A FOUR-ECHELON TWO-STAGE STOCHASTIC MODEL FOR BLOOD PRODUCTS SUPPLY IN DISASTERS

In recent years, the location of mobile facilities has been highly regarded in design of health system. One of the important areas in this field is design of blood collection and distribution systems. Since blood is as perishable and vital goods and donation of blood is a voluntary work, supply blood and blood products is one of the most challenging issues in the supply chain in emergency and non-emergency situation. In this paper, we propose a four-echelon two-stage stochastic model to supply whole blood and its products in disaster. The hospitals, regional blood centers, local blood centers and bloodmobile facilities are considered as main elements of blood supply chain. It is assumed that the blood collection from the donors just done in bloodmobile facilities and local blood centers. Also, processing operation of blood products from collected whole blood are done only in the regional blood centers. In the designed supply network in this paper, every hospital could provide needed blood products from other near hospitals in the emergency situation as well as other fixed and mobile blood centers. One of the most important issues in the blood supply in disasters is delivery time of blood, so here; we consider an upper bound to blood delivery time to the affected areas and the objective of our presented model is blood supply in the standard time so that the total supply cost be minimized. In the other hand, the time solution is very important to obtain an acceptable solution in a reasonable time. We present a heuristic approach based on genetic algorithm and exact method to solve the proposed MIP model. The locations of fixed and mobile facility are computed through genetic algorithm then other variables are calculated by solve model with CPLEX. In order to validate the proposed approach, we generate 8 examples with different sizes and numerical results are presented. Also the results of comparison our approach with exact and metaheurisitc method are presente.