ML20203N084

Summary of ACRS Extreme External Phenomena Subcommittee 860806 Meeting in Washington,Dc Re Workshop to Review Importance of Seismic Risk to Nuclear Power Plants

ML20203N084
Person / Time
Issue date:	09/08/1986
From:	Advisory Committee on Reactor Safeguards
To:	Advisory Committee on Reactor Safeguards
References
ACRS-2453, NUDOCS 8609230132
Download: ML20203N084 (32)

Text

. _ _

t B N MRS -Mf3 Olp $d a 4 , t i TPE DATE ISSUED: 9/8/86 g/tf-gf 7 N ACRS Meeting Minutes on the Extreme External Phenomena Subcommittee August 6, 1986 Washington, DC Purpose The ACRS Subcommittee on Extreme External Phenomena met on August 6, 1986 at 1717 H Street, N.W., Washington, DC. The purpose of this meeting was to conduct a workshop to review the importance of seismic risk to nuclear power plants. The emphasis was on seismic hazard. The Subcommittee heard presentations from representatives of the NRC Staff, industry, and the scientific community. The meeting began at 9:00 a.m.

and was adjourned at approximately 7:00 p.m., and was held entirely in open session. The principle attendees were as follows:

ACRS Meeting Participants D. Okrent, Chairman L. Reiter, NRC Staff M. Carbon, Member D. Bernreuter, LLNL J. Ebersole, Member I. Wall, EPRI W. Kerr, Member J. C. Stepp, EPRI H. Lewis, Member D. Perkins, USGS J. C. Mark, Member P. Thenhaus, USGS C. Michelson, Member G. Bollinger, VA Dolytechnic Inst.

D. Moeller, Member C. A. Cornell, Stanford Univ.

G. Reed, Member R. Holt, Weston Geophysical C. P. Siess, Member M. Trifunac, Univ. of So. Calif.

D. Ward, Member D. B. Slemmons, Univ. of Nevada C. Wylie, Member J. Reed, Jack R. Benjamin, Inc.

R. Savio, Staff G. Thompson, Stanford Univ.

Highlights

1. L. Reiter provided an overview of seismic hazard methodology and the NRC approach to defining seismic hazard. Mr. Reiter noted that the problem associated with understanding seismology differs in the 92 2 860908 2453 EEUICIIAIE3 C2ICI"AL PDR C: .: ?;_ f

l P

.+

EEP Meeting Minutes August 6, 1986 Western U.S. and Eastern U.S. Large ecrthquakes occur frequently in the Western U.S. and evidence of the faulting is frequently seen at the surface. There are fewer large earthquakes in the Eastern U.S. and they appear to be related to structures not readily correlated to surface features. The mechanisms for past and potential earthquakes in the Eastern U.S. are presently not well understood and competing scientific theories exists. Very large earthquakes have occurred in the Eastern U.S. and have affected much larger areas that similar magnitude earthquakes in the Western U.S. The NRC criteria described in Part 100 Appendix A and Regu-latory Guide 1.60 were used at the Construction Permit stage for most operating reactors. A number of older reactors for which this methodology was not used were reevaluated in the Systematic Eval-uation Program. Zion was, in addition, analyzed with the method-ology developed in the SSMRP research work. The NRC Staff has more recently been using the " site specific spectrum" methodology to reevaluate selected plants in the Eastern U.S.

Mr. Reiter noted that the site conditions may be very important in accessing the seismic hazard.

2. D. Bernreuter described the recently completed LLNL work (Seismic Hazard Characterization Project) on the characterization of seismic hazard for the Easter U.S. The LLNL methodology is based on the systematic evaluation of a spectrum of . expert judgement. Specif-ically, the LLNL work used two panels (a Seismicity Panel, consist-ing of 11 experts, and a Ground Motion Panel, consisting of 5 experts) to develop a "best estimate" characterization of seismic hazard. The work was an extension of similar work performed for SSMRP and the Systematic Evaluation Program. The final product represents a combined, through not necessarily consensus, represen-tation of the various expert's opinions. The principle insights which LLNL obtained from the work were:

9 I

EEP Meeting Minutes August 6, 1986 (1) The differences between various experts on source area zona-tion and the choice of ground motion model was large, and the differences resulted in a large uncertainty in the LLNL characterization of seismic hazard.

(2) Small earthquakes contribute significantly to the hazard in the LLNL characterization of seismic hazard.

(3) The uncertainties in the characterization of seismic hazard are such that it is unlikely that this uncertainty can be significantly reduced in the near future.

