Question # -I · The mesh is presented in Figure 03.09.02-169-2, and its three types of orifices in Figure 03.09.02-169-3. This mesh has been optimized for computational purposes;

• AAR EVA

September 28, 2012NRC:12:048

Document Control DeskU.S. Nuclear Regulatory CommissionWashington, D.C. 20555-0001

Response to U.S. EPR Design Certification Application RAI No. 508, Supplement 6

Ref. 1: E-mail, Getachew Tesfaye (NRC) to Dennis Williford (AREVA NP Inc.), "U.S. EPR Design CertificationApplication RAI No. 508 (6005,6000,5994), FSAR Ch. 3," August 26, 2011.

Ref. 2: E-mail, Dennis Williford (AREVA NP Inc.) to Getachew Tesfaye (NRC), "Response to U.S. EPR DesignCertification Application RAI No. 508 (6005,6000,5994), FSAR Ch. 3, Supplement 5,"September 20, 2012.

In Reference 1, the NRC provided a request for additional information (RAI) regarding the U.S. EPR designcertification application. Reference 2 provided a schedule for a technically correct and complete finalresponse to the remaining question of RAI 508 (Question 03.09.02-169), and a history of the priorsupplemental responses.

Enclosed is a technically correct and complete final response to the remaining question of RAI 508, as shownin the table below.

AREVA NP Inc. (AREVA NP) considers some of the material contained in the attached response to beproprietary. As required by 10 CFR 2.390(b), an affidavit is attached to support the withholding of theinformation from public disclosure. Proprietary and non-proprietary versions of the enclosure to this letterare provided.

The following table indicates the respective pages in the enclosed response that contain AREVA NP's finalresponse to the subject question.

Question #RAI 508 - 03.09.02-169

Start Page

2

End Page

18 -I

AREVA INC.3315 Old Forest Road, P.O. Box 10935, Lynchburg, VA 24506-0935Tel.: 434 832 3000 www.areva.com

Document Control DeskSeptember 28, 2012

NRC:12:048Page 2

This concludes the formal AREVA NP response to RAI 508. There are no questions from this RAI for whichAREVA NP has not provided responses.

If you have any questions related to this submittal, please contact Mr. Darrell Gardner by telephone at704-805-2355, or by e-mail to [email protected].

Regulatory AffairsAREVA NP Inc.

Enclosures:1. Proprietary and Non-Proprietary versions of RAI 508, Supplement 62. Signed and notarized affidavit

cc: G. TesfayeDocket No. 52-020

AFFIDAVIT

COMMONWEALTH OF VIRGINIA )) ss.

CITY OF LYNCHBURG )

1. My name is Russell D. Wells. I am U.S. EPR COLA Licensing

Manager for AREVA NP Inc. and as such I am authorized to execute this Affidavit.

2. I am familiar with the criteria applied by AREVA NP to determine

whether certain AREVA NP information is proprietary. I am familiar with the policies

established by

AREVA NP to ensure the proper application of these criteria.

3. I am familiar with the AREVA NP information contained in "Response

to Request for Additional Information No. 508, Supplement 6," and referred to herein as

"Document." Information contained in this Document has been classified by AREVA NP

as proprietary in accordance with the policies established by AREVA NP for the control

and protection of proprietary and confidential information.

4. This Document contains information of a proprietary and confidential

nature and is of the type customarily held in confidence by AREVA NP and not made

available to the public. Based on my experience, I am aware that other companies

regard information of the kind contained in this Document as proprietary and

confidential.

5. This Document has been made available to the U.S. Nuclear

Regulatory Commission in confidence with the request that the information contained in

this Document be withheld from public disclosure. The request for withholding of

proprietary information is made in accordance with 10 CFR 2.390. The information for

which withholding from disclosure is requested qualifies under 10 CFR 2.390(a)(4)

"Trade secrets and commercial or financial information".

6. The following criteria are customarily applied by AREVA NP to

determine whether information should be classified as proprietary:

(a) The information reveals details of AREVA NP's research and

development plans and programs or their results.

(b) Use of the information by a competitor would permit the competitor to

significantly reduce its expenditures, in time or resources, to design,

produce, or market a similar product or service.

(c) The information includes test data or analytical techniques concerning

a process, methodology, or component, the application of which

results in a competitive advantage for AREVA NP.

