ML20244B090

Rept 2 to Acrs,Nine Mile Point Nuclear Station Power Increase

ML20244B090
Person / Time
Site:	Nine Mile Point
Issue date:	01/27/1971
From:	US ATOMIC ENERGY COMMISSION (AEC)
To:
Shared Package
ML20244B088	List: ML20244B086 ML20244B090
References
NUDOCS 8904190004
Download: ML20244B090 (22)

Text

---

----

...L{ ,

1

,% ! January 27, 1971 Docket No. 50-220 Report No. 2 to the ACRS

.N

. '} ,

43

+

.4 6,

NINE MILE POINT NUCLEAR STATION i

Power Increase _

.l. ,

t

. . .i

).t U. S. Atomic Energy Commission Division of Reactor Licensing i >

l 8904190004 710127 0 ,

DR ADOCK 050

..... , : y g. y

._ u f

@p.- - . -_.u q u . __ v ano i

.9~

. f ts, . .

.l-p-L l .4:&.

L .r ,1 ,

ABSTRACT r +

• Our November 23, 1970 report to the Connaittee, regarding 4 I

the increase in the licensed power level of the Nine Mile Point Nuclear Station from 1538 We to 1850 We,  !

l

' * -' ' indicated that our evaluation of the performance of the emergency core cooling system at the proposed power level of 1850 We had not been completed. We have performed extensive additional reviews of the calcula-tional models now used by GE and a model developed MK - independently by INC. . Dif ferences between the two

+ models have not yet been fully resolved.

-li;A The applicant has provided assurance that the NMP core

+' 7 .

spray system can achieve rated flow reliably in 35 seconds or less, instead of the value of 60 seconds assumed in pj,7, j OW1 previous calculations. Taking this change into consid-eration, we have determined that, for a loss-of-coolant 47, accident resulting from a recirculation line break, the peak clad temperature calculated by either the w' -i f GE method or the more conservative INC method is less than 2300'F. For accidents involving small breaks, (n. GE calculates peak clad temperatures of 2226*F.

4.W:<

~

~

We have concluded that the ECCS performance is accept-able for operation of Nine Mile Point at a power level

' ' ~

of 1850 W t.

.s,.r,. ..

4 ab

S,8 f } .-

,, * [*..

4,p f

4.~

l t

j OFnt;n m

v e M_ _

___._,m m any gp assa pee *.

~

e. j

u ,

,. l

" :I .,

EVALUATION OF ECCS FOR 1850 W e .I 9

- NINE MILE POINT In our. evaluation of the Oyster Creek power increase to 1690 Wt,

~

4 we used the FLECHT BWR test results (Er-2K) as an experimental basis for assessment of the adequacy of the ECCS. The' indicated test conditions for the Zircaloy bundle test (Zr-2K) were essentially upper limits for -

' ?, *h,

. 6 the 1690 We power level, so that the test results could be regarded as a demonstration' test. In that case, we did not require additional

~

l analytical model formulation as a basis for accepting the adequacy of

{

.i.* : e , the ECCS for:the Oyster Creek application.

The power increase to 1850 Wt for the Nine Mile Point reactor, however, appears to excee d the test conditions so that a similar l

I approach is not possible. The relationship of the Nine Mile Point conditions to -the FLECHT test (Zr-2K) conditions is indicated in

! the accompanying tables. The total power levels and some of the spray initiation

.. ~1 temperatures obtained in the FLEQlT bundle test are exceeded by those f ,~,

n expected in the Nine Mile Point core at the higher power . level, although

[

the linear power density appears favorable. ' On balance, the indicated

/

dif ferences between the conditions of the test and of the Nine Mile J'

Point reactor at 1850 MWt require a systematic extrapolation of the test results by means of an appropriately developed model formulation based on the full range of experimental data available.

