| dc.description.abstract | This thesis presents a nurse scheduling problem tailored to characteristics common of
Norwegian hospitals. The problem involves allocating nurses to specific shifts to ensure
coverage of demand, while respecting work regulations and accounting for balance of
workload, general nurse preferences and fairness. We formulate models for the nurse
scheduling problem in line with the scheduling principles at Haukeland University Hospital
and solve them using mathematical programming techniques. The purpose lies in the attempt
to present a more efficient approach to the problem, compared to the manual scheduling
approach currently utilized at the hospital.
The method involves formulating a mixed integer programming model, which is implemented
computationally in an optimization software. Multiple decision models are produced to
represent two different forms of schedules, one cyclical and one calendar based. The model
for the cyclical schedule can be optimized directly using the solver of the software. The model
for the calendar-based schedule is more complex and is therefore solved by designing a
decomposition heuristic approach to find a good solution in a reasonable computational time.
The conclusion is that the schedules derived from the decision models are viable, with
emphasis on the considerable time savings compared to the current scheduling approach at
Haukeland. Currently, the hospital uses a manual method which takes approximately four to
six weeks to create a schedule, whereas the models proposed in this thesis are able to derive
an optimal solution within two hours. The models manage to effectively account for many
criteria, including work regulations, fairness, balance of workload and preferred practices. The
work in this thesis has been conducted through close cooperation with a representative from
the staffing department at Haukeland, and the solutions derived from the models are able to
capture their considerations in practice to a large extent. Our work has contributed to giving
staff at Haukeland insight on how optimization and computational tools can be used to deal
with their complex work scheduling problem. | en_US |