(d) The information reveals certain distinguishing aspects of a process,

methodology, or component, the exclusive use of which provides a

competitive advantage for AREVA NP in product optimization or

marketability.

(e) The information is vital to a competitive advantage held by AREVA

NP, would be helpful to competitors to AREVA NP, and would likely

cause substantial harm to the competitive position of AREVA NP.

The information in the Document is considered proprietary for the reasons set forth in

paragraphs 6(d) above.

7. In accordance with AREVA NP's policies governing the protection and

control of information, proprietary information contained in this Document has been

made available, on a limited basis, to others outside AREVA NP only as required and

under suitable agreement providing for nondisclosure and limited use of the information.

8. AREVA NP policy requires that proprietary information be kept in a

secured file or area and distributed on a need-to-know basis.

9. The foregoing statements are true and correct to the best of my

knowledge, information, and belief.

SUBSCRIBED before me this

day of September, 2012.

Kathleen A. BennettNOTARY PUBLIC, COMMONWEALTH OF VIRGINIAMY COMMISSION EXPIRES: 8/31/2015Registration No. 110864

KAT.LM AN BENNWTTNOta" Pubki

CommnW l of Virginia110864

MY Cwoss, n bpw Aug 31, 2015

Response to

Request for Additional Information No. 508, Supplement 6

8/26/2011

U. S. EPR Standard Design CertificationAREVA NP Inc.

Docket No. 52-020SRP Section: 03.03.01 - Wind Loading

SRP Section: 03.07.03 - Seismic Subsystem AnalysisSRP Section: 03.09.02 - Dynamic Testing and Analysis of Systems Structures and

Components

Application Section: 03.03.01

QUESTIONS for Structural Engineering Branch 2 (ESBWR/ABWR Projects) (SEB2)QUESTIONS for Engineering Mechanics Branch 2 (ESBWR/ABWR Projects) (EMB2)

AREVA NP Inc.Response to Request for Additional Information No. 508, Supplement 6U.S. EPR Design Certification Application Page 2 of 18

Question 03.09.02-169:

Follow up to RAI 422, Question 03.09.02-131:

The FIV analysis of the RPV upper internals reported by the applicant in CVAP Report Rev. 0(see Section 4.5.3) utilized thermal hydraulic conditions determined from one dimensionalanalysis. The results of this analysis indicate that several of the components would fail both thehigh cycle fatigue criteria (2800 psi, rms) and the vortex shedding stress criteria (13,600 psi, 0-peak) as reported in the markup accompanying the May 3, 2011 response to RAI 422 Question03.09.02-131 (see Table 4-20). The applicant has, since the issuance of CVAP Report Rev. 0,performed a three dimensional CFD analysis of the U.S. EPR and has used the results toupdate the thermal hydraulic conditions employed in the RPV upper internals FIV analysis. Theapplicant has stated in both CVAP Report Rev. 0 Section 4.5 and in the mark up accompanyingthe response to RAI 422 Question 03.09.02-131 (see Section 4.5.3) that the CFD approach hasbeen benchmarked against the ROMEO 1/5 scale flow testing, but the applicant has providedno information from that analysis. Further, the updated predictions substantially reduce thepredicted stress for the RPV upper internals, in some cases by more than an order ofmagnitude, resulting in all of the upper internals meeting the stress criteria by wide margins.The applicant is requested to provide the discussion of the CFD models and the ROMEO testswhich addresses the following points.

1 a. The applicant is requested to address the procedure used to validate the CFD model ona system reflecting the degree of complexity of the RPV upper internals, including themetrics and reference planes or locations used.

lb. The applicant is requested to address the sensitivity analysis performed to ensure thatthe grid size of the model is sufficiently small such that further grid refinement would notaffect the CFD results.

The information should be included in the CVAP Report.

Response to Question 03.09.02-169:

Item a:

The response to this question includes the following subsections:

" Computational fluid dynamic (CFD) model details.

* Code convergence.

* ROMEO test details.

* CFD analysis validation procedure.

CFD Model Details

The CFD simulation represents the upper part of the U.S. EPR vessel as shown in Figure03.09.02-169-1. Its computational domain is limited by the upper core plate (UCP), the uppersupport plate (USP), and the end of the hot legs. It is composed of:

" The UCP, with its orifices of three different types.

* The upper plenum, with its different elements.


