ML20206Q459
| ML20206Q459 | |
| Person / Time | |
|---|---|
| Issue date: | 01/08/1999 |
| From: | Bennett Brady NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | Mays S NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| References | |
| NUDOCS 9901140213 | |
| Download: ML20206Q459 (18) | |
Text
..
_.m. _ _. -.. _ _ _. _ _. _.. ~ _..._ _.
f; y
9 Mauq g
UNITED STATES
- 5
'j NUCLEAR REGULATORY COMMISSION f
k.....[2 WASHINGTON, D.C. 20066-0001 January 8, 1999 MEMORANDUM TO: Steven E. Mays, Sectiori Chief Reactor Risk Assessment Section Reliability and Risk Assessment Branch Safety Programs Division Office for Analysis and Evaluation of Operational Data FROM:
Bennett Brady Reactor Risk Assessm nt Section Reliability and Risk Assessment Branch Safety Programs Division
' Office for Analysis and Evaluation of Operational Data
SUBJECT:
MEETING OF EPIX AD HOC WORKING GROUP, DECEMBER 8, 1998 i
On Tuesday, December 8,1998, I attended the second meeting of the EPIX Ad Hoc Working Group at INPO. As part of the agenda for the meeting (Attachment A), I gave a presentation to the Working Group on the proposed NRC Reliability and Availability Data System (RADS). A copy of the slides that I presented is attached (Attachment B).
' Submittal of Voluntary Data in the June 13,- 1997 Staff Requirements Memo (SRM) the Commission directed the staff to accept the voluntary approach to providing reliability data and to continue to work with industry representatives to improve the content of the voluntary data. Items discussed at the meeting related to the Commission's directions are noted below:
Provision of voluntary data:
o Joe Bishop opened the meeting by presenting the, July 25,1997-letter from Ed
/jC Jordan to Ralph Beedle, NEl, with the attached SECY 101,' recommending G
the voluntary altemative, and the June 13,1997 SRM. He explained the history J
of the reliability data effort and why they are being collected for EPIX.
o The reliability data, i.e., the demand and run-hour data for the key components in the maintenance rule risk significant systems, was originally due to INPO on July 1,1998. At the time of the meeting,44 of the 60 utilities have submitted some reliebility data. Of these 44, about 30 are complete.
o:
The delay in submitting Me data is primarily because much of the support for y
EPIX activities in the utilities is in their maintenance rule organizations. The utilities have been unable to put their resources on EPIX st this time due to
' maintenance rule schvities such as maintenance rule self-assessments, +O O & flb 0r 99011'40213 990'100 9F-ce9 l
PDR u
D, maintenance rule inspections, and maintenance rule inspection deficiencies.
1 I
The late submittals are also due to utilities in outages having to put their resources on outage activities and to some tumover in EPIX coordinators.
According to the EPIX status report, "the orincipal challenge has been motivating o
the plants to provide the periodic demand and run-hour data." In September, the President of INPO sent a letter to the Executive Points of Contact addressing the late submittals of these data.
The utilities estimated that the resources required for providing the one-time o
estimates of demands and run hours are from six to 12 man-weeks. The utility representatives present at the meeting estimated that their plants had required from two to six man-weeks.
o One of the members of the group reminded the meeting that this cost was small in comparison to the savings possible from regulatory reductions through risk-informed regulation. I reiterated this point m my presentation.
]
Joe Bishop stated that a decision on any reduction in the reporting of reliability o
data could not be reached by a consensus of this group but would need to go to NEl.
Following a presentation on the SSPl, some of the members of the group o
questioned why they were reporting demands and run hours to both EPIX and 1
SSPl if INPO and NRC were not using the SSPI reliability indicator data. I reminded the group that this was part of the voluntary agreement and that NRC may be using the reliability indicator information in the near future.
o Due to the late submittals and the need to QA and compare the data with that of sister plants for consistency, INPO will not provide the EPIX data to NRC until March 1999.
To provide high assurance that the data will be ready in March, INPO has asked o
the utilities that could not provide their reliability data by December 15,1998 to provide a plan for completing the data submissions.