The methodology was used to obtain estimates of the seismic hazard for 10 reactor sites in the Eastern U.S. A sample of the results is shown on Figure 1 of Attachment A.

3. J. C. Stepp described the EPRI methodology for the characterization of seismic hazard. The approach used was similar to that used by LLNL. EPRI, however, used teams whose expertise spanned the full range of issues and developed consensus positions within each team.

l The methodology was used to evaluate the seismic hazard at a number of reactor sites. Uncertainties, as in the LLNL, calculations were very high. There were differences between the LLNL and EPRI estimates, most of which appear to be related to the credibility (i.e., probability that they were correct) associated with the various ground motion models. Mr. Stepp believes, as does LLNL, that uncertainties in seismic zonation are not resolvable in the near future. He does however believe that the ur. certainty in the attenuation models can be very measurably reduced or eliminated in a year or two if proper use is made of available information.

4. P. Thenhaus described recent USGS work in which the consequences of f different interpretations of the Eastern Coast seismic zones were

F EEP Meeting Minutes August 6, 1986 evaluated. The models are derived from the current understanding of Phanerozoic and Precambrian structures and historic seismicity.

The potential source zones are shown on Figures 2 to 15 of Attach-ment A. The figures are arranged in groups, the first figures in a group describes the physical model (basin model, simple hazard model, and uplift model) and the translation into source zones with the following figures showing the peak ground acceleration corresponding to a 500 year reoccurrence interval.

5. Mr. Perkins discussed techniques for developing seismic hazard models which were not sensitive to the choice of not-well-known parameters, accepting as necessary the limitations of this models.

This would tend to lead to models which predicted " upper" limits of seismic hazard. Mr. Perkins made the following suggestion.

(a) The selection of lower maximum possible magnitude events tends to destabilize seismic hazard models. High magnitude events should be used. This is consistent with the current under-standing of Eastern U.S. seismicity.

(b) The seismic hazard models can be sensitive to the choice of source zones boundaries. The boundaries should be " smeared" probabilistically in a manner consistent with the actual knowledge relating to their location.

(c) The various seismotectonic models lead to a number of differ-ent seismic zonations. The range of hazard predicted by the various models needs to be examined and the issues which are significantly impacting on the prediction of seismic hazard for the sites of interest need to be examined.

EEP Meeting Minutes August 6, 1986 (d) Ground motion models need to be refined. The existing data needs to be more carefully analyzed and additional data gathered.

(e) The "a" and "b" values need to be derived from regional (rather than very large) data sets. The values obtained will be more appropriate to-the practical source zone and will be less sensitive to the particular method of calculation.

6. A panel discussion was carried for the purpose of exploring the current understanding of large earthquakes, and the level of uncertainties associated with the prediction of their temporal and spacial occurrence. The highlights of this discussion were as follows:

(a) Mr. Bollinger discussed a recent evaluation of seismic network data obtained in the southeastern U.S. The data is extensive and indicates that in Giles County, Virginia the seismic energy is being released by a predominately strike-step faulting jn a near vertical northeastly-trending zone below the Appal' a chian decollement. The seismic activity in central Virginia is, by contrast, from a mixed dip-slip and strike-slip faultin,g in a zone above this detachment. Mr. Bollinger noted that tdchniques had been developed for the excavation and dating of sand-blows and that this technique had potential uses in obtaining information on the location and dating of large earthquakes which had occurred prior to the accumulation of the historic record.

l (b) Mr. Cornell indicated that he believed that there is currently a reasonably complete characterization of the seismological

! problems existing in the study of the Eastern U.S. Areas in l

which there is a lack 6f knowledge have been identified and no l

EEP Meeting Minutes August 6, 1986 major revisions of the seismologists perception of overall

. Eastern U.S. seismicity is expected. Unexpected events are thought to be a small contributor to the seismic hazard of a particular site. Mr. Cornell estimates the PGA values of 0.15g have return periods of from 1 to 3 x 10'4/yr in the Eastern U.S. Events of from 2 to 4 times the SSE at 10-5/yr; and events of 4 to 6 times the SSE at 10-6/yr. At the SSE level, the difference between the median and the 90 percent confidence bound is about three and the difference between the 90 percent and 10 percent confidence bounds is about ten.

These uncertainty bounds increase for the larger events to about a factor of five larger in the 10-6/yr range. Mr.

Cornell noted that in the studies done, the median tends to follow the upper bound, making it in some sense, a captive of the upper bound models. Mr. Cornell stated that the thesis that the occurrence frequency of larger earthquakes can be predicted from the historic record of smaller earthquakes needs to be carefully examined.

