ML18051A922

Safety Evaluation Supporting Justification of Xnb Critical Heat Flux Correlation

ML18051A922
Person / Time
Site:	Palisades
Issue date:	06/07/1984
From:	Office of Nuclear Reactor Regulation
To:
Shared Package
ML18051A920	List: ML18051A919 ML18051A921 ML18051A922
References
NUDOCS 8406130115
Download: ML18051A922 (27)

Text

UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 (Enclosure 1)

SAFETY EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION JUSTIFICATION OF XNB CORRELATION FOR PALISADES

1.0 INTRODUCTION AND BACKGROUND

The Exxon Nuclear Company's (ENC) XNB critical heat flux (CHF) correlation as described in XN-NF-621(P) Revision 1 (Ref. 1), has been reviewed previously by the staff (Ref. 2) with technical assistance from Idaho Nuclear Engineering Laboratory.

As a result, the staff had concluded that XNB with a minimum departuFe from nucleate boiling ratio (DNBR) of 1.17 is acceptable for licensing calculations when it is used with the XCOBRA~IIIC code and is applied within it~ applicability range.

The CHF test data provided in XN-NF-62l(P), Revision 1, as a basis for the development of XNB, consist of test sections representative of fuel as~emblies designed by various fuel vendors such as Exxon Nuclear, Westinghouse and Combustion Engineering.

Since this data base does_not include explicitly a rod bundle prototypical of the Palisades fuel design, the application of XNB to the Palisades reload could be outside of the *XNB applicability range.

By letter dated November 1, 1983 (Ref. 3), Consumers Power Company submitted an ENC report, XN-NF-709, "Justification of XNB Correlation for Palisades" to justify the use of XNB for Palisades fuel. In XN-NF-709, an additional CHF test section, ENC-204, is provided.* This test section is representative of the Palisades fuel design.

The staff evaluation of the applicability of.lNB to the Palisades fuel is addressed in the following section:

~

2.0 EVALUATION The CHF test data presented in XN-NF-621(P), Revision 1, consist of test sections representative of various fuel designs and a variety of axial power distributions expected for power operation.

The ENC-204 test data presented in XN-NF-709 for Justification of the application of XNB to Palisades fuel consist of uniform axial power shape data only. Therefore, rather than using the ENC-204 data alone to derive the DNBR limit, it must be shown that the ENC-204 test data belongs to the same population of the data which were used in the XNB development.

In addition, it must be shown that the DNBR limit of 1.17

• is correct or conservative limit relative to the ENC-204 data base~.

The CHF data reduction for the ENC-204 test section is performed using XCOBRA-IIIC thermal-hydraulic code for the determination of subchannel fluid conditions.

The same computer code and method were used previously in the data reduction for XN-NF-62l(P), Revision 1, and had been found acceptable.

Therefore, this safety evaluation will be concentrated on the statistical analysis of the CHF data.

In the treatment of CHF test data; the statist~cal method used by ENC was to evaluate the predicted-to-measured (P/M) ratios of CHF data.

This is a 94tJ6/~/l//S p

• deviatton from general practice of using the measured-to-predicted (M/P) CHF ratios~"" However, the previous staff review (Ref. 2) had determined that this statistical characterization of the CHF data is acceptable.

The same method is used in the treatment of the ENC-204 data and is also acceptable.

In the staff evaluation of XN-NF-621(P), Revision 1, a One-Way analysis of variance was p~rformed by the staff technical consultant at INEL on the ungrouped test data to test the equality of means of the P/M ratios.

The results of this analysis revealed three separate populations among the test data presented in XN-NF-621(P), Revision 1.

The DNBR limit of 1.17 was the highest limit obtained among the three populations and could be a conservative limit when compared to the one-sided tolerance limit derived from a particular population.

For example, the one-way analysis of variance had determined that the test sections ENC-3, ENC-4, ROSAL-4 and

~.JH*+~;:l64 were of the same population.

The DNBR limit derived from this population is 1.15 with a 95 percent probability at 95 percent confidence level of avoidin~ DNB.

These test sections consist of axial power distributions ranging from uniform, cosine U and U sine U shapes, and ro.d diameters ranging from 0.374 to 0.422 inches.

In response (Ref. 4) to a staff question on whether the ENC-204 data belongs to the"-population of data presented in XN-NF-621(P), Revision 1, the licensee performed a one-way analysis* of variance for the test sections ENC-3, ENC-4, ROSAL-4 and WH-164, which had been determined previously to be of the same*

population, and the test section ENC-204.

The result showed a F-statistic of 2.50 for all 5 test sections compared to an F-statistic of 3.02 for the 4 test sections including ENC-204.

Therefore, it is reasonable to assume that the ENC-204 test section also belongs to the same population of the other four sections.

The staff has also performed an independent calculation by~

combining the means and standard deviations, respectively, of the four test sections to form a group mean and standard deviation of.the combined data.

An F-test and a t-test are performed to determine the equality of variances and equality of means between the ENC-204 data and the combined data from the four test sections. *The results show an F-statistic of 1.024 and a t-statistic of 0.964.