The March date for receipt of the date will be the latest date at whicn INEEL can receive the data and meet their schedule for production of the system. This will also mean that we will not be able to test whether 1) the workaround components will be identifiable from the EPIX database,2) failure modes and conditional failures in EPIX can be correctly interpreted and 3)
EPIX failures can be related to their workaround demands.
Actions beino considered to improve the content of the data that are of interest to NRC:
o Add planned unavailability for either SSPI or all risk-significant equipment. It was noted that a potential barrier to this is the need for a co'nmon definition between Maintenance Rule, SSPI, and EPIX.
o
' Add demand / operating hour data for additional types of maintenance rule risk-significant key components (e.g., heat exchangers, electric heaters, and transformers).
o Change the fourth failure reporting criteria from reporting technical specification declared inoperability of any maintenance rule risk-significant component to read "any demand failure of any risk-significant key component." The new criteria l _
would insure that failures of key components that do not result in a system l
l functional failure because of redundancy (e.g., six parallel injection valves where
[
only two are required for the safety injection function) are reportable.
L o
Create a mapping of the present EPIX failure modes into PRA failure modes.
They asked NRC's assistance (n providing PRA failure modes of interest.
-.._y
.- ~ -.
_,- - -- e
~.. -. - -
L-L o
Possible changes to improve the quaiXy of the data, l
Run comparisons studies of reliability data scoping and completeness between sister plante Develop more strict enforcement of the coach QA requirements for data reported Develop metrics and output reports for quantifying and trending data quality Since the last meeting of the EPIX Ad Hoc Group, INPO has done considerable work in developing training material for the EPIX program and conducting additional training classes for coordinators. These activities should further improve the overall quality of the data.
l Follow-uo to meetina
. During the meeting, the Group established two subcommittees, one to deal with issues in the short term and the other to advise on long-range issues. The topics and time frames of the two subcommittees are shown below:
l-Subcommittee Agenda Time Frame Short-term
- 1. Definitions - unavailability, review of standard Around January l
Subcommittee definitions 19 -21,1999
- 2. Calculations - proposed calculation of reliability parameters
- 3. Engineering characteristics - minimum needed i
1 j
- 4. Component engineering analyzes l
l Long-term.
- 1. Goals and objectives review (mission and To be l
Subcommittee vision) coordinated L
- 2. Functionalrequirements with NEl L
- 3. Scoping review l
- 4. Data review L
The next meetirig of the Ad Hoc Group will be in the May 1999 time frame. As follow up to the meeting, I will provide INPO with a list of the Pi Peer Groups for comparison of data submitted to EPIX for consistency and a list of the failure modes for mapping EPIX failure modes into commonly used failure modes.
i L
l-i l-Attachments: As stated cc satts.:
l.
Joe Bishop, INPO
' Glenn Masters, INPO
+
i l
... ~.
I j
Memorandum dated ' 01/ 08/99
SUBJECT:
MEETING OF EFIX AD HOC WORKING GROUP, DECEMBER 8, 1998 Distribution w/atts.:
PBaranowsky
- CRossi JRosenthat DAllison
- MFederline AThadani NRC Reliability and Availability Data System Coordination Group RRAB RF
- Public TKing, RES i
- H
- \\BMB\\ EPIX-D98.WPD To receive a copy of this document, Indicate in the box "C" copt w/o attachlencl "E" copf wlattachtenci "N" No copy OFFICE RRAS/RRAB c' RRAS/RRAB C
NAME BBrady ph SMays p DATE j / 7 /99 f / 5' l99
' OFFICIAL RECORD COPY.
-r
NRC RELIABILITY AND AVAILABILITY DATA SYSTEM L
COORDINATION GROUP t-1 Name Og Telenhone Room Mail l
Pat Barnowsky AEOD 415-7493 T4 T4 A9 1
l Steve Mays AEOD 415-7496 T4 T4 A9 j
i l
Dale Rasmuson AEOD 415-7571 T4 A9
'Bennett Brady AEOD 415-6363 T4 AS T4 A9 Steve Long NRR 415-1077 08 H2O 010 E 4 Frank Talbot NRR 415-3146 09 F 19 09 A 1 Gareth Parry NRR 415-1464 09 B 9 08 E2 Ed Rodrick RES 415-5871 T10 E 59 T10 E 50
. Mary Dreuin RES 415-6675 (61897)
T10 F 7 T10 F 13 Harold Vandennolen RES 415-6236 T10 E 65 T10 E 50 Bill Carrier OCIO 415-5778 T6 E 51 T6 C30
~
cC Don Marksberry AEOD 415-6378 T4 A17 T4 A39 Kenneth Russell INEEL 208-526-9592 i
9 L
l:
9
- 7..
