Using cross‐impact analysis for probabilistic risk assessment

Cross-impact analysis is widely employed to inform management and policy decisions based on the formulation of scenarios, defined as combinations outcomes relevant uncertainty factors. In this paper, we argue that use nonprobabilistic variants cross-impact problematic in context risk assessment where usual aim produce conservative estimates which may exceed but are not smaller than actual level. Then, building characterization probabilistic dependencies, develop an approach (i) admits several kinds statements about factors their dependencies; (ii) maps such into constraints joint probability distribution over all possible scenarios; (iii) provides support for preserving consistency elicited statements; (iv) uses mathematical optimization compute lower upper bounds overall This approach—which illustrated with example from nuclear waste repositories—is useful it retains informativeness while ensuring these interpreted within coherent framework theory.