ML20249B572
| ML20249B572 | |
| Person / Time | |
|---|---|
| Issue date: | 06/19/1998 |
| From: | Bennett Brady NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | Baranowsky P, Mays S NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| References | |
| NUDOCS 9806230318 | |
| Download: ML20249B572 (9) | |
Text
l l
1 t
UNITED STATES l
g g
j NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. 20e06-0001 s.,*****/
Jme 19,1998 MEMORANDUM TO: Patrick W. Baranowsky, Chief Reliab;lility and Risk Assessment Branch Safety Programs Division Office for Analysis and Evaluation of Operational Data l
I Steven E. Mays, Chief Reactor Risk Assessment Section Reliabilility and Risk Assessment Branch Safety Programs Division Office for Analysis and Evaluation of Operational Data FROM:
Bennett Brady Reactor Risk Asse sment Section Reliabilility and Risk Assessment Branch Safety Programs Division Office for Analysis and Evaluation of Operational Data
SUBJECT:
MEETING OF EPIX AD HOC GROUP, JUNE 11,1998 Thursday, June 11,1998, I attended the first meeting of the EPIX Ad Hoc Group at INPO. The invited participants (Attachment A) had been selected to represent a broad range of interests -
PSA users, systems engineers, engineering management, industry representatives, EPIX coordinators and NRC.
Data to be Provided Under the Voluntarv Anoroach Piior to the meeting, INPO sent participants sections of the February EPIX guidance that we reviewed earlier. This guidance is consistent with our agmement on voluntary data. There were no proposed changes to EPIX nor recommendations made during the meeting that would not meet the proposed voluntary approach that we recommended accepting in SECY 97-101.
l l*
However, INPO is running slightly behind schedule. The March 1997 letter from NEl had promised NRC access to EPIX data in early 1998. About 80 pucwt of the utilities have entered failure data into EPIX. Failure data for 1997 were due April 1,1998; failure data for first quarter 1998 were due by May 30,1998. INPO anticipates that utilities may also be late in ey submitting the reliability information that is due June 30. An INPO representative informed me p
l that they had expected to provide NRC reliability data by mid-1998 but they may not be able to meet this.
d H 4 5 -3 4 2.c 31
- ' !" ^
jgTV y 04 4 -(2 Hubs-l 9006230318 990619
~
l PDR ORG EPSINPO i
- f. 1 Wcsic toward NRC t==== diam--4 in SECY 97-101 Also, prior to the meeting, INPO provided participants their master list of about 30 problems and
. work items remaining to be completed for EPIX. The proposed meeting agenda (Attachment B) had been devoted almost completely to setting priontes on these and other implementation issues. However, the meeting evolved into an open discussion of the participants' concems and issues and then prioritization of their issues.
During the discussion, I presented NRC's primary issues - categonzation of demands by type of demand, c^:;-:-6R-i of run hours by nature of operation, planned unavadability for non-SSPl
- systems and assurance of the quality and wirCteness of EPIX data. After the meeting, I provided a copy of these issues and proposed changes (Attachment C) to Bruce Keener. Of the issues discunaad by the group, INPO agreed to 22 top pnority issues. These included three of NRC's areas of concem - better definition of demands and run hours and quantification of quality. The INPO i.pi::: Ative said that reliability data will be looked at closely in the coming
~ years to see if they can better define these data. However, no commitment was made to the specific changes proposed in Attachment C. No comment was made on the proposal for collection of planned unavailability for additmal systems.
I Other areas of improvement that may enhance reliabdity data Of the remaining 22 issues agreed to, the following may enhance the quality of the reliability information from EPIX:
e Better definition of functional failure Clarification of component boundaries (PRA boundaries versus maintenance rule e-boundaries)
New field to designate whether a functional failure was a PRA failure e
Other charw== in EPlX that may be of interest to NRC Two recommendations for changes to EPIX (Attachment D) were to drop two of the fields for maintenance rule implementation - tracking of systems' or functions' maintenance rule status (a1/a2) and tracking of systems' or functions' maintenance rule scoping criteria.
Follow-up to meeting INPO asked members for input in writing guidance on reliability data information and for comments on Chapter 1-4 (guidance on failure and reliability data) of the February guidance.
