The refugee camp management: a general framework and clean water network design

Abstract

Motivated by the increasing need to provide support to refugees, which remains as a pressing issue in the agenda of many countries, we consider the refugee camp management problem. It is of utmost importance that asylum-providing countries co-operate with the Non-Governmental Organizations (NGOs) while managing the process and use the resources e ciently and e ectively. The Refugee crisis is one of the challenges of today's world mainly due to the di culties such as geographical discrepancies and unknown duration of refugee camps. In the rst part of the thesis (Chapter 2), we consider the refugee camp management process with an Operations Research (OR) perspective and provide a framework. In the proposed framework, the overall refugee camp management problem is rst categorized into two main phases: the establishment phase, which consists of one-time decisions such as infrastructure design; and the administration phase, which focuses on routine decisions that are taken on a periodic basis like aid distribution. We then detail each phase via sub categories and link each decision made in these sub categories with an appropriate decision problem from the OR literature. The proposed framework is aimed to be general enough to cover the problems that might arise in di erent geographies. In the second part of the thesis, the water network design problem encountered in establishment phase of refugee camps is considered. We propose single and bi-objective models that determine the locations of water well and water distribution units (WDUs), and design the general network structure (transmission pipes). As objective functions; total cost as a function of pipe length and total walking distance of refugees to the closest water source are utilized. The minimal requirements de ned by various NGOs are incorporated into the models. One of them is ensuring that the distance between a tent and its closest WDU should be below a certain threshold, which is imposed on the models as a covering constraint. The data set acquired from Gaziantep refugee camp in Turkey is utilized when solving the models. We solve the single objective models using CPLEX. While solving bi-objective models, the epsilon constraint approach is applied in order to nd the optimal Pareto frontier. In our computational experiments, we observed that for smaller threshold values, solution times increase in an exponential manner. Thus, we modify the non-dominated sorting genetic algorithm (NSGA-II) and use it to solve such problems. The quality of the Pareto solutions acquired by NSGA-II and the epsilon constraint method are assessed via performance measures de ned in the literature.