ML20126L357

Nonproprietary Version of, Responses to NRC Concerns on Applicability of CE-1 Correlation to SONGS Fuel Design

ML20126L357
Person / Time
Site:	San Onofre
Issue date:	05/31/1981
From:	ABB COMBUSTION ENGINEERING NUCLEAR FUEL (FORMERLY
To:
Shared Package
ML13323A152	List: ML20126L357
References
CEN-165(S)-NP, NUDOCS 8106020439
Download: ML20126L357 (38)

Text

-- - . . _ _ . . . _ _ . _ _ _ _ ___ _ _ _ _ _ _ . _ . _ .

0 I*

l l l

l SAff OfiOFRE UllITS % Afl0 3 00CKETS 50-361 and 50-362 I

l CEN-165-(S)- llP

• l l

I RESP 0flSES T0 flRC C0flCERilS Oil APPLICABILITY OF THE i CE-1 CORRELAT10tl TO THE S0iGS FUEL DESIGft May, 1981 l

COMBUSTION EilGIflEERiflG, INC,

' fiUCLEAR POWER SYSTEMS P0',!ER SYSTEliS GROUP i

- WillDSOR, C0fniECTICUT 06095 8106020 -4

,,g.y, , , , , . _ .

LEGAL NOTICE THIS REPORT WAS PREPARED AS AN ACCOUNT OF WORK SPONSORED BY COMBUSTION ENGINEERING, INC. NEITHER COMBUSTION ENGINEERING NOR ANY PERSON ACTING ON ITS BEHALF:

A. MAKES ANY WARRANTY OR REPRESENTATIOS, EXPRESS OR IMPLIED INCLUDING THE WARRANTIES OF FITNESS FOR A PARTICULAR PURPOSE OR MERCHANTABILITY, WITH RESPECT TO THE ACCURACY, COMPLETENESS, OR USEFULNESS OFYHE INFORMATION CONTAINED IN This REPORT, OR THAT THE USE OF ANY INFORMATION, APPARATUS, METHOD, OR PROCESS DISCLOSED IN THIS REPORT MAY NOT INFRINGE PRIVATELY OWNED RIGHTS;OR B. ASSUMES ANY LIABILITIES WITH RESPECT TO THE USE OF, OR FOR DAMAGES RESULTING FROM THE USE OF ANY INFORMATION, APPARATUS, METitOD OR PROCESS DISCLOSED IN THIS REPORT.

, 1 l

1

l 1

1

. ABSTRACT

. l This docunent develops in detail, data which supports the position that CE-1 l is applicable to the SONGS fuel design. This is accomplished with statistical analyses of critical heat flux (CHF) data, mechanistically based explanations of the CHF data obtained and supporting documents in the literature.

• 1 The following conclusions are made:

- The presence of the HID-7 spacer grids' has a beneficial ef fect on critical heat flux. Extrapolation of BWR experimental data to PWR conditions is invalid. -

9

- When applied to non-uniform axial power distributions, the CE-1 correlation produces conservative results. The reasons for this conservatism are known.

- Reduction of the data base for CE-1 to include only the primary indication of DNB,results in essentially the same statistics for the correlation.

- Detailed statistical analysis of the data base for CE-1 denonstrates that there is no statistically significant difference attributed to the change from 14x14 to 16x16 fuel designs.

The results conclusively The docenent addresses all currently known NRC concerns.

demonstrate the validity of the CE-1 correlation for a MDNBR of 1.13 but presents a still more conservative value of 1.19. They also show that there is no technical basis for further analytical or test efforts.

s b

1 4 T

i

TABLE Of CONIEtlTS

.Page ABSTRACT i I.

SUMMARY

l CRITICAL HEAT FLUX Iti THE PRESEtlCE OF 3 l II.

SPACER GRIOS III. EFFECTS OF SPACER GRIDS Illf10tl- 10 UtiIFORM APO CHF TESTS ,

IV. EFFECT OF MULTIPLE DATA P0lflTS 17 V. STATISTICAL EVALUATI0tl 0F CHF DATA 19 VI. REFEREtlCES- 33 I

I l

l 11 l

l. . .- ._ , -- .- -

~

l I.

SUMMARY

During the recent review of the SONGS 2/3 licensing submittals, a number of NRC staff questions have been raised relative to the applicability of the CE-1 critical heat flux (CHF) correlation to the high impact grid design of the SONGS 2/3 fuel assemblies.

The particular concern is the applicability of CE-1 to C-E's HID-2 grid design.

The design characteristics of this grid are compared to other C-E grid designs in Table I-1.

This document provides responses to several concerns that have been expressed by the staf f. The first concern is that the generally larger HID-2 grid causes a flow stagnation effect upstream of the grid to such an extent that the minimum departure from nucleate boiling ratio (MDNBR) is degraded. This is an effect .

which has been discussed in the literature under BWR conditions.Section II of this document addresses this concern and cites a number of references which support C-E's contention that the alleged effect is inconsequential . Based on this literature we conclude that it is not clear for the BWR conditions whether the effect of grids is beneficial or adverse. Furthermore, the conditions wnich might cause the effect to be adverse in BWR's do not occur for the HID-2 grid under PWR conditions. All experimental evidence from PWR tests shows grids to have a beneficial ef fect. This beneficial effect is believed to occur via the mechanisms discussed in Section III. This same mechanism provides the explanation of why C-E's non-uniform axial power distribution (APD) data has more scatter and is more conservative than.the uniform APD data.

The second concern expressed by the staff was that "mul tiple data points have been used for some heater rods with quadrant instrumentation in non-matrix subchannels".

This concern is addressed in Section IV which provides the results of a reanalysis of the data witnout use of the multiple data points. The results of this reanalysis show the revised statistics to compare very favorably with the original CE-1 data base statistics and continue to support an overall 95/95 MDNBR of 1.13.

This is further evidence that the design MDNSR of 1.19 is adequately conservative to account for any uncertainties associated with the SONGS 2/3 high impact fuel design using HID-2 grids.

Another concern expressed by the staff is that C-E's non-uniform APD data from 14xl4 and 16x16 designs "may not come fron the same population". This concern is addressed in Section V. This section provides the details and the conclusions of an analysis of variance that was performed for both uniform and non-uniform APD data. This analysis shows statistically conclusive results supporting the fact that the variance between bundle types (i.e., 14x14 vs. 16x16) is very small, and the refore , the effect of the 14x14 vs.16x16 design on the measured to predicted ratio is not statistically sic;nificant.

The attached sections address all of the concerns discussed to date and provide continued evidence that the CE-1 correlation is applicable to and in fact conservative fo- the SONGS 2/3 fuel design. It is also concluded that the method of application of the CE-1 correlation and the Tong F-factor as applied to S0nGS 2/3 with a MDNBR of

. 1.19, provides substanti- conservatism relative to the C-E committed DNBR criterion that the minimum DNDR shall provide at least a 951 protability with 95r confidence that DNB does not occur on a fuel rod having the MLNBR during steady state operation and anticipated transients of moderate frecluency.

j. . . .- .

? -

i -

1, TABLE I-l t

i-1

.16 X 16 SPACER GRID CHARACTERISTICS f

i ,

s

-i .

SAN Oio.:RE

~ '

j- ARK /t!SAS .

HID-1 HID ,

s

f I GRID SPACING, INCHES 14.8 35.7 15.7 CLEAR SPACE BETWEEN GRIDS _

. INTERICR STRIP, INCHES

.r'o PERIFETER STRIP, It4GiES i

i i

PERIt'.ETER STRIP TiiiCRIESS, ItiCHES l

! HEIGHT, INCHES i

i INTERICR STRIP ,

THICC;ESS, INCHES ,

f. '

HEIGHT, INCHES . ,

1, . .

