Vous êtes sur la page 1sur 15

Progress in Nuclear Energy 114 (2019) 31–45

Contents lists available at ScienceDirect

Progress in Nuclear Energy


journal homepage: www.elsevier.com/locate/pnucene

Baffle jetting: CFD analysis of plain jets impinging on fuel rods T


a,∗ b
Ulrich Bieder , Alexander Rashkovan
a
DEN-STMF, CEA, Université Paris-Saclay, F-91191, Gif-sur-Yvette, France
b
Physics Department, NRCN, P.O.B. 9001, Beer-Sheva, 84190, Israel

A R T I C LE I N FO A B S T R A C T

Keywords: The risk of baffle jetting exists mainly in pressurized water reactors, which are designed with a counter flow
Baffle jetting configuration in the core bypass region. Baffle plates are foreseen between core barrel and core, providing the
Rod failure core's lateral restraint and allowing a core bypass flow. A pressure differential establishes between the coolant
CFD flow in the core baffle and the upward flow through the core. A part of the coolant bypass flow is directed
LES
through the gaps of the baffle plates towards the core. Coolant cross flow through enlarged baffle gaps can result
Validation
in high velocity jetting in the case of significant pressure differences between core and baffle region. These water
jets impact onto the fuel rods and may induce rod vibration that can lead to fuel rod failure.
Modelling baffle jetting induced vibrations of fuel rods is still a challenge in the field of fluid structure
interaction. The present work aims to extend the understanding of the unsteady flow in rod bundles to situations
where baffle jetting occurs. As a first step, the used LES modelling method is evaluated on the example of a well
documented test case related to fuel rods in cross flow: the flow over a circular cylinder at Re = 3900. In the
successive step, the validated modelling method is applied to the CEA PANACHET experiment, mimicking baffle
jetting related phenomena. This analysis helped to understand better the flow field of a corner baffle jet and
further validated the modelling approach.
The flow field in the core baffle corner is finally analysed for realistic reactor conditions. A plain narrow slot
jet is injected parallel to a baffle wall through a 1 mm width gap perpendicular to the axial flow of a corner
assembly. The analysis of the flow field showed that: (i) Swirling flow exist close to the corner rod. The cal-
culated force coefficients lift and drag of this rod show a dominant fluctuation frequency of about 80 Hz. (ii) The
baffle jet impacts on the third rod in jet direction (passing the first two rods without impact). The force coef-
ficients of the third rod do not show a dominant oscillation frequency but show increased oscillation amplitudes.
The distinguished fluctuation frequency detected for rod n°1 and the direct impact of the baffle jet on rod n°3
might explain the rod failures that have been detected for these rods.

1. Introduction between intersecting baffle plates becomes enlarged, the differential


pressure across the baffle plate creates a water jet that impinges on
According to NRC Circular Notice 80-17 (1980), failed fuel pins adjacent fuel rods. As shown in Fig. 1c, this jet can impinge onto the
were detected in two assemblies during the past operating cycle. It was fuel rods either tangentially, forming corner baffle jets, or perpendicu-
reported that failed fuel pins were located adjacent to a joint in the core larly, forming central baffle jets. This water jet impingement causes flow
baffle, and that the failures had resulted from vibration caused by water induced vibration and can potentially result in fuel rods damage.
jet impingement on the fuel pin, the baffle jetting. Baffle jetting can The IAEA-TECDOC-1119 (1999) summarized that at the beginning
occur in reactor pressure vessels (RPV) designed with a counter flow of the 1980s, jets of water through the baffles created damage to the
configuration in the bypass region. In this case, a significant pressure fuel elements in various power plants all over the world (six in France
differential across the baffle plate is established. Fig. 1a shows sche- including Bugey 2 and Fessenheim 2). This effect was related to the
matically such counter courant flow in the core baffle. Fig. 1b presents a pressure difference between the baffles with water down flow in this
horizontal cut plane of the RPV internal structure. The fuel assemblies, location. Modifications in the baffle formers (see Fig. 1a), assuring
which can be affected by baffle jetting (cross × marked in Fig. 1b), are water up flow have reduced the pressure and suppressed baffle jetting.
located in the periphery of the core, adjacent to the core baffle. If a gap This modification was later implemented practically in all reactors of


Corresponding author.
E-mail address: ulrich.bieder@cea.fr (U. Bieder).

https://doi.org/10.1016/j.pnucene.2019.02.006
Received 28 September 2018; Received in revised form 13 February 2019; Accepted 26 February 2019
0149-1970/ © 2019 Elsevier Ltd. All rights reserved.
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 1. Formation of baffle jets: Downward baffle flow in (a), location of potentially affected fuel assemblies in (b) and schematics of baffle jets impinging on fuel rods
in (c).