GE had previously presented a spray cooling model development for the Oyster Creek application for 1690 Wt based on the FLECHT l-J

.mu- je NF (yy_mj nmu ,;;; Q J}'

~

' .m m n u w "

u <

i-  !

l- l j

i

.j ammnw - - - - -

4,z ,

%T m a , .. , m~

l 4

• l I tests. At that time, we concluded that there had not been demonstrated sufficient bas'is on which to completely accept the GE model'for spray

.. 4 ~

cooling. Several subsequent meetings have been held with representatives r; of GE to discuss their model formulation in greatar depth. In addition, independent ef forts to develop a model have been initiated by INC. I

'.};h Recently, a meeting was held at GE in San Jose between members of the

' .Mr D.

..t";

regulatory staff, INC, and GE to discuss detailed calculations obtained from the models and to er.plore conceptual differences that have been form-r .T'r

,.A- ulated in .the two models. Details and differences involved in the models are discussed in the attached appendix.

Both methods represent attempts to produce rational and systematic The approaches to a very complicated thermal-hydraulic phenomenon.

t models have helped to understand the importance in the spray cooling

.') phenomenon of certain parameters, such as channel quench time, channel

/"'.D film coef ficient prior to quenching, heat transfer to the spray fluid, u

• radiation heat transfer, and grouping of the fuel rods. However, several dif f erences in the treatment of these parameters exist between the models that have not been resolved at this time. Numerical eval-l uations of the Nine Mile Point reactor conditions have been made with both models for a recirculation line break with an assumed 60-second 3,

initiation time for the spray system (i.e., achievement of rated flow). These evaluations predict a peak temperature of 2180*F by the GE method and WYN$' 50$

1

'l (

~

. - . , ~ . -

h, . "

Md.s is w f .i.m w m,

^

_ h' . ,

t ,

t

. i.m ,

calculations and 2220'F by the INC method and calculations. This comparison of peak temperature alone, however, is not suf ficient to h.

W.; appraise the models as the time at which the peak temperature 4.s pre-

.y .. .

dicted to occur may also be a significant indication of the intrinsic validity of the model formulation. In the case cited above, the peak is predicted to occur 4 minutes af ter the break by the GE model and

/ "." 10 minutes by the INC model, a substantial difference. A similar t dif ference has been exhibited between the model predictions for the

.G ;. ' .

Zr-2K FLECHT data; the GE prediction underestimates the time at which

> the experimental peak temperature occurs while the INC model appears I

to be substantially better in this regard. Both models predict the

- i peak test temperatures reasonably well but the prediction of an earlier peak temperature is less conservative.

If somewhat more conservative values of the channel heat transfer dI '.

('(,) coefficient (h) and channel quench time are considered (reducing h from 20 to 10, and increasing the quench time from 3 minutes to 4 ' minutes), the INC model predicts a peak clad temperature of about 2500'F. GE ha; made a comparable calculation with the same result, but GE pref ers to calculate a "best estimate" by reducing some of the l

specif'ic conservatism included in their calculation of a 2180*F peak.  !

The "best estimate" peak temperature is about 2000'F. If the reduced 1 conservatism in this latter calculation were included in the INC cal- I y

culations, the conservative INC result would be reduced to approximately I

2320*F. Although these computations are useful, the uncertainties in i ^

l Q =_ _

Q ==

. - _ _ . _ .Q."i}= _

i

.m H, - ,

. . 4-t . , .

j. , - s

,a the model' formulations

require additional consideration; p ,y There are significant' difficulties in properly modeling the complex

( '

~

. phenomena : involved. in the' apt ay cooling phase of the Nine Mile Point BWR.

g The; different methods of extracting' and applying generalized heat transfer gif O .I-Lparameters from the' FLECHT tests as represented by the INC and GE approaches.

1

/ '; are not unreasonable, but some fundamental differences appear to exist 'at

(* 3:%'j: ' .this time which .have not yet.been resolved. At present, . then, we do not s..l-rely solely on the.GE model for.the evaluation of the spray cooling phase; ofi the recirculation line break in the Nine Mile Point LOCA, but also rely .

~

' .7a L on estimates of anticipated performance based on the more conservative 7 p ~

results ob'tained by the INC calculations.

