ML20052F320
| ML20052F320 | |
| Person / Time | |
|---|---|
| Issue date: | 05/07/1981 |
| From: | NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Bernero R NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| Shared Package | |
| ML19219B002 | List: |
| References | |
| FOIA-81-417 NUDOCS 8205120345 | |
| Download: ML20052F320 (12) | |
Text
.
/0y/cr
'e UNITED STATES
[ h.,q, ',7.
NUCLEAR REGULATORY COMMISSION 3
waswie.oroes. o. c. aones
,- P / e
\\;.M.... /
May 7, 1981 MEMORANDUM FOR:
Robert B. Minogue, Director Office of Nuclear Regulatory Research FROM:
Robert M. Bernero, Director Division of Risk Analysis (DRA)
SUBJECT:
AE00 REPORT: SAFETY CONCERNS ASSOCIATED WITH PIPE BREAKS IN THE BWR SCRAM SYSTEM, MARCH 1981 RE: (E00-10372) 001337 Commissioner Ahearne's meno of April 8, 1981 to W. 01rcks, EDO asked for PAS (now DRA) comments on the subject report. In his meno Commissioner Ahearne indicated that the seriousness of the problem is significantly affected by the probabilities oescribed in the subject AE00 report. We offer the following comments and observations on the subject safety concern.
1.
AE00 correctly acknowledges the r.omplexities and difficulties involved in balancing the need for a reliable BWR scram system design against perceived needs for reliable isolation of containment and the primary coolant system for postulated pipe ruptures.
We strongly urge that care be taken not to lessen the reliability of the scram system because of the scram system LOCA toncern.
2.
Probabilistic estimates that have been made previoutly by others and ourselves regarding unavailability of scram; and the ATWS initiated core damage events appear to be comparable to or larger than the AE00 proba-bilistic estimates (i.e.,10 5/RY) for ccre damage from unisolated piping ruptures in the scram volume discharge system.
Because the ATWS initiated core damage events (and precursors to an ATWS event have already been indicated) have been estimated to result in some of the largest off-site consequences, we believe that the main NRC safety emphasis should remain on assuring reliable shutdown means for the BWR if risk reduction improvements are to be sought. Said yet another way; our present judgment is that the overall BWR risk is dominated by ATWS initiated events and is not importantly impacted by the scram system LOCA concern.
3.
The AE00 absolute estimate of 10.s/RY as the procability for core uncovery is not and should not be taken as a statement of public risk.
i Our Consequence esticates would also be needed for a risk statement.
present judgment is that the off-site consequences given the AE00 core damage scenario would be less than those previously predicted for the BWR l
design by perhaps a factor of 5 or more.
I l
f 3205go345g11103 l
hNAP 417 PDR l
i L
R. B. Minogue 2
it should be noted that action guidelines 2 t
Regarding the 10 5/RY estimate we have previously suggested to NRR in connection with possible backfit decisions would indicate that no precipitous actions on the postulated AE00 scenario would be required.
4.
The present configuration of the BWR scram volume piping and valves as depicted by the AE00 report appears to clearly violate the intended isolation valve redundancy requirements set forth by GDC 54 and 55 of Appendix A to 10 CFR 50. This factor alone - absolute estimates on probt.bility notwithstanding - appears to warrant NRC actic9sa to either (1) ensure compliance to the GDC of 10 CFR 50 cc, (2) change applicable regulations or, (3) grant a specific BWR exemption on defined bases and justifications.
We recommend the latter course of action and the adoption of the postulated scram piping rupture as a design basis event to be assessed in each BWR design. Those actions recommended
- Dy AE00 that would lead to increased piping surveillance, improved diagnostics, procedures and operator training for correct response and, as needed, additional environmental protection of equipment should suffice to assure a continuing low level of risk from accidents involving failure of the scram volume discharge system piping.
5.
The AE00 estimate of 2 scram challenges would appear non-conservative with respect to available operational data.s,a From NUREG-0020 an average of about 12 forced outages per year can be seen to be applicable to U. S.
LWRS. It is the practice of the operating BWRS to manually trip the RPS (for scram test purposes) for those outages not involving an automatically actuated trip. Assuming about 2-3 planned outages per year plus acout 12 forced outages, the challenge rate (or scram loading) on the scram discharge piping outboard of the BWR scram valves would be, on average, about 15 challenges 1 Commission policy issued subsequent to recommendations of the Risk Assessment Review Group Report (Lewis Committee) advised staff against high reliance on absolute probabilistic arguments.
8See R. Bernero memo of 7/22/80 to R. Mattson indicating that potential core damage accidents in the range of 10
- to 10 5/RY may require no backfit actions.
8Here it should be recalled that plant shutdowns were required by NRC when inadequacies in seismic design of piping were suspected - a situation of perhaps lesser risk or legal significance.
