ML19309E865
| ML19309E865 | |
| Person / Time | |
|---|---|
| Site: | Indian Point, Zion  |
| Issue date: | 01/15/1980 |
| From: | Taylor M NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Rowsome F NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| Shared Package | |
| ML19309E842 | List: |
| References | |
| FOIA-80-186 NUDOCS 8004240518 | |
| Download: ML19309E865 (17) | |
Text
-
>m usg[*g UNITED STATES 5.g%gyN.,E wff a NUCLEAR REGULATORY COMMISSION ENCLOSURE 5 WASHINGTON, o. C. 20555 s wy/
a
%v
.AN 15;ggg J
/
MEMORANDUM FOR: Frank d. Rowsome, Deputy Director Probabilistic Analysis Staff, RES TERU:
Gordon E. Edison, Head b
Syste=s Engineering Section Probabilistic Analysis Staff, RES FRCM:
Mat A. Taylor Probabilistic Analysis Staff, RES l
SUEJECT:
" EVENT V" - ZION AND INDIAN POINT UNITS Enclosed are some perspectives about the " Event V" probabilities in the Zion and IP units. These are in accord with your recent request for a quick look at the " Event V"* situation in these plants.
For the Zion Units, the probabilistic estimates herein suggest that a ratchet to provide for testing of the LPI/RER check valves vould be an app cpriate one to initiate to reduce the everall accident risks.
For the Indian Point units, the rather unique ECCS configuration appears, on the surface, to prc.-ide a so=ewhat more favorable situation regarding
" Event ?".
This is so because the more probable break location should i
be i= side containment as opposed to outside as envisioned by the " Event V" sce:ario. However, it appears,that either break locatien could negate I
the LPR-EPR function. For the = ore probable break inside containment, l
it appears that the RER heat exchangers could be da= aged; thereby risking j
da: age to the ACS (Auxiliary Coolant System) which provides support ceoling to =any of the ESF pumps. Based en a cursory look, there vould also appear to be several 1_prevements that ight be initiated to i= prove on thn ability to cope with such an interfacing system LOCA and to i= prove on some single point vulnerabilities in the ECCS.
Such i= prove =ents are, bevever, beyond the present assignment and are best incorporated into IRIP. The more direct way to L: prove against such an interfacing syste= LOCA in the IP design would be to require testing to assure that the LPI/RER check valves are in fact in-place and functioning as pressure isolation barriers throughout the majority of ti== that the plant is at pressure and power cperations.
Testing could significantly reduce the probability of the interfacing syste= LOCA along with its potential for EST interactions; thus, a " ratchet" for testi=g is also recc== ended as being appropriate to the IP units.
e 9
9 9
e 1
D hM0
Frank H. Revsome '
30th the Zion and IP units have a 2 check valve configuration in the l
LPI/RER discharge lines similar to that of the WASH-1400 FWR design,.
i The IP units each have four legs with this configuration; each Zion unit has rvo and the WASH-1400 FWR design had three, otherwise the interfacing system LOCA probabilities would be quite comparable-with tge IP units being at a somewhat hieher erobability level (e.e..
6x10 /RT based on a 5 year averaggL The enclosed information addresses a Zion unit vs. the WASH-1400 PWR i
design in some detail. This information can also be used to infer the
[
i interfacing system LOCA probabilities for the IP 2 units.
Enclosures Figure 1 illustrates the two kinds of configurations being used in the Zion LPI/RER discharge lines. Note that the cold leg LPI lines include
[
a 3 check valve configuration whereas the discharge lines to the hot legs contain a 2 check valve configuration similar to the WASH-1400 PWR d es ign. It is this latter configuration that represents the do=inant pathway for the " Event V" scenario in Zion and, of course, should receive the priority attention regarding installation of test provisions.
Figures 3 and 3.1 illustrates the IP 3 LPI/RER configurations as reflected i
in the FSAR. Based on a more detailed print obtained from DOR, IP 2 i
appears to be very similar.
l l
Table 1 provides a comparison to the WASH-1400 estimate on the probability of " Event V" (i.e., probability averaged over a 5 year exposure interval).
Figure 2 illustrates the accumulation of exposure to " Event V" in a Zion unit assc=ing that no specific tests vould be undertaken over a 40 year reactor lifetime.
