ML20059J192

Responds to Containment Emergency Procedure Guidelines Issue on Heat Capacity Temp Limit Portion of Open Issue F18.1-1, Transmitted by

ML20059J192
Person / Time
Site:	05200001
Issue date:	01/20/1994
From:	Fox J GENERAL ELECTRIC CO.
To:	Poslusny C Office of Nuclear Reactor Regulation
References
NUDOCS 9401310392
Download: ML20059J192 (9)

Text

.

I GENuclearEnergy r.

GeneraiElectic Company US Curmer Avenue. San Jxe. CA 951?S January 20,1994 Docket No.52-001 Chet Poslusny, Senior Project Manager Standardization Project Directorate Associate Directorate for. Advanced Reactors and License Renewal Office of the Nuclear Reactor Regulation

Subject:

Submittal Supporting Accelerated ABWR Schedule -

Containment Emergency Procedure Guidelines Issue on IIcat Capacity Temperature Limit (IICTL)

Reference:

R. W. Borchardt letter to J. F. Quirk,"GE ABWR Containment systems and Severe Accident Review Issues",

December 29,1993

Dear Chet:

This letter responds to the HCTL portion of Open Issue F18.1-1 transmitted by the above reference. The issue as stated in the reference is:

"On order for the staff to find the IICTL curve acceptable, as proposed in Amendment 32 of the SSAR, GE must demonstrate that large continuous steam plumes do not occur within the suppression >ool such that the containment liner integrity could be jeopardized the sudden unstable collapse oflarge steam bubbles. Large steam bub les appeared to have been observed in a saturated pool during the sub-scale experiments performed by Chen and Sonin as discussed in Dr. Sonin's p, aper published in Nuclear Engineering and Design (1981). The staff will fmd acceptable a suppression pool operated near saturation if the applicant can demonstrate that the Cross-Quencher proposed for ABWR can produce a stable steam bubble when a steam discharge could occur into a suppression pool operating near the saturation temperature. Stable steam bubble size would then be defined as that size steam bubble or group of bubbles that may drift into a cooler region of the pool and condense such that suppression pool wall pressures do not exceed those wall pressures previously defined for unstable condensation oscillation loads from SRV actuations or a LOCA."

r

()[

9401310392 94 PDR ADOCK^05

\\\\

g g

e De attached response has been prepared by Dr. Ain A. Sonin under contract to GE Nuclear Energy. Dr. Sonin was a co-author of the paper referenced in the NRC's statement of the issue. In summary, ion that the condensation process with SRV -

response which supports the conclus discharges through quenchers into the suopression pool results in low amplitude loads for all suppression pool temperatures. *IIe attached has been reviewed by Dr. Fred J.

Moody, Saul Mintz and J. E. Torbeck of GE Nuclear Energy, and we agree with Dr.

Somn s response.

Please provide a copy of this transmittal to A. D'Angelo and Mike Snodderly.

Sincerely, k

n Jac Fox Advanced Reactor Programs cc:

Joe Quirk GE)

Alan Beard GE)

Norman Fletcher DOE)

Cal Tang GE) 1 6

y i

i 6

JNIN44W

..F

1 On quencher operation at high pool temperatures Ain A. Sonin**

Deparunent of Mechanim! Engineering MIT 19 Jannary 1994 Subject NRC concern: "large continuous steam plumes (may give rise to) large bubbles (that) driftinto a cooler region of the pool and mddenly collapse _."

Response

1.

At high pool temperamres Chun & Sonin's* quencher dieharge gests showed a long, steady, turbulent, forced plume which cnnAad of a random two phase mixture of entrained water and steam bubbles. It was not a transient flow sheriding Inrge coherent bubbles which might drift away and collapse in a colder region of the pool The long plumes existed at suboolings below about soc, and were char =~*rd by very quiet quencher operation and low pressure flumadans in the pool, the measured pressure signals being either below or barely higher than the amplifier noise.

The steam bubbles in such a plume are typically relatively small, randomly disuibuted, and, unlea they have already mehed very small size (below I cm, say, which would render them innocuous), they must be of nonspherical shape since they will be distorted by the mrbulcut fluctuadons in the plume. In fact, the plume appeared to be a random mixant of steam and near-sannated waterligarnmes typically with quite low steam voinme fraction. Even if condenotion raies were high enough (see the discussion below, which indientes the enntrary), the condmodon of the steam within such a mixture will not give rise to the kind of focnsd symmetric bubble collapse which could tran5mit significamloads to the bonndaries.

• • Consultantto GE NuclearEnergy

• Seed-atcodofpaper.

2 2.

The plume occurs only when a quencher discharges into water with Islatively umil subcooling. a quencher's hole spacing is so configured that pool wateris enmuned g

efficiently into each of the small steam jets which emanate from the holes. The two-phase mixxure so produced forms a buoyant jet, or forced pinme, dhocle outward'fmm the quencher. The outward flow is driven by the momenmm flow impaned at the quencher; an upward flow also results because the steam bubbles impart buoyancy to the mi.w.u c. The larrer causes the plume to cmve vertically upward as the mixmre moves away fmm the quencher, as observed by Chun & Sonin..Because mrbulence entrains subcooled water into the plume from its sides, the plume rapidly loses icmporature and steam volume fraction with increasing dienm from the quencher. See, for example, Turner's monograph and references therein for a discussion of jets, plumes, entrainment, etc.

t i

3.