relevant design. For instance, Framatome converted the reactor lower CEA has initiated the experimental program CHRISTINE to better
internals of the Tihange 1 pressurized water reactor in Belgium from a understand the combined action of axial and cross flow on the vibration
down flow to an up flow configuration in September 1986. excitation of fuel rods (Boulanger et al., 1997). Baffle jetting was found
Nevertheless, in recent years, several incidents related to baffle to induce early large vibratory amplitudes of fuel rods at low turbulence
jetting were reported. The Slovenian Nuclear Energy Administration kinetic energy levels and thus failure from fretting wear (Tekatlian
reported in 2013 that the Krško nuclear power plant and the fuel sup- et al., 1999). The PANACHET single effect analytical experiment was
plier (Westinghouse) consider that the damage to the fuel rods at the set up in 1996 in support to the CHRISTINE program (Jacques et al.,
outer edge of the core has occurred mainly due to the stream of water at 1997) in order to better understand early, large amplitude, vibratory
the reactor baffle plates, which caused excessive vibration of fuel pins. instability. Numerical analysis with CFD was associated to the PANA-
Small cracks were detected on the pin surface through which the pri- CHET program (Jacques, 1999).
mary coolant system water has penetrated the fuel sheath (primary In the study on local cross flow induced vibrations, it was shown
damage). Two nuclear fuel rods were found damaged during North that vibration mechanisms are very similar to those observed in uni-
Anna power plant refuelling in 2014 (Licensee Event Report No. 50- form cross flow (Fujita et al., 1990). However, additional vibration
339/2014-002-00). The fuel pellet loss occurred because of baffle jet- mechanisms can appear, which are related to the small jet width
ting on the fuel assembly. The affected fuel rods had their top springs (Fujita, 2009). The jet position has proved to have a major influence on
dislodged and fuel pellets were able to escape the fuel rod. Fragments of vibratory amplitude (Seki et al., 1986). Although the axial flow tends to
fuel pellets were found within the associated fuel assembly and on the stabilise the tube array vibrations, it was shown that when rod arrays
core plate. are subjected to both cross flow and uniform steady axial flow, a fluid-
Baffle jetting is not classified as a significant safety risk, as the in- elastic instability can occur.
creased radiation level due to clad damage can be easily detected and Baffle jetting flow induced vibration of fuel rods is still a challenge
enclosed in the primary circuit. Additionally, the formation of baffle in the field of fluid structure interaction. The present work aims to
jets and the associated fuel rod vibrations are detectable in the signals extend the understanding of the unsteady flow field near fuel rods
of the neutronic noise (Bernard et al., 1985). However, following fuel subjected to baffle jetting. The study is thus dealing with predicting the
pins damage, significant costs can arise from cleaning and fuel re- complicated flow field around tube bundle in situations mimicking the
placement actions. Nevertheless, only little research was performed to real life nuclear reactor geometry features. Therefore, in a first step, the
better understand the phenomena of rod failure associated with baffle numerical and physical model is evaluated on the example of a well
jetting. documented test case related to fuel rods in cross flow: the flow around
Uniform cross flow induced vibrations, being usually encountered in a circular cylinder at Re = 3900. Then, the validated calculation
heat exchanger tube bundles (e.g. in shell and tube heat exchangers), method is applied to the PANACHET experiment (jet Reynolds number:
were extensively investigated in the past and the vibration mechanisms 5000–10000). In this context, the flow field of a corner baffle jet is
are well understood (see e.g. Tanaka and Takahara, 1981). The phe- analysed and calculation results are compared to the experiment.
nomena of heat exchanger tube bundles vibrations and those in the Finally, the flow field under realistic reactor conditions is discussed and
nuclear fuel rod bundles assemblies baffle jets induced vibration differ pressure coefficients are evaluated for two rods (jet Reynolds number:
in a number of ways. The velocity (5–10 m/s) of the flow impinging 50000).
onto the fuel rods is not uniform in the cross section (0.5–1 mm). Hence,
the fuel rod behaviour is dependent on the jet-to-rod relative location.
The shape of the jet velocity profile will greatly affect the emerged flow 2. The numerical scheme
induced vibration (Seki et al., 1986). Additionally, as the baffle jets are
not exclusively directed in the direction normal to the fuel rods axis, the The fluid is assumed to be Newtonian, incompressible and the flow
resulted flow situation is complicated by the combination of the in- is treated as turbulent. The flow around the fuel rods is described using
teraction of both cross and axially directed coolant flow with the rods in large eddy simulation (LES) (see e.g. Sagaut, 2001; Pope, 2000). The
the fuel assemblies situated in the periphery of the reactor pressure application of a filtering operation (symbolised by ∼) to the equation of
vessel, close to the core baffle plates. mass conservation and Navier–Stokes equations gives:

32
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 2. Cylinder test case: boundary condition and meshing in (a) and zoom to the mesh on the cylinder wall in (b).

