Comparative Assessment of Severe Accidents Risk in the Energy Sector: Uncertainty Estimation Using a Combination of Weighting Tree and Bayesian Hierarchical Models

This study analyzes the risk of severe fatal accidents within the full fossil energy chains causing five or more fatalities. The risk is quantified separately for OECD and non-OECD countries. In addition for the Coal chain, Chinese data are analyzed separately because it has been shown that data prior to 1994 were subject to strong underreporting. In order to assess the risk and its uncertainty, a Bayesian hierarchical model was applied. This allows yielding analytical functions for frequency and severity distributions. Furthermore, Bayesian data analysis inherently delivers a measure of a combination of epistemic and aleatory uncertainties, through the a priori distribution and likelihood function that compose the Bayes theorem. In this study, in order to reduce the epistemic uncertainty related to the subjective choice of the likelihood function, Bayesian Model Averaging (BMA) is applied. In BMA the final posterior distribution is a weighted combination of the posterior distributions assessed for different likelihood functions (models). The proposed approach provides a unified framework that comprehensively covers accident risks in energy chains, and allows calculating specific risk indicators, including their uncertainties, to be used in a holistic evaluation of energy technologies.