Как выбрать гостиницу для кошек
14 декабря, 2021
The LOCA frequencies in this Base Case Report are derived from service data on Code Class 1 piping. In this section we investigate how the LOCA frequencies relate to two data issues: 1) completeness of the pipe failure data collection, and 2) data interpretations. The former remains an ever-present issue in probabilistic safety assessment. Completeness is addressed by having in place an active and rigorous data collection process (c. f. Appendix A). Two sensitivity cases (SC:s) are defined to demonstrate how changes in the input to the failure rate calculations affect the estimated LOCA frequency. The sensitivity cases are defined as:
1. SC1: A small leak (< T. S. limit for unidentified RCPB-leakage) is assumed to have occurred in a pipe-to-safe-end weld in a BWR NPS28 reactor recirculation pipe during the time period 1988 — 2002. This evidence is used to modify the posterior weld failure rates.
2. SC2: This sensitivity case is concerned with an assumed large leak (= Cat0 LOCA) in a NPS28 BWR reactor recirculation pipe. Again, the large leak is assumed to have occurred in the time period 1988 — 2002. This evidence is used to modify the posterior weld failure rates and the conditional failure probability.
The results of the sensitivity analysis are summarized in Table D.21. These sensitivity cases are hypothetical in that they do not account for effects on piping reliability by the anticipated industry and regulatory actions that invariably would arise in response to the results of root cause analysis to determine the reasons behind a significant RCPB degradation such as defined by SC1 or SC2.
Base Case |
LOCA Frequency — Statistical Mean [per Reactor-year] |
||||
Flow Rate Interval [gpm] |
|||||
Catl 100 < v < 1500 |
Cat2 1500 < v < 5000 |
Cat3 5000 < v < 25,000 |
Cat4: 25,000 < v < 100,000 |
Cat5: 100,000 < v < 500,000 |
|
Base-1 |
8.24E-06 |
7.64E-07 |
3.07E-07 |
1.22E-07 |
3.05E-08 |
Base-1 — SC1 |
8.70E-06 |
8.07E-07 |
3.27E-07 |
1.29E-07 |
3.29E-08 |
Base-1 — SC2 |
1.30E-05 |
1.17E-06 |
4.77E-07 |
1.87E-07 |
5.63E-07 |
The following tables list some summary data from the non-pipe database.
Degradation Mechanism (see legend)
|
DM________ DM Description___________
MA Material Aging
FDR Fabrication Defect and Repair
SCC Stress Corrosion Cracking
LC Local Corrosion
MF Mechanical Fatigue
TF Thermal Fatigue
FS Flow Sensitive (includes FAC and E/C)
UNK_______ Unknown
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
PP-ID |
Piece Part |
FLG-fbs |
Flange Bolts |
LIV-bbs |
Bonnet Bolts |
LIV-bdy |
Valve Body |
LIV-bon |
Bonnet |
MSIV-bbs |
Bonnet Bolts |
MSIV-bdy |
Valve Body |
MSIV-bon |
Bonnet |
Pipe-w |
Weld |
PIV-bbs |
Bonnet Bolts |
PIV-bdy |
Valve Body |
PIV-bon |
Bonnet |
Pzr-bbs |
Pzr valve bonnet bolts |
Pzr-brv |
Bolted Relief Valves |
Pzr-hsl |
Heater Sleeves |
Pzr-mwb |
Manway Bolts |
Pzr-mwy |
Manway |
Pzr-noz |
Pzr Nozzles |
Pzr-rvb |
Relief Valve Bolts |
Pzr-shl |
Shell |
RCP-bdy |
Pump Body |
RCP-fwh |
Flywheel |
RCP-noz |
Pump Nozzles |
RCP-sel |
Pump Seals |
RecP-bbs |
Bonnet Bolts |
RecP-bdy |
Pump Body |
RecP-hx |
Pump cooler |
RecP-noz |
Pump Nozzles |
RecP-sel |
Pump Seals |
RPV-crc |
CRDM connections |
RPV-crd |
CRDM |
RPV-hbt |
Head Bolts |
RPV-hdb |
Head (bottom) |
RPV-hdt |
Head (top) |
RPV-ici |
In-Core Instru. |
RPV-noz |
RPV Nozzles (incl. Instr.) |
RPV-pen |
Penetrations |
SG-mwb |
Manway Bolts |
SG-mwy |
Manway |
SG-noz |
SG Nozzles |
SG-shl |
Shell |
SG-tbs |
Tube Sheet |
SG-tub |
Tube |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The database relies upon LERs submitted by plants under the requirement of 10 CFR 50.73. Of the LERs reviewed for this effort, the two most commonly cited reporting requirements (each LER must reference the requirement that necessitates the LER.) are 50.73(a)(2)(i)(B) and 50.73(a)(2)(ii)(A). These are described below.
50.73(a)(2)(i)(B) — Any operation or condition which was prohibited by the plant’s Technical Specifications. Westinghouse Standard Tech Specs (NUREG-1431, Vol. 1, Rev. 2, June 2001, Section 3.4.13) related to RCS leakage are as follows.