∇⋅ũ = 0 (1) related geometry: a circular cylinder in cross flow. One of the best
documented cylinder experiments, which has been analysed with a
∂u˜ 1
+ (u˜ ⋅∇) u˜ = − ∇P˜ + ∇⋅((υ + υt )⋅(∇u˜ + ∇T u˜)) wide range of modelling approaches and numerical methods, is that for
∂t ρ (2) Reynolds number Re=3900 (Norberg, 1987; Wallace and Ong, 1996;
In this study, ũ is the filtered velocity vector and P̃ is the filtered Parnaudeau et al., 2008). This flow was analysed with LES by boundary
pressure. Filtering is implicit; the cut-off length is the grid characteristic fitted curvilinear and unstructured finite volume methods (Breuer,
size Δx. Scales smaller than Δx are not resolved. Only their effect on the 1998; Afgan et al., 2011), spectral methods (Kravchenko and Moin,
resolved, larger than Δx scales, is taken into account by the sub-grid 2000) and discontinuous Galerkin methods (De La Llave Plata et al.,
scale tensor. As in most LES models, the sub-grid scale tensor is ex- 2017).
pressed by using the concept of eddy viscosity with the definition of the Turbulent flows past bluff bodies in general and around circular
turbulent viscosity νt. The first model applied is of Smagorinsky (1963), cylinders in particular are very complex (Breuer, 1998) including
that gives the classical expression of νt. Nicoud and Ducros (1999) have phenomena such as separation, reattachment and vortex shedding with
shown that this model suffers from severe limitations near walls, in large two- and three-dimensional vortical structures. Although the
shear layers and in the turbulence transition process. Thus, their wall- geometry is quite simple, the flow phenomena are extremely rich and
adapting local eddy-viscosity model WALE (Nicoud and Ducros, 1999) have been investigated in a large number of studies (Zdravkovich,
is used in this study. 1997). Recently, modelling fluid-structure interaction was also ad-
Simulations are performed using the TrioCFD code (Angeli et al., dressed (Gsell et al., 2017, 2018). The flow regime for Reynolds number
2015; TrioCFD), developed at the French Alternative Energies and between about 2000 < Re < 40000 is called transition in shear layer
Atomic Energy Commission (CEA). Angeli et al. (2017) have demon- (Zdravkovich, 1997). A laminar boundary layer develops on the cy-
strated the accuracy of the discretisation scheme employed in TrioCFD. linder wall. Transition to turbulence occurs in the wake due to ampli-
Eqs. (1) and (2) are solved using the pressure projection method of Hirt fied sheer layer instabilities (Singh and Mittal, 2004).
et al. (1975). Time marching is explicit and relies on the classical 2nd The unsteady flow around a circular cylinder of diameter D is
order Adams-Bashforth method. A finite element based finite volume analysed with LES using the finite element based finite volume method
method is used for spatial discretisation on tetrahedral elements with described in chapter 2. Concerning the expected quality of the analysis,
non-conforming P1 elements for velocity components that are located it should not be forgotten that the objective of this study is not to re-
in the centres of the faces of an element. Conforming P0/P1 elements produce as exactly as possible experimental values, but to evaluate a
are used for pressure discretisation where the pressure nodes are lo- calculation strategy in an engineering framework. Fig. 2 shows the
cated in both centres and vertices of an element (Angeli et al., 2017). calculation domain of 40 × D, 20 × D and π × D in stream wise (x-
This discretisation may be viewed as a generalization of the Marker and direction in Fig. 2a), transverse (y-direction in Fig. 2a) and span wise (z-
Cell (MAC) method of Harlow and Welch (1965) for unstructured me- direction) directions, respectively. The physical properties dynamic
shes. The MAC method is very well suited for incompressible flows viscosity (μ) and density (ρ) as well as the boundary conditions are
(Ferziger and Peric, 1997) being both accurate and stable. In particular, added to Fig. 2. A global view of the meshing as well as the close to the
the method does not suffer from spurious node-to-node pressure modes cylinder wall region meshing are given. Maximum cylinder wall non-
(Fortin, 2006). Convection terms are handled with an adaptation of the dimensional values of the mesh are of the order of y+ ≈ 2 in wall
flux corrected transport formulation of Kuzmin and Turek (2002). normal direction. The mesh resolution in stream wise direction is in the
Ducros et al. (2010) have reported the main features of this formula- order of x+ ≈ 10 and in span wise direction of z+ ≈ 30. After various
tion. The resulting convection scheme (called ef_stab) is both robust and tests on methods for meshing and mesh refinement, the final meshing
accurate since a minimal amount of numerical diffusion is added to was realized with the octree method of ICEMCFD. This mesh comprises
ensure the computation stability of the centred scheme (Kuzmin and of 22 million tetrahedral cells corresponding to about 44 million control
Turek, 2002). In this work, the critical time step depends on the con- volumes for the velocity. The wake of the cylinder was discretized with
vection operator for which the Courant–Friedrichs–Lewy (CFL) number elements, which are 8 times smaller (Δx = 0.05 m) than the elements in
is set to 0.8. Bieder and Rodio (2019) presented more information of the the periphery of the calculation domain. Three prisms layers (cut into
numerical scheme. Bieder et al. (2015) has discussed the code realiza- tetrahedrons) were introduced close to the cylinder for properly resol-
tion on High Performance parallel Computers. ving of wall boundary layer - see Fig. 2b.
The vortex street stabilizes after about 10 vortex shedding cycles.
Turbulence statistics have then been collected in the span wise mid-
3. Evaluation of methodology for large eddy silmulations plane (z = π × D/2) for about 50 shedding cycles. Only time averaging
was performed, without geometrical averaging in span wise direction.
High quality experimental data are not available to evaluate a LES Time averaged velocity statistics along vertical lines downstream of the
modelling strategy applicable to corner baffle jets. Guidelines for such cylinder at x/D = 1.06, 1.54 and 2.02 were collected. These lines are
calculation methodology are developed and evaluated for a physically

