ML051870070
| ML051870070 | |
| Person / Time | |
|---|---|
| Site: | Millstone, Surry, North Anna  |
| Issue date: | 06/30/2005 |
| From: | Grecheck E Dominion Nuclear Connecticut, Dominion Resources Services, Virginia Electric & Power Co (VEPCO) |
| To: | Document Control Desk, Office of Nuclear Reactor Regulation |
| References | |
| 05-328 | |
| Download: ML051870070 (31) | |
Text
h Dominion June 30, 2005 United States Nuclear Regulatory Commission Serial No.05-328 Attention: Document Control Desk NL&OS/ETS: RO Washington, D.C. 20555 Docket Nos. 50-280/281 50 -33813 39 50-336/423 License Nos. DPR-32/37 N PF-4/7 DPR-65/NPF-49 VIRGINIA ELECTRIC AND POWER COMPANY (DOMINION)
DOMINION NUCLEAR CONNECTICUT, INC. (DNC]
NORTH ANNA AND SURRY POWER STATIONS UNITS 1 AND 2 MILLSTONE POWER STATION UNITS 2 AND 3 REQUEST FOR ADDITIONAL INFORMATION ON TOPICAL REPORT DOM-NAF-2:
REACTOR CORE THERMAL-HYDRAULICS USING THE VIPRE-D COMPUTER CODE INCLUDING APPENDICES A AND B In letters dated September 30, 2004 and January 13, 2005 (Serial Nos.04-606 and 05-020, respectively), Dominion and DNC submitted the Topical Report DOM-NAF-2, Reactor Core Thermal-Hydraulics Using the VIPRE-D Computer Code and associated Appendices A and B, for NRC review and approval. VIPRE-01 is a core thermal-hydraulics computer code developed by EPRI, approved by the NRC, and currently in use throughout the nuclear industry. VIPRE-D is the Dominion version of VIPRE-01, which has been enhanced by the addition of several vendor specific CHF correlations.
In a May 19, 2005 letter, the NRC requested additional information to complete their review of VIPRE-D and the associated appendices.
On May 25, 2005 Dominion and NRC held a public meeting on VIPRE-D licensing issues at which Dominion discussed the proposed scope for the RAls. The NRC staff agreed that the proposed scope for the responses was acceptable. The attachment to this letter provides the detailed responses discussed in the scope at the public meeting, including appropriate references and supporting information.
If you have further questions or require additional information, please contact Mr. Thomas Shaub at (804) 273-2763.
Very truly yours,
/
Eugene S. Grecheck Vice President - Nuclear Support Services Virginia Electric and Power Company Dominion Nuclear Connecticut, Inc.
Serial No.05-328 Docket Nos. 50-280/281/338/339/336/423 Page 2 of 3 Attachment Commitments made in this letter: None cc:
U.S. Nuclear Regulatory Commission Region I I Sam Nunn Atlanta Federal Center 61 Forsyth Street, SW Suite 23T85 Atlanta, Georgia 30303 U.S. Nuclear Regulatory Commission Region I 475 Allendale Road King of Prussia, Pennsylvania 19406-141 5 Mr. J. T. Reece (w/o Att.)
NRC Senior Resident Inspector North Anna Power Station Mr. N. P. Garrett(w/o Att.)
NRC Senior Resident Inspector Surry Power Station Mr. S. M. Schneider (w/o Att.)
NRC Senior Resident Inspector Millstone Power Station Mr. V. Nerses NRC Senior Project Manager - Millstone Unit 2 U. S. Nuclear Regulatory Commission One White Flint North 1 1555 Rockville Pike Rockville, MD 20852 Mr. G. F. Wunder NRC Senior Project Manager - Millstone Unit 3 U. S. Nuclear Regulatory Commission One White Flint North 11 555 Rockville Pike Rockville, MD 20852 Mr. R. E. Martin NRC Lead Project Manager - North Anna and Surry U. S. Nuclear Regulatory Commission One White Flint North 1 1555 Rockville Pike Rockville, Maryland 20852
Serial No.05-328 Docket Nos. 50-280/281/338/339/336/423 Page 3 of 3 Mr. Stephen R. Monarque NRC Project Manager - Surry U. S. Nuclear Regulatory Commission One White Flint North 1 1555 Rockville Pike Rockville, Maryland 20852 Mr. J. Honcharik NRC Project Manager - North Anna U. S. Nuclear Regulatory Commission One White Flint North 1 1555 Rockville Pike Rockville, Maryland 20852
Serial No.05-328 Docket N 0s. 50-28 0/28 1 /338/339/336/423 COMMONWEALTH OF VIRGINIA
)
)
COUNTY OF HENRICO 1
The foregoing document was acknowledged before me, in and for the County and Commonwealth aforesaid, today by Eugene S. Grecheck who is Vice President -
Nuclear Support Services, of Virginia Electric and Power Company and Dominion Nuclear Connecticut, Inc. He has affirmed before me that he is duly authorized to execute and file the foregoing document in behalf of these companies, and that the statements in the document are true to the best of his knowledge and belief.
Acknowledged before me this 30% day of 9-
, 2005.
My Commission Expires: &
u a z 3 /, 2 o a.
m u
- h. f$&
Notary Public (SEAL)
Attach me nt (Serial No.05-328)
Request for Additional Information Topical Report DOM-NAF-2 Reactor Core Thermal-Hydraulics Using the VIPRE-D Computer Code Virginia Electric and Power Company (Dominion)
Dominion Nuclear Connecticut (DNC)
PROPOSED TOPICAL REPORT DOM-NAF-2 VIRGINIA ELECTRIC AND POWER COMPANY (DOMINION)
DOMINION NUCLEAR CONNECTICUT (DNC)
NRC Question 1 Departure from Nucleate Boiling Ratio (DNBR) is sensitive to the turbulent mixing coefficient, Please justify the value of the turbulent mixing coefficient. Stating it is conservative is insufficient. Please show the sensitivity of the DNBR to the turbulent mixing coefficient.