Therefore, both null hypotheses of equal variances and equal means can not be rejected at a 5 percent significance level.

We, therefore, conclude that ENC-204 test section belongs to the same population and the ENC-204 data~can be incorporated with the other four sections.

The new combined data mean of P/M ratios and standard deviation are 0.95646 and. 0.10188, respectively, for a total 257 data points.

The one-sided tolerance DNBR limit derived from these data would be 1.141 with a 95 percent probability at 95 percent confidence level of not experiencing DNB.

The 95/95 DNBR limit derived frbm -

the ENC-204 data alone is 1.169. Therefore, use of DNBR limit of 1.17 is conservative.

3. 0 CONCLUSION The NRC staff has reviewed XN-NF-709.

Based on this review, the staff concludes that the XNB correlation is acceptable for application to the Palisades fuel with minimum DNBR limit of 1.17. This acceptability is subject to other restrictions imposed in the staff safety evaluation report (Ref. 2, copy attached) on XN-NF-62l(P), Revision 1..

REFERENCES

1.

R. B. Macduff, "Exxon Nuclear DNB Correlation for PWR Fuel Designs,"

XN-NF-621(P), Rev. 1, Exxon Nuclear Company, April 1982.

2.

Letter from C~ 0. Thomas (NRC) to Dr. Richard B. Stout (Exxon Nuclear Company), "Acceptance for Referencing of Licensing Topical Report XN-NF-621(P), Revision 1, Exxon Nuclear DNB Correlation for PWR Fuel Designs," April 12, 1983

~

3.

Letter from B. D. Johnson (Consumer Power) to D. M. Crutchfield (NRC),

"Docket No.

50-255 - License DPR Palisades Plant - XN-NF-709,

'Justification of XNB Correlation for Palisades,' May 1983," November 1, 1983.

4.

Letter from B. D. Johnson (Consumers Power Company) to D. M. Crutchfield, May 11, 1984.

• 1 UNITED STATES

/

I

/t.,:._L./)-*

Dr. Richard B. Stout, Exxon Nuclear Company 2101 Horn Rapids Road P. 0. *Box 130 Richland, Washington

Dear Dr:

_Stout:

CLEAR REGULATORY COMMISSION

_,./.

. WASHINGTON, 0. C. 20555 APR 12 B33 Manager 99352

Subject:

• . Acceptance for Referencing of Licensing Topical Report XN-NF-621 (P), Revision 1, "Exxon Nuclear DNB Correlation for PWR Fuel Designs" We have completed ou'r review of the subject topical.report submitted May 5, 1982 by Exxon Nuclear Company (ENC) letter GF0:034:82.

We find this report is acceptable for referencing in license applications for LWR Plants to the extent specified and under the limitations delineated\\in the report and the associated (NRC) *evaluation which is enclosed. 'The evaluation defines the basis for acceptance of the report.

We do not intend to repeat our review of the matters described in the report and found acce.ptabl e when the report appears as a reference in 1 icen*se appl i cations except to assure that the material presented is applicable to the specific plant involved.

Our acceptance applies only to the.matters described in the report.

In accordance with established procedures (NUREG-0390), it is requested that ENC publish accepted versions of this report, proprietary and non-propri etary; within three months of receipt of this letter. The accepted versions.should incorporate this letter and the enclosed evaluation between the title page and the abstract.

The accepted versions shall include an -A {designating accepted) following the report identification symbo 1.

( '

~r. Richard APR 12 'B83 Should our criteria or regulations change such that our conclusions as to the acceptabiljty of the report are invalidated, ENC and/or the* applicants ref erer:ici ng the topi ca 1 report wi 11 be expected to revise and resubmit their respective documentation, or submit justi fi ca ti on for the continued effective app 1 i cabi1 i ty of the topi ca 1 report without revi.s ion of their*

respective documentation.

Enclosure:

As stateti Si nee rely,

~a.

Cecil O. Thomas, Chief Standardization & Special Projects Branch DiYision of Licenstng

r-:'.*.'.*

y 1

INTRODUCTION In XN-NF-621, R~vision 1, Exxon Nuclear Company (ENC) presented the XNB critical

• heat f!µx (CHF) correlation which will be used to assess the thermal margin of pressurized water reactors (PWRs).

The XNB is an empi:ica.1 relationship which specifi.~ CHF (i.e.., the heat flux-at-which departure from nucie-ate boiling, D~~,. occurs) as*a fu~~ti6ri of local coolant ~onditions and fuel assembly geometry.

It is based on 14 test series with a total of 714 data po.ints and three different PWR fue! vendor designs.

The 14 test series include variations in grid design, heated length, grid span, rod diameter, and axial and radial power distributions.~*

n.e local coolant conditions in the rod bundle were calculated using the XCOBRA-IIICJ computer code which is described in XN-NF-:75-2l(P) and the,range of cgolant conditions tested were typical of an op~rating PWR.

Bri.sed on the XNB' s a~il i ty to predict the test data*, Exxon has proposed a

• departure from nucleate boiling ratio (ONBR) limit of l.17 for the correlation.