(c) Mr. Holt noted that it was important to characterize the site conditions and the 10 CFR Part 100 Appendix A has specific requirements for doing this. Attenuation of the different response spectrum frequencies in the structure, as displayed at Perry, is different.

(d) Mr. Perkins reemphasized the importance of evaluating and taking account the site conditions at a particular site. He j '

stated that be believed that the attenuation models could be significantly improved by the proper treatment and evaluation of existing data. He also believes that studies of existing ),

data will lead to improved (and lower) choices of maximum magnitude for much of the Eastern U.S. He, in addition, noted that the study of sand blows should produce valuable

_ . - - . a (

e s i y <

s e ~, 1 t 3 ih

8 4/ ,

. 4-

, A kt j' 1y ~

,3 ,

s,, ,  % i

- 4 ,

1 EEP Mseting Minutes August 6, 1986

b. ,

'b ' '

s y ,

, 1.nformation'as

-> l to the potentfal ,for and pattern of large s - earjhquakesir.itheEasternU.S.

s

% a (e) Mr. Reiter kre'sented a comparison of the recent LLNL and EPRI at e '

'l seismic hazard stupies. A summary is displayed on Figure 2 of h' 1 itth,chmbntA.,TherecurrencefrequencyofSSE'stendtobein

' u 'I/

' \the'ra'nge of 10-3/yr and 10-4/yr for most sites. There are, 1 ,

t ,

however, some exceptions. Mr. Reiter believes that there o s doufd b'e discoveries that will well alter the current pre-cehtionofEasternU.S.seismicityandgivestherecent discoveries on,t'ie Meers fault in Oklahoma as an example.

, q \*

(f) Mr. Siepp noted that the recent EPRI work indicates that n . [rscurrence' frequency for the current SSEs for Eastern U.S.

Ih viants are Th the 10-4/yr to 10-6/yr range, and that at a Necurrencefrequencyof10-4/yr, the SSE is about 0.2g on the e

East, Coast and about 0.1g in the Midwest. A 10-0 recurrence 4 ffrequency requires ground motion of a few times the SSE.

Attenuation models, particularly for high frequencies, contin-ue to be a problem. Site conditions and the associated resporse need attention.

L 1 (g) Mr. Trifunac believes that Intensity should be used in atten-p, uatikn model correlations. He does not believe that the current data set for the Eastern U.S. is large enough to j _ adequately describe attenuation and use the work done for

!s California as an example of what is required. He stated that

!' he believed that both regional and local characteristics were w

important. Properly located network instruments are viewed by ys / ,f Mr. Trifunac as ar important technique for gathering data

[

5' which can be used to resolve some of the current issues. This j use of physical models should be an important feature of data i analyses. The engineering analysis should include nonlinear

>3)

-(:.y

EEP Meeting Minutes August 6, 1986 analysis to the extent possible to achieve a more realistic analysis of the effects of ground motion.

(h) Mr. Thompson stated that we may see developments which would significantly change the current perspective of the Eastern U.S. and that current thinking should account for that pos-sibility.

7. Mr. Slemmons discussed some of the important features of the seismology of the Western U.S. Surface geology is important in understanding the pattern of seismicity and the data base obtained from observations and measurement of earthquake phenomena is much larger than in the Eastern U.S. New data is continuing to contrib-ute to the understanding of Western U.S. seismicity and sometimes changes the perception of the seismic potential of specific re-gions. Consideration of tectonic setting in the evaluation of the data is leading to significantly improved modeling. Diablo Canyon 1s currently the subject of an extensive PG&E/NRC reevaluation.

8. F. Rowsome discussed PRA insights into the comparison of external and external events. Seismic risk has been found to be a dominant contributor to risk in several of the PRAs in which seismic risk was considered. Mr. Rowsome stated that only in the case of Indian Point 2 was the contribution disturbingly large. The design for Indian Point was approved prior to the implementation of 10 CFR 100 Appendix and was evaluated in the PRA for earthquakes well beyond the original design basis. Mr. Rowsome believes that the principle benefit of this type of analysis lies in the insights it provides rather than an absolute values of risk it predicts. Relay chatter is a controversial issue and requires further study.