- We refer to recommendations #1 through #5, pgs. 21-24 in the subject AE00 report.
Recommendation #6 pertains largely to future plants and care should be taken not to substantially increase the chances of pipe rupture via increased quantity of piping.
sSee NUREG-0020 (Gray Book) 8See NUREG-0460 on ATWS matters and related EPRI Report with data on anticipated transients experienced in BWRS and PWRS.
1 l
R. 8. Minogue 3
per year. From NUREG-0020, U. 5. 8WRS have accumulated about 225 calendar years of experience. Worldwide the total BWR experience is about 400 calendar years. Assuming an average BWR plant availability of
- 65%, the actual time of operation (at or above critical opsration when the scram discharge piping w uld be subject to dynamic scram loads) is about (0.65)
(400) : 260 reactor years at power. This translates to about 3900 dyn mic challenges that have already been experienced by the scram discharge ofping outboard of the scram valves. Thus far we are unaware of any cracking c,e rupture evidence reported for the BWR scram discharge piping.7 From this we could reasonably infer with about 95% confidence (chi-square) that the frequency of cracked scram discharge piping would be
- 8 x 10 4 or less per scras loading. Similarily the 50% confidence level would be about 3 x 10-4 or less per scras loading.
6.
Vesely and Bushe have reviewed infonnation gained through actual nuclear plant operations for pipe rupture precursors. They have used this information for estimation of pipe rupture probabilities. Their assessments indicate that for the ductile piping systas in use in the nuclear industry, evidence of cracking before rupture (i.e., leak before break) exists. The probability of rupture appears to te lower than the probability of cracking by a factor of 101 to 10 2 Bush has used 5 x 10 2 (i.e.,PRUPT : 5 x 10 2 PCRACK). Vesely suggests
- 3 x 10 2 as being representative of a 50 percentils value.8 Considering the above, a best estimate probability'of a pipe rupture out-board of.the scram valves per scram loading would be:
P 9 50% : 3 x 10 2 PCRACK : 3 x 10 2 (3 x 10 4) : 9 x 10.e RUPT A reasonable upper bound estimate for rupture would be roughly 3 x 10 5 per scram loading.
' Note: It seems reasonable to assume that NRC would have gathered such evidt. ice if it existed in light of the considerable pipe crack experience of the BWR c,;
the regulatory focus on such cracking experience.
'See Paper IAEA-SN-218/11, Reliability of Piping in Light Water Reactors, S. W. Bush, USNRC, ACRS, Washington, DC, USA (Inte 9ational Symposium on Applications of Reliability Technology to Nuclear Power Plants, Vienna 10-13, October 1977).
ePersonal communications with W. Vesely, ORA /RES.
em pe..
fM
R. B. Minogue 4
Taken together with the BWR scram challenges, a best estimate on the probability for rupture of piping outboard of the scram valves would be 15 (9 x 10.s) :: 1.4 x 10
- per reactor year. This estimate is not inconsistent with AE00's (i.e., 2 x 10 *) although our approach differs somewhat.18 It is possible that a lower probability for pipe rupture exceeding the sump pump capacity would exist. However, data are not sufficient to enable further refinements in this regard.
It is also unlikely that the number of scram challenges would vary by more than a factor of about 2 (from data inference and because of plant economics) so that a reasonable upper bound assignment for such'a piping rupture should be
- 3 x 10 4/RY.
7.
The probability of failure to cope (Pfailure) given a rupture in to cope the scram discharge piping outboard of the scram isolation valves is complex and dependent on plant design and layout; the operator diagnostics training, and procedures; the nature of events causing the scram; the environmental interactions (steam, flooding) to equipment caused by tts rupture; operability of the scram valves in the adverse environment and the various alternative ways available at the particular plant to carry-but the necessary coolant makeup and decay heat removal functions. AE00's overall estimate on the probability of failure to cope given the scram piping rupture was
- 5 x 10 2 Our experience in general, indicstes that the BWR design has many alternate ways of providing for the coolant makeup and the heat removal function though use of either safety or non-safety related equipment. Past predictions 1 stemming from the very severe fire 1
at Browns Ferry Nuclear Plant (BFNP) #1 that caused loss of essentially a_ll safety related equipment for coolant makeup and heat removal suggests to us that the AE00 predictions may be Quite conservativo, oerhaos by as much as an order of macnitude or more. Tne past predictions were cerived through a systematic application or the event / fault tree logic techniques which time (and lack of information detail) has not permitted us to do tere. On the other hand, if our past predictions are taken to be reasonably representative of a best estimate on the probability of failing to cope with a rupture of the scram piping, then the overall probability of core uncovery would be 181t should be noted that the traditional Regulatory thinking about scenarios that are to be treated as design basis accidents (DBAs) has been that those scenarios at or above a probability of $10 */RY should become a (DBA). On this basis, the postulated scram discharge volume piping rupture (outboard of scram valves) would appear to be marginal as a CBA choice.
ssSee Appendix XI to WASH 1400 pgs. XI 3-50 through 3-58.
