ML20041G560

Forwards Review of Zion Probabilistic Safety Study Sys Analysis

ML20041G560
Person / Time
Site:	Zion  File:ZionSolutions icon.png
Issue date:	03/10/1982
From:	Varga S Office of Nuclear Reactor Regulation
To:	Delgeorge L COMMONWEALTH EDISON CO.
Shared Package
ML20041G561	List: ML20041G560 ML20042A000
References
ACRS-CT-1420A, NUDOCS 8203220532
Download: ML20041G560 (6)

Text

y*. '

5 Distribution MAR 10 i'382 Docket File Docket Nos. 50-295 NRC PDR 50-304 Local PDR ORB #1 Rdg D. Eisenhut OELD OI&E (1)

Mr. Louis 0. DelCeorge DWigginton Director of Nuclear Licensing CParrish Commonwealth Edison Company NSIC Post Office Box 767 ACRS (10)

Chicago. Illinois 60690 JHel temes

Dear Mr. De1 George:

Enclosed is a short report from the Stidta National Laboratories containing comments on the Zion Probabilistic Safety Study. This review was performed under cor, tract to the NRC. This information is provided for your infor-mation and coments if you wish to do so.

Sincerely, l

^h Steven A. Varga, Chief Operating Peactors Branch fi Dhision of Licensing

Enclosure:

As stated cc: See next page g

C MilC2fyra 9

V-:

MR19199g;,.F

("S$M2 spa ics l/

'b l

C' 8203220532 820310 PDR ADOCK 05000295 P

PDR

. DL.th,].

.,DL:0RB#1,

omcr >

sunums)

.U.3.!.9.9.i nto n :frs

,,1 3M.182 3

82

,.I omr>

c.nc ronu ain oa ao3 nacu eno OFFICIAL RECORD COPY usom an-n,* j

.i Mr. Louis 0. DelGeorge Comm3nwealth Edison Company cc:

Robert J. Vollen, Esquire r

109 North Dearborn Street Chicago, Illinois 60602 Dr. Cecil Lue-Hing Director of Arsearch and Development Metropolitan Sanitary District of Greater Chicago 100 East Erie Street Chicago, Illinois 60611 Zion-Benton Public Library District 2600 Emmaus Avenue Zion, Illinois 60099 Mr. Phillip P. Steptoe Ishan, Lincoln and Beale Counselors at Law One First National Plaza 42nd Floo-Chicago, Illinois 60603 Susan N. Sekuler, Esquire Assistant Attorney General Environmental Control Division

'86 West Randolph Street, Suite 2315 Chicago, Illinois 60601 U. S. Nuclear Regulatory Commission Resident Inspectors Office 105 Shilon Elvd.

Zion, Illinois 60099 James P. Keppler Regional Administrator - Region III U. S. Nuclear Regulatory Commission 799 Roosevel Street Glen Ellyn, Illinois 60137 i

S e

r.-

> rm :r".:,

Sandia National Laboratories A 3 0 ' C.! f I O u e. N t v. i.' e s 0 0 & * *. I 5

,'b I.

e r s,a k ' --'

__.r,_.---+.sro February 15, 1982 Pro:essor David Okrent 5532 Ecelter F.all p...,.,,.... v.

c' C a '

....d a as

'sc s Angeles Schcri cf Engineering and e

e n ; a. n e e.

s e,Oe -

cs e.n g e.t e s, Ca n.zornia a

er o.zessor Okrent:

sear to Mr. Gr letter of D..ecember 18, 1951, a crae.,.iesmever's. personat coservation about I n. response review anc

..ve enclosec.

2

. a.en r.v,..

,.n o.n e.

.crovices : ooc _ o r th o u cht.

the c

Sincerely, l/

e.WW p/w=/

/

/

n J a CI=, K. hiChman, SupervisCr

'; u c i c-a r Fuel Cycle Systems Safety 21._sier 4412 C.. ' : il 12 : Ep Enclosures C e r s-i c :-

- s : 1 <....

=...

c e.-- e.e e.c,

..c.. : -

us

. ~.

l

'l E:: A C F

-:...o.cs c e a.-

e..

u.

nadani n

L.c..