At high pool two=#hnn & Sonin observed steady plumes which suctched away from the quencher (which was a single tube with rows of holes on both sides) in both directions to some disemce away in the pool The plumes were relatively short at lower pool temperanues, and stretched further and further into the pool as the pool temperature rose. 'Ihe fmthest point of the pInme was simply the point where all the latent heat and superheat of the steam had been absorbed by the water which was mrbulently tmxed with the steam by entrainment.

4.

In their small-scale emnidons, Chun & Sonin observed very efficient mixing of the pool as a whole. A emaety parameter for modeling buoyancy effects can be written either as a Rayleigh number based on turbulence intensity instead of mean velocity s'

gradient (which is converuionally used) or as a ratio of an appropriately defined Monin-Obitkhov length scale and i e pool depth. (See Turner, for example, for scaling parameters used in buoyant, naculent flows.) "Ihe buoyancy emRarity parameter could not be propedy scaled in the tests of Chun and Sonin. At full scale, buoyancy effects will be more significant and will tend to give rise to vertical thermal stratification, with a warmer (mixed) layer near the surface and a colder region below. The effects of vertical sgrnrifiertrion are dicenm d in whalfolloWs.

5.

Dmmg the extended discharges which are required to bring a pool to "high" temperature, the pool will be mixed by the nubulence in Anrad by the discharges, much as in the experimen" of Chun & Sonin. The pool tmbnlence is created by the quencher pinmes themselves (e.g. see Rodi for tmbulence cr-M byjets andplumes) as wellasby

3 the secondary flow field caused in the pool as a whole by the entrainment. The flow field in the poolis sketched qnnlitntively in figme 1, which shows (i) the rehtively high speed, buoyant plume regions addng from the qnencher and curving up toward the surface and carrying warm water to it. (ii) a layer of warm buoyant, mixed fluid under the surface which thicken as the dieharge puw.ds, (iii) cuculadng regions (the circulation c,n<ed by entr*inmem) at intermadiare elevanons and intermediate temperamre, and (iv) the colder regions near the floor, below the quencher elevadon. Cireninnry drift modon is caused in a11 regions by the entraininent into the plumes, but the circIlatory mean flow velocrties are Smnil reladve to those in the the plume (e.g. see Rodi).

We note that in a mrbulent, thermally stratified pool, horizontal mixing occurs much easier than vertical unxing. Thermal stratification with temperature decreasmg l

from top down is dynamien11y stable. Unrimnal thermal gradients are not. Witness, for example, the well known phennmenon of a cold weather front colliding with a watIn one-the cold front will slide under the warm one and spread over the ground while the warm one slides over the its top, the two mixing at their interface (e.g. see Tener). The end resn1r tends thus toward a vemcally erHied conditinn; which is stable.

When a pool traches high temperatme after an mended discharge,it will thus be relatively weII mixed horimmally in the areas affected by the pinma and their seenndary flow field. 'Ilms in the region where bubbles exist the stIntHicatinn will be primarily vertical, with highest temperamte in the warm buoyant layer near the surface and the colder wuwsamres near the bottom, below the Icvel of the quencher. Water will be entrained into the quencher discharge from the colder parts of the pool in the gnenebss vicinity. That is, quencher operation is conttulled by the colder pool temperamres in the quencher's vicinity. As a result, an mandeA plume of the sort observed by Omn and Sonin will occur only if the colderparts of the pool reach high tempetantre. The plume with its bubbles is buoyant and will rise upward, into the warmer water. He condidon where the plume, with its steam bubbles, moverfrom a warm region of the pool to a signrficandy colder region is therefore implausible.

The objection may be rased that the foregoing depicts the prcicess in quasi-steady terms. Couhi there be a nandent flow, resulting for enmple when a remote cold pool region which has not been mixed by the quencher flow begins to slide under the hot, j

thermally stranfiedregion of the poolin the region affected by the quencher? Thc cold region would advance toward the plume (diding under the warm region and mixing with it ) at a speed of orderi ATgh orless (where p is the coefficient of thermal expansion of the water, AT is the rnarimnm q.nue difference in the pool, g is the acceleradon of gravity, and h is the pool depth) or at the speed of the secondary flow caused by the i

4 1

quencher, whichever is larger. Both these speeds are slower than or of the sa:nc order as the mrbulem velocity Onm= dons in the gneneber plume, and a cold Tunt will typically have a hadvant=1 extent which is larger than the plume's transver.se dEnensinn Any such cold front wordd thus make its presence felt on a time scale which Is slower than, or at at most of the same order as, the turbulent mixing time in the plume itself. This means that the plume's response will be locally quasi-steady. As the local pool wmyu Anne changes, the plume will simply adjust itself to an opw. dug mode conszstent with the lower pool temperamre in its vicinity, i.e. Its length will thrink and it will behave dynamicaHy as in the somewhat lower pool temperatme data of Chnn & Sonin, which showed modest nn m adans.

