Optimization of Blood Supply Chains with Stochastic Demand and Supply: Formulations and Solution Approaches

Abstract

Blood Supply Chain inventory management is a field that is quickly gaining in popularity due to the increasing unmet demand for blood products and their wastage (spoilage). This mismanagement of the blood supply chain calls for supply chain models that focus on the unique aspects of blood components to account for agedifferentiated demand and other variances between traditional and blood supply chains. This thesis presents two new supply chain models and a solution algorithm to improve management of the system. Both models were applied to the case of New York City (USA). These models aid in the coordination of blood supply and demand by studying the multi-echelon network of blood supply (from blood drives), blood storage and processing (blood banks) and distribution (hospitals). Given that the USA has over 600 institutions that conduct blood drives, determining a systematic schedule to manage the synchronization between blood collection and its demand at the hospitals could help improve the efficiency of the system and reduce both shortage and outage costs. The first model is a two-stage stochastic model for Red Blood Cells, platelets and plasma in the Blood Supply Chain. This model is based on stochastic demand and supply and accounts for the perishability of the blood products and the different types of Red Blood Cells available in the four echelons of the supply chain. The model aims to integrate the supply chains for the different blood components given that each of the useful subcategories of blood (plasma, red blood cells, and platelets) have different aging properties. The second model, also a two-stage stochastic model accounts for a truckload transportation times and costs to more accurately represent the existing BSC iii system. Both models are tri-objective models that are converted into a single objective model using the ε-constraint method since they are not solvable using commercial software for large instances. Therefore, A high-efficiency solution algorithm is presented in this thesis to mitigate the increased complexity of the mathematical model along with two additional algorithms that could be efficiently utilized for medium-sized instances.

Date of Award	Dec 2017
Original language	American English