This limit corresponds to a 95% probabi 1 ity of not experiencing ONB at *a ~95%

confjdence level.

The comparable value for. the W-3 correlation, which is presently us*ed by ENC,.is 1..30.

Exxon SER 1-1

~*

~*-*

. 2 DESCRIPTION OF CORRELATION The basic fonn of the XNB correlation is *as follows:

q 11

= A + B 11: HLOC uncorrected where A= f (pressure, mass velocity, inlet subcooling)

B = f (p.ressure, mass velocity, local enthalpy)

HLOC = Reduced local enthalpy

= local Enthalpy/906.00 eq. (1)

All of the parameters used in the XNB are reduced using the critical properties of "Water* (i.e., the "'ater properties at the cri ti ca 1 pressure, 3208. 2 psi) and using the above method for HLOC.

Additional factor.s are used. as part of the correlation to account for non~

uniform axial power distributions, geometry differences such as spacer pitch and mixing ~ane loss coefficients, and differences in_heated lengths. *The

• final form of the XNB is:

q" crit i ca 1 = (q 11

)11: Correction Factors uncorrected eq..(2)

T*he procedure for using the XNB is to initially calculate the heat flux u_sing

~""'

~_qua~ion (1), determine the appropriate correction factors, calculate CHF using equation (2), and determine the DNBR, which is the ratio of the actual heat flux to predicted CHF.

The ranges over which Exxon is requesting the XNB be applied (Chandler; January 6, 1983) are:

Exxon SER Pre~sure (psia) 1395 - 2425 Local Mass Velocity (Mlbin/hr-ft2) *a.92 - *3.04 Local Enthalpy (Btu/lb) 594.85 - 821.24 Local Quality Heated~ Length (inches)

Grid Spacing* (inches)

Inlet Subcooling (Btu/lb) 2-1 +0.3 66 - 168 14.3 22.0 37.2 - 336.34

)*I.

It will also be used for the f~l1owing geometries:

Vendors:.

Fuel Design:

Equivalent Hydraulic Diameter (inches)

Equivalent Heated Diameter. (in_ches) __ _

Exxon Nuclear Combustion Engineering Westinghouse Non-Mixing Vane Mixing Vane 0.177 - 0.612 0.463 - 0.528 The test series and their associated fuel rod arrays are:

Vendor Rod Array Test Series

• westinghouse/

1.4xl4, 15xl5 ENC-3,

• 4, and Exxon ROSAL-2, 4, 7, Exxon 17xl7 E.NC-6 Cumbustion 16xl6 CE-4.7, CE-59 Engineering.

Westinghouse 17xl7

\\olH-162 and 164 Exxon SER 2-2 5

and 8

• 3 STAFF EVALUATION 3.1 Scope of Review The staff review of XN-NF-621, Revision l included an independ~nt audit of the subchannel calculations performed to determine the local coolant conditions in the rod bundle..for. ~11. 714 data_ points_. __ Jhis was performed us:t-ng the COBRA-IV*

.computer c-ode which* was derived from and is an ancil 1ary of the-COBRA-II IC p.rogram.

Our review also included a statistical analysis of the calculated results and a review of the methodology used in combining the XCOBRA-IIIC code and the correlation.

During the review, requests were made for data clarifica-tion and additional or correctep information was received in seve-ral areas.

The above reviews were performed by the Idaho Nationa 1 Engineering Laboratory (INEL) unde~ the direction of a cognizant staff member.

-3.2 Results of Audit Calculations The results of the INEL audit calculations are presented in Tables l and 2.

Table l is a comparison of the local conditions at which CHF was predicte~ as

• determined by the XCOBRA-IIIC and COBRA-IV codes for a limited number of data points.

The* comparison indicates good agreement between the two codes and either could be used to establish the local conditions required for t~e develcip-ment of a CHF correlation.

Table 2 is a comparison of the mean and standard deviation for each of the data sets and the total population.

This comparison shows good agreement for the overall values but contains discrepancies in many of the individual data set~. The possible ram~fications associated with these differences are de-scribed in the statistical analysis discussion contained in this report.

EX)Con SER 3-1

• During our review, the staff requested that Exxon provide a description* of ti.~w* the local conditions for the XNB were determined including a discussion of the subchannel code used, subchannel modeling, axial nodalization,.and input assumptions.

Exxon responded that the XCOBRA-IIIC code was used to calculate the local coolant condi~ions.

XCDBRA-IIIC is a derivative of the CDBRA-IIIC code which was developed at Battelle Pacific Northwest Labora~pry.* The modifications made by Exxon to CDBRA-IIIC include minor improvements in the solution technique, the addition of calculational options, and operational modifications such as streamlining code input.

-~ -

E~xon furt-her stated "that the friction factors used were determined from pressure drop measurements performed on 'ENC test sections or estimated for geometries for which ENC does not have detailed t.est data.

These loss coefficient ~stimates are based on the experience gained fro~ measuring actual fuel bundl.es of Westinghpuse or Combustion Engineering (C-E) desig_ns.

They also.reported that sen_sitivity studies of CHF test data sho...,ed n~gli-:

.. gible influence on predicted conditions when the form loss coefficients *.were varied *by:, as* much as 15%.