, e

• s'

11. CRITICAL HEAT FLUX IN THE PRESEHCE OF SPACER GRIOS Introduc tion Experiments have been performed to establish the effect of spacer grids on the Critical Heat Flux (CHF) in both BWR's and PWR's. At first glance, DUR test data appear to be inconclusive. Some data support the contention that spacer grids have a beneficial effect on CHF while other test data seem to show an adverse spacer grid ef fect. On the other hand, PWR test data consistently indicate that CHF is greater in the presence of spacer grids. The postulated mechanism associated wi th the BWR test data showing an adverse spacer grid effect will be examined in this section to determine whether such a mechanism could occur under PWR conditions.

One of the major thennal-hydraulic (T-H) di f ferences between BUR's and PWR's ,

is the typical flow regime that is present in the region of boiling crisis for the reactor. Subcooled and bubbly flow are usually encountered in PWR's while annular flow is predominant in BUR's (1).* The two types of flow regimes dif fer substantially, as shown in Fig. II-1.

Subcooled and bubbly flow regimes in PWR's are typically characterized by a relatively thin bubble layer on the fuel rods and a liquid care. On the other hand, annular flow is characterized by a thin liquid film and a vapor core, in BWR's, the liquid film associated with annular flow coats the fuel rods while the remaining subchannel area is filled with vapor.

Spacer grids are used in both BWR's and PWR's to maintain subchannel geometry.

These grids disrupt the local flow in subchannels and thereby influence the T-H performance of the core. Since the flow regimes di f fer in BWR's and PWR's, and spacer grids af fect T-H performance by disrupting subchannel flow, it is likely that the T-H effects of spacer grids will be different for 3WR's and PhR's.

Results from several BWR and PUR tests will be discussed The and postulated conclusions will be made regarding the ef fects of spacer grids in PWR's.

mechanism associated with adverse grid ef fects in BWR's will be examined While spacer grids to determine whether such a mechanism could exist in PWR's.

disrupt local flow and thereby give rise to hydrodynamic and heat transfer ef fects, attention will be focused on the impact of these ef fects on CHF in the following discussion.

fffects of Spacer Grids in BWR's Tescs with 9 and 16 rod BWR test sections have shown that boiling crisis occurs preferentially upstream cf spacer grids ( 2 )( 3 ) These test data indicate that spacer grids may either promote or delay boiling crist'.,

thereby decreasing or raising CHF, dependent on the coolant quality and the axial location of the grid . The authors concluded that CHF is af fected by the flow redistribution caused by the presence of spacer grids, but determined that local disruption of the liquid film by the spacer grid

• Parenthetical numbers refer to similarly numbered documnts in the REFEREHCE section.

' did not contribute to boiling crisis. .

Adiabatic air / water tests were conducted at low pressures to determine the

- hydrodynamic ef fect of various flow obs tructions (2) (4 ) (13). These tests simulate the flow conditions that occur in the annular flow regime, where a thin liquid film blankets the fuel pin and the remainder of the subchannel flow area is filled with vapor. Dry patches formed in front of the flow obstructions below critical air / water flow rates. These patches grew as the film flow rate was reduced. Dry patches also were observed just downstream of the obstructions.

Lahey and Moody have proposed that the upstream dry patches are caused by horseshoe vortices which form upstream of the flow obstruction, as shown in Fig. II-2.The vortices improve local mass transfer just upstream of the grid to the point where all the liquid is transferred from the surface

• to the free stream air flow. This results in the formation of a dry patch just upstream of the grid, The mechanism for boiling crisis in annular flow is dryout of the liquid film (6). Hence, the hydrodynamic dryout which occurred in the adiabatic air / water tests is postulated to be the mechanism by which boiling crisis occurs upstream of BUR spacer grids.

Thus, BWR test data indicate tnat spacer grids may have either beneficial or adverse ef fects on boiling crisis. The postula ted mechanism associated with the adverse effect of spacer grids is the formation of a horseshoe vortex upstream of the grid. It is important to note that this postulated mechanism best explains test data from idealized adiabatic air / water tests and may therefore have only limited applicability to actual diacatic BWR condi ti ons .

Effects of Spacer Grids in PWR's Many researchers have reported that the presence of spacer grids in PWR test sections has a beneficial effect on CHF because the grids deplete the bubble layer near the fuel and enhance turbulence and mixing. These three ef fects, depletion of tne bubble layer, enhanced turbulence, and enhanced mixing will be referred to collectively as enhanced heat transfer.

. Chang and Dean (7) noted the beneficial effect of spacer grids in their discussion of the ef fects of flow blockages on core power limits. Tong (8) has included a beneficial spacer factor in a CHF correlation to account for the beneficial effects of enhanced heat transfer on CHF. Other researchers (9) (10) have found that the heat transfer coef ficient is increased in the presence of spacer grids, An extensive survey by Gr: ld and Yousef (11) of experiments on the effects of spacer grids concludeo ..at.there is no deleterious upstream ef fect associated wi th spacer grids. Tests have, however, shown that boiling crisis occurs preferentially just upstream of PUR spacer grids. Groenveld and Yousef conclude that this preferential occurrence of boiling crisis is a result of the enhanced heat transfer that occurs downstream of the spacer grid, since l

4s

m. -,, , - . . , - - - , - .-+-,n. - - , - -e,--,. -. y u. , , ~ . , ,,,e,c ---- * ,. ---# ~.- - , . -5 <a ,. ,-

s

- at the upstream edge of a spacer grid, the enhanced heat transfer associated with the previous spacer grid is at a minimum.

~

C-E has conducted uniform axial power distribution CHF tests with a spacer grid near the end of the heated length (12) . These tests have shown that the presence of a spacer grid just downstream of the location of CHF has a negligible impact on the magnitude of CHF.

Hence, a considerable body of evidence indicates that PWR spacer grids:

1) have a beneficial ef fect on CHF downstream of the grid

2) have a negligible ef fect on CHF upstream of the grid The CHF benefit associated with the presence of spacer grids is the result of enhanced heat transfer as the coolant passes through the grids. As ,

explained earlier, subcooled and bubbly flow regimes are typically encountered near the boiling crisis in PUR's. A layer of bubbles clings to the fuel rods in this flow regime while the remainder of the subchannel is filled with sub-cooled liquid, as shown in Fig. II-1. A sketch of the projected area of a spacer The flow obstruction presented grid in a matrix subchannel is shown in Fig. II-3.

oy the intersection of grid strips in the center of the subchannel diverts " cold" water from the subcooled core of fluid in the subchannel toward the fuel rods.

It is noteworthy that the flow obstruction presented in a typical subchannel by C-E spacer grids differs greatly from the obstructions The obstruction usedbyinthe presented theC-E adiabatic grid air / water tests discussed earlier.

lies primarily in the center of the subchannel while the obstructions used

• in the air / water tests and shown in Fig. II-2 were in contact with the channel wall over a large area.

The flow diversion associated with C-E grids has beneficial ef fects on heat transfer and CHF performance for several reasons. The civerted coolant must The combination of higher travel at a higher velocity to preserve continuity.

velocity and cooler liquid near the fuel rod will strip the bubble layer from the fuel rod. Furthermore, heat transfer will be enhanced by the mere presence of cooler liquid near the fuel rod. Redistribution of coolant through the grid also increases the turbulence of the coolant, thereby increasing mixing and improving heat transfer.

