ML20211H423

Comment Supporting Draft NUREG-1602 & DG-1061 & Encourage NRC to Carefully Consider Comments as Well as Encl Comments

ML20211H423
Person / Time
Site:	Seabrook
Issue date:	09/30/1997
From:	Callendrello A NORTH ATLANTIC ENERGY SERVICE CORP. (NAESCO)
To:	Meyer D NRC OFFICE OF ADMINISTRATION (ADM)
References
FRN-62FR34321, RTR-NUREG-1602, RTR-REGGD-XX.XXX, TASK-*****, TASK-RE 62FR34321-00021, 62FR34321-21, NYN-97100, NUDOCS 9710060370
Download: ML20211H423 (7)

Text

i f7J61^

(-.2 F< JYJLJ

/7Animhrr)

Je'n.1s; /997

['y, ;yorgy MW Nont, ra,,aa ng sawa.,,, ann RECEIVED in na= 3*

'$ A(lad (IC Seabrook, Nil 03874 J/

1791 OCT M 8: 46 (*3) **

AULES & DR C.%NCH USfBC September 30,1997 NYN 97100 United States Nuclear Regulatory Commission Attn.: David L. Meyer Chief, Rules and Directives Branch Division of Administrative Services Office of Administration Washington, DC 20555-0001 Seabrook Station Comments on Dran NUREG-1602 and DG-1061 On June 25,1997 (62FR34321), the NRC issued for comment, a series of documents describing-how risk insights would be incorporated into the regulatory decisionmaking process. The documents included:

NUREG 1602 -

The Use of PRA in Risk-Informed Applications

~ DG-1061 An Approach for Using Probabilistic Risk Assessment in Risk-Informed Dccisions on Plant-Specific Changes to the Current Licensing Basis Draft SRP Chapter 19 Use of Probabilistic Risk Assessment in Plant-

[I Specific, Risk Infonned Decisionmaking DG-1062 and SRP Insen' ice Testing Chapter 3.9.7 DG-1065 and SRP Technical Specifications Chapter 16.1 DG-1064 Graded Quality Assurance North Atlantic Energy Service Corporation (North Atlantic), the operator of Seabrook Station, has reviewed this material and we have several general t id specific comments on NUREG-1602 and DG-1061. These comments are enclosed for your consideration ~. Although our comments are confined to these two documents, several cf the comments apply to other of the documents as ngony

/) t

)

[1tLNNbh lElllB)EIlEII 9710060370 970930-PDR NUREG 1602 C PDR

]

United States Nuclear Regulatory Commission Attn.: David L Meyer:

- NYN 97100/Page 2 i

well.~ 'In general,= we feel that these documents will provide useful guidance and direction in f

several areas,-however, several aspects are likely to contribute-to confusion on the NRC's expectations and others will significantly increase the required level of effort in the PRA area

- without 'a corresponding increase in the value of the analysis for improving plant safety. North

-Atlantic has also reviewe'd the industry comments provided to the NRC in;the'NEl letter of September 22,1997.- We support those industry comments and encourage the NRC to carefully consider these comments as well as our enclosed comments.

lf you have questions regarding our comments, please contact me at (603) 773-7751.

Very truly yours,

- NORTil A'LANTIC ENERGY SERVICE CORP,

/

2 fn'tiionygallendrello~

+_

n Licensing Manager cc: BiffBradley Nuclear Energy Institute 1776 i Street, NW

- Suite 400 Washington, DC '20006-3708 Mr.- Albert W. De Agazio

- Senior Project Manager,

- Project Directorate 1 ' Division of Reactor Projects U.S. Nuclear Regulatory Commission

- Washington, DC 20555-0001 U.S. NRC Resident Inspector (49-NR)

- NAESCO PO Box 300-Seabrook,NH 03874 w

Mww r

m v

W---

-s-r-et-m-

y.

~

y.-

y y-gw c

tr"*-

Comments on DG 1061

1. Section 1.3 states that the policy is adopted because of the " inherent uncedainties in PRA". The NRC PRA policy statement clearly indicated that while PRA explicitly exposes these uncertainties, they are also present in detenninistic analyses. It would be helpful to reiterate this position in DG-1061.

2. Section 2.1, the seventh bullet, states that the staff expects independent peer review of the PRA if used to support proposals to change the Current Licensing Basis (CLB).

This type of review is not required for deterministic analyses and should not be a requirement fbr probabilistic analyses. Note that independent peer review for PRA would typically require review by an outside organization since PRA is typically performed by a small group at each utility. This is not the same requirement as a typical independent engineering review.

3. Section 2.12.1-The NRC is redefining the tenn " full scope" as applied to PRA. The NRC definition is substantively different from the current usage within the industry and will cause unnecessary confusion.

5. Section 3.3-The documentation required as part of a submittal appear to go well l

beyond the requirements for an equivalent submittal based on deterministic analyses.

j There is no apparent justification for a substantially different treatment of l

probabilistic analyses. A summation of the results and key assumptions would be appropriate.