.. The mixing values (~s) chosen were based on spacer. design and are dependent

.on a particular fuel type.

These values were determined experime.ntally.*for the

£NC *designed fuel while for non-E.xXon fuel a lower bounding value was us.ed for mixing vane grids.

For example, in analyzing, the Westinghouse "L" gr5d.design a lower value of 0.010, which was obtained.from.WCAP-8030-A, -was used.

Based on our review of the above information, the staff concludes that the apprciach taken by Exxon in determining the local conditions used in developing the XNB correlation are acceptable.

The XCOBRA-IIIC code is still under staff review, and any limitations resulting from this review will be addressed in our safety evaluation report-on XN-NF-7S-2l(P), Revision 2.

The INEL audit ~alculations were performed using the same friction factor correlation, two-phase flow correlation, crossflow resistance, momentum turbulent mixing factor, pitch to length parameter, inlet enthalpy ~nd inlet mass velocity as Exxon.

Exxon.SER

Our review also included an an~lysis of the correction factors used in the XNB development and the determ:ination of these factors in actual reactor application.

. B*a*s*ed** on this review, we have concluded that the method used to calculate

• these parameters and their values used in determining the*DNBR limit are 9cceptable.

However., it is the opinion of the staff and our consultant that a change in these parameters, *such as determining their values using a prototype and then a fu11 scale bundle, may increase the uncertainty in both the code's pr~diction.

_of 1oca1 *cool ant co Adi ti ons and-the ccrrre*1 at ions prediction of-CHF.

This may. :

• • si;gnifican-tly ~,.ter. the st~tistical analyses -on which the ONBR limit is based.

Therefore, we conclude that the values of these parameters used in the develop-ment of the XNB must be used in licensing analyses.

For the uni form heat flux tests., ENC used the.end of the heated

~ ength as.* the

• CHF *location whi}e the experiments*shbwed

upstream of \\the end of the heated length.

that for the same tests, CHF.' Jccurred When asked to justify *using this

technique iri determining the DNBR Exxon responded that the worst 1 oca l condi-tions calculated for a bundle having a uniform axia-1 power distribution (APD)-

~are at the end of the heated length.

In order to maintain a consistent path

"'bet'.Ween test_ anal~sis and reactor design and based on the fact that the DNBR location in a reactor is determined by the code and is not k.nown aprior~, the procedures used to determine the ONBR for those tests where* burnout occllr':"ed upstr:eam of the heated length is acceptable.

We have reviewed the additional inf ormat i ori pr*o~i ded. by ENC :and have* cone 1 uded that the method used by Exxon in determining DNBR is acceptable since the DNBR limit is dependent on the ability of the subchannel code to predict local conditions which produce CHF.

An additional.area of concern raised by the staff on the uniform heat flux tests w2~ why CHF occurred *at the thermocouple upstream of the end of the heated length rather than at the end of the heated length where the highest quality region should occur.

Exxon stated that burnout is a function of.the Exxon SER 3-3

locatio~ of the spacer,gri~ and that the grids will improve heat transfer for a distance of 20 or more rod diameters downstream of the spacer.

Because the spacer was located slightly downstream of the end of the heated length, heat transfer above the spacer would improve while the local hydraulic conditions.

downstream of the grid would be *more severe.

Therefore, for the experimental data in question, the effects of the spacer grid dominated the occurrence of CHF ev~n though a higher quality may occur at the end of the te~t bundle.

The staff has revie.wed.this information and concludes that ENC has acceptably addressed our concerns on this issue.

Finally in the area of test procedures, the staff requested that Exxon provide a d1scussion on how the rate of ~ower w~s increased, what post-test *inspections were performed, and what, *if any, duplicate runs were made to establish continued integrity of the test bundle.

In response to this toncern, ENC stated that the power was manually*raised in the CHF.tests by an increment of less than 1%

and held ~onstant until conditions be~ame stable.

This process was rep~ated

,.,until CHF occurred.

They further stated that dup 1 i cate runs were made to

• establish c6ntinued integrity.

As an example, they c1~ted the ENC-6 tests,

• • Where replicate points were taken during the test and one in between poi*nt was

. *taken at the end of the test to confirm continuity and consistency of the test

.:data from beginning to end.

At the end of the tests, post-test i nspection*s

• • .were performed and, for example, on the ENC-6 bundle there were no visible

• signs of hot spots on the rods.

Based on our review of this infonnation,:-*the staff has concluded that the CHF tests were performed in an acceptable manner; Our review of the statistical characterization of the XNB results dealt mainly with the method used by Exxon to statistically analyze the data and a review of the analyses.

The statjstical method used by ENC was to evaluate the predicted-to-measured (PIM) ratio of CHF data.

Since in previously approved correlations, the measured-to-predicted (M/P) ratio was used to determine the 95/95 limit, Exxon was as~ed t.o justify their technique.

ENC responded that the procedure used.in detennining*the 95/95 limit assumed a normal..

distribution.

Transforming the data from P/M to M/P yields two distribu-tions for comparison, both of which may be normal or both may depart from normality.

