ML20213E097

Submits Documentation of UHS Pond Analysis.Techniques Explained in Rept Will Be Useful in Ultimate Design of Heat Sink Spray Plants

ML20213E097
Person / Time
Site:	Columbia
Issue date:	03/05/1982
From:	Codell R Office of Nuclear Reactor Regulation
To:	Lear G Office of Nuclear Reactor Regulation
References
CON-WNP-0486, CON-WNP-486 NUDOCS 8203240077
Download: ML20213E097 (25)

Text

m 3~

c n

c,

.g.

7 DISTRIBUTI0fl Docket File I

HGEB P,eading f.!AR 5

'982 Docket flo. 50-397 ffEi!0RAflDU'l FOR: George Lear, Chief Hydrologic and Geotechnical Engineering Branch Division of Engineering Ti!RU:

flyron Fliegel, Leader, Hydrologic Engineering Section Hydrologic and Geotechnical Engineering Cranch Division of Engineering FR0ft:

Richard Codell, Hydrologic Engineering Section Hydrologic and Geotechnical Engineering Branch Division of Engineering

SUBJECT:

DOCU'iEllTATI0fl 0F UfiP-2 ULTI!! ATE IIEAT SIllK Plant flame: WflP-2 Docket flumber: 50-397 The analysis I perfomed for the UflP-2 Ultimate Heat Sink was unusual in several respects, so I have taken the extra steps necessary to docunent what I did. The techniques explained in this report will be useful in the event.of similarly-designed ultimate heat sink spray plants.

I believe that TVA have proposed similar types of spray ponds for one or nore plants.

/

/c /

,<./ /

Richard Codell, Senior Hydraulic Engineer

!fydrologic and Geotechnical Engineering Branch Division of Engineering g

[

Enclosure:

As stated

<3 Il RFM*

cc:

J. Knight M. F1fegel 0 1982 #

G. Staley q

~

$ gqQuLT } /

R. Wescott R. Gonzales R. Codell 3

6'

,i D 4

xc

,1, 6 (82GWNhWN

}S99 or,,co J.i.G E,0,;,0,E,,,,,,,,,,,,,hGkB,jdE,,,,,,

m.mo

.R,C,9,d,e,))[me,,,,,flj,i,e,g,ej,,,,,

. 3 L ;2- /? 2......... 32I28.?.....

n..

6 8203240077 G20305 hnc, W ADOCK 050003

>FFICIAL RECORD COPY usem-mm

I Hydrologic Engineering Section, HGEB Analysis of Washington Public Power Supply System Project No. 2 Ultimate Heat Sink, Docket No. 50-397 i

by Richard B. Codell 6

Introduction The WNP2 plant uses as part of its ultimate heat sink a circular spray pond of novel design which uses oriented sprays to induce a vertical draft.

This type of spray pond has not been previously used for nuclear power plants, nor for any other commercial application. No large scale tests on the design have been performed under full heat load, although limited testing has been per-l1 formed at the site without external heat loading (except that provided by operation of the pumps).

In addition, experiments with a prototype develop-mental spray system have been performed under heat loading at a site i.n South Carolina (Ref. 1).

The staff does not consider the results of either or both of the experiments on the spray pond, nor the emperical models developed from these data, to l;

necessarily predict the the design basis water loss or heat rejection for the WNP2 ultimate heat sink.

The staff has developed and used mathematical models to predict the maximum 30 day water loss rate and highest return water tempera-ture for spray ponds of conventional design (Ref. 2).

These models have been shown to accurately predict the performance of experimental spray ponds, but I

the models are not well suited to the circular oriented spray ponds used at the WNP2 site.

The staff has nevertheless used modified versions of

.7

_ j._,

the computer codes of NUREG 0733 with onservatively chosen model parameters, the applicant's experimental data and the vendor's models and data to bound the performance of the spray pond.

Description of WNP2 Spray Ponds 6

The WNP2 ponds are shown in Figure 1.

There are two square ponds of 250 feet i

width which are about 15 feet deep.

The ponds are connected by a siphon so l

that water stored in the non-operating pond is available to the operating pond.

The sprays in each pond are arranged in a circular array of spray 4

trees.

The circle is 1 0 feet in diameter, with 32 spray trees, each having seven arms with spray nozzles attached to the end of each.

