Statistical software for risk analysis at the Savannah River Site

This paper describes statistical software developed at the Savannah River Site (SRS) to analyze event time of occurrence data extracted from fault tree data banks and/or user defined input data files. Five different distributions can currently be fit to the empirical data: normal, lognormal, exponential, Weibull and loguniform. Two goodness of fit tests, the Kolmogorov-Smimov one-sample test and the Chi-squared test, are used to determine how well a particular distribution fits the observed data. In addition, a comparison across fitted distributions is done to determine the most likely distribution fitting the data. A number of graphics can be generated illustrating the important characteristics of the data and how well each theoretical distribution fits the data. The theoretical distribution which best fits the observed data, the expected occurrence rate, and the probability of occurrence are used in fault tree analyses. Results from the SRS developed software were compared with commercially developed and tested software, SAS.