Date

2023-1-23

Author

Akdoğan, Muharrem Altan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

179
views

106
downloads

Cite This

In this thesis, emergency medical service (EMS) vehicle location problem with uncertainties in demand, travel times, and incident handling time is studied in three layers. The performance measures of EMS systems are evaluated with discrete event simulation models due to the uncertainties incorporated. Firstly, we focus on the equity in service quality resulting from vehicle location decisions in emergency medical services. We address the unbalanced service quality among regions with respect to various mathematical models including conventional ones. An extensive numerical study is conducted to show the effect of modeling approaches and network features on equity. Several observations are drawn and it is shown that the use of overall performance measures in the objective functions of mathematical models ignores other essential criteria, and there is room for improvement in terms of equity. Secondly, we propose decomposition methods based on queueing theory to assess the performance measures of the EMS system under a given location solution without needing to construct computationally cumbersome queueing or simulation models. The proposed decomposition methods require a set of nonlinear equations to be solved simultaneously, yet they are still favorable to exact queueing or simulation models in terms of the computational burden. A genetic algorithm is proposed to use the decomposition method in the optimization problems and find solutions for the EMS vehicle location problem based on queueing theory. It is shown that the proposed methods perform well in assessing the performance measures and evaluating close enough solutions the best solution for the optimization problems. Lastly, we propose mixed integer nonlinear problems (MINLP) with various objective functions incorporating closed-form formulations for the performance measures of the system based on decision variables. With MINLP models, there is no need for estimating problem parameters such as busy probability of vehicles in advance. The proposed models are easier to construct and solve with respect to approximation algorithms or decomposition methods in the literature where stochastic processes are incorporated. Hence, the proposed MINLP models enable decision-makers to incorporate uncertainties in the problem environment directly in the estimation of the parameter of the models based on queueing theory while still keeping the models relatively easy to solve since only the objective functions are nonlinear.

Subject Keywords

emergency medical services, ambulance location, simulation, stochastic processes

URI

https://hdl.handle.net/11511/102550

Collections

Graduate School of Natural and Applied Sciences, Thesis