.The calculated peak clad temperatures can be reduced by taking into

.(

account. the f act that the Nine Mile Point reactor ECCS is to be' operated j

and maintained so that the core spray system can achieve rated flow in 35 seconds or less instead of 'the value of 60 seconds used in the~ calcul-

. . ' .V

' !- ation. Representatives of the Niagara Mohawk Power Corporation have assured

-.f .: .

the staff that the required startup. and operation of the' diesels for this

]:

Under these conditions, we have concluded

j. purpose is. feasible and reliable.

t that the. peak clad temperature calculated by either calculational model

] i will remain below 2300*F; GE estimates a peak temperature of approximately I

2000*F, and INC approximately 2200*F.

l On the basis of a spray system which achieves rated flow in 35 seconds l1 or less, we conclude that the ECCS for Nine Mile Point reactor should provide adequate core cooling performance in the event of a recirculation line break at power levels up to 1850 MWt.

4  ;

4

• ~

m-,.--..,, -. , ,

. . . - ~ .

w ,-_

5-7'i'.'d VM .

In Amendrent.No. 5 to.the application for power increase (received 1/26/71), the' applicant has documented ' calculations by GE for small-break 8 1 9

' LOCAs. The calculated peak clad temperature is 2226*F. We are reviewing these calculations and will report our conclusions orally .to the sub committee.

'

..IJd.i  ?;

. :- l e

e:p 9

f 6

e .

l r /

va rswa.. a =_--_mueH

,g,

=

. % := wbfM 7 4 2 0 o(/ - 1 1

. P w9 1 '1 yk

_ _ rt T" 9 4 9 3 ai H2

. es C -

2 3 1 1 nn) E r 1 1 1 1 i eELZ LDCF" e n l" o

, u - d2i t

. - nn

) iura)

... BZi e

g-y a

r sr" s td i n  %

5 2

0 5

f 3 p eoT no ctHI c

, s xcC e 1 6 3 7 2

E aE ys 5 1 1 e eL a

. .~? e rRFr0 + + + +

s e p6 wPrS(

. A 0

6

(

t Po M- vt 9 OA e

. C T .+ u O w L

0

.- 1 T

N 5

8 n I 1 ro )

4 0 7 4 E O ei P L P T wt M .W 3 0 4 2 B A oa - (

k 2 0 6 3 A E Pi 9 1

. T L N yi t

I O M I an t T cI w

._ E A e Mt n n N

i

l. D y n ei I I a

• o5i n A l r T" ono N

V E

ao tS H2)

C W 5

4 0

0 5

• 2 4 ita46. Po di o E rk 6 7 5 i e nt F Tt LZ( 5 8 5 2 tfl aa

[: ,. - F A F" 1 i oi ,i A n M t T ir- 4i S e9 n

' s yw pi aor rpo u

oy i

rP 0 1 0 0 p f ra gr 1 9 8 sg

. eoM gt -

3 nt p 1 1 0 0 ti w nst

+ ac9 atM i s

• ra a t e eF t r5 sat v ae2 d Ag

• ep o ,e

. p inTH "2 7 8 0

0 9

1 7

h o5 r3l d

ekC - 9 ykd d pn oaE r reLZ 0 1 0 0 aan euu GPF" cea ep l ob i r d 2 a g2 t - f -

% nr oiZ nr

. n 2.

4 aI l

eZ w

tdt

. o .

vH oH i niC srC t E

- a 1 2 3 4 oui qI t

cdL

- dc p p p p d eF eeF oi e ll Rf u 'u u u o sd r fi e

- i o o o ano eah t r r r rg Rft n g g g P af e )

d 1

(

• I Il,

p;jl;1 ,

c - -

. a r e 5

- e pi n n P 6

. _- mT - 4 -

. . e 9 ^ 1 _

T h c

._ xn oo -

._ B u Q

l e&

n T "2 E 0 ne c- 0 ar Er LZ 4

1 hu Ct F" .

f

.m_

e d P 5 3

0 3

0 6

0 8

1 r - 9 9 0 tl. mu ut 9 '

1 1 2 2

)

y na ir 28 a xc aa T"

