Operations Research Models for Blood Supply Chain Network Design for Disaster Planning

Abstract

A blood supply chain (BSC) deals with the collection, processing, storing, and distribution of blood collected from donors and delivered to patients at demand points (DPs) through a network of several temporary mobile blood units (MBUs) and permanent local blood centers (LBCs). In general, the strategic and tactical design of a BSC network determines response time and blood availability and accessibility for various health care services that extend and improve lives. Typically, BSC networks in the context of disaster management fall within the scope of humanitarian logistics networks in which profit, the main objective of commercial supply chains, is replaced by the objective of timely and proper delivery of post-disaster aid. Yet, most of the existing literature on the strategic design of BSCs aims at minimizing costs of the supply chain. Therefore, the main objective of this dissertation is to develop mathematical models to minimize total travel time for the distribution of blood units within a BSC and improve response time to deliver blood to DPs. Moreover, the approach used in this research explicitly incorporates uncertainty associated with disasters in the design and planning of BSCs. First, a deterministic mathematical programming formulation is developed to model the design of a BSC with the specific goal of minimizing travel time to distribute blood to DPs. The model is solved to identify the optimal location and relocation of LBCs and MBUs. Then, the deterministic model is extended into a stochastic programming model by incorporating uncertainty in blood supply, blood demand, and travel times associated with disaster scenarios. The resulting two-stage stochastic programming (TSSP) model is tested on small, medium, and large instances considering several disaster scenarios. Finally, a Sample Average Approximation (SAA) method is utilized to address the computational complexity of the TSSP model and efficiently solve a large-scale BSC network with a large number of scenarios within a reasonable computational time. Based on the insights obtained from computational experiments completed to test key parameters (e.g., donation rates, collection capacities, threshold for demand satisfaction, and facility installation budget) that may affect the BSC network design and the effectiveness of blood distribution, we conclude that there is a trade-off between total travel time and demand satisfaction for all BSC network sizes. Mainly, it was observed that reduced donation rates and collection capacities decreased demand satisfaction while increasing total travel time due to the need for deploying a considerable number of MBUs to collect blood. Another observation is that increasing the threshold for demand satisfaction at DPs considerably increases total travel time given the additional blood that must be distributed to DPs. Further, we noted that although a higher facility installation budget had little effect on reducing total travel time, it allowed for increased demand satisfaction which significantly improved service within the BSC network. Additionally, the value of deploying MBUs along with LBCs is demonstrated in the computational experimentation where optimal BSC networks utilize MBUs more frequently in disaster scenarios to allow for additional blood collection to meet demand in a timely manner. In this regard, we observed that the need for multiple shipments from different LBCs to a single DP reduces when MBUs are deployed. Overall, MBUs have proven to be effective in decreasing total travel time as well as increasing demand satisfaction. Finally, there is evidence that available blood supply had a significant influence on location decisions for LBCs and MBUs more than the location of DPs and their associated demand. Most importantly, the computa