Thermal-hydraulics system codes uncertainty assessment: A review of the methodologies

By revising the ECCS licensing rules in 1989, the USNRC has allowed the use of 'best estimate' thermal-hydraulics computer codes (such as RELAP5, TRAC, and TRACE), with the requirement that uncertainty analysis accompany the results. Several methodologies have been developed for the quantification of the uncertainties of such codes. These methodologies are either input-driven or output-driven. They disagree in definition for the uncertainty range, qualification and quantification steps, types of uncertainty sources considered, methods of assignment of uncertainty distribution or range to various parameters, approach to propagation of uncertainty, and the way the dynamic characteristics of TH codes are handled. The IMTHUA methodology, developed by the author, is a hybrid approach where an input-driven 'white box' method is augmented with output correction based on experimental results relevant to code output. This paper offers a comparative assessment of uncertainty analysis methodologies for thermal-hydraulics transient calculations. The methods will be compared based on their approaches for treatment of input, propagation, and code models and correlations, as well as output. Comprehensiveness, approach to data treatment, and interpretation of results are among the criteria for comparison. Several examples are provided to clarify the differences.