, r 0 2 p E2

_ s McT C -

2r

+

0 0

[7 8

'., e LZ 8 6 1 3 1

s e F" 7 1

7 1

0 2 2 3*. A O _

C 6 . . Ia ! i l;:-

O ( ,

L l _

T aF 0 5 0 T W i* P M

0 3 2 9 3 2

I M t 9 8 6 5 I i e - 1 1 1 ~

O 0 nr 9 1 E 5 I u L P 8 t B 0

._ A E L

1 ea gr f 3 92 U

T 5 "2 43 92 I T A

cn re C -

01 [7

_ M e-c Er 0 3 0 4 1 5 7 E N ve LZ 7 7 6 N O AT F" 6 1 1 I I 1 1 _

N T A

U _

L _

A P y) .

U V E hra. erc 9 vpe

~ F OSs F n A iet0 T

l a6  %  %  %

S sd (  % 0 5 sn2 5 2 .

eu cEro n

1 6 3 7 )

l _

1 2 x 1 Zi S e Er t

+ + + + v oTa e _

- rtHi l ._

ecCt waEi oeLn '6 0 e $

PRFI ,

( i e _

n

_ a _

l p

_ d i _

n m e

_ o -

. i

) ) ) r t )

2s 3 s 4s o

• a 1s d d c -

_ d c po d pod po po e

o i R f ur ur ur ur o h i c o o  : t t r4 g(

r0 g2 r6 g1 g(9 n ( ( t -

e A d

• I

!l f!

jyi fc l i (lllIL  :!

- e e.

o' g. y-- - --

.; M ys .-- m _

APPENDIX

h*x:4

' NINE MILE' POINT IDCA EDEL

'1.- For Nine Mile Point, the loss-of-coolant accident is divided

,,.4 into three time periods: (1) blowdown, (2) core heatup, (3) core i jg

.;y

, ' i. spray. 'Ihe blowdown and com heatup periods of the IOCA presented ,

. : ; .~

^ '

by GE appear to be reasonable. 'Ihat is, the use of the 1.8 second  :

L

%* 5 dryout time, the experimentally based dryout heat transfer, and no credit for steam cooling are warranted in this case. 'Ihe deriva-Qf:<  ;

- n.

- T' ?

tion of acceptable core spray models and their application to core

. f .'. h y,'y jg heatup calculations is of particular concern in this plant.

' 7" '" I. CDRE SPRAY FDDELS

. .p Starting with the FLECHT SS-2N data, General Electric and Idaho Nuclear Corporation have embarked on similar procedures.

.. e ,

?' Each has developed computer programs to " extract". generalized v:.

%d heat transfer parameters from the data. 'Ihese parameters were then used to develop heat transfer correlations. 'Ihese correlations for

m. heat transfer coefficient and channel wetting time are used in

.. J.

'vy,, ,

"L conputer codes to predict the results of the FLECHT stainless and I

- ,?

.7' zircaloy tests, as well as the IOCA for NMP.

U4 A. Extracting Core Spray Heat Transfer Parameters I$,4 is .

D The first step in GE's extraction procedure is to calculate f

,YQ

,i grey body factors for the 49 rod plus channel box array for a constantemissivityusingtheirGREYcode(forSS-2N,(=0.6).

'[p

.a

'Ihese grey body factors are then combined to ratch the desired q

grouping of rods for which grey body factors are reeded.

{ "'.k!biW a~ - ., 'h 3 i

_ - - - I

W

"' Next, the code FIUC0 is used to solve the rod energy balance, the only unknown now being the heat transfer coeffi-cient, since rod surface temperatures am known and the fluid temperature is assured to be at saturation. FILM 00 neglects conduction ud does not calculate channel box heat transfer i coefficients. For extracting coefficients from SS-2N the 16 l

" northwest'? cods in the bundle were considered and grey body factors cmbined accordingly.*

.. }

1

'Ibe INC extraction procedure begins by using the DATAR

. 6 .]! code to solve for a " total" heat transfer coefficient from

~

each rod and channel surface. The coefficient includes all energy leaving the surface including any radiation. It is 1 the total heat flux divided by the temperature drop between the surface and the same saturation temperature that GE uses.