RCS operational leakage shall be limited to:
a. No pressure boundary leakage
b. 1 gpm (3.8 lpm) unidentified leakage
c. 10 gpm (38 lpm) identified leakage
d. 1 gpm (3.8 lpm ) total primary to secondary leakage through all steam generators (SGs), and
e. 500 gallons (1,900 liters) per day primary to secondary leakage through any one SG Pressure Boundary Leakage is defined as leakage through a non-isolable fault in an RCS component body, pipe wall, or vessel wall (except SG leakage). Leakage past seals and gaskets is not considered pressure boundary leakage.
50.73(a)(2)(ii) — Any event or condition that resulted in: (A) The condition of the nuclear power plant, including its principal safety barriers, being seriously degraded; or (B) The nuclear power plant being in an unanalyzed condition that significantly degraded plant safety.
NUREG-1022, Rev. 2 clarifies statement (A) as:
This criterion applies to material (e. g., metallurgical or chemical) problems that cause abnormal degradation of or stress upon the principal safety barriers (i. e., the fuel cladding, RCS pressure boundary, or the containment). Abnormal degradation of a barrier may be indicated by the necessity of taking corrective action to restore the barrier’s capability. . .
PWR tech specs also contain reporting guidance (via LERs) associated with the plants SG tube surveillance program. Typically, this reporting requirement is triggered when an inspection reveals that greater than 1% of the tubes in a SG are found to be defective (i. e., greater than 40% thru-wall crack).
Submitted by Professor Larry Hochreiter of Penn State University
Comment: We have revised the data for pipe breaks and leaks following the suggestions from the NRC staff and have made comparisons to the NRC expert solicitation panel’s break spectrum failure frequency plots that were developed in support of the revisions to 10CFR50.46. We had discussion with C. Gary Hammer and other members of the NRC staff on our original comparisons and they suggested that we limit our comparisons to just the class 1 piping for the reactor since that was the class of piping systems that the NRC break spectrum frequency study were originally developed for. Therefore, we eliminated all piping failure, leaks, and cracking data from the mixed data base that we had originally developed such that only the class 1 piping information remained. A CD with the revised data is included with this letter. We also reviewed the class 1 systems with the NRC to make sure that we isolated only those systems which should be compared to the estimated break frequency curves. When comparing the remaining data to the NRC break frequency curves, we used approximately the same break size bins that were used in the NRC study. We also normalized the individual bins of data using the approach recommended by the NRC in which the number of breaks or breaks and leaks were normalized on the numbers of effective full power days.
We have separated out "Breaks" from the total data base as separate plots for both PWRs and BWRs as shown in Figures 1 and 2. As these figures indicate, there are no large breaks in the class 1 piping. However, for the smaller breaks, the data clearly lies above the estimated break frequencies estimated in the NRC solicitation study. The data is above the 95% percentile as compared to the NRC break frequency plots indicating that even smaller breaks are more frequent as compared to the estimated frequencies from the solicitation study.
We also have combined "Breaks" and "Leaks" together on separate plots for PWRs and BWRs as seen in Figures 3 and 4. The rational for this is that leaks are really breaks since the pipe has failed and is leaking. We used the size of the pipe for when combining the leaks with the breaks. This may not be the best method of comparing the data to the break frequency distribution and it may appear to bias the results since there are only leaks for the larger pipes and not breaks. However, it is not clear that, given a set of conditions, a leak in a large pipe could grow to a full pipe size break. Say for example, under a safe shutdown earthquake condition. Also, this grouping could be viewed as being conservative since the original premise is that the pipes should not leak in the first place.
All the data, breaks and leaks, when combined in this form indicate that there is a significant difference between the existing data and the break spectrum failure frequencies from the NRC expert solicitation study. This indicates to me that we should not be revising 10CFR50.46 by introducing a "transitional break size" and reducing the mitigation capabilities of the plant’s ECC systems and defense in depth for the larger break sizes. I believe that we need to maintain the full functional capability of the robust ECC systems that we have in the current operating plants for all possible break sizes up to and including the full double-ended break of the reactor coolant piping. The break and leak data for the class 1 piping
indicates to me that we need the full functional ECC systems and their supporting systems and they should not be compromised or sacrificed by revising 10CFR50.46.
Response: There are several issues with the analysis performed by the commenter with respect to both the pipe “break” frequencies plotted in Figures 1 and 2 and the pipe “break and leak” frequencies in Figures 3 and 4. Figures 1 and 2 above were created by identifying “breaks” from a database of piping failures to construct separate PWR and BWR break frequencies. As depicted in Figures 1 and 2 above, the resulting plot clearly lies above the estimated break frequencies estimated in the expert elicitation documented in NUREG-1829.
The biggest issue pertaining to the commenter’s results is with the integrity of the database used in the analysis which was supplied as part of the commenter’s public comment. The database appears to be similar or identical to the database developed in Bush, et. al, “Piping Failures in US Nuclear Plants: 19611995,” SKI 96.20. This can be surmised from the fact that there are only two events as recent as early 1996 in the database, and the database contains the similar erroneous information to the Bush et. al. report for selected audited events. The Bush et. al. database was never validated and it was also the foundation of EPRI TR-111880. An independent review of this database[19] found that a large percentage of the database records involved non-piping failures. The EPRI report was subsequently withdrawn due to requests by EPRI members.