* The outlet nozzles and the four hot legs, which are 10 diameters long (so that outletconditions are far enough from the plenum).

Inside the plenum, the control rod guide assemblies (CRGAs), the first two guide plates, as wellas the lateral openings (toward the plenum volume) are modeled. The column internalcomponents (tie rods, continuous guides, split tubes and rods) are not represented becausethey do not have any impact on the flow patterns around the columns. The flow is blockedslightly above the second guide plate (in both the plant design and mock-up) as most of the flowgoes outside, passing through the gussets. The support columns have also been represented.

The numerical CFD model has been developed with a mesh of 15.4 million hexahedral cells.The mesh is presented in Figure 03.09.02-169-2, and its three types of orifices in Figure03.09.02-169-3. This mesh has been optimized for computational purposes; it is a relativelycoarse mesh when the flow is easy to calculate (for example, the free orifices group) and arefined mesh when the flow field is more complex (for example, at the bottom of the CRGAgroup where many openings are located).

In the plenum, a boundary layer has been meshed around obstacles such as CRGAs, columns,aeroballs and plenum envelop, in order to accurately calculate the attenuation of the fluidvelocity profile near them.

The hot legs are meshed with an "O-mesh" to have a good compromise between the number ofcells and the computing time. Hot legs 3 and 4 are finely meshed, with 3 million cells for each ofthem, whereas hot legs 1 and 2 present a relatively coarse mesh, with 310,000 cells for each ofthem. For the refined mesh hot legs, the first cell boundary layer has a thickness of about 0.02inch (0.51mm).

This thickness of the boundary layer has been determined in order to confirm the use of the"standard wall function" approach. The dimensionless wall distance y+ is defined as(y+=pUty/p), with y the height of the centroid of the first cell, p is the fluid density, p is thedynamic viscosity and Ut is the shear velocity due to wall friction. This distance is equivalent toa Reynolds number based on the distance to the walls, and was validated to be between 20 and300 for the wall functions.

Due to the complexity and the size of the whole mesh, a non-conformal grid interface has beenmodeled between the upper plenum and the outlet nozzles.

Code Converqence

To demonstrate code convergence, five points were chosen in the computational domain:

" Three points located in the hot leg 3 near the scoop section.

" Two points located in the plenum between hot leg 3 and hot leg 4 (where there may bestrong fluctuations).

The local velocity components (in m/s) as well as the temperature (in OK) have been monitoredduring the iterations for each probe, and their evolutions were plotted. The convergence criteriawere fixed at 10-4 for the momentum equation and 10 3 for the continuity equation, and wereachieved at 6200 iterations.


The condition on the range of y+ is satisfied for the four hot legs, and especially for the refinedmesh of the hot legs (the average value is about 45).

ROMEO Test Details

The goal of the ROMEO tests is to characterize the upper plenum hydraulics of the U.S. EPRreactor. The ROMEO mock-up represents the upper part of the reactor vessel at a scale of(1/5.2). The simulated part is located between the UCP, the USP, and the end of the hot legs.A sketch of the mock-up is presented in Figure 03.09.02-169-4.

The core effect is taken into account by additive head losses through the UCP. The CRGAs,columns, and aeroballs lance fingers are geometrically represented.

The UCP of the mock-up consists of two plates separated by 241 ventures (of which aim is tomeasure flow rates through the UCP). Each plate is drilled with 241 holes and diaphragms areinserted in the upstream holes. These diaphragms are of three types, CRGA, column, and freeorifice, as there are three different outlet groups. The diameters of the diaphragms weredetermined to represent the upstream effect of the core on flow distribution. The followingmethod was used:

" Realization of a computation of the reactor upper plenum hydraulics with simulation of fuelassemblies, using porous media and singular head losses. The length of represented coreis about one meter or 39.37 inches.

" Realization of a computation of the reactor upper plenum hydraulics with an empty corecavity upstream UCP (with representation of UCP holes, but without representation of fuelbundles).

" Determination of additive head loss coefficient to put into UCP holes in the empty cavitycomputation in order to obtain the same flow rate distribution as the one obtained withsimulation of fuel assemblies.

The mock-up has been designed to obtain a uniform pressure distribution upstream of the UCP.In this way, the UCP flow field is only created by downstream effects and in particular by the hotleg suction effect.