33
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 3. Comparison of flow statistics to experimental results of Parnaudeau et al. (2008): time averaged velocity < u > in (a) and < v > in (b) as well as time
averaged Reynolds stresses < u'u’ > in (c) and < v'v’ > in (d).

shown schematically in Fig. 2a. The statistics are compared in Fig. 3 to meshing is probably not sufficiently fine to resolve correctly the fine
the measured values reported by Parnaudeau et al. (2008). The time sheer layers. However, when comparing the calculated Reynolds
averaging is designated by “ < > “. Fig. 3a shows profiles of the ve- stresses further downstream at x/D = 3 and 5 to the measurements
locity component in stream wise direction, < u > , while velocity reported by Ong and Wallace (1996), a good accordance was achieved
component in the transverse direction, < v > , is presented in Fig. 3b. (not presented here for brevity).
The velocities are normalized by the bulk velocity Uc (3900 m/s). The The length of the recirculation bubble in the wake of the cylinder is
temporal mean values of the velocity components represent very well estimated from Fig. 4a, where the time averaged stream wise velocity in
the measurements. The corresponding profiles of the time averaged the cylinder wake along the centre line (in the span wise mid-plane) is
Reynolds stresses < u'u’ > and < v'v’ > , which are normalized by Uc2, presented. The length of the recirculation zone is determined by the
are shown in Fig. 3c and d, respectively. Differences between calculated axial location, where the stream wise velocity changes from backflow
and measured Reynolds stresses exist close to the cylinder, as the (negative velocity) to positive values. The calculated time averaged

34
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 4. Comparison of the LES to experimental profiles: time averaged velocity in stream wise direction in the wake of the cylinder in (a) and time averaged local
pressure coefficient in azimuthal direction along the cylinder wall in (b).

velocity in stream wise direction (< u >) is compared to experiments Fig. 5 shows the power spectra of transverse velocity fluctuations, Evv/
of Parnaudeau et al. (2008), Ong and Wallace (1996) and Lourenco and (Uc2D), at two locations on the centre line in the wake of the cylinder:
Shih (1994). The data of Lourenco and Shih (1994) were obtained from x/D=3 and x/D=5. Samples of the transverse velocities were collected
Kravchenko and Moin (2000). The length of the recirculation zone is over about 50 vortex shedding cycles. The spectra, calculated from
slightly overestimated but still well predicted by the LES. these time series, were averaged in the span wise direction to increase
In order to validate the correct pressure distribution around the the statistical sample. Since an adaptive time stepping technique was
cylinder, the time averaged local pressure coefficient Cp(γ) in the mid used for the time integration (varying time steps), the collected time
span plane is compared to the experiment of Norberg (1987). The time series were interpolated to an equidistantly positioned time vector of a
averaged local pressure coefficient, Cp(γ), is calculated from eq. (3) as a power of 2 length. To obtain the spectra, the periodogram technique
function the temporal mean value of the wall pressure < P(γ) > at was used (Welch, 1967) employing the GNU program Octave
angle γ: (OCTAVE). The frequency was normalized by the calculated Strouhal
shedding frequency (fvs) of 0.21. The power spectra (normalized by the
< P (γ ) > − P∞ bulk velocity squared) are displayed together with the experimental
Cp (γ ) = 1
2
ρUc2 (3) results of Ong and Wallace (1996), which have been obtained from
Lysenko et al. (2012). A −5/3 slope is shown as well. The overall
The reference pressure P∞ is set to zero. The correct prediction of agreement between experimental and numerically predicted spectra is
the pressure distribution along the cylinder wall is shown in Fig. 4b. good at both locations.
The angle turns from the frontal stagnation point on the cylinder at 0° to This first study has shown the good performance of the developed
the cylinder back face at 180° as shown in Fig. 4b. LES methodology by comparing experimental and calculated
In analogy to Parnaudeau et al. (2008) and Lysenko et al. (2012),