The conditional pro-bability of core damage given the common caused failures by the severe fire experienced March 1975 at BFNP #1, was estimated to be approximately 1/300 vs.
the AE00 estimate of 1/20 given the rupture of scram piping.
a i
)
i 5
approximately:
P,9 3,3 1.4 x 10 4 (3 x 10 s)
=
P P,,,
f g
rupture uncovery of to scran piping cope
- 4 x 10 7/RY number of causes is not a novel concern to NRC.tz,tal equf;oent by a 8.
The possibility of common interactions with vits Equipment quali-fications for various steam and water environments caused by LOCA and by high energy pipe ruptures outside of containment have been of concern to NRC for some time. The evidence available to date suggests that even unprotected elactrical equipment in the BWR reactor building should have a reasonaole chance of surviving the high humidity conditions likely to result from a ruptura of scram piping. For example Sandia information
- 1 on unprotected electrical terminals (in a 100% humidity environment for 5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> duration) indicates a average probability of breakdown failure in the range of 101 to 10 2 (For protected terminals the probability of breakdown failure was improved by rougnly an order of ragnitude or more.)
Not all terminals in the reactor building would be exposed to the same severity of environment, nor would all terminals at the same severity of environment be expected to fall at the same time. (For example inferences from other data indicate that a 10-20% chance may exist for the simulta-neous failure of similar ccaponents in a redundant system exposed to a common environment.) In other words the presence of a common environment would be expected to accelerate failure of a component but a rancaness in the failure times would be expected. The net effect of this randemness would be a gain in time for operation of makeup and heat removal systems and in turn a gain in time for taking corrective or alternative actions.
From the above information, we would also consider AE0G's estimate of failure probabilities for various components subject to the pipe rupture environment to be conservative by perhaps an order of magnitude This further reinforces our overall judgment that the AE00 or more.
estimates on the probability of failing to cope with a scram piping l
rupture is conservative as is the overall probability estimated for core i
t uncovery.
18See IE Circular Bulletin #78-06 (5-23-78) transmitted to all holders of OLs and cps.
Subject:
Potential Common Mode Flooding of ECCS equipment rooms at BWR facilities.
s See NUREG-0588, " Interim Staff Position on Environmental Qualification of Safety Related Electrical Equipment" - notices in Federal Register, Dec.1979.
(Pertains to environmental qualification of equipment.inside and outside containment.)
7
- See NUREG/CR-1682 (Sand 80-1957) Electrical Insulators in a Reactor Accident Environment, Jan.1981 Sandia National Lacoratories. 04ote: Results are depicted and discussed on pgs. 54-59 for breakdown of protected and unpro-tected electrical terminals for 100% humidity and 5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> expJsure duration.
f
--,,.,-r, y..,
y,_
y. - :- gM.
~g-4W
. 4>-.
w-R. S. Minogue 6
In summary, we consider that the AE00 estimate for the probability of core uncovery (i.e.,10.s/RY) from a rupture of the scram discharge piping is conservative by perhaps an order of magnitude or more. Even if the AE00 probabilistic estimate was taken to be a correct absolute value for the postulated accident, we believe it would not represent a dominant contributor to risk. Notwithstanding the reassurance provided by these estimates of absolute probability, the present configuration of the BWR scram discharge piping and valves appears to violate the intended isolation requirements set forth under GDC 54 and 55 of 10 CFA 50. This latter fact seemingly warrants NRC actions to ensure compliance with GDC or to grant an exemption on a defined basis. However, on the basis of probabilistic' and risk based arguments there appears to be no need for precipitous action in this regard.
eoriginal515'J A.I'*-
Robert M. Bernero, Director Division of Risk Analysis DISTRIBUTION:
Central File RES CIRC /CHR0rl RMBernero FHRowso ne MATaylor a
P dbf kil!
lY en.c. p,DRA:RRB CRA:,RRB, D,RA,: D D.,,
,DRA:Dn meme..mmm...mmm..
..m.. m..
.x.GB,...
.u uB.,....
.ukm...
..u..zw...
~+
=c om m e.io. oi.cu ca o OFFICIAL RECORD COPY
" " =-m*-
VERSUS DISTANCE GIVEN* VARIOUS CO SES E
~
s'( {TC,AEY RESASELINE
\\
[N 5 EY 5 Er
'1 b
f 10-3 3
w TQUVy'
~~
10-4 AEa, 5 Ea (DF=100)-
~ AEy' 3
5E' g:
~
2,
~
i i
g 5
10-5
(
<O 4
'Y
~
=
4s 5
,q g.,, u W u.a A
'/:au4 J/' 7.~"',
N 9,,, a C.