.s 1 ~ 2 :-

R.

G.

Easter ing 4410 D.

J McCloskey 4412 D.

D.

Carlson 4412 G.

J.

Kolb

Sandia Laboratories em:* c.

% r.w.:e cate-February 9, 1982 L.vemo... ce.w...

to:

J. W. Hickman, 4412

./M u,c "Om F.. G. Easterling,1223 Cc,.e-:s en :ne 2icr 1r.ferfacir.; LOCA Analysis 2:.+ :'

(;;.1.3-72,77) is cominated, by PLG's (Pickarc, Tne ir:e-f acine LOC Lowe, anc Garrick) estimates, by the rupture of two mctor-c; era:ec vaives Tne scenarie is that first the upetrear and then j

in the RMF. suction path.

- the downstream valves rupture sone time during a ene year period between Let i denote the hcurly rate of valve rupture (assumed refuelinc outaces.

to be constant).

Suppose further that both va.1ves are subject to the same Then 'the scenario pretability is rupture rate.

J 0 = 1 - e-^ 2 (1 + AT ),

resuit fer the ca:e c'

• t.ere ! = E760 nrs. (inis is a casic re'iabilits erre,ecusly u - (.-),.

s.

,r t.e or sr.a:,.. n sca dtv recuncancy.)

~

r

. i e

. ;.;. i, c, a.

ce,.se ~.

..y se.

..e. e. e.

. - c y.,1. t.

.. =. <... o c. e_

w e

v w-n e

.n..

=

w...

.e. ;. r e.

..... 2. i. e

e.. g c. wi;. ). c. e. *.,.. -

ci e

,2.

r. e. -....,. a -, < e r.

. n o.

2 c

e,j

..c.e.

A....

g2(

-+

n-

.e:

n,4Ag c-]*,,

4 c.

t...

e g s..-gre.--

F,r.-*

..... e.

-u

+ e c......c,.

.c.

4r.

(

m'.

e...n 6-~..- -. v g.

c.

That In PLG's Ea.vesian anti.vsis 1 is assignec a prior dictribution.and a variance of y

cistributigt is legnertal *ith a nean cf 2.66 x 10-C-w...

4..

.e c_ _;y:..,

s.e

~.3c<<..... -.. w. 4. e. r <. c.. r..... :.,

4...

e a...

..n-::.

• +, --

e 2. n. - r.-

.. s.. c.

c. :,.. c. ; c..c.

5c. g c e.., a - n. s.... c v.

g... g *. p g *be.-

y

. h.e.

C.*.-.

g..

C..**.-

~.

.e

.n

.. e 9.,y

.; *. g

..;,. y :.

.,.....-g.C'!.

e. e

+w s.

4.

=.e....

, e.. c.... ; n_ e. e.

c.,,-..n...:,.

..e.a...i.-,..,

..4..,

.e,-

c,.

c..c ~e.

.y

c Tnis :rier distributien is nc: mcdifiec by Zicn ca:a, in centras:

resul*s.

OLG's us.ai procecures, so it is the basis for ineir.~subsecaen:

Given 0 = ( AT)2 a, and the assumed distribution of 1, the mean cf Q is

/

mean(Q) = (T,-/4) mean (14)

(T /4) [mean2(1) - va-()))

2

=

E.7 x 10-6

=

(Ancther scenarie censidered is a' leai. of ine upstrean valve followed Adcing its gstimatec prcbability by ructurr. cf the downstream valve.to the abcVe yielcs PLG's estidat

.4.

J. W. Hickman, 4412 February 9,1982 -

of the interf acing LOCA, p.1.3-77). The second equality above is a standard relationship among moments of a random variable and is not a consequence of the lognormality assumption.

PLG's stated methodology was to treat WASH-1400 bounds as the.20th and Suppose we do this here.

.50th percentiles of a legnomal distribution.

Then the prior mean pf 1 would be 4.2E x 10-7 and the variance of 1 wculd be 3.3C x 10-Id.

Substituting these values into the abov:

expressic for mean (O' yields

~

re e -d O ) = E. E

2. i0-0, Pirt of a s.;r;-isin; five Orcers cf a;nitude 'arter than PLG's result.