Fig. 5. Power spectra of the velocity component in transvers direction at two distances from the cylinder: x/D = 3 in (a) and x/D = 5 in (b).

35
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 6. The PANACHET facility: view on test channel in (a), geometry and dimensions of the test section in (b).

turbulence statistics (mean values, Reynolds stresses and power scheme. A periodic channel of 0.005 m height and 8 dh length is con-
spectra) of the flow around a single circular cylinder. The peripheral nected to the inlet boundary of the test section in order to impose a fully
distribution of the cylinder wall pressure was also well predicted. The developed turbulent channel flow. The coupled simulation of test sec-
LES methodology is thus applied in a second step to small scale PAN- tion and periodic channel advance in time jointly. The unsteady velo-
ACHET experiments related to tangential baffle jets. city distribution of the periodic channel is imposed at each time step on
the test section inlet. Bieder et al. (2014) have discussed this procedure
4. LES analysis of baffle jetting for experimental conditions in more detail for LES in fuel assembly flow. Physical properties cor-
respond to water at ambient conditions (density ρ = 1000 kg/m3 and
4.1. Description of the experimental facility PANACHET dynamic viscosity μ = 10−3 Pa s). The used time step of about
2 × 10−6 s respects the Courant-Friedrichs-Levy stability criteria.
CEA has initiated the experimental program CHRISTINE to better
understand the combined action of axial and cross flow on the vibration 4.3. Qualitative analysis of the flow in the experimental facility
excitation of fuel rods (Boulanger et al., 1997). The PANACHET single
effect analytical experiment was set up in 1996 in support to the The flow in the PANACHET test facility is discussed qualitatively for
CHRISTINE program (Jacques, 1999) in order to investigate early, large Reynolds number Re = 5010 based on the hydraulic diameter dh of the
amplitude, vibratory instability. The small scale facility was built to inlet channel (mean inlet velocity = 0.53 m/s). The flow field in the test
mimic the fluid structure interaction for the case of baffle jetting. Only section is hydraulically fully established after a transient of about 3.5 s.
tangential jets were investigated. The deflection of the jet by the axial The flow distribution in the test section is visualized in Fig. 8. The time
flow in the corner assembly was not investigated in analytical experi- averaged velocity magnitude contours are shown (averaging period
ments. The experimental model consists of an array of 36 inline posi- from 3.5 to 4.5 s of the transient). Half of test facility is removed vi-
tioned 6 × 6 cylindrical tubes as presented in Fig. 6a and b. The tubes sually from the figure to better illustrate the flow close to the bottom
are placed in the square cross section channel of 0.08 × 0.08 m2 build wall.
in Plexiglas as shown in Fig. 6a. The flow entering the test section through the inlet channel is de-
The geometry of the test section is shown in Fig. 6b. Concerning the flected upward by the 1st rod. In fact, the flow splits in two main
arrangement of rods, rows are defined in stream wise directions (x-di- branches after impinging onto the corner rod. One moves horizontally
rection) and columns in transverse direction (y-direction). The rod's in the gap between the 1st row of rods and the bottom wall in the
diameter is 9.5 mm (D) with a pitch-to-diameter ratio of 4/3 (P/D). The positive x-direction. The second branch moves upward in the gap be-
vertical and horizontal (Yf and Xf respectively) distances from the tween front wall and 1st column of rods, turns right (90°) in the corner
centre of the closest rods to the channel walls was set to 8.4 mm, between the front wall and the upper wall and leaves the rod bundle
leaving a rod-to-wall gap of 3.65 mm. The experimental set-up of essentially in the gap between upper wall and the 6th row of rods. A
PANACHET allowed to perform experiments with the rods rigidly held significant diagonal flow develops downward between the rods. This
and loosely fixed. Only rigid rods are considered in the present study. A downward flow balances the upward flow near the front wall. A
plane wall jet is injected into the test section by an injection channel of backward flow is formed in the wakes of the first three rods of the 1st
5 mm height (y-direction) and 80 mm width (z-direction). Downstream row. Secondary flow structures are seen in the wake of the rods of the
of a convergent, this channel has an injection length in x-direction of 1st row.
about 40 hydraulic diameters dh with an almost constant cross section. Fig. 9 presents contours of the instantaneous velocity magnitude in
The ratio of rod-to-wall gap to the height of the inlet channel (l) is 0.73. the span wise mid-plane of the PANACHET facility. The velocity mag-
nitude (Fig. 9a) and velocity vectors close to the exit of the injection
4.2. Setup of the calculation and boundary conditions channel (Fig. 9b) are presented. The deflection of the flow on the 1st
rod into two main streams as well as the formation of an extended re-
The unsteady corner baffle jetting flow in the PANACHET facility is circulation zone between 1st rod and front wall are clearly seen in
analysed by LES with the finite element based, finite volume method Fig. 9a and b, respectively.
described in chapter 2. The meshing of the calculation domain is shown The time averaged velocity magnitude is visualized in Fig. 10 for
in Fig. 7. The grid of 32 million tetrahedrons was realized with the two planes. Plane (a) is a vertical plane in y-z direction in 1 mm dis-
advancing front method of ICEMCFD. The mesh characteristic is similar tance from the front wall and plane (b) is a horizontal plane in z-x
to that of the single rod analysis of chapter 3. direction in 2 mm distance from the bottom wall.
No slip walls were used on all rods’ surfaces, test section- and The velocity distribution in the vertical plane of Fig. 10a shows that
channel walls (y+ < 2 was assured). Neumann boundary conditions the velocity in the inlet channel is not constant in span wise direction as
with a constant pressure outlet were used at the outflow face of the test supposed by the experimentators. Rather the flow accelerates in the
section. In order to stabilise the flow near the outlet, a first order up- centre of the channel due to momentum deficit at the edges of the inlet
wind convection scheme replaces close to the outlet the centred channel caused by wall friction on the lateral walls. Therefore, the

