During the elicitation, each panelist will determine the biggest contributors for each LOCA frequency threshold category (Table B.1.1) separately for BWR and PWR plants. The information summarized in Tables B.1.2 — B.1.8, and discussed in the previous sections, is simply intended to identify those issues and variables which contribute to the LOCA frequency distributions. Each panelist will determine the magnitude and likelihood of each variable separately, but more importantly will determine the importance of the interrelationships among the variables.

Each panel member will not be asked to provide absolute LOCA frequencies. All questions will be structured so that relative differences with a specific base case will be queried. The base cases will be associated with absolute frequencies and quantitative LOCA estimates will be derived from these values and the relative relationships provided during the elicitation. The group spent quite a bit of time and effort both understanding the role of the base cases in the elicitation and assessing their importance. Ideally the base cases are chosen to represent significant contributing conditions to the total LOCA frequency estimates in order to minimize the extrapolation required during questioning. Representative, significant base cases should therefore theoretically improve the elicitation accuracy. Once this concept was understood, the group settled on several base case conditions for PWR and BWR systems. Pete Riccardella provided the original suggestions which were largely adopted by the panel after the analysis framework was clarified. The base case discussion evolved over the Wednesday and Thursday meeting days. All of the discussion will be summarized in this section for consistency.

Two base cases were developed for BWR piping systems (Table B.1.9). The first case will examine the recirculation system piping. All the various piping sizes will be considered, and original 304 stainless steel material will be assumed that has not been replaced during plant operation. The safe end is non­creviced Alloy 600 which is connected to the piping and vessel by Alloy 82/182 weld material. Only the IGSCC (subcategory of SCC, Table B.1.2) will be considered as the degradation mechanism. The loading will consist of pressure, residual stress, and dead weight nominal components. Transients to be considered include SRV loads and seismic. The base case will assume that NWC is used in the system.

The next BWR base case will examine the feed water system. A 12 inch diameter carbon steel pipe will be analyzed. The safe end and weld materials were not specified and will need to be defined by the base case analysis team (to be discussed subsequently). The degradation mechanisms for this base are FAC and TF. Loading sources include pressure, thermal, residual stresses, and dead weight nominal components. Thermal fatigue loading from stratification and possibly striping will provide alternating loads, and water flow velocities will also be included to assess fluid loading. Transients for this base case will include water hammer and seismic. Once again NWC will be assumed.

Three base cases were constructed for PWR systems (Table B.1.9). The first will examine the hot leg in the reactor coolant piping system. A 30 inch diameter Type 304 stainless steel pipe will be considered with Alloy 600 safe ends and Alloy 82/182 bimetallic welds. This base case will examine thermal fatigue and PWSCC. Loading will again include pressure, thermal, residual stress, and dead weight nominal loads and thermal fatigue alternating loads. Transients will include seismic loading and a pressure pulse transient, the magnitude and duration of which is still to be determined.

The next PWR base case is a 10 inch diameter surge line. The surge line material is Type 304 SS and an Alloy 82/182 bimetallic weld will be included at the pressurizer. No safe end materials will exist. Once again, thermal fatigue and PWSCC will be considered. Loading will include pressure, thermal, residual stress, and dead weight nominal loads and thermal fatigue alternating loads. Transients will include seismic loading and a pressure pulse transient, the magnitude and duration of which is still to be determined.

The final PWR base case was the most ill-defined case because it was added after initial group discussions. It was added to provide a base case for a smaller diameter piping system. This base case will need to be defined more completely prior to analysis. A 4 inch diameter high pressure injection/makeup (HPI/MU) line will be examined for thermal fatigue degradation. The piping material, welding, and safe end materials still need to be specified. Nominal loading is once again provided by pressure, thermal, residual stress, and dead weight loads. Thermal alternating loads will be defined and seismic and pressure pulse transients will be considered.

