ML19268B899

Review of Info Needs for Design of Magnesium Oxide Core Retention Device.

ML19268B899
Person / Time
Site:	Atlantic Nuclear Power Plant
Issue date:	05/23/1979
From:	SANDIA NATIONAL LABORATORIES
To:
Shared Package
ML19268B898	List: ML19268B899
References
NUDOCS 7907190014
Download: ML19268B899 (15)

Text

7 I IR i

i i

l S

i

REVIEW OF IfiF0F."AT105 NEEDS FOR DESIGN ,

I a ,

i j

OF A Mg0 CORE RETENTION DEVICE  :

1 1

l SANDIA LABORATORIES l MAY 23, '979

-i;

'b 1

9500044 i,

+ 79071900/

*****-*~~*N

• g*-. ,,, , ...,,.. .. .. -= * * ** " a- -= ' /

d~

t Infor a tion ::aeds for Core 0.etention revice Design and .'e : I y s i s

,Section V of offshore 7cwer Systems Topical Report Number 26A59 does a creditable job in identifying areas of uncertainty roncerning melt / core r etention ma terial interactions. \The reviewers ve~re able to identify some cdditional concerns and had ccm ents on items cited in these docunents. -

Prior to discussing information needed for design of a core retention device, it is important to realize that inclusion of such a device in a power plant would have an impact on the entire ,

cour se of a hypothetical meltdown accident. For instance, a core d

• ajor conclusions of the review are as follous:

1) The documents submitted for review together do identify mos c areas of uncer tainty. The most important of these, i 1

i and the additional areas of unear tainty identified by  ;

the reviewers, are felt to be:  !

l

. a) crust formation and upward heat flux from the celt, l u

b) oxfoliation of brick layers in the retention device,  !

thermalhydraulics of the molten core materials, l c) .

d) mechanism of melt attack on the refractory material, and e) influence of retention device geometry on local refractory erosion.

2) The ' design and desigr. analysis of the core re tentien device submi tted by Of f shore Pcwer Systems places an unjustifiable reliance on the Icv temperature experi-ence of the steel industry. Other relevant industrial experience does not appear to have been considered.

I 9500<1345

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

retention device vould greatly reduce the rate of aerosol gen-Sotstantial

- eration by gas sparging of caterial fro: the celt.

reductions of aerosol genera tion f r o non-fuel sources ..suld occur. At the same time a variety of heat removal techanisms available 'to the melt vhile in contact with concrete vould not Aerosol be available to a relt within the core r etention device.

generation by vaporization of fission prciucts from this botter melt would incr e ase. The net result mruld be a decrease in aero-sol generation and a decrease in the rate of aerosol sedimenta- ,

tion within containc.ent. The aerosols within containment vould i f

come primarily from fuel cources rather than non-fuel sources as in the case of celt / concrete interactions. '

The reviewers did not attecpt to identify colla teral incacts  !

i on meltdown accidents caused by the inclusion of a core retention l Such de terminations can best be done in conjunction with  !

device.

vith accident modeling such as that being performed at Battelle ,

Attentions I Memorial Institute.

were directed, instead, toward identification ci design informa- l tion necessary to meet other goals of a core retention devica, ,

namely retardation of ex-vessel melt movement and gas genera ticn.

1 A) General Co ents from the Reviewers

1) The heavy reliance on steel industry experience is not war r anted since the temper a tur e r anges involved I

l in steel manufacture (1350 - 1700*C) are on the I

' lov end of u a melt te..perature range expected

~

i during a light-water reactor core meltdc-n 3ccident (1350 - 2500*C). Torperatures cited in the Offsherc I

l 9500g.54 6

t. y,,.. .

.v -e m e . m a . :: ,.,. , ,..q .,.. < a, y r.... m ...... .s s

v. w . m ,e u eg.a.,,v. .. .

r< ~ ra .n . v . . s -r .

4 .,m .. . . r :. . . .o.

... . .. .. ..... ., c

Power System report corprised the reviewers since they seemed quite low. During the fir st day of a a.e l t/c o n c r e t e interaction melt tenper atures f all rapidly from about 2600*C to about 2000'C. For the next five days melt temperatures smoothly decline over the range from 2000*C to 1700*C.