The nozzles are oriented toward the center of the ring to create an inward and upward draft through the spray field in the absence of wind.

The service water operational flow rate is expected to be approximately 10,200 gallons per minute. The top nozzle is about 20 feet above the normal full pond water level.

The spray t

field (area containing water droplets) appears from photographs of operation to extend about 25 feet above the normal full pond level and have an annular thickness of about 24 feet.

I i

j The staff's concerns about the design basis water loss of the plant are based l

g on the following information:

(1) The height of the spray field is about double that for normal SPRACO* spray pond designs used at most other nuclear plants with spray ponds.

• Spray Engineering Company, Burling ton,MA 2

q r

k i

l l

p i

e i

I WP { 43s' 4 ' *.

• 8" 27' RihG ME ADE A s

TAtt AI$$ A W $PA AYS itvP 90A 32 Alst A$

CAiENT SPAAv N0Z2tts TQWAA05CENitA07 Ai4G htADEAs

/

?.or A ii,v itv.,

t cir so~ irve, s'##e flf 1

\\

[

l l

r i

si. om w swimi O

1 l

l 1

1 I

t Figure 1, WNP2 Ultimate Heat Sink Spray Ponds 3

1 I

ii (2) The nozzle pressure is much higher than the SPRACO design end should have' a finer spray drop size distribution because of their higher energy density (however, vendor-supplied drop size distributions do not show i

this expected trend).

i j

(3) The oriented spray nozzles create their own updraft even in the absence of WiCd or heat load.

i j

All of these properties of the oriented spray pond contribute to intimate contact between dry air and the water droplets which lead to high cooling 2

efficiency. Along with the high cooling ef ficiency, however, is also a higher i

rat.e of water use for the following reasons:

(1) The height of the spray field, updraft, a,1d the fineness of the spray distribution would contribute to the drift loss of the water droplets, expecially under windy conditions.

(2) The intimate cent. set between dry air and water also leads to high rates of evaporaf.fon.

In the limiting case of perfect contact tetween a f al-ling water drop and ai, tne water would only reach the wet bulb temperature because of the equilitrium betwe.en convective heat transfer and evaporatice (Ref. 3).

Even thocgn no further coolirg :f the diop

.vould occur, water would still be ev4porating frcm the drep. The latent j

heat of vaporization of the water would essenti.sily be cooling the air at this point rather than tne water in the crep.

Inis pnenomenon has been

[

called "the swamp cooler effect" after a type of air conditioning used in dry Cli, nates.

1 4

.-. - ~.

n i

ij The WNP2 applicant and several other spray pond users have erroneously con-i; l;

sidered that spray pond water use would be no greater than that which would j!

occur if all of the heat load went to evaporating water (roughly 1000 STU/lb water). This relationship is frequently adequate in humid climate, but may lI seriously underestimate water use in hot, ory climates because of the above mentioned " swamp cooler" effect.

I 4

Applicant's Analysis of Pond Derformance

'I

• 1 The applicant's UHS maximum temperature analysis (FSAR Section 9.2.5, Ref. 4) i was based on tne vendor's emperical performance model (Ref.1), cerated I

i 10 percent to cover data scatter. The water loss analysis assumed a drift loss rate of 1.02 percent of the water sprayea plus evaporation due to 100*. of the heat load. Only one spray pond was used for tre maximum temperature analysis.

I They predict a maximum service water temperature of 88.6*F.

Two spray ponds I

were assumed for the maximum water loss analysis, but when teqiperature in one of the ponds drop below 80"F the spray header is bypassed to reduce watra l:

loss.

The maximum 30 day water loss was predicted to be 8,871,020 gallons f :

leaving a margin after 30 days of 3,761,661 gallons in the two ponas.

Details

!}+

l of these analyses are contained in FSAR Section 9.2.5.

1 t

Staff's Analysis l

I t

i The staff's analysis is based on nodification to tne mecels of NUREG-0733, i

combined witn the applicant's data from the onsite tests described in Ref. 5.

i I

5 4

l

4

,,.--.a-L ll Orift 1.oss Orift loss is described as the fraction of sprayed water which is lost because of being physically carried by wind beyond the boundaries of the pond. Factors I

which can lead to high drift loss are fine spray diameter distribution, sprays l,

high off the pond surface, small catchment basin area for returning drops and strong updrafts.

j The drop diameter distribution presented for the Ecolaire spray nozzle is i

shown in Figure 2.

