ML20044G954
"Draft Meeting" is not in the list (Request, Draft Request, Supplement, Acceptance Review, Meeting, Withholding Request, Withholding Request Acceptance, RAI, Draft RAI, Draft Response to RAI, ...) of allowed values for the "Project stage" property.
| ML20044G954 | |
| Person / Time | |
|---|---|
| Site: | Grand Gulf  |
| Issue date: | 05/21/1993 |
| From: | Mcgaha J ENTERGY OPERATIONS, INC. |
| To: | Gillespie F NRC - REGULATORY REVIEW GROUP |
| References | |
| CNRO-93-00022, CNRO-93-22, NUDOCS 9306070086 | |
| Download: ML20044G954 (7) | |
Text
P po c.
1
,s.c.
=== ENTERGY.
- 3=s Jackson.MS 39286-1995
'4 Tel 601984 9740 John R. McGaha Vce Prescent
. C4eratons LJppOrt 3D-MLk May 21,1993 Regulatory Review Group U.S. Nuclear Regulatory Cainmission Washington, D.C. 20555 Attention:
Mr. Frank P. Gillespie, OWFN 12 D21
Subject:
Comments on Draft NRC Report " Risk Technology Application" CNRO - 93/00022
Dear Mr. Gillespie:
Representatives of Entergy Operations, Inc. attanded the May 6,1993, public.
~
l meeting held by the Regulatory Review Group. We feel that this meeting was a valuable experience for both the industry and the NRC in developing realistic ~
approaches to risk-based regulations and programs. At the meeting, the draft'-
NRC report " Risk Technology Application" was distributed and ' comments were requested. ' We have reviewed this report and are providing the attached comments for your consideration.
1 We appreciate this opportunity to express our views on the draft report and your consideration of the our comments. Please contact Mr.. Kenneth Hughey (601--
984-9756) or Mr. Herbert Kook (601-984-9766) of my staff should you have any questions regarding this matter.
j Sincerely, k n. M A A-I JRM/hek 9
attachment D
cc:
(See Page 2)
O t
\\
r, ann -
9306070086 930521 e
PDR ADDCK 05000416
~
P PDR
?
E
r' -
Y i
Comments on Draft NRC Report " Risk Technology Application" May 21,1993 CNRO-93/00022 Page 2 of 2 cc:
Mr. T. W. Alexion Mr. J. L. Milhoan Mr. R. P. Barkhurst Mr. P. W. O'Connor Mr. R. H. Bernhard Mr. N. S. Reynolds Mr. R. B. Bevan, Jr.
Ms. L. J. Smith Mr. J. L. B!ount Mr. D. L. Wigginton Mr. S. D. Ebneter Mr. J. W. Yelverton Mr. E. J. Ford Central File (GGNS)
Mr. C. R. Hutchinson DCC (ANO)
Mr. H. W. Keiser Records Center (WF3)
Mr. R. B. McGehee Corporate File [8]
d i
se r
cm.- n n e -,
.n-.w,,--
,-.nr
[
V Comments on Draft NRC Report " Risk Technology Application" 6
May 21,1993 Attachment to CNRO-93/00022 Page 1 of 5 Entergy Operations, Inc. Comments on Draft NRC Report " RISK TECHNOLOGY APPLICATION" Section 5.4 Consideration of a graded approach to QA programs based on risk is a very positive development in allowing the industry to utilize resources commensurate with importance to safety. However, the benefits to be gained are a function of how the QA groups are defined. The proposed groupings throughout section 5.4 appear overly conservative and will limit-the benefits from adopting such an approach. Specifically, including any SSC found in "A PRA," that is, in any similar plant, in the most important QA group will unnecessarily reduce the benefits of a graded program and will conflict with PRA application to the Maintenance Rule.
Although the NSSS design of plants are similar to other plants of their class, there are significant design differences, particularly in support and balance-of-plant systems, such that the validity of applying the conclusions of one plant PRA to another is doubtful. For examples a support system at Grand Gulf niay exhibit a significantly different risk measure than the same system at another BWR/6; requiring Grand Gulf SSCs to be determined by the characteristics of another plant is inappropriate. An attempt to differentiate these SSCs is made in the document but results in the creation of another category of SSCs.- SSCs that are shown to be not important on the basis of a plant specific IPE do not warrant being included in a graded QA category higher than SSCs not important in any PRA.
in addition to the above concern, some plants such as Grand Gulf have utilized their IPE results to rank SSCs for the Maintenance Rule implementation. Adopting a different approach to a graded QA program would create inconsistencies between these two programs, while they should have the same basis.
If the goal of including risk results from other plants is to ensure that all.
necessary SSCs have been included with a minimum of review effort, the same result could be attainable through a licensee comparison of plant-specific risk rankings against " class-average" rankings and requiring justification for any discrepancies which might be of concern. Most plants have already compared their IPE results with other facilities of their class
\\
a
F y
Comments on Draft NRC Report " Risk Technology Application" t
May 21,1993 Attachment to CNRO-93/00022 Page 2 of 5 and understand the reasons for such differences. Therefore, little additional burden would be created. This appears to be a much more preferable approach than mandating unnecessary conservatisms which would be in effect for the remaining life of a plant.
t Section 5.4.1 4th paragraph-The need to be able to normalize the importance measure is understandable but this appears to put plants with better core damage frequency numbers at a disadvantage to those with poorer results. For example:
Plant #1 has baseline CDF of 1E-5 with an SSC that has an Achievement importance Measure of 4E-4. The unavailability impact for this SSC is a factor of 40 which would be considered relatively important for Plant #1.
