DRAFT: Thermal-hydraulics and Conjugate Heat Transfer Calculation in a Wire-Wrapped SFR Assembly

Similar documents
Thermal-hydraulic simulations of a wire spacer fuel assembly

Application of computational fluid dynamics codes for nuclear reactor design

RECONSTRUCTION OF TURBULENT FLUCTUATIONS FOR HYBRID RANS/LES SIMULATIONS USING A SYNTHETIC-EDDY METHOD

Lectures on Applied Reactor Technology and Nuclear Power Safety. Lecture No 6

Turbulent Boundary Layers & Turbulence Models. Lecture 09

EasyChair Preprint. Numerical Simulation of Fluid Flow and Heat Transfer of the Supercritical Water in Different Fuel Rod Channels

International Journal of Scientific & Engineering Research, Volume 6, Issue 5, May ISSN

Numerical Simulation of the MYRRHA reactor: development of the appropriate flow solver Dr. Lilla Koloszár, Philippe Planquart

CHAPTER 7 NUMERICAL MODELLING OF A SPIRAL HEAT EXCHANGER USING CFD TECHNIQUE

COMPUTATIONAL FLUID DYNAMICS ANALYSIS OF A V-RIB WITH GAP ROUGHENED SOLAR AIR HEATER

Laminar flow heat transfer studies in a twisted square duct for constant wall heat flux boundary condition

MODELLING TURBULENT HEAT FLUXES USING THE ELLIPTIC BLENDING APPROACH FOR NATURAL CONVECTION

Simplified Model of WWER-440 Fuel Assembly for ThermoHydraulic Analysis

Lecture 30 Review of Fluid Flow and Heat Transfer

Principles of Convection

Numerical simulations of heat transfer in plane channel flow

PREDICTION OF MASS FLOW RATE AND PRESSURE DROP IN THE COOLANT CHANNEL OF THE TRIGA 2000 REACTOR CORE

Explicit algebraic Reynolds stress models for internal flows

Studies on flow through and around a porous permeable sphere: II. Heat Transfer

Available online at ScienceDirect. Procedia Engineering 90 (2014 )

Validation of MARS-LMR Code for Heat Transfer Models in the DHRS of the PGSFR

Computation of turbulent natural convection with buoyancy corrected second moment closure models

Status and Future Challenges of CFD for Liquid Metal Cooled Reactors

CFD STUDIES IN THE PREDICTION OF THERMAL STRIPING IN AN LMFBR

CFD SIMULATION OF SWIRL FLOW IN HEXAGONAL ROD BUNDLE GEOMETRY BY SPLIT MIXING VANE GRID SPACERS. Mohammad NAZIFIFARD

Unsteady RANS and LES Analyses of Hooper s Hydraulics Experiment in a Tight Lattice Bare Rod-bundle

Wall treatments and wall functions

SENSITIVITY ANALYSIS FOR ULOF OF PGSFR

CFD Analysis for Thermal Behavior of Turbulent Channel Flow of Different Geometry of Bottom Plate

Problem 4.3. Problem 4.4

Analysis of Heat Transfer in Pipe with Twisted Tape Inserts

CFD SIMULATIONS OF THE SPENT FUEL POOL IN THE LOSS OF COOLANT ACCIDENT

Explicit algebraic Reynolds stress models for boundary layer flows

THERMAL HYDRAULIC REACTOR CORE CALCULATIONS BASED ON COUPLING THE CFD CODE ANSYS CFX WITH THE 3D NEUTRON KINETIC CORE MODEL DYN3D

Computation of Unsteady Flows With Moving Grids

Experimental investigation of the pressure loss characteristics of the full-scale MYRRHA fuel bundle in the COMPLOT LBE facility

Three-Dimensional Simulation of a Simplified Advanced Gas-cooled Reactor Fuel Element

The mean shear stress has both viscous and turbulent parts. In simple shear (i.e. U / y the only non-zero mean gradient):

EFFECT OF DISTRIBUTION OF VOLUMETRIC HEAT GENERATION ON MODERATOR TEMPERATURE DISTRIBUTION

There are no simple turbulent flows

Theoretical Study of Forced Convective Heat Transfer in Hexagonal Configuration with 7Rod Bundles Using Zirconia-water Nanofluid

On the transient modelling of impinging jets heat transfer. A practical approach

UNIT II CONVECTION HEAT TRANSFER

NUMERICAL SIMULATION OF FLOW THROUGH TURBINE BLADE INTERNAL COOLING CHANNEL USING COMSOL MULTIPHYSICS

CFD analysis of the transient flow in a low-oil concentration hydrocyclone

Model Studies on Slag-Metal Entrainment in Gas Stirred Ladles

Comparison of two equations closure turbulence models for the prediction of heat and mass transfer in a mechanically ventilated enclosure