Many of the heat removal mechanisms available during.

melt / concrete inceractions--such as convective heat tr anspor t by gas gener ation and endothermi,c decompo-sition reactions of ' concrete--are not available during melt interactions with core retention materials. The melt temperatures ought then be at Ica st as g reat dur- ,

ing melt / core retention material interac:icas as these i encountered during =elt/ concrete interactions.

I Melt temperature becomes important because at 12ast a portion of the refractory erosion expected during ~

melt / core retention material interaction is due to chemical reaction. Since the rates of chemical reac-tions are sensitive and non-linear functions of tem-perature, it is most hazardous to ex tr apolate encoura-ging experience at low temperatures to core meltdown situations.

t

2) Heat generation in the melt and environs was restricted j .. .

• o fission product decay hea t. No c0ns'0*r3 tion was i

given to heat produced by exidation of retallic phases 95002347 4 .. . _ _ _ . ..

.m . . . _ . . _ .

" * ' * " ' ' *h*8m. .'+a.ie>+e.=*. . - . , , , , , _

of the core melt. Saukal et al.,* have sh:un that heat due to oxidation of zirconium and chro:ium in a core malt can be significant in e .parison to fission product decay heat. Generation of this heat mi.e~ht be slower--and consec.uenti.v more crolon=.ed-- .

during melt / core-retention-material interactions than Juring melt / concrete interactions since gas tr anspor t through the melt would be more limited.

H0 wever, the oxidizing environment of a light-water reactor accident do,es assure that this chemical heat source will be available.

3) A significant source of industrial experience was neglected in the Offshore Power System (CPS) report--  !

the glass-making industry. Though again the tempera- l t

t ture regimes this industry empicys are reuch lower than  !

i the core meltdown temperature regimes, the industry i I

has had to deal extensively with oxide melt / refractory i

interactions.

4) The reviewers did not feel competent to address questions concerning mechanical damage to a retention -

' device as a result of debris impacting the device.

The reviewers felt that industrial ex,0erience citad in the CPS repor t was par ticularly per tinent and

%'. Saukal, J. Nixdorf, R. Skootajan, and F. Winter, "Irvestiga-

- tion of the Relevancy and the Taasibilitv of M2asurement of

[ Che:-ical Reac tions Dur ing Cor e Mel tdown on the Integral Esat Content of Molten Cores," 5.M ?T- RS -19 7, Eattele Institut., e.v.,

i f Frankfort am Main, ?.E. Germany, June 1977, English translation N'.' .;I 3/T R - 0 0 4 7 , October 1973.

9500048

. . ~ . . . ~ . . . . , . _ _

. .s . . , . - - . . . , . . . . . . . . e s ~. _ : .; o r- . - 3 7. . .- - -; s . ,.,;.

,,_,,3.,4 ~, ,----.-------: ., ,.

. _ . . ~ . . .. _

realistic to these questions of mechanical damage.

' At some point re: ore definitive data than the anecdotal

' accoun ts in the OPS repor t should be cade available.

5) The reviewers felt that chemical interactions of oxidic melts with ref ractory bricks were not adequately treated. Little data are available on this point.

~

Cata tha t are known indicate .that assumptions of uniform attack and neglect of melt convection may be seriously in error. The reviewars agreed with the OPS report that the nature of the chemical interac-tions was a major area of uncertainty.

5) Little data are available concerning celt behavior under conditions of interest. Crust for:Ition over the sur f ace of the melt or other phenomena that would impede upward heat flux from the melt were neglected in the CPS report. The reviewers do not share the OPS confidenca that more cocplete understanding of the i

cur f ace behavior of the melt could only lead to a greater margin of' safety for the r e tention device.

i bs e

A'hy r educ tion in upward he at tr ans f e r r a te from the i.