. INPO proposed meeting twice a year but no deosion was made.
4 O
7
. As follow-up to the meeting, I plan to e;
Contact INPO early in July on their anticipated schedule for industry to complete.
submission of 1997 failure data and reliability information e.
Also discuss with INPO their decision or_when they expect a decision on the particular changes proposed by NRC e'
Provide the maintenance rule group copies of " Focused EPIX Emergent Recommendations - June,1998" and coordinate their comments on the proposed changes to maintenance rule fields
- ~
if time pomwts, provide comments on Chapters 1-4 of February guidance and definitions.
for a list of "PSA draft input" (Attachment E).'
Attachments: As stated cc INPO Distnbution:
RRAB RF CERossi NRC Reliability Database Coordination Group SDP RF DHickman
- File Conner JRosenthal PDR- ~
DRasmuson DOCUMENT NAME: H:\\BMB\\ EPIX-J98.TWO r
To recohe a copy of this document, inacate in the bou "C" cop r wie attachfenol "E" copy wIsttachfenci "N" No copy OFFICE RRAS/RRAB 8
RRAS/RRAB c.
RRAB c
NAMEi BBradyhl[
SMays %
PBaranowskyM 4
DATE.
[ / /7 198
(,/s) /98
(,//1/98 w _ _x = _-
= _ _ _ = _ __ _ _
I Attachnent A EPIX Ad Hoc Group Members Name Organization Area of Expertise JoAnn Amick Summer EPIX Coordinator Jerry Andre Westinghouse NSSS representative Bennett Brady NRC NRC representative l
Ken Bych Diablo Canyon PSA
' Sam Cummins Nine Mile Point EPIX Coordinator Mike Davis TMI system engineering Ernie Hite Duke Power engineering management John Holmes Palo Verde engineering management Greg Kmeger PECO PSA Jim Maddox Watts Bar engineering management Jim Mercer Comanche Peak EP!X Coordinator Paul Planing Dresden engineering management Fran Rosch NMAC NMAC representative Steve Rowe-ANO EPIX Coordinator Jim Schottel Clinton EPIX Coordinator Tim Scoggins Callaway system engineering Jim Soletski Kewaunee system engineering Kavi Walia Vogtle system engineering Lucky Wroniewicz VAP EPIX Coordinator I
l
}
d
.J
Attactrent B M
EPIX Ad Hoc-Group June 11,1998 Starting at 8 a.m.
Room 216, Institute of Nuclear rower Operations k.
Agenda; Topics
- ,, w; Welcome and intmductions Ed Hux 8 s.'m06
Si30 a.m.i EPIX overview Bruce Keener Glen Masters 9:15 a.m.
Break 9:30 a.m.
Discussion of master list items (and possible All' additions / changes) 11 a.m.
PSA usage considerations All 11t45 a.m.
Lunch Provided by INPO 12:45 p.m.
Continued discussion of master list All Other implementation issues All 1:30 p.m.
2:30 p.m.
Break i
2i45 p.m.
Getting info into the hands of the users All 3:30 p.m.~
Summary and adjourn Bruce Keener Glen Masters
f Attactinent C
[.!..
ISSUES WITH VOLUNTARY DATA Data on planned unavailability is optional for non-SSPI systems e
impacL-Data are not availability on planned unavailability twt are needed for o
NRC to verify total unavailability used in PSAs and risk-informed PSA a
applications use in approving requests for Tech Spec. changes such as allowed a
outage times determine instances and probability of concurrent outages of two a
or more trains to input into the NRC Reliability Database, and a
For industry, to use in making industry wide comparisons for use in maintenance e
rule implementation NRC must make estimates of equipment unavailability from other sources of data (such as SSPI ) and based on conservative assumptions, upper bounds on the concurrent train unavailability These estimates may be conservative Corrective action - Make the optional reporting of planned unavailability a l
o requirement for the most risk-significant systems (sLc to 16) as determined by each unit during maintenance rule scoping Data on demands in EPIX are not specified in sufficient detailed categories e
Impact - NRC cannot determine if data from test demands and non-test demands o
are homogeneous, i.e., come from the same population as determined som statistical tests and engineering judgement, to be combined to estimate component demand failure probability for use in PSAs and in risk-informed regulatory applications input into the NRC Reliability Database o
Corrective action - Categorize demands by a
the number of valve-stroke tests versus valve flow tests a
the number of EDG manual starts (and manual loads) versus automatic starts (and automatic loads)
E the number of return to service tests versus other tests, and a
for non-test demands, the number of actual / spurious demands versus operational demands (e.g., filling accumulators versus injection to RCS)
Running hours for operating components in non-SSPI systems will be provided as one e
time estimatas; hours performing or testing the c=_mt's risik-significant function are combined with hours performing routine opcations Impeg - Data necessary for verifying the operating failure rate of operating o
. -... ~. -. _ -.. _...