EEP Meeting tiinutes August 6, 1986

9. J. Reed discussed some of the probabilistic and engineering in-sights which have bee a recurred from existing PRA's. The most important benefits have come from: (1) The performance of plant walkdowns, (2) the focus on safety-related components, (3) the consideration of system related aspects of components, (4) the focus on the calculation of best estimate capacities. The charac-terization of a current commercial seismic PRA and the dominant components and core melt scenarios from the PRA's already performed are summarized on Figures 16 and 17 of Attachment A. The lessons-learned and conclusions are summarized on pages 18-20 of Attachment A. These PRAs have identified classes of low-capacity components and have been useful in quantizing seismic margin.

10. I Wall summarized the results of recent EPRI assessments of seismic PRA's. Previous EPRI work had been reported in EPRI NP-3562 (June 1984) and EPRI NP 4168 (July 1985). The PRA's reviewed were for Indian Point 2 and 3, Millstone 3, Oconee 3, Seabrook, Zion, Limerick, and Shoreham. The basis methodology of these PRAs is similar but the scope, depth, and basic assumptions often differ in significant ways. A summary of the results, conclusions, and recommendations is given on pages 21 to 24 of Attachment A.

NOTE: Additional meeting details can be obtained from a transcript of this meeting available in the NRC Public Document Room, 1717 H Street, N.W., Washington, DC, or can be purchased from ACE-Federal Reporters, 444 North Capitol Street, Washington, DC 20001, (202) 347-3700.

-n n -

• • 8 5 66 6= &. O w...

- u

_3 4

5

• -oes n@4 Ls 4 ,

a LN e

a y

e uo s ys M 9%

t~

= $

.EQ e ( *. bd O E N4h0 $ 40 Tr j i gN i M y h (! ~

/// -

!w g g .t .

1 i

g ' .

sga my x 4

I * -

.- a1 Bg g

~

'V b

i 5

~

• m E

E _ -

z y -

20+3 4M - '

v, s:

o=

Te ve ve 1,

4" re

1. -

m= - -

6 LL.I =

d5 y EE d E 6

' i y n 5

L m R

EzE Sm 5 - -

WE a ea fc .

sa$ 5=

pg g- - * ~

S 35x 55 0 i M5

= .-

b m

l$

g &W 0h -

h "k

Mj EN o e  !

Ms w r l*

5 C$ @g '

"5 e

~

kk$ ,

ew -

CO+3 L

. . .#-2_ '.._.

O , O C To To

I. <  %. ,;i.:Ir

• s n 4 g.W.

1. SI!J -- g,n'# K. ~..s.- 2* j ,

, $f -

. T.-'J.M..,~

"'.*d'nefa'-

J';;

%e d' I.6.. :".?"' "

- sesettaa

'e\

f 't' 't' "' '

i de A A == R  %.

1 s  %. #

1  %

e 1

w'l' .

5

, s..

'l l

./

J l

D['

)

j I

ji N

w **.

4 }

l

.r, -

h

• %r r- .

v . ef $;,j' .-

l h ~; ,

I  %

sw.

f

, ~.

._ o

.A

%. . 4

(

2 L ._

• e I

! . s.

+ ,

. .~,9." .

. \

i l

I i

l l

T g

?

E0 th NO EM

'f' "' " ,; \ p I

1 i

y

(a Y N

'~

( .

8a

,sk d

)

\

f8 i V

N, 9 4

<l l\

~

/

/

. - _ O

.g 5. #

l .

L _....m. ,, .. ,..,..,_:. .. - ~.~... - ,,,

l

64-27-7-1-0-0 N/yr/km2(x10-6) 1.1008 (33 7%)....... 1.7235 b-value-1.10 at .

O l

• N 12-4-1-0-0-1 N/yr/km2 (x10-6) eI jf 0.31208 (9.6% ) . . . . . . . . . . . . 1.9098 Rtgional rate 79-30-5-1-0-0 0-0-0-0 M 3 4.3 1.1600 (35.65)....... 1.7833 .00654 (0.25)............. 3 069 4

s 4T 2-0-0-0 36-16-5-2-0-1 0.1863 (5.75)............ 7.5042 3 2624 ** AfA 0.93349 (28.65) d' 12-2-1-0-0-0 T<< 4 jf 0.082 35 (2.55 ) . . . . . . . . . . . . 1.1772

', 36-16-5-2-0-1 ,4' 3-2-0-1-0-0 1.0016 (30.75) 0.1564 (4.25)............ 2.2978 9, 4-0-1-1-0-0 AREA Dl8TRIBUTIO14 *7 0.18983 (5.85) . . . . . . . . . . . . 4.5585 C *,

eS

%e lh 0.0058847 (0.185)......... 2.5198

'a 0-0-0-0 0.021870 (0.685)......... 2.5239 0.06811 (2.15) 0.03003 (0.935) . . . . . . . . . 2.5 207 AREA DISTMIBUTION 0.010319 (0 325)......... 2.5216

$ ~: ^..