Numerical investigation of swirl flow inside a supersonic nozzle

Computational and Experimental Studies of Fluid flow and Heat Transfer in a Calandria Based Reactor

Turbulence Model Affect on Heat Exchange Characteristics Through the Beam Window for European Spallation Source

Calculations on a heated cylinder case

The effect of geometric parameters on the head loss factor in headers

COMPUTATIONAL SIMULATION OF THE FLOW PAST AN AIRFOIL FOR AN UNMANNED AERIAL VEHICLE

ENERGY PERFORMANCE IMPROVEMENT, FLOW BEHAVIOR AND HEAT TRANSFER INVESTIGATION IN A CIRCULAR TUBE WITH V-DOWNSTREAM DISCRETE BAFFLES

CHAPTER 7 SEVERAL FORMS OF THE EQUATIONS OF MOTION

WALL RESOLVED LARGE EDDY SIMULATION OF A FLOW THROUGH A SQUARE-EDGED ORIFICE IN A ROUND PIPE AT RE=25000

INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 5, ISSUE 09, SEPTEMBER 2016 ISSN

CONVECTIVE HEAT TRANSFER

In order to optimize the shell and coil heat exchanger design using the model presented in Chapter

Manhar Dhanak Florida Atlantic University Graduate Student: Zaqie Reza

CFD Analysis of Forced Convection Flow and Heat Transfer in Semi-Circular Cross-Sectioned Micro-Channel

Heat Transfer from An Impingement Jet onto A Heated Half-Prolate Spheroid Attached to A Heated Flat Plate

Advanced Simulation: applications for fast reactors

Uncertainty quantification for RANS simulation of flow over a wavy wall

DNS STUDY OF TURBULENT HEAT TRANSFER IN A SPANWISE ROTATING SQUARE DUCT

Overview of Turbulent Reacting Flows

AN UNCERTAINTY ESTIMATION EXAMPLE FOR BACKWARD FACING STEP CFD SIMULATION. Abstract

HEAT TRANSFER IN A RECIRCULATION ZONE AT STEADY-STATE AND OSCILLATING CONDITIONS - THE BACK FACING STEP TEST CASE

CFD ANANLYSIS OF THE MATIS-H EXPERIMENTS ON THE TURBULENT FLOW STRUCTURES IN A 5x5 ROD BUNDLE WITH MIXING DEVICES

NUMERICAL SIMULATION OF LDI COMBUSTOR WITH DISCRETE-JET SWIRLERS USING RE-STRESS MODEL IN THE KIVA CODE

Helical Coil Flow: a Case Study

Wall Effects in Convective Heat Transfer from a Sphere to Power Law Fluids in Tubes

Due Tuesday, November 23 nd, 12:00 midnight

CFD in Heat Transfer Equipment Professor Bengt Sunden Division of Heat Transfer Department of Energy Sciences Lund University

CONVECTIVE HEAT TRANSFER

SIMULATION OF THERMAL CHARACTERISTICS OF RADIATORS USING A POROUS MODEL. YETSAN Auto Radiator Co. Inc Çorum, Turkey NOMENCLATURE

Numerical Investigation of Thermal Performance in Cross Flow Around Square Array of Circular Cylinders

Simulation and improvement of the ventilation of a welding workshop using a Finite volume scheme code

Publication 97/2. An Introduction to Turbulence Models. Lars Davidson, lada

THERMAL ANALYSIS OF A SPENT FUEL TRANSPORTATION CASK

AT A LOCAL SCALE. Technical University of Lisbon, TU Lisbon, Av. Rovisco Pais, Lisboa, Portugal 2

CFD Simulation of Sodium Boiling in Heated Pipe using RPI Model

CFD SIMULATION OF THE DEPARTURE FROM NUCLEATE BOILING

Computation of turbulent Prandtl number for mixed convection around a heated cylinder

Contents. I Introduction 1. Preface. xiii

Numerical Heat Transfer Study of Turbulent Square Duct Flow through W-Type Turbulators

Numerical Prediction Of Torque On Guide Vanes In A Reversible Pump-Turbine

Comparison of Turbulence Models in the Flow over a Backward-Facing Step Priscila Pires Araujo 1, André Luiz Tenório Rezende 2

PERFORMANCE SCREENING OF A LOUVERED FIN AND VORTEX GENERATOR COMBINATION

Turbulence Modeling I!