The mechanism for boiling crisis in PWR's is departure from nucleate boiling Boiling crisis occurs when bubbles generated at the fuel surface cannot (DNB).

escape from the surf ace f aster than they are generated.

This results in the The stripping formation of an insulating blanket over the surface of the fuel rod.

away cf bubbles by flow diverted by the spacer grids thus will postpone the onset af DNB.

t Inappl!cability of Adverse BWR Conclusions to PWR Conditions e

It is *nclear whether presence of spacer grids in BWR test sections has an advers; or beneficial ef fect on boiling crisis. PWR test data indicate that s;,acer grids have a beneficial ef fect on boiling crisis. The postulated mechan'sm associated with the adverse ef fect of spacer grids in BWR's is the formathn of a norsesnoa vortu just upstream of the spacer grid This and the l corresppnding dryout of tne liquid film at the " fuel" surface. I

\

l 4

- mechanism will now be examined to determine whether such an ef fect is possible in C-E PWR's.

Formation of a horseshoe vortex in adiabatic air / water tests (6) improved mass transfer between the water film and air to the extent that dryout of the water film occurred. It is unlikely that such a vortex will form upstream of the obstruction presented by C-E spacer grids, since the grid obstruction dif fers greatly from the type of obstructions used in the air water tests. A comparison of Figs. Il-2 and 11-3 shows that the air / water test obstructions were in contact with the " fuel" surface over a large area while the obstruction presented by the spacer grid is located primarily in the center of the flow channel with -

relatively 1ittle fuel surface contact. '

While the difference in geometry makes the fonnation of a vortex at the leading edge of the grid unlikely, such a vortex would have a beneficial effect on CHF at typical PWR conditions. A vortex upstream of a PWR spacer grid would inprove heat transfer between the fuel surface and bulk coolant flow in a manner analogous to that in which mass transfer was improved in the air / water tests and also by bringing colder This improved heat transfer would retard water closer to the fuel rod surface. Hence, the formation of a vapor blanket at the fuel surface thereby increasing CHF.

if a vortex did form upstream of a PWR spacer grid, it would have a beneficial ef fect on CHF.

Formation of vortices upstream of spacer grids is postulated to have an adverse effect on CHF in BWR's. It is unlikely that such vortices could form in PWR's because the obstruction created by the spacer grids in subchannels differ greatly from the types of obstructions considered in the air / water tests.

However, if such vortices did form, local heat transfer would be improved and CHF would thus increase in PUR's. Therefore, none of the BWR test data can -

be used to demonstrate that spacer grids will decrease CHF in PWR's. ,

Conclusions The flow regimes and mechanisms associated with boiling crisis are different in BWR's and PWR's. BWR's typically exhibit annular flow; dryout of the liquid film on the fuel rods is the mechanism for boiling crisis in this flow regime. Subcooled and bubbly flow are encountered in PWR's and DN8

.or vapor blanketing of the fuel is the mechanism for boiling crisis in PWR's, BWR test data show both beneficial and adverse effects of spacer grids on CHF.

f The postulated mechanism for the adverse ef fect is the formation of a vortex upstream of the grid which accelerates dryout of the liquid film on the fuel.

Because of the difference between the type of obstructions used in the tests that support the formation of vortices ar;d the obstruction presented by C-E spacer grids in flow subchannels, it is unlikely that the vortices will appear in the flow subchannels. Even if vortices did form, they would improve heat

~

transfer between the fuel surface and bulk coolant thereby increasing CHF in PWR's.

All PWR and some BWR test data demonstrate that CHF increases in the presence of spacer grids. It is unlikely that the postulated mechanism (vortex) associated with the adverse ef fect of spacer grids in CUR's can exist upstream of PWR

. grids. If this mechanism did exist in PUR's, it would be expected to increase There fore ,

CHF because the PWR flow regime dif fers from the BUR flow regime.

all experimental evidence indicates that spacer grids increase CHF in PWR's.

s BWP, PWR .

AWAA2 RCW SU800AfD 4 auflu stCW tl

' 01 ~g" f, 0 4

^

,5 06 ~/ - - ,

h85 I 5 'S ff (

i ::"  : Ti:'r

[t{ ; " l; l.

g- "u ,

f - 0,  ; .! ,

Wh to t til 0 til e 00 r- es t 0 rit c Ou [j  ; ggycy, P,- ,

,'E I . '

'. s: a 1:' , $h i

'5' ':r

, 1 s ., .

s' wutsu i e si c'

c wcicut

UOv!) Cott c, 5, b 5. f l

I {} >Ayt3 I AvtulUS l

Fig ~ure II-1: Comparison of Flow Regirres flear Boiling Crisis in PWR's and BWR's (from Reference 1) l l

l t

l t

I s

FLOW _

08Sf AUCT10N ICY LIN O( Al

• Hon $($ HOE VORTEX L10VIO FILM L10VIO FILM OAY PATCH f 4- O ~> -4~ O ->

Figure 11-2: Effect of Flow Obstruction on Liquid Film in Annular Flow (from Reference 5)

G 4

o

l J \

- l t

.4 .

. . . s.

9

/,.r.. -

/

/

/

l

(

Figure II-3: Projected Area of C-E Spacer Grids in a Itatrix Subchannel l

9 h

e e ,

-9

~

. III. EFFECTS OF SPACER GRIOS Illfl0il-Uflir0Rf1 APO.CHF TESTS Introduction Spacer grids enhance heat transfer and have a beneficial effect on CHF in PWR rod ,

bundles, as explained in Reference 8. Hence, observed CHF in PUR test sections is greater in the presence of spacer grids. The following discussion will demonstrate that the effect of spacer grids is amplified in non-uniform axial power distribution (APD) test sections. Enhanced heat transfer downstream of a grid postpones the occurrence of 0llB until just upstream of the following grid. Thus, two effects are combined in non-uni form CHF tests. First CHF is greater because of the presence of the spacer grids; second, because CHF is delayed until just upstream of a spacer grid, the local heat flux is typically lower than would have been found in the absence of the grid. ,

The CE-1 correlation is based on uniform APD test data from test sections with standard C-E spacer grids. The correlation therefore implicitly contains the beneficial effect of these spacer grids in increasing CHF in uniform APO bundles. Tests have shown that boiling crisis occurs at lower local heat fluxes with non-uniform APO test sections than with their uniform APD counterparts.

However, the CE-1 correlation does not include models to explicitly account for the effects of non-uniform APO's or the local ef fects of spacer grids within the grid span. The Tong F-factor is used with the CE-1 correlation to account for the effects of non-uniform APO's. However, the coefficients involved in the F-factor have not been re-optimized for use with the CE-l correlation; '

nor have the correlation coefficients been re-optimized to accommodate the F- fac to r. ,

The use of the F-factor with the CE-1 correlation, w*th no explicit modeling of local grid effects in the correlation gives rise to conservatism and increased scatter in the data. The conservatism and increased scatter arise because of. the real effects of spacer grids in the non-uniform test sections. The relationships between the ef fects of spacer grids and the increased conservatism and scatter will be explained qualitatively. This explanation demonstrates that the use of the CE-1 correlation in conjunction with the Tong F-factor yields conservative CHF predictions and is therefore suited for use in design analyses.