I

) An inverse conduction model is used in DATAR. Since all

& 7. < experimental tenperatums are not recorded simultaneously, j the values used in calculating are interpolated fmm the data

' to the time desired.

'Ihe RAmT code uses the DATAR results and the experimental I

surface tenperatums to extract coefficients from surface to fluid. 'Ibe rod matrix used in RADHT is the entire bundle plus i

i

• This entire procedum is explained more fully in the appendix

. of GEAP-13086 " Heat Transfer in a Simulated BWR Fuel Bundle I Cooled by Spray Under less-of-Coolant Conditions," June 1970.

'***6**me m___.___...

u

, 4 V ,. .

. 1 I _3_

.4

!' l f'

the channel box. If temperatures are not available for each rod, its " mirror image'.' along the northwest to southeast. diagonal is W.l used. A temperature-dependent' emissivity is progransned into RADHT and the matrix solution for body-to-body radiation solved for each body for each time step. 'Ihe emissivity calculated

,g,y <

for stainless in RADHT is almost always about 0.89.

The INC method is somewhat more rigorous than the approach adopted by GE since it includes a conduction model, a temperature interpolation routine, a variable emissivity, a radiation

}

solution for the entire matrix for each time step, and extracts channel box heat transfer coefficients. In general, the INC

[

method yields lower h's than the GE method especially for the

]

outer rods and the channel box. For these reasons, it seems a i

more conservative method and, therefore, preferable at this 4

time.

I B. Correlation of Extracted Heat Transfer Parameters t

GE was able to correlate their results as a temperature function (Roger's correlation) strongly dependent on rod loca-tion, the outer rods yielding higher coefficients. INC did not find a strong temperature or location dependency but rather a I

dependence on channel wetting time. After can quench, INC i

h's were between about 0.5 and 1.5; GE values were between

j. )I I about 0.5 and 10. The ability of GE to correlate well 4 j with the temperature function (T4-Ts )/(T - T s) indicates a 4

5 4 , , , -

- - p

l m

.~

2h' 4

1 strong radiation component of the heat transfer coefficient .j Ihis (T, is the sink temperature, assumed to be saturation).

component is too large to be accounted for as radiation to j}' I steam or droplets, so GE claims it is radiation to the water film around the wetted channel. Even though some radiation is now absorbed by water surrounding the channel, an appro-priate decrease in the amount of radiation to the channel is

" not made. That is, body-to body radiation is not properly

. ,;". w- Conservation decreased to account for radiation to the film.

of radiant energy requires that the sum of all shape f actors d

for a given body to all other bodies is 1.0. This is main-tained prior to channel wetting. However, after channel n..

wetting, the radiation portion of the Roger's correlation 1 INC causes the sum to be larger than 1.0 for all rod groups.

, ..i r

estimates that in the GE case of a corner rod, the sum of

.1 the shape f actors can be as high as 1.5 to accommodate r.he l ,

additional rod-to-film radiation. It can be argued, however, that a high total view factor is compensation for a low value of emissivity. But without a calculation to demonstrate that hypothesis, the issue remains unresolved. The procedure of l

allowing an excess amount of radiant energy to be transferred

' r .3 4

.s to a constant sink appears non-conservative. i

.. 4*" INC did a parametric study of the effect of emissivity l using their method. It appears that about half the difference

4.

a .

>J I

---n-,

' 1

'I

- aW..

1 in extracted h's between the GE and INC methods can be attri-buted to the values chosen for emissivity. The intrinsic method of extraction then appears to account for the balance of J.b  !

'~ the difference.

GE has correlated channel wetting time as a function of vn. 3. . 4

' a Yamanouchi* parameter containing channel wall temperature and temperature gradient. The GE correlation with this para-

. ~l meter is linear, although the data are insufficient to be

. . .g .