The authors reviewed the larger pipe events that could be considered “breaks” and used in constructing Figures 1 and 2. The authors examined events classified as a breakage, rupture, failure, or severance in the database for 2 inch diameter or greater pipes in both BWR and PWR plants. This approach was necessary because the commenter neither definitively identified the data used to construct Figures 1 and 2, nor provided details supporting the calculations. Events were checked against the OPDE Database Rev. 0.e, dated 24 March 2004 which has been verified. In several cases, the original failure references were obtained for verification.
Nineteen events were reviewed in all and this data verified many of the reported problems with the Bush et. al. database. Almost all of the database records contain some error or inconsistency. Many of the listed events cannot be referenced to a known event. Common problems include incorrect event dates, references, pipe sizes, or break sizes. The failure classification type (i. e., leak, rupture, severance, etc.) was also found to be both inconsistent and inaccurate in several cases. In spite of these problems, the authors searched the OPDE Database for events that likely correspond to the break events identified in the commenter’s database. Approximately 15 events were identified which either could be definitively matched to an event in the commenter’s database or at least represented by a pipe break at the listed plant in a similar system. None of break events occurred in an unisolable segment of the reactor coolant pressure boundary piping. This is the fundamental definition of a LOCA, and the objective of the elicitation was to estimate LOCA frequencies. The only basis for comparison with the elicitation results is an estimation of LOCA frequencies. No other basis, including a comparison of piping failure frequencies, is consistent with the elicitation results. Hence, the Figure 1 and 2 comparisons with the elicitation results are invalid without further consideration.
However, there are other inconsistencies and errors in the commenter’s analysis. The largest “failed” PWR piping in the database is for a 4 inch diameter pipe. Leak data is reported in the commenter’s database for PWR piping up to 13 inch diameter. However, Figure 1 above depicts breaks up to 12 to 15- inch diameter. The basis for reporting PWR break data greater than 4 inch diameter is not supported by the supplied database. There are other, less significant, but still consequential problems with the commenter’s analysis. Contrary to the commenter’s statement that all the breaks and leaks represent “Class 1 piping” systems, most of the events occurred in lower grade piping. This is not a trivial distinction because Class 1 piping is subject to more rigorous design, testing, and inspection requirements than other piping systems within the plants. In addition, the commenter’s database contains nearly twice the number of carbon steel failures in PWR plants than in BWR plants. The finding is heavily biased by the consideration of non-ASME Code, FAC-susceptible piping in the commenter’s database. Also, several of the rupture sizes appear to overestimate either the actual pipe size or the rupture size which occurred. The distinction is important because the size of the rupture, not the piping size, determines the rate of loss of system fluid. A partial failure of a large pipe can lead to substantially lower fluid loss rates than expected if the pipe completely burst.
The leak data comparisons presented in Figures 3 and 4 are similarly misleading. The most substantive issue is the commenter’s contention that “leaks are really breaks since the pipe has failed and is leaking.” This is simply untrue. Pipe breaks occur once a flaw reaches a critical size such that it grows unstably at the applied load level. The piping can then not support required operational loading, a readily-apparent hole or breach forms, and internal fluid is released at a rate beyond the reactor water make-up system of the plant. Conversely, piping having leaking joints or through-wall cracks typically remains intact and continues to support internal and external stress without imminent failure. These cracks may continue to grow until imminent failure occurs, but the difference between the leaking crack size and the rupture crack size increases with pipe size. Hence, larger pipes provide more margin against failure after a leak appears. The NRC’s LBB approach for large pressure boundary piping is based on the premise that a leaking through-wall flaw can be detected at normal operating loads via the plant’s normal leakage detection systems before the flaw reaches a size that could grow unstably at emergency or faulted load conditions. The commenter’s database records support this premise as the larger diameter, Class 1 piping events are leaks that were found and repaired prior to a catastrophic failure. The LBB approach has used to ensure compliance of GDC-4 requirements for over 20 years.
Additionally, the commenter’s analysis of the leak data using the supplied database suffers from similar issues as the break data. A majority of these leak events actually occurred in secondary-side plant systems and the pipe leak did not result in loss of pressure boundary coolant. Therefore, longer term failures of these pipes (presuming that they remained undetected and the associated flaw continued to grow to failure) would not have resulted in a primary LOCA event.
In contrast to the commenter’s analysis, a review of some of the more well-established and well — recognized piping failure databases (PIPEex, OPDE, and SLAP) has found 1 event that could be characterized as a SB LOCA in PWR plants and no events that could be characterized as SB LOCAs in BWR plants (See Section 7.10 of NUREG-1829). No MB or LB LOCA events have occurred. As previously discussed (See responses to GC3, GC4, GC5, GC6, and GC7) the LOCA elicitation results for the present day estimates are consistent with this operating experience. The comparison of the elicitation results and operating experience is also provided in Section 7.10 of the revised NUREG report.