Numerical simulation of fluid flow in a monolithic exchanger related to high temperature and high pressure operating conditions

DEVELOPMENT OF CFD MODEL FOR A SWIRL STABILIZED SPRAY COMBUSTOR

Fluid Flow and Heat Transfer Characteristics in Helical Tubes Cooperating with Spiral Corrugation

NATURAL CONVECTION HEAT TRANSFER CHARACTERISTICS OF KUR FUEL ASSEMBLY DURING LOSS OF COOLANT ACCIDENT

Thermal Dispersion and Convection Heat Transfer during Laminar Transient Flow in Porous Media

Detailed Outline, M E 320 Fluid Flow, Spring Semester 2015

Study of Forced and Free convection in Lid driven cavity problem

Process Chemistry Toolbox - Mixing

2.3 The Turbulent Flat Plate Boundary Layer

Numerical Heat and Mass Transfer

Transcription:

DRAFT: Thermal-hydraulics and Conjugate Heat Transfer Calculation in a Wire-Wrapped SFR Assembly C. Péniguel, I. Rupp EDF R&D, 6 quai Watier 78401 Chatou Cedex, France Email:Christophe.peniguel@edf.fr S. Rolfo School of MACE, The University of Manchester, Manchester M60 1QD, UK M. Guillaud INCKA, 85 Avenue Pierre Grenier, 92100 Boulogne-Billancourt, France Abstract Fast reactors with liquid metal coolant have recently received a renewed interest owing to a more efficient usage of the primary uranium resources, and they are one of the proposal for the next Generation IV. In order to evaluate nuclear power plant design and safety, 3D analysis of the flow and heat transfer in a wire spacer fuel assembly are ongoing at EDF. The introduction of the wire wrapped spacers, helically wound along the pin axis, enhances the mixing of the coolant between sub-channels and prevents contact between the fuel pins. The mesh generation step constitutes a challenging task if a reasonable amount of cells in conjunction with a suitable spatial discretization is wanted, especially if in the near future, industrial cases with up to 271 pins needs to be tackled as shown in this paper. Quite complex global flow patterns are found using either k-ε or preferably Reynolds Stress turbulent models with a strong influence of the number of pins. Global parameters like friction factor or Nusselt number are compared against experimental correlations. Likewise exploratory conjugated heat transfer calculations using a coupling between the finite element thermal code SYRTHES and the finite volume CFD code Code_Saturne are also shown. I. INTRODUCTION In France, fast reactors with liquid metal coolant have recently received a renewed interest due to a more efficient usage of the primary uranium resources, and they are one of the proposals for the Generation IV reactors. In order to evaluate nuclear plant design and safety, three-dimensional numerical studies are on going at EDF. Fuel bundles of fast reactor are arranged into a triangular configuration and pins are wrapped with wire spacer, which follows a helically pattern around the rod axis. The primary reason of the wire is to avoid collision between adjacent pins. Moreover the presence of the wire is also reducing vibrations and avoiding the trapping of the liquid metal coolant (in general sodium). From the thermal-hydraulic point of view the wire is creating a very complex secondary motion enhancing mixing between subchannels. From a historical point of view the effect of the wire via experimental correlations, which provide the friction factor as function of geometrical and hydraulic parameters. A first example is provided by Novendstern 14, where the usual Blasius formula for pipe flow is corrected taking into account several parameters like the number and the hydraulic diameter of the different type of sub-channels which can be found in the fuel assembly. Another famous correlation was given by Rehme 18 where a shape factor F, which take into account the pitch-over-diameter ratio P/D and the helix-over-diameter ration H/D, is introduced. A milestone in the experimental evaluation of this type of flow was presented by Cheng and Todreas 4 were two sets of correlations were presented. The detailed version are taking into account several geometrical parameters and different hydraulic parameters, making the correlations suitable for many configurations, but also very difficult to use. A simplified version was also presented and the two version are converging toward same values as the Reynolds number increase. Because the interest of this work is toward fully turbulent, relatively high Re number