~ definitive and a less f avorable correlation conceivably could be supported by the limited data. On the other hana, the approach developed by INC requires that the wetting time be treated arbitrarily and handled parametrically in their pre-dictions. In formulating the heat transfer correlation after  !

- . .(

spray initiation but before channel wetting, GE calculates the

.~ .a '

radiation to the water film component and uses a fraction of the i

fully developed coef ficient as expressed by the ratio of (t - t3)/ ,

(t g- t,), where t is the time at which the coefficient is being calculated, t, is the time of spray initiation, and r q is the channel quench time. Note that t,4 t c tq. Neither the INC nor the GE data correlate too well with time or temperature I

bef ore channel quench. This is of special concern where tempera-ture turnarounds are predicted to occur at about the same time

[

I l -- 1 *Yamanouchi, A. , "Ef fects of Core Spray Cooling and Stationary State Af ter i

LOCA", Journal of Nuclear Science and Technology, October 1968.

C __

~

4tt y as channel quench; that is, most of the transient has occurred during this period before channel quench. The diffemnces in correlation have given rise to different interpatation of the'

.. mechanism operating during this period. Inasmuch as the heat

' ' transfer inproved prior to and just 'at channel quench, E Li+d believes that the phenomenon is due to inproved emissivity of a .in '  ;

the wetted channel. INC, on the other hand, attributes the l

,. / g .

improvement to improved convection due to splashing and slaking I' di of the water from the quench front. Without msolving the

?H.y differences in the two calculational models, it would be diffi-n~d N

cult to coment on the interpretation of mechanics.

'h( ,

• J If just the inner can face is considend, E estimates heat transfer coefficients of about 20 for the channel prior tc N' J L can quench. INC actually extracts coefficients that are abrost g

always less than 10. The entire E correlation which allows an

[Q additional amount of radiation to be transferred to a low tempera-

'*U I The ture sink prior to and after quench is non-conservative.

j i

higher channel coefficients which msult in lower charmel X temperatures provide a better radiation sink for the hot rods.

i ,.  :

Most subsequent calculations including some where rod coeffi-Ci 6 :n'~ <

1 cients were zero show that radiation ultimately to the channel i 4 i even before channel quench is what arrests the temperature transient. Therefore, any mechanism which msults in lower i l

e-l l

l

. 7-

.:. l channel temperatures prior to quench should be evaluated most conservatively.

II. HEATUP PEDICfIONS_

' E uses the code CHAST for bundle heatup calculations. The code can be used for predicting a IDCA or results such as FIECHf.

R.M';

'Ihe 49 rod bundle is luTped into 4 groups plus the channel box

- for ease of calculation. Group 1 is the 4 corner rods, group 2

' 4 is the outer rows minus the corner rods, group 3 is the second i

rows, and group 4 is the center nine. Grey body factors are l

1

?

appropriately :cmbined so that each group and the channel has one interchange factor with each other group, the peaking factors are averaged, and each group has a corresponding heat transfer corre-f

( lation extracted from the EWR FLECh7 tests (Rocer's correlation) .

INC's revised ! OXY code does the heatup calculation. 'Ibe 49 rods are treated individually in contrast to the simplified GE approach, i.e., individual peaking factors are used and the entire 9

1 radiation matrix is solved for each time step. This allows varia-

.; tion of emissivity with time for each body. GE uses a single 1

The INC emissivity for the entire matrix for the entire time.

method then is more flexible and rigorous. Thus far, INC has treated GE's groups 1 and 2 as a single group with respect to heat I transfer coefficient correlations.

9 NON bN NIM h$

yn y n - n __= g r u umu

- ~ ~ - - _ - - - - _ _ _ _ _ _ _ _ _ _ _ __-

p .,:

.- -= ,,

&n . -

--:~.~-

n- .a w .__g-

" _ 8-

.. a.(

FLECHT Predictions g

A.