This distribution for 17 PSIG (the nominal pressure of the top nozzles) has been tabulated in Table 1 along with the distribution for

,i SPRACO 1751A nozzles operating at its normal 7 PSIG pressure for ccmparison.

The updraft in the spray field has been experimentally determined to be between 10 and 20 feet per second in Ref. 1.

i l

As a first approximation of the drift loss, the fraction of the sprayed drops 7

l which fall at a speed slower than the updraft are assumed to be lost as drif t.

Figure 3 shows the settling velocity for spheres of water in air at 70*F.

l Frem Table 1 and Figure 3, it can be seen that the minimum drift loss for a l

t j;

10 foot per sec and updraft would be about 0.7% for the vendor distribution i

i t

i and 3 percent for the SPP.ACO distribution respectively.

For a 20 foot per second updraft, this loss would be aoout 19% and 25% for the Ecolaire and j

SPRACO distributors, respectively.

This analysis oversimplifies the problem, t

however, because not all of drops would fall into the high velocity portion of the vertical plume frem the spray. Also, once the suspended drop fell outside

[

i 4

i!

of the plume, the vertical updraft would diminish and it would fall.

l!

6

-._,..._,___._,-,_c

~_ -..- --

- = _. -- -. _

t i

e

'l

)

i i

I a

l N

.1 l

IO s

/

i l

i f

4 F

]

1 Io Psi 9 3 -

IS Psi 9 l

E IO o

<c i

u e

~

c

~

<c'

?

l U

g i

O Psig

/

~

I 2

e 10

~

/

a 30 Psly

~a 40 Psig 4

<c a

4 a.

l.

10 g

g I

-3

-2

-I O

l 2

30 to go i xio go 10 ACCUMULATED VOLUME, PER CENT i

i Figure 2 Particle Size Vs. Volume Percentage for Ingersoll-Rand (Ecolaire) Ref. 1 I

I i

i t

i I

m-3-, -&-.-.me

,,yv,-..-.-wve,

-p..,

+-e-.

,-v-ev w

.,w v.

--,,-.-,...,-y

f Table 1 Orop size distributions for Ingersoll-Rand (Ecolaire)* and SPRACO** sprays Mean diameter in fraction, microns Fraction of Cumulative total fraction Ingersoll-Rand SPRACO (17 psig)

(7 psig) 0.15 0.15 4500 4000 0.15 0.30 4000 2600 0.2 0.50 3400 2800 0.1 0.60 2800 2290 O.1 0.70 2400 2000 0.1 0.80 2000 1650

+

0.05 0.85 1700 1340 i

0.05 0.90 1500

.1190 0.03 0.93 1300 1000 0.03 0.96 1200 855 0.02 0.98 940 640 0.006 0.986 820 580

<0.004 0.99 760 520 0.003 0.993 700 460 O.002 0.995 650 425 0.001 0.996 610 400 t

0.001 0.997 580 365 0.001 0.998 530 330 0.001 0.999 420 200 0.005 0.9995 260 260 0.005 1.0 230 200

• From Ref. 1

• • From Ref. 2 t

l t

Ee,.m,eent.ste.ad.c.ed T.

x,.

**'d "** -*g g sapm o 7 7::a 4 wi +sadAA g g g 3, gm,ifliiTMiliiililllllli o

,ev

,. ! I lillfi I i IIMI lills i Ipd

. ~ _ -m

, ---,v.

=

~..,.

, / /// //,, -

+......

I Ililpi Illi!Il /W//f(6i M 1 / i; J J

-/

,.....,,., i ris i s... / //

t lui'!i i I/X////iiluvrf/_A n

=;;;;;;,),i;Q,4 i i!

.4 / / s/ / /.i /T / / // / ' / / :>l illile///N/////hyfMinal m ~._

4 1 ll;..'

E

• 4 v r is g')h i

g a

y O"

i ii E

h10'8 g*f _e

• S -[--

H'

. 2 :;;;,:,;

}

/ /// / j n / /1,f f / / /.

...s

.. +

l,,..///NMNNN/IIi!M9iin$

i i

w..

re/i...

4..

L Numcare en cmes resmeet

>g

=

{ / / e i 4 / / i / cf l e*.e(not svn er eccove )

,,-.Ni JNNA,Un,'blM."