Dominion/DNC Response:
The turbulent mixing coefficient (ABETA) is obtained empirically by the fuel vendor.
Both AREVA and Westinghouse have conducted several subchannel mixing tests in pressurized water loops at Reynolds numbers similar to that of a PWR core under single phase and two-phase flow conditions (References 13, 14, 15) to determine ABETA. It has been determined experimentally that the value of ABETA is a function of grid spacing, and as such, it is dependent on the fuel product design. A turbulent mixing coefficient value of 0.038 has been validated by the NRC for analyzing Westinghouse 17 x 17 and 15 x 15 fuel assemblies with mixing vane grids having a spacer span of 26 inches or less (Reference 15). The Dominion submittal for the AREVA Fuel Transition (Section 4.2.5 in Reference 5) documented a turbulent mixing coefficient of 0.038 for the AREVA Advanced Mark-BW fuel. Therefore, a turbulent mixing coefficient value of 0.038 is applicable to current AREVA and Westinghouse fuel designs, and will be used by Dominion in VIPRE-D models for current fuel products. The turbulent mixing coefficient used by VIPRE-D models for future fuel designs will be provided by the fuel vendor.
As agreed during the May 25, 2005 Dominion-NRC public meeting, the sensitivity of the DNBR to the turbulent mixing coefficient has not been provided herein.
NRC Question 2 Please describe the basis for the use of the drift flux correlations employed as part of the Electric Power Research Institute (EPRI) void model. Please describe the flow regimekegimes that the EPRI drift velocity correlation is applicable to and show that this applies to the flow regimes experienced during DNB in the plant analyses. Comparisons of the drift velocity correlation/correlations to void data in rod bundles and small pipes would be desirable. This could be done using all other particular inputs and correlation choices (and code corrections) included to show the effecvability to continue to predict the test data presented in Volume 4 of the VlPRE Manual entitled l'Applications" dated 1987. Since Page 1 of 26
the EPRI void model appears to employ only the drift velocity correlation applicable to churn turbulent bubbly flow (for void fractions less than 0.3), please explain and justify why this correlation is applied to slug and annular flow where Critical Heat Flux (CHF) can occur. Define the limitations of the drift flux correlations (i.e. pressure range, flow conditions, etc.). What distribution parameter is assumed? Since more voiding may occur near the walls of the hot rods, how does the distribution parameter account for this condition in the drift flux modeling? Please explain.
Dominion/DNC Response:
VIPRE-01 has been approved by the NRC (References 10 and 11). VIPRE-D, which is based upon VIPRE-01, MOD-02.1, was developed by Dominion to fit the specific needs of Dominions nuclear plants and fuel products by adding vendor specific CHF correlations and customizing its input and output. Dominion, however, has not made any modifications to the NRC-approved constitutive models and algorithms in VIPRE-01. Therefore, none of the models for bulk void, subcooled void and two-phase friction factor present in VIPRE-01 has been modified in any way in VIPRE-D, including the EPRl models.
The EPRl models for bulk void, subcooled void and two-phase friction factor are described in detail in the VIPRE-01 documentation (Section 2.6 of Reference 16 and Section 2.7 of Reference 7), including their underlying assumptions and ranges of validity. Sections 2.5 and 3.0 of Reference 9 document the VIPRE-01 comparisons performed by the code developer. VIPRE-01 two-phase friction factor models were evaluated against experimental data from the FRIGG rod bundle test loop. VIPRE-01 void models were evaluated against experimental data from the FRlGG rod bundle test loop, ANL void test and Martin void measurements at high pressure. The comparisons show that although all the models available in VIPRE-01 match the experimental data reasonably well, the EPRl set of correlations compare more favorably with the measured data. In consequence, the EPRl models for bulk void, subcooled void and two-phase friction factor are the default selections in VIPRE-01 for PWR analysis.
Based on the evaluation of the benchmark calculations discussed above, NRC staff concluded that the EPRl models for bulk void, subcooled void and two-phase friction are acceptable for use in licensing calculations (Reference 10).
The benchmark studies mentioned above, as well as the NRC approval of the EPRl models provided Dominion a good starting point for the selection of void and two-phase friction multiplier models. In addition, Dominion performed another sensitivity study to determine the most suitable set of models for Dominion applications. This sensitivity analysis, which is summarized in Section 5.4 of DOM-NAF-2, provides justification for Dominions modeling selections, thus fulfilling condition (3) of the SER for VIPRE-01 MOD-01 (Reference 10).
Dominion performed a detailed analysis of the Correlations available in the code, and four sets of correlations were chosen based on the compatibility of the modeling assumptions used in deriving the various correlations. The selected sets use together only those correlations that have consistent or complementary bases and take advantage of previous industry experience and vendor recommendations. The four cases studied were:
Page 2 of 26
Case 1 (EEE)
Subcooled Void Model: EPRl Bulk Boiling Void Model: EPRl Two-Phase Friction Multiplier: EPRl Case2 (LSE)
Subcooled Void Model: LEVY Bulk Boiling Void Model: SMITH Two-Phase Friction Multiplier: EPRl 0
Case 3 (LHH)
Subcooled Void Model: LEVY Bulk Boiling Void Model: HOMOGENEOUS Two-Phase Friction Multiplier: HOMOGENEOUS Case 4 (LSH)
Subcooled Void Model: LEVY Bulk Boiling Void Model: SMITH Two-Phase Friction Multiplier: HOMOGENEOUS Although other relevant parameters such as void fraction or quality were verified throughout the benchmark evaluation, Dominions acceptance criterion was based on the DNBR performance of the analyzed cases. The EPRl models for bulk void, subcooled void and two-phase friction factor provided the best DNBR comparison for all the following:
VIPRE-D/BWU against the CHF experimental database for the AREVA BWU CHF correlations (DOM-NAF-2 Appendix A).