Using their respective codes, both organizations have done a good job of predicting temperatures for the stainless steel t

.. X s FLECHT tests frcm which the data were extracted. Agmement in

' h^ .a. this "mfitting" process is necessary but not sufficient or

.' surprising. In trying to predict the Zr-2 tests, both do g . . . . .

surprisingly well in view of the extensive experimental diffi-

%y ,

')- culties discussed previously. INC's peak temperature predic-e

.! tions are generally closer to the experiment than E's, usually 7,s::,

slightly higher than the experiment but lower than GE's prediction.

-l The tine of turnaround is also usually better predicted by INC.

E's turnaround times are almost always too early by significant g

- factors. The shape of the I:iC tegerature curves are generally i flatter around' the peak as are the Zr-2 data, but usually this 4

is not the case with the E predictions which tend to be more

-.e.\

c, sharply peaked near the maximum tenperatums.

B. Nine Mile Point Predictions In predicting cladding tenparatures for NMP, E used the CHAST code by first appbing their experimentally based dryout correlations to the blowdown, then an " adiabatic bundle" heatup from end of dryout until the core spray maches rated flow.

i

" Adiabatic bundle implies radiant energy interchange among the Iod groups and the channel walls but with a convective coeffi-o

^

cient of zero. The E Roger's correlation was then used to

_ -m y g--hmmup M ia u.;m a m u iq J

g .

9.

. .a n. .

4 model the period of spray cooling which includes the channel

.)

s wetting time correlation. The results are presented in the NMP addendun. 3 cited earlier which indicates the peak clad

, .i,,,,

temperature of 2180*F for the largest double ended break and 2225'F for the. worst intermediate break of 0.14 ft2,

,;. ,5,.;,

INC used the MOXY code to parametrically study spray cooling. effectiveness in NMP. The temperatures at time of g

- spray initiation as calculated by GE were used as initial T ~'4 temperatures in these calculations. The results are shown

} The result most nearly equivalent to the GE in Table 3 predictions are for a 60 sec spray initiation time,180 sec

.h.

quench time, and a channel h of 20. GE predicts a peak of 2178'F with a quench eine of 172 sec af ter spray and a time 1

of peak at 154 seconds af ter spray. The peak temperature by

,a -

, M0XY is 2220*F at 520 seconds af ter spray. The longer times

. predicted by INC are ccasistent with their Zr-2K predictions and the Zr-2K data. The longer times and flatter profiles ..

are consistent with lower heat transfer coefficients which <

l are not strongly temperature dependent. Since INC's extrac-g 1

l tion procedure yielded channel h's generally of 10 or less, l whereas GE asserts that these values should be in the order l

I; of 20, identical calculations were made with channel h's of 10 and 20. In the Zr-2 test, 5 of 6 T/C's on the hot can f aces ranged from approximately 1200*F to 1400*F prior to

, ym e u ai ggy

,' PTT 41TT M " _^"--?-_-:__- _

t_r= tv - Tmjil M R.a R Ih sju O l h -IE '

.(

I * -bd _ _

_---_- ---- -wue wa - a  ;

quench, 3 of those 5 quenched at between 3.7 and 4 minutes, l.. the other 2 quenched at about 2.5 minutes, one of those was just barely at 1200*F at quench. The sixth T/C which was about 1000*F quenched at about 3.5 minutes. GE calculates a

, /;,

channel temperature of 1467' at spray initiation for Nine Mile Point reactor which f alls to about 1350* at quench under

- the influence of the constant channel coefficient of 20.

When INC uses a constant coef ficient of 10 during this period,

.t

'1 the can temperature does not fall. Under these circumstances I I

then, a quench time of 4 minutes appears quite reasonable.

The 10XY prediction for this case of a 4-minute quench and a can coefficient of 10 yields a peak of 2496*F. Re-calculation of this worst case using the calculated GE clad temperatures

{'

' at 30 seconds rather than 60 seconds to simulate earlier spray j

initiation yields a pe.ak value of 2340*F. This INC value is artificially high for a 30-second spray since the channel

) wetting time was not reduced from 4 minutes to a more appro- j priate value of about 2 minutes for a 30-second spray. There-fore, it would appear that a 30-second spray initiation will successfully arrest the clad temperature rise with reasonable choices of channel parameters. In fact, the INC prediction l

using a 3-minute channel wetting time is only 2268'T at j 49 sec.

i emmwn +~ ~mm m m nr

) Q)_T }_ ' & _5x n a k25tb[. ll_ _3-u

- ir . - _

" "r - wruw %%u.o wWu

.s..,

T w. ,

m. ,

III. ~ NINE MILE POINT MODEL EVALUATION

.?*

As was the case with the Er-2 predictions, the INC temperatures

'for NMP have a flatter profile near the peak than the GE predictions.