5.'t'".*h'.'

1 Saches-hwhagpsm estr-

. esten tec*er e incluty ter j j?j';'j:f'j j. rs.N '"

i P9ynsesi procert'es used

//[/[

Tomsk Visces.tv Csasity 30.g 31 creesse 15 res 't.

Air 70 00!$1 00749 Woter 70 Q SSI 6~ 3

/ / / /4/,

e e e pin,

+ e t i eta e f t e nir

,./A// ipa i l iiMI I t ilp'il i lluml 4

10 CO iC00 C.C00 wee o.am.ier, micrea.

i Figure 3 Settling velocity of spheres of various density in fluids of higher density (From " Perry's Chemical Engineers' Handbook," by R. Perry, C. Chilton and S. Kirkpatrick, Fourth edition, Figure 5-74, McGraw Hill, New York, NY, 1963.

9

i, The vendor applied a mathematical model which predicts drift on the basis of I

drop trajectories from 3 levels in the spray field, and takes the updraft into account.

I I

I The report on the drif t loss model is difficult to follow.

Furthermore, the vendor's results appear to be too small and have not been backed up with field data.

Results from the field tests at the WNP2 site (Ref. 5) and photographs showing highly visible plumes (Figs. 4 and 5) strongly suggest that the vendor's l

drift loss predictions are optimistic.

}

l In order to predict a drift loss versus wind curve suitable for determining the design basis water loss for WNP2, the staff modified the DRIFT code in NUREG-0733 to approximate the WNP2 pond geometry.

The spray field was modeled as an array of spray nozzles whose spray apogee was 300 cm. from the pond surface, which is roughly half of the observed spray field height.

The noz-zles were arranged in a circle of 140 feet diameter in a square pond 250 feet l

on a side.

The total spray was assumed to be initially concentrated along the l

locus of the 140 foot diameter circle at a height of 300 cm, with zero vertical or horizontal velocity. Several combinations of constant, uniform vertical updrafts and either the SPRACO or Ecolaire drop size distribution were tried.

This model neglected the true nature of the complicated drop trajectories and the 3-dimensional vertical plume, but was tried for lack of any other available 3

model. Orift loss versus wind speed for several cases has been plotted in Figure 6 along with field data from the WNP2 tests.

l 10 j

I

.--a m

• =.m ma

_.u 6ON N

_d, og,,

" ~ ~

..e

,l l

?.

"*?A'

• +

'y.'... ,,, g' l y 4 _-. ; u w FIGURE 4 AERIAL VIDI 0F 11tlP-2 SPRAY P0flD lil OPERATI0tl I

• 5 L

1 _. s. N M W-d u,# .~..... T. -N f I m Tfy l e I q Ew 4 '?-c 3,. l - -qh lt ~. g ,.y g ). ;, -.i,, ,,7..,.. h, 1 am.-----A. ..-.-ama.2 ew .m_...a__-__. - _. i 1 I . j ;f I I o 1 . h] b*- g4,, a, Jl a i I h A j 1, P M ' ^ .g.,glW=i } ~ g

f;u...1%p*:.W.W m>-

f-! ' ' " i ' - ~ I .T T-Q j i u. n..,,$ N.. ] CW'ild. t k y mm k,..,. S_h I~I.-I _,h,.. l j I ~ Dj ~ . ' }$p, Je - r [* f . w..:. 3, 2 i 1: [p/f! l J y l e-

n us

g 4

%* :.,9...o.. fy* S'M, -".u-'.<*r* ::j';%. i,.r;,p & } n ,h. Yt. e;

c... *D.5 4

."*, *.af.

,. *4.g s+.. !;'.?'.

n. 4 mu 1 M. T.. p-s~*.m.vi" ' J 'g[ s ' ]I?.':- '., c, ~ 1 2,Y *'"b$' 5 he ' ^ E,f ' Y_ g}f-l a .w 1. - M %g.g7..,;g, y . c, f,, is -n 5= g-: _... J * .I i ). , " '5h}j;$.T T- 'i.< *.%%'g$g".:E- .'6 me 0= l L= 55

z.,,

-J o - l .'i^ 5...- x< ._.pt g. d, w 5 E= = .., :. M fifi.,.... ' # 3, s. l m= $$mE 1 '.[(h-.93p2'[. ' : ?'

