ML20205K902
Text
_
J' i
- o m
Carolina Power & Light Company OCT 25 1988 i
SERIAL: NLS 88-243 United States Nuclear Regulatory Commission ATTENTION: Document Control Desk Washington, DC 20555 BRUNSVICK STEAM ELECTRIC PIANT, UNIT NOS.1 AND 2 DOCKET NOS. 50 325 6 50 324/ LICENSE NOS. DPR 71 6 DPR 62 SUBMITTAL OF ADDT"If"'AL INFORMATION FOR PROBABILISTIC RISK ASSESSMENT REVIEW (NRC TAC N 4ERS 67123/67124)
Centlemen:
Representatives from the Nuclear Regulatory Commission, their consultant EC&G and Carolina Power and Light Company met on August 25 and 26, 1988 to discuss the status of review of the probabilistic risk assessment (PRA) including the fire scer. rios.
The meeting provided the opportunity for a good technical exchange and overview of the PRA models.
Several requests for additional information and supporting documentation were made during the meeting.
Responses to these requests are provided in the following attachments: :
Set of Alternate Safe Shutdown Procedures :
Set of Emergency Oper ting Procedures :
Containment Venting Diagrams :
Responses to Questions 1,5,8,9 and 10 frorn Mr.
i P. R. Davis.
(NOTE: Responses to questions 2,3,4,6,7,11,12 and 13 were provided during the meeting.) :
Additional Copy of CP&L Letter dated January 4, 1988 transrnitting Probabilistic Fire Analysis g
can r aa w
e.. nae i
- ,s.4.. m
<a n l
(e.,im y es4 a <.w
,n
.. o
.,,....,,.. s e..,.
c..,,,,. n...,,, c,,,,c,
'==== ~ ~ ~ = ~ ===
g,pOg00CKC0000324 03ee se102,,
P L
P8tg
Document Control Dack NLS 88-243 / Page 2 Please refer any questions regarding this submittal to Mr. Stephen D.
Floyd at (919) 836 6901.
Yours very truly, 7
Leonar I.
lin Mariage Nuclear Licensing Section SDF/sdf(\\cor\\praques)
Attachments cc:
Dr. J. Nelson Grace (w/o attachments)
Mr. W. H. Ruland (w/o attachments)
Mr. B. C. Buckley (w/o attachments)
Mr. E. Chelliah (NRR)
Mr. M. Sattison (EC&G)
Mr. S. Mayes (EC&G)
Mr. P. Davis (EC&G)
Mr. D. Ward (RII)
e e 4 9
ATTACHMENT 3 CONTAINMENT VENTING DIAGRAMS 2
4 I
J
g N
>4 b
Wh v an
<ti u
.n$.
s u
1 4
i 1-r L
b
- h h
5n d
d j -
5
/
\\
u
')
p te h,
g 4.__
y 32 [**?.
ased.
a..-
<b!?.D
- !!j h%;
J
!g !5w>
5 Nm 2~-
s.
4
!aEu [*AddI..
s _.. - 7
,g, g
c a 4& kyd y
' 'i. c. s w-
~
J V
"N s
I i
I h
6 3
0-!l0-]:!
- h. )-[
n :=
e-6
! d, h, leN 04!
0-):k.
a 0-d se,,
e
- m g
'M g
O O
2
~
a p >
n{ f s
l
$ :54 i
T_
l lH gg
> =c ' a d.
d::
e e
e-0 se gaaaal s
A
>~_s,
__3 e
.t.! w g
}
V N D
(.
~
n c w c
W
$ 4D 3
9 6 3
+4 *e w 8
'f s,-
u 4
e g
7 2 ' ~g a
1
=
4 W
D U it) i
%P
- p
'""~Y 9"**N jj f
I i
1:*5 ; 3%y f
t' a
y I ' ' L,',.
MWeth gg y
e
% ~
S.J di JQ ef yJ g
_ - --. - =._
a32 et
{.