VIPRE-D/BWU against the LYNXT/BWU code results for a set of representative operating conditions (DOM-NAF-2 Section 5.0)
VIPRE-DNVRB-1 against the CHF experimental database for the Westinghouse W RB-1 CHF correlation (DOM-NAF-2 Appendix B).
VIPRE-DNVRB-1 against the COBRANVRB-1 code results for a set of representative operating conditions (DOM-NAF-2 Appendix B Section B.7).
Consistent with the results of these external and in-house benchmark studies, Dominion has demonstrated that the selection of the EPRl set of void and two-phase friction models is acceptable for application in DOM-NAF-2.
During the May 25, 2005 NRC-Dominion public meeting, Dominion agreed to provide some graphic comparisons to show VIPRE-Ds performance for void fraction and quality.
Figures 1 and 2 provide a comparison between the VIPRE-DNVRB-1 and the COBRWRB-1 codes. In particular, these figures show the void fraction and quality axial distributions in the hot channel at the time of minimum DNBR (3.0 seconds) for a Locked Rotor Transient event at Surry Power Station. Additional description and results associated with this transient were included in Section 8.7.3 of Appendix B to DOM-NAF-
- 2. In addition to the VIPRE-DNVRB-1 with the EPRI-EPRI-EPRI set of models, Figures 1 Page 3 of 26
and 2 show VIPRE-DNVRB-1 with the LEVI-SMITH-HOMOGENEOUS set of models, which is the set of bulk void, subcooled void and two-phase friction models closest to COBRANVRB-1 modeling in Reference 12 (LEVY-SMITH-BAROCZY).
Figures 1 and 2 show that the different void and two-phase friction models result in comparable void fraction and quality trends for COBRA and VIPRE-D. However, as noted above, DNBR comparisons are the main criterion used to evaluate the performance of VIPRE-D against other thermal-hydraulics codes and models. These comparisons demonstrated the acceptability of the EPRl models for bulk void, subcooled void and two-phase friction factor for use in the Dominion VIPRE-D models.
Figure 1: Locked Rotor Transient Equilibrium Quality Results at Time 3.0 Seconds 0.2 0.1 0
-0.1
- 5.
B 4 4.- -
?9
-0.2
-0.3
-0.4
-0.5 7
7 120 I
VlPRE EEE Axial Height (in)
Page 4 of 26
0.35 0.3 0.25 0.2 C.-
4-0.15 E
0 >
0.1 0.05 0
-0.05 Figure 2: Locked Rotor Transient Void Fraction Results at Time 3.0 Seconds VIPRE EEE 1
--L 60 100 120 1
Axial Height (in)
Page 5 of 26
NRC Question 3 Please explain the basis for choosing the EPRl bulk void model. Please describe the transient test cases employed to determine the values given in Table 5.4-1 and show the void distributions for the VlPRE and LYNXT codes at the initiation of DNB in some example cases. Also provide justification for the choice of the subcooled boiling model as well as the two-phase friction multiplier. Please show comparisons of the VIPRE-D code to data for these models. Please also describe the limitations and identify the ranges of applicability of each of these correlations.
DominiodDNC Response:
The basis for choosing the EPRl bulk void, subcooled void and two-phase friction multiplier models is provided in the response to Question 2.
The transient test cases employed to determine the values in Table 5.4-1 of DOM-NAF-2 are the same statepoints that were used in Section 5.1 of DOM-NAF-2 to benchmark VIPRE-D/BWU against LYNXT/BWU, and were also used to support the Licensing Amendment Request (LAR) for the Framatome ANP Fuel Transition (Reference 5). These statepoints were obtained from the UFSAR Chapter 15 events including the reactor core safety limits, axial offset envelopes (AO's), rod withdrawal at power (RWAP), rod withdrawal from subcritical (RWSC), control rod misalignment, loss of flow accident (LOFA), and locked rotor accident (LOCROT) events. These various statepoints provide sensitivity of DNB performance to the following: (a) power level (including the impact of the part-power multiplier on the allowable hot rod power FAH),
pressure and temperature (reactor core safety limits); (b) axial power shapes (AOs); (c) elevated hot rod power (misaligned rod); and (d) low flow (LOFA and LOCROT). These statepoints cover the full range of conditions and axial offsets in UFSAR Chapter 15 evaluations (except for MSLB that was evaluated separately), and were specifically selected to challenge the three BWU CHF correlations.
In the May 25, 2004 NRC-Dominion public meeting, Dominion indicated that LYNXT models use different (proprietary) void models, and that as a consequence, void distribution comparisons for VIPRE-D/BWU and LYNXTBWU would not be expected to match. A comparison of the void and quality distributions for COBRANVRB-1 and VIPRE-DNVRB-1 is provided in the response to Question 2.
NRC Question 4 In section 4.12, please explain what "nearly identical" means. Identifying the percent difference between the two results would be helpful or show the plot of the two DNBR calculations.
Dominion/DNC Response:
In the few occasions in which VIPRE-D did not converge when using the default values of the convergence criteria and damping factors, and these criteria and/or damping factors Page 6 of 26
were adjusted to ensure numerical convergence, the difference in the DNBR results reported by the code was 0.5% or less.
NRC Question 5 Section 5.3 describes a comparison with the COBRA code but states the Minimum DNBR (MDNBR) results are different because the analyses use different fuel types and CHF correlations. Please provide the latest comparisons between the codes using the same fuel type and CHF correlations. Please show the channel void distribution and quality at several selected times during the events. Show the steamline break, feedline break, and loss-of-flow events.
Dominion Response:
Section B.7 in Appendix B to Topical Report DOM-NAF-2 includes comparisons of VIPRE-DNVRB-1 with COBRAANRB-1 for Surry Power Station. These comparisons use the same fuel type (Westinghouse 15x15 SIF) and the same CHF correlation (Westing house W RB-1 ).