The following tables display the detailed results of Base Case #1 calculations. The legend of degradation categories is shown in Table E.3. LOCA data are presented on the tables listed below:
PWR pipe Table E.4
PWR passive non-pipe Table E.5
BWR pipe Table E.6
BWR passive non-pipe Table E.7
Table E.3 Degradation Categories
Deg Mech |
DM Description |
MA |
Material Aging |
FDR |
Fabrication Defect and Repair |
SCC |
Stress Corrosion Cracking |
LC |
Local Corrosion |
MF |
Mechanical Fatigue |
TF |
Thermal Fatigue |
FS |
Flow Sensitive (includes FAC and E/C) |
UNK |
Unknown |
E. 1 “Reactor Safety Study: An Assessment of Accident Risks in U. S. Commercial Nuclear Power Plants,” WASH-1400, U. S. Nuclear Regulatory Commission, October 1975.
E. 2 Beliczey, S., and Schulz, H., “Comments on Probabilities of Leaks and Breaks of Safety-Related Piping in PWR Plants,” International Journal of Pressure Vessel and Piping, Vol. 43, pp. 219 — 227, 1990.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
For question set 3B, Panelist A has previously listed the following important PWR piping systems and their individual contributions to Category 1 LOCAs after 25 years of plant operation (see Table J. A.1). For this example, this panelist does not expect the relative contributions to change after either 40 or 60 years of plant operation.
Table J. A.1 PWR Piping System Contributions for Category 1 LOCAs
Case |
Piping System Lines |
Percentage Contribution |
1 |
Instrumentation |
50 |
2 |
Drain Lines |
10 |
3 |
Reactor Coolant Pressure Hot Leg |
10 |
4 |
Chemical Volume Control System |
10 |
5 |
Safety Injection System Accumulator |
10 |
The reactor coolant pressure hot leg has an associated base case. The base case geometry is a 30 inch diameter pipe, manufactured from Type 304 stainless steel with an Alloy 600 safe end. The safe end to pipe weld is a nickel-based bimetallic weld. The base case degradation mechanisms are thermal fatigue and PWSCC. The loading is pressure, thermal, residual stress, dead weight, with a pressure pulse transient. Panelist A next needs to estimate the ratio between all Category 1 LOCAs in the hot leg compared to those represented solely by the base case conditions. His results are summarized in Table J. A.2.
Table J. A.2 Panelist A’s Ratio Between Entire Piping System and Base Case Contributions for
Base/ref case |
25 Years |
40 Years |
60 Years |
||||||
5% LB |
MV |
5% |
5% LB |
MV |
5% |
5% LB |
MV |
5% |
|
UB |
UB |
UB |
|||||||
SL BC |
1 |
2 |
3 |
1 |
3 |
4 |
1 |
4 |
5 |
Panelist A believes that the total hot leg Category 1 LOCAs are twice the number represented by those captured by the base case conditions after 25 years of service (present day). However, this ratio increases with time until after 60 years, the total hot leg Category 1 LOCA probability is 4 times what is predicted by the base case conditions. This initial difference is due to the fact that Panelist A believes that an equal number of Category 1 LOCAs will be due to thermal fatigue of other piping materials than are represented by the base case (specifically cast stainless steel and stainless steel clad carbon). This panelist also believes that these mechanisms will become more apparent with time than either thermal fatigue or PWSCC of stainless steel material. Future increases are also included for unanticipated mechanisms.
The conditional LOCA probability for a given piping system, degradation mechanism, and emergency faulted load can be determined by multiplying the likelihood curve (L, red in Figure J. B.1) by the conditional piping failure probability (P, black in Figure J. B.1) and then integrating over all the possible damage states. This conditional LOCA probability will likely be a function of LOCA size (pipe size), piping system, applied emergency faulted load, and degradation mechanism. Figure J. B.1 below provides a schematic for a fixed ASME Service Level B (SLB) load, piping system (instrument lines), LOCA size (Category 1), and degradation mechanism (PWSCC). The curve shapes/trends in Figure J. B.1 are an illustration and do not represent any panelist opinion. A separate set of likelihood and conditional failure probability curves exists for each unique combination of these four variables.
50% TW Perceptible Leak Tech. Spec. Leak |
Figure J. B.1 Conditional Failure Probability for Service Level B Loading for Category 1 LOCAs in
PWR Instrument Lines due to PWSCC
The elicitation question first asks each panelist to identify only the significant piping system and degradation mechanisms to conditional LOCA for each LOCA size category (EQ 5A.1). Then, the panelist will pick seismic base case conditions for either the PWR hot leg or BWR feedwater seismic base cases described in the “Seismic Base Case” document (EQ 5A.2).
Next, the panelist will provide ratios of the relative likelihood and conditional failure probabilities (Figure J. B.1) among three different damage states (a 50% through-wall crack, a perceptible leak, and a technical specification leak) at a fixed load for each LOCA size category. All the estimates are initially for a
Service Level B load. Here are the ratios that will be provided: Ptsi/Pbc (EQ 5.A.3.1), Ppi/Ptsi (EQ 5A.3.2), and P50/Ppl (EQ 5A.3.3). The Pbc value is defined in the “Seismic Base Case” document while all other variables are defined in Figure J. B.1. In this way, all answers are linked back to the case quantification of the base-case conditions. The next elicitation question (5A.4.1) asks for the relationship between the Service Level D and Service Level B event for the given degradation mechanism and system which is the ratio of Ptsl at SLD/Ptsl at SLB. The other questions ask for the ratio of Ppl/Ptsl (EQ 5A.4.2), and P50/Ppl (EQ 5A.4.3) for solely the SLD loading event.
