The Impact of an Improved Flaw Model on a Pressurized Thermal Shock Evaluation

Abstract

The current regulations for pressurized thermal shock (PTS) were derived from computational models that were developed in the early-mid 1980s. The computational models utilized in the 1980s conservatively postulated that all fabrication flaws in reactor pressure vessels (RPVs) were inner-surface breaking flaws. It was recognized at that time that flaw-related data had the greatest level of uncertainty of the inputs required for the probabilistic-based PTS evaluations. To reduce this uncertainty, the United States Nuclear Regulatory Commission (USNRC) has in the past few years supported research at Pacific Northwest National Laboratory (PNNL) to perform extensive nondestructive and destructive examination of actual RPV materials. Such measurements have been used to characterize the number, size, and location of flaws in various types of welds and the base metal used to fabricate RPVs. The USNRC initiated a comprehensive project in 1999 to re-evaluate the current PTS regulations. The objective of the PTS Re-evaluation program has been to incorporate advancements and refinements in relevant technologies (associated with the physics of PTS events) that have been developed since the current regulations were derived. There have been significant improvements in the computational models for thermal hydraulics, probabilistic risk assessment (PRA), human reliability analysis (HRA), materials embrittlement effects on fracture toughness, and fracture mechanics methodology. However, the single largest advancement has been the development of a technical basis for the characterization of fabrication-induced flaws. The USNRC PTS-Revaluation program is ongoing and is expected to be completed in 2002. As part of the PTS Re-evaluation program, the updated risk-informed computational methodology as implemented into the FAVOR (Fracture Analysis of Vessels: Oak Ridge) computer code, including the improved PNNL flaw characterization, was recently applied to a domestic commercial pressurized water reactor (PWR). The objective of this paper is to apply the same updated computational methodology to the same PWR, except utilizing the 1980s flaw model, to isolate the impact of the improved PNNL flaw characterization on the PTS analysis results. For this particular PWR, the improved PNNL flaw characterization significantly reduced the frequency of RPV failure, i.e., by between one and two orders of magnitude.