only the simplified version is considered. Last correlation, which will be used herein, was presented by Engel et al. 5. In this work a modified version by Bubelis and Schikorr 2 is used. This last work present a very wide and methodic comparison of several well accepted correlations with some available experimental and numerical data. This paper is a perfect introduction of one of the key issue encountered during this work: which is the accuracy of the results? Experimental correlations have validity ranges depending on the experiments used for the definition. Pure experimental data are more reliable but very scarce. In this contest CFD can play a role in supplying a vast and very specific amount of data. An example is again present in Bubelis and Schikorr 2 were experimental correlations are compared with RANS (Reynolds Average Navier-Stokes equations) turbulence models from Gajapathy et al 8. Now a problem arises in Gajapathy et al 8 the CFD results are validated against correlations finding good agreement and in Bubelis and Schikorr 2 the same correlations are validated against the CFD: but now which are the data to be trusted and used as reference? A possible solution could be provided by LES (Large Eddy Simulation) and DNS (Direct Numerical Simulation) as the one presented by 7. LES and DNS are able to provide a very broad and very accurate, if a proper code is used, amount of data, which can be very difficult to obtain with experimental techniques. Instantaneous flow field and extensive average results (for example Nusselt distribution along the fuel rod) will be available, making more rigorous the validation of RANS models. Because of the extremely time consuming and cost (very powerful High Parallel Computing, HPC, facilities are necessary) of LES and DNS, they are still limited to reduced geometry and moderate Reynolds number, whereas usual RANS will be devoted to the study of more industrial cases. Some RANS study are also starting to appear like Raza and Kim 17 and Smith et al 22. In general those type of paper present studies carried out with commercial unstructured codes, employing reduced geometries with a limited number of pins and results compared with well establish experimental correlations. A even more difficult task is to find heat transfer correlations for the evaluation of the Nusselt number. Several studies were conducted during the sixty and the seventy for several projects. Pfrang and Struwe 15 presents a review of the outcome of several studies. In this work the Nusselt correlations of the EUTATOM project 9, of Karim and Carelli 12 and Mikityuk 13. The aim of this work is firstly to investigate the ability of Code_Saturne to study these types of flow, underlining which are the important parameters to take into consideration (mesh configuration, turbulence models, etc ). This validation is carried out using reduced geometries composed by only seven and nineteen pins with only one helix pass. Consequently periodicity is used in the streamwise direction, in order to reduce further more the domain. Global results of pressure drop and global Nusselt number are compared with the previous mentioned correlations for both friction factor and Nusselt number. A second kind of objective, more industrially related, has also begun and is briefly discussed. The first aspect concerns the feasibility of applying the CFD approach to real configurations counting up to 271 pins. This leads to huge meshes for which HPC aspects are required. The second aspects which needs to be tackled concerns the coupling of Code_Saturne with the thermal code SYRTHES, in order to study the heat deposit in the solid part of the domain. In this case fuel an inlet/outlet problem formulation need to be employed and the computational domain requires several helix passes. For both aims the mesh generation step constitutes a fairly challenging task. The wire induces a very large number of singularities in the geometry due to the fact that the wire attached to each pin is almost in contact with the surrounding pins. After several attempts with commercial mesh generators, the homemade procedure described in Peniguel et al 16 is followed. Several variants regarding the way to handle the connection between the wire and the pin have been investigated. It leads to an almost structured mesh with a very good control on the number of cells across two adjacent pins. Here all meshes used have at least 8 cells between two pins. II. NUMERICAL METHODS In this simulation, conduction, convection heat transfer must be solved simultaneously. The model used and equations solved are described thereafter. II A. Code_Saturne In this study the CFD code Code_Saturne 1, developed by EDF. The code is able to solve steady or transient, single phase, incompressible, laminar or turbulent flows. The Reynolds Average Navier-Stokes equations are discretized using a co-located finite volume approach. Velocity and pressure coupling is insured by a prediction/correction method with a SIMPLEC algorithm 6 and the Poisson equation is solved with a conjugate gradient method. A Rhie and Chow interpolation 19 in the correction step is also used in order to stabilize the solution. The code is completely unstructured and able to handle any type of cell including polyhedral and embedded refinements. Since 2006 Code_Saturne has become open source and a very extensive work of Validation and Verification (V&V) has been carried out. A very good overview about V&V and why it is extremely important in nuclear engineering is givem by Chabard and Laurence 3

Governing equations for periodic calculations. In a fully developed turbulent flow pressure and temperature can be decomposed as and with, x 3 being the periodic direction. The terms and represent respectively the pressure drop and the rise of enthalpy in the periodic direction. The equation can then be written as: The parameter is recomputed at every time step in order to keep constant the mass flow rate, from a simple energy balance: (1) (2) is instead obtained where is the wall heat flux, is the wall heat surface and the mass flow rate. Turbulence modeling. Two different turbulence models have been used: the twoequation model of Jones and Launder 11 and the Reynolds Stress model of Speziale, Sarkar and Gatski (SSG) 23. In the first case, two additional transport equations, one for the turbulent kinetic energy k and the second for the rate of dissipation ε, are resolved and used to compute the turbulent viscosity as: where (4) (3) is a constant equal to 0.09. One reminds that the value of the constant has been tuned in order to match experimental results in simple cases. This model has many advantages from the user point of view: it is stable and easy to implement. The model is quite accurate in a wide range of simple flows, but for more complex cases it might be not accurate enough because of its inability to take into account the anisotropy of the Reynolds stress tensor. In the SSG the closure is obtained by solving a transport equation for every component of the Reynolds Stress Tensor. The equations assume the following formulation: Where is the viscous diffusion, is the turbulent diffusion, is the pressure strain rate correlation term, is the production term and is the turbulent dissipation rate term. The turbulent dissipation, the pressure strain rate correlation and the third order correlations in the turbulent diffusion require further modelling. In the SSG model a non-linear quadratic model is used for pressure strain rate correlation. The wall modeling is based on the so-called scalable wall functions 10. The main disadvantage of the standard wall function, apart the non-general validity of the log law, is the difficulty to place the in a specific range ( ), as required for a good wall modeling. This problem is even more challenging when the geometry is quite complex and characterized by a huge variation of the wall distance. In the scalable wall function the minimum value is limited to a value of 11.06, so the value of the velocity gradient at the first cell will be the same as if it were at the edge of the viscous sub-layer. The turbulent heat fluxes are modeled using a gradient diffusion hypothesis and the eddy diffusivity employ a turbulent Prandtl number analogy: II.B Heat Transfer at the Fluid/Solid Interface At the interface, every time step, the thermal coupling is performed. Let T s be the temperature of an internal solid node, T w the temperature at a node that belongs to the interface, and T f the temperature of a fluid point (located generally in the log layer). At time t (n), the CFD tool Code_Saturne provides after calculation: h (n) : the local heat exhange coefficient at time t (n) T f (n) : the local inside fluid temperature at time t (n) using these data, the flux to be applied to the solid is : Then, using this flux or the exchange conditions h (n) (calculated by a log law) and T f (n), SYRTHES 21 can solve the heat conduction equation inside the solid. At the same time, a similar procedure is used in the fluid to update the fluid temperature field. (6) (8) (5)