@9 Also, as in Zr-2, the. predicted ' times are longer in the INC-MOXY Although a peak may be predicted at about 500 seconds predictions.

' .-.t@ in MOXY, it typically may have been within 100*F of that peak at

..$ 250 seconds or about the time the GE peak is predicted. Experimental o

work at Oak Ridge

• and INC** suggests that extended periods of time, N, .8# like 8 to 10 minutes, at modest temperatures, like 2200*F, may be j

just as likely to cause brittle fracture as a short time at a more elevated temperature such as 2700'F.***

k If the total amount of heat transferred at the surface of a

'.a. :  !

rod is about the same in the INC and GE extraction methods and both l

have about the same amount for " pure" convection crnonent, then the GE correlation which has a larger body- to. fluid radiation d

i

'., component is non-conservative. As temperatures are elevated, the strong tempr.tature dependence (roughly3T ) for the Roger's corre-

-lation is removing more heat from the bundle. Although the body-q.

,- to-body component would be larger for the INC extraction method,

d. both prediction methods used the same emissivity for zircaloy.

1 *Rittenhouse, P. L. , Progress in Zircaloy Failure Modes Research, ORNL-TM-3188, December 1970.

j

• • Herzel & Meservey, ' Trans. Am. Nucl. Soc. 12 (1), 355-356, 1969.

• • • Using Baker-Just Kinetics, 8 minutes @ 2200*F or 40 seconds at 2700*F locally reacts about 10% of BWR cladding.

i

.s tftP = = ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ' _ ^ ^ ~7 u ,

9 wraaswni'au. V = w SSA-

'* O* l

_____________J

c -

e,.- -- - ,,

.. . .[

' D'n-d-wwn e- -

• :[1 - ,,

12 -

~

4. u :- .

Therefore, the body-to-body component is not as such larger. as the

- INC predictions, if at all, than it was in the extraction procedure.

At any rate, a large body-to-body component does not remove heat s,..

.from the bundle but merely redistribute 44 it within the bundle. A strong temperature dependence for the body-to-fluid radiation

'*1&.'i component seems to be a non-conservative self-limiting feature.

e p..

4 e e:

.=0 .E 9

,',7* ),

T e 49

. fe <

. ,' .Y*.

o

1. e4 se l

l p

i U -M"H k I[IPTNY '

. :nLi*Hu ew-. - - . .

~

- T ey Pa r 0 0 0 0 0 0 f p 1 2 4 9 6 6 oS 2 5 2 4 5 2 er me s it Tf

,. A F

,e -

s

, d e ar 6 8 8 0 l u 7 0

.. -. Ct 1 2 9 6 4 4 a 3 2 4 2 2 3 kr 2 2 2 2 2 2 5 ._ ' - ae

• ep Pm e ~

6 T

- T N

,- I

.s. O F P

• k* 2;u .I r2

'r.,. E eT

.- * ,a 1

L fF

. I s -

M nR

-- aH E r/

N TU I T 0 0 3 N tB 0 0 0 0 a- 1 2 1 1 1 1 E R et L O Hn

- B F li e

A T S ec

a. .

N O

ni nf af I

. T he C C o

. I C D

E

~

u. .

R P

e Y m .

X ie O TeS

- M h-cy na er 0 3 0 0 3 3 up 8 8 4 8 8 4 QS 1 1 2 1 1 2 l r ee nt

- nf aA h

C n

- os id

- t n ao i c t e

. iS 0 0 0 0 0 3 n 6 6 6 3 3 3 I -

ye am ri pT S

iJ!

ell j !io