Risk-Averse Distributionally Robust Optimization for Resilient Biomass Supply Chain Network Design

This study investigates designing a resilient biomass supply chain (BSC) network to mitigate disruption risks under uncertain data. It addresses disaster events causing demand uncertainty without relying on precise probability distributions. A demand-based backup assignment strategy is proposed to strengthen the supply chain against disruptions. An environmental penalty cost is also integrated into the network design for emission mitigation. The study introduces a novel BSC design combining resilience, risk management, and environmental assessment under data ambiguity. First, a risk-averse two-stage stochastic programming model is proposed using the mean-conditional value-at-risk (CVaR) measure. Then, to tackle the ambiguity in the probability distribution of demand, a distributionally robust optimization (DRO) model based on the worst-case mean-CVaR is developed. The box and polyhedral ambiguity sets are used to model uncertainty in DRO. Computational results show that DRO models lead to more conservative supply chain decisions, preventing higher-than-expected costs. The demand-based backup strategy strengthens the BSC against disruptions, and environmental integration adds significant value to carbon emission reduction.