Optimization models for multiperiod probabilistic ambulance location

The objective of a medical emergency service (EMS) system is to provide immediate medical care to the population. EMS managers constantly deal with the problem of improving the system performance, in particular the response time to calls. A careful strategical and planning phase is a major prerequisite for the success of a system. In this Master thesis work we consider the problem of ambulance location, and we develop a multiperiod probabilistic model. Our research is carried out in cooperation with AREU emergency center of Milan, within the project DECEMBRIA. The objective of the model is to determine the minimum number of ambulances needed to meet a predetermined level of reliability for the system, together with their locations. We propose a multiperiod model, defined on a set of consecutive time periods. Each period represents a time cluster, in which system conditions can be considered as stationary. This way, we take into account the variability of system conditions with respect to time. We also consider the probabilistic aspect of the problem. The chance of a system congestion is modeled with the introduction of the possibility that an ambulance be already busy when called for a service. A heuristic algorithm based on Lagrangian Relaxation and neighbourhood search is proposed as an instrument to solve the problem. We report results from the application of the model to the emergency medical system of Milan.