III PERIODIC FLUID CALCULATIONS. Periodic boundary conditions are in general used for refined turbulent calculation in order to reduce the domain and keep the mesh size reasonable. The Reynolds number is varied between 5000 and 50000 (based on the hydraulic diameter and bulk velocity). The wall heat flux is constant and equal to 6x105 W/m2 which makes the Peclet number ( ) ranging from 25 till 400. Two different helix-to-diameter ratios H/D are used, which take the value of 22 and 17.As expected, the flow field presents no symmetry in the plane perpendicular to the stream-wise direction. The presence of the wire is inducing a global swirling motion on the edge and corner sub-channels. Moreover the location of the maximum of the stream wise velocity is rotating following the pattern of the helices. In all wall channels, characterized by a high axial velocity, a big secondary vortex is also visible. It is interesting to notice that the wall sub-channel, characterized by the big bulk velocity, has also the maximum of the swirl flow velocity. On the opposite wall sub-channels, where the velocity is lower, this secondary structure is not so obvious. Central sub-channels are characterized by a secondary vortex for all the wire angles. For those subchannels the unbalance of the stream-wise velocity, typical of the wall sub-channels, doesn t appear clearly. Fig. 1. Velocity in the streamwise direction and secondary flow for seven pin configuration P/D = 1.1 H/D = 22 and Re=10000. Results obtained with a k-ε model Fig. 1 and Fig. 2 present the comparison between mean streamwise direction velocity and secondary motion for the two turbulence models employed herein. Both flows look very similar even in their quantitative comparison. This could lead us to the wrong conclusion that the turbulence model has a minor effect on the results. As a matter of fact if the velocity at the wall is compared (Fig. 3 and Fig. 4) the two models are showing quite a substantial difference, making the estimation of the pressure drop quite different. Fig. 2. Velocity in the streamwise direction and secondary flow for seven pin configuration P/D = 1.1 H/D = 22 and Re=10000. Results obtained with a Rij model. Fig. 3. Contours of the norm of the velocity at the wall for P/D = 1.1, H/D = 22 and Re=10000. Results obtained with a k-ε model. Fig. 4. Contours of the norm of the velocity at the wall for P/D = 1.1 H/D = 22 and Re=10000. Results obtained with a Rij model. The variation of the helix pass has a great influence on the solution as can be appreciated in Fig. 5, where results are obtained using a k-ε model. The helix-to-diameter ratio H/D = 22 correspond to a new EDF design, whereas H/D = 17 is the old design used in the SuperPhenix reactor. The flow features are still the same, but a higher velocity and a stronger secondary motion can be appreciated. If now the now the k-ε is compared with the second moment closure Rij of Fig 8, a very large difference can be appreciated. The k-ε model is giving an almost 20% underestimation of the maximum axial velocity respect to the Rij, and also the secondary motion is less strong.

