Otimização da escala mensal de motoristas de ônibus urbano utilizando a heurística Variable Neighborhood Search

One of the last stage from public transportation planning concerns to defining the scale of urban bus drivers for short term period, called Crew Rostering Problem (CRT). This problem aims to generate sequences of daily shifts, includ-ing weekdays, Saturdays and Sundays, respecting labor laws and operational constraints. Moreover, a good crew roster should provide a better division of workload among the crews and still to reduce the overtime costs paid by the company. The model proposed in this paper is able to generate solutions satisfying a fixed scheme called 5/1 of day-off, beyond the labor laws and operational constraints imposed by the company. The Variable Neighborhood Search metaheuristic (VNS) was implemented using different neighborhood structures, varying the number of modifications performed on the current solution. The implementation was tested with data from a midsize company and the results show significant improvements compared to the solution adopted by the company.