Description of Phenomenon' The CE-1 correlation (15) is based upon 731 data points from uniform APO sections with standard spacer grids. The distance between the uppermost spacer grid and the end of the heated length in these test sections was purposely made equal to the nominal grid spacing in order to provide the most conservative DNB test results. This design ensured that DNB occurred at the end of the heated length. Thermocouples in these test sections were located

-0.5" below the end of the heated length. Depending on the test section under consideration, the thermocouples were located between[ ]down s tream of the downstream edge of the uppermost spacer grid in the test sections. Spacer grids enhance heat transfer downstream of the grid; however, this enhancecent decays with distance downstream of the grid. Hence, although the CE-1 dat" include a benefit resulting from the enhanced heat transfer do'.instrean of the grid, the benefit is small, since the thennocouples are relatively far dcenstreem of the spacer grids.

10-

- e t+g-->t -=-

y +.- e.--y +e,- - a- ><e.<--,.---e.- --.m.- e.-- - .4 -vr 9- yr- --wtg --

-re-- -e

The ef fect of enhanced heat transfer on CHF for uniform APD's is illustrated in Fig. III-1, which shows a plot of CHF and local heat flux vs. distance from the test section inlet. Fig. III-la shovn the hypothetical CHF profile for a bare test section with no spacer grids. The CHF in this hypothetical test I l

section starts out at its maximum value and decreases with distance from the inlet as quality increases. The power level is chosen such that DilB is observed l at the end of the heated length for an average heat flux Pj . l The CHF profile for a uniform APD test section with grids is presented in Fig. III-lb.

The CHF profile without grids is superimposed on Figure III-lb for reference. The effect of the spacer grids on CHF gives rise to the " sawteeth" in Figure III-lb.

Immediately downstream of the spacer grid, the enhanced heat transfer increases the local CHF. Since the enhanced heat transfer decays with distance from the However, the spacer grid, local CHF decreases with distance from the spacer grid.

effects of the enhanced heat transfer do not disappear completely before reaching the downstream grid. The enhanced heat transfer benefit associated with the grids gives rise to a higher CHF value at the end of the heated length. The test section with no spacer grids experienced CHF at the end of the heated length with an average heat flux P1 . The enhanced heat transfer associated with the presence of spacer grids causes the test section with grids to reach an average heat flux P2 before experiencing DNB at the end of the heated length. The fact the P2 IS greater than Pj is a direct result of the enhanced heat transfer associated with the spacer grid.

In summary, the presence of spacer grids in the test section has the following two effects on the observed CHF in uniform APD tests:

1) increase

2) increase ininpower CHF (P in vs p

crossing)a Pj spacer grid (" sawtooth ef fect")to DNB resul tin at the end of the heated length.

The ef fects of grids in non-uniform APD test sections are more complicated than in uniform APD test sections because the location of DNB shif ts in the non-uniform tests as a result c,f the spacer grid heat transfer enhancement. CHF profiles for two non-uniform APD test sections are presented in Fig. III-2. The CHF profile for a bare non-uniform test section without spacer grids is shown in Fig. III-2a.

A CHF profile for a non-uniform APD test section with spacer grids is shown in Fig.III-2b. The bare test section CHF profile is superimposed on Fig. III-2b for reference. As in the discussion of uniform APD test sections, the local ' sawtooth" ef fect and the increased power level to DNB resulting from enhanced heat transfer in the presence of grids are apparent in Figure III-2b. A third effect also appears in the non-uniform APD test with grids.

DUB occurs at the intersection of the local heat flux profile and the CHF profile in Figs.III-2a and III-2b. The curves intersect at point A for the bare test section and point B f0P the test section with grids. A comparison of the CHF curves for the bare test section and the test section with arids in Fig. III-2b shows that both the power level at DNB and the location of DNB change because the grids are present.

- If the location of DNB did not chanqe, the averaqe heat flux could be increased to Pj 'before the local heat flux and CHF profiles intersected at point A' . However, as power is increased above P), the local heat flux and CHF profiles intersect at pnint B at an averar;e heat flux P9 Thus, the location of O'S shif ts due to the presence of grids. Three ef fects" occur as a result of the presence of grids in non-uniforn APD test sections:

.11-

1) local CHF increases in crossing a spacer grid (" sawtooth ef fect")

2) average power to DNB increases as a result of the enhanced heat transfer associated with spacer grids

3) location of Otl8 shif ts to just upstream of a spacer grid because of the enhanced heat transfer associated with spacer grids.

Effect of Phenomenon on CHF Correlation The CE-1 CHF correlation is based upon uniform APD test data from test sections with standard spacer grids. Part of the increase in power to 0:1B associated with the presence of spacer grids is therefore implicitly included in the CE-1 correlation, as explained earlier.

Measured and predicted CHF values for the CE-1 correlation with the Tong F-factor were compared for 4 test sections with nonuniform APD's in Reference 16. An

• evaluation of the measured to predicted (M/P) ratios showed that D.e predicted CHF was consistently lower than the measured value and that there was more scatter than for the uniform data (i.e., the standard deviation was larger). Both of these observations can be explained by the ef fects that arise from the presence of grids in non-uniform APD test sections.

As explained earlier, the presence of spacer grids increases the power to ONB and changes the location of DiiB in non-uniform test sections. The consistent underprediction of CHF by the CE-1 correlation (with the Tong F-factor) is explained by the additional increase in CHF associated with the presence of spacer grids not accounted for.by CE-1 and furthermore by the fact that the coefficients in the Tong F-factor were not recorrelated for use with CE-1.

The increased scatter in data for the non-uniform test data is also explained by the effect of the spacer grids. As already explained, the CE-1 correlation does not explicitly model the local ef fects (i .e. , the " sawtooth ef fect") of spacer grids on CHF. Consequently, the correlation nay ont accurately predict the axial location of CHF in non-uniform test sections. Application of the CE-l correlation in design analyses recognizes this and thus, Minimum DNBR is used as the design criterion regardless of the location. The location of DNB may be predicted at any axial level by the CE-l correlation in the non-uniform test sections. However, the presence of spacer grids in the test sections causes DNB to occur just upstream of spacer grids. Consequently, when N/P ratios are computed for a particular test point, the predicted location of DNB l generally differs' from the observed occurrence of DNB. This difference in axial l location results in the increased scatter found in the M/P data for non-uniform

) test sections.

The effect of the difference in predicted and actual location of DNB is shown in Fig.III-3. The CE-1 prediction line is shown tangent to a non-uniform heat flux profile labeled q"2 (z) at the predicted DNB point. However, the actual CHF line, with spacer grid ef fects included, is above the prediction line except at locations irrmediately upstream of the spacer grids. Another non-uniform heat flux profile labeled q)" (z) is shown intersecting the actual CHF line at grid #2, the actual DNB location. The distance "X" between the two heat flux profiles represents the magnitude of CHF power underprediction that results from not including the spacer orld effect explicitly in the CE-1 correlatiori. The example shown in Fig. III-3 represents a case where the predicted DNB location l

l l

.l l

l I

If the predicted location f is approxinately midway between two spacer grids. Because the were at a spacer grid, the underprediction would be zero. i predic ted locations will be more or less randomly distributed with respect to the spacer grids, the set of underpredictions, X. , will have a mean greater than 1.0, and a variance. It is estimated that thid mean and variance are of adequate tagnitude to explain part of the conservatism of CE-1 when it is applied The remainder of to the non-uniform data as well as most of the increased scatter.

the conservatism is due to the F-f actor.

