Rail Crew Scheduling and Rostering Optimization Algorithms

Abstract

Train crew rostering involves the development of a duty timetable for each of the drivers of a rail transport organisation. This duty timetable is spread over a period known as the roster planning horizon. We propose an optimisation approach for an instance of a train crew rostering problem arising from a practical application. The problem involves developing continuous and cyclic work lines (or rosters) for train drivers spread over several depots in a rail network. The rail timetable includes over 1300 trips a week. The rosters for all the crew must conform to complex industrial regulations and work rules. The main objective in this model is to minimise the overall roster cost accrued from using the available crew while providing the requisite number of crew for each train trip in the schedule. The rosters must also satisfy quality standards for all drivers, by attempting to satisfy their personal preferences. We describe the problem and provide optimisation formulations and solution approaches. We also present some computational results.