Models and algorithms for balanced rostering with limited skills

The rostering problem consists in finding an optimal assignment of shifts to workers in a given time horizon. When the total workload should be evenly distributed, we have a balanced rostering problem. We consider a practical application to the junk removal company in Crema, Italy. In this case, predefined daily shifts must be assigned to the drivers in a fair way. However, the drivers have limited skills, that is they are qualified to perform only a small subset of the possible shifts. Here, we propose three alternative formulations for the problem, and corresponding algorithms to solve them.