Based on the r.rrent time in operation, each of,ghe l
Zics units have accumulated an exposure to " Event V" of roughly 10 (this is close to the average probability /RY over 40 years).
Figure 2 also allows one to esti= ate other exposure assumptions until such time a test program might be i=plemented.
Based on internal discussions (DOR et al) and from looking at detailed j,
prints on the LPI/RER system, it is understood that no specific LPI/RER -
pressure isolation barrier tests are being perfor=ed at either the Zion or the IP units.
Pending any added thoughts /suggestiens fres you, I v111 plan on contacting the Zion and IP people (via DOR) to confirm or deny this understanding.
t Te s ting Table 1 also shows that i=plementation of tasting can have an i=portant value in reducing the probability of " Event V" at Zion.
The degree to which the p cbability of " Event V" should be reduced remains an unresolved f
t
=
i
~.-
Track H. Revsome question. An answer to this vould, or course, help guide the kinds of check valve testing to be sought.
For exa=ple, the folleving kinds of testing might be sought--these are listed in an order of preference that reflects their potential value in reducing the " Event V" probability:
1.
Continuous monitoring and readout capability for status of check valve position.
2.
Tests conducted to assure check valves are seated in-place after every known disturbance of the valves (e.g., tests after any disturbance caused by the LPI/RER flow following cold shutdowns or perhaps an unanticipated cooldown event where EPI flow disturbances =ay occur.
Typically, this might require on the order of 1/2 dozen tests per year that would be keyed to the actual plant operations).
3.
Tests sinilar to #2 above but conducted on a monthly basis.
(This vould not be keyed to plant operations.)
4 Tests similar to #2 above but conducted on half yearly or yearly basis..
5.
Periodic tests (e.g., yearly) conducted for the specific purpose of leak rate measurements (as may eventually be required by the 5RC) for ce=pliance with ASMI In-Service Inspection requirement..
(Conduct of these tests, as may eventually be required for all operating reactors, could serve as #4 above providing the tests are directed teward assuring the barriers re=ain in-place / seated and are not disturbed soon after the tests and prior to ascension in pressure and power. In other words, " heroic" measures to verify or de=onstrate existence of very lov valve leak rates are not required for purpose of reducing the " Event V" probability and such " heroic" measures =ay serve to unnecessarily increase occupational exposure levels.)
I trust that the above perspectives meet tBe intent and schedule of your recent note regarding the " Event V" scenarios in the Zion and IP units.
Please keep =e advised of any further needs or initiatives undertaken in this matter.
i Mat Taylor Probabilistic Ana*.ysis Staff Office of Nuclear Regulatory Research l
l Inclosures:
As Stated ec:
R. Bernero, PAS T. Murley, RSR L. Tong, RSR S. Fabic, RSR C. Relber, RSR r
V. Nerses, DOR J. Murphy, PAS W. Vesely, PAS l
J. Carry, PAS l
l e
- 6 m *
-Wh e e 'em e
- e. m e a e
se
.e e o
e
~
DRAFT RBlond 02/01/80 The following engineering insights from the Reactor Safety Study are pertinent to an appreciation of reactor accident risks and indicate areas where improve-ments could be made in reactor power plant design and operation:
1.
Core melt accidents dominate risk; 2.
Small-LOCAs, transients are the dominant accident initiators; 3.
Dcminant system failure modes entail human errors, test / maintenance outages, and system interactions--particularly unreviewed inter-dependencies among systems; 4.
We have assumed (perhaps conservatively) that all core melt accidents will ultimately lead to centainment failure.
These findings could potentially have a major impact on the regulatory process since c6rrent practices do not focus in these areas as they are limited to considerations dictated by design basis accidents.
Another very important finding of the Reactor Safety Study was that core melt accidents would not necessarily lead to catastrophic offsite consequences.
In fact, the most probable consequences from a core melt accident at Indian Point or Zion (large dry containment) would not be conspicucusly greater than the consecuences from the accident at Three Mile Island-2 (except from possible liquid pathways).
Since all core melt accidents are believed to lead eventually to containment failure, the containment failure mcde and prevailing weather will determine the magnitude of the public consequences.