1 I
h 5
ah,e]M3
? +
i i
i ;
i l
60-i j $1:* : 1 0:k,
~
jfy,
a,,,
e, sm te 2
f I
+1:e s
> :t s
E*[#
'g
$ :54 V
I
~
~
?
< A6, 6
4
- > =O'
- Lj' filly
/6agi FW, 9
g666-L
)-as 0
+
,p 4
a d
v
'4 "4 e b +i a
o a.
u E
a h
(,
A 3
?
?!
y e
y 4 [- g a
- 4....+g ' -
)
.Q g
e !v6
- 4 h%
[eb
- i i
r i
'i E
!4 M W H!t.
i y-
.L_
,o g i
o v%.,
.,*e
- F ]' ' '
N 9
y d
n V
d m
['
g*
K.T 8
5e D-!: 3
, a!
H
- Rc; i
O E
3
$h; D-] 9 a_];h'-
y e+
,a s.
- O
- ? ::
n;>
5 i
2{
e s
i I
enan e
5 M*
S l
\\
~
l
?
ih59E
)
_v o., J..
i e
3<
e,_
/\\
"*'m /
3 65
%I
)b
'f g
> =A>-
iy-w I' ]a q.
I
%w M
e,.
e 1
a
~
a Al
,g
't
~~.
m3 9
7 i
'f o
a y
b E
t h
b
+ !
'ji i
h%
,8
?
e 5
hW 4kfo g
s i
=.
t s+ :
ea !
t i
C 2
V v
m, 9
e..
n'
,;-mi f
l a
i.x.
3,dM S E e
,W b O i
i is a
4 g
3 m::k.
?*
i 0 di;,
$N N:
0
_f.
se
- ze n; su f
9 0
m > He y
es s
X y
1 B *,1 s
2 E
k 3 g3.g:v @ g..>
gg ;
J 3<
4s m/
Qo j
eN. A
Y 1%
8
$I 2~
1,
_3 ibd
> h >a '
3 3}
o e
a 3
k 1
A s
x f {+s 1
1 ss z
.,*o a
j 2L
+
i
,3 i 1
?
E!jg y
u s
sll [~2h
{
< * = = - Yh'..
.=***-$
g f
?c gL9 4:*3; h%
x
~
e pii h.
d 5
q i
,f@i I
..P, -
~
f g
=.
e n=
m,
@6 q
d j'-y r-I N
5 r: a. o-
- D m
i d) O E
3
.3 h b
b e
te a; ;<
5
Y y
l Ho i
a mga m:g s
I Ed y
1
$11@@I
@ - j" Aa l
)
e
) "aab a X' g *-
5<
/aN W
55
%l g555'7 gi
> =O
P--
gS i
t g+2 3
a b
a ki8 E
E q, l
.R f
5.
a T
y e!
4 Q
>U Y
m e
- w a
{g
,,, Iy,g.
-@ p-+&
gg g
t
! e t,.
u; I
N e
l' r
n 5,
g I
,-n$$"kld I
~
-- ng=
1 g
a 7,
A 9
- 1.,gy 6
i r
Y
\\
- u. -
E 5 rWWs i
3 ld
<a60-N d
7 I
i I yib, jN H3p>p?%-
y-te
=,,
y g"
2 3
8 8
H'? E 0-X}
E 1
Ed
$ :!;6 xY I
?
A4994 e
i<
h
> 249 *
- j' kiiEI!
e a
a6666-
>= %
O 4
G 1
l ATTACHMENT 4 RESPONSES TO OUESTIONS 1.5.8.9 AND 10 i
i I
I l
I f
l I
4 1
The following responses are provided to questions from Mr. P. R. Davis which were not answered during the review meetings on August 25-26, 1988. All questions are related to Volume 3 of the Brunswick Plant PRA.
Question 1 (p. 3-2)
It is not clear why the wind data is restricted to two time periods (1872-1971 0 Wilmingtoa and 1876-1894 0 Southport).
Isn't more extensive and more recent data available? Also, please explain how a 140 mph wind speed return period of 1000 years was derived from the Southport data which includes only 18 years of very old data?