The comparisons include:
0 A statepoint safety analysis evaluation of 164 statepoints obtained from the Surry UFSAR Chapter 14 events, including the reactor core safety limits, axial offset envelopes (AO's), rod withdrawal at power (RWAP), rod withdrawal from subcritical (RWSC), control rod misalignment, loss of flow accident (LOFA), and locked rotor accident (LOCROT) events.
0 A Main Steam Line Break statepoint application.
0 A Feedwater Malfunction Transient (FWMAL).
0 A Locked Rotor Transient (LOCROT).
Selected examples comparing channel void and quality distributions were provided in the answer to Question 2.
NRC Question 6 The qualification document identifies the DNBR limits for the correlations for several pressure groups. Please explain what DNBR limit is applied or how the situation is handled when the range of validity is exceeded for the other parameters identified in Table A.5.2. Please also define the quality range for the correlation.
Dominion/DNC Response:
The qualification of the BWU CHF correlations with the VIPRE-D code was performed consistent with AREVA's approved Topical Report (Reference 4). The evaluations performed for the various pressure groups as well as the resulting limits were included in Appendix A for completeness. However, Dominion will only apply the DNBR limits listed in Table A.5-1 :
Page 7 of 26
I VIPRE-D/BWU-Z DNBR limit at or above 594 psia IDNBR limit below 700 psia I
1.59 1.18
~~~
IDNBR limit 700 - 2,400 psia I
1.20 DNBR limit below 1200 psia DNBR limit at or above 1200 psia I
VIPRE-D/BWU-ZM 1.39 1.22 IDNBR limit below 594 psia I
1.59 BWU-2 BWU-ZM BWU-N 400 to 2,465 0.36 to 3.55 Pressure
[psia]
Mass Velocity
[MI bm/h r-f t2]
The ranges of validity for the three BWU CHF correlations, including quality, are listed in Table A.5-2:
400 to 2,465 0.47 to 3.55 788 to 2,616 0.25 to 3.83 Thermodynamic Quality at CHF Less than 0.74 Less than 0.68 Less than 0.70 Mixing Vane Mid-Span Mixing Non-Mixing Vane I
Grids 1
Grids Applicability I
Grids The Low Flow Main Steam Line Break (MSLB) statepoint has been identified as the only event that might fall outside of the range of validity of the BWU correlations (for the mass velocity). This issue was already identified in the Framatome Fuel Transition Submittal (Reference 5). The Low Flow MSLB statepoint results in minimum DNBR values of the order of 5.8, which are obviously very far from being limiting. In this case Dominion will use the DNBR limit corresponding to the pressure at which the event takes place.
NRC Question 7 CHF is also sensitive to the axial power distribution. Since the correlations were developed from data with uniform or symmetric power distributions, please justify applicability of the correlations to the asymmetric power distributions that may be limiting in the plant calculations. Explain how the correlations are applied and describe any correction factors that may be applied to accommodate skewed distributions.
Dominion/DNC Response:
This issue is a generic concern with all CHF correlations, since these correlations were developed using a limited number of uniform and non-uniform axial power distributions.
Both the AREVA BWU set of correlations (qualified with VIPRE-D in Appendix A) and Page 8 of 26
the Westinghouse WRB-1 correlation (qualified in Appendix B) include a correction factor, the F-factor proposed by Tong (Reference 8), that is used to correct for non-uniform axial power shapes. The Tong Factor has been extensively verified by both Westinghouse (Reference 6) and AREVA (Reference 4) with numerous CHF test data, that over the years have included uniform and non-uniform power shapes. The use of this F-factor has been previously approved by the NRC staff.
In addition to the verifications performed by both vendors, Dominion performed a statepoint safety analysis evaluation using symmetric and non-symmetric, positively and negatively skewed axial power shapes to benchmark VIPRE-D/BWU with LYNXT/BWU and VIPRE-DNVRB-1 with COBRANVRB-1. The results of these evaluations do not show differences in the performance of the CHF correlations dependent on the form or uniformity of the axial power distributions.
NRC Question 8 Please justify applicability of the steady-state DNB correlations to steamline breaks since these events have rapid depressurizations where the steady-state correlations may not be applicable. These transients may also transition through slug and annular flow. As such, please justify the use of the EPRl bulk boiling drift flux model since it only applies to bubbly flow.
Do mi nio n/DNC Response:
The issue of the applicability of steady-state CHF correlations to transient analysis is generic to all thermal hydraulics codes and CHF correlations. Section 6.7 of Reference 9 provides a study to determine the applicability of steady-state CHF correlations to transient analysis for VIPRE-01. The NRC staff review concluded that the studies have shown that the transient CHF for power ramp and flow coastdown transients are higher than the steady-state CHF, and that, except for very rapid depressurization events (LOCA), the use of CHF correlations developed with steady-state CHF data can correctly or conservatively predict the transient CHF when the instantaneous local fluid conditions are used (Reference 10).
Dominions Main Steam Line Break (MSLB) DNBR evaluations are performed at the limiting statepoint. Dominion does not perform full DNBR transient analysis for this event. For MSLB all the DNB limiting statepoints occur after the pressurizer has drained and the upper head has flashed. Therefore, pressure is changing relatively slowly and the quasi steady-state assumption is appropriate.
The use of the EPRl bulk boiling void correlation was justified in the response to Question 2.
NRC Question 9 Please discuss whether the slip option will be used and if so justifL the slip ratio employed in the DNBR calculations.
Page 9 of 26
DominiodDNC Response:
Dominion VIPRE-D models do not use the slip option.
NRC Question 10 Since the conduction model will not be used, please explain how the stored energy in the rod is accounted for. Please explain why use of the dummy rod model is conservative since the conduction model does not include the effects of gap conductance and initial stored energies. Please also describe how the heat flux is calculated for use as input to the VIPRE-D code.
