Correlation in probabilistic safety assessment uncertainty analysis

摘要

Uncertainty analysis is essential part of every complete risk assessment. Increased usage of Probabilistic Safety Assessment (PSA) models for various practical applications in the nuclear industry makes uncertainty analyses even more important. Essential steps related to the PSA model uncertainty determination are: parameters uncertainty identification, uncertainty propagation, and uncertainty analysis. One special issue related to the PSA model uncertainty analysis is about parameter uncertainty correlation. PSA model parameters uncertainty might be correlated because of various different reasons (i.e., same data source, similar component design, same related function, same operating environment, etc.). There are two interesting and important questions related to the parameter uncertainty correlation: (1) how to determine correlation level? and (2) what is the influence of parameter correlation to the PSA model uncertainty? This paper discusses second question: assuming parameter correlation existence and various levels, the potential influence to the PSA results uncertainty is investigated.