ML20045B272
| ML20045B272 | |
| Person / Time | |
|---|---|
| Issue date: | 11/23/1990 |
| From: | Carr K NRC COMMISSION (OCM) |
| To: | Chilk S NRC OFFICE OF THE SECRETARY (SECY) |
| Shared Package | |
| ML20042D089 | List:
|
| References | |
| FRN-57FR14514, REF-GTECI-B-56, REF-GTECI-EL, RULE-PR-50, TASK-B-56, TASK-OR AE06-1-064, AE6-1-64, VALB-901123, NUDOCS 9306170115 | |
| Download: ML20045B272 (2) | |
Text
AEo& -f Qpg h
NOTATION V0TE G j,gg
@ Mu-RESPONSE SHEET
@ 444' dC 7&c 6 L44+4r T0:
SAMUEL J. CHILK, SECRETARY OF THE COf+1ISSION FROM:
CHAIRMAN CARR
SUBJECT:
SECY-90-340 - RESOLUTION OF GENERIC SAFETY ISSUE B-56, " DIESEL GENERATOR RELIABILITY" l
l ABSTAIN APPROVED w/ comments DISAPPROVED REQUEST DISCUSSION NoT PARTICIPATING C0ffiENTS:
I approve the staff's proposed actions, subject to the change indicated on the attached draft and to the addition of the following footnote to Tables A1, A2, A3, and A4:
"These practices are not intended to replace current EDG i
reliability maintenance activities that have proven to be reliable; rather, they reflect current proven practices 1
\\
SIGNATURE RELEASE VOTE
/
/
kI N
- DATE WITHHOLD VOTE
/
/
f ENTERED ON "AS" YES No 115 930422 g61 gg {
g.mm ~.
c y
APPENDIX A (M
RELIABILITY PROGRAM CONSIDERATIONS Inc discussions tha follow are considerations Emergency Diesel Generator Performance and'inustrative examples (MTables A1, A2, and Monitoring A3Emitare not intended to replace current emer-performance monitoring 2 applied to equipment gency diesel generator reliability maintenance activi-that is run on a continual or a near-continual basis.
ties that have proven f ective. Rather, these com-De purpose is to monitor certain parameters on un if mems refIcct current proven practices that are ongoing basis in order to obtain information about discussed in more detail in NUMARC's topical report the state of physical conditions that may impact the i
which is referenced ~ > Appendix D (5-2-90) to operability of a piece of equipment and that coald be NUMARC-8700.
gh6 used for trending purposes. Rese trends may signal a degradation in a particular condition. Such evalu-l Emerguicy Diesel Generator Surveillance Plan ation may detect onset of failure and allow corrective i
actions :o be taken before failure occu-s.
The formulation of a surveinance plan considers the following factors:
Equipment that is normaUy in a standby condi-tion, such as an emergency diesel generator, can only 1.
De effect that emergency diesel generator be monitored on a limited basis. Monitoring crideal support and auxiliary systems have on over-operating parameters is usuaUy performed during all emergency diesel generator reliability.
monthly operational testing. In order for this moni-2.
Failures caused by surveillance, toring to be effective, the foDowing conditions should apply.
3.
Frequency and nature of surveillance testing b
- *
- I* C' C or Parameter should be a effects on emergency diesel generator reli-measurable condikon that is known to be re-abiEry and unavailability.
lated to an important failure mode.
4-The types of faDures that can be detected by 2.
He characteristic or parameter should be a surveillance program.
able to be measured conveniently and practi-Detection of faDures by parameter monitor.
caUy.
ing versus testing.
3.
De characteristic or parameter should be l
6.
The ability of specialized tests to simulate accmately mosored actual operating conditions.
4.
Parameters recorded should be measured un-der the same conditions (i.e., load) to the Figure A-1 iUustrates typical corr.ponents and extent possible.
support systems that should be considered when de-finmg an emergency diesel generator boundary.
De acmal vabes c,f the conditions should be re-I Dose components whose function is solely to support corded rather than, simply verifying that they are the emerEency diesel generator should be viewed as within a specIc range. A comparison between the l
wnhin the emergency diesel generator boundary. He values obtained from successive readings can then be systems that provide support to the emergency diesel made to awertain the possibility of a degrading con-gemenmr and perform other plant functions are diu,on.
shown outside the boundary, with the understanding Emergency Diesel Generator Maintenance that the boundary interface function must be main' Program tamed.
l An important contributor to emergency diesel IEEE Std 387-1984 and ANSI /ASME OM-16 generator reliability is the manner in which both pre-provide emihr definitions of components and system ventive and corrective maintenance are performed, h=M=+< and may also be used.
Generally speaking. an emergency diesel genera-Tables A1, A2, and A3 illustrate periodic surveD.
tor maintenance program is based on the foUowing:
Iance actrvuies that have proven effective. When per-1.