• f.%C l$.!

' j =m = r .... ; _, t . f a=:w

1. e;-(i[.I,N$.h-

[hhb I d .. j, - 4.. '. ',,.;,.- i. ,,,. c { = .g, s _-g e = - - = = = + <

=

• a i

o.1 a. c) Applicant's Data (ref S) f O'1 ~ flRC model(ref 2) 300 cm/sec updraft a SPRACO distribution b flRC model(ref 2) 300 cm/sec updraft Ecolaire distribution { o.2 c Drif t loss used in lif(P-2 evaluation E d Drift loss used in Palo Verde y, evaluation (SPRACO design) e Ecolaire predictive model O./f lf}a ?, u J L) 0, / 4 / o g o.o s-l 0 1 l E]' -4 O i> s i I i i i i r a 5-y /,- 2s 2 e-is Jr +o +r WIflD SPEED MPil FIGURE 6 - DRIFT LOSS 140DELS AtiD DATA FOR tlflP-2 EVALUATI0fl

Discussion While it is recognized that the drift loss model is probably deficient, the trends demonstrated appear to.be in the correct direction. That is, one would expect higher drift loss for higher wind speed, higher updraft and finer drop size distribution. The family of curves in Figure 6 probably therefore have the correct shape. Therefore, a similar curve which envelopes the test data would probably be a suitable and conservative model to extrapolate drift loss beyond the range of the test data wind speeds. The bold curve in Figure 6 has therefore been " eyeball" fitted to the test data and is the drift loss used in the staff's performance model. The accuracy of the drift loss data, however, is subject to criticism for several reasons: (1) Drift loss was determined by subtracting an assumed evaporative loss from water loss calculated by observing a change in pond water level over the period of the test. The evaporation was calculated as simply the water loss due to the heat loss divided by the heat of vaporization. This t' procedure does not truely account for evaporation and could be either an overestimate or underestimate, depending on meteorological conditions. (2) Water loss should have been corrected for thermal expansion during the f test as described in Refs. 6 and 7. If temperature decreased over the period of the test, however, true water loss would be overestimated l because the water level had dropped due to thermal contraction. i i j 14

..____...__.m_.__. Because water level changes of only several millimeters were observe'd for the tests, thermal contraction could cause a substantial error, i (3) The magnitude of vertical updraft for high heat loading could be substantially greater than the relatively low heat loads used in the test. Furthermore, thermodynamic properties of water such as density, viscosity, and surface tension decrease for higher temperature while air viscosity increases. These factors might lead to increased drift for l high heat loads, i The drift loss curve, therefore, is highly uncertain. It is doubtful that a much better approximation could be made without extensive field tests on the prototype under high heat loads and high wind speed. Another uncertainty about the drift loss computations is the vendor-supplied crop size distribution. The vendor's distribution for 17 PSIG shows a j relatively small amount of fine drift particles compared to the SPRACO 1751A i nozzle distribution, even though the latter operates at only 7 PSIG. This result is conterintuitive since high pressure and high velocity of the vendor's nozzle would appear to provide conditions favorable for the creation of fine drops (i.e., high energy density). Cooling Performance Model The spray performance model of NUREG-U733 was modified to predict the cooling performance of the Ecolaire spray pond. The NUREG-0733 models cannot account for the updraf t induced by the oriented spray design of the pond. Furthermore, 15

i the NRC models consider the sprays arranged in a rectangular array, with all ~ nozzles pointed straight up. The staff's approximation of the Ecolaire pond was performed in the following fashion: (1) The donut shaped spray field was " broken" at one point and stretched into a rectangular volume. The dimensions of this volume were 20 feet high, l 25 feet wide and 440 feet long. I (2) The spray nozzles were assumed to be in the center of the spray volume at a height of 9.0 feet above the pond. The cone angle of the nozzles was 55 degrees from horizontal with an initial velocity of 26 feet per i second. The mean drop radius was determined from the venrior's dis-tribution to be 0.08 cm 'using the suggested weighting procedure of NUREG-0733. Discussion t I The predicted spray performance for the high wind speed (HWS) and low wind speed (LWS) models was compared to field data on the vendor's pond prototype l collected at Canadys, South Carolina (Ref. 1). This ccmparison is shown in Figure 7. The agreement between the LWS predictive model and the data is very good, while the HWS model results are optimistic. The apparently optimistic HWS performance prediction could have resulted from i the configuration of the vendor's test pond. The pond consisted of a l 16 1

