Blood supply chain design in disaster management: a comparison of risk-averse measures in robust optimisation approaches
研究了灾害响应型血液供应链设计，通过两阶段随机优化框架比较风险规避措施，发现均值-CVaR方法在平衡预期绩效和风险暴露方面优于其他方法，并开发了分解算法以处理大规模场景。
Disasters severely disrupt blood supply chains, challenging healthcare systems to ensure timely and equitable access to blood products. This study investigates the design of disaster-responsive blood supply chains through a two-stage, multi-period stochastic optimisation framework that explicitly incorporates decision-makers' risk aversion. The model aims to minimise unmet demand under uncertainty and budget constraints, using mean-Conditional Value-at-Risk (CVaR) and worst-case risk measures. Computational experiments conducted on synthetic instances of varying scenario structures show that the mean-CVaR formulation outperforms alternative approaches in 95-99% of the tested cases in terms of balancing expected performance and risk exposure. To address computational complexity, decomposition-based solution methods are developed, enabling the solution of large-scale instances with up to 300 scenarios, which are otherwise intractable using monolithic formulations. A sensitivity analysis, complemented by a case instance, provides insights into the influence of risk parameters and budget levels on preparedness and response decisions. The results underline the critical role of budget availability in determining system performance, while demonstrating that risk-averse formulations allow explicit control of extreme shortages without compromising average outcomes. The study contributes to the literature on humanitarian logistics and healthcare operations by integrating risk-aware modelling and scalable solution techniques for disaster preparedness planning.