We make available a new set of real world instances for the Nurse Rostering problem. These seven cases are from Norweigen hospitals, collected through the software vendor Gatsoft AS.

SINTEF ICT: Our test cases are synthetic instances but they derived from detailed discussions with the hospitals.The use of synthetic data is justified by the present lack of high-quality real 24-hour demand data suitable for our purposes; our reference hospitals have plans for improving and standardizing their 24-hour demand data. The 24-hour demand data is subdivided into time slots of 15 minutes, and the demand values in a single time slot are for each skill type. The 24-hour demand data can be different for different weekdays. According to our data, the workforce (nurses, head nurse, and so on) is subdivided in up to 5 groups, each group having the same skill set. The skills in a skill set can be of 5 different skill types.

We make available a new set of benchmark problem instances for the admission planning problem. The problem formulation considers the allocation of day and operating room to a set of surgeries, as well as a specific time of surgery. Constraints include the feasibility of the chosen surgeon's individual schedule, as well as adherence to a "Master surgery schedule" for the operating rooms. The exact problem formulation, as well as computational results may be found in the paper A. Riise and E. K. Burke, Local search for the surgery admission planning problem, Journal of Heuristics.

SINTEF ICT was given data of  3 years of surgery  by SAB (Sykehuset Asker og Baerum HF), a major hospital in the city of Oslo, for a project called MSS-Adjust. We would like to thank Tone Jensen at SAB not only for providing the data but also allowing us to make it availible for the research comunity.