Driver scheduling for vehicle schedules using a set covering approach: a case study

Abstract

For transport companies, one important aspect is to decrease their operational costs. This can be done by somehow optimizing the routes of their public transport services. Usually the overall operational cost of the transportation consists of the cost of the vehicles used and their drivers. Solving such a task is rather complex and usually the optimized planning process is divided into several phases. These might be the vehicle scheduling, driver scheduling and driver rostering. First the journey tasks are assigned to vehicles, and then these vehicle schedules are divided into driver schedules, which are usually shorter, because of driving time restrictions. Next, the driver schedules are assigned to individual drivers. In the article, we will focus on the second part, i.e. with driver scheduling for given vehicle schedules. It is NP-hard to find the optimal driver schedules [8]. Here, we will use the well-known set covering approach to model the problem. It will be formally expressed as an integer programming problem, which can be solved by column generation. To generate the new columns, a time-space network based generator network is used, as proposed by Gintner et al. [8], and Steinzen et al. [14]. Here, we present a case study of how we applied these techniques to real data supplied by the Szeged public bus transport company.