Analysis of locations of existing fire stations in Ankara in comparison to optimized locations

Abstract

This thesis aims to evaluate the locations of the existing fire stations in the city of Ankara. Vertex restricted p-center problem is solved via set covering problem and optimal locations of fire stations are found. We evaluate the fire stations in two ways. The first study aims to locate the same number of fire stations with the existing case. We use an algorithm proposed by Tansel (2011) which solves a finite series of set covering problems to reach the optimal solution of the p-center problem. We use three approaches for the p-center solution which constitutes our first study. In the first approach, we optimally locate the fire stations while keeping their allocation sets the same as those of the existing stations.In the second approach, we locate the fire stations to minimize the maximum of the distances between demand nodes and their closest stations. In the last approach, we take the population of the nodes into account and solve a weighted p-center problem where the weights of demand nodes are defined by their composite populations. We also give a methodology to compute the composite population counts for nodes. In the second study, we locate a minimum number of fire stations while ensuring that each demand node is covered within a pre-determined upper bound on the service time (distance). The bounds are taken to be 4 minutes and 8 minutes in accordance with the internationally accepted norms. Optimally located fire stations are found for both time restrictions. There are two subcases for this study; one of them aims to locate the fire stations without considering the existing fire stations while the other one locates the new fire stations in addition to the existing fire stations. Optimal solutions for all the cases explained above are compared with the existing fire stations with regard to the maximum distances, populations, and covered regions. Assessments are made on existing fire stations by benchmarking their performances against optimal solutions obtained via pcenter or covering solutions as outlined above. Geographical information systems are used for obtaining the demographic data and presenting visual analysis.