General Comments on NUREG-1602

1. This document was clearly written not as the minimum set of requirements or even the nominal expectations, but as the upper limit with regard to scoping, analysis, and documentation required. The total set of requirements would be appropriate only for use in gerieric applications by the NRC in setting regulations. For any plant-specific application, the requirements are so burdensome that few, if any, applications would be worth the effort required. It was clearly the authors' intent that different levels of PRA Standard would be needed for different applications, depending on their

" complexity". However, no guidance is provided on what would require a " higher standard" versus a lower standard. Without explicit guidance, regulators that have responsibility for specific programs are almost certaia.a think that their program requires the highest standard. At the August workshop, an NRC representative stated that NUREG-1602 was a collection of best practices not requirements. We recommend that this philosophy be clearly stated in the introduction. We also recommend that two major sections need to be added to t'iis document or other documents in the set. First, we recommcnd a section which provides a delineation of several Standards Levels-what minimum requirements are needed to meet Standards 1 through 4, for example. Second, we recommend a section to match up specific 1

l 1

~

applications with Standards Levels so that the requirements are explicitly specined for each apphcation.

2. In the " application impact Considerations" sections throughout, the reference is to changes to the plant's Current Licensing 11 asis (CLB). While this is clearly the area of interest to NRC, it should be noted that the plant can make changes that have no impact on the CLB that overwhelm the significance of the changes to CLB. Thus, CLB does not control risk.

Specific Comments on NUllEG-1602

1. Section 2.1.1-The distinction between internal and external events-putting fire and flood in with internal events-doesn't make sense. There are several important distinctions between internal events and " hazard" events. Ilazard events typically have much higher level of uncertainties than internal events because they usually extrapolate from small events (small fires) to large scale events that cause the damage of interest. Also, hazard events are usually less amem. ole to risk management. They typically are built into the design of the plant and are not influenced by on line maintenance, for example.

2. Section 2.1.1.1, Paragraph 18 " Credit for motor operated valve. "-This section requires " analysis" or " testing" before taking any credit for isolating an ISLOCA.

PRA always relies to some extent on judgment as to whether such a component would function outside its normal operating conditions. It should be sufficient to credit expertjudgment in such an application.

3. Section 2.1.2, " Establishing Success Criteria"-The distinction among various definitions of core damage may be ofinterest from a scientific standpoint, but from an engineering /PRA view, it is not important. Typically the time from high core temperature to melt is short compared with the time from initiation to the high temperature.

4. Section 2.1.2, Paragraph 8 & 9, "The use of realistic success criteria..."-This section discusses a key issue, the use of realistic success criteria versus licensing-bases criteria. This appears to fly in the face of CLB. For example, if realistic success criteria depends on non-safety equipment, it would be outside the scope of CLB.

5. Section 2.1.3.1, Paragraph 3, "An example of where..."-This section uses the term

" permissible" in discussing the use of a simpi!fied fault tree. This wording and simihir words turns this standard into a " cook book". We recommend that wording such as "might be reasonable" would be better than "is permissible".

6. Section 2.1.4.1, Paragraph 6, "The failure to run rate..."-This section (last sentence) requires determining whether tests are a true measure of an actual demand for the 2

component. While this is an important issue, it also is well beyond the scope of an individual plant analysis. This is a generic research issue that should be pursued at a higher level, e.g. EPRI or NRC research.

7. Section 2.1.5.1, " Screening / Excluding Iluman Events"-This section requires previously screened human events to be reconsidered for "every risk-informed regulatory application of the PRA". This could be interpreted as requiring that the PRA be redone for every application.

8. Section 2.1.6.1, " Selecting Truncation Values"-This section states that truncation is done to " simplify the quantification process and make it less time intensive." Our experience is that truncation is done to make the problem solvable-without truncation, the computer codes cannot solve some of the more detailed fault tree and event trees.

9. Section 2.1.6.1, " Estimating Uncertainties"-This section discusses the importance of quantifying and understanding the important uncertainties in the PRA. While we would agree that it is important to have a general sense of the uncertainty band and to know the important uncertainties --those that, ifincreased or decreased, influence the results significantly-- blindly quantifying sequence uncertainties is not very useful.

First, the uncertainty band is usually determined by the first 25 to 50 sequences.

Second, sensitivity analysis is much more useful to determine what uncenainty actually matters. Third, the mean value reflects the upper bound uncertainty, Finally, the general uncertainty bands can be predicted based on what we know about the analysis inputs. Internal events generally have range factors (5th to 95th) of around 10 while external events have RFs of 100 or more. While that is interesting, it is unclear how this would change a decision.

10. Section 3.1.1.1, Paragraph 5, " Recoverability of failed systems..."-This section says that we give no credit for repair of failed hardware in Level 1 PRAs. On the contrary, electric power recovery includes hardware failures implicitly in the recovery curve.