Response
The external flooding analysis is a screening study; therefore, the Southport data were used to be conservative.
e The source used for windspeed versus return period was published in 1982 and was considered to be the best source at the time.
A more recent source (Neumann et al, July 1965) is listed on page A-37 of the final report of NUREC-1032 (June 1988), but this data source was not known to be available at the time the flooding analysis was performed.
Question 5 (p. 7-31)
What is the basis for the gasket and flange failure rates quoted here? Also, it appears the operating hours per year for the plant was assumed to be 5256 hours0.0608 days <br />1.46 hours <br />0.00869 weeks <br />0.002 months <br /> (60 percent availability). What is the basis for this numbert i
)
Response
First, the means for gasket and flange failures for both data sources in Table A.3-2a were converted to medians and a geometric average of these values was taken. This provided a composite median failure rate for gaskets and flanges. This value was then converted back to a mean. Then the value was divided by 100, reficcting the assuantion that 1/100 of gasket / flange failures provided catastrophic leakage. The resulting value of 6.0 E-8/hr. :epresents the catastrophic large leakage failure frequency.
It should be noted that the WASH-1400 data source shown for gasket leakage in Table A.3-2a (page A.3-23) is incorrect. The correct values used to derive the 6.0 E-8/hr. value are shown on the attached markup.
The number of operating hours per year for the plant (5256) was chosen on the basis of an average plant availability of 60 percent per year (0.6 x 8760 hr/yr = 5256 hr.).
The systems utilizing these gaskets may well operate a greater fraction of the yeart however, since core damage events during shutdown conditions were excluded from this risk assessment, 5256 represents t
the number of hours per year that these systems will be required to prevent core damage from events during power operation.
Question 8 (p. 11-19) t Please explain how the use of generic fragilities provides conservatisN.
(A1.so claimed on page 11-33.)
a rsv& o n.n.
e e
l i
i 1
if f
it it i
j,; i = :.
s c sl
.].
i,l s1
,1
..e s,.
g...
ga o
. si I
i.
se
.s
.s s
L-a
.I I
.t.
i 5.
i in ii e 3 s
t
" f I$
q g
4; a
a a q
Iea a
a gY
}J
]
.r.
3g 11 1.
l
),1f.4
=3 a
a s
I I'. I.'
- a. !
l
! Ik k
k.
-j ll
. s.e a
u wi s.
M
- s. 4
.e ea y
.a
,1 A
l 4 i
- 3j
(
f i
11, s' '!,
,3 j'.<
g 1
g..
p n
i:.
((
e I
O M
. 4 M
l l
$. il
,j 1
.i l
I at l
l
,UI
}
.' 1 3]3 f..
'g l
?
'll til
}:ll! L
)I !}].I 1p-lji.i
}s 3j]
I i
s J
l A.3 23 L
Response
Seismic fragilities assigned to the components based on the review and walkdown of Brunswick were generic in nature.
For each component type, the seismic fragilities from published and unpublished seismic PRAs have been compiled in Reference 1.
These fragilities were derived using a combination of generic and plant specific analyses, fragility and qualification test data, and actual earthque'.e experience data. A review of these fragilities was done for appropriatenest to the Brunswick equipment. The median capacity A was assigned using the lower range of the median capacities of the equipme,nt type reported in the above reference. The variabilities, B and B wrre assigned R
u using the higher range of those reported in this reference.
The combination i
of low median capacity and large variabilities makes the generi fragility assignment conservative.
Reference 18 Campbell, R.
D., M. K. Ravindra, A. Bhatia and A. C. Hurray, "compilation of Fragility Information from Available Probabitistic Risk Assessments," Lawrence Livermore National Laboratory, Livermore, CA, UCID-20571, September 1985.
Question 9 (p. 11-26)
Please explain the trimming process described in Section 11.3.3.2.