DominiodDNC Response:;
Conduction models are typically used to perform fuel temperature calculations and to simulate delay of energy transport. Dominion does not plan to use VIPRE-D for these applications. The use of the dummy rod model is consistent with previously approved Dominion methodologies (Reference 12).
The use of the dummy rod model requires the user to provide the fuel rod surface heat flux as one of the operating input conditions. For steady-state statepoint analysis this value is easily calculated based on core thermal power. Fuel rod surface heat flux forcing functions for transient calculations are provided by an NRC-approved transient system code (e.g., RETRAN). Transient system codes account for fuel conduction, gap conductance, and delayed energy transport effects. Therefore, the use of the dummy rod model appropriately includes the relevant effects into the analysis.
Page 10 of 26
Appendix A to DOM-NAF-2 NRC Question 1 The appropriate statistical analysis of the data, which form Tables A.3.1-1, A.3.2-7, and A.3.3-1 is an analysis of variance of a mixed-effects model.
a) Give the appropriate analysis of variance tables for these mixed-effects models.
b) Formulate the appropriate statistical hypothesis tests to justify the values for M/P and owp used in q. A. 1.1 based on the data in Tables A.3. I - 1, A.3.2-1, and A.3.3-7, c) For those cases where individual DNBR design limits were developed for each low pressure group, how were the results of the above analysis of variance taken into account?
DominiodDNC Response:
As clarified in our May 25, 2005 meeting with the NRC staff, this question refers to Tables A.4.1-2, A.4.2-2 and A.4.3-2.
a) The qualification of the BWU CHF correlations with the VIPRE-D code was performed consistent with AREVAs approved Topical Report (Reference 4). An analysis of variance (ANOVA) test was performed for each one of the three BWU correlations. This analysis was not included in the original submission because it was not deemed to provide additional substantial information to the qualification.
An ANOVA test divides the database in several groupings according to a given variable and then evaluates whether or not the distributions for each one of the groups appear to belong to the overall distribution. If all the groupings belong to the same distribution it can be deduced that the total population does not show a bias with respect to that particular variable Even though it was recognized that ANOVA tests cannot be used as the sole measure of the performance of a CHF correlation, they can be useful to indicate an extremely bad mismatch (very large F statistic).
The variables analyzed were mass velocity, pressure, quality, test cell type and axial flux shape type.
The results of the ANOVA tests for the BWU-Z, BWU-ZM and BWU-N correlations are provided in Tables 1 through 5. The results for the BWU-ZM (Table 3) and BWU-N (Table 4) correlations prove the ANOVA hypothesis: all the groups analyzed belong to the same distribution, i.e. there is no bias of the results regarding the analyzed variables. The results for the BWU-Z correlation (Table 1) show F values slightly above the critical value, but still reasonably small. Following the AREVA approach in Reference 4, an additional ANOVA test was performed excluding the low pressure data from the BWU-Z database. This treatment is consistent with the fact that the low pressure data were taken separately by determining a separate DNBR design limit. Table 2 summarizes the results of this second ANOVA analyses.
While the values of the F statistic do decrease somewhat in most cases, they are still slightly above the critical value of F for the appropriate number of degrees of freedom. An additional ANOVA test was also performed for BWU-N without the low pressure data (Table 5), and the trends observed are the same as those shown in Table 4. For all cases, these are the same trends showed by AREVA in Reference 4.
Page 11 of 26
Table 1 : M/P CHF Performance by Independent Variable Grouping using the entire BWU-Z database at 95% confidence level 0.5 Mlbm/hr-ft2 1
9 I
I I
I I
I I
0.9559 I
0.1 147 1.I083 0.7833 Number Average I of Data I M/P Unit Cell 21 8 Guide Tube 170 Other 140 1.0077 0.0894 1.I 683 0.7669 0.9849 0.0917 1.I 693 0.7261 0.9877 0.0899 1.2974 0.6980 Fdistribution = 3-6760 Fcritic,1(2,525) = 3.01 28 Page 12 of 26 All Data 528 0.9950 0.0907 1.2974 0.6980
Number of Data Average Standard Maximum Minimum M/P Deviation M/P M/P c 10%
116 1.01 86 10%- 15%
122 0.9859 15% - 20%
115 0.9737
> 20%
111 0.9746 Fdistribution = 6.7560 0.0839 1.1693 0.7261 0.0887 1.1576 0.771 7 0.0787 1.1479 0.81 26 1.I41 1 0.6980 0.0951 Fc,itiC,1(3,460)
= 2.6243 Page 13 of 26 Unit Cell 188 Guide Tube 150 Other 126 1.0024 0.0880 1.I683 0.7669 0.981 0 0.0931 1.1693 0.7261 0.9761 0.0805 1.1457 0.6980 Fdistribution = 4.1 660 Fcritica\\(2,461) = 3.01 53 All Data 464 0.9883 0.0883 1.1693 0.6980
Table 3: M/P CHF Performance by Independent Variable Grouping using the entire BWU-ZM database at 95% confidence level Number of Data Average Standard Maximum Minimum M/P Deviation M/P M/P
< 1000 psia 1000 - 1500 psia 1500 - 2000 psia
> 2000 psia 11 0.9758 0.0545 1.0749 0.8832 15 1.0202 0.0840 1.21 10 0.91 25 53 1.0249 0.0982 1.1 844 0.81 42 69 1.0099 0.0833 1.2299 0.7793 Fdistribution = 1.0528 Fcritical(3,144) = 2.6674 c 10%
11 1.0004 I
0.0622 1.1249 0.91 95 10% - 15%
15% - 20%
> 20%
Page 14 of 26 39 1.0111 0.1023 1.1 699 0.7793 35 1.0082 0.0737 1.1 680 0.8783 0.0897 1.2299 0.8496 63 1.0209 Fdistribution = 0.2794 FcritiCa1(3,144) = 2.6674 Unit Cell Guide Tube 76 1.0230 0.1000 1.2299 0.81 42 72 1.0041 0.071 5 1.1747 0.7793 Fdistribution = 1.7240 Fcritical( 1,146) = 3.9059 All Data 148 1.0138 I
0.0875 1.2299 0.7793
Number of Data Average Standard Maximum Minimum M/P Deviation M/P M/P Page 15 of 26
