Quantified PIRT for Thermal Hydraulic Computer Code Validation

摘要

US Nuclear Regulatory Commission (NRC) has revised the reactor licensing rules in 1988 to allow the use of realistic best-estimate (BE) computer code, such as RELAP5 [1], if its uncertainties are quantified following the rigorous Code Scaling, Application and Uncertainty (CSAU) methodology [2]. To identify major sources of uncertainty in best-estimate code calculations, CSAU methodology introduced a systematic approach called Phenomena Identification and Ranking Table (PIRT), which is based on expert panel subjective conceptual judgments that may be overly conservative. To support and examine the traditional PIRT with objective quantified judgments, this study proposes a quantitative approach, the Quantified PIRT (QPIRT), together with an example application to a 2" Large Break Loss-of-Coolant Accident (LBLOCA) in APEX test facility at Oregon State University (OSU).