Plant #2 has baseline CDF of 1E-6 with an SSC that has an Achievement importance Measure of 4E-5. The unavailability impact for this SSC is again a factor of 40 which would also be considered relatively important per the definition.
However, the change in CDF associated with Plant #1's SSC (Achievement importance Measure - CDF) is 3.9E-4, while the change in CDF associated with Plant #2's SSC is 3.9E-5. There is an order of.
magnitude difference in the two but the defined Achievement importance Measure impact ratio makes them appear equal. There should be another measure to capture importance without penalizing those plants that have a relatively low CDF. Possibly, the above could be offered as an example with some desirable attributes, such as providing a method to ensure that existing levels of safety are maintained.
Section 5.4.3 As discussed above under Section 5.4, the inclusion of all SSCs found to be important in "A PRA" is not necessary if justified. Also, as now written, the description of Group 1 SSCs appears to envelop all Group 2 SSCs, which appears confusing. A possible alternative phrasing (which also addresses the preceding comment) might be:
" Group 1 - Those SSCs meeting one of the following criteria:
-Found to be important in the plant-specific PRA
-Found to be important through deterministic methods.
f
c-t Comments on Draft NRC Report " Risk Technology Application" t
May 21,1993 Attachment to CNRO-93/00022 Page 3 of 5 Group 2 - Those SSCs meeting one of the following criteria:
-Found to be important in a PRA of a plant of similar design r
and which has not been shown to be not important by plant-specific means
-Found to be important to plant or system availability through deterministic or other means."
For Group 1 SSCs, generally the same standards as now are used for SSCs requiring " full QA" would be used for design, procuremelt, receipt, storage, installation, testing, maintenance / surveillance, etc. For Group 2, only limited activities such as design control and initial testing should be imposed. Other programs and requirements (i.e., the Maintenance Rule) would ensure that these SSCs would continue to operate to the level identified as appropriate by the IPE. For Group 3, these would be subject to the same minimal requirements as now applied to "nonsafety-related" or nonnuclear equipment.
Section 5.4.4 Human reliability analysis (HRA) offers valuable insights to a PRA study.
By using HRA analysis, past PRAs have identified a number of problems in procedures and training that had not been found using " traditional regulatory" approaches. HRA techniques can help to measure many effects which influence the reliability of an action, thereby establishing the relative importance of human failure events. This assists in prioritizing training activities and showing the effectiveness of verification activities, allowing the most effective use of resources. HRA techniques offer the best, if not only, means of evaluating the significance of training, stress, F
hesitancy, conflicting priorities, etc.
If an Human Error Probability (HEP) of 1.0 is used, the effectiveness of using safety prioritizations determined by the IPE is weakened. Most application work and sensitivity work done (by utilities) with PRA is done by manipulating the cut set results. Therefore, many cut sets are lost (i.e.,
truncated) once an HEP has been added to the cutset. Evaluations done by the manipulation of dominant cut sets usual!y can be performed fairly quickly. However, requantification of an entire PRA is much more time consuming.
[
l i
i 1
F-
.s Comments on Draft NRC Report " Risk Technology Application" s
May 21,1993 Attachment to CNRO-93/00022 Page 4 of 5 1
If the PRA should be requantified with all HEPs (both pre-accident and post-accident) set to 1.0, this may well cause important cut sets to be lost because of limitations of the computers used in quantifying the accident sequences.
That is, important hardware cut sets may be truncated because of the large number of cut sets generated because of the overly conservative HEP l
of 1.0. Rather than setting all HEPs to 1.0, a reasonable screening value should be used. Different screening values should be used for pre-accident and post-accident HEPs. The pre-accident screening value could be relatively low as these actions are usually controlled by procedure and well understood. Post-accident screening values of approximately 0.5 could be used since these are usually cut set dependent and time considerations come into play. Values or ranges of values should be specified by the NRC or the industry for acceptance without specific t
additional plant-specific justification.
Screening values are often used in PRA studies to solve the plant modelin order to protect against dependencies, high stress, and other influences which could invalidate a human failure event probability. An adequate consensus should exist resulting from this experience to develop industry screening values. The NRC Accident Sequence Precursor study method j
values for human failure events might also be used; these are regularly used by the NRC staff in regulatory evaluations. Another option would be the development of conservative probabilities developed for actions required within various time intervals. Such an approach would allow the distinction between actions required to be performed in ten minutes from those required to be performed in several hours.
Section 5.5.1 Page 33, third line from the bottom of the page, "successfuylly."
Figure 5.5-1, label the horizontal axis.
Section 5.5.2 Page 35, first full paragraph, fourth line, second "be" should be "by."
t a
I s'
Comments on Draft NRC Report " Risk Technology Application" May 21,1993 Attachment to CNRO-93/00022 Page 5 of 5 Although the concept of using CDF to increase STis for relatively non-important SSCs appears sound, the assumption of unavailability as directly proportional to time x failure rate may be flawed. Some component failure rates, such as those for diesel generators and normally nonenergized solenoid valves, may be more valid when computed per number of demands rather than per time period. Also, the equation used does not address unavailability due to out-of-service tirne for maintenance and surveillance activities, which might be greater in some cases than that due to failure rate x STI. For example, the equation seems to imply that reducing an STI by one-half would reduce unavailability one-half; there does not seem to be any consideration of a " break-even" point for continuing to reducing the STis of important SSCs.
Section 5.5.3 See previous comments on Section 5.4.4.
F