Table 5: M/P CHF Performance by Independent Variable Grouping using the BWU-N database without the low pressure data at 95% confidence le Page 16 of 26
b) As clarified in our May 25, 2005 meeting with the NRC staff, this question refers to Eq. A.4.1.1 and Tables A.4.1-1, A.4.2-1 and A.4.3-1. The validity of this equation is based on two assumptions: 1) the average M/P is 1.0 and 2) the M/P distribution is normal. These two assumptions were demonstrated in Appendix A:
CORRELATION BW U-Z BW U-ZM M/P DISTRIBUTION NORMAL AVERAGE M/P = 1.0 Table A.4.1-1 Page A-1 3 Table A.4.2-1 Page A-23 Page A-33 I
BWU-N I
Table A.4.3-1 (Hypernormai distribution, also seen by AREVA in Reference 4) c) Different DNBR limits were developed for low pressure groups for the BWU-Z and BWU-N correlations following the guidance in Reference 4. The plots (not the ANOVA tests displayed in Tables 1 and 4 respectively) seemed to indicate a poorer performance of the correlation at low pressures, and for that reason a separate DNBR limit was calculated at low pressures. In those cases, a new ANOVA analysis was performed excluding the low-pressure data (Table 2 for BWU-Z and Table 5 for BWU-N). These ANOVA results were similar to the previous results. All Dominion results were similar to AREVA results in Reference 4.
NRC Question 2 As in Appendix B, you state that the plots show that there are no biases in the M/P ratio distributions, and that the performance of the CHF correlations is independent of the three variables of interest. The plots show a mostly uniform scatter of the data and no obvious trends or slopes. The plots again suqaest but do not demonstrate that the claims made in those sentences are true. Please give the appropriate statistical analvsis, that demonstrates the truth of the claim.
Domi n io n/DNC Response:
As discussed in the response to Question 1 for Appendix A, an ANOVA analysis was performed for each correlation to formally demonstrate that the performance of the BWU CHF correlations is not biased by the three independent variables present in the correlations (mass velocity, pressure and quality). Please refer to the response to Question 1 for analyses demonstrating no biases in the M/P distributions.
Page 17 of 26
NRC Question 3 Please show the empirical probability density functions for the M/P values used in the analyses together with the estimate of the 95-percent fractile for each correlation.
Dominion/DNC Response:;
The probability density functions for the BWU-Z, BWU-ZM and BWU-N correlations are shown in Figures 3,4 and 5.
The 528 datapoints of the VIPRE-D/BWU-Z M/P distribution summarized in Section A.4.1 of DOM-NAF-2 Appendix A were used to create the empirical probability density function. These datapoints were distributed among 24 equal bins that covered the entire range of M/P in the VIPRE-D/BWU-Z distribution, and the frequency of data in each bin was determined. The resulting empirical probability density function for the VIPRE-D/BWU-Z M/P distribution was then compared with the probability density function of a normal distribution of mean 0.995 and standard deviation 0.0907, which are the mean and standard deviation calculated for the VIPRE-D/BWU-Z M/P distribution in Section A.4.1 of DOM-NAF-2 Appendix A. Figure 3 also displays the obtained 95% fractile (1.13) for the data and the VIPRE-D/BWU-Z DNBR limit obtained in Section A.4.1 of DOM-NAF-2 Appendix A (1.20).
Figure 3: VIPRE-D/BWU-Z Probability Density Function 14.00%
12.00%
10.00%
F 8.00%
0 z
w 3
a LL 6.00%
4.00%
2.00%
0.00%
7 95% FRACTILE = 1.13 I '
I I
3.65 0.68 0.71 0.74 0.77 0.80 0.83 0.86 0.89 0.92 0.95 0.98 1.01 1.04 1.07 1.10 1.13 1.16 M/P I
1.19 1.22 1.25 1.28 1.31 1.34 Page 18 of 26
The 148 datapoints of the VIPRE-D/BWU-ZM M/P distribution summarized in Section A.4.2 of DOM-NAF-2 Appendix A were used to create the empirical probability density function. These datapoints were distributed among 27 equal bins that covered the entire range of M/P in the VIPRE-D/BWU-ZM distribution, and the frequency of data in each bin was determined. The resulting empirical probability density function for the VIPRE-D/BWU-ZM M/P distribution was then compared with the probability density function of a normal distribution of mean 1.0138 and standard deviation 0.0875, which are the mean and standard deviation calculated for the VIPRE-D/BW U-ZM M/P distribution in Section A.4.2 of DOM-NAF-2 Appendix A. Figure 4 also displays the obtained 95% fractile (1.I 6) for the data and the VIPRE-D/BWU-ZM DNBR limit obtained in Section A.4.2 of DOM-NAF-2 Appendix A (1.1 8).
Figure 4 VIPRE-D/BWU-ZM Probability Density Function 12.00%
10.00%
8.00%
r_ g 0
z W
6.00%
3 s
E 4.OO0/o 2.00%
0.00%
45% FRACTll 16
--I---
I I
I I
E I
1 I
1 I
c I
I t
I I
IIPRE-~/B\\
INBR LilMli I
1 I
I r
075 077 079 081 083 085 087 089 091 093 095 097 099 101 103 105 107 109 111 113 115 117 119 121 123 125 127 WP Page 19 of 26
The 1090 datapoints of the VIPRE-D/BWU-N M/P distribution summarized in Section A.4.3 of DOM-NAF-2 Appendix A were used to create the empirical probability density function. These datapoints were distributed among 26 equal bins that covered the entire range of M/P in the VIPRE-D/BWU-N distribution, and the frequency of data in each bin was determined. The resulting empirical probability density function for the VIPRE-D/BWU-N M/P distribution was then compared with the probability density function of a normal distribution of mean 1.001 8 and standard deviation 0.1038, which are the mean and standard deviation calculated for the VIPRE-D/BW U-N M/P distribution in Section A.4.3 of DOM-NAF-2 Appendix A. Figure 5 also displays the obtained 95% fractile (1.16) for the data and the VIPRE-D/BWU-N DNBR limit obtained in Section A.4.3 of DOM-NAF-2 Appendix A (1.22).