Absolute LOCA frequencies will be developed for each base case and for each threshold leak rate category defined in Table B.1.1. There are six leak rate categories and five base cases; therefore at least thirty separate calculations will be required to fully define the base case frequencies. The base cases will include analysis of many welds and other piping components. The LOCA frequencies for the system will obviously be the summation of the contributions from all system components. The frequencies will also be determined as a function of time. Three time periods will be evaluated: 25 years after plant startup (current-day), 40 years after start-up (end-of-plant-license), and 60 years after start-up (end-of-plant- license-renewal).

The panel decided that seismic transients would be handled as part of a sensitivity study. As mentioned previously, seismic-induced LOCAs will not be determined as part of this elicitation for several reasons: the PRA models that these estimates will be used for do not consider seismic loading; there has been significant work in developing seismic LOCA estimates; and the group has no specific expertise in seismic analysis. However, many panel members are experienced in conducting analysis when the seismic loading history is provided. Many panel members also seem comfortable with comparing other loading histories to seismic events. Therefore, it was decided that the elicitation would ask for

comparisons with conditional seismic loading (probability of occurrence of 1) separately in order to conduct a sensitivity analysis of the effect of seismic loading.

Originally, the group determined that the seismic loading magnitude will be 0.3 g for the sensitivity study. Pete Riccardella suggested changing the 0.3 g criteria for the seismic condition to an ASME Code faulted-stress condition. By doing this soil conditions, damping characteristics, etc. do not have to be considered. While there was some discussion on the merits of this suggestion, the issue was not finalized.

The base cases will also perform a sensitivity analysis to determine the effect of the frequencies both with and without ISI. Standard ISI techniques will be considered in the analysis. Credit will also be given in this analysis for leak detection. The leak detection threshold for the base case analysis will be leak rates which are commensurate with the data defined in the SKI-pipe database. The specific leak rate threshold associated with the database must be defined.

A base case team was established to develop four separate LOCA frequency estimates for each of the five base cases. The base case team will consist of Vic Chapman, Bengt Lydell, David Harris, and Bill Galyean. The team will model a specific piping system and define plant operating characteristics for each base case. Then, the team will develop input for each of the five LOCA variables (material, geometry, loading, degradation mechanism, and mitigation/maintenance) within the parameter constraints identified in Table B.1.9. The team will share information to ensure that each analysis is considering the same nominal conditions for each base case.

Each base case team member will develop their own LOCA estimates for each system using whatever methodology they choose. The likely general approaches for each team member are summarized in Table B.1.10. Each specific methodology will require additional assumptions. It is incumbent that each team member catalog required assumptions and document the methodology used to arrive at their base case LOCA estimates. This information will then be rigorously, yet concisely, presented to the remaining panel members. This should allow each panel member to completely understand the assumptions, methodology, and results generated by each base case team member. It needs to be stressed that once the general conditions are developed, the base case members should independently develop their estimates without further consultation. This step is necessary to retain realistic sample uncertainty in the calculated results.

The base case team would collaborate to ensure that the PFM analyses accurately capture that leaking pipe operating experience. This is the one aspect of the exercise that contains plant operating experience data. Initial PFM calculations will be conducted based on best-estimate assumptions and the current leak rate frequency predictions will be compared with the operating experience. At this point, PFM input assumptions may be changed in order to match the operating history. Each PFM model should accurately document the input variables, any model changes, and results both before and after benchmarking. This benchmarking exercise will help the remaining panel members gauge uncertainty in the calculations.

The panel will supply background information to the base case calculation team as required. All requests for background information will be coordinated by Rob Tregoning to ensure proper cataloging and dissemination to the group. Some volunteers already offered certain background information. Bruce Bishop has run his own PFM models for seven Westinghouse plants that could ultimately be used to help verify the base case calculations.

It was also stressed by the panel that it is important to make the PFM modeling conditions as close as possible to the postulated service conditions so that the various base case approaches can be directly compared to assess uncertainty and possible inaccuracies. For instance, many existing PFM models assume that all repairs are perfect (no defects). However, many repairs introduce new defects and most large flaws are associated with repairs. This fact (and other similar issues) is naturally captured within the operating-experience database, and needs to be considered within the PFM modeling if possible.