Thus the underprediction of CHF by the CE-1 correlation in non-uniform test sections results from the presence of spacer grids in the non-uniform test section and the fact that the coef ficients of the Tong F-factor were not The increased scatter in M/P re-optimized for use with the CE-1 correal tion. ratios for non-uniform test includes no grid effects model and, consequently, the axial location of predicted and observed DMB differ because of the presence of spacer grids in the test section. Increased scatter also results from the use of a non-optimized F-factor which increases variance, as shown in Section V.

d 5

s 4

\

CRITICAL HEAT FLUX HEAT FLUX LOCAL HEAT FLUX DISTANCE FROM INLET

1. Fiaure.III-la CHF PROFILE WITH UNIFORfA APD AND NO GRIDS

\. -

s CRITICAL HEAT FLUX

's  % L (TEST SECTIOfJ WITH SPACER GRIDS)

N \ s s's s s Nj s HEAT FLUX

's s ss \' s ,

f CRITICAL HEAT FLUX ss ,

(TEST SECTIOrl WITH NO SPACER GRIDS) 's's **

s's% .*

l , ' 8 l

LOCAL HEAT FLUX s'*s' P

[ (TEST SECTION WITH SPACER GRIDS)*%,'

2 LUGAL HEAF f LUX

! p . . . . . ( (TEST SECilOff WITH NO SPACER GRIDS) '*m DISTANCE FROM INLET l

i

. SPACER GRID LOCATIONS Fiaure III-lb CHF PROFILE WITH UNIFORM APD AND SPACER GRIDS g- -..,,'e , _r+yww.,w.,..,.=,y..g, r, , - - , ,,,,.~,.~.y,,..-,,...,,7,.,yr +ye,,..,, ,c, , . ,,-,,-,---,.g.-, . . , , - , , w..- yr.ww,. , - . . - .=, r.,--- , , ,',,,,v.

9 CRITICAL llEAT FLUX LOCAL llE AT FLUX IIEAT A FLUX

• AVERAGE POWER e, ----- ------------------------ .

t OlSTANCE FROM INLET Fiqure III-2a CHF PROFILE WITil NON. UNIFORM APD AND NO GRIDS f '

CRITICAL FIEAT FLUX (TEST SECTION WITH GRIDS) llEAT FLUX

~~

• %s g CRITICAL HEAT FLUX (TEST ' ~ ., A, SECTION WITHOUT GHIDS) '~~, ,,,

• 8 LOCAL HEAT FLUX WITil GRID IF DN8 AT A'

/.- -

~'

~ , " . .

... ,, f .' , . - ---.y .- g..

AVERAGE I A ..

POWER P

k('WlT LOCAL HEAT F LUX (TEST SECTION b .,,

p 2g',', ~ ~ * *[..['d H SPACEI),GltIDS) _,_,_,_,

y'*j ' / _ LOCAL HEAT FLUX (TEST SECTION WITilOUT SPACER GRIDS) \,N

/* \

DISTANCE FROM INLET SPACER GRID LOCATIONS Figure ill-2b CHF PROFILE WITil NON UNIFORM APD A.'!D SPACER GRIDS

-TS-2

. - - . . - , . . - - - - .-w.,_,v. , , - . . ..,,-,m-m,,.....,_w,-,-,...,,m.. , , , . . , .,,w,7.._-n,, . . , , , _ . , , - , - . , , - . _ , ,, . -. , , , , - , c- me . -

04

\

\ ACTUAL DNB

\ .

LOCATION N

'f N Q

'Deo N

-_ X Y'

PREDICTED Di18 -

,} ..

LOCATION N

Q(L X = UNDERPREDICT10fl GRID //2 GRID #1 Figure III- 3:Ef fect of Dif ference Be tween Predicted and Actual Locations of DNB

.. _ _ - . - .~ _. -

IV. EFFECT OF MULTIPLE DATA POINTS A stated concern of the NRC staf f has been the use of multiple data points from the CHF tests. Multiple data points are those CHF indications during a test that arise because of the multiplicity of instrumented rods and thermo-couples within a rod. The thermocouple which gives the very first indication of CHF during a test provides the data point identified as the primary indication.

In most tests CHF indications occur at other thermocouple locations slightly later. These are identified as secondary indications. The use of primary and secondary indications in the development of CE-1 has prompted the following question on another docket. "In the development of the CE-1 correlation with uniform axial power distribution (CENPD-162), multiple data points have been used for some heater rods with quadrant instrumentation in non-matrix subchannels.

Therefore, the statistical analysis should be re-performed by eliminating the -

redundant data points. Provide, based on the reduced data, new minimum DNBR limits for 14x14 and 16xic rod bundles, separately and jointly".

A technically consistent response to this question would require re-correlation of the new subset of test data (i.e. , the subset with " redundant" data points eliminated) to produce revised values for the constants in CE-1. In lieu of this, a conservative approximation to the requested information has been obtained by performing a statistical analysis of primary CHF indicatiens only, using un-revised CE-1. This approach is conservative because the constants in CE-1, not being optimized for the data subset being examined, will introduce additional scatter into the resulting measured / predicted CHF ratios.

In Table IV-1 means and standard deviations for the ratio of measured / predicted CHF are shown using data for only the primary CHF indication in each test run.

The corresponding 95/95 MDNBR limit is given separately for 14xl4 and for 16x16 fuel types, as well as for the complete primary indication data set.

For comparison, the same information is provided for the total CE-1 data base (including multiple indications).

It can be seen that the primary-indication-only statistics compare well with the total CE-1 data base statistics, and continue to support an overall 95/95 MDNBR limit of 1.13. This is further evidence that the design MDNBR limit of 1.19 is adequately conservative to account for any uncertainties associated with the SONGS high impact fuel design using HID-2 grids.

i

- - - - - . . . -- - . . 7 ,=

~

M iV-1 ,

.COMPARIS0ft OF [0TAL CE-1 OATA BASE .

,~

WITil PRIMARY ONB Ifl01 CATI 0tl SUBSET Type 14x14 16x16 Ho, Rods 21 21 25 ' 21 21 25 Length in ft. 7 12.5 '7 7 12.5 7 TOTAL

~l41 99- 51 169. 157 Il4 731 Al1 s n. Data -

. Number of Data Points 70 52_ 55 331 .

Primary 72 45 .37 Data .

4 a

1 v Mean of CilF Meas.

'CITF Pred.

0 ^

Standard '

l Deviation

l. 1. of

/r d -

1.133 All 1*132 1.139 95/95- Data l HOllBR 1.123 1.136 1.122 l Primary I. . Data o

' e 9

4 3

1. D

r18-1 .

c V. STATISTICALEVALUAl'IONOfCHfDATA Introduction In response to NRC requests, a statistical analysis has been performed on j' critical heat flux (CHF) data from C-E CHF experiments. The data evaluated are those from experiments with both uniform and non-uniform axial power distri-butions ( APD). The analysis was performed to determine if there is a statistically' j significant effect of bundle type (i.e.,14x14 vs.16x16) and to establish the significance of test section length for the uniform data using the methodology of Reference 17.

l To perform this calculation, analysis of variance is used. iiiis approach allows one to -extract from the total variance for the experiments, the components of variance between groups (e.g. bundle type), between test sections and within test sections. The F-test is then applied to determine the statistical significance ,

of the observed variances.

Statistical Methods and Analysis ,

Since an analysis of variance is a method for separating the total variance into its components, the data are classified according to the postulated causes of ,

va ria tion . It is possible to classify the data with respect to each source of i

variation and the complete classification of these sources is a necessary first i step. This can be done with a hierarchic classification. When CHF tests are performed such that data is taken with many test sections of two different designs, it is possible to separate the variance of the data due to the design difference,

, the variance of the data between test sections and the variance of the data ,

within a test section. For the case where the variance due to 14x14 and 16x16 l designs is needed, the hierarchic classification can be shown schematically as in Figure V-l.