CFD Analysis Validation Procedure

The CFD analysis of the U.S. EPR upper internals is performed using the AREVA computerprogram "STAR-CD" Version 4.02. The results of the computer runs performed for thevalidation of STAR-CD are compared with analytical or semi-analytical solutions and also withmeasurements obtained from experiments where both the integral and separate effects wereverified.

The validation includes the comparisons of the results obtained with the computations andexperiments that were specifically dedicated to flow distributions and thermal mixing in thereactor vessel internals." These experiments include U.S. EPR upper internals hydraulics,represented at scale 1/5.2 in ROMEO mock-up.


Item b:

The CFD analysis did not include a sensitivity study of grid size; however, the modelingmethodology was demonstrated by the following methodology:

1. The model is validated with ROMEO mock-up, which is representative of the U.S. EPRupper internals geometry at -one-fifth scale. This validation relies on comparisons ofexperimental and calculation results of:

- The velocity fields in the hot legs (by means of Particle Interference Velocimetry(PIV)/Laser Doppler Velocimetry (LDV) measurements).

- The thermal mixing in the hot legs by means of temperature measurements.

- The flow distribution in the upper plenum (by means of brine concentrationmeasurements).

2. Other comparisons between the analytical solution and the test results that reinforce thevalidation of the CFD model are:

- The pressure field over the upper support plate, which shows over-pressure in the centerof the plenum and under-pressure near the hot legs, is very consistent between thenumerical model and the tests and demonstrates a good prediction of the suction effect.

- The hydraulic loads measured on CRGA column supports during steady state flowconditions are predicted very well by the numerical models. Based on instrumentedCRGAs, the relative difference between measured and calculated loads is lower than 10percent on average. The maximal load is calculated with a precision of 5 percent.

3. The similarity of the Reynolds number of the mock-up when compared to the numericalmodel validates the numerical model. This validation allows for the reliable numericalmodel at reactor scale.

4. The mesh was specifically refined for the study in hot legs 3 and 4 at reactor scale inorder to assess the flow field in the hot legs. Such a refinement increases the mesh sizeto 15.4 million cells which is sufficient to handle the flow and thermal mixing in the upperplenum.

5. Calculation at the wall boundaries has been checked by post processing the y+parameter in the computational domain (equivalent to a Reynolds number at wall). Therecommended value is reached globally, which provides confidence in the validity of thestandard functions, and subsequently, in the computational results.


Additional NRC Comments (Reference 1)

Several points of clarification for Question 03.09.02-169 are requested.

1. AREVA states on pg 3/10 of the draft response that the validity of the CRGA model wasconfirmed by comparison to the evolution of the piezometric line from the MAGALYtests.

a. AREVA is requested to expand on the definition of the piezometric line.

b. AREVA is requested to explain what the metric for confirmation of the CFD model.What constitutes a sufficiently close agreement?

2. In the response, AREVA states that no grid resolution study was conducted. AREVA isrequested to explain what is meant on page 3/10 by the mesh has been optimized.

3. The definition of y+ is non-standard. The definition as written is simply the Reynoldsnumber based on y. What does AREVA mean by y+ and where does this definitioncome from?

4. On page 3/10, under code convergence, AREVA lists five points in the computationalplane that were selected for convergence studies. AREVA is requested to explain whythese points were selected and why this set is sufficient to determine convergence of thesolver. Are the points in regions of the flow field that are particularly important to theachieving the goals of the analysis or in particularly unsteady flow locations, etc...

5. On page 3/10, AREVA states the convergence criteria for momentum and continuityequations. AREVA is requested to clarify if these criteria are for each time step?

6. On page 4/10, AREVA states that STAR-CD Version 4.02 was used to compute theupper internals. AREVA is requested to clarify if STAR-CD is a commercial solver, whothe manufacturer is, and

7. AREVA is requested to provide locations in the CVAP and FSAR where descriptions ofSTAR-CD are found, or to justify why this software is not discussed in the CVAP andFSAR.

8. AREVA references several experimental mock-ups that are used to benchmark the CFDcode and analysis approach (ROMEO, JULIETTE, CREARE, DUPLEX, HYBISCUS,UPTF, ALAIN AND KATHY). AREVA is requested to identify which of these mockupswere conducted by AREVA as specific to the U.S. EPR, which are programs conductedby AREVA but not necessarily for U.S. EPR and which were conducted by otherorganizations for use as experimental validation data sets.

