ML20112D556
| ML20112D556 | |
| Person / Time | |
|---|---|
| Site: | Shoreham File:Long Island Lighting Company icon.png |
| Issue date: | 08/07/1984 |
| From: | AFFILIATION NOT ASSIGNED |
| To: | |
| References | |
| I-SC-LP-062, I-SC-LP-62, OL-4, NUDOCS 8501140368 | |
| Download: ML20112D556 (9) | |
Text
,
. - ~.
"hk SB-BatA OL-OxkETED USHRC v.
^
s 15 jai-8 Mi N
~
'i ICE M ECRt. :i.*'
rI>
/
r OCKETING A SE9V!r.l
,L aftANCH s
ATTACH' NT E
I u
9
~
^
NUCLEAR CEGULATORY COMMISSION
.....m-...
~ dd
- 7. Official Exh. No.
Docket N1 d / d da
_le#M AAd)d k la the cutter of IDENTlflED
$!rgf RECENED App:kar.t
. ROECTED
/
Intervenct Cergoffr
_DATEhY1l$Y C,ontractor
.raaess D_ E N A>> "I-sW
/
w C,/~ /d)'A[LS other
~
$Llxq.,
. /,
-U m/
Si
\\.1 *,k p?iD,
s
~
y
'l
~
f-
-71 "g(s /n
+~
a.$
19e e
s e' s
' /)S s !. N/
l m y(.-
. /
k e
h
c.u:- w... =,.,1
.: =.r a..
...-4 a -..
....z---
~
Attachment'E SENSITIVITY STUDIES'AND ADJUSTMENTS TO SAI METHODOLOGY AND DATA Some of the data-an'd assumptions used'by SAI in per-forming its Low Power PRA for LILCO could be improved or made more accurate.N The two most significant items are (1) the frequency of occurrence of the loss of offsite power transient at the Shoreham facility, and (2) the as-sumed means of restoring offsite power via the 69 KV switchyard.
It also appears that-slight changes are nec-essary in the probability of restoring power following a loss of offsite power and in the conditional availability of the 138 KV switchyard following the occurrence of a N loss of offsite power.
We have recalculated the frequencies of core vulnerable conditions due to loss of-offsite power, as set forth in Table' 1 of our testimony, using corrected data as described below.
First, we used a loss of offsite power-frequency of 0.25 events per year instead of.n82 events per year as was used.by SAI in both the Low Power PRA and its 1983 E-1 1
8 9
a
...;ma.
n-n.
t j,
=
PRA.
SAI's loss of offsite power frequency value is based on data concerning only the LILCO grid.
Thus, its value of :082/ year does not.take into account the probability of failures within the Shoreham. switchyard resulting in loss of offsite power.
In our opinion, the failure to account for such failures makes the SAI value' unrealistically low.
N The.25/ year frequency of loss of offsite power, which we believe is more realistic, is from a Brookhaven i
National Laboratory assessment of the frequency of loss of of fsite power for the nuclear reactors found in the Reliability Council region to which LILCO belongs.
See
' Table E-1.
We consider this figure to be conservative, but more realistic than SAI's, because it takes into l
account the contribution to losses of offsite power from failures in the switchyards of nuclear power plants.
Such failures are a major contributor to loss of offsite power Altho'gh we believe that a val even higher than events.
u the.25 figure.might be appropriate for a plant such as Shorehan which will be operated at low power by relatively inexperienced operating sta f f using equipment subject to break-in type failures, we did not increase the Brookhaven frequency in performing our calculation.
E.-2
. TtJ--
- - - -- ' -- - ;2 s--
- 2. _. ; - - - -- ; ;;--- _.::--- ;n--- :;n; ;c-----
_. a su Second, our recalculation also corrected what we believe to be an error in the SAI model for offsite power availability.
The SAI low power hvent tree for the loss of offsite power trans,ient takes into account the possi-bility that offsite power will be restored at different times after the transient, with varying probabilities.
SAI also assumes, however, availability of offsite 69 KV power with a probability of 0.99985, after the occurrence of the loss of offsite power transient.
We believe this second assumption is improper, and amounts to double-counting, because the probability of restoring offsite 69 KV cower is already included in the event tree in the time varying probabilities for restoring of fsite power.
We have eliminated this double counting in our recalculation.
The final major change we made was to consider the possibility of repairing the gas turbine and the EMDs fol-lowing a failure.
The SAI Low Power PRA did not discuss the possibility of-repairing the EMDs and gas turbine.
Thus, to the best of our knowledge, the values in Table 1 of our testimony reflect comparable assumptions of no repairs for both the EMDs and gas turbine, and the TDIs.
If the SAI Low Power PBA did include repairs of the EMDs and gas turbine, then the difference between the core E-3 6
9
- a..
~.
~--.:-.