Fig. 5. Velocity in the streamwise direction and secondary flow for seven pin configuration P/D = 1.1 H/D = 17 and Re=10000. Results obtained with a k-ε model. Fig. 6. Velocity in the streamwise direction and secondary flow for seven pin configuration P/D = 1.1 H/D = 17 and Re=10000. Results obtained with a Rij model. As the number of pin increase the future of the flow does not change as can be establish from Fig. 7. It is interesting to notice that as the pins number increase the edge and corner sub-channels start to loose their predominance and the flow field is more homogeneous. Fig. 8. Dimensionless temperature field for seven pins configuration P/D = 1.1 H/D = 22 and Pe=55. Results obtained with a Rij model. The temperature field has also a very complex pattern as can be seen from Fig. 8. On the fuel rod surfaces a simplified boundary condition (constant temperature) is imposed (Dirichlet BC), whereas walls of the external case are adiabatic (Neumann BC with wall normal gradient equal to zero). In reality of course, as explained in the last part of the paper, only the pin itself contains fuel and therefore heat deposit. The heat flux is taken into account using a sink term into the temperature equation as described in eq. (2). Also in this case the field can be divided into central and side sub-channels. The first-ones are characterized by higher temperatures and in particular in the gap region between two adjacent pins and by a quite uniform temperature distribution. Side channels are more influenced by the location of the streamwise velocity maximum, which correspond to the temperature minimum. The influence of the turbulence model in the heat transfer is very limited probably due to the use of scalable wall functions. Movies about the results presented in this sections can be 20 found in. Friction factor and Nusselt profiles. In order to compare the results with available and wellaccepted correlations global parameters have to be evaluated. The friction factor f is defined as: (9) Fig. 7. Velocity in the streamwise direction and secondary flow for nineteen pins configuration P/D = 1.1 H/D = 22 and Re=10000. Results obtained with a Rij model. where is the hydraulic diameter the periodic length in the streamwise direction the bulk velocity and the pressure drop. The evaluation of the pressure drop presents a very delicate situation because several option are available and the results can vary considerably. For example the pressure drop can be evaluated from the wall shear as: being the wall shear and the total wall surface. The wall shear can be estimated from the friction velocity given by the wall function as:

(10) with I being the intersection between the wall normal direction through the wall face centre F and the projection of the cell centre on that line. K and C are two constant in equal to 0.42 and 5.2 respectively. The wall dimensionless distance is evaluated from a turbulent velocity as In general the Reynolds stress model is giving higher value of f respect to the eddy viscosity model, but value tend to converge as the Reynolds number increase. Indeed the same effect can be seen as the number of pins increase because of the minor importance of the edge channels on the global flow field. The data are in the range given by the experimental correlations and they seems to agree better with the one of Cheng and Todreas 4. (11) where d is the distance between I and F. Another possibility could be to use the imposed pressure drop β of eq. (1) obtaining: Δ p = βl z (12) Fig. 11. Nusselt number 7 pins configuration, P/D = 1.1 H/D = 22. The Nusselt number is instead evaluated as: Fig. 9. Friction factor f for the seven pins configuration, P/D = 1.1 H/D = 22. In the first case also pressure redistribution on the cross section are taken into account. The value are quite different, in particular at low Reynolds number, as can be seen in Fig. 9 for the seven pins bundle and in Fig. 10 for the nineteen. The method of eq. (10) is labelled as, whereas the one of eq. (12) as QDM. The comparison against the experimental correlation is plotted in Fig. 15 for the seven pins configuration. In this case the difference between the two turbulence models is almost negligible, but slightly increase with the Peclet number. The reason could be the wall treatment; consequently more accurate modeling is required. CFD profiles are increasing with a relatively steep slope. On the other hand the experimental correlations have almost the same and not very rapid grade of growing, although they have different starting points. Effect of the mesh configuration. Despite the fact that the geometry can be described using few geometrical parameter, the model is quite difficult and very difficult to mesh as described in Peniguel et al. 16. The wire is introducing in the geometry many singularities that are skewing the mesh. Now an obvious question arises: with which grade of accuracy the mesh has to reproduce the geometry? Can a simply geometry still produce the same flow features and accurate global flow parameters? Fig. 10. Friction factor f for the nineteen pins configuration, P/D = 1.1 H/D = 22.

However this conclusion is valid mainly for global dynamic aspects. It is likely that from the local thermal point of view the hybrid approach and its better geometry approximation may have an influence that needs to be investigated. Fig. 12. Base mesh used for the study. In blue the mesh for the fluid, others colors represent the different solid part of the pin Fig. 14. Comparison of the axial velocity for the configuration P/D = 1.1 H/D = 22 and Re=25000. Results obtained with k-ε model. IV. PERPECTIVES Mesh with high blending of the Improved mesh with triangular wire elements around the wire tips Fig. 13. Different meshes approach to handle the wire The base mesh used in this study is visualized in Fig. 12. An easy way to improve the orthogonality is to blend the wire with the relative pin creating what is shown in Fig. 13 left. A more accurate solution is to hybrid hexahedral and prism with triangular base around the wire obtaining a better geometrical description (Fig. 13 right). Fig. 14 shows a visual comparison between the blended and the triangular (around the wire) meshes. The flow features seems pretty similar, the triangular mesh is only showing very low velocities at the tips of the wire, whereas these areas are chopped out in the blended mesh. Table I, that reports the value of the friction factor f for the different meshes and the perceptual difference with respect to the base mesh, also confirms this fact. The difference between the models is around 11%, while between the base and the blended mesh is less than 5%. The previous calculations and tests have indicated that a reasonable confidence could be placed in Code_Saturne results. Moreover, it is also clear that the number of pins has a very strong influence on the global behavior. Therefore EDF is interested to know if it is possible to tackle real cases with up to 271 pins and understand what happens for example when local hot spot or degraded inlet velocity occur. Several meshes have been generated counting 61 pins and several helices and 271 pins on 1 helix and five helices so far. The latest meshes counts above 100 millions cells and needs to be handled thanks to EDF s mainframe with 256 or 511 processors to get reasonable CPU times. TABLE I Comparison of f for different mesh at Re=25000 using k-ε. Mesh Type Base Base Rij Triangular Blended f eq. () 0.287 e -1 0.319 3-1 0.283 e-1 0.274 e-1 Difference [%] 0.00 11.1 1.39 4.53 Fig. 15: Domain decomposition of a 271pins and 1 helix case in 256 independent domains.

