Developing a generalized flaw distribution for reactor pressure vessels

Abstract

The US Nuclear Regulatory Commission (NRC) is re-evaluating the guidance and criteria in the code of federal regulations as it relates to reactor vessel integrity, specifically pressurized thermal shock (PTS). Recent ultrasonic examination of considerable vessel material at Pacific Northwest National Laboratory (PNNL) and industry experiences with Yankee Rowe have provided the NRC with a better understanding of PTS issues. The re-evaluation of PTS will consider a risk-informed approach to the PTS rule and also provide important benefits for licensees considering license renewal. Pressurized thermal shock transients can lead to reactor vessel failure. These transients have occurred at operating reactors but, to date, they have not resulted in vessel failure. To properly determine the potential or probability for vessel failure from a PTS event, an accurate estimate of fabrication flaws is necessary. The characteristics of the fabrication flaw are inputs to fracture mechanics structural calculations that will determine the probability of vessel failure during a PTS event. Also, the results will indicate the sizes and locations of flaws that are most likely to cause failures. This information is also an integral input to the overall pressure vessel safety program. In order to obtain an accurate estimate of fabrication flaws to address PTS events for all classes of reactors, a generic flaw distribution must be developed. An expert judgment process will be used in conjunction with empirical data from PNNL, reactor pressure vessel studies and modeling (RR- PRODIGAL Code) in developing generalized flaw distributions. This paper will demonstrate the important relationship between reactor vessel integrity and flaw distributions in reactor pressure vessel material, discuss the PNNL work to date on developing flaw density and distributions for domestic RPVs, and describe the expert judgment process that was used to verify that a generalized flaw distribution can be properly developed and then assist in developing a generalized flaw distribution.