The application of fuzzy mathematics in safety analysis

In the general consideration of probabilistic safety analyses (PSAs), inputs may be stochastically variable, subjectively uncertain and/or dependent. Conventional techniques (e.g. probabilistic calculus or various forms of Monte Carlo analysis) can account for stochastic variability and correlation. However, basic PSA methodology may not apply well to some situations, most significantly when the inputs lack precise definition. This suggests that new approaches could be helpful for realistic modeling. Fuzzy mathematics is one of the most appropriate approaches available for processing subjective uncertainty, but incompatibilities inherent in combining fuzzy analysis with conventional PSA must be addressed. In this paper, a technique based on fuzzy mathematics is described that accounts for arbitrary combinations of stochastic variability, subjective uncertainty and dependence