Finally, the likelihood of the 3 damage states is estimated. The likelihood base case to be used for each piping system and degradation mechanism listed by each panelist will be the leaking frequency for that system over the piping size range of interest for all degradation mechanisms (Lbc). Obviously, this frequency will be more than the frequency for any single degradation mechanism, but dominant degradation mechanisms may provide a ratio close to 1. This frequency has not been quantified, but it will be after the elicitation once a complete listing of important systems and degradation mechanisms has been provided by the panelists. However, leaking rate propensity has been provided as part of the piping base case analysis for those systems (i. e., see Bill Galyean and Bengt Lydell’s results). The elicitation questions ask each panelist to provide the following three likelihood ratios: Lpl/Lbc (5A.5.1), Ltsl/Lpl (5A.5.2), and L50/Lpl (5A.5.3). These variables are defined in Figure J. B.1.
The curve is developed for the 3 damage states and 2 loads in an attempt to capture the most significant contributions to the conditional LOCA probability. These events will be interpolated and extrapolated as necessary to develop continuous relationships as a function of damage state and loading magnitude. This information can then be combined with a plant’s seismic hazard curve as well as knowledge about the relationship between the hazard curve and actual piping stresses to determine actual LOCA frequencies due to a seismic event. They could also be used to determine the LOCA contribution of other large transients knowing both the transient frequency and the relationship between the transient applied loading magnitude and the ASME service level loading magnitudes.
PIPING BASE CASE RESULTS OF VIC CHAPMAN
Summary of Benchmarking Analysis
Carried out Using ‘RR-PRODIGAL’
G. 1 General Background to RR-PRODIGAL
RR PRODIGAL is a basic fatigue failure probability model developed by Rolls Royce for the Naval Nuclear program. When analysing a weld, it first simulates the weld construction in order to determine a start of life defect distribution and density for both buried and surface breaking defects.
A failure probability using standard linear elastic fracture mechanics methods is then evaluated for both the buried and surface breaking defects (assumptions about break through of buried defects to surface defects are based on the ASME criteria). Failure is achieved when the defect either exceeds the R6 failure criteria or simply grows through to the full thickness. The failure probability for all initial defects is then combined to form the total failure probability.
For non-weld areas, a probabilistic crack initiation analysis is carried out with a correlated crack growth analysis to failure. This correlation means that short times to crack initiation imply that a fast crack growth follows this initiation. There is no positive data to confirm or deny this proposition. It was chosen simply because it is pessimistic.
The modelling contains a routine to assess the growth of the defect around a welded pipe at the same time as the defect grows through the weld thickness, however, this part was not used in this assessment. RR-PRODIGAL does not, at present, contain a routine to evaluate the crack growth of a through wall defect around the outer surface of a pipe weld.
At failure, the model evaluates a critical through wall defect size based again on the R6 criteria.
At present RR-PRODIGAL does not contain a verified and validated assessment of the PWSCC degradation mechanism.
There are several publications that describe RR-PRODIGAL, which include a recent benchmarking exercise as part of a European initiative. References G.1 and G.2 should provide sufficient information for any readers wishing to obtain further information on this code.
The detailed results from the elicitation procedure are presented in this appendix. Each panelist’s quantitative elicitation responses can be found through the “Electronic Reading Room” link on the NRC’s public website (http://www. nrc. gov/) using ADAMS. The documents are found in ADAMS using the following accession number: ML080560005. Both the quantitative (i. e., numerical LOCA frequency estimates) and qualitative results (i. e., rationale) are presented in this appendix. The quantitative results are often presented in the form of box and whisker plots. Box and whisker plots (often referred to simply as box plots) are a graphical representation of a data set. A box plot is typically based on five data points: the minimum and maximum of the data set plus the median, lower quartile (LQ), and upper quartile (UQ) associated with the data set. The LQ and the UQ are the 25th and 75th percentiles, respectively, of the data set.
An example box plot is shown in Figure L. 1. The region defined by the LQ and UQ is drawn as a shaded box with a vertical line through the median. In this appendix, the 10th and 90th percentiles are also indicated in the box plots with short vertical lines. Note, for the plotting program used to generate these box plots, the vertical lines (representative of the 10th and 90th percentiles) and the associated horizontal connecting them only appear when there was at minimum of nine points in the data set being plotted. Finally, the box plots are also overlaid with a horizontal scatter plot of the data set, with the left-most point being the minimum value and the right-most point being the maximum value in the data set. Note that the 10th and 90th percentiles are not necessarily points in the data set. The range in the data set encompassed by the shaded box (i. e., the range between the 25th and 75th percentiles) is referred to as the interquartile range (IQR) of the box plot. In the box plots in this appendix, letter designators are often included for the minimum and maximum values (e. g., the letters G and D in the example shown in Figure L.1). The letters designate the code for the panelist whose estimated value was either the lowest or the highest value of all the panelists who provided responses for the quantities in the data set.