As a verification on the 95/95 limit for the P/M data,. Exxon Exxon SER 3-4

performed a distribution free estimate of the limit and determined the value to be 1.177.

For the reverse ratio, and using their original statistical

'.approach; Exxon calculated that 95/95 limit for the WP data, when a normal distribution is assumed, is 1.191.

ENC further stated that the non-parametric estimate of the 95/95 limit, 1.177, does nQt make complete use of the actual distribution, and therefore this limit will bound the 95/95 limit obtained from the actual distribution.

By considering the first four moments of the P/M data ENC found that the actual

.-d1:stributi.Pn is-a g~a distribution. - On-the other hand, the.use of the.M/P data is overly conservative sine~, the actual value of the 95/95 limit for the P/M data, when the appropriate distribution is used, lies at some value below the non-parametric limit of 1.177.

ENC also stated that the DNBR reported for licensing analyses is defined as P/M*ratio.

Based on our review.of the above information, the staff has concluded that the analysis of the P/M data is*

. acceptab 1 e.

'1 J

As. part of the review, the staff requested that Exxon -demonstrate that each of the samples, e.g., test series, belong to a single population.

ENC responded-

.'.by initially performing a Bartlett test for homogeneity of variance (Chandler;

August 26, 1982).: The breakdown was based on both vendor design and fuel

• assembly geometries.

The results of this test showed that the variances do differ among geometry types.

Exxon also performed a K-sarriple Squared Ranks test of variance using the above groupings (Chandler; August 26, 1982).

Results for the population of 6 samples and 5_ degrees of freedom indicated that at 1 east two of the variances were unequal.

By removing the ROSAL, ENC-1, and 2 data, Exxon found that there exists a significance level between 2.5% and 5.0% that the remaining data were from thi same population.* Finally, ENC removed the ENC-3, 4, and 5 data and analyzed the remaining population.

Based on the results of the third analysis, Exx6n concluded that the data comprised of 3 samples and 2 degrees of freedom were likely identical.

Exxon SER.

3-5

An analysis of the means ~n~ a*comparison of variance analysis showed that for an equivalent sample size of 83,7 with 378.7 degrees of freedom the mean is 0.98502 with a standard deviation of 0.09847.

Based on this mean and standard d~viation the 95/95 DNBR limit would be 1.168.

The final analysis performed by ENC was the determination of a DNBR limit exclud~~g that data which had the greatest possibility of being from a different population.

For all sections less the ROSAL and ENC 1 thru 5 data the DNBR limit was 1.169 while for all sections less* the ENC-6, WH-162, WH-164, CE-47,

~nc149 *da.ta, th.e DNBR limit w-as-1.176.- ---

The results of the above tests lead ENC to conclude that the data could be tT"'eated. as a single population and that the 1.17 DNBR limit would cover any deviation within the data sets.

.In order to ascertain the validity of these conclusions, INEL performed a*

.,series of F-,tests to i cient i fy any systematic vari at*i on among the test series.

I I

The tests were performed at a 99% confidence level.

Based on the F-test, INEL concluded that there was a variance among tests of different geometries.

4~dditiona11y, INEL performed a one-way analysis of variance using the ungrouped

• test series."

For the one-way analysis, INEL used the groupings reported by ENC and calculated a F-r~tio of 24.03 for six samples with fiv~ and. 708 de~ree~ of freedom for the numerator and d~nominato~.* This result shows that there is a variance among the tests when they are grouped by geometry type.

Removing data sets WH-1~2, WH-164, ENC-3, 4, and 5 resulted in an F-ratio of 2.40 with three and 392 degPees of freedom for the numerator and denominator.

This indicates that the remaining data have a probability o.f between 5% and 1~ of being in the same population.

A second one-way analysis of varfance was performed on the ungrouped data:*

The results of fhis test are presen~ed in Table 3 and indicate that ENC-1, ENC-2, ENC-6, ROS~L-2, ROSAL-7, ROSAL-8, 'WH-162, CE-47 and CE-49 are probably of the same population while test series ENC-3, ENC-4, ROSAL-4, and WH-164 are Exxon SER 3-6

of a second population.

ENC-5 i~ a untque test series and does not fall into either popul~tion.. Using the above ~opulation~, a DNBR limit of 1.21 for the

.ENC-1, ENC-2, etc. population was determined while the ENC-3, ENC-4, etc.

~opulation has a 95/95 limit of 1.133.

Figure 1 is a histogram of the total_ data set and it shows that the *overall populatJon is approximately normally distributed.

Histograms for the individual

.samples (EGG-HTAP-6167) show that ENC-3, ENC-4, ENC-5, ROSAL-4 and WH-164 are skewed t6 the left of the population mean.

-F~rtryer analyses-we.re-pe~f ormed to determine if there was a reason for the grou~ings bbtained from the one-way analysis of variance.

A number*of groupings were examined using different bases such as rod diameter, grid spacing, radial power distribution, ~xial power distribution, KLOSS; and an unhe~ted guide tube in the bundle. <These stud_ies showed no ~niqueness in either grouping..

