An Uncertainty Analysis Software to Treat State-of-Knowledge Correlation in PSA

The result of a PSA includes uncertainty. The sources of uncertainty are data uncertainty, model uncertainty and completeness uncertainty. Uncertainties are distinguished as being either aleatoric or epistemic. Aleatoric uncertainty comes from 'random variability in some parameter or measurable quantity', and the epistemic uncertainty comes from 'an imprecision in the analyst's knowledge about models, their parameters, or their predictions'. The aleatoric uncertainty can be estimated in quantitative values using the Monte Carlo approach. Recently, ASME PRA Standard requires propagating parameter uncertainties and those model uncertainties explicitly characterized by a probability distribution using the Monte Carlo approach. In addition, it requires also propagating uncertainties in such a way that the 'state-of-knowledge' correlation between event probabilities is taken into account. The same kinds of components are categorized as one group, for which probabilities are calculated from the same data. There exists the correlation between event probabilities because they are calculated from the same data. This correlation is called state-of-knowledge correlation. The grouping of the common cause failures (CCF) should be also considered in state-of-knowledge correlation. There are several software to treat the correlated uncertainty such as KIRAP. But, they are not suitable to treat the correlation in CCFs. This article describes a software developed to treat correlation between independent events as well as correlation due to CCFs