Vendor-recommended maintenance actions
~
formmg such surveillance, it is important to capture and schedule for implementation.
l the actual values of critical parameters since such 2.
PrkrMes for mairitenance actions are based l
would be extremely useful for faDure analyses, II as for long-term emergency diesel generator n such factors as repair time, severity, he-c tition monitoring. It should also be noted that lihood of reemtence.
arte-specific activities may vary from those shown in 3.
The reliability characteristics of the emer-these_ tahlec.
gency diesel generator subsystems and 1.9-14
// 2 o 6 - l W
i N0TATION V0TE gC f/
g e WTG~
RESPONSE SHEET
- 6) 7 Mar
@ l%>try T0:
SAMUEL J. CHILK, SECRETARY OF THE C0fEISSION FROM:
COMMISSIONER R0GERS
SUBJECT:
SECY-90-340 - RESOLUTION OF GENERIC SAFETY ISSUE B-56, " DIESEL GENERATOR RELIABILITY" APPROVED cl 4.
DISAPPROVED ABSTAIN as io.u reo.
NOT PARTICIPATING REQUEST DISCUSSION l
COMMENTS:
sce rrre ts s e-rs.
sw.
i l
1 l
1
@4 bA 1
SIGNATURE O~
RELEASE VOTE
/
/
o 2A, /?2D-DATE' I
t>
WITHHOLD VOTE ~/
/
ENT.ERED ON "AS" YES NO _
7fL 4
N4fp.
3-l.-
i
/
i
.. i.
- (LR) jfy,,,y g (SR) 3)
EDG Reliability
=
I l
The above equations produce point estimates f individual EDG reliabilities with attendant uncertainties.
Care should be taken in using such numbers in comparing plant performance with the EDG trigger values, particularly when using the last 20 demands data set.
Estimates of EDG reliability for a nuclear unit should utilize individual EDG performance data, which are then combined in a manner representative of the EDGs assigned to a specific.
nuclear unit.
NUMARC-8700, Revision 1, Appendix D, Table D.2-1, provides guidance for combining data _from individual EDG performance to arrive at a nuclear unit reliability estimate.
3.3 Maintaining EDG Reliability:
Maintaining EDG reliability should include the following:
(1) maintaining data on successful and failed EDG start and load-run demands.
evaluating nuclear unit reliability indicators-(2) for thg ly t J and the last 100 demands,as well as, individual EDG performance over the last 20 demands.
(3) relating calculated EDG performance and i
reliability indicators to trigger values established for selected target reliabilities.
(4) taking remedial actions for individual failures and for exceeding one or more trigger values.
I The sample size and action levels are based on the assumption that the minimum surveillance testing interval for each EDG is i
once per month.
The f;11cwing feilmue tote Liigg;re ohnnld be-u:Od tv a s o e s s" hw unhuu.
EDf puufvuwauvo and Lv duLuAwiso vv&&vvLive acLivue wv
/& 8 9
18 e*
J 9
---t--q-w e wtr--
y
-4 we tu y v
-w-g*"*7 T'*Mvo'-'M'M' = = esr
-w'r9' 7^we'"9k N'
D,-N'h'a 7
've-'*'t B"
- a'--'*r-
?
-P-W W e-F*reer
-==+47e P-""v'
r EDG TRIGGER VALUES Selected Target Failures in Failures in Failures in Reliability 20 Demands 50 Demands 100 Demands 0.95 3
5 8
0.975 3
4 5
The sele,cted t~af t~ reliab]1TE 71~s tha[t' eWcEed fo tihe stptionl
~
~
~
~
I ng th[ plying
]
nde k4d facko2ft coping alysis Th' value represents copino lit needed for dete nucip'ar unit,EDG relia) blackout, igure 1 defines actions that C-duration for a station should be undertaken when one' more of the triggers shown above are exceeded.
A more detailed discussion of actions related to exceeding one or more of these triggers can be found in Section D.2.4 of NUMARC's Appendix D.
3.4 Problem EDG A problem diesel generator is defined as an individual EDG_
Amu that has experienced 4 or more failures in the last. 25 demands.
Should this case arise, the actions taken in response to exceeding a single trigger value as defined in Figure 1 would apply.
Following completion of reliability program corrective actions, restored performance of the problem EDG should be demonstrated by conducting seven consecutive failure-free start and load-run fqshne 6
demand tests per Regulatory Position 2.3.3.
The monthly
- ?? d surveillance test schedule should not be resumed until the seven TFdd consecutive tests are successfully completed.
All starts and load-runs performed during this period should be included in the nuclear unit EDG reliability data set so long as the EDG is declared operable.
This process of evaluating recent demands and taking appropriate action on the individual EDG experiencing recurring failures is a key element in providing reasonable assurance that EDG performance is restored to an acceptable level.