In accordance with the hierarchic classification, an analysis of variance table is constructed as shown in Table V-1 (Ref.17 pg. 165). The true variance j

between dif ferent gropps (bundle typesi is 2 , the true variance between subgroups l (test sections) is o j and the true variance within subgroups (test sections) is o g . These quantities are unknown and must be estimated from the data.

jj and M is an estimate 2

Hg is an estimate of a g ,M j is an estimate of ga 2+no 2 The F test is applied to ratios M j/Mgand M2 /H1 '

of o 2+n#21 +i 2 g

If a null hypothesis is formulated which states that there is no test section 2 2 effect or bundle type ef fect, then the values of a j and 2 are zero. Three 2

independent estimates for a g can be obtained;one from the mean square between test bundle types, (M ), ne from the rean square between test sections (Mj ), and 2

one from the mean square within test sections (Mg ). If we formulate the ratio of M /N , and of li j/Mg, the magnitude of these ratios quantifies the extent 2 1 f, to which the means square deviate fren each other and thus the extent to v;hich the estinates of a 2 and a deviate from zero. Obviously, large deviations 1

i i - 19 '-

l ,

1 -.. . . -.. -. . - .-..

from zero are a measure of the weakness of the null hypothesis of o1 and2 "2 2

, that. oj and "2 are zero.

Thus, the F-test is formulated as:

=

F j Mg y =

2 n

1 and a large value of F rejects the null hypothesis at a particular significance icvel. Tabulated values fnr significant magnitudes of F for various probabilities and degrees of freedom can be found in many references (see Ref.17). .

A similar procedure is applied to examine the effect of bundle length for the uniform axial power distribution experiments.

The evaluated quantity is the ratio of measured to predicted CHF, therefore, the statistics become a function of the CHF prediction method. In the following analyses, the F test is applied to several sets of data generated in different

. ways from various types of experiments. It is therefore convenient to subdivide the analysis into cases. These are identified as Cases A thru D. Data points outside the range of the correlation are excluded.

CASE A This case presents the analysis of variance for uniform axial power distribution data analyzed with TORC /CE-1. The components of variance between bundle groups according to bundle type (14x14 versus 16x16), between test sections and within test sections are estimated. An F-test is applied to the estimates.

CASE B This case presents the analysis of variance for uniform axial power distribution data analyzed with TORC /CE-1. The conpcnedts of variance between bundle groups according to bundle length (12.5 f t vs 7.0 f t), between test sections and within test sections are estimated. An F-test is applied to the estimates.

CASE C This case presents the analysis of variance for non-uniform axial power distri-bution data analyzed with TORC /CE-1 and the Tong F-factnr. The measured to predicted ratio is calculated at the axial position of mininun predicted DNCR. The components of variance between bundle groups according to bundle typc (14xl4 vs 16x16), between test sections and within test sections are estimated. An F test is applied to the estimates.

CASE O This case presents the analysis of variance for non-unifona axial power distri-lGion data analyzed with ici":/CE-l without the Tong F-factor. The neasured to predicted ratio is celculated at the axial position of minimum predicted DNBR.

-20'

e The components of variance between bundle groups according to bundle type (14x14

~

vs 16x16), between test sections and within test sections are estimated.' An F-test is applied to the estimates.

Resul ts

" The basic statistics (number of points, mean, standard deviation) for all four cases are given.in Table V-2.

The analysis of variance as described above has been applied to the noted cases.

When the technique is applied to Case A, results as shown in Table V-3 are obtained,

= 30.34 is much b l t d value lhe of 2.37 calculated (takenvalue of Fj.95 at the confidence andlevel)

F =greater than less

.0066 ic much the than ta u the ae This shows that tabulated value of 7.71 (taken at the .95 confidekce level).

the variation from test section to test section is significant but the variation from bundle type to bundle type (14x14 versus 16x16) is not significant.

Therefore, the hypothesis that bundle types have no ef fect cannot be rejected on statistical grounds.

When the technique is applied to Case B results as shown in Table V-4 are obtained. The calculated value of F = 12.71 is much greater than the tabulated value of 2.37 (taken at the 0.95 confidence level) and 2 F = 5.49 is less than the tabulated value of 7.71 (taken at the .95 confidence 1evel). This shows that the variation from test section to test section is significant but the variation from bundle length to bundle length (12.5 f t. versus 7.0 f t.) is not significant. Therefore, the hypothesis that bundle lengths are not dif ferent cannot be rejected on sta tis tical grounds. The statistical statement does not necessarily preclude the existence of a length effect. The value of F2 = 5.49 is relatively close to the tabulated value of 7.71 which may suggest a weak dependence.

The length dependence, however, was not iricluded in the CE-1 correlation for the following reasons:

'l. Any dependence on heated length is weak.

2. A dependence on heated length is inconsistent with the " local conditions hypothesis" which has been used extensively and successfully for correlating CilF data.

3. The source data include a substantial body of data for the most " adverse condition", i .e. , a heated length of 12.5 feet.

4. The true effect of heated length, if any, is likely to be dependent on axial flux shape.

The two cases above are applied to experiments where axial power distribution is uniform. A similar analysis is done for non-uniform axial power distribution V-5 are da ta . When the technique is applied to Case C, results as shown in Table obtained. The calculated value of F = 45.26 is greater than the tabulated value

=.24 is much less than the tabulated of 3.07 (taken at the .95 confidence) level) and F value of 18.5 (taken at the .95 confidence level)2 This shows that the variatio from test section to test section is significant but the variation from the Therefore bundle type to bundle type (14/14 versus 16x16) is not significant.

the hypothesis that the bundle types have no ef fect cannot be rejected on statisti'.al grounds.

- Similarly results for Case D are shown in Table V-6. The calcule.i: value of Fj - 100.96 is greater than the tabulated value of 3.07 (taken at : N .95 confidence level) and F is much less than the tabulated alue of 18.5 (taken a't the .95 confiben=c.062 e level). This shows that the variati'.- from test section to test section is significant but the variation from bun:^e type to bundle type (14x14 versus 16x16) is not significant. The re fo rt , r1 hypothesis that the bundle types have no effect cannot be rejected on statis.':al grounds.

Conclusions Several conclusions can be drawn from the results of the analyses :erformed here. The first and most important is that.when using a variety t test and analysis ccnditions. the hypothesis that the bundle types have .o effect cannot be rejected at the .95 confidence level. This is proven by the consistently small values obtained for F2 '

The large values obtained for F), show that there is a significant iariation between test sections both for uniform and non-uniform data. This :ehavior is observed quite consistently in C-E's CHF data as well as that ' ri other vendors (Reference 18). An analysis of variance of the Reference data has shown large values of Fj leading to the conclusion that significar: variation between test sections is a random effect inherent in CHF testing. or the uniform data, the between section deviation given in Tables V-3, 4 is ver/ :onsistent with repeatability analyses in the literature. This isTheto non-uni be expe :ed because f;~, data has of the randomness of the causes for the deviations.

a greater deviation, however, it is emphasized that the CE-1 corre:ation was not refit to this data base.

Other aspects of the comparison between the uniform and non-unifor- data should be discussed. From Trbles V-3, 4, 5, 6, it can be seen that the c.erall deviation is greater for the non-uniform data than for the uniform data. Fv:hermore, the overall deviation of the non-uniform data is greater with the Tong : factor than without this factor. As described in Section 111 the increased deciation within test sections is explained in part by the grid spacing of the C-E casign and the favorable ef fect of spacer grids in non-uniform CHF tests.