G •1 • і |
|||
0 |
——— 1—— 10 |
——— 1—— 20 |
——— 1—— 30 |
——— 1—— 40 |
——— 1—— 50 |
60 |
Data Array Values
Figure L.1. Example “Box and Whisker” Plot
Generally, the source of the qualitative responses (i. e., the rationale) came from the individual elicitations although there were some opinions expressed during the various plenary panel meetings that were also included. For each of the individual elicitations, minutes were taken. Minutes were also taken at each of the plenary panel meetings (see Appendix B). In addition, the participants often provided a handout to lead the discussion at their individual elicitations. After each elicitation, most of the participants also provided formal written responses to the elicitation questions. It was from these minutes, handouts, and written responses that the rationale provided below was extracted. Finally, each of the elicitations was audio taped and each meeting was video taped to provide a permanent record of the proceedings.
Most of the panelists believe that precursor events (e. g., cracks and leaks) are a good barometer of LOCA susceptibility. This is reflected in the fact that almost all the panelists anchored their responses against some form of the available operational experience data. A distinct advantage of the operational experience data is its inclusion of all degradation mechanisms which have emerged to date, while the PFM approaches only address selected degradation mechanisms. The advantage of the PFM approaches is that they are best suited for addressing LOCA size and operating time effects. A number of panelists used the PFM results as a basis for adjusting the operational experience data in this manner.
A major assumption made in the elicitation procedure is that all components, both piping and non-piping, were fabricated in accordance with applicable code standards, e. g., there were no counterfeit bolts used.
There is no service data associated with Cat0 LOCA events. Therefore, the estimation of conditional failure probabilities is based on zero-failure statistics. Since not all flaws propagate through-wall if left unattended, an alternative to the approach in Section D.5.2 (constrained noninformative prior) would be to use Jeffrey’s noninformative prior and to assume all flaws (non-through wall and through wall) as pressure boundary integrity challenges. The result would be conditional failure probabilities that are closely approximated by the power law (Equation D.8), however. It is acknowledged that this is just one way of representing the current state-of-knowledge with respect to gross Code Class 1 pipe failure. It is not a physical model of flaw propagation given its interactions with certain loading conditions and pipe stresses.
An application of a parametric attribute/influence method has yielded results as summarized in this section. Central to the method is the processing and interpretation of service data on Code Class 1 piping. A Markov model of piping reliability is used to develop time-dependent LOCA frequencies.
D.7.1 Discussion of Assumptions
A parametric attribute/influence method is applied to five base cases. Three types of assumptions are made in the analysis; global assumptions (applicable to all five base cases), BWR-specific assumptions and PWR — specific assumptions:
Global Assumptions
• Pipe failure results from observable degradation mechanisms and loading conditions. A statistical evaluation of service experience data therefore provides a sufficiently accurate basis for piping reliability analysis.
• The PIPExp database is of sufficient completeness and depth to support an application of the parametric attribute/influence methodology. This database addresses piping performance in response to both anticipated and unanticipated loading conditions.
• The effect on piping reliability from pressure, deadweight, weld residual stresses, thermal loading, and thermal stratification is implicitly accounted for in the PIPExp database. This database also accounts for the effects from inadvertent over-pressurization and relief valve actuation, water hammer and seismic[12] events.
BWR-Specific Assumptions
• The BWR-specific LOCA frequencies are assumed to be representative of a plant with IGSCC Category D and E welds operating with normal water chemistry (NWC). The pipe failure database includes plants with hydrogen water chemistry (HWC) and NWC. This study did not differentiate between plants with weld overlays and HWC versus plants with weld overlays and NWC, however. This study shows improved water chemistry together with weld reinforcements to lower the weld failure rates by about a factor of ten (10).
• Because of service conditions and piping arrangements, flow accelerated corrosion (FAC) is not viewed as a significant degradation mechanism affecting Code Class 1 feedwater piping.
Degradation involving wall thinning is therefore not viewed as having an effect on the time — dependent LOCA frequency.
PWR-Specific Assumptions
• The estimation of RC-HL weld failure rates is based on the assumption that the observed (in 4th quarter 2000) weld degradation at V. C. Summer is a circumferential flaw in the RPV nozzle-to-safe-end weld. This assumption is believed to result in an over-estimation of the actual weld failure rate.
• Relative to PWRs of Westinghouse design, the pipe failure database includes no records on through-wall flaws in large-diameter pressurizer surge line welds. The analysis assumes that the piping is susceptible to thermal fatigue of sufficient magnitude to potentially cause a flaw in the through-wall direction.
The SCSS is an NRC-sponsored database maintained by Oak Ridge National Laboratory (ORNL). It is a web-accessible database of LERs that can execute searches using a variety of criteria. It can be accessed at:
The following search criteria were used to generate the LER portions of the non-pipe database.
