostering. Like many problems in scheduling and logistics, the combinatorial complexity of cyclic rostering causes exact methods to fail for large-scale practical instances. In case of the FCRP, this is accentuated by the additionally imposed fairness requirements. Hence, heuristic methods are necessary. We present a three-phase heuristic for the FCRP combining column generation techniques with variable-depth neighborhood search. The heuristic exploits different mathematical formulations to find feasible solutions and to search for improvements. We apply our methodology to practical instances from Netherlands Railways (NS), the main passenger railway operator in the Netherlands Our results show the three-phase heuristic finds good solutions for most instances and outperforms a state-of-the-art commercial solver.}, language = {en} }