A second evaluation revealed that the modeling of the guide tube ~as an influence in determi ni:ng the above grouping.

For those bundles -containing an unheated

~u~de tube, CHF experimentally occurred in a channel that contained the ~uide-

'tube; however, in predicting CHF, Exxon often reported burnout in a*channel other than the one with the.guide tube.

Sin~e the guide tube is an unheat*d

• .. wall; CHF occurs at less severe local i:onditions and has a lower value. lf CHF is predicted in a typi ca 1 channel, four heated rods, when it actua 11y'-

occu~red in a guide tube channel, this would be nonconservative.

The reason for this is that the predicted local conditions are greater than the conditions which experimentally produced CHF; therefore, the analytical results show that you can go to a higher power than you actually achieved.

Table 4 presents a summary of the test series that have one or more unheated guide tubes.

For all of the series reported in Table 4 ENC predicted CHF in I

the COBRA hot channel rather than the experimental channels listed in the*

tab1e.

This indicates that the reason ENC-3, ENC-4, and ENC-5 do not belong to the populati6n may be the difference in the channel for the predicted and measured CHF.

Test series ENC-6 do.es not fall from the population because the Exxon SER 3-7

difference between-the-COBRA-IV experimental hot channel and t-he guide*tube channel is only 3.0% and tfie sample mean is closer to *the expected mean of

l. 0.

In addition to the above analyses, the INEL audit calculations revealed that the ENC-1, 2, 3, 4, 6, CE-59, and ROSAL-8 tes~ series were biased..with inlet pressure.

For pressures less than 1800 psia the correlation predictions tend to be scattered about som~ value less than 1.0 while for data above 1800 psia the data is randomly scattered about 1.0.. This indicated that the correlation

.under pred_i cts.**cHF. for:* the 1 ower pressures-bu~ is reasonably accurate for p~esiures above 1800 psia:

Based on-this review, the staff has concluded that altho~gh these test series statistically belong to one of the two populations, excluding the ENC-5 population, the fact that they are biased with pressure may preclude them-from being placed in either population.

.. :Also, the staff statistically analyzed the six different geometry types reported I

.by Exxon.

Table 5 contains the results of our analysis based on a geometric i:..::.

characterization.

These results show that for the ENC-1 and -2 populati:on the

~mean, standard deviation, and 95/95 limit are much greater than the mean,

• standard deviation_, and 95/95-limit of the remaining populations when they are_

• compared to the sam~ parameters of the. total population.

Based* on our review of the ENC statistical ~nalyses, ou~ consultant's analyses, and the result of th~ staff'~ _statistical analyses, we requested additional information from Exxon which justified treating the 14 samples as one population.

In response to our concerns, Exxon provided plots of DNBR versus inlet pressure for those test series that the staff felt were biased with pressure (Chandler; December 16, 1982).

Based on their own pressure plots ENC concluded that there was no significant systematic.trend~ with pressure.

We have reviewed the.information submitted in the.December 16, 1982 letter and have concluded that there is a small trend with pressure; however, the trend is random.in nature and does not exhibit any sysfematic characteristics.

Therefore, the staff concludes that the ENC-1, 2, 3, 4, 6, CE-59, and ROSAL-8 test series Exxon SER 3-8

-4

• need not be treated as a single pop~lation due to the trends in pressure, since these trends are not systematic.

With respect to the statistical analyses, Exxon requested that the data be reviewed as two separate populations (Chandler; December 22, 1982).

One of the popul~tions would be iomprised of the test series representing 16xl6 and

.17xl7 irrays (CE-47, CE-59, WH-164, WH-162, and ENC-6).while the s~cond popula-tion would represent the 15xl5 bundles.

As justification for requesting* this breakup,.ENC p~ovia~d ~he range of te~t-~o~ditions and axial power distri-butions fo*und in each *population,*

A review of the 16xl6 and 17xl7 data base showed that only a chopped cosine and uniform axial power distribution (APO) were present. lt is the position of the staff that* al~ possible power distributions expected throughout an operating_cycle be used in the development of any CHF correlation.

Since the 16xl6 and 17xl7 do not include either an upsk~w or downskew APO, Exxon cannot remove thos~ test series, e.g. the 15xl5 array, that ~ave the upskew APOs.

. Therefore, the 15xl5 test series must remain in the data base until ENC pro-vides additional data for the 16xl6 and 17xl7 test series which contain an

. up,skew and/or downskew APO.

ln a modified response (Chandler; January 3, 1983) Exxon requested that t~st series ENC-1 and ENC-2 be removed from the data base.

The reason for elimin-ating this data-was.that ENC-1 contained minimu~ grids that were not repre-sentative of any grid being manufactured by ENC, Westinghouse or CE while ENC-2 had a uniform axial and radial power distribution that was atypical of actual reactor conditions.

ENC further stated that a statistical analysis of the data was performed using the populations reported by INEL.

The results of these e~aluations showed that the worst 95/95 limit was 1.17 for the population containing the CE-47, -59, \\ilH-162, ENC-2, ROSAL-2, -7, and -8 test series.