In addition, the C-E design dininishes the upstream history effect odelled by the Tong F-factor because of the presence of the grids and the relatively close grid spacing. The F-factor is based on data where no grids ere present.

l l

The same discussion applies to the conservative shif t of the non-u.iform data i relative to the uniform data. This conservative shif t is evident J.on comparison of Figures V-2,3.

It is concluded that the cambination of the above noted ef fects (grids, grid spacing, F-factor) are the causes of the laroer deviations and hig'er means noted in the non-uniform data, and do not cause any concern raer the

- validity of the correlation of the other significant parameters. :t is also concluded that the method of application of CE-1 and the F factor proposed for SONGS provides sonstantial conservatism relative to the C-E 0liCR c iterion.

~ This criterion states that the inim I'mR shall be such as to L ',ide at least a 95n probability with 95; confidence that departure fron tvieate bo,iling (DiB) does not occur on c fuel rod having the ninice OE

during This steady-state operation and anticipated transients of moderate f reg ancy.

conservatism is evident in figures y-2, 3, 4.

. _ ~ _ _._ __ ._. ._ _ . _ . _ '" E . _._. _ . _ _ _ _ . _ _ _ . . - __._..

~

1 l

TABLE V-1 l

Degrees of Mean Quantity Estimated Sources of by the Mean Square j Sum of Squares Freedom Square _

Variation Between Groups (Bundle Type S I2 N' 2 +"2 1 +I"2 2

14x14 vs 16x16) i Between Subgroups 2- 2

' Test Sections) 3 1

I l "1 "o +"I 1 2

Within Subgroups S f Mg o g

o o (Test Sections) where: o g = true variance within subgroups (test sections) o = true variance betvieen means of subgroups (test sections) z

= true variance between means of groups (bundle types) c2 ny ,E = estimates of the average number of observations (data points) 2 in subgroup (test sections)

Ti = estimate of the average number of observations (data points) .

per group (bundle type)

S "2

= S)

Mj S

o

>1, =g-o 6

TABLE V-2 Bundle Type Axial Pcwer Bundle No. of Mean of Std. Dev.

Distribution Length Data Points Ratio of of Ratio :

Measured to Measured t:

Predicted CHF Predicted C Case A,B 14x14 Uniform 7.0 ft 51 1.001 .0293 14x14 Uniform 7.0 141 1.028 .0550 14x14 Uniform 12.5 99 .963 ' .0637 16x16 Uni form 7.0 114 1.036 .0732 16x16 Uniform 7.0 169 .991 .0a53 16x16 Uniform 12.5 157 .980 075' T' 1.119 .lC5 Case C 14x14 1.68 Top Peaked 12.5 74 14x14 1.68 Bottom Peaked 12.5 82 1.287 .130 16x16 1.46 Symmetric 12.5 108 1.237 .122 16x16 1.47 Top Peaked 12.5 106' l.254 .053 Case D 14x14 1.68 Top Peaked 12.5 74 1.000 .085 14xl4 1.68 Bottom Peaked 12.5 32 1.170 .091 16x16 1.46 Symmetric 12.5 108 1.070 .lC3 16x16 1.47 Top Peaked 12.5 106 1.155 .066 i .

1 .

i TABLE V-3 Case A: Uniform Axial Power Distribution Analyzed with TORC /CE-1 Between Bundle Variation Due to Bundle Type Effect Estimated Source of Sum of Degrees of Mean Standard Deviation i

4 Variation Squares Freedom Squace (o)

.0008 .0008 0 Betuacn Bundle 1 Type

, (14x14 vs 16x16)

Between Test .4700 4 ' .1175 .0 31

. Sections DS '

8 Within Test 2.8273 725 .0039 .062 Sections 4

M I

117.4 - 30.34 (Tabulated value for F for confidence n) = F) - M .95 is 2.37) 0 n = 126.7 2 M 5' = 350.3 F2 _ 2 .0066 (Tabulated value of F for confidence

.95 is 7.71 M)

= r TABLE V-4 i

Case B: Uniform Axial Power Distribution Analyzed with TORC /CE-1 Between Bundle Variation Due to Length Effect Estimated Sou rce Sum of Degrees of Mean Standard Deviation of Variation Squares Freedom Square (c)

.2725 1 .2725 .025 Between Bundle Length (12.5' vs 7.0')

.1983 4 .0496 .020

, Between Test y Sections

~

Within Test 2.8273 725 .0039 .062 Section M

nI = 109.6 Fl - I - 12.71 (Tabulated value of F for confidence M

.95 is 2.37 )

n, = 134.7 p N - 332.7 F2 _ 2 _ 5.49 (Tabulated value of F for confidence .

M)

.95 is 7.71)

^ '

TABLE Y-5 .

Case C: Prediction Method - TORC /CE-1 with Tong F-factor Prediction at Minimum DNBR Location

-Estimated Scurce of Sum of Degrees of Mean Standard Deviation '

i Variation Squares Freedom Square (c) 1 .1311 0 Between Bundle .1311 Type l (14x14 vs.16x16) l 4'

, 7" Between Test 1.1133 2 .5567 .077.

l Sections l

4.5120 366 .0123 .111 Within Test Sections M

I nj = 92.4 F) - - 45.26 (tabulated value of F for confidence .95 is 3.07)

M n2 = 90.3 M-

__ z N = 0.5 F 2 * !!~

(1

= .24 (tabulated value for F for confidence .95 is 18.5)

TABLE V-6 Case D: Prediction Method - TORC /CE-l laithout Tong F-Factor Prediction at Minimum DNBR Location.

Estimated Sum of Degrees of Mean ' Standard Deviatica Source of Variation Squares Freedom Square ( o)

.0465 1 0465 0 Between Bundle Type (14x14 vs. 16x16)

Between Test 1.5106 2 .7553 .090 Sections k

? 2.7380 356 .0075 087 Uithin Test Sections

=

U 1

1 = 92.4- F) -100,96 (Tabulated Value of F for confidence M

0

.95 is 3.07) n2 - 90.3 M II= 180.5 F -

2

=

062 (Tabulated value of F for confidence M .95 is 18.5) 1 t

g_2_,

. _ , _ ___-- -__e__=-_-_m=___==-=_--_m-

- -_ -- =- . - - = _ _

_ __-~.____v____=.=_-____ _

==__ _--=-----m__._t

_ - _ - _ _ - - - - _ _ - _ , . = _ _ , _----___m.___.- -

_m.. _ _ . _ _ _ _ __ . _ . - - - . _ - -

.._._.._,_.m__ -- -

. . _ . _ _ . . = _ _

- - _..,._,~=_.m, __ , _._

j All CHF Data ,

14x14 16x16 (GROUPS)

Data Cata

\

/

l Test Sections (SUBGROUPS) Test Sections

1 1 2 gj3  ;,1 lf2 }F 3

!'. I'. s\ h\ l \\\ , l Individual CHF Data Points l

Hierarchic Classification FIGURE V-1 l .

I l

--. - g -e. ,-<m -mp- . + + , m- -m - ,,e ,++-v y y- - ---i-

+

J

- 1.0 4 i I i N -

g i .- ..

g .

~

I '- ..-

a. 0,8 _. .

i ., .: .

r- .

cQ-o 2. 8 .'<'. ' . '

a .

n. . ...::.

~ > . . , .

• ., ~p; e y "

=a 0.6 -- .- --' .I.: ; . -

u ' .<,

. 7. 4U '.'.

• 1.19 DilBR p

<C .sh.!h ' .. ~ Limit .