11. Section 3.1.1.1, Paragraph 9, "The mapping from the Level 1..."-This sentence states that the mapping to plant damage state is done "at the cutset level, not the accident sequence level." This is ambiguous. Does cutset level mean system cutsets?

It also seems to be in error. Mapping IS done at the event tree sequence level.

12. Section 5.1.1.1, Paragraph 11, "In the final PRA model..."-The last sentence adds

" aging effects" to the analysis for component and structural fragility evaluation. This may be useful research, but is well beyond what should be covered in a plant PRA.

13. Section 5.1.1.2, " Application Impact Considerations"-This section sounds like we need a seismic expert to look at every change to the CLB. While it is reasonable to hire a seismic expert for a one time review of the plant, it is not reasonable to expect to have such a person available to review every change.

3

14. Section 6-Throughout this section the linking of low power operation with shutdown does not seem to make sense. Low power still requires essentially the same functions and the same Technical Specifications as full power, in shutdown, the initiator of interest is loss of decay heat removal. Event LOCA is a subset of loss of DilR. Also, the Mode 5 and 6 Tech Specs look much different from the Mode 1 - 3 Tech Specs.

15. Section 6, Paragraph 4-With regard to units that risk should be expressed in, this section says to include the fraction of time that the plant is in a given plant operating state. While this is proper for some applications, if th application is measuring the difference between on-line or offline maintenance, ' ;onditional risk" is the proper value to use.

16. Appendix A, Section A.2.1.1, Paragraph 6, "The FV and RRW importance measures..."-This paragraph is unclear. What is the purpose of first addressing "all components within the scope of the regulatory requirement" and then addressing "the remaining components"? Why do this two step process?

17. Appendix A, Section A.2.1.1, Paragraph 9, "This equation shows the close relationship.."-This paragraph recommends use of an absolute measure, such as Birnbaum, to compare importance results for different sensitivities. Why is an absolute measure useful for such an application?

18. Appendix A,

Section A.2.1.2, Paragraph 9,

" Consideration of implicit Contributors"-This section effectively brings in the whole set of BOP components-most if not all of which are not controlled in the CLB-through contribution to initiators. The only way to do this effectively is to have a component-level model of the BOP, related to the initiators. While this may be pursued for plant efficiency reasons, it does not belong in regulatory based applications for PRA.

19. Appendix A,

Section A.2.1.2, Paragraph 11, " Consideration of implicit Dependencies"-Consideration of instrumentation or monitors not modeled-their impact on modeled actions. This appears to be beyond the scope of a typical PRA.

20. Appendix A, Section A.2.3, Paragraph 3, " Multiple Component Considerations" -

The method identified in this paragraph-to ensure that at least two " higher category" components are in all minimal cutsets by arbitrarily declaring a component " higher risk"-makes no sense -except in a risk-avoidance climate.

21. Appendix A, Section A.2.3, Paragraph 4, " Consideration for Defense-in-Depth"-

This method-reviewing each sequence cutset to assure that there are at least two component failures for which requirements have not been relaxed-appears to.be highly labor-intensive, and may have the effect of discouraging beneficial risk-based changes.

4

1 A

9

22. Appendix A, Section A 2.3,1 Paragraph 5, " Consideration for Allonable Plant Configurations"-This method _ does not address the function of on line risk management to protect against/high risk plant configurations. These are many configurations allowed by the Tech Specs that would not be permitted in the current risk management program. PRA must be a best estimate-not a bounding analysis of what could be, but a best estimate of what will be.

23. Appendix A, Section A.2.3, Paragraph 9," Component Specific Sensitivity Analyses"

-This sensitivity-using the upper bound failure rate for all low ranked components INDIVIDUALLY-makes little sense. Is there any basis for such an assumption? It is impractical to run individual cases for perhaps hundreds of components.

24. Appendix A, Section A.2.3, Paragraph 10, " Sensitivity Analyses for a Component Group"-This sensitivity-using the upper bound failure rate for a similar group of components - also makes little sense. Is there any basis for suspecting these components might not be performing acceptably? If not, then such a sensitivity tells us nothing of use.

25. Appendix A, Section A.2.3, Paragraph 15, " Sensitivity analyses can be used..."-This section recommends removing all credit for recovery Again, there is no basis for such a step. There is no justification for why recovery would not be available. This assumption would be similar to assuming that all valves don't work. The analysis can be done, but the usefulness of the results is questionable.

26. Appendix A, Section A.3, Paragraph 3, " Compliance with decision..."-This section brings in emergency planning and late containment failure for consideration. This raises the question ofjust how far the analysis should go. A reasonable bound must be placed upon the scope of the analysis.

27. Appendix B-This section begins with stating an " independent peer review...". Then Section B.2, Paragraph 2 states that the reviewer should also have a " good knowledge of the specific plant and its operation." That is a difficult position to fill-independent yet knowledgeable of the specific plant. The term " independent" needs to be carefully defined to exclude only someone who provided significant contribution to the analysis.

5