Risk-averse Dynamic Bayesian Network for Supply Chain Risk Evaluation under Ripple Effects and Risk Dispersion

Severe disruptions in the supply chain (SC) trigger a ripple effect, leading to significant temporal and structural propagation with lasting impacts. Mitigating these effects requires a thorough SC risk assessment, focusing on performance fluctuations to inform long-term crisis strategies. Although prior studies often overlook dispersion characteristics, particularly from a risk-averse perspective, this research introduces a novel approach to evaluate multi-echelon SC risk under data scarcity. The proposed method: (i) models temporal and structural propagation using a risk-averse dynamic Bayesian network (RA-DBN) and probability interval method (PIM); (ii) formulates the problem with a risk-averse nonlinear programming (RA-NLP) model; and (iii) employs a quadratically constrained quadratic program (QCQP) technique for efficient resolution.