36
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 7. View on the meshing of the PANACHET facility: global view on the lateral side in (a) and zoom on the first two rods in (b).

upward flow between front wall and 1st rod (at y ≈ 0.01 m) shows averaged over 1 s (solid lines in Fig. 11).
significant diversions in span wise direction that are directed from the The wide dispersion of the measured velocity poses some doubts on
span wise mid-plane to the lateral walls. the quality and validity of the experimental results. However, as long as
The velocity distribution in the horizontal plane shows the forma- no better data is available, the PANACHET experiments are taken as
tion of stream wise secondary flow structures in the gap between reference. For both Reynolds numbers, the overall comparison of the
bottom wall and 1st row of rods. The stripes in span wise direction wall bounded jet (y < 0.006 m) seems to be of reasonable accuracy.
represent regions of lower velocity. As these regions are located close to Unfortunately, it is not possible to conclude about the formation of the
the rods, the beginning of the formation of significant stream wise recirculation zone between the front wall and the 1st rod and the
secondary flow structures is estimated to start at the 4th row. This backward flow in the wake of the 2nd rod due to the location of the
phenomena is also observed Fig. 8. profile in the centre of the gaps, where the x-component of the velocity
vector is very small. Measuring the y-component of the velocity vector
in this location would have significantly enriched the experimental
4.4. Quantitative analysis of an PANACHET experiment database.
The distribution of the time-averaged pressure (1-s temporal mean
Particle image velocimetry (PIV) measurements of the stream wise value) is shown in Fig. 12a for the span wise centre plane with a zoom
(x-direction) velocity components have been performed in the span wise on the first two rows and first two columns. The Reynolds number is
mid-plane (Jacques, 1999) of the test section of the PANACHET facility. 10300. The points of high pressure on the first row of rods represent the
In an internal report of CEA, the error of the measurement was reported velocity stagnation points on the rods wall. The low pressure zone be-
to be below 1%. Unfortunately, it is not possible today to confirm this tween 1st row and front wall is formed due to already mentioned re-
reported high precision. Spatial profiles in transverse direction circulation zone. Fig. 12b shows the peripheral wall pressure distribu-
(0 < y < 0.013 m) were realized in the gap between front wall and tion of the corner rod (1st rod). The experimental values, the calculated
1st rod (x = 0.001825 m) as well as in the gap between 1st rod and 2nd mean pressure (averaged over 2 s) as well as the standard deviation of
rod (x = 0.01325 m). Profiles of the time averaged stream wise velocity the calculated pressure fluctuations (std) are presented in Fig. 12b. The
of the baffle jet are compared in Fig. 11 for the Reynolds numbers calculation globally represents well the experiment.
Re = 5010 (two left plots in Fig. 11) and Re = 10300 (two right plots in Neglecting viscous forces, the force coefficients lift and drag were
Fig. 11), corresponding to bulk inlet channel velocities of 0.53 and calculated for the 1st rod according to eqs. (4) and (5), respectively.
1.1 m/s, respectively. The experimental values are marked with symbol
“+” and are compared to the time averaged values obtained with LES,