- malt translates into greater erosion rates of the

- retention device. -

I i 7) The OPS retention device has a cinimun ver tical thickness of S feet 3 inches. It has a Tinicum

, lateral thickness of 3 feet 3 inches. Since, to i

! a first approximation, erosicn rates in the vertical and la ter al cirections due to thermal attack ara 95003349 I

. e == - ..aspe esas e omkam _aa e e.asm a g a= -e -

• ~ * - * * * = .-,-w. .,qm,,_,,

considered equal,, i t appears thai the sidewalls are

. the weak link in the design. La te r al pene tr a tion, ra ther than ver tical penetr ation, would be the expected failure mode. Prediction of the failure time must then include the effects of both chemical and thermal erosion by the mel t and the effects of material streng ths at elevated temperatures. .

Cata provided by CPS indicate that the core retention device would fail laterally in no more than 60 hours6.944444e-4 days <br />0.0167 hours <br />9.920635e-5 weeks <br />2.283e-5 months <br /> even if only 50% of the decay heat were to go into refractory erosion.

4 B) Areas of Concern Sot Considered by OPS ,

- 1

1) Reflooding of the melt by water was not considered i I

i by OPS. OPS did cite unsc.ecified emargencv. actions i should a meltdown accident occur. Celiberate reficod- ,

ing of the celt could be a..ong these. Re flooding has -

been indicated by accident analyses (P. Cybulskis, Battelle !!emorial Institute, Columbus, Ohio) to be a hazardous under taking. Reflooding, whether deliberate or accidental, has not been experimentally studied and appears to be an impor tant area of uncer tainty.

2) As a corollary to 1) O?3 d id no t cons ider -pr essur e g en-era tion produced when melts contact a water-saturated retention device. Since the '!;O bricks described in the CPS report are about 17; porous, they could retain 3

as much as 2030 ft of us ter. The only design festure L

95008350

.e a=*==**

. - . . . . . . . . . .n e..

.....e_s 4 3 . g.

.A. * ****~~:'

,...n,p.,,......e..g..

,.=.~;.,..,,...~..=.-.

• --*=.~.=.....-**..o.

" ,oo -s ,y _,_

to assure that water does not become entra.ac-d ia d.e bricks is a 1/4" steel liner of uncer tain description.

- '/aporization of entrained wa ter could be a s ign ificant source of containment pressurization. The reviewers are aware of only a single, scoping, tr ansien t experi-d cent in which a high temperature me lt was s reame t onto a water-saturated brick. (D. A. 70wers, Meeting with Experts on the Technology of Sacrificial Mater,ials for Delaying core Melt-Through, Augus t 29-30, 1978, Sethesda, MD) This transient test indicated only rela-tively smooth vaporization of entrained water. .

3) The CPS design of the retention device includes tongue- .

and-groove bonding of r ef ractory bricks to pr avent brick flo a ta tion . This design will function satisf ac torily l

}

only if an entire course of bricks remains intact. 1 I

Should localized attack penetrate a few bricks, the entire course might exfoliate and float to the tcp i of the pool. This uncer tainty adds special c:phasis >

to the uncer tainty of localized r ather than uniform l

i I attack on the refractory.

l ,

! 4) Creep of stressed refractory at high temperatures i

! was no t addressed in the OPS study t'aough they pec-vided data for ref ractory creep a t icw temper atures

(~ 1500*C). Creep of high purity MgO is significant

.)

at 1300*C and at 1:wer temperatur es for -a terials of j ,

lower purity. Creep rates are expenential functions i

i I.

I 5

95009351

. gr y p,,, , e*',4 *. 3 ..s...

_-.4 . . eu.e = *.- = y seew. *=s-i. e,s. . e f Aw 64m es wa w m =W . . *.= e .. qw

e. -e

, w =7 - .e . . . . . .

e , . . ; . .. y ey g .e * .

• -e =.v....,.ss...,.,.e.,wyw-....im,......,...,..-........,,

.- . . . . , , s, 4 gw- .e.. . . . _ . . . , . - , . , ,

d of temperature And sensitive to composition.* The core

- mel t places sur f ace bricks in the retention device under loads of about 4 psi. Sricks on the bulkhead a

valls above the retention device are under loads of abou t 18 psi . At 2000*C these loads are sufficient to produce significant deformation of refractory struc-tures. Should the bricks be contaminated by solid , state dif f usion of melt materials into the bricks, even greater creep rates may develop.*

5) The OPS report negrects thermal-hydr aulics of the me)

The analysis of Mgo erosion is conducted by a thermal ,

ablation model assuming uniform attack on the refractory. 1 The reviewer s could find no basis for this assumption.