I i
a v,=
.--,n a
Attachment A
_ Preliminary Agenda EPIX Ad Hoc Group l
December 8,1998 l
Room 202, Institute of Nuclear Power Operations l
Agenda Topics 8:00 a.m.
Welcome and '
Gary Fader introductions r-8:30 a.m.
EPIX Action Item Status
.ioe Bishop Items from 6/11/98 e
meeting 1999 Product / Priorities 9:15 a.m.
Break 9;30 a.m.
Implementation Issues Tim Scoggins/All 10:30 a.m.
Feedback on Usage Glen Masters
'll:15 a.m.
PSA Program Issues Jim Maddox/ Ken Bych 11:45 a.m.
Lunch Provided by INPO l
l 12:45 p.m.
PSA Program Issues Jim Maddox/ Ken Bych (cont.)
1:30 p.m.
NRC Reliability Data Bennet Brady l
Activities l
2:00 p.m.
WANO SSPI Activities At: Szczepaniec l
l 2:30 p.m.
Break 2:4$ p.m.
Aligning EPIX Glen Master Definitions / Rules with other Programs 3:30 p.m.
Summary and Adjoum Jim Maddox Joe Bishop
3 i
i m
i g
i 1
i 5
NRC RELIABILITY AND AVAILABILITY DATA SYSTEM (RADS) i f **%,s 4
A, D
o
- +,,,
BENNETT M. BRADY (301-415-6363, bmb1@nrc. gov)
RELIABILITY AND RISK ASSESSMENT BRANCH OFFICE FOR ANALYSIS AND EVALUATION OF OPERATIONAL DATA U.S. NUCLEAR REGULATORY COMMISSION EPIX AD HOC GROUP MEETING DECEMBER 8,1998
i i
i i
4 RADS - INTRODUCTION i
i i
l e
- Purpose of RADS i
e Uses of reliability data e
Output 6
l e
Analysis Methods i
e Status and schedule 4
l I
4 i
i i
2 1
[
4
i i.
I 3
j PURPOSE i
l e
Compute estimates of reliability parameters for use in risk-informed applications l
i e
Uses of reliability and availability data j
Risk-based indicators of plant performance Performance-based, risk-informed inspections l
Monitoring maintenance rule implementation l
Reliability studies of selected risk-significant systems and components l
Improving accident sequence precursor analyses Available to NRC and licensees for risk-informed regulation (plant-l specific licensing actions) 1 i
s 3
l I
4 4
NRC RELIABILITY AND AVAILABILITY DATA SYSTEM l
I LERs Search -
m Module -
A V
O Analysis l
4 Input Data and m
m Module Module Estimation ca Module i
-$(
SSPI i
A V
Reports m
Other Module M
Data 4
n-,
, - -, -. r
~
m,
.,.,,w..
-nn
W i
l l
re C
U T
U P
F S
P r
a t
r l
q a
p e
n a
a o
i u
n p
n p
n u
n b
l d
e d
a n
r d
a i
l r
i r
e d
n e
e b
b i
a b
n n
d i
r y
w" l
t o
e u
i a
t e
u p
d t
f y
c n
a n
e a
o i
r u
a o
r l
k o
d a
n v
d u
f a
m o
m a
a u
r r
i r
e fa o
m n
e v
l i
a n
r i
t a
l u
o c
e b
g a
u i
n n
o r
a l
i t
r l
o e
e d
n s
i b
ty p
c c
i e
o a
u l
i f
t r
n o
u r
r s
r y
a e
e t
d i
e R
s n
o a
n m
A e
f t
le u
a D
i l
c u
n n
S t
r a
d e
e v
O d
e a
U i
c v
l a
T o
e b
P m
n i
t U
l i
p s
t o
y T
n S
e n
ts 5
l I
i I!
