Existing service experience with piping systems shows a strong correlation between failures and presence of an active degradation mechanism in combination with service conditions and transient loading conditions. It is therefore possible to estimate piping reliability parameters through statistical analysis of service experience data. Such analysis includes data processing whereby the appropriate reliability attributes are correlated with influence factors as described in SKI Report 97:26 [D.11].

In this Base Case Report the technical approach to LOCA frequency estimation builds on statistical analysis of service data associated with ASME XI Class 1 piping in the BWR and PWR operating environments. The study accounts for two kinds of uncertainties in piping reliability analysis, namely data uncertainty and state — of-knowledge uncertainty. The pipe failure database on which this study is based is called PIPExp [D.12], which is the extended version of the OPDE pipe failure database [D.13]. A description of PIPExp is included in Appendix A. The uncertainty analysis is performed by using a Monte Carlo merge technique to develop the LOCA frequency distributions. A commercial software package called Crystal Ball (Version 2000.2.2), which is an add-on for Microsoft Excel, is used to perform this Monte Carlo merge operation. Time- dependent LOCA frequencies are developed using a Markov modeling approach [D.14].

The BWR Base Case analysis is based on the degradation mechanism analysis as documented in Reference [D.5], and it builds on insights from an earlier BWR LOCA frequency pilot study [D.15-D.16]. The PWR Base Case analysis is based on the degradation mechanism analysis as documented in References [D.6, D.8], and builds on insights and results from an earlier sensitivity analysis performed in support of a risk informed

inservice inspection (RI-ISI) evaluation [D.12]. That sensitivity analysis addressed the impact of using a different pipe failure database on the RI-ISI weld selection.