19
'~/
Insert A The calculations discussed above will be point estimates of the reliability and will have inherent uncertainties because of the sample size available.
A point estimate reliability calculation for a 50-demand sample that falls below 92 percent, or for a 100-demand sample that falls below 93 percent, is an indication that the true underlying reliability may have fallen below 95 percent.
A point estimate reliability calculation for a 50-demand sample that falls below 94 percent, or for a 100-demand sample that falls below 96 percent, is an indication that the true underlying reliability may have fallen below 97.5 percent.
l Insert B For purposes of monitoring nuclear unit EDG reliability levels against those selected for the Station Blackout Rule, and identifying corrective actions for maintaining target levels, the following combination of failure rates (or triggers) has been developed to provide statistical approximation of the selected H
target reliability level:
Insert C It is recognized that large statistical uncertLinaies accompany point estimates made from a 20-demand sample size.
However, the most recent tests provide an insight into the apparent operability of the EDGs.
i i
\\
)
I l- - - - - - - -
- ~I N 0 T A T I'0 N V0TE f'
RESPONSE SHEET T0:
SAMUEL J. CHILK, SECRETARY OF THE COtEISSION FROM:
COMMISSIONER REMICK
SUBJECT:
SECY-90-340 - RESOLUTION OF GENERIC SAFETY ISSUE B-56, " DIESEL GENERATOR. RELIABILITY" w/cormh APPRovEo X DISAPPROVED ABSTAIN,
NOT PARTICIPATING REQUEST DISCUSSION COMMENTS:
hh"J feg effeCYf Y COMW ko I approve staff's issuing the generic letter on GI B-56, as modified in the ED0's memo to me dated November 8, 1990, with some minor additions (attached). Since this generic communication expects licensees to review Revision 3 to Reg. Guide 1.9, that guidance should be an enclosure to that letter.
[//// SIGNATURE RELEASE VOTE
/X/
/9 Ady Md i
DATE WImHOLD VOTE
/
/
ENTERED ON "AS" YES /
NO ___
/OS CN
l w;-
~
?
v.,
i i
2 I
i able operation of onsite emergency Ac power sources should be ed by a reliability program designed to amintain and monitor i
f ane reliability of each power source over time for assurance that selected reliability levels are being achieved.
It was noted in i
the Federal Register Notice (53 FRN 23218) that: "The resolution i
of B-56 will provide specific guidance for use by the staff or 1
industry to review the adequacy of diesel generator reliability programsconsistentwiththeresolutionofUSIA-44."f og,,M RG 1.9, Rev. 3 provides such specific guidance.
INFORMATION REOUIRED TO BE SUBM m zu BY ADDREBBEEBr sasue esong h anee sie+4 to c FR., 30. G 2 ** ys reon e vCb s A b. 50 In order to(determine whether any operating license or j
i construction permit for facilities covered by this request should j
j be modified, suspended, or revoked, you are required, pursuant to Section 182 of the Atomic Energy Act and 10 CFR 50.54(f), to i
mTibmit informatien to the NRc under oath or affirmation, within r
{
180 days of the date of receipt of this letter, regarding your current plans for assuring adequate diesel generator reliability.
j For your guidance the staff has identified an approach for resolving GSI B-56 which, if implemented, would provide an 3
j acceptable method for monitoring and amintaining EDG reliability levels as required for compliance with 10 CFR 50.63.
That j
approach consists of implementation of NUMa e a Initiative 5A and
{
Regulatory Positions C.3, C.4, C.5 and C.6 of Regulatory Guide j
1.9, Revision 3.
The staff recognizes that equally effective programs for monitoring and maint=4ning EDG reliability may have already been implemented.by licensees.
In such instances, if a 1
j program has successfully demonstrated its effectiveness in i
achieving and maint=4ning the target levels of diesel generator reliability, such a program, would be considered by the staff to j
be an acceptable equally effective program.
In such instances, i
no changes to an existing approach would be. considered necessary.
1 j
In your required response to this letter, addressoas are i
requested to indicate their commitment to implement either the j
suggested acceptable program consisting of NUMARC's Initiative SA j
and Regulatory Positions C.3, C.4, C.5 and C.6 of Revision 3 to i
Regulatory Guide 1.9 (or equivalent h N guidance as indicated
)
in Table 1 of Regulatory Guide 1.9, Revision 3), or an equally j
effective program for monitoring and maintaining EDG reliability l
j levels for compliance with 10 CFR 50.63.
If you plan to l
implement an approach other than the suggested acceptable approach, your response should describa your program, including j
justification that it represents an equally effective approach.
Your response should also provide a commitment to imple:2ent your j
program at the earliest possible date, but in no case later than 270 days from the date of receipt of this letter.
i You should retain all documentation supporting this response consistent with the records retention program for your facility.