Based on these results, we have concluded that the proposed grouping of data which results in a DNBR limit value of 1.17 is acceptable..

Exxon SER.

3-9

'~.. -.. :.

4*

CONCLUSION The staff has rev1ewed XN-NF-621, Revision l.and the additional ~upporting

. information submitted by Exxon Nuclear Company.

Based on this review, we have conclua~d that XNB ~orrelati~n is acceptable for use in reactor licensing applications.

We have also concluded tha~ the 95/95 DNBR limit of 1.17 reported -

.b~_Exxon*i_s actept.abl:. *T~ese conclus-ion*s-are based on the foiJowing:

(1)

The subchannel code used, XCOBRA-IIIC, is acceptable for predicting local coolant conditions used in the development of a CHF correlation.

This is based on a comparison of XCOBRA-IIIC with the staff's audit code

~

COBRA-IV.

Since the XCOBRA-IIIC is still under staff review, any limi.ta-

• . tions resulting from its use will be addressed in our safety evaluation report:on the code.

I (2)

An independent audit, performed by our consultant INEL, using a different subchannel code yielded similar results.

• .(3)

• Th,e DBNR data has been statistically characterized in an acceptable.

manner.

(4) The 95/~~ limit is based on three separate populations that were recom-mended by our consultant; therefore, the 95/95 limit of one population will be conservative when compared to the limit of a population containing all of the test data.

We will require that the correction factors used in analyzing the CHF test data and the mixing factors used i~ the data reduction be used in reactor design applications, since a _c~ange in'these factors may alter the code and correlation uncertainties associated with the prediction of CHF.

This in turn may raise or lower the 95/95 DNBR limit.

Therefore, if any of these parameters are changed, ENC must provide a description of the c_hange an_d Exxon SER

• 4-1

s*i.Jfficient justificatio_n which warrants making this change.

Additiona*ny, Exxon should provide the test data which justifies using the XNB on fuel

'designs not contained in the data base or acceptable justification on why the XNB is applicable to this fuel type.

For example, Exxon manufactured fuel for CE reactors is not present in the data base.

ENC must provide additional test data for these fuel bundles or a quantified justification of the XNB's appli-cabili~y to this bundle type.

Finally,.it sho.uld. ~e ~oted.~ha~ the D_NBR.Jimit does not include.any adjustment*

which is required when a mixed core, e.g. a core with geometr1cally different fuel types, is analyzed.

Exxon SER.

4-2

5 REGULATORY POSITION The staff concludes that the XNB CHF correla~ion as described in: XN-NF-621, Revision 1 is acceptable for use in licensing application when it is used with th~ XCDBRA-IIIC code and within the range of application reported in Section 2.2 of this safety evaluation report.

We also conclude that the

• 95;95 limi-t. of*l.17. associated with the XNB is acceptable.* Use-of the correlation should be within the limitations described in the previous section.

Based on our review,.the staff finds XN-tff-621, Revision 1 an acceptable.and referential report with the restrictions noted in the above paragraph.

Exxon SER.

5-1

~.

. i '.

Table 1!

Comparison of Local Conditions Enthalpy Quality Void Fraction,.

Mass Flux Case..

XCOBRA-III COBRA-IV XCOBRA-:-I_ll COBRA-IV XCOBRA-II IC COBRA-IV

• XCOBRA-11 IC COORA-.IV ENC-3-63 656.57 656.49 0.077 0.077 0.610 0.594 1.9046

1. 9434

~

ENC-4-28 703.28 705.78 0.167 0.167

0. 709.

0.712

1. 4897

1. 5210 e

ENC-6-42 616.44 620.00 0.00 0.007 0.318 0.350 2.8655

2. 8990'.

ROSAL-2-18 612.57 627.50 0.001 0.027 0.550 0.554 L8674

1. 8809 ROSAL-2-9 622.44 636.52

0. 018. *.

0.043 0.561 0.566

1. 9409

1. 9601

. 1.*

. l'.

Table 2: -

Test*

Number of Section Data Points CE-47 95 CE-59 89 WH-164..

53 WH-162 53 ROSAL-2 28 ROSAL-4 26 ROSAL-7 11 ROSAL-8 32 ENC-1

. 28 ENC-2 24 ENC-3 73 ENC-4 80 ENC-5 59 ENC-6 62 Total Population 714 Comparison of Mean and Standard Mean (Meas./Pred)

XCOBRA-IIIC, COBRA-IV 1.028

1. Q.300 1.023

1. 0'500 0.950 0.9727 0.992 1.0032 0.976 0.9995

-o. 933

.:.0.-9689

-. 0.970

1. 0383 1.001

1. 0586

1. 040

1. 0504 0.993 1.0119 0.994 0.9458 0.985

0. 9712 0.911 0~8956 0.995 1.0071 0.985 0.99614 2

Deviation Standard Deviation XCOBRA-III COBRA-IV 0.0741 0.0804

. 0. 0820 0.1020

0. 0677.

0.0682 0.0845 0.0736 0.118 0.0990

0. 0843*

. 0. 0832 0.104-3 0.1210 0.0987 0.1070 0.1212 0.1220 0.10$3 0.1090 0..1029 0.0923 0.1196 0.112 0.0848.

