GENERAL APPROACH AND PHILOSOPHY OF EACH PANEL MEMBER

In this appendix the general approach and philosophy that each panelist followed as part of this elicitation exercise is presented.

K-1

BRUCE BISHOP

For PWR piping frequencies, the median probability of a 5,000 gpm (19.000 lpm) leak after 40 years of operation comes from the average point estimate for 7 plants that used the PFM methodology for the WOG Piping RI-ISI (WCAP-14257, Rev. 1-NP-A, Supplement 1). These seven plants were selected to provide a representative sampling of all plants with a Westinghouse NSSS design. Characteristics considered in the sampling included number of primary loops, old and new design vintage and foreign and domestic utility operators. The variability in 40-year probability with leak-rate comes from a WOG supported sensitivity study that reflected both the decrease in probability with increasing leak rate of one pipe size and the number of pipes of a given size that could contribute to a given leak rate. All piping leak probabilities consider the effects of LBB with a minimum detectable leak of 1 gpm (3.8 lpm) per typical plant tech-spec requirements. The increase in failure probability in going from 40 years to 60 years of operation is based upon another WOG sensitivity study. This study and its results are described in a paper presented at the 1999 Pressure Vessel and Piping Conference of ASME and included in PVP — Volume 383.

Non-Piping Frequencies are based upon the degradation mechanism of PWSCC initiation and through — wall growth, which is currently the primary cause of unexpected leaks in non-piping components in the primary system. Most other degradation mechanisms are being effectively mitigated. The relative frequencies by component type are based upon a proprietary best-estimate of PWSCC susceptibility by Westinghouse experts for unmitigated Alloy 600/182 base/weld metal. The uncertainties are based upon the variability between the best-estimate susceptibility for PWSCC and observed leak experience.

VIC CHAPMAN

In order to derive a basic set of failure probabilities, the values generated by the ‘Base Cases’ analysis were initially considered. However, in the end, a decision was made to use the results from some previous work that involved a ‘Risk-Informed ISI’ application. That work considered a full plant assessment using fatigue as the basic degradation mechanism. Initially, the results from this full plant assessment were compared with the appropriate base cases in order to ascertain whether they were in general agreement with each other. Once it was decided that the two sets of results were in agreement, it was decided to proceed with using the full plant assessment results. These results provided a set of pipe weld failures over a full range of pipe weld sizes that could be considered as a form of global values for each weld size. Factoring would then be from this base set.

Since the leak rate, given a failure, is independent of the failure probability, this can be evaluated separately to obtain a conditional probability. The basic method developed for the base case was expanded to include lower and upper estimates at each step. These basic steps are as follows:

1 Use expert judgment to estimate the COD, up to full rupture, as a function of defect size.

2 Evaluate the defect cross-sectional area for a given defect size using its associated COD.

3 Evaluate the leak rate from a given defect size using some data supplied by the USNRC.

4 Use expert judgment to assess the distribution of the defect length at failure.

5 Combine Steps 3 and 4 to obtain the conditional probability of a leak rate greater than the prescribed leak rate.

The final probabilities were obtained by combining the conditional probability above with the basic fatigue failure probability.

The effect of leak detection was introduced via a factor that was a function of the leak rate. This reduction factor varied from about 5 for a Category 1 leak, up to about 50 for a Category 6 leak.

For non-weld areas, such as the pump bowls and nozzle crotch corners, the basic probabilities were first factored. Next, the basic steps to derive the conditional leak rates as discussed above were followed to adjust the distributions as appropriate.

The effect of PWSCC was introduced as a multiplying factor on the basic fatigue failure rates. It was assumed that for small pipes, 2 inch diameter, that they would still have a significant contribution from fatigue, but that for the largest pipes, the full three orders of magnitude implied by the PWR-1 Base Case (i. e., hot leg base case) should be applied.

Finally, the failure rates for each system were derived by simply summing over all the elements within a given system.