- _ - ~
'y'
.g -
~.
f~
~ vulnerable frequencies for the TDIs and the EMDs and gas turbine is understated in Table 1 to our testimony, be-cause adding the repairability assumption to the TDI values would further reduce the probability of reaching a core vulnerable' condition.
We took values from the SAI 1993 PRA to determine the core vulnerable frequencies assuming the TDIs could be repaired.
To be conservative, we used the same TDI repair values used by SAI in our EMD and gas turbine event trees to determine core vulnerable frequencies for the alternate system.
The results of our recalculations are summarized in Table E-2.
Increasing the frequency of loss of offsite power increases the estimated frequency of core vulnera-bility due to loss of offsi te power by an equal factor of about 3 for both the alternate and the normal AC power systems.
Thus, the impact of this adjustment is only in the overall core vulnerable frequency, and the adjustment does not affect the frequency for one system relative to the other.
The elimination of redundant consideration of offsite power restoration results in a grester increase in the probability of core vulnerability for the alternate configuration'than for the normal configuration.
This E-4 e
h
_ i.
,. u i..c:~ G:..:..:
a.~.
2 - ~;
.a.....L.a.
- :. 5.
..t..
.2..-.-..L.
E...
'would reflect the greater dependency of the alternative system on the 69 KV switchyard availability.
Explicitly considering repair of the gas turbine and EMDs reduces the estinated probability of core vulnerabil-ity due to loss of offsite power for the alternate system.
The TDI analysis showed a comparable reduction in core vulnerable frequency when repairability was included.
This is expected because the system components might be returned to operation even though they may have initially failed to operate.
Combining. the corrections in data and nethodology de-scribed above, and assuming the possibility of repair for both the alternate and _ normal systems, the probability of core vulnerability due to loss of offsite power, is still about a factor of 4 higher for the alternate system.
Fur-thermore, assuming the accuracy of SAI's estimate of 1.6 E-6 for the annual frequency of core vulnerability from all other initiating events during 5 percent ope'ation r
(SAI 1983 PRA at Table 4-4-1), the likelihood that the Shoreham. plant would experience an event leading to core vulnerability during 5 percent operation is approximately
.8 times greater under the alternate configuration than it is under the normal configuration.
E-5 9
e k-
z TABLE E-1 PLANT-SPECIFIC POSTERIOR PROBADILITY FOR Tile FREQUENCY OF TIIE LOOP (Events Per Year) j RELIABILITY COUNCIL - NPCC PLANTS IN SITE N
T MEAN 5 PERC 56 PERC 95 PERC
- 1. Fitzpatrick 2
5.55 2.cE;01 9.6E-02 2*4EE0i 5.4E-01
- 2. 'inna 3
10.57 2.6E401 1.0E-01 2*.2E-05 4.6E-01 c
- 3. staddam Neck 5
13.72 3.0E-01 1.3E-01 2'.7E101 5.0EiO1
- 4. Indian Point 2 & 3' 4
7.94 3.5E-01 i.4E-01,
3'.6E-Oi 6.2E-Oi'
- 5. Main Yankee 1
7.62 2.0E-01 5.3E-02 l'.7E-Oi 3.8E-01
- 6. Hillstone 1.62' 1
10.47 1.7E-01 4.5E-02 l'.5E-01 3.2E-01
- 7. Nine Mile Point 1
11.32 l'.6E-01 4.3E-02 l'.4E101 3.1E 01 1
- 8. Pilgrim 4
7.96 3.5E-01 1.4E-01 3'.0E-01 6.2E-01
.)
- 9. Vermont Yankee 1
8.19 1.9E-O'1 5.1E-02 1*.6E-Oi 3.7E201
- 10. Yankee Rowe 1
20.70 1.2EE01 2.9E-02 l'.0E-01 2.2E-01 AGGHEGATE 23 104.04 2.5E-01 4.4E-02 l'.9E-01 5.8E-01
[
Source:
-I.
A. Papazoglou et al, Bayes Analysis Under Population Variability With An Application to the Frequency of Loss of Offsite Power in Nuclear Plants, DNL Peport, Feb., 1983
TABLE E-2 REQUANTIFICATION OF SAI EVENT TREE FOR CORE VULNERABILITY DUE TO LOSS OF OFFSITE POWER TRANSIENT (Frequency Per Reactor Year)
\\
Gas Turbine /EMD Diesels TDI Diesels Type Non-Repairable Repairable Non-Repairable Repairable 1
2.3E-5 1.0E-6 1.4E-6 6.4E-8 2
1.9E-5 1.7E-6 1.2E-5 1.lE-6 3
4.0E-6 2.0E-6 7.0E-7 3.5E-7 4
5.6E-6 1.3E-6 6.8E-7 1.6E-7 5
8.7E-6 2.6E-6 1.2E-6 3.6E-7 Sum 6.0E-5
.87E-5 1.6E-5
.21E-5 Note:
Column totals may not exactly equal the sum of the figures in each column due to rounding.
9 4
9 h-
- _ _,