Supplier selection and collaboration for determining joint facility location for humanitarian relief distribution

Humanitarian logistics activity is essentially made up of three main stage in the disaster life cycle: (pre-disaster) mitigation stage, (post-disaster) response stage, and recovery stage. Relief vendor selection and cooperation is a very significant section of the pre-disaster phase in order to get through to eliminate the troubles in the response phase that the government cannot handle on its own. The primary aim of this thesis is to choose the most suitable relief vendors in the mitigation phase according to determined criteria. In order to accomplish this aim; First, the criteria are determined as a result of literature review, brain storming, and face-to-face surveys. Then, Interpretive Structural Modeling (ISM) is applied to clasify and sort the criteria and discover the mutual effects between them. Among 15 (fifteen) determined criteria, 7 (seven) of them are evaluated to be more significant and affecting the other criteria. Second, Analytic Network Process (ANP) is applied to obtain the weights of the criteria, which are choosen through the Interpretive Structural Modeling (ISM). Then, the potential candidate relief vendors are determined. In this study, Asian side of Istanbul is assumed as the affected zone in the case study. Whereupon candidate relief suppliers are evaluated and ranked thanks to weighted criteria by means of the Rating technique. Third, mathematical model is constructed with four objective functions comprising of minimizing the impact of over and under supply in the affected zones, maximizing satisfied demand rate, minimizing joint facility location, and minimizing the distance between the joint facility locations and the other relief suppliers who have not been selected as joint facility locations. Whereupon, Non Sorting Genetic Algorithm (NSGA-II) is applied so as to solve this mathematical model, whereupon the solutions obtained after using NSGA-II are clustered using k-means algorithm. Ultimately, the proposed model's solution that is obtained as a result of multi-criteria decision making, and Pareto optimal solutions, which are obtained after using NSGA-II and k-means algorithm, respectively are compared. In this respect, if the host government collaborates with only highly collaborative relief suppliers under determined circumstances, our proposed model's solution lies among the Pareto optimal frontiers as a result of the NSGA-II. On the other hand, if host government collaborates with not only highly collaborative relief suppliers but also with normal collaborative relief suppliers, our propose model's solution is not situated within Pareto optimal frontiers. Keywords ⎯ Analytic Network Process (ANP), Humanitarian logistics, Interpretive Structural Modeling (ISM), Supplier selection, Non- Sorting Genetic Algorithm II (NSGA-II), k-means algorithm.