A multi-period and bi-objective approach for locating ambulances: a case study in Belo Horizonte, Brazil

TL;DR

This paper enhances ambulance location models in Belo Horizonte by introducing multi-period and bi-objective optimization, significantly improving coverage and flexibility through relocation strategies and decision-maker tools.

Contribution

It develops a novel multi-period, bi-objective model for ambulance location that incorporates relocation, improving coverage and decision-making in EMS systems.

Findings

Coverage increases up to 31% with deterministic models.

Relocation improves coverage by up to 21%.

Bi-objective approach aids decision-makers with trade-offs.

Abstract

This work aims to apply the Facility Location Problem in the Emergency Medical Service (EMS) of Belo Horizonte, Brazil. The objective is to improve two previous optimization models from literature to handle base locations and ambulances allocation/relocation problems. The proposed multi-period models introduce the concept of relocation to the local EMS, which allows ambulances to move among bases in different periods to raise the system coverage. This paper also proposes a bi-objective approach aiming to minimize the number of bases and maximize the coverage of demands, which is solved using the epsilon-constraint method. Results show that coverage levels increase by up to 31% for the deterministic approach and up to 24% for the probabilistic approach. Ambulances' relocation optimization might improve coverage levels by up to 21% for the deterministic approach and up to 15% for the probabilistic approach when change scenarios from static to multi-period. Also, since the multi-period model solutions result in installing a larger number of bases, the bi-objective approach is a powerful tool for the decision-maker. Bi-objective results suggest modest increments in the objective function when the stations' number exceeds 28 for the deterministic approach. The probabilistic approach increments in the objective function start to narrow above 30 installed stations.