a t
i 11 equipment used in PSAs and risk-informed applications are not directly available from EPIX alone NRC will develop estimates ofoperating times from LERs and Tech Specs These estimates may be conservative and contain more uncertainty than actual data on operating hours Corrective action - Categorize operating hour data by the nature of the operation -
o test, routine operation, or safety function Assurance of the quality and completeness of the data is somewhat limited e
Impact - Risk-informed applications may be subject to greater scrutiny without an o
agreed upon process for quality controls Users cannot assess quality of EPIX data and make statistical adjustments to estimates derived from EPIX Conective -*iaa - Collection of a small statistical sample ofrecords from utilities o
on periodic basis to compare with data in EPIX to verify completeness and accuracy and to develop estimates of error I
O
___m_______
h a
nr b t e a m
a n n
nc e
r u s
r h lbt uci o l
e po a
a c wb v e
t miiotnny
,a s
y c eum a
d d b e
a us s r e g-is oa or inn e
s o r
nl c
r r
a es r
t f ai pn c
r a pa e uo d
on oh ch) n c
mi lt o
i ems t
e i
f r
t eh o s
ph od i
d t
nsi gn d gt f
ow e
none n
s c a c om ma t
n t
n n n no c s a ik it
,t n
vb la mr l
ne ai e e y is t a om at nt l
i nu c
n c
oc s r
ci ine c ntaVe o e
lyp rof n e o i
gnl a sOf p d s a n f r u f
el f f
y a
r o c e
s s n e r ef h un inl t
p ehAa m is l
ri a e ed Ri isi t
ga s o e
pid r e
l eb l
nt f ch k mt c
t, t
n
.ic et s e oea n
s a
.V hf n a v
c d e e
a g
o i
r a
r gO c o igt iwa v ol e nt ki c
c t
e if s
u n(eMh p n o yh le c el a s is r i
n i t o
r r u a
nm li r
t y eR ihl gs wu e c rh mlns e
,e h
o t
t r s r u
e wi mie mrl a
i s c e o oene d e b
t
,r n
h f
s e a ui t
g h cih 8
o t
g k
wc eh d
a s e w l
e h
s t
o n
t o nin n
nt si i
t n
f in nw yt waMf i
e 9
s f
s s
y i
i a s a a y
c t
r s os i s s s n o
e 9
n nd pmn c Wd s
s e e s t
pi e s
ei t
e f w
m w m wnmt w n An n
s o
o u
u md wree 1
noa F
s e e
ie l
k r us c el r oub mi lont ot ot t n
us al e R
s l
s l
oo g l la sll y t c el l l
/ e n
a a e n sl a r l
e u
B di uAf pmis ApsAs Aro yM A m Wu lp i
y s
r a
n g
Rii l
s u
n 2
1 2
Rfah in J
O t
2 l
I
-s wn n
e e o o
k
's c
e
)
iis n
l e
v sl
,t eis s
ma ui a
s i
i s ush r t e n rRh RP w A b o
r t
m i.