LER SYSTEM EVENT SEARCH CRITERIA
Primary System(s) =SAB, SAF, SAE, SAA, SAD, SAI, SAH Interfacing System(s) =Any Include Trains/Channels =Yes Include Components =Yes Happening(s) =Any Event Cause(s) =Any
Event Effect(s) =BH, BF, BE, BI, DE, BN, BL, BK, BP, BX, BC, BB, BA, BD Event Timing(s) =Any Detection Methods(s) =Any Nuclear Plant =Any Beginning Event Date =01/01/1990 Ending Event Date =1/1/2003 Maximum LERs =2000
This search returned 1,036 LERs. Basically, this search criteria looks for any leaking or cracking event associated with any primary coolant related system. The above search criteria rely upon the coding effort performed by the staff at ORNL as part of the SCSS program. In that effort, each LER is reviewed and characterized for possible relevance to each related system. This characterization includes both actual and possible system failures. Therefore, these search criteria returned both pipe and non-pipe failures, as well as many “non-failure” events. Each of the returned 1,036 LERs was reviewed and approximately 80% (823 LERs) were judged to be non-failures and coded as not-applicable (NA). Most of these NA events were of the type where an engineering review or some other analysis was performed by the plant, and it was found that a pipe was inadequately (compared to the design requirements) constrained such that if an earthquake were to occur, there was an increased chance that the pipe might fail. Another common “non-failure” example is of a problem unrelated to the integrity of the primary coolant system, which would have adversely affected the ability of the plant to respond to a loss of coolant accident (i. e., a failure of the primary coolant system). These potential or possible issues were judged to not be actual failures and hence were deleted from the list. A further 34 LERs were removed from the set of LERs when they were found to document problems associated with pipe defects (or pipe-weld defects).
In general, a point estimate of the frequency of pipe failure (where ‘failure’ includes both small and large leaks and through-wall cracks, but excludes partial-through wall cracks), A, is given by the following expression:
A = nF — NT
Where:
nF = the number of failure events including both small and large leaks in the operational experience data;
T = the total time over which failure events were collected;
N = the number of components that provided the observed pipe failures.
A point estimate of the total frequency of flaws (cracks and leaks), ф is given by the following expression:
Where:
nC = the number of crack or flaw events
f = the fraction of welds inspected for cracks or flaws
PFD = the probability that an inspected weld will find an existing flaw
Nearly all through-wall leaks are found from independent observations such as routine leak inspections and not from NDE inspections. However, part-through cracks are only typically found by NDE and thus the number is a function of the number of inspection locations. In Equation H. C.2 we account for the observed cracks in the data base and the fact that only a fraction (f) of the welds in the database are inspected according to ISI programs looking for cracks. The number of flaws actually discovered in ISI is subject to a finite NDE reliability, which is characterized by the factor PFD.
If we now take the ratio of ф to A, we get an expression for the factor by which to multiply the pipe failure rate to obtain the flaw (non-through wall crack) rate:
Where:
Rc/f = Number of non-through wall cracks per leak event:
One approach to assess the RC/f ratio is to evaluate those records where both cracks and leaks were found during a single inspection of a component of interest. Ideally, the best data would be found in those instances where the component was 100% inspected. Without complete inspection, some assumptions about the inspection coverage, f, are required to assess this ratio. An example of this approach and the effect of the inspection coverage and POD is provided by analyzing the database for CRDM nozzle failures. For the component type ‘CRDM Nozzles’ in B&W PWR plants the database includes 6 LERs (= 6 database records) as identified in Table H. C.1. A detailed review of each of these LERs revealed multiple failures and degradation. Equation (C.3) together with an assumption about the inspection scope f) makes it possible to estimate RC/f.
Table H. C.1 B&W CRDM Nozzle Failures in ‘Non-Pipe’ Database
Plant |
Date |
LER Number |
No. Components Leaking |
No. Components Cracked |
Population |
Comment |
Oconee-3 |
2/18/2001 |
2001-001 |
9 |
N/A2 |
69 |
Expanded inspection of an additional 9 nozzles. No recordable flaws. |
Oconee-3 |
5/2/2003 |
2003-001 |
2 |
N/A |
69 |
RVH replaced |
Crystal River-3 |
10/1/2001 |
2001-004 |
1 |
N/A |
69 |
5 flaws found in CRDM Nozzle #32. Expanded inspection of 8 nozzles found no flaws. |
Three Mile Island-1 |
10/12/2001 |
2001-002 |
5 |
7 |
69 |
Inspection scope included 12 nozzles |
ANO-1 |
3/24/2001 |
2001-002 |
1 |
N/A |
69 |
Visual inspection only of remaining nozzles. |
ANO-1 |
10/7/2002 |
2002-003 |
1 |
6 |
69 |
NDE of all nozzles |
Totals: |
19 |
13 |
Estimates of RC/f for the data set in Table H. C. 1 are presented in Table H. C.2 for different assumptions about the POD and fraction of welds inspected. This analysis also assumes that the inspection criteria and the cracking characteristics of the events listed in Table H. C.1 are representative of the entire population. The fraction of welds inspected is a function of the ISI program requirements. In Table H. C.2, a LB for f is calculated using insights from piping reliability studies. This low f estimate results in the high RC/f estimate presented in the table.
