Comment Number: 4-1

Submitted by Joseph Conen of the BWR Owners Group

Comment: Table 4.1 shows a six-order of magnitude difference between the PFM and the field history estimates of through-wall cracking frequencies for the BWR-2 base case. Although the report suggests that service history data could be analyzed to resolve this difference, it is not clear if this was actually done. This also points out to the need for a rigorous examination of the statistical methods used to translate field leak data or service experience data into pipe break frequencies.

Response: The complexity of translating field leak data or operating-experience data into pipe break frequencies was part of the rationale for conducting the elicitation (Section 1). It was also a major consideration for performing multiple base case analyses so that results from different approaches could be compared. The small break LOCAs and through-wall cracking frequencies represent the categories where operating experience and PFM comparisons are most meaningful because the least extrapolation is required. For the small break LOCAs (Category 1), the results are typically within 2 orders of magnitude for most of the base case team members. However, the total spread in both the through-wall cracking leak frequencies (Figure 4.1) and SB LOCA frequencies (Figure 4.2) can be greater than three orders of magnitude. As the commenter indicates, the BWR-2 Category 0 frequencies (Figure 4.1) vary significantly (4-5 orders of magnitude in final NUREG)

Furthermore, while the PFM and the operating-experience predictions of the base case analyses did not always readily agree, this simply reflects the current uncertainty in calculating these estimates.

Differences among these estimates reflect the different assumptions and approaches used in the various analyses. Specifically, the BWR-2 operating-experience frequency estimate was based on 20 reported incidents in BWR feedwater systems. However, none of these events resulted in through-wall leakage. Therefore, the operating-experience estimate is undoubtedly conservative.

The elicitation reported in NUREG-1829 did not attempt to resolve these discrepancies; rather, the elicitation’s goal was to reflect the scientific uncertainty in the technical community. Therefore, the purpose of the base case estimates was not to obtain convergence among the predictions. Rather, it was to understand how the assumptions and approaches associated with particular analyses contributed to the disparity in the results. These analyses and their differences were clearly presented to the elicitation panelists so that they could judge their effect as part of their elicitation. As a result, most panel members chose to anchor their current-day LOCA Category 1 responses using operating experience and used PFM, if at all, to inform their assessments for LOCA sizes greater than Category 1 or for future LOCA frequency estimates. These are the domains where relevant operational experience data does not exist.

Key elements of this response were incorporated into Section 4.2 of the revised NUREG.

Comments Related to Section 5 of NUREG-1829

None