9. AREVA refers to ROMEO as a 1/5.2 geometric scale model of upper internals. AREVAis requested to explain what the relative Reynold's Number is between the ROMEO andU.S. EPR conditions and at what temperature and pressure the ROMEO experimentalstudy was conducted.

10. In item b-1 of page 5/10, AREVA is requested to explain

a. what constitutes a good comparison between computation. and experiment for thethree sub-bullets provided.

b. In general, how was brine concentration modeled for comparison to the experimentalmock-up.

11. AREVA is requested to explain the source of the suction effect.


12. AREVA is requested to explain further what is meant in item b-3 on page 5/10.

13. AREVA is requested to explain why the CVAP will not be modified to contain at least asubset of this information.

Response to Additional NRC Comments

The MAGALY testing is not pertinent to the description of the CFD model. Therefore, theresponse has been revised to remove reference to the MAGALY tests and piezometric line.The response to NRC questions la and 1 b are provided below.

la. The piezometric line is defined as the axial evolution (i.e., along axis Z) of thepiezometric pressure along CRGA axis:

Ppio = p, + ,gz, with

p, : Static pressure

pg:: Hydrodynamic pressure

The differential piezometric pressure (difference between..ocal and outlet values) is used1 2

and normalized by the dynamic pressure P4 PVF,-2 at the inlet of the CRGA, with2

VFAthe mean velocity through a cross sectional flow area of a fuel assembly.

lb. The piezometric line reflects the distribution of the flow rate among lateral openingscorresponding to gussets orifices, at the bottom of the CRGA, and outlet (i.e., physicalconsistency in piezometric line prediction demonstrated that the numerical modelaccounts for the flow behavior through CRGAs).

Figure 03.09.02-169-5 compares the computed piezometric line to the experimentalmeasurement. Results of this comparison are provided below:

+ Obstruction of guide plates increases the upstream piezometric pressure. Themagnitude of this increase is well predicted which indicates that the head lossesgenerated by geometrical singularities are handled by the model.

* Above the end of lateral openings, downstream of the second guide plate, flowdiversion from CRGA bottom to upper plenum is no longer possible. From thislocation, measured and calculated piezometric pressures are very close as can beobserved in Figure 03.02-69-05; therefore, flow redistribution at the bottom of theCRGA is correctly predicted.

Based on these comparisons, the numerical model of CRGA agrees with the MAGALYresults with respect to the hydraulic behavior at the bottom of the CRGA (i.e., the effectof the guide plates is well predicted by the code). The average deviation betweenexperimental and calculated piezometric pressure is lower than 10 percent of themaximum pressure heterogeneity measured, which confirms the validity of the CRGAmodel.

2. The response was revised to clarify that the mesh has been optimized for computationalpurposes.


3. The "dimensionless wall distance" for boundary layer thickness y+, under focus whenevaluating wall mesh quality, is defined as a Reynolds number based on the distance yto the wall:

Y = + with,

.v: Thickness of the first cells layer

it.: Friction velocity at the wall

This definition is implemented in the CFD code Star-CD, which is used in the presentstudy. It is in agreement with the generic formulation of y+ parameter. It allows forcalculating the wall friction (as well as the attenuation of the velocity profile close to thewalls) based on a generic wall function (i.e., standard logarithmic wall function). Thiswall function is used in CFD codes to avoid too many cells near walls and to allownumerical stability.

4. The numerical sensors are distributed in the computational domain as follows:

* Using three sensors in hot leg 3 at the vicinity of the scoop section.

Four scoops that enable temperature measurement in the hot leg are implanted atthis location. Therefore, the iterative resolution process does not introducenumerical oscillations, which could affect the reliability of the measurements.

Local fluctuations potentially generated by the complex geometry in the upperplenum are supposed to be filtered when entering the hot legs. Thus, even if thefluctuations remain locally in the computational domain, the sensors located in thehot leg (i.e., two sensors in the upper plenum between hot legs 3 and 4) provide arelevant proof of the global convergence of the calculation.

* Using two sensors located in the plenum between hot leg 3 and hot leg 4.

Fluctuations may appear at this location, since the main flow rate is split in two subflow rates. Consequently, these sensors are put in location where there is turbulentand unsteady flow. Therefore, these five sensor locations demonstrate flow fieldconvergence in regions of local unsteadiness and downstream of the upper plenumin hot leg 3 and hot leg 4, respectively.