0.08.11 0.0749 0.0868 0.09847 0.1030

"\\.

Tabl.e 3:. One Way Analysis of Variance Test Series Grouping ENC-1, -2, -6 ROSAL-2, -7, -8 WH-162:* CE-47, -59 ENC-1,. ~ 2, -4,._-6

• _

.ROSAL -2, -.:.4, -7, -E -.

WH-*lfr2, -164 CE-47, CE-59 ENC-1, -2, -3, -4, -6 ROSAL-2, -4, -7, -8 WH-162, -:t.64

.CE-47, -59

)

'ENC-3, -4, -5

.ROSAL-4, WH-164 ENC-3, -4 ROSAL-.4, WH-164 F-Ratio 2.47 5.57 7.84 7.39 1.23 3

Probability of Being in Same Population 1 - 2. 5%.

>10%

y

Table 4:

Comparison of Test Series With Unheated Guide Tubes Number of Experimental CHF,Predictions COBRA-IV Channel Test COBRA-IV Other Than Series Hot Channe1 1 Hot Channel

~ Exp~anation WH-162 All ENC-6 20 ENC-3

.18 ENC*:4 30

.ENC-5 CE~47

• . 82 CE-59

.85 53

-o-42 S3

. 50 14 4

iENC predicts all CHFs in this channel.

4 As expected.

The 42 channels are 3% cooler than*

the hot channel.

Five of the indications occur in a channel with 5% less power, 21 in a channel with 0.4% less power and the remaining in a channel.with 23%

1 ess power.

Seven of the 50 indications were.

in a channel with 0.20% less* power while the remaining 43 were in a channel with 22% less power.

Twenty-five of the 53 indications*

occur in a channel with 0. 9%-.less power while the remaining 28 are in a channel with 22% less power.

The 14 indications occur in a channel with 0.3% less power.**

The:4 indications occur in a channel with 0.1% less power.

' c'

,i. __,;._..

)

Tables:.

Geometry Grouping CE-47, CE-59 WH-162, 'WH-164 ENC-6 ROSAL-2, 4, 7, 8

.*EHC-1, ENV-2 ENC-3, ENC-4, ENC-5 Total Population Comparison o.f 95/95 Limit Based.on Geometry Standard Mean Deviation 95/95 Limit 1.0256

0. 0778 1.169 0.9710 0.0791

• l.123

.0.995 0.0749 1.146 0.9720 0.1021 1.169

.l. 0183

- 0.1173

-l. 259 0.9503 0.0865 1.109 0.985 0.0985 1.163 5

6 REFERENCES 6.1 Topical Reoorts XN-NF-621, Revision l "Exxon Nuclear DNB Correlation for PWR Fuel Designs,'.' Exxo~ Nuclear Company, Apri ~ 19.8~.

XN-NF::-75-21(?), Revision 2, 11 XCOBRA-IIIC:

A Computer Code to Determine the Distribution of Coolant During Steady-State and Transient Core Operation," Exxon Nuclear Company, September 1982.

WCAP-8030-A, "Application of Modified Spacer Factor to L Grid Typical

.and Cold Cell DNB, 11 Westinghous.e Electric Corporation, January 1975.

)

'o~ 2 Other References

Ambrosek, R. G., C-C Tsai, and R. D. Wadkins, "Exxon Nuclear DNB Correlation

Review, 11 EGG-NTAP-.6167, EG&G Idaho,. !NC., February 1983.

J.C. Chandler (ENC) to H. Bernard (NRC),

Subject:

"XN-NF~62l(P), 'Exxon Nuclear DNB Corr*elation for.PWR Fuel Designs,' "Revision l," August 26, 1982.

J.C... Chandler (ENC) to J.J. Holonich (NRC),

Subject:

"XN-NF-621, Revision 1, 'Exxon Nuclear Company DNB Correlation for PWR Fuel Designs,'

Apri 1 1982," December 9, 1982.

• J.C. Chandler (ENC) to L.E. Phillips (~RC),

Subject:

"XN-NF-621, Revision l,

'Exxon Nuclear.company DNB Correlation for PWR Fuel Designs,' April 1982, 11 December 16, 1982.

Exxon SER 6-1

J J... C. Chandler (ENC) to L.E. Phillips (NRC),

Subject:

11 XN-NF-621, Revision 1, 1 Exxon Nuclear Company DNB Correlation for P'WR Fuel Designs,* April 1982, 11

• December 23, 1982.

J.C. Chandler (ENC) to L.E. Phillips (NRC),

Subject:

11 XN-NF-621, Revision 1, 1 Exxon Nuclear Company DNB Correlation for PWR Fuel Designs, 1 April 1982 1 11 Januar~. 3, 1983.

J.C. Chandler.~ENC) to L.E. Phillips ~NRC),

Subject:

11 XN-NF-621.(P),

• Revision 1, 1 Exxon Nuclear DNB Correlati-On for PW'R Fuel Designs-,' April 1982, 11 January 6, 1983.

• Exxon SER 6-~