I i f 9p - LilS MODEL 6 ilWS MODEL O O Oo 0 e sa '3 80 o O u E O o E 5 O 6 !E O O FIGURE 7 N 6 COMPARISON OF NRC MODELS FOR COOLING (REF 2) TO ECOLAIRE MEASURED DATA AT oy 6 7, CANADYS.SC EXIPERIMENT e O ^ f' a Y O O E' a 5 O O O E a u go g,, pp Sc) 9 CANADYS OBSERVED POND TEf1PERATURE - F

full-sized, pie-shaped segment of the true final spray ring, with air tight baffles on the radii of the segment. These baffles may have isolated the pond i segment from the effects of the ambient wind, and thus prevented higher pond efficiency when the wind was blowing. Therefore, the apparently optimistic behavior of the staff's HWS model, which has not considered these baffles, may in fact be representative of the WNP2 pond. Normally, both the LWS and HWS models would be used in the performance asses-sment, but in this case, given the good agreement of the LWS model alone and the uncertainties of applying the NRC model to this situation, only the more conservative LWS model will be used. Determination of Maximum Pond Temoerature Maximum pond temperature was approximated from a procedure modified from NUREG-0733 (Ref. 2). A TDF-1440 meteorological tape was obtained from the National Climatic Center, Asheville, North Carolina. This tape contained meteorological data from Yakima, Washi'ngton from 1948 to 1981, and was used to represent the long-term offsite data base. These data were scanned with the SPSCAN code of Ref. 2, using the LWS performance model and a steady heat load of 6.47x10 BTU /hr which is the 30 day average heat load. The maximum pre-dicted pond temperature for the steady heat load was 85.05'F with one pond operation. Output from the scanning model was used with program SPRPND (Ref. 2) and the actual heat load curve, determined from Ref. 4 and shown in Figure 8, to predict a peak pond temperature of 87.7'F for the Yakima data. 18 4 i

.-y-- -. - - - - c ma.., 2 u_ y,y -A 'MA9MM

• &&9 OW O'O'

[ /2. z e 4 e /t ~ i /v A k 4 ( 7 (3 llEAT LOAD USED IN WNP-2 SPRAY POND ANALYSIS FIGURE 8 x N 8 g( + t. 7 f~ I i i i 6l / /t* /s'o setto inner ~' h.>.> r-s

l 1 I Correction for Onsite Versus Offsite Data Differences The onsite-offsite correlation procedure suggested in NUREG-0733 was not i followed. Rather, the correlation between onsite and offsite data was determined in the following way: (1) The applicant's analysis of maximum UHS temperature agreed with the timing of the staff's analysis. That is, the applicant predicted that the peak pond temperature would occur on the same c 3te as the staff predicted, even though the applicant used Hanford on-site data and the staff used data from Yakima. A conservative correction factor was determined by evaluating a steady-state version of the spray cooling model for hourly values of the worst day ineteorology at Hanford (Ref. 4) and Yakima. A conservative correction factor of 1.2*F was added to the staff's maximum temperature prediction on the basis of this analysis. The staff's peak pond temperature estimate nas therefore been determined to be 87.7 + 1.3*F = 89.0*F. The applicant's estimate of the design basis peak pond temperature is 88.6*F which is essentially identical. The staff's analysis is l probably conservative because the effects of wind-driven currents (HWS model) l have been neglected from the temperature calculations. We therefore concur with the applicant's analysis-for peak pond temperature. l 20

l Water Loss Modeling ,il Water loss was modeled with according to the procedure described in NUREG-0733.* The scanning of the Yakima data with the 30 day average heat load of 7 6.47 x 10 BTU /hr gave a maximum 30 day uncorrected water loss of 7 1.28 x 10 gallons for continuous one spray pond operation. The 30 day average drift loss was I determined to be about 1.7 percent of the water sprayed, corres-ponding to an average wind speed of about 10 miles per hour. Since the 7 design basis pond capacity is only 1.25 x 10 gallons, and two pond operation would be expected to use more water than one pond operation, the model predicts that the Reg. Guide 1.27 requirement for a 30 day water supply to available on site has not been achieved. Discussion Actual operation of the spray ponds would not necessarily be continuous. One or both of the ponds could be run during high heat loads, and as the heat loads fell, could be shut off or cycled between a minimum and maximum l temperature range to conserve water. I Four cases of cycled and uncycled spray pond operation were tried using more detailed SPRPND model (Ref. 2) and the actual 30 day heat curve of Figure 8:

• 0nly the LWS model was used for this scanning because it gave lower cooling efficiency than the combined LWS-HWS model.