If, during a core melt accident, the containment failed directly to the atmosphere.(i.e., by steam explosion,
-2 overpressurization, hydrogen explosion, or inadequate isolation) a substantial amount of the core inventory would be released, resulting in substantial offsite consequences.
Even with these circumstances, there is less than a 5 percent chance of early fatalities at Indian Point or Zion. That is, it takes the worst release categories coupled with very bad meteorological conditions (e.g., precipitation) to get offsite whole body doses above 300 rem. However, given these conditions, an accident at Indian Point or Zion will as likely kill thousands of people as kill one.
It is this point that sets these two sites apart from other sites.
If, on the other hand, the containment failed via a meltthrough of the basemat and depressurizes through the earth or a man-made filter, the offsite consequences would tend to be much smaller.
To place the problem in proper perspective, the relative importance of the containment failure mechanisms to risk is given in figure 1.
This figure is taken from NUREG/CR-165, "A Value-Impact Assessment of Alternate Containment Concepts." As was stated in that report, the interfacing systems LOCA (check valve rupture (Event V)) sequence and containment overpressurization failures dominate the risk in the Pressurized Water Reactor.* Hydrogen burning and vessel steam explosions have a minor contribution to the risk.* Therefore, reducing the probability of over-pressurization by filtered-vented containment in and of itself will do
' Assuming a PWR aesign similar to VEPCO's Surry Virginia reactor (large dry containment).
. little in reducing accident risk, unless steps are also taken to reduce the probability of Event V (see attached memorandum from M. Taylor to Frank Rowsome).
Indian Point and Zion Figure 2 presents a range of site specific risk curves for Early Fatalities.
This figure was taken from SAND 78-0556, "An Investigation of the Adequacy of the Composite Population Distributions Used in the Reactor Safety Study."
The Indian Point and Zion sites clearly dominate the early fatality risk curves as compared to the other sites presented.
These curves' in'clude the contribution to early fatalities from all of the WASH-1400 PWR accident release categories. The accident release categories consist of the dominant accident sequences having similar release character-istics. To account for the chance of an accident sequence having different release characteristics, a smoothing technique was used. Ten percent of the accident probability was assigned to the adjacent categories. Although it was felt that the smoothing technique would account for much of the uncer-tainty in the release categories, in order to identify the containment failure mode contributors, it is necessary to eliminate the probability smoothing between categories.
Figures 3 and 4 show the early. fatality curves without smoothing for Indian Point and Zion, respectively.
In addition, these figures present the curves for the contributions to the total from each of the containment failure modes. The interfacing system LOCA (check valve) contributes about 60 percent to the total. Overpressurization containment l
l
o F
. failure contributes about 30 percent, followed by hydrogen burning and vessel steam explosion with about a 10 percent ccmbined contribution.
Once again, it should be pointed out that these curves assume a Surry type PWR design at Indian Point and Zion. To perform the analysis properly, the specific systems interactions for the Indian Point and Zion designs should be factored into the problem. However, for an initial cut, it is not anticipated that the design differences would substantially change the results.
i Figure 5 presents a comparison of the total unsmoothed risk versus the expected risk reduction from filtered-vented containment and assuming a reduced probability for the interfacing systems LOCA for both Indian Point and Zion. The assumptions used in generating this figure are the same assumptions used in WASH-1400, but the release categories do not include smoothed probabilities.
In addition, the pressure vessel steam explosion containment failure mode has been given a much smaller probability of occurrence than in WASH-1400; this is due to a better understanding of the phencmena since the completion of the study. The vent'ed-filter is assumed 100 percent efficient, to be able to accommodate the effects on hydrogen burning and the interfacing system LOCA probability is reduced by a factor.
of 5.
If more credit can be taken to reduce the interfacing system LOCA, 22n the curves can be directly shifted to reflect the expected improvements.
By reducing the probability of Event V and the overpressurization contain-i ment failure modes, the risk from Indian Point and Zion can be reduced at least an order of magnitude. Thus, by making these fixes, the societal S
I
risk of early fatalities from Indian Point and Zion sites becomes much more similar to the societal risks from early fatalities of the other nuclear power plant sites.
