Kan bankası şebeke tasarımı: Tesis yerleşimi, envanter ve rotalama kararları

In this thesis, we design a supply chain network for blood bank distribution by integrating strategic, tactical and operational decisions. These decisions are usually studied separately in classical literature. However, we formulated a mixed integer nonlinear programming (MINLP) model to combine three decisions to minimize total system cost. In blood distribution network of Istanbul, hospitals keep their own inventory and procure bloods from main blood banks via weekly shipments. In the proposed model, some of the hospitals are selected as local blood banks (LBBs) and serve the hospitals which are assigned to them. Thus, our MINLP model solves a complex problem which aims to find optimal number and location of LBBs, assignment of hospitals to opened LBBs and the weekly and daily routes from the main blood bank to LBBs and from LBBs to hospitals. We use exact and approximate solution methods to solve this NP-hard problem. Firstly, small sized instances are solved by using commercial solvers. However, for mid and large sized problems, exact solution failed to find solutions in polynomial time due to increasing complexity. Therefore, we propose a tabu search based heuristic approach to find optimal and near optimal solutions. The performance of the solution methods are analyzed by comparing with each other and the current system costs on 65 test instances