A mathematical modeling approach for robust nurse rostering

Unexpected work absences occur for a variety of reasons such as illness, excessive workloads and accidents. These shortages may cause an initially valid roster to become infeasible. A reactive procedure which assigns unscheduled employees to cover the shortages may be adopted in order to repair the infeasibility. Although this method is frequently employed, it is often inefficient due to employee unavailability, disruptions regarding employees' personal lives and also high costs. This work investigates robust nurse rostering and addresses the following research question: is it possible to organize a robust nurse roster which is not only less susceptible to disruptions but also has a lower cost to repair the roster to feasibility when disruptions do occur? The proposed methodology is validated by way of an integer programming model and a three-step simulation method. First, an initial solution is generated considering different proactive strategies. Second, disruptions are randomly generated over the solutions. Finally, the infeasible solutions are repaired and the final costs are evaluated. Several strategies are investigated employing proactive, reactive and a combination of these two strategies. Computational results demonstrate that a trade-off between proactive and reactive strategies performs best, reducing overall final cost by the most significant margin. Further insights regarding the integer programming model and robust rostering strategies will also be presented.status: accepte