Quite the contrary, available data and industrial  ;

experience suggest that localized attack is a major mode of r efr actory erosion. Some photographs of refractories exposed to glass melts are shown in Figure 1.

I f

The most important variables in determining the rate

' of localized attack appear to be gecEe try, melt ccm-position, temperature, and fluid phase convection.

Another uncertainty related to melt hydraulics is whether small perturbation 3 in the refractory i

8 i

s t

' E. Yasuda a'd S. Eirvia, agigh Te.perature Creep of Mygne5[2 Wfyh j Minor Add.itives,* Proc. Cxy# ;?_s EefractaireS 90"T f_i,t_iOI33 ,tE$E11~ _

{

ticues de H3ute Tt psrsture, OdeillO, ::ance, vune &c- ss-2 -t

. mm%

.1977.

2 95006.552 t -

.c.~.-- .-

awha e., efagen.-= w +me-. *,am%se. ._og.y._ =

. . . su

• ~ s *.- - . .

• n ' < y t t +% mc -- m'4 "

• ;:.:*.p:;. ,7. .,Tj .s.. f. ,uA. ,w.,s* a w y., x. - . ~

p-. .-.s e .- en.

. - . = ~y*- ' . ev.' e .

pt 4

e.s's' r c. '* M'.vr . 's :.c-  ! ,

• ['Y *..: jiy  :.. N '

f' l$ ,ll !!Id.3 M'_ '3 j',,?, ,.;.,. sf' * . 'f.

. h.M. d, ! ..

1...'

b, .\ s . li ). .f'. lJ \.(;

c i

L m .' 4  :

?

,-. ,sN ,... 4. .i x .. 4,,..l .9./.s

.A

. g l !1, ';

,e s ..

1 4.,

i,.

' , ..c , r h.

j s .

s.i,.. f g. 7 e

,,'. % ' .'s ..'~ ;; . ::

• ,- g ,(..

3 . l.m* % , '1

." d ,(.R E:s.e

<J -/.i ,4 y't , '< .

r i

r ~'

3 '.',N5/ C;'=9::2 4 ~- :~ *.-:-.c ~, : - Y.'.- {f .

bitij . N's...,.. .) .: ' ,. , '.h.

i i

i .

. \. f ,s . A'.J Qg e 'a

.t ~~ i d U; ard.ta:ing s u r'a == frw "t

a l ,

f ., d y l

v.h;;h the c'a> deant.g -4ss s u a;p :r.t'y

.d.k *. [ b./,Id, '.I, . .). /

. , t, J r.ept as fast a formed, with the resJt 6t

. 2 .. .

a) Note stepped crosien e.nd enhanced crosien at ioints a bet..een bricks.

. . _, . s a w - - - .r ,

i p 6v g 57 . ; -

1) .

., a.

. k..,; e 43 ,,

.u. 4 .

' o A.r, p N gn.:..s~,, .

e p. '. A,,. , , . .g . i 't e,

g ;;

oy p, t '

,j

,11- ,

d ' M.: .' 35 l,-

- '.y; <,.t gp i ..

%.w.

• N u.49is. Oy . /d 4, ,- f.rf .-

. t M.;g . ',:. . x . .,!d..C. M W ;-l'

• g [ ,-

c. # *

.'~ d,} p 1 ',., s.. "' 3 *. Y.. . . M.4

.T :.

>. e s.'ra.'

1. ' se '-

• v ~

( f f .;b.}.6l'Irhlgy

' l.*I *qS.,1n.. sL ,s .M e-s fjJ'h,NV.-gl M?

~

, I Y.*t..(*n & - , p,

,.. t<. ' &s

.; ~i'z!. ll5 h,.D E?