We dic this dif ference is cue to approximating i - exp(- AT/2) by AT/2.

a 5000-run Mpnte Carlo and estimatec' the mean of (1 - exp(- AT/2))2 to be 2.9 710-", sc there are still four orcers of magnitude attributable change from 5/9E bounds to 20/80. This change to the seemingly innocen results frcn the change in var (1) and the way.in which var (1) contributes to PLG's mean value, their point estimate.

Fer PLG's Let u and c denote the mean and standard deviation of In(l).

'i i ld:

2.74 In b :h cases a = in(10-E)

g the 20/80 assumpt ons y eThe nean cf a legnormal distrib assu.ptions c = 1.40.

Usin 15.12.

c /2).

Ey increasing c frcn i.a0 tc 2.72, the near is in:reased I

e ex;(U By a fac c; cf Ic., Tne variance Of a ic$n:rmai cistributic. is e r t ( 2.: - c4)(ex0(C#} -l).

Here the incre2se in C resul:C ir.' incre a s ing ite V2riance by a f3C:c[ Cr 77.71E.

2n

• e abOVe Cal 0 Id '.Ons TCP Dian(C),

l i#fe* ente i! C::ai ed.

I sa-(,) C'. tis",adCws rea-I[?) ard thus :".e la ;e c~

.ne ficu-e relC+ snDws the tw..10$nt-nai C'.s!-ir.:'.Cns...s C*stusseC, Crawn Cn arithrdtiC scale.

Innocent-lC0linE LCrail Cis:#ibutens on I O;-

scale transforn te greatly skewed distribu:1cns on arithmetic scale.

Whether they accerately cepic; anybody's state of knowledce is open to

-cm inis 'igure, the lar;e ircrease in the variance cf 1 usinc a

eil a;:gren:. However.

l

.;es;; ~.

5e 20/50 assen;u r.s. raine- : tar 5/55, is n::

M-' i- :ne f0-rer case.e-sus :ne

~

) =

nc 2N c' : e c's:-i;.: ic-Seyr :

le leVirage.

E:.in *ne lat:Er exerts CCns".Ce*~ J Tnis analysis shows the er:reme sensitivi:y cf PLG's results to their assumac prict cistributions (and the acvisabili:y cf havir.; cata) and als; illustrates an unrecognized, though conservative, flaw in their Bayesian To see this, su; pose ene b gan with a noninf ormative prior distribution for A, then mod;fied it by cata consistinc of f f methodol ogy.

I f/T and a variance of A'/T.

The pesterior mean of A4 i

i neurs.

a mean c# l' -

e woult De nean(14) = i'4 - 1*/T i

l

\\

r o

o.*

4 J. b'. Hic kman, 4412,

-3 February 9, 1982 If we regard this as a point estimator of 12 and take its expectation with respect to the sampling distribution of f, the result is E [mean(1 )3,x2 + 2A/T.

2 f

l Thus, as an estinator cf x2, the costerior mean is positively biased A!: cent-ive and can be seripusly se, cs the above results illes:rg:e.

estica:crs c' 14 basec en ncn-Eayesian ne: hoes wpid de 12, if I'/T, if one one wintec r. maxinu, inelincoc estimate, cr-A*4 wanted an enbiased estimator (: hough this can :grn cu: net.a:ive, se it migh: net be used in such cases).

RGE:122f dw-b y

-w-_,,

2.

U-t fn i

{_hNI'l bW

- ~ ~ ~ ~,

Dyftt s.

• i*

.it* *. 0 1

r-w i 4

'F*

i l

I I

b'd i

.$ \\

an s (( e'.

f 5 e= I '/k

[.wff [s e,'f,

T *// 4

. * $ 75 v)I*., k *' / [ ' #44 e

en EA[A

. -. c, :.

.: :.:..c'

..' t a-l f

%r r"

^6

" * <

• a ;,. --

~'2'~"

- se o5 s.J i,

,, (., s -es -

t

,.m

. v,.,,:

./

,. - u,;.

u~

age

\\

\\

f i

q,$ f i

I i

hfp6 I

O Jg h

m m

m

. - =