e
\\
RADS ANALYSIS METHODS 4
i Classical analysis (point estimate and confidence interval)
Bayesian analysis (distribution, mean, interval)
Empirical Bayes
]
Tests of homogeneity j
Weighted least squares trends f
ensi k
6
_y.
y y
d 4
i i -
STATUS AND SCHEDULE e
May 1998 - NRC Reliability and Availability Data System Requirements Specification completed e
September 1998 - Preliminary System Design Review Meeting e
January 1999 - Demonstration of Prototype Model i
e August 1999 - Beta test system at NRC e
November 1999 - Implementation of system at NRC 7'
0 0
b e
9
>0 7
C T
CD r-Om M
I i
I l
1
)
i CD
I BACKGROUND i
j e
October 1994 - NRC initiated rulemaking to require reporting of reliability and i
i availability information for risk-significant systems l
i e
October 1996 - INPO provided a sample of SSPI data for NRC evaluation i
i l
e January 1997 - Meeting with industry on SSPI evaluation Unable to make meaningful estimates of demand failure probability or operating failure rate l
Could estimate unavailability i
e March 1997 - Industry modified EPIX to address issues 3
e May 1997 - Staff recommended that the Commission accept the voluntary approach l
Advantages (less cost, receive data earlier, linked to maintenance rule, more components) outweigh disadvantages (some data not available, must
]
develop ~workaround" that introduce additional uncertainty and bias, less j
assurance of completeness) j
BACKGROUND (continued) e May 1997 - Staff recommended that the Commission accept the voluntary j
approach (continued)
If approved, staff will continue to work with INPO and industry to improve i
voluntary data, develop a reliability database to estimate parameters for
~
j PRAs and risk-informed applications, withdraw rule i
e June 1997 - Commission approved the staff's recommendation. Directed the j
staff to l
Continue to work with industry to improve the content of the voluntary data I
to minimize Uncertainty in the data i
The use of "workarounds" to derive parameter estimates Update Commission on Efforts to work with industry to address shortfalls and limitations of data 1
?
Whether the voluntary approach remains viable 10
i i
4 i
l SOURCES OF DATA Outputs Analyze by Data from Analyses Check homogeneity across units and across systems, if ReIiabiIity Component EPIX, for failures and run times.
parameters, EPIX or NRC estimates (from LERs, appropriate j
9,, 4, 4,,
Tech. Specs., IST, etc.), for demands.
Classical (MLE and conf. Interval)
Bayes (distribution, mesa, interval) e l
Empirical Bayes if approp.
e Unplammed Train Umavailable dates and times from Check homogemeity across units and across systems, if unavailability EPIX, for components appropriate parameter, Exposure hours from SSPI or from Classical (MLE and conf. interval) l f,,
MORP/ OUTINFO Observed unavailabilities by train and by calendar e
f quarter i
Planned Train Planned unavailable hours from Check homogeneity across units and across systems, if l
unavailability SSPI, for SSPI tralas.
appropriate
- Parametn, NRC-estimated naavailable hours, Classical (MLE and conf. laterval) e q,
for other tralas.
j Trend for g.,4, Data for the parameter, by year, Weighted least squares l
4,,,g quarter, or month or Concurrent Train Usavailable dates and times from Concurrent trala unavailability from unplanned trala EPIX, for components malatemance j_
usavailability Unavailable hours from SSPI, for Upper bonad on total concurrent trala unavailability, e
Consmoom Cause Comnoment Failures from EPIX Identify candidate CCF events using search criteria e
type 4
11 T
I -
.O'-
....,.,,n
INITIAL COMMENTS ON EPIX STRENGTHS e
Reliability data for emergency diesel generator:: appears to be complete e
Linking of key components to their subcomponents, support components, and piece parts is a useful tool not available previously WEAKNESSES e
Some difficulty in extracting failures for use in reliability studies and PSA applications o
Guidance is needed on definition of failures to distinguish a complete failure from a conditional failure e
inconsistent reporting among units, e.g.
o RCIC pump reliability information provided for one unit and not another o
RCIC pump reported as a key component at one unit, not in the device list for other unit 12
.,...