Fig. 8. Time averaged velocity magnitude in the PANACHET facility.

37
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 9. Instantaneous velocity in the span wise mid-plane of the PANACHET facility: velocity magnitude in (a) and velocity vectors (coloured with velocity mag-
nitude) close to the exit of the injection channel in (b).

0.88 × L 2π
Fl ∯< P (γ ) > ⋅sin(γ )⋅dγ⋅ds 1 order to increase the statistical sample, time series of the force coeffi-
CL = 1 2
= 1
ρUc2⋅D⋅hc
=
D ⋅h c
∫ ∫ Cp (γ ) cients were collected over approximately 2 s at 13 equidistantly dis-
2
ρUc ⋅D ⋅h c 2 0.12 × L 0
tributed locations in span wise direction that are located between
⋅sin(γ )⋅dγ⋅ds (4) 0.01 m and 0.07 m. The spectra calculated from these time series were
0.88 × L 2π
then spatially averaged. Since an adaptive time stepping technique was
Fd ∯< P (γ ) > ⋅cos(γ )⋅dγ⋅ds 1 used for the time integration (varying time steps), the collected time
CD = 1 2
= 1
ρUc2⋅D⋅hc
=
D⋅hc
∫ ∫ Cp (γ)
2
ρUc ⋅D ⋅h c 2 0.12 × L 0 series were interpolated to an equidistantly positioned time vector of a
power of 2 length. To obtain the spectra, the periodogram technique
⋅cos(γ )⋅dγ⋅ds (5)
was used (Welch, 1967) with the GNU program Octave (OCTAVE) al-
The time averaged local pressure coefficient, eq. (3), is integrated ready mentioned in chapter 3.
circumferentially around the rod of diameter D and in span wise di- It is interesting to note that a dominant shedding frequency does not
rection between z = 0.01 m (0.12 × L) and z = 0.07 m (0.88 × L). L is exist for the two Reynolds numbers. Both spectra show a rupture of the
the total length of the rods. The span wise extension of the integration slope at about 50 Hz for Re = 5010 and at about 100 Hz for
domain has been shortened in order to suppress the influence of the Re = 10300. Such slope ruptures are characteristic for the spectral force
lateral walls on the force coefficients. These walls are located at z = 0 m coefficient of rod bundles in cross flow (Axisa, 2001).
and z = 0.08 m. The resulting cylinder axial length hc represents only
the central 75% of the total length of the rod. The mean inlet velocity of
the jet is used as reference velocity Uc. Lift and drag coefficients are 5. LES analysis of baffle jetting at reactor conditions
averaged in time over 2 s (from 3.5s to 5.5s). The resulting mean force
coefficients lift and drag are summarized in Table 1. It is interesting to When baffle jetting occurs, two flow directions that are perpendi-
note that lift and drag diminish with increasing Reynolds number. cular to each other, are superimposed: the main flow in the reactor
Fig. 13a and b shows the spectra of the force coefficients lift and drag core, which is aligned with the fuel rods and the cross flow initiated by
for the Reynolds numbers Re = 5010 and Re = 10300, respectively. In the baffle jet (see Fig. 1). Although related experiments for reactor

Fig. 10. Time averaged velocity magnitude in two planes: close to the front wall in (a) and close to the bottom wall in (b).

38
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 11. Comparison of transverse profiles (0 < y < 0.013 m) of the time averaged stream wise velocity at two Reynolds numbers and two locations: between front
wall and 1st rod (x = 0.001825 m) and between 1st rod and 2nd rod (x = 0.01325 m).

conditions exist (Boulanger et al., 1997; Drumont et al., 2018), ex- defined, which is shown in Fig. 14, to particularly treat the turbulent
perimental data have not been published. Thus, the complex flow in- flow close to the fuel rods n°1, n°2 and n°3 (see Fig. 14b). The calcu-
teraction of a corner baffle jet with fuel rods and reactor core main flow lation domain was inspired by the numerical analysis of the AGATE
is analysed here only numerically. A simplified calculation domain was experiments presented by Bieder et al. (2014). The numerical model is

Fig. 12. Distribution of the time averaged pressure in Pa: field in the span wise mid plane close to the inlet channel exit in (a) and pressure course around the corner
rod in (b).