9 6
cm
in W
-H
~
W
]
W C
W H
p
<w l
2 O
M H<
3 W
Q 4
O H
A Z
Z w
D Q
A
~
m H
O
~
Z m
W w
A o
DCM t
1 0
ZZww
>>ZZww
- ZZWw 6
O ooag UUo Oc ooze 3
u ww
- oag W w m-p3
-g3 wW- -32 O
m ax w a
mm mm mm-OCad aa O O a.O d
Q O.s a < s<
--O
~-oo--
g
-OOaJ <<
a<<
Z A 4 x
wZggww w Z A a. w w w ~ *A g
xxZZ g
xxgh
=yw522 e
E 3www 9 9-Og WW o e
c a < w<
- 5 sm22cc m 3 <*
- 9 0 m2 33
=
e<<
a 3
mww55 m
w g m *w
- O O w a * *w
.J w
w
~
z wk 8-Co g
w OO 4
I
- a. w WW m g
m m mWmm
(
M Wom g g m
g g
~ ~
~
- o..J >
>gwgaa w "A a w " 8 w *A d
- w w " w " d"'
- Z
C W Z
>W*
W'd d d d O W " 2 *3 h W
Q>*W m
W O
>m W
W O>W C > 3 s. W 2
z >.3 Q
C*E>
OW 33 D
> Exp w
<mw
<=w
<Tw
-o5
- Og g-
- O3 O
c",
2 Z
y Z
C O
O U
O U
H u
W c
~
in
.J 0
C5 w
?
H
~
h O
a 4
T 4
Q H<
~
g s
4 o
h C
Z M
g W
- U J
Q A
e O
H m
I.aJ A
~
$3 a
~
7_
n o
f 4
2 wwoZZww
- "O22Ww W W c2Zww E9 I
3
- Z Ocza kh=ZOczg b3 Z e $ z $.
33- -333 EG333 N.
ENhC= Odd CO e
-m 0--
A m=O
- a. m mw2 w g=g A=ww 2 w a *a z < <
U *x a "da. 3 w# w" O
3 E
- 3 a
A x>Zk a.
x >- 2,_
A xwg%
wz2 w*Oz wgzw[9w Z
W z2w w9wg
> ww D w w
>w
>co<dcW w
a
>cI
>c2 a > c2w d
d
<W >2 sf 2
w 2 a< z 4
3 E
$O0j3 oo$
g
- z e
OE o<
o <-.
t O,m O m t
om
$~
u
>d Z
8 Ca an UD a
2 a
Z Z
N EC Zw w
w W
w
~
a 2 0
-w O
m Z
u U
m wQ U
W U
u a
w w
20 W
2o y
20 EZ Z
5Z 1
a z
Z Z
0 O
O o
u U
l l
R?NGE OF SPECIFIC SITE PROBAD!t.! TIES
~8 10 2
t p1
-7 to y
m
<e Oe
-8 4
10 5
2 10,
+0
+1
+2
+3
+4
+5 to 10 to 10 to 10 EARLY FATALITIES Figure 2:
Log-Log plot of probability (per reactor-year versus early fatalities showing the dispersion of site specific CCDF's about the Reactor Safety Study CCDF.