5. The convergence criteria are for each iterative step.

6. The AREVA NP response to RAI 167, Supplement 3, Question 15.06.05-34 and 37discusses STAR-CD. Also, the NRC evaluated the use of STAR-CD in Section15.6.5.3.4.2 of the SER for U.S. EPR FSAR Tier 2, Chapter 15.

7. See item 6.

8. Only the ROMEO testing is pertinent to this CFD model. The response is revised toremove references to other tests. Mock-ups ROMEO, JULIETTE, ALAIN, and KATHYwere conducted by AREVA NP. Mock-ups CREARE, DUPLEX, HYBISCUS, and UPTFwere conducted by non-AREVA organizations, the results of which are used asexperimental validation data sets.


9. Temperature of water supply is about 68°F; injection temperature is about 140°F forthermal tracing tests. The pressure is maintained at 2 bars (29 psia) in the mock-upfacility. In nominal operating conditions, the ROMEO mock-up is fed with water at500L/s (approximately 793 gpm) and the Reynolds number characterizing the flow in thehot legs is nearly 106. At reactor scale, it reaches nearly 108; thus, the scale factor ofReynolds number between ROMEO and U.S. EPR is 100. The Reynolds number valuein the mock up is large enough to account for the flow behavior occurring at reactorscale. The Reynolds number similarity has also been verified experimentally, whichvalidates the representativeness of the mock-up.

10a. Validation with regards to:

* Velocity fields in the hot legs

Comparison between experimental PIV/LDV and calculation results is based onmacroscopic observations of the velocity field in the hot legs. The flow patternmatches and is in agreement regarding its evolution along the pipe.

Figure 03.09.02-169-6 compares the calculation results with the experimentalmeasurements. The scale adopted is 0.2 m/s (0.66 ft/s). The blue zones stand forlower velocities, near Om/s, whereas the red ones correspond to higher velocities,near 2m/s (6.6.ft/s).

* Thermal mixing in the hot legs

For comparison with thermal tracing tests, the temperature distribution is simulated.Figure 03.09.02-169-7 presents the comparison of the normalized calculatedtemperatures with the measurements in hot leg 4 for two thermal tracing tests (M6orifice and R4 orifice) in the first and the last cross section. Tests results arerepresented by red circles and calculation results are represented by full blue circles.The diameter is directly proportional to the corresponding value.

Qualitatively and quantitatively, computational results are in good agreement with thetest data based on the following:

- For each traced orifices, the location of hot spots calculated is consistent with thetest results.

- CFD calculations enable prediction of the correct location of the hot spot and itsevolution along the pipe.

- Average errors between tests and CFD calculation are lower than thethermocouples uncertainty measurement (±0.50C).

* Flow distribution in the upper plenum

Figure 03.09.02-169-8 provides a comparison between tests and calculations of thebrine distribution in a well-balanced mode. Tests are blue colored; calculations arered colored. Calculated distributions are correlated with the tests within 3.5 percent.

1Ob. The brine is introduced into the mockup and its concentration in the mock-up ismeasured. These measurements are used to identify the flow patterns. The flowpatterns are compared to the flow patterns in the numerical model to provide adequatevalidation. This process is similar to a dye tracer test.


11. The acceleration of the fluid at the vicinity of the hot leg nozzles is due to the pumps andthe outlet conditions in the calculation. This so-called "suction effect" exists in the UCP,with an over-pressure at the center where velocity is lower, and under-pressure at theperiphery close to the hot leg nozzles, where fluid velocity is higher.

12. See revised response to item b.3.

13. ANP-10306 will be revised to reflect this RAI response.

References:

1. NRC e-mail dated May 7, 2012, Getachew Tesfaye (NRC) to Dennis Williford (AREVA NP),"Comments on Draft Response to Question 3.9.2-169."

FSAR Impact:

The U.S. EPR FSAR will not be changed as a result of this question.

Technical Report Impact:

ANP-10306P, "Comprehensive Vibration Assessment Program for U.S. EPR Reactor InternalsTechnical Report," Revision 0 will be revised as described in the response and shown in theattached markup.


