Uncertainty Analysis as Applied to Probabilistic Risk Assessment

Abstract

There are three basic stages in any uncertainty analysis. The first is data analysis, including the use of expert opinion. This task is the development of probability distributions for the basic quantities of interest, such as equipment failure rates and repair times. These distributions can be either generic or plant-specific, depending on the purpose of the analysis and the availability of plant-specific information. The second stage involves propagating the distributions for the basic quantities of interest to obtain distributions for the desired overall performance measure (e.g., the frequency of a hazardous release from a chemical plant). This can be done using either Monte Carlo simulation or the method of discrete probability distributions. Finally, the results of the uncertainty propagation must be interpreted and their implications for decision making clarified. This stage involves significant interaction between the analysis and the decision makers.