., L u~ f.
.b A' INN b
_c.c
... 7 6
g,,
10-J 2.5 2.0 1.5 -
U 1.0 10 0.5 0
DISTANCE (MILES)
TWy-s E PROBASILITY) ASSIMED s
ASSUMPTIONS: 0
- RELEASE PROBASILITY (f.e. SEQUE TO BE UNIlY PER REACTOR YEARARCH-CORRAL I
O RELEASE MAGNITUDES ESTIMATED FROM USE l
CODES QUENCES WITH O EXPECTED CONSEQUENCES FRGt 91 WEATHER SE
/
i 3200 MWT POWER LEVEL POSURE 0 ENTIRE CLOUD EXPOSURE + 24 HOUR GROUND F SHIELDING BASED ON NORMAL ACTIVITY I
v l
i
~
q RISK 0F EARLY FATALITY TO AN INDIVIDUAL VERSUS DISTANCE FOR VARIOUS CORE MELT SEQUENCES 10'3 TCy
~
\\
RSS-BWR DESIGN REBASELINE
%w.
N D
O'4 g b
TCy'
=
2 T
b 2
ss
- TQUVy W s
n#
N
- AE i
.~4
~
q.
~
TWy'
/
/
N 10-5 i,
=
C Z
3
_c y
_9a f., a L as:i. s la L.. e vc s=r : ~,.
n
,g. 7
.e a.
- i;s a n-.
- I 10-6
=
2 AEs
_~
\\
S E.
j
\\
N 3
10~7 -
7 2
I
\\
\\
2
\\
EFFECT OF POOL DF=100 g
I
-8 I
O 0.5 1.0 1.5 2.0 2.5 DISTANCE (MILES)
ASSUMPTIONS: O OVERALL CORE MELT PROBABILITY ASSUMED TO BE 10~4/REACTCR YEAR O SMOOTHING ELIMINATED, EXPLICIT ACCIDENT SEQUENCES l
1 O EXPECTED CONSEQUENCES FROM 91 WEATHER SEQUENCES WITH 3200 MWT POWER LEVEL O ENTIRE CLOUD EXPOSURE + 24 HOUR GROUND EXPOSURE SHIELDING BASED ON NORMAL ACTIVITY i
5 i
l I
1
-j,g
~-
s.
e.
o 11 q'
~.
s.
m.
h @ ><
o e
X di t
4 ts 9
R u
m y
[41%h@u! R 3
Y-
't i
l b
We l
g g
4 q
9t v N
AR Ng
.s 1
% #f M 9
Akik }, !
c'
- 3 7e is o" a
j l
f?
w
{y4 1w3 lI) h 4
y ea t
q u 3!
}ef
$*[d E N
!i $kh1 il n-w e
s 9
$$D 4 %
4 $1 s
@ ti di 0
j
i l
poQir,4ce,,9}
FAicu,w
,43
@ Au. V#
f LW
) S c t. A T A Oca en f oAJO so11 emnrW.M
/
6 i
i r'ditu w 's SoLA1 l
FAet.c?o P4IDwr va C.
CCMt r4trAJ-r[.13y FAtsus4 saJ e
9h%s4sv CoMfa
%'l osuseo** <rM4. vers I
ter7A m Oe GGeM sci'a AM YAs. V(C\\
PMQ T*64e l
SCV Neur f v As.v es D< A eas j
l ^
f
, pontG& pptN g 4.
e4 FAsGuidt'~ \\ofS(TAM fSC-***
f crew
?
c4 LaMi!
-rus e.
,,, l I
N uee,a 5"U,r r 3
Von
^
? A3W04 PJ W'MG sW ue %Ost \\
vwwa v
oS 50v M. AO Git,I 34M
'STE'*'
mg S'qp
,4 svo i
8" S4 ro iSDgg vmGuas 300 i 98FiM4 r vacvc yog
\\gvne o.r %ardu ve4r V
NO SO Ge r
A Sov reo Nd A Odd E ScR Am
+
'V A4Vit$
I y
os eges i-
.,~,t g 8' 8
-rseM t
8 Le M C '
yA,<mo.
g$. mom UNd o
o4. i sw "i"h A M i i
/.
t i Je -
A t
k I
i l
i r
l I
f f
i r
.,. _ - 1 2
_C9h,W-Ms.#
a e_.
1 F
&s 4a aa 0JS1d 4
aa4 s
h s w s g D* W n" ts e e en er e
naa kh sksxKs
-I E
E E
9 N 2
1 0 l 2
I b
i e
W*s O
i 4
W s
N
.t p
.kk d
a 4*
N.k l
2d
}.
w 34
- y
{g
}.I E
x i
h k" iS S
w 3
s-4 a %*
2 g
f f
4 QnN
't R
4s a
a j
b )l e
e e
s 8
9
)