As shown on figure 16 (corresponding to only one helix case) the velocity field confirm that the edge and corner sub-channels start to loose their predominance and the flow field is more homogeneous for the inner pins. Fig 18: Scalar field of a small spot (set to 1 at the inlet) Fig. 16 : Velocity norm for the 271 pins case The second type of test currently done at EDF corresponds to the investigation of the influence of the wire wounded around each pin on a small spot (here a scalar set initially to zero and set to one in a small portion of the inlet). One clearly note on Fig 17 and 18, both a deviation (even after only one helix) and a diffusion of the scalar field. Similar calculations on a five helixes case are in progress. The second industrial aspect that EDF wants to investigated in the near future concerns the possibility to tackle the conjugated heat transfer taking place between the solid and the fluid to have access to the local cladding or fuel temperatures as well as the sodium temperature. In order to do that the solid parts, ie pins (with fuel and cladding), wires as well as the hexagonal can are explicitly taken into account. The heat deposit is set only on the central part where fuel is located. For the time being the procedure is tested with success on the 7 pins case and should be extended in the near future to assemblies with a larger number of pins (up to 271). Fig. 19: Solid and fluid meshes used for conjugate heat transfer Fig 19 presents a view where both fluid and solid meshes are gathered. One underlines that in order to provide more flexibility, meshes at the interface do not need to coincide. The preliminary results obtained (The configuration presented on Fig. 20 corresponds to a 3 helices case) suggest a strong local influence of the spacer wire. Fig. 17: Scalar field of a small spot (set to 1 at the inlet)

ACKNOWLEDGMENTS Authors express their thanks to D. Laurence (EDF) for fruitful discussions on turbulent modeling and D. Monfort for advices in Code_Saturne usage. S. Rolfo is grateful to the UK Engineering and Physical Sciences Research Concil for funding under grant EP/C549465/1 Keeping the Nuclear Option Open. NOMENCLATURE Fig. 20: Cladding and wire temperature (7 pins case ) V. CONCLUSIONS In this paper thermal-hydraulic of wire wrapped fuel bundle was investigated. Two different configurations, with seven and nineteen fuel rods, were taken into account, finding that the main flow features remain unchanged as the number of pins increase. Two different turbulence models were tested finding good agreement with experimental correlations. On the other hand the evaluation of the heat transfer requires more investigation. Experimental correlations for Nusselt number are quite scattered, making difficult the assessment of the CFD. Indeed the wall modeling for the heat transfer is only fist order accuracy, which can make quite dubious the Nusselt evaluation. Better modeling employing low Reynolds models could be required and comparison with high Reynolds approach is on going. Another solution to the problem, which can avoid the use of refined mesh in the near wall region, could be the use of more advance wall functions as the one presented in 24. As the number of pins increase the influence of side and corner sub-channels is becoming less important and flow is more homogeneous. In general it was difficult to find data to compare with. In this contest refined LES and DNS could play a big role providing a large and reliable amount of data for RANS modeling evaluation. ρ : density U i : velocity component k : turbulent energy ε : turbulent dissipation rate term Cµ : constant (0.09) Pr T : turbulent Prandtl number p* : pressure g : gravity x i : coordinates t : time µ : viscosity C p : specific heat k s : thermal conductivity φ : volume heat source q w : wall heat flux S wh : wall heat surface h : heat transfer coefficient T : temperature β : coefficient for periodic model γ : coefficient (energy balance) m : mass flow rate REFERENCES 1. Archambeau, F., Mechitoua, N., & Sakiz, M. (2004) Code\_Saturne: a finite volume code for the computation of turbulent incompressible flows - Industrial Applications. Int..J Finite Vol. 1,. 2. Bubelis, E. & Schikorr, M. (2008) Review and proposal for best fit of wire-wrapped fuel bundle friction factor and pressure drop predictions using various existing correlations. Nuclear Engineering and Design 238, 3299--3320. 3. Chabard, J. P. & Laurence, D. (2009) Heat and fluid flow simulations for deciding tomorrow s energies. Proceedings of the Sixth International Symposium on Turbulence, Heat and Mass Transfer, 1--16. 4. Cheng, S. K. & Todreas, N. E. (1986) Hydrodynamic models and correlations for bare and wire-wrapped hexagonal rod bundles--bundle friction factors,