1 = Indian Point (2985 MWt) 2 =
lon (3150 MWt) 3 = Palo verde'(3713 MWt) 4 = Millstone SWR (1956 MWt) 3 = San Onofre (1290 MWt)
+ = Reactor Safety Study 4
.e e. man
n t
n 0
1 NO I
S N
O O
L I
P T
x 0
U E
I 1
B i
N I
R M
G h
R N
U A
T B
E I
N H
T O
T N S C
0 E
0 G
L E
H O
E D
R S
S
- 0 D
S O
S H
g T
1 E
Y E
U H V E
N I
O T
R l
H 3
i I
U T
L L
I A
I W
T A
E A
F E
R L
E F
V U
A T
L S
T 0Y 1
A S
l I
O i
l 1
L 0
E V
E T
R P
M R
(
A a
K P
E l
l l
c R
A A
E E
I T
V i
i D
N C O l
O l
I C 1 2 0
8 g
1 3
ER U
G I
3 F
2_
1 0
0 1
7 O
'0 0
0 1
1 1
1 0S0p2m E
e l1I I
1 l
li i
i
+ g %g,
- 3, EYb %y?{h
$4 D
,9
/ g \\xNN'(
/e[4 4-
\\\\*NN77 if^
4);
\\///
c <
49 y
gff YW,
%h 4
,M ee Ev <e 1,em TEST TARGET (MT-3)
I 1.0
!!a HM W M Bu-
,; as t 22 lllll=20
\\
l-l L
\\
=
I.8 1.25 IA 1.6 h
4 6"
y MICROCOPY RESOLUTION TEST CHART 4%
+ //p A g-5,,,, xx f '
S 4
4Ah4 y,
. s <y
/
s
.L 1
-6 10 Lo i8 TOTAL WITHOUT SMOOTHING
~
P 2
r
-a
@g
-8 3
jg 5
Ig
~
I
-9 4
10 I
I I
l I
0 t
2 3
4 5
10
,g
,g go 10 10 EARLY FATALITIES FIGURE 11 Z10R C0itTAllliiENT FAILURE MODE CONTRIBUTI0:1 1 tilECK VALVE 3 IlYDROGEN BURN:'
2 OVERPRESSURE Il VESSEL STEAM EXPLOS10N
~
I 1
i l
l i
j P
i i
t T
I m
C i
1 ez i
-=
l--o or m
o
- =.
a$
-x c.
s-z o
o g
o y,
w a_
s.
25 z
a N
oz o
w i
N 1-o o
e s >-
=
a m
._z.
. >o e
o e
a-i--
e w
3 A
y r
n o
=e tz o
z e
j z;
c-o z
.y o
zo.<
P oo o'
"E o
LS Nm
=
l w
em e
r 1
C Lt.
C e,
n, e.
m.
O C
O C
(2033) S3111IIEVEC'dd
~
R. Slond, RES/ PAS 02/07/80 By usfc-e insights from the Reactor Safety Study and recognizing the s
attendant uncertainties of the methodologies and techniques, an initial attempt was made to determine the relati' a societal risk posed by Indian Point Units 2 and 3 and Zion Units 1 and 2 as compared to the total population of reactor sites. The basis for the evaluation comes i
primarily from analyses performed for the NRC Office of Nuclear Regulatory Research and include the following:
MITNE-NUREG-205, A Method for Risk Analysis of Nuclear Reactor Accidents, M. Maekawa, January 1978; and SANDA78-0556, An Investication of the Adecuacy of the Comcosite Poculation Distributions Used in Reactor Safety Study, J. Spring, October 1978; and from insights gained from the staff's own analyses.
The most important assumption made throughout the analyses is that Indian Point i
Units 2 and 3 and Zion Units 1 and 2 can be adequately represented by VEPCO's Surry Unit design.
It was found by the analysis that Indian Point 'and Zion represents a signifi-cant portion of the total societal risk frcm reactor accidents.
Three measures were used to estimate the societal risk. They are early fatalities; latent cancer fatalities; and property damage.
1.
Eariy Fatalities Early fatailities are caused by very high doses (above 300 rem whole body). Such doses generally would be limited to the exposed population within about 10 miles of the reactor. The 6
O g.
g R. Blond, RES/ PAS 02/07/80 analysis estimates that Indian Point and Zion represent about 30 percent of the total early fataility risk due to their unusually large populations within about 10 miles.
2'.
Latent Cancer Fatalities The number of latent cancers fatalities is determined by the total population dose (person-rem). The major determinant of population dose is the total number of people living within about 250 miles of the reactor. At such large distances, the populations at most reactor sites tend to look very similar.
Although Indian Point and Zion rank as having greater popu-lations within 250 miles than most other sites, the societal risk posed from latent cancer fatalities by Indian Point and Zion, probably represent less than 10 percent of the total from all reactors.
3.
Procerty Damage Finally, the risk for property damage would be dominated by the contaminated areas where people would have to be relocated and the land interdicted or decontaminated. Such areas could probably extend out to as much as 50 miles from a reactor.
Indian Point and Zion represent a substantial portion of the risk from I
property damage due to their proximity to very large population centers.
It is estimated that these plants represent more than 30 percent of the total property damage risk.
e
_