. x r. .

JMMh,@It .

( ,se 7# a n'

' AA' no" q

\ \ s%

+ a.

+ .

e Y

p 18 N 33

\

\i4 20

p. 13 4

to) . . .$

12 - ed

.. -(a ,

{ ,

s0 u to

) T" .

s

's "h:F6 f~ .

\

, { 9 Y.

(

..Y . . {/

) .

8f 1'.. . .r ' ,e -

4-f ~J_ , " *, , .. g4

.-d.

- /, .,9,

~ f 16  %

% j

! ,i ^ukm x

.. .. .. M--7 -

14 y 15 .

\, h.. N

- 38 A

I 16 t 14 l

Y j\'

_, g- d. -... , / +

g . i.. ...=ne.

j e s.o 3.o stisemeses.

- If

+

+

1 s \

1 to' as' so" 7 s' f

,. S.

$ C ,

M ,h :

\

'Eh['[.,EMa*' )

O. D,s'f"t.M..!!.!"t%

. f <

t .:

?  ? 'f* *

4. e. e. "==

?

. me ..= .... ,

./k .i. ..

( 4 i /

I

)

. ~... .

fs~.I; 1

r pgi)"llrg jw :w -

f C EM YWENT G

h:+g,h /

J

,g S. <,

i l

l 6

e , ,

S.

( # 4 e** >

• ?lV 33 l

l

• l r.

O tee sio 350 as *4.

j

' /

, s

. 1 4

( .

S!

\

a

(

\

s f

l -

i Y

- ns O

./

j s

's.

, t

(

l l

l 1

7

. .-.. . . , ... ..,, g,....:=

. - ~ ; r g --- -

ao. , n. tr as-an.

[

+

I t i

+

.w .

.f. x.-

// st '

o

~1o

@h 4

,g [e

• g '

39

• h4l I

4l (_

f.. 7*

l l l 3 .

- ed

+

-, . . , . . 17 i

e 1 . IN o

'l A. . . -

l * .

J t

4 T

g

)+.1~ *

• 1

( H R Wi

)  % Y.

~

.-J f~g

.- s Aff

. + M

, 4o

.._. s

, ' 22

/ p~ 17

~

11 A

' y

.._.. ., t

' 17 , Y

\.23 .

A -

Q

, +

I ( 10 y . __ _A._ . .

+

+

._ I* .

'?*

sde n. .....,,

et g +

+

I I I i 75' 80' w as*

~ -=-'- _- __:  :: - _;-c ~ ~ : - :- ~; -- _ _ __

e

~ ( ,.?

'E S.

j 4

VERTICAL NCYEMENT RATES (mm/yr) g D a= S -i O .. '.. .. g. .. ..s e.. .. .. g. ..

- t .,. ,e ...

i ea*~ .l ,,,

!l' d'  !

+

. ~ i< ' s

. .. g.

O. .i .s..-

9 W

> +. * *r

. ==

s g

.. [

' g,

- N 5 ,_ . vs i

.. - n\ .

M q

• A \ s r'

< r '

] ...

.J7 ,

\

P >l s 9 y

.__ y g=W =,

~

• .I; j .

\.l1

- .e

'.:.::i:

. .- N Q; .\,

s. h.

e

. s. ,t.

f  %*ehy*

h

. ,j.1 *.' ' .

..e

\ esovgtagery as0 te0vttstet?

10 SLOW Sh ow $W$etasCE

. FT lasCS .

L 1 f $ #f 886L48 *,. ,

4 e 34. 34.

.- l

.- o r

' 'O. ,

)

n.

j r.

g

(.\

( ..-

h l

.'-'...' .' . )

3y . . ..

.y s

)

, ,e l .. .

4 l

x$,:' l;.

l i

k a I

t g

f g/ *

.g S.

(

IC

8 e

( ' as. ** r s. va a s>

I I + i

[*

+

+ *a

, '=

j .

+ i

)'t

--y i R se ' .I I

%v -p. I+

, I

_s 11 f .

- .e W +

~a

- T -- ,

o l i ..,,- e t' '+ / ,,

(

1 7

}

) .wY.  !

p~</.r' ?.

\ +- w t

...-.. J 7.. ,,

4 40 s

40 ,- / -

3 23 -

A 4f, .. ., + ..