a r n t
t o e s e o/ es eO f n yw r
S u p
d d ci si cE o P s n
s o
nn. n n
W p f a a a ga in e
f e
R m oM) n s
o y
.n
's e (e ed n iut t
r g
2 R o n 's a mt r
m t
e e r
a n a
/ c i
l W y k nl ei nsd qa m
a niui e p m
o a t
a c
r m
e ci yMt Mia ld a u
s e
s A
at (
s c k
r c s s
n o
R a
ef a e nginis u s s
t t
n u d 's t
c n
d ut c
d t
e e or d
s e e
e ir i iu/o iut a
wo e c r r s io t
mmsug pe s t
r s
r R
u f
eai wec u
u e
a qi qdl qc pt c
t r
t o
n nos ci ef o
u e
t e
e g e n uc ye r
RaR Rf s sb m NopF i
n F
R s eg e
e e e e r
o m
n s
m oe m
et u o
h s e
h s t
r a s u
io e
h t it n
et u o
r t
r r e eh t
r m
t o u h
n a ss ly
,f s y
, f yil m
t r
a e
s ol s e e e r a a s s ol s e e it l
E w d c
o t
t l
ia a
pid e r a a
s if r
aa c c c a a b
o s
c c c t
e e mv it k
X bd i c n Xh c
e a n f
sI
=d s n g
l t
o o c a y
c a paI a p X
i a t
t t
= u nn in p
t i
t a
ynP8 i
a yip r
ndl c p rh r
eE a u rh l
r I
o psi P g eE t
a o n
t e
c o o n pd di ly P g et d s et d s ul ea t d i ul P
d s ir g r ge u wi e
a s cd ai f is le opc n le u le opc nl o e s u r
s e
o e i
E c
u us u no r e ups e a u not is u us t
c t
d cr uApe R
pa u R ina R
pa uR na s lc s
t s ee c e e c d
e cf iul o nn c a a no c nt c a a noc nt lyn e e t
ib i
ly s af r sS s t
d b
t t
et a ent t
t et a a r sk o
e r t
i Pf e
u i
s e o qe s o
nt h on n oa nt h enn oa n on nd c
s a cd af oi s
t c
a cd ai t cs n ns e Ats it b s ged u gldi ai t
n ns e u
ns ya o ee s u
nnr oon ua ns ya o e nr e e s e nr o la q t
c o elid me r i om t
t ot si t r
s a a ni u nc s a a ni u S t
ot si t
r r
l t
u o o
n pel e c r iat di s a qh iat di uo P u
nc nmd oAmduod f uo u
t i pc t
o u
t i
s s s a qh eel Mri Mo nhf Mr i it r aty o
S n n I
F a o o
e n e n mWPNmai Mnitnhf nit C c s r cil I
s
- e ei is nl t b
t a
i a A B
l 2
3 1
2 1
2 Of r
n
)
n W
e2 e
a c
nl n
n A
c s
c a t
o Sin e
r o
R i
al a
it o
R na n
P s i t t
t p
d t s r a
/
e(
e c
t f l r
i ob oe d
t t
n n n oe peh n
We n
t s
n n
u oe r
s u
e nf e w y
s e
Ae at gm ia F
it i
m ot m
Rm Mt r
er a
/mgammi M
r s a e ec a n
e i
r ine m
e um S
r o
pi pek ir r
e r
uv u ud pet po u lcl i n om u
u ol s el s c
o ol c of q q
r ua q r uyc r u n e neaE ps e
r e
r e DRS R
Dr DRt i r f r
si r e
se e
2 3
4 5
1 l
j e
i Attachment E l
PSA DRAFT INPUTS 6/11/98 l-Definitions 1.
Unavailability 2.
Failure to Start.
3.
Failure to Run 4.
Common Cause Failure (CCF) 5.
Demand (testing) 6.
Demand 7.
" Planned" Urmailability 8.
" Unplanned" Unavailability 9.
Out of Service (OOS)
- 10. Initiating Event (Frequency)(IEF)
- 11. Human Error (Probability)(HEP)
- 12. Risk Achievement Worth (System / Train / Component)
- 13. Risk Reduction Worth (System / Train / Component)
- 14. Basic Event (ties to boundaries)
- 15. Key Component Data Collection A.
Breakers, Heat Exchangers, Check Valves, Electric Heaters - Need demands, run hours and failure mode for these " passive" devices B.
Emphasize rotating " actual" data in data entry whenever possible (better than " estimated" data)
- C.
System, train, component unavailability
_.. _. _ _ _.