J i
i
\\
Afo(a.-l Taylor
~
Jniezek
,fga a' %*'o UNITED STATES Thompson NUCLEAR REGULATORY COMMISSION Blaha
{,
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS Beckjord 0,
g W ASHINGTON. D. C. 20555 g p]gy
/
Bernero an November 23, 1990 l
The Honorable Kenneth M.
Carr Chairman U.S. Nuclear Regulatory Commission Washington, D.C.
20555
Dear Chairman Carr:
This is being written as an individual letter, since it deals with a subject - statistics - that holds more interest for me than it i
does for the other members of ACRS, or indeed for the NRC staff.
I see no good reason why,the NRC shouldn't do its statistics-correctly.
~
Since my " additional comment" on the ACRS diesel generator letter of August 14, 1990 has been treated with benign neglect by the staff, and has had no visible effect on the draft Regulatory Guide 1.9, Rev.
3, it may be useful to spell out the statistical point a bit more expansively.
The matter is not trivial, since the purpose of the entire exercise is to establish a level of reliability for a licensee's diesel generators (either 0.95 or 0.975), and to provide a mechanism through which the licensee can demonstrate maintenance of that level.
In our meeting of November 8, 1990, Commissioner'Curtiss emphasized this point. The trigger value method given in the draft Regulatory Guide provides no such demonstration.
Any statistician could have made this point to the staff, and I am embarrassed to have to do so.
l I will use only one of the cases cited in the draft Regulatory Guide, the trigger value of three failures in twenty attempts, for a presumed reliability level of 0.95.
Presumably, failure to trigger is to be construed as the necessary demonstration that the desired reliability level of 0.95 is being. maintained (otherwise, why does the failure trigger any action?).
I will give these-numbers a standard elementary statistical analysis.
First notice that the statement "three or more failures in twenty l
attempts" is ambiguous. There is a considerable difference between l
the case in which one studies twenty attempts to see how' many failures have occurred and that in which one waits to count three failures, and counts how many attempts it took to get there.
They have different probabilities.
In this case, since the event is to be used as a trigger for action, I will assume that the latter is L
what was meant.
)
bL0 SPP-p
l l
f t
i The Honorable Kenneth M.
Carr 2
November 23, 1990 In that case, the precise statement of the situation is that at each failure the licensee should check how many have occurred in the previous nineteen attempts.
If the number is zero or one, that will be construed as demonstration that the ra'4 Tbility objective rwo or more the I
of 0.95 is being maintained.
If there a)
- , discussion about i
demonstration has failed.
(We could have the case.
In addition, l
any of the other thresholds - this is just onc l
the term "two or more" is redundant - if theit: are more, then the l
trigger has already been pressed.)
The fact that there ha's been zero or one failure to start in the previous nineteen attempts provides information that can be subjected to a standard and classical statistical
- analysis, assuming that the failures are independent.
Such an analysis, which is trivial and can be performed by any statistician, leads i
j to the conclusion that the 90% confidence bounds on the underlying diesel reliability are 0.774 and 0.981.
Only such information can be drawn from those data.
That falls rather short of the desired demonstration that the reliability is greater than 0.95.
(If none of the three triggers - 3/20, 5/50, 8/100 - are pressed, the 90%
i l
confidence bounds on the underlying reliability become 0.892 and 0.98?,_again hardly a demonstration that the reliability has been maintained above 0.95, as required.)
l I have no problem with an aspirational statement about diesel reliability, and none with arbitrary trigger levels for attention l
to the diesels.
But these are not connected in the way stated in the draft Regulatory Guide, and it is simply error to say that they are.
Finally, I would think it useful for the NRC staff to include a few l
statisticians, and for the engineers to take statistics seriously.
l There is no sin in doing mathematics correctly.
In fact, it would be easy to construct a monitoring and evaluation program that provides a running estimate of each diesel generator's underlying reliability, and of its uncertainty, and provides a means of l
detecting trends.
Any statistician could devise such a program, l
and it could provide a sound basis for regulatory attention.
Additional comments by ACRS Member J.
Ernest Wilkins, Jr.,
are presented below.
Sincerely, j
kW M & Q48[T Harold W.
Lewis, Member Advisory Committee on Reactor Safeguards m
s e-
y The Honorable Kenneth M.
Carr 3
November 23, 1990 Additional Comments by ACRS Member J.
Ernest Wilkins, Jr.
I concur with Professor Lewis' concerns in this specific case, as well as with his general observations about doing " statistics correctly."
I believe, however, that the NRC staff already includes at least one competent statistician.
The problem, therefore, may be that neither his professional services nor those of any other such individual are utilized for the purposes described in this letter.
O i
l l
l u
i
. - _,