This paper addresses the Crew Rostering Problem (CRP) in a transit bus company. The CRP attempts to obtain a sequence of daily journeys for weekdays, Saturdays and Sundays, for a given planning horizon. Any model that represents the problem should consider both operational and labor constraints, such as minimum rest time at home and appropriate days off. In this work the problem is solved as a sequence of assignment problems for the daily journeys. This model builds a rostering as balanced as possible, minimizing the amount of overtime and crew idleness. Starting with the solution provided by the assignment model, a new model is presented in order to include the days off to the crews. This model assures that each crew has its days off properly assigned, either by the maximum days worked consecutively, or in order to have a Sunday free each seven weeks. The objective function is to minimize the number of crews covering the days off for regular crews. The models were tested with real data from two bus companies, and the results were compared with the solutions adopted by the companies. Thus it was possible to verify reductions in the number of crews ranging between 8% and 54%, plus reductions of 13% in the overtime paid by the companies.
public transportation; monthly rostering; crew rostering problem