Figure 03.09.02-169-1 U.S. EPR CFD MODEL


Figure 03.09.02-169-2 U.S. EPR CFD MODEL MESH


Figure 03.09.02-169-3 CFD MODEL MESH OF THE U.S. EPR UCP ORIFICES


Figure 03.09.02-169-4 SIMPLIFIED SCHEME OF THE ROMEO MOCK-UP


Figure 03.09.02-169-5 Comparison between the Computed Piezometric Line to the"- Experimental Measurement


Figure 03.09.02-169-6 Comparison of the Calculation Results with the ExperimentalMeasurement


Figure 03.09.02-169-7 Comparison of the Normalized Calculated Temperatures with theMeasurements in HL4


Figure 03.09.02-169-8 Comparison between Tests and Calculations of theBrine Distribution

AN P-10306PComprehensive VibrationAssessment Program for

U.S. EPR ReactorInternalsTechnical Report

Markups

AREVA NP Inc.

Comprehensive Vibration Assessment Program for U.S. EPR Reactor InternalsTechnical Report

ANP-10306NPRevision 1

Page 4-142

Acceptance Criteria for High Cycle Fatigue

The ASME fatigue curve "AC" shown in Figure 4-21 is applied to the off Alck in the response of

the structure. The allowable high cycle fatigue stress of [ ] (0-peak) at 1011

cycles for fatigue curve "AC."

4.5.3 Response of the Column Supports

The primary fluid velocity and density throughout the RV upper internals is determined with a-three dimensional CFD model onc dimensinal thcrmal hydraulic m odc! for the full power normaloperating condition.

CFD Model Details

The CFD simulation represents the upper part of the U.S. EPR vessel as shown in Figure 4-72-U.S. EPR CFD Model. Its computational domain is limited by the UCP, the USP, and the end ofthe hot legs. It is composed of:

" The UCP, with its orifices of three different types.

" The upper plenum, with its different elements.

" The outlet nozzles and the four hot legs, which are 10 diameters lonq (so that outletconditions are far enough from the plenum).

Inside the plenum, the CRGAs. the first two guide plates as well as the lateral openings (towardthe plenum volume) are modeled. The column internal components (tie rods, continuous guides,split tubes and rods) are not represented because they do not have any impact on the flowpatterns around the columns. The flow is blocked slightly above the second guide plate (in boththe plant design and mock-up) as most of the flow goes outside, passing through the gussets.The support columns have also been represented.

Code Convergence

To demonstrate code convergence, five points were chosen in the computational domain:

* Three points located in the hot leg 3 near the scoop section.

" Two points located in the plenum between hot leg 3 and hot leg 4 (where there may be strongfluctuations).

The local velocity components (in m/s) as well as the temperature (in "K) have been monitoredduring the iterations for each probe, and their evolutions were plotted. The convergence criteriawere fixed at 1 OA for the momentum equation and 10-3 for the continuity equation, and wereachieved at 6200 iterations.

•Question 3.9.2-1 691

AREVA NP Inc. ANP-10306NPRevision 1

Comprehensive Vibration Assessment Program for U.S. EPR Reactor InternalsTechnical Report Page 4-143

ROMEO Test Details

The qoal of the ROMEO tests is to characterize the upper plenum hydraulics of the U.S. EPR

reactor. The ROMEO mock-up represents the upper part of the EPR vessel at a scale of (1/5.2).

The simulated part is located between the UCP, the USP, and the end of the hot legs.

The core effect is taken into account by additive head losses through the UCP. The CRGAs,

columns, and aeroballs lance finqers are geometrically represented.

The UCP of the mock-up consists of two plates separated by 241 ventures (of which aim is to

measure flow rates through the UCP). Each plate is drilled with 241 holes and diaphragms areinserted in the upstream holes. These diaphragms are of three types, ORGA, column, and freeorifice, as there are three different outlet groups. The diameters of the diaphragms were

determined to represent the upstream effect of the core on flow distribution. The method used is

the following:

" Realization of a computation of the reactor upper plenum hydraulics with simulation of fuelassemblies, using porous media and singular head losses. The length of represented core isabout one meter or 39.37 inches.

" Realization of a computation of the reactor upper plenum hydraulics with an empty corecavity upstream UCP (with representation of UCP holes, but without representation of fuelbundles).

" Determination of additive head loss coefficient to put into UCP holes in the empty cavitycomputation in order to obtain the same flow rate distribution as the one obtained withsimulation of fuel assemblies.