Experiments to predict maximum water loss indicate virtually identical results with the LWS-HWS model. 21 l

l (1) One pond run continuously for 30 days. (2) Both ponds run continuously for 60 hours, gradual throttling to one pond operation at 108 hours to 720 hours (30 days). (3) One pond operation. Continuous operation for 3 days, then cycled operation between minimum 80*F and maximum 85'F from 3 to 30 days. Sprays are bypassed directly into pond, f (4) Same as Case 3, but two ponds cycled instead of one. The resulting water losses for these four conditions of spray pond operation are, respectively: 7 (1) 1.27x10 gallons (2) 1.34x10 gallons 6 (3) 8.52x10 gallons 6 (4) 9.42x10 gallons i l Therefore, the resulting 30 day water losses with cycled pond operation are i less than the 1.25 x 10 gallons available. I i I Conclusion i The staff has analyzed the performance of the spray ponds for the WPN2 plant and reaches the following conclusions: I t 22

(1) The maximum pond temperature determined for the period of record is likely to be no greater than 90*F. This agrees closely with the appli-cant's prediction of 88.6?F. The staff considers this prediction to be conservative based on comparisons of its model with prototype data. I i (2) Water loss for continuous one or two pond operation may exceed by a small amount, the 30 day onsite supply without makeup. This analysis did l not consider the effects of seepage, which is unquantified. Cycling the ponds after the initial period of high heat loads would be effective way to reduce the water loss to acceptable levels. The accuracy of the water loss models, particularly the drift loss model, is questionable. Final resolution of the water loss issue should be made, but can be delayed until after plant operation commences in order to facilitate the operation of the ponds with design basis heat loads, which would be diffi-cult or impossible to achieve before the plant begins operation. The staff concludes that the ultimate heat sink is acceptable provided that the following actions are taken: (1) There should be a technical specification for procedures which would .I allow cycling of the spray ponds to conserve water. These procedures

I l

should be implemented whenever sources of makeup water are interupted, or there is a threat of interuption. l I } (2) Confirmatory tests of spray pond operation under design basis heat loads should be made over a wide range of meteorological conditions in order to r 23 f

r s-- test the heat rejection and water loss performance for the ponds. If performance is shown to be more adverse than the model predictions, it should be brought to the immediate attention of the staff. Conversely, if it can be shown that water loss is more favorable than the staff's predictions, the above technical specification may be modified or withdrawn. References 1. Ingersoll-Rand Company, " Oriented Spray Cooling System (OSCS) for Ultimate Heat Sink Applications (UHS)," Topical Report IR-100-P, January 1977, Ingersoll-Rand Company, Condenser Division, PO Box 483, Phillipsburg, NJ 08865 (presently called "Ecoiaire" Corporation). 2. Codell, R. B., " Analysis of Ultimate Heat Sink Spray Ponds," NUREG-0733, U.S. Nuclear Regulatory Commission, Washington, DC 20555, August 1981. 3. R. 8. Bird, W. E. Stewart and E. N. Lightfoot, Transcort Phenomena, John Wiley and Sons, Inc., New York, NY,1960. 4. Applicant's FSAR, Section 9.2.5. 5. K. R. Conn, "1979 Ultimate Heat Sink Spray System Test Result," Washington Public Power Supply System Nuclear Project 2, WPPSS-EN-81-01, 1981 i'; 6. R. K. liadlock and O. B. Abbey, " Thermal Performance and Water Utilization Measurements on Ultimate Heat Sinks - Cooling Ponds and Soray Ponds," USNRC, NUREG/CR-1886, PNL-3689, May 1981. 4 l 7. A. L. Godbey, " Evaporation Determined from Energy and Water Balances at Two Heated Ponds," Master's Thesis, flassachusetts Institute of Technology, Cambridge, MA, June 1981. 1 i i i 24 ._}}