< . d.%

g rl,,a e.1 . o y , t.g.,Q. ve .Mr.u ;;;fq,9 @)

~ '

.-.- N. $

w% , .

.- u i .

.-=(.s.......e','.z.v'.a,

,. .=/ e. . ;.</'. ~ c 2

, ~.

• . . a_ * ' v.f -(s. .

t3 e

.g'.

c ..

.: .., , .r p 2c, i

• _* L . p;s ,;

e* . j ., -

h d{ a w,,D*]*( $((JJ

~

&. G % .

A . -

i b) An Ex mple of Local Attack These two b:xta dde by side ic .e.,ri:e, the face of the .d;*3t hand b'xk did not crael ar d shewed p.-acti:2"y nc, uting. 'I'he face of the :elt hand t':<:1 shrvoi a*.4 cr.n 1 e4 Ladly in ust: upward eating star:ed in these defec:s and p acti:a"y ha'f the i 1.lA s a. dirois ed a s ay; dd a!! b*a:ia, tottle g'asa tar.k.

9 I

~,

. Figure la f rc: . F. C. Flint and A. R. Payne J. Amer. Cer n. Soc. 9,_ 6 1 3 (1926)

Ficur? lb , c from D. 1'. Ross a

2 J. Am e . . Cer s. 500. 9,641 (l? ::6 )

1 1

F i g '.: r c 1. E X 3.~ p l e s o f Ch e~"ii C a l A t t a c'<. C n Ee f r a c t Or i 3 3

.' by G'. ass l2lts 9500%;S..,.)

e

%., -. . _ . _ 7 e < a. .p o. . w ,,, w , w y _

.. .-==....- -.. ...... .. . . ,,. .

o. .me.,. ,s.+4=-- m . p. se

21-

- surface grow pre'ferentially or are healed during

< fur ther attack. Industrial experience su;;ests that

< both results ate possible.

6) Volumetric expansion of the molten pool is not con-sidered by OPS. Crude calculations by the reviet.'ers indicate that the expansion is not likely to be sig-nificant i f no additional material falls from the reactor pressure vessel or the bulkhead valls into the molten pool. The ef fective volume change associ- ,

a'ted with heating a 17% porous brick from 25'C to melting

• and a'ssuming a' volume change on melting of

+5% (exact value is not known) is only +0.6%. How-ever, if steel from the icwer head of the pressure vessel is added to the melt, the me,lt volume would increase by at least 24%. Oxidation of metal phases in the pool or collapse of bulkhead walls would fur-ther expand the pool to the point that little safety margin would exist in the core retention device.

C) Areas of Concern CPS Treated as Adecuately Understood j 1) The model for Mgo erosion used by CPS was a sixple i

i thermal energy balance using the classical steady I

! state ablation formulation. Heat flux applied to I

the, refractory surface was treated as simp'le frac-tions of the fission product decay heat (see I-A-1 above) and were independent of the thermalhydraulics of the melt. The model neglected any chemical comp;nent attack on the r ef r actory. The maiel 9500dL554

.. ,,4.f..y_ . .,..

. . 3

. - .. -- . --,w---- , . --7-, . . . , ,

..g6. 4 .g .-w. ,,he.g,=. - - - . ---.... -

conse:uently assumed that the refractory must b2 heated to a critical melt temperature before it tras eroded. The reviewers could not ascertain the basis f.or the OPS confidence that this siaplified model of refractory erosion was verified.

2) The tongue-and-groove construction used for csr m-

~

bling the core retention materials does appear ade-quate to prevent brick floatation provided:

-- all bricks rema,in in place and exfoliation of the brick layers cannot occur (see I-S-3 above).

-- the tongues do not shear due to thermal or mech-anical shock.

3) Thermal shock of refractory bricks is most definitely  ;

an area of uncer tainty. All tests to date involving  :

i prototypic melts deposited on MgO bricks have been of I

a transient nature. In every case the bricks suffered ,

i catastrophic fracture after the melt solidified.

cecause of the transient nature of the tests, it is I

' not known whether the fracturing uas due to surface cooling of the bricks or delayed heating of the brick interior.