\. -

T 17

\

'7 h. g A -

Q ,, y i 4 I ..

+

.~ . . . , , , ,

i l

!* ' .'?' " 858 =si.....,,

st

+

t

+

I I 1 f I 1s-

w as- so* Ih

g a

6 15' 10

' g s

\ N \ e$

.~

4*

+ -

. .h:

Q . .<

~

, k

'\

a.

h s s ,t 1

.. ,1 b g

... ** f .. .( #

..- k O

s i

\

. 4 . .. ,. .

(. .

+L.t J.

~. .

J

) / I r ,<

- k-

d. i .'..q, '*

.V /

=

jI .. -

. sykm.-

A

\  % ,

\

/..- '

s.

/

y

\ i

.s* -4 g

4 A

4 e

i, s?*"H**

/

1. I

(

n ,n +

k

\ g \

\

to' as' se is' A

\r

l e

' as* noa i s* to c

a s*

\ \ q \ \

4 -3

+ y e ct f

.e/ k ,,

t . \

) - - 8 pk '

( .. 4 04

) . .w - - 4 1 , o

\ 4.+..

(

i 9

f _.Y

[7

. ,/ IQ 2

h / / no haf'A),

^'

~

s Lk, I A

j ,, __./- ,

l 24 4 27 T

%4 A - 38 l .

+

1 .~.

+

i ei. sie n u.......

& F N * - rf

(

t i + .

ena as' se 75' baC E

l

( w 88* 1 s* 7F ss*

I l 4 t \

g:  :

+  ;;;

+

e e

i

/ 17 kne i I V *$

) EI t A'

ll 1e

( -

+ l I ,

.. g ,

" W

..- ,,  ! 7 t  :

. .y . .

f se I

+ 10 o

1 l 1.+.q.

7 f' y

( I ~

g l ,,

W*' %Y..}. . ". ~ /* * *

,4F .

. y y ,, , .{ZJ

._: s d34 sa A

u .. .' ,-

,o . .

k 17 ,

y so h *. 4 23 A - g y

I 4 I . . _ . ..

^

~

• 'E' 858 x er. ...,,

. M A W A,4r m f4&C4

' <'O M ,,,y g g

• E$

( N?eQ[g3

+

l l I \

av yp w as' l

- ~- . . . ~ ~ , '*=,,e , ' - ~-~c ' ~ ~ - - _ _ .

e ,

e I

as* *** ss 7e ' 5'

( I g .g.

+

+ i.s

/ +'t .

y 'Y

+

e

...S Ylf ,

i V. ,., ,

+

-T t* j

{,

i

, - ev t .._..- 4

.g.

.. l sy .

3 I b

~ ~~~}~ " ,

N I

f ,)  :~ .

l t'

( i

) et p/.rwX. '

+- ..

- &- si a-

• ~~

, 'r to 1 44 s

u +. ,. .Pf 4 U..

te (.'~

+

• 4 \, t

,o

\ ~

- ,e 4

). . _.4

. ~ .

+

s p e =,,,,

.~

l I* 858 '?'sie=,,,,,,,,,,

$f*)p y .

Y & ce, c c.'*s yg WO //hg, ,

Mebfq +

+

i I I I 85* ao* 7s*

90* 1

- _ - - . . . ,. ,. n , . -

\

l l

OWACTERISTICS OF ClERENT C(ftERCIAL SEISMIC PRA NAZARD ANN.YS1S e ESTAN.lSHED IET W OLOGY 9 LARGE UNCERTAINTIES FRAGILITY ANN.YSIS e SitT'LE LOGNORML f0 DEL 0 (LNERIC NO PLANT SPECIFIC DATA e ORIGINAL DESIGN ANALYSES 9 DESIGN AND CONSTRUCTION DISCREPANCIES (NOT INCLUDED)

G IEDIAN PAPRETER VALUES (ON CONSERVATIVE SIDE)

S OJCERTAIKTY (ON LOW SIDE)

SYSTEMS ANN _YSIS e FAULT TREES S NAZARD/fRAGILITYINTEGRATION l 9 DJMINATE CONTRIBUTORS DIFFERENT e RESULTS C0f(TROLLED BY 9 h DIN 1 FRAGILITY (AND UNCERTAINTY NIEN CAPACITIES ARE HIGH) e UNCERTAlHTY IN I M l

f Jock R. Benjomin & Associo1es,Inc.