Figure 5: VIPRE-DIBWU-N Probability Density Function 14.00%
12.00%
10.00%
8.00%
+
0 z
w 3
0 6.00%
L 4.00%
2.00%
0.00%
95% FRACT~LE = 1.16 I I
064 067 070 073 076 079 082 085 088 091 094 097 100 103 106 109 112 115-118 121 124 127 130 133 136 139 MIP Page 20 of 26
NRC Question 1 The appropriate statistical analysis of the data that form Table 6.6-1 is an analysis of variance of a mixed-effects model.
a) Give the appropriate analysis of variance table for this mixed-effects model.
b) Formulate the appropriate statistical hypothesis tests to justify the values for M/P and owp used in q. B.6.1.
DominiodDNC Response:
a) The qualification of the WRB-1 CHF correlations with the VIPRE-D code was performed consistent with Dominions approved Topical Report for COBRAMlRB-1 (Reference 12) and Westinghouses approved Topical Report (Reference 6). An analysis of variance (ANOVA) test was performed but was not included in the original submission because it was not deemed to provide additional substantial information to the qualification. As mentioned in the response to Question 1 on Appendix A, it was recognized that ANOVA tests cannot be used as the sole measure of the performance of a CHF correlation, but they can be useful to indicate an extremely bad mismatch. The variables analyzed were mass velocity, pressure, quality and test cell type.
The ANOVA results for VIPRE-DNVRB-1 (Table 6) exceed the critical values of F for some comparisons, but other comparisons prove the hypothesis that all the groups belong to the same distribution, i.e. that there is no bias of the results regarding the analyzed variables. These are the same trends discussed in the NRCs Safety Evaluation of WRB-1 (Reference 6).
Page 21 of 26
Table 6: M/P CHF Performance by Independent Variable Grouping using the entire WRB-1 database at 95% confidence level Number of Data Average Standard Maximum Minimum M/P Deviation M/P M/P Analvsis bv Mass Velocities 58 Below 1.25 MIbm/hr-ft2 0.9849 1.01 25 0.9930 131 247 203 159 1.25 - 1.75 MIbm/hr-ft2 MI bm/h r-f t2 Mlbm/h r-ft2 MIbm/hr-ft2 1.75 - 2.25 2.25 - 2.75 2.75 - 3.25
~
0.0753 1.1158 0.8005 0.0970 1.3069 0.7632 0.091 4 1.271 1 0.7600 MI b m/h r-f t2 Below 5%
5% to 10%
10% to 15%
15% to 20%
20% to 25%
262 1.oooo 0.0762 1.3069 0.801 7 199 1.01 19 0.0778 1.2034 0.8226 247 1.0090 0.081 0 1.271 1 0.8005 169 0.9927 0.0942 1.231 2 0.7600 68 1.0222 0.0922 1.2234 0.7662 Fdistribiition = 2.4209 Fcriticrri(4.940) = 2.381 4 1.0129 I 0.0688 I 1.1715 I
O. 8 2 r 1 Thimble I
207 I 0.9910 I 0.0771 I
1.2229 1.0074 0.0755 1.1858 0.8337 1.01 37 0.0787 1.2356 0.8017 0.7873 All Data Page 22 of 26
d) The validity of equation B.6.1 is based on two assumptions: 1) the average M/P is 1.0 and 2) the M/P distribution is normal. These two assumptions were demonstrated in Appendix B:
CORRELATION WRB-1 M/P Dl ST R I BUT 10 N NORMAL AVERAGE M/P = 1.0 Table B.6-1 Page B-12 NRC Question 2 On page B-14 you state, These plots show that there are no biases in the M/P ratio distributions, and that the performance of the WRB-1 CHF correlation is independent of the three variables of interest. The plots show a mostly uniform scatter of the data and no obvious trends or slopes. The plots suqqest but do not demonstrate that the claims made in those sentences are true. Please give the appropriate statistical analvsis that demonstrates the truth of the claim.
Dominion/DNC Response:
As discussed in the response to Question 1 for Appendix B, an ANOVA analysis was performed to formally demonstrate that the performance of the WRB-1 CHF correlation is not biased by the three independent variables present in the correlation (mass velocity, pressure and quality). Please refer to the response to Question 1 for an analysis demonstrating no biases in the M/P distribution.
NRC Question 3 Please show the empirical probability density function for the M/P values used in the analyses together with the estimate of the 95-percent fractile.
Dominion/DNC Response:
The probability density function for the WRB-1 correlation is shown in Figure 6.
The 945 datapoints of the VIPRE-DMIRB-1 M/P distribution summarized in Section B.6 of DOM-NAF-2 Appendix B were used to create the empirical probability density function.
These datapoints were distributed among 31 equal bins that covered the entire range of M/P in the VIPRE-DNVRB-1 distribution, and the frequency of data in each bin was determined. The resulting empirical probability density function for the VIPRE-DNVRB-1 M/P distribution was then compared with the probability density function of a normal distribution of mean 1.0051 and standard deviation 0.0827, which are the mean and standard deviation calculated for the VIPRE-DMIRB-1 M/P distribution in Section B.6 of DOM-NAF-2 Appendix B. Figure 6 also displays the obtained 95% fractile (1.14) for the data and the VIPRE-DMIRB-1 DNBR limit obtained in Section B.6 of DOM-NAF-2 Appendix B (1.1 7).