F
Response
Tae fault tree trimming process discussed in Section 11.3.3.2 is used to remove those non-seismic events that would not be risk significant from the fault trees before quantification. The screening criteria for trimming non-seismic events (such as equipment unavailabilities, operator errors, or common J
cause failures) were developed for NUPEC/CR-4826, Seismic Margin Review of the I
Maine Yankee Atomic Power Station. They are more conservative than the single criterion of 0.01 given in NUREC/CR-4482, recommendations to the Nuclear Regulatory Commission on Trial Guidelines for Seismic Margin Reviews of Nuclear Power Plants. The criteria are based on experience with past seismic PRAs and knowledge of the relationships among the seismic hazard curve, component seismic fragilities, and non-seismic unave.ilabilities.
i The system fault trees from the internal events analysis were modified for the seismic analysis by first adding the seismic failure events to the fault 1
trees. Then, non-seismic events that had unavailabilities lower than the 1
screening criteria were removed from the fault trees.
If the event to be removed was an input to an "0R" gate, then the event could be simply removed l
from the tree, with some consolidation of the tree gates for clarity.
If the event was an input to an "AND" gate, then that gate and all of the subtrec
[
below the gate was removed from the fault tree. All non-seismic events were j
reviewed for trimming, including non seismic common cause failures (treated
]
with beta factors), operator errors, test / maintenance outages, and equipment unavailabilities. The trimmed fault trees were then linked and evaluated to generate the Booleaa expressions used to quantify seismic core damage i
frequency.
(346DED/ cme) l
Question 10 (p. 11-27)
Please explain the alleged conservatism in the assumpcion that the component failures are statistically independent both in terms of the randomness in the capacity and the uncertainty in the capacity.
Response
A review of the Boolean equations for seismically induced dominant plant damage states and core damage at Brunswick indicates that the failure events (i.e., components) are in series. Assumptions of independence in the randomness gives the upper bound on the failure probability of the plant damage state (or core de,iuge) when the components are in series or singletons. Assumption of independence in the uncertainty of median capacity results in spreading the tragility curvec for the plant damass states thereby increasing the annual frequency of plant damage states at higher confidence values. Some sensitivity studies were performed to confirm these observational Table 1 below shows the results for the plant damage state PDS1A with the upgraded fragility of the battery charger.
TABLE lt ANNUAL FREQUENCY OF PLANT DAMACE STATE PDSIA Case Mean Median 95% Confidence Randomness Uncertainty 1.
Indep. - Indep.
4.4 E-5 4.4 E-5 1.1 E-4 2.
Dep. - Dep.
1.9 E-5 7.9 E-6 6.7 E-5 3.
Indep. - Dep.
3.1 E-5 1.4 E-5 1.0 E-4 4.
Dep. 50% correlated 3.1 E-5
- .2 E-5 7.4 E-5 5.
Indep. 50% correlated 4.1 E-5 2.7 E-5 1.0 E-4 l
Figures 1 through 5 show the plant dam 4:e state fragility curves for these curves. The conservatism in the assumption of independence in randomness and uncertainty can be judged by inspecting Figures 1 and 2 where the fragility curves for the independent-independent case are shifted up and to the left with respect to those for the dependent-dependent case.
A review of the other plant damage states developed in the seismic PRA of Brunswick has shown the following The plant damage state PDS1B has singletons, and doubletons and tripletons containing seismic failures with non-seismic unavailabilities.
There is no reason to believe that seismic and non-seismic failures are correlated.
It is 1
i 346NE 0/1 ah )
Judged that the seismic singletons contribute most to the frequency of this damage state. The assumption of independence in randomness and uncertainty in these singleton failures is conservative.
The plant damage state PDSIC has seismic doubletons, and tripletons containing seismic and non-seismic failures.
It is seen that doubletons consist of failure of either the reactor internals or CRDM pupports ar.d block wall or MCC anchorage. There is no reason to.xpect correlation in the failure of reactor internals and the MCC anchorage, given the earthquake level.
Note that the major common cause source of large earthquake ground motion damaging both these components has been treated consistently by keeping the seismic hazard outside the integration.