The mock-up has been designed to obtain a uniform pressure distribution upstream the UCP. In

this way, the UCP flow field is only created by downstream effects and in particular by the hot leg

suction effect.

CFD Analysis Validation Procedure

The CFD analysis of the U.S. EPR upper internals is performed using the AREVA computer

program "STAR-CD" Version 4.02. The results of the computer runs performed for the validation

of STAR-CD are compared with analytical or semi-analytical solutions and also with

measurements obtained from experiments where both the integral and separate effects were

verified.

The validation includes the comparisons of the results obtained with the computations and

experiments that were specifically dedicated to be representative of physical phenomena

occurring in a pressurized reactor vessel. These experiments include U.S. EPR upper internals

hydraulics, represented at scale 1/5.2 in ROMEO mock-up.

The CFD analysis did not include a sensitivity study of grid size: however, the modeling

methodology was demonstrated by the following methodology:

•-Question 3.9.2-1691

AREVA NP Inc. lQuestion 3.9.2-169 ANP-10306NPRevision 1

Comprehensive Vibration Assessment Program for U.S. EPR Reactor InternalsRo

Technical Report Page 4-144

1. The model is validated with ROMEO mock-up, which is representative of the U.S. EPR upperinternals geometry at -one-fifth scale. This validation relies on comparisons of experimentaland calculation results of:

- The velocity fields in the hot legs (by means of Particle Interference Velocimetry (PIV)/Laser Doppler Velocimetry (LDV) measurements).

- The thermal mixing in the hot leas by means of temperature measurements.

- The flow distribution in the upper plenum (by means of brine concentrationmeasurements).

2. Other comparisons that are performed between the analytical solution and the test resultsthat reinforce the validation of the CFD model are:

- The pressure field over the upper support plate, which shows overpressure in the centerof the plenum and under pressure near the hot legs, is very consistent between thenumerical model and the tests and thus demonstrates a good prediction of the suctioneffect.

- The hydraulic loads measured on CRGA column supports during steady state flowconditions are predicted very well by the numerical models.

3. The similarity of the Reynolds number of the mock-up when compared to the numericalmodel validates the numerical model. This validation allows for the reliable numerical modelat reactor scale.

4. The mesh was specifically refined for the study in hot legs 3 and 4 at reactor scale in order toassess a better calculation of the flow phenomenology in the hot legs which were underfocus. Such a refinement increases the mesh size to 15.4 million cells which is sufficient tohandle the mixinq phenomenology in the upper plenum according to our engineeringfeedback.

5. Calculation at the wall boundaries has been checked by post processing the y+ parameter inthe computational domain (equivalent to a Reynolds number at wall). The recommendedvalue is reached globally, which provides confidence in the validity of the standard functions,and subsequently, in the computational results.

The U.S. EPR upper plenum thermal hydraulic model is qualified based on benchmarking withflow tests performed with the ROMEO mock-up which simulates the upper internals at a scale of1:5.2. The magnitude of vclecity and density thatis evaluated for the columnupport• ... is

rcprcscntativc of a location ncar the hot leg outlet nozzles (V= +±-]fusee-eftd-p-

=--- -] W-.-T-sThe worst located column support is evaluated. As noted in

Section 4.5.1, the cross flow velocity distribution along the length of the column supports isincreased by 10 percent to account for the RCP over-speed transient conditions. The distributionof vclocity at the hot leg nozzle •. l.vtion is pr-jctcd upon tho c....sponding floW arca of thecolumnR SUPPorts to providoe a velocity pro~fllc for the analysis.

Sincc the time at which the fudll scale @anays~is of the col)umnR supports wero orFiginally cva uated,the velecit' profile along the length of the cclumn supports has becn rcfincd using throc-dimensional thcrmal hydFaulic . F.,dling and analysis of the flew distribu.tion in the upp.. plcnum.

AREVA plans to rcvise the FIV analysis of these columnR supports to sonsidcrthe rofined vcleeity

AREVA NP Inc.

Comprehensive Vibration Assessment Program for U.S. EPR Reactor internalsTechnical Report

ANP-10306NPRevision 1

Page 4-222

Figure 4-72-U.S. EPR CFD Model

•k-Question 3.9.2-1 69]

Documents

Question # -I · The mesh is presented in Figure 03.09.02-169-2, and its three types of orifices in Figure 03.09.02-169-3. This mesh has been optimized for computational purposes;