subchannel friction factors and mixing parameters. Nuclear Engineering and Design 92, 227--251. 5. Engel, F.C., Markley, R.A. and Bishop, A.A. (1979) Laminar, transition, and turbulent parallel flow pressure drop across wire-wrap-spaced rod bundles. Nucl Sci Eng 69, 290-296. 6. Ferziger, J. H. & Peric, M. (1997) Computational Methods for Fluid Dynamics. Springer. 7. Fischer, P., Lottes, J., Siegel, A., & Palmiotti, G. (2007) Large Eddy Simulation of wire-wrapped fuel pins I: Hydrodynamics in a Periodic Array. Joint International Topical Meeting on Mathematics & Computation and Supercomputing in Nuclear Applications (2007), Monterey, California 8. Gajapathy, R., Velusamy, K., Selvaraj, P., Chellapandi, P., & Chetal, S. (2007) CFD investigation of helical wire-wrapped 7-pin fuel bundle and the challenges in modeling full scale 217 pin bundle. Nuclear Engineering and Design 237, 2332--2342. 9. Graber, H. & Rieger, M. (1972) Experimentelle Untersuchung des Warmeubergangs an Flus- sigmetall (NaK) in parallel durchstromten Rohrbundeln bei konstanter und exponentieller Warmeflussdichteverteilung. 19,. 10. Grotjans, H. & Menter, F. R. (1998) Wall functions for general application CFD codes. Computational fluid dynamics'98, 1112--1117. 11. Jones, W. P. & Launder, B. E. (1972) The prediction of laminarization with a two-equation model of turbulence. International Journal of Heat and Mass Transfer 15, 301-314. 12. Kazimi, M. S. & Carelli, M. D. (1976) Clinch River Breeder Reactor Plant heat transfer corre- lation for analysis of CRBRP assemblies. Westinghouse, CRBRP-ARD-0034. 13. Mikityuk, K. (2009) Heat transfer to liquid metal: Review of data and correlations for tube bundles. Nuclear Engineering and Design 239, 680-687. 14. Novendstern, E. H. (1972) Turbulent flow pressure drop model for fuel rod assemblies utilizing a helical wire-wrap spacer system.. Nuclear Engineering and Design 22, 19-27. 15. Pfrang, W. & Struwe, D. (2007) Assessment of Correlations for Heat Transfer to the Coolant for Heavy Liquid Metal Cooled Core Designs. FZKA 7352. 16.Péniguel, C., Rupp, I., Juhel, J., Guillaud, M., Gervais, N., & Rolfo, S. (2009) Three Dimensional Conjugated Heat Transfer Analysis in Sodium Fast Reactor Wire- Wrapped Fuel Assembly Three Dimensional Conjugated Heat Transfer Analysis in Sodium Fast Reactor Wire-Wrapped Fuel Assembly Three Dimensional Conjugated Heat Transfer Analysis in Sodium Fast Reactor Wire-Wrapped Fuel Assembly. Proceedings of ICAPP 09Tokyo, Japan, May 10-14, 2009, Paper 9311. 17. Raza, W. & Kim, K. Y. (2008) Comparative Analysis of Flow and Convective Heat Transfer between 7-Pin and 19-Pin Wire-Wrapped Fuel Assemblies. Journal of Nuclear Science and Technology 45, 653--661. 18. Rehme, K. (1973) Pressure drop correlations for fuel element spacers. Nuclear Technology 17, 15-23. 19. Rhie, C. & Chow, W. (1983) Numerical study of the turbulent flow past an airfoil with trailing edge separation. AIAA Journal(ISSN 0001-1452) 21, 1525-- 1532. 20. Rolfo, S. (2010) Wire wrapped fuel rod bundle: animations available from: http://cfd.mace.manchester.ac.uk/twiki/bin/view/cfdt m/respub229 21. Rupp, I. & Peniguel, C. (1999) Coupling heat conduction, radiation and convection in complex geometries. International Journal of Numerical Methods for Heat and Fluid Flow 9, 240--256. 22. Smith, J. G., Tokuhiro, A., Pointer, W. D., & Fischer, P. F. (2009) Predictions in CFD Simulations of Wire- Wrapped SFR Fuel Assemblies. Proceedings of ICAPP 09Tokyo, Japan, May 10-14, 2009. 23. Speziale, C. G., Sarkar, S., & Gatski, T. B. (1991) Modelling the pressure-strain correlation of turbulence-an invariant dynamical systems approach. Journal of Fluid Mechanics 227, 245--272. 24. Suga, K., Craft, T., & Iacovides, H. (2006) An analytical wall-function for turbulent flows and heat transfer over rough walls. International Journal of Heat and Fluid Flow 27, 852--866.