The current ASME Section XI requirements are to inspect 25% of the Class 1 pipe welds and 7.5% of the Class 2 pipe welds. The current inspection practice for most if not all plants calls for the same welds to be inspected each inspection interval as opposed to randomly selecting a different set of welds for each interval. When cracks or significant flaws are found, the ASME code requires that an expanded search be made; however, the frequency of flaws and failures is so rare that this requirement adds very few additional inspections to the total population of inspected welds. Using data from an operating 4-loop Westinghouse PWR unit on the number of Class 1 and Class 2 welds of 1,605 and 1,800, the following estimate of the parameter f is obtained for Westinghouse PWR plants:
N/A in Table H. C.1 means that a full-scope NDE was not pursued.
1, 605(0.25)+1,800(0.075) _ 0 f_ (1,605 +1,800) _ ‘
We assume this estimate of f to be representative of the non-pipe components. With additional assumptions about the reliability of the NDE we get the results as indicated in Table H. C.2.
Table H. C.2 Estimates of RC/f for B&W CRDM Nozzles
Parameter |
Data Source |
PWSCC |
||
High Est. |
Median Est. |
Low Est. |
||
Number of cracks |
Table H. C.1 |
NA |
13 |
NA |
Number of leaks |
Table H. C.1 |
NA |
19 |
NA |
Fraction of components inspected, f |
Equation (H. C.4) — LB |
1.0 |
0.5 |
0.157 |
Rc/f with Pfd = 0.5 |
— |
9.72 |
3.74 |
2.37 |
Rc/f with Pfd = 0.75 |
— |
6.81 |
2.82 |
1.91 |
Rc/f with Pfd = 0.9 |
— |
5.84 |
2.52 |
1.76 |
Hence, the relative number of flaws and leaks observed does not predict the relative frequency of flaws and leaks at a given weld. The estimate for the ratio of cracks to leaks obtained in Table H. C.2 reflects the degree to which components are exposed to PWSCC and inspection coverage.
REACTOR VESSEL LOCA PROBABILITY
BASE CASE ANALYSES
(BWR VESSELS AND PWR TOP HEAD NOZZLES)
Submitted by Westinghouse Owners Group
Comment: Method of Aggregating Individual Frequency Responses into a Group Response. For this study the geometric mean aggregation method was used instead of the arithmetic mean method or mixture distribution methods, which would give higher mean values of LOCA frequency. The reviewers concur that the geometric mean is most representative of the consensus of the group (expert panel). As an example, consider possible individual responses the different group means for the distribution of factors on a baseline frequency in the following table (see WOG response for table):
Number or Responses |
Value of Factor |
1 |
0.01 |
2 |
0.1 |
3 |
1.0 |
2 |
10 |
1 |
100 |
For this example, the arithmetic mean value for the 9 responses is 13.69 while the geometric mean value (average of the logarithms) is 1.0, which seems to be much more representative of the group’s opinions. Part of the reason for this is that the probabilities and frequencies of failures of structural components, such as piping, are normally expressed as orders of magnitudes much less than one. Uncertainties on these values are also expressed as factors instead of differences because the physical contributions to structural failures (leaks and breaks), such as flaw sizes and crack growth rates, are also known to be log-normally distributed. Use of logarithmic distributions and geometric means is also consistent with NRC Guidance on Risk-Informed In-service Inspection for Piping (Draft Report NUREG-1661, January 1991). Figures 3.3 and 3.4 of this guidance show the range of frequency estimates from expert elicitation, plotted on a logarithmic scale, for failure of auxiliary feedwater system components and the reactor pressure vessel, respectively. Figure 4.6 of this same report shows the uncertainty in the best estimate (median value) of piping failure probability, calculated using probabilistic fracture mechanics methods, to also be logarithmically distributed. Scanned copies of these figures are provided in Appendix A[20] of the WOG’s public comment document.
Response: The authors agree that the geometric mean aggregation supports the elicitation objective of developing a consensus group estimate by reducing the effect of either wide differences in the individual estimates, or the effect of a single estimate which is significantly higher than the others. As documented in Section 7, geometric-mean aggregation, in this study, produces group estimates which approximate the median of the individual estimates. In the example cited by this commenter, the arithmetic mean estimate of 13.7 is larger than all but one of the responses and does not approximate the median of the individual estimates.
However, as indicated in the Executive Summary, because the alternative aggregation methods can lead to significantly different results, a particular set of LOCA frequency estimates is not generically recommended for all risk-informed applications. The purposes and context of the application must be considered when determining the appropriateness of any set of elicitation results. This position is consistent with the recommendation of the NRC’s Advisory Committee on Reactor Safeguards (Letter from W. J. Shack to D. E. Klein dated December, 20, 2007, Subject: Draft Final NUREG-1829, “Estimating Loss-of-Coolant Accident (LOCA) Frequencies Through the Elicitation Process,” and Draft NUREG-XXXX, “Seismic Considerations for the Transition Break Size”, ADAMS Accession Number ML073440143).