Consulting Engineers -9 l4

Iml%1T COMP 0fENTS MD CORE PELT FRE0TNCIES DOM!riNn '

PMAN FREQUErCY OF EL/1{I {0MPONEriTS CORE PtLT (PER W AR) 210N SERVICE VATER PmPS 5.6x10-6 AUXILIARY RUILING IfRER-CONNECTING PIPING e CRIB fDUSE 125 VDC BATTERIEF RACXS IriDIAN Polta 2 C0fGROL BUILDING 1.4X10-4 INDIN4 Polta 3 DIESEL GEf4ERATOR FUEL 3.1x10-6 OIL TANKS C0ffTROL BUILDING llMERICK 440-V Bus /SG BREAKERS 5.7X10-6 440-V BUS TRNiSFORMER BREAKER 125/250-VDC PUS 4-KV BUvSG DIESEL GENERATOR CIRCUIT BREAKERS M1LLSTONE AlfXILiARY BUILDING 9.4x10-5 125 VDC DISTRIBlfTION PANEL DEMINERALIZED WATER STORAGE TN1K REACTOR CORE GEOMETRY-SERVICE WATER PIPING ESF POILDING EMERGEfCY GENERATOR ENCLOSLRE C0f(TAlf0 Era CRNiE WALL Jock R. Benjomin & Associates,Inc.

Consulting Engineers B

,d

1 LESSONS FROM PAST SEISMIC PRA l CAVEAT: THESE t%__Y (COULD) HAVE BEEN LEARNED USING APPROACHES DIFFERENT FR0f4 FORfML PRA.

e SYSTEl% TIC EXN4INATION OF " SAFETY-RELATED" COf PONENTS.

e EVALUATION FOR EARTHQUAKE MOTIONS GREATER THAN THE SSE.

e FORfMLIZED WALKDOWN (WALK-BY) OF C0f POf1ENTS IN THE PLANT.

9 IDENTIFICATION OF RELATIVELY STRONG AND WEAK C0f PONENTS, E.G.,

STRONGER WEAKER CIVIL STRUCRIRES TANKS PIPING UNREINFORCED BLJDCK WALLS CABLE TRAYS BATTERY RACKS NSSS SUPPORTS DAMS VALVES DISPLACEMENT BETWEEN STRUCTURES DIESEL GDJERATORS DIESEL GENERATDR PERIPHERALS PLFPS (WELL ANCHORED) RELAYS t

l Jock R. Benjamin & Associates,Inc.

l Consulting Engineers B t ;.-

e OLDER PLANTS POTENTIALLY HAVE MORE PROBLEMS THAN NEWER PLANTS.

6 FOR EAST COAST PLANTS, MEAN FREQUENCY OF CORE MELT GENERALLY CONTROLLED BY ACCELERATIONS 2 TO 4 TIMES THE SSE.

e UNCERTAINTIES IN THE Nt.f4ERICAL PROBABILITY RESULTS ARE LARGE SUCH THAT FACTORS OF LESS THAN 5 TO 10 ARE NOT SIGNIFICANT (PRIMARILY DUE TO SEISMIC HAZARD).

e FREQUENCY OF FAILURE IS NOT GREATLY SENSITIVE TO CHANGES IN COMPONENT CAPACITY.

9 DETERMINATION OF THE MEDIAN CAPACITY REQUIRES A HIGH PROPORTION OF JUDGEMENT VERSUS USE OF ANALYTICAL OR EXPERIMENTAL DATA (CALCULATIONS MAY BE CONSERVATIVELY BIASED).

I 9 IT IS IMPORTANT THAT C0fM)NENTS ARE DUCTILE.

O COf4TRIBLTTION FROM SEISMIC PRA APPARENTLY IS COMPARABLE TO OTHER CONTRIBUTORS (E.G., INTERNAL EVENTS).

8 IT IS DIFFICULT TO GENERALIZE FROM PLANT TO PLANT.

Jock R. Benjamin & Associates,Inc.

Consulting Engineers D rI/

J CONCLUSIONS FROM SEISMIC PRA

(#0 OTHER EXPERIENCE) e StISMIC CAPACITY IS GENERALLY HIGH (A CLIFF DOES NOT EXIST It44EDIATELY ABOVE THE SSE).

e OLDER PLANTS ARE MORE LIKELY TO HAVE RELATIVELY WEAKER C0f PONENTS COMPARED TO NEWER PLANTS.

e CERTAIll CLASSES OF C0f PONENTS ARE POTENTIALLY WEAKER THAN OTHERS.