6.

Some further insight into steam bubble behavior at high pool temperamres may be gained by considering the collapse of one of the steam bubbles in a plume when the water temperance is high. Mass conservation dier'r" that df 4xR3T 2

gpv 3 J=-m4nR (1) where R is the bubble radius, pv is the vapor density in the bubble, and m is the condemnrion mass finx across the vapor-liquid interface. Brown et al have shown uguuuentally that for a bubble bounded by tmbulent, subcooled water, m-0.020Pr433 v'c(Ts-TyL (2) p where p is the water densiry, v' is the Ims value of the tmbalent velocity on the liquid side, e is the specific heat of the water, Ts is the santration Emisanne of the steam (the water temperarme at the interface), T is the local bulk water kmgu4uuw L is the latent heat of ennde== dan, and Pr is the Prundd 1 A, of the water.

The collapse of a steam bubble is driven by the internal pressme reduedon caused by condensation atits bonndary, and resisted by tb inertia of the water which surrounds the bubble. An upper-bound esnmate for the collapse velocity can therefore be obtained by negIccting the inertial effects associated with the water, and considering the limidng case where the vapor in the bubble is simply removed at essentially constant (saturation) pressure at the rate diW by condensation at its interface. In this limiting case, the dynamics of the water outside has negligible effect on the bubble collapse, and Eqs (1) and (2) imm-durdy yield the coIIapse velocity

-r-

~

j-

]

t.

u

]

t-Creef [

'l b

g, A _(; M;f u ir m n

~growingins'apa f

Je 4-p.3 N hj fff

.t 57c,>

[,, s/

1 1

~

~. )/

3

% W J. & Q'q:jf K

~

t mgN v

,/

h

+

g g

/ ' / / / ' '

// / // / / //

/ / r ry s i / /,, 7 7 /~/,,,,

--s HA.c,k er.

f MB**cker~

urett t"'~

co/A+

i ugs.,

Co/de's k w k

rqun,

~

l

.l 5

F h=0420Pr0331Y iv (3)

(Ps; 1

)

i where Psis the sannation value of the vapor denstry at the pool pressure. For pool muyestme of 950C, this bamm% in SI units, dR 7 = 0.05 (Ts-T)v' (4)

The rms velocity v' is related to the average value U of the turbulent mean velocity in the plume. Typically, v'a03U in ajet or a plume. If we take U=4m/s, say, and a subcooling of soc [see wmad.s in point (1) above], we find dR/dt=03 m/s". This indicates that the condensation rate tends to limit the collapse speed to modest values under typical plume condirinns*. In any case, the plume's zeaction to pool temperature is quasi-steady, as we have argued above, and Chun and Sonin have indicated experimentally that at quasi-steady conditions the pressure fluctuations are low at all subcoolings, for reasons which are discussed in point (1) above.

References J. S. Brown,D C Khoo & A. A. Sonin, " Rate correlation for ennde.nsation of pure vapor on turbulent, subcooled liquid", Int. J. HearMars Transfer 33 (1990) 2001-2018

" If we =wne that the mrtain rare is inertally connu!!cd. that is, the bubble pressme is insnnanmnply brought down (by an infinite mad== dan raze) frtxn the local pool pressure p to the satmanon pressure ps(T) at the local pool temperature, and the subseqarxzt bubble mhpce is inertially controlled, the collapse spcmd can be shown to be of order Dps)/P. This gives a figme of about 8m/s for the case at hand. Since tbc mmni rate of conapse will be controGed by the slowest of the two rate m@"We (condensation and sacrua), the mad==dnn rimired equanan is the cocrect one for the case at hand. Eqrorinn (3) ir.r-an upper bound in the sense that the acmal ram canot err-f this value (harnng, ofcourse, a breakdown of the 9mh=h rate mm hdna. as ocx:ars for example very near the steam hjamna point-see below). If the enir e is incruamemw Eq. (3) will overnem= the speed by a large factor.

m

• Note that while the values usaf in this ermple are ht-in the regions ofthe plume away frnm the direct steam dacharge, they are not appropoare for the regum close to the neam discharges. Here, violcat mixing occurs, the steam. water interface is broken up, and the ce=6an rate of Brown et al does not apply. Higher hear transfer rates will prevail, so high in fact that there is complete condmudnn of the steam near the gnereher disdiarge at all mhmnf% cxcept very small values.

6

+

J.-HJ Chai & A. A. Sonin, "Small-scale simbrinn of vapor discharges into subcooled liqmd pools", Nuclear Engineering & Design 85 (1985) 353-362 l

W. Rodi, TurbulenrEaoyantlezr andPhaner, Pergamon Press, Oxford,1982 J. S. Turner, Buoyancy Efects in Fluids. rmMagc Universay Press, r= Mage,1973 P

h i

w

.