Dynamic Ambulance Redeployment and Ambulance Dispatching

In this thesis, we address the problems of ambulance dispatching and ambulance redeployment. That is, deciding which ambulance to send to an accident and choosing a base for the ambulance to return to after it finished service. The goal in these problems is to minimize the fraction of late arrivals. As an alternative to the well known closest-idle policy, we propose a dispatching policy that makes a weighted choice between distance to the accident and the coverage an idle ambulance currently provides. For the ambulance redeployment, we alter an existing dynamic solution that was already shown to improve performance compared to a static benchmark. We alter it in such a way that every time an ambulance becomes idle, a base is chosen by a trade-off between the coverage an extra ambulance at that base will provide and the distance from the idle ambulance to that base. We evaluate our performances by a simulation of a realistic case study in which we measure the fraction of late arrivals. We compare the performances to a benchmark that uses a static redeployment policy and the closest-idle policy. We show that our dispatching policy has a relative improvement of 9.4%compared to the benchmark. Furthermore the original redeployment policy has a relative improvement of 12.3% compared to the benchmark static solution. Our alteration to this policy does not improve the fraction of late arrivals, but lowers the fraction of time an ambulance spends on the road compared to the unaltered policy. Together the policies result in a relative improvement of 17.8% which can be achieved without costs for extra crew shifts or extra vehicles.