4) Upward heat flux from the melt is,an area of uncer-tainty. Crust forma tion, or spalled refr ac tory float-ing on the melt sur face will depress the melt surf ace temperatures and consequently the upward heat flux.

Heatup of valls and rea: tor internals above the melt will also depress upward heat f_ex. Any reduction in i

\

95101st S5 e

1

--.- . ~ .... u e .eam-g n._s.a.er. mmmerm s ** t-e py. p*v:,y,;-

" "" e--. + e Wm e ,4, _ , ,

upward heat flux results in more heat being available for ablation of the core retention material. ,

P.;0-basaltic concrete interactions were not cddretsad 5) in 'the OPS report. These interactions may occur at the bulkhead wall coated with 4" Mgo bricks. Once basaltic concrete under this coating reaches 1100*C it will begin to melt and the MgO coating will lose i ts s tructur al integr ity. (See also Section I-3-4.)

Molten basaltic concrete vill be f ree to flew into the core melt pool .and to attack the core retention material.

D) Areas of Uncertainty Considered by OPS The reviewers agreed with the listings of uncer tainties presented in the OPS r epor t. These areas need not be discussed f ur ther here. The reviewers felt, however, that there might be some misunderstanding of oxide chemical attack by liquid oxides--sometimes termed " slag-line" attack or " flux-line" attack.

fI

! Chemical attack is mass transport dominated erosion of the i

i

?

r e f r ac tory--as opposed to the hea t transpor t dominated erosion considereo in the OPS report. Dissolution as opposed to ablation

} ,

\

! of the refractory occurs because of favorable free-energy

) rela tionships among constituents of the melt / refractory sys tem.

i

! T'.;e r ate of dissolution is given by expressions of the for::

\,

J I

I i

.i 9500'4356

. :t

.. . ...:. . . :- ..; m-:zv:.-- ..- ::":,.;, ; ::-c;gy;g3qm;;3=7q-7 p. .:.:7--?.

g'7.-

rate = ko(C,T) exp (-E(c)/RT) (C-C3(?)]

where T= temperature C= fluid phase composition fluid C3 = saturation composition of the K and E = kinetic parameters dependent on temperature 3

and fluid coeposition ,

Rt univer sal gas constant chemical attack is important,.because it can occur at low tempera-tures (even in the solid state) and it can be responsible for i non-uniform attack.

The rate of attack is very sensitive to tenperature. Because of the non-linear nature of the rate expression, it is difficult  !

to xtrapolata data from low temperature experiments to predict '

high temperature behavior. Further, the above rate expression refer s only to the net dissolution of refractory. Erosion of  ;

solid refractory can occur even when the net rate is zero, pro-vided pr ecipitation of r efractory-containing species from the fluid phase occurs. When these precipitated species are of low density--like MgC--and can be swept out of the system--say by floating to the top of an immiscible phase overlying the attack-ing fluid--this cero net ra te dissolution is quite likely.

Iron oxides frequently arise in discussions of refractory attack since they form icw melting species with most refractory oxides. Ecwever , chemical attack on re f r actor ies is not restricted

- to iron oxides.

9500!1457

Iron cxides are especially i portant in discussions of light

'? uater rea: tor accidents since these occidents involve high te -

pera,ture molten steel in very oxidizing environ:ents. Ts s t e s o f i steel oxida tion in these conditions can be quite high unless the s te el i s c.ov e r ed b- a reasonably thick, viscous slag layer.

(See.

also 3ection I-A-2.)

Photographs of refractories subjected to chenical attaqk shovn in Tigure 1 filus tr a te the non-unifor. na ture of chenical ctteck, Cer tainly one of the uncer tainties that cust be addressed in core retention device design is dhethe: vertical or horizontal surfaces ,

a r e n.or e g r ievously a f f ec ted by chert.ical a ttack.

3 t

t 9500fASB

...__.m.. ... . .. . _ . _ .. _ .._ . _ .. ........ _ _ , _ . . , . _ _ . , . .

. _ . . - . . - _._ _ _ . . _ . _ _ . _ . . . . .