THESE C0f PONENTS REQUIRE THE ATTENTION.

e PAST SEISMIC PRAS GENERALLY ARE CONSERVATIVE (MY JUDGD1ENT) .

e POTENTIAL FOR DESIGN AND CONSTRUCTION ERRORS AND EFFECTS OF OPERATOR STRESS AND RELAY CHATTER HAVE NOT BEEN ADEOUATELY RESOLVED.

e SEISMIC MARGIN ANALYSIS MAY BE MORE USEFUL IN DECISION MAKING THAN SEISMIC PRA.

e CALCULATED VALUES FROM SEISMIC PRA SHOULD NOT BE USED WITH0tJT THE ASSOCIATED FINDINGS FROM THE ANALYSIS.

Jack R. Benjamin & Associates,Inc. I Consulting Engineers D n-'

C'

SEVERE FUEL EARLY CONTAINMENT LATE CONTAINMENT -

DAMAGE FAILURE OR BYPASS OVERPRESSURE FAILURE l-d

! OVERALL SEISMIC  % OVERALL SEISMIC  % OVERALL SEISMIC  %

LIMERICK 2.4-5 5.8-6 24 6.9-7 5.3-7 77 4.2-5 3.2-6 8-b 6.5-5 2.6-6 4 N/A N/A SHOREHAM 1

IP-2 1.4-4 2.0-5 14 3.7-7 7.7-9 2 5.2-5 7.7-6 14 1.9-4 5.7-6 3 5.8-7 --

<1 1.1-5 6.7-6 26 IP-3 l

5.9-5 8.9-6 15 2.1-6 1.1-7 5' 8.7-6 8.1-6 93 MILLSTONE 3 I,

j 2.5-4 6,2-5 25 2.2-6 1.2-6 55' l.7-4 6.3-5 38 OCONEE 3 I ^l[~1...,

A e c.,..o. ,,

l 2.4-6 5.7-7 2 1.7-4 2.4-5 11 i

SEABROOK 2.3-4 2.3-5 10 i

j 5.7-6 1.0-6 --

<1 5.8-6 5.6-6 98 l ZION 5.2-5 11

) //

L __

l l

l l

CONCLUSDONS

• PRAs, ESPECIALLY SEISMIC, ARE VERY COMPLEX '

- REPORTS ARE VERY DIFFICULT TO UNFOLD BASIC METHODOLOGY IS UNIFORM

- SCOPE, DEPTH AND ASSUMPTIONS VARY GREATLY PERCENTAGE SEISMIC CONTRIBUTION SOMETIMES MISLEADING

- SHOULD EXAMINE ACTUAL PROBABILITIES AND CONTRIBUTING MECHANISMS o

CONC!.USIONS (cont)

ALL REVIEWED SEISMIC PRAs IDENTIFY BYPASS OR EARLY FAILURE OF CONTAINMENT AS DOMINANT CONTRIBUTOR TO EARLY FATALITY RISK

- NO STUDY ANALYZED BYPASS SEQUENCES IN ADEQU ATE DETAll MECHANISMS LEADING TO EARLY RELEASES l ARE VERY PLANT-SPECIFIC )

ALL IMPORTANT COMPONENTS ARE " GOOD" FOR AT LEAST 2

• SSE FOR SITES HAVING SSE BELOW 0.5g COMPONENTS WITH CONSISTENTLY LARGE SEISMIC CAPACITY

- PIPING - CONTAINMENT

- CABLE TRAYS  !

COMPONENTS WARRANTING CAREFUL REVIEW i

- RELAYS - MASONRY WALLS

- ANCHORAGES - DIESEL-GENERATORS

- NSSS SUPPORTS - VERTICAL TANKS I

4 O'

9 0

RECOM M EIOATIONS .

o NEED MORE SYSTEM ATIC IMPLEMENTATION OF EXISTING M ETHODOLOGY o SEISMIC HAZARD METHOD HAS BEEN RELATIVELY UNIFORM EPRl/LLNL WORK ASSURES CONSISTENCY AND ROBUSTNESS FRAGILITIES METHOD IS UNIFORM BUT NOT NECESSARILY ROBUST ASSUMPTIONS AND SCOPE REQUIRE SYSTEMATIC APPROACH m <

o SEISMIC ACCIDENT, SEQUENCE DEFINITION AND QUAN-TIFICATION ARE NOT UNIFORM

- ASSUMPTIONS AND SCOPE REQUIRE SYSTEMATIC APPROACH NEED ADDITIONAL REVIEWS AND SENSITIVITY STUDIES

-_ _ __- -