Plant damage states PDS2 and PDS3B have seismic failures and non-seismic failures appearing as doubietons. Again, there is no reason to treat the non-seismic and seismic failures as dependent.
Plant damage state PDS3C has non-seismic failure (#17) with the union of several singletons (seismic failures). The union of these seismic singletons has been evaluated with the conservative assumption of independence in randomness and uncertainty.
In the plant damage state PDS4, the major er tribution comes from the doubletons of reactor internals or CRDM supports and fallere of any one of electrical cabinets, pumps, heat exchangers, tanks. etc.
The union of singletons has been quantified using the independence assumption which is conservative. As stated before, there is no reason to believe that failures of reactor internals or CRDM supports are correlated wi"h the failures of other components such as pumps and electrical cabinets and treating them as independent is justified.
It is therefore concluded that r.he assumption of independence in the randomness and uncertainty in the component failures is conservative for the dominant damage states Considered.
assvE0nsu
BRUNSWICK - POS1A PLANT LEVEL FRAGILITY CURVES MEDIAN AND CONFIDENCE LIMITS 0
LIJ x to E
04-06-80 15: 25: 17 D
[
g s.oA
}
i-4 LL lt O
- " 95% Confidence Curve Median Fragility Curve 5% Confidence Curve
-/
-/
H x
_J H
. tr-D<m O
E o.
.4--
J<
Zo m
.2-F-
H O
Z O
U
.0
^
.0
.2
.4
.6
.8 1.0 0 x 50 PEAK GROUND ACCELERATION (g)
Case 1
BRUNSWICK - PDS1A, FULL CORA.
IN RAND.
S UNCERT.
PLANT LEVEL FRAGILITY CURVES MEDIAN AND CONFIDENCE LIMITS 0
11 x sn 09-07-88 08: 41: 04
]
s. o--
H
<t Lt_
EL O
8' V--
95% Confidence Curve g
H
" Median F. agility Curve m
/
/
5% Confidence Curve 3
O E
ct
. s--
__I
-1 Z
O H
.2--
H H
Q l'
Z O
L3
.c
.o
.2 4
.s
.e s.o,
X 10 PEAK GROUND ACCELERATION (g)
Case 2
BRUNSWICK - PDS1A.
FULL CORR.
IN UNCERTAINTY PLANT LEVEL FRAGILITY CURVES MEDIAN AND CONFIDENCE LIMITS 0
tij x 30 09-07-88 08: 45:32 E
D
.o--
to tL O
e--
955 Confidence Curve -
t-Median fragility Curve J
5% Confidence Curve 3
C<
LD O
E O_
.+-
J<
y Z
O H
.2--
H Q
Z "o.
.0
.2 4
.6
.8 1.0 0 X 30 PEAK GROUND ACCELERATION (g)
Case 3 1
BRUNSWICK PDS1A:
RAND.- FULL CORR; UNCERT.- PARTIAL C PLANT LEVEL FPSP.ILITY CURVES MEDIAT' AND CC.s. iDENCE LIMITS 0
UJ x 10 09/07/88 11: 11: 15 E
3 1.o--
H 4
tL o
. e--
F-H 95% Confidence Curve--*
H
.&-Median Fragility Curve r-5% Confidence Curve r
m O
E O_
4--
J 4
Z O
H
.2--
P H
O Z
O U
0
.0
.2 4
.6
.8 1.0 0 x 10 PEAK GROUND ACCELERATION (g)
Case 4 J
i BRUNSWICK - POSIA.
PARTIAL CORR. UNCERTAINTY PLANT LEVEL FRAGILITY CURVES MEDIAN AND CONFIDENCE LIMITS W X se '
09/07/88 11: 16: 45 E
D J
s. e--
H 4
It O
.e--
95% Confidence Curve MedimfragilityCurve!
)
u 5% Confidence Curve (D
4 (D
O
[
0
. s,-
J<
l Z
O H
. z-_
1 F-H 9
i o
O U
.o
.e
.2 4
.s
.a s.e X se PEAK GROUND ACCELERATION (g)
Case 5
-