.w .. ..

.x

!>.s,..

.- =#n .- .. y. .

o 0.4 - ,.:4.w- >.

p .

, ,gS..

.x . . .:~'..-

u .

O ,/ ~

w e 0. 2 -

a m

w

E

' ' ' i O

0 0.2 0.4 0. 6 0,8 1.O 0 2 PREDICTED CRITICAL HEAT FLUX,10 BTU /HR-FT l

i Fie;ure V-2: ficasured and Predicted Critical Heat Fluxes from Uniform CHF Tests and the CE-1 Correlation l

P l .

l.0 ^/

w I. I I I W ao a

- E a

- 0 - "

. >g; 0. 8 - -

S 8 go x_[. gg *

• a o cd 0.6 - -

1.19 Oll8R

}--

b

_ -A or Limit '

g Lelf-

<C

$ 0.4 - .

E 0" o

b m '

D 0.2 - -

5 2

4 I I I 0.0 I

, 0.0 0.2 0.4 0.6 0.8 1.0 2

PREDICTED CRITICAL HEAT FLUX,106 BTUlHR-FT ,

0 14 x 14 ASSEMBLY GEOMETRY, FLUX PEAK TOWARD OUTLET A 14 x 14 ASSEMBLY GEOMETRY, FLUX PEAK TOWARD INLET o 16 x 16 ASSEMBLY GEOMETRY, SYMMETRIC AXIAL FLUX SHAPE 0 16 x 16 ASSEMBLY GEOMETRY, FLUX PEAK TOWARD OUTLET

! Figure V-3: tieasured and Predicted

!. Critical Heat Fluxes from ?!onuniform CilF

{. Tests and the CE-1 Correla tion f

\ .

l .

. .  : i ,

- - - - - , i 4 , i

, f.o .

' ' pp l

- . i

.9 .

i i

. . I i

i ;; e ,

++

' l. $ #

b8  !

. Ik>;J n th ,t

+

v l ,

i '

+b t t to @Ifh+ - -

o,

- i i

- +

. ,4lt ,

) ,' m.+, 4, lig*,, .

's 7 '

., ,i ,' ti' ,

ytph;4t l l' vi i . t +I ,t .t g th pI f4'  :

x, . ,

, . 4

• qh' J S .YS#

19 DNBR y, G -

t i

tre

'.,fth'+lfM

++ , , * , * -

9,  ! ,

, , , , .,1

+ . L. imi t i. ,

,,b e ' < . +'- q $+>.

a. , . I ). ' -

Q* ' '

' ' I 5;.5 , b i$v. ena e l

-j , -

he ,'d,g,'gN;I.).l/NNi d [,',R ./l . ! ,

l,'g ,

o ,

p: ,wt -

,. +' 4 , ,q r t i ,

' ., , . j, E I

, & ,'l ,

, i

- '*INrf;if+'g4 j!'# j ,

l

.i  ; ;., ,:

e

+ ti (I'Jff f ,. [ ?,+5 .

. 'l <i

- l ;. {

. . .g.. .

e ce o ,3 -

p + ,n.d,,, f , 5ip p , .

1. ,c g s't ' ,i p '

1 f,lf/r  ;'

i . .

i

+

. j. I sk -

W ;p ,t, ,.

1 ,

c ' '

,, '+ , . ,

,z p '

1 I

t

• p i 9

./  :

i .

/

/

0 ,9

.O ,/ , . 2. ' .3 & .5 .4 *T . 8 /. O PR ED)C,TE D CRtrtC AL HE AT Flux - (,10' Bvv/H.v Fv8) +

Figure V-4: Measured and Predicted Critical lleat Fluxes for Coth Uniform and Nonuni forn les t Da ta with the CE-1 Correlation e

t

VI. REFEREtlCES

. 1. L. S. Tong, G. F. Hewitt, "Overall Viewpoint of Flow Boiling CHF Mechanisms",

ASME paper 72-HT-54,

2. E. Janssen, "Two-Phase Flow and Heat Transfer in Multired Geometries",

Final Report, GEAP-10347, March 1971.

3. E. Janssen, F. A. Schraub, et al, " Sixteen Rod Heat Flux Investigation, Steam-Water at 600 to 1250 psia", presented at the 1969 Winter Annual ASME Meeting, Los Angeles in session entitled "Two Phase Flow and Heat Transfer in Rod Bundles".

4. B. S. Shiralkar, R. T. Lahey, Jr. , "The Effect of Obstacles on a Liquid .

Film", Trans. ASME, flovember,1973, pp. 528-533; ASME paper flo. 72-HT-31.

5. R. T. Lahey, Jr., and F. J. Moody, The Thermal Hydraulics of Boiling Water Reactors _, AUS, 1977, pp. 97-99.

6. ibid., p. 92.

7. S. A. Chang, R. A. Ocan, "Effect of Flow Blockage on Core Power Capability",

ASME paper 69-WA/NE-21.

8. L.S. Tong, "A Phenomenological Study of Critical Heat Flux", ASME paper 75-HT-68.

9. G. J. Kidd, Jr. , H. W. Hoffman, W. J. Stelznan, "The Temperature Structure and Heat Transfer Characteristics of an Electrically Heated Model of a Seven-Rod Cluster Fuel Element", presented at the 1968 A5ME Winter Annual Meeting, New York.

10 S. C. Yao, L. E. Hochreiter, U. J. Leech, " Heat Trans fer Auonentation in Rod Bundles Hear Spacer Grids", presented at the 1980 ASNE Winter l

Annual Meeting, Chicago.

l l

11. D. C. Groenveld, W. W. Yousef, " Spacing Devices for fluclear Fuel Bundles:

l A Survey of Their Effect on CHF, Post-CHF Heat Transfer and Pressure Drop",

l Proceedings of the ANS/ASME/flRC International Topical Meeting on Nuclear Reactor Thermal Hydraulics, Saratoga Springs, New York, October 5-8, 1980, pp. 1111-1125.

I 12. "CE-1 Applicabili ty to San Onofre Units 2 & 3 HID-2 Grids", ffarch,1981, Cell-155(5)-P, response to question 1; dockets 50-361 and 50-362.

l

13. B. S. Shiralkar, "Two-Phase Flow and Heat Transfer in Multirod Geometries:

A Study of the Liquid Film in Adiabatic Air-Water Flow wi th and without <

Obs tacles", GEAP-10248, October,1970.

.e

s'.

P

14, L. S. Tong, " Boiling Crisis and Critical Heat Flux", AEC Critical Review Series, August 1972; Library of Cungress Catalog. Card Number 72-600190.

15. "C-E Cri tical Heat Flux: Critical Heat Flux Correlation for C-E Fuel Assemblies wi th Standard Spacer Grids, Part 1, Uniform Axial Power Dis tribution", CENPD-162-P-A, September,1976.

16. "C-E Critical Heat Flux: Critical Heat Flux Correlation for C-E Fuel Assemblies wi th Standard Spacer Grids, Part 2, Nonuniform Axial Power Distributions", CENPD-207-P, June, 1976.

17. " Statistical Methods in Research and Production", 0. L. Davies, P. L.

Goldsmith, 4th Edi tion Longmans Group Ltd. , London,1976. ,

18. " Joint Meeting of the Subcommittee on Reactor Fuels and the Subcommittee on Emergency Core Cooling", Transcript ACRS Meeting, May 5,1981,

19. " Evaluation of DNB Test Repeatability", K. W. Hill, F. E. Motley, F. F. Cadek, WCAP-8201-A, April 18, 1975.

1 i

_ -.