Page 23 of 26
Figure 6: VIPRE-DMIRB-I Probability Density Function 0.12 0 1 0.08 0
2 w 0.06 3
s E
0 04 0.02 0 -L I
1 95% FRACTILE = 1.14 i
VIP~E-DNURB-I DNBR LIMIT = 1.87
- - i 1
1 074 076 078 080 082 084 088 088 OW 092 094 096 098 100 102 104 106 108 110 112 114 116 118 120 122 124 128 128 130 132 134 MIP 0.12 0.1 0.08 0.06 0.04 0.02 0
Page 24 of 26
References
- 1. Letter from S. R. Monarque (NRC) to D. A. Christian (Dominion), North Anna and Surry Power Stations, Units 1 and 2 and Millstone Power Station, Units 2 and 3 -
Request for Additional Information Rearding Topical Report DOM-NAF-2, Reactor Core Thermal Hydraulics Using the VIPRE-D Computer Code (TAC Nos. MC4571 through MC4576), Serial No.05-328, May 19, 2005.
- 2. Letter from L. N. Hartz (Dominion) to Document Control Desk (NRC), Virginia Electric and Power Company (Dominion), Dominion Nuclear Connecticut, Inc.
(DNC), North Anna and Surry Power Stations Units 1 and 2, Millstone Power Station Units 2 and 3, Request for Approval of Topical Report DOM-NAF-2, Reactor Core Thermal-Hydraulics Using the VIPRE-D Computer Code Including Appendix A, Qualification of the F-ANP BWU CHF Correlations in the Dominion VIPRE-D Computer Code, Serial No.04-606, dated September 30, 2004.
- 3. Letter from L. N. Hartz (Dominion) to Document Control Desk (NRC), Virginia Electric and Power Company (Dominion), Dominion Nuclear Connecticut, Inc. (DNC)
North Anna and Surry Power Stations Units 1 and 2, Millstone Power Station Units 2 and 3, Request for Approval of Appendix B of Topical Report DOM-NAF-2, Qualification of the Westinghouse WRB-1 CHF Correlation in the Dominion VIPRE-D Computer Code, Serial No.05-020, dated January 13,2005.
- 4. Topical Report, BAW-lOl99P-A, The BWU Critical Heat Flux Correlations, B&W Fuel Company, August 1996, including Addendum 2, Application of the BWU-Z CHF Correlation to the Mark-BW17 Fuel Design with Mid-Span Mixing Grids, Framatome Cogema Fuels, November 2000.
- 5. Letter from L. N. Hartz (Dominion) to US NRC Document Control Desk Virginia Electric and Power Company, North Anna Power Station Units 1 and 2, Proposed Technical Specifications Changes and Exemption Request, Use of Framatome ANP Advanced Mark-BW Fuel, Serial No. 02-1 67, dated March 28, 2002. (Proprietary version).
- 6. Topical Report, WCAP-8762-P-A (Proprietary) and WCAP-8763-A (Non-proprietary),
New Westinghouse Correlation WRB-1 for Predicting Critical Heat Flux in Rod Bundles with Mixing Vane Grids, Westinghouse Electric Corporation, July 1984.
- 7. Technical Report, EPRl NP-2511-CCM Volume 2, Revision 4, VIPRE-01: A Thermal-Hydraulic Code for Reactor Cores. Volume 2: Users Manual, February 2001.
- 8. Technical Report, TID-25887, Boiling Crisis and Critical Heat Flux, L. S. Tong, AEC Critical Review Series, 1972.
- 9. Technical Report, EPRl NP-2511 -CCM-A Volume 4, VIPRE-01: A Thermal-Hydraulic Code for Reactor Cores. Volume 4: Applications, April 1987.
10.Letter from C. E. Rossi (NRC) to J. A. Blaisdell (UGRA Executive Committee),
Acceptance for Referencing of Licensing Topical Report, EPRl NP-2511 -CCM, VIPRE-01 : A Thermal-Hydraulic Analysis Code for Reactor Cores, Volumes 1, 2, 3 and 4, May 1, 1986.
Page 25 of 26
- 11. Letter from A. C. Thadani (NRC) to Y. Y. Yung (VIPRE-01 Maintenance Group),
Acceptance for Referencing of the Modified Licensing Topical Report, EPRl NP-251 1 -CCM, Revision 3, VIPRE-01 : A Thermal Hydraulic Analysis Code for Reactor Cores, (TAC No. M79498), October 30, 1993.
12.Topical Report, VEP-NE-3-A, Qualification of the WRB-1 CHF Correlation in the Virginia Power COBRA Code, R. C. Anderson and N. P. Wolfhope, November 1986.
- 13. Topical Report, WCAP-7667-P-A (Proprietary) and WCAP-7755-A (Non-proprietary),
Interchannel Thermal Mixing with Mixing Vane Grids, F. F. Cadek, Westinghouse Electric Corporation, January 1975.
- 14. Topical Report, WCAP-7941 -P-A (Proprietary) and WCAP-7959-A (Non-proprietary),
Effect of Axial Spacing on lnterchannel Thermal Mixing with the R Mixing Vane Grid, F. F. Cadek et al., Westinghouse Electric Corporation, January 1975.
- 15. Topical Report, WCAP-8298-P-A (Proprietary) and WCAP-8299-A (Non-proprietary)
The Effect of 17x17 Fuel Assembly Geometry on lnterchannel Thermal Mixing, F.
E. Motley, et al., Westinghouse Electric Corporation, January 1975.
16.Technical Report, EPRl NP-2511-CCM-A Volume 1, Revision 4, VIPRE-01 : A Thermal-Hydraulic Code for Reactor Cores. Volume 1 : Mathematical Modeling, February 2001.
Page 26 of 26