39
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Table 1 treated as walls, on which Reichardt's wall function is applied (for de-
Calculated mean force coefficients lift and drag on the 1st rod. tails, see Bieder and Costa (2018)). To account for the predominant
Re = 5010 Re = 10300 axial flow inside of the corner assembly, symmetry conditions are used
at the boundaries inside of the assembly (called assembly faces), which
Force coefficient Force coefficient Force coefficient Force coefficient are located on the opposite sides of the core baffles. The symmetry
lift drag lift drag condition will significantly amortise turbulent fluctuations at these
4.12 5.57 3.17 4.83 boundaries. However, this drawback was accepted to allow the flow to
accelerate in axial direction inside the corner assembly, away from the
baffle walls, to compensate flow deceleration close to the baffle walls
identical to that presented in chapter 2. This model has been validated due to the wall friction. It was verified by increasing the calculation
by the single rod test case (chapter 3) and the PANACHET experiment domain from 5 × 5 to 6 × 6 rods, that the presence of the symmetry
(chapter 4), respectively. planes has no effect on the turbulence characteristics close to the target
A fuel assembly without mixing grids of 5 × 5 fuel rods is con- rods n°1, n°2 and n°3 (Fig. 14b). The thermal hydraulic conditions are
sidered, located in a core baffle corner. An axial assembly length of 40 also given in Fig. 14c that lead to the following constant physical
dh is modelled. The flow rate in the assembly is imposed at the assembly properties of the fluid: density ρ = 644 kg/m3 and kinematic viscosity
inlet plane. In order to create turbulent fluctuations at this location, a ν = 7.5 × 10−5 m2/s.
periodic precursor-domain, called periodic box, of an axial length of 8 dh For the precursor periodic box, Fig. 15a shows the distribution of the
was created upstream of the fuel assembly. Periodic box and fuel as- time averaged velocity magnitude in colour scale with superimposed
sembly are shown schematically in Fig. 14a. The front face of the core velocity vectors. The velocity vectors are projected onto a horizontal
baffle is transparent for visual reasons. The periodic box has the same plane (normal to the z-axis). Both, the deceleration of the flow in the
cross section as the fuel assembly. Starting from an initially constant corner formed by the core baffles and the acceleration of the flow in the
axial velocity of 5.35 m/s, a transient in the periodic domain was si- symmetry planes are good visible in the figure.
mulated until laminar/turbulence transition occurred and the turbulent Secondary flow structures due to anisotropic turbulence do not
fluctuations were stabilized. The initial flow rate in the box was con- develop under the used boundary conditions in the sub channels of the
served by adjusting at each time step a momentum source in form of rod bundle as observed in 5 × 5 rod bundles enclosed by walls (Bieder
linear pressure gradient. The instantaneous velocity, calculated at each et al., 2014, 2015). Secondary flow structures develop only in the core
time step at the periodicity plane of the periodic box, is imposed at each baffle corner and close to rod n°1. Arrows mark these vortices in
time step at the assembly inlet plane. The used time step of about Fig. 15b, which is a zoom on rod n°1 in the corner of the core baffle.
2 × 10−6 s respects the Courant-Friedrichs-Levy stability criteria. Fig. 16 shows the velocity distribution in a vertical plane, which is
A zoom on a section of the assembly is shown in Fig. 14b, this oriented parallel to the bottom face of the core baffle (see Fig. 14). The
section is marked in Fig. 14a. The diameter of the rods and pitch-to- vector in the lower left corner of Fig. 15b indicates the baffle jet di-
diameter ratio are identical to those in the PANACHET test. The dis- rection and thus also specifies the orientation of the plane. The distance
tance between rods and baffle walls is 0.0035 m. The baffle jet has a of the visualisation plane to the baffle wall (y-direction) is 0.0005 m;
width of 0.001 m and is injected tangentially to the core baffle called this means that the plane is placed in the centre of the baffle jet. Both,
bottom face. The jet forms a wall bounded jet when developing along the instantaneous and time averaged values of the velocity magnitude are
core baffle. Since the gap flow is considered as the flow through a small shown in colour scale in Fig. 16a and b, respectively. Instantaneous
orifice, which suppresses large scale turbulent fluctuations, a spatially velocity vectors, which are projected onto the visualisation plane, are
and temporally constant flow velocity of 5 m/s is assumed at the baffle shown in Fig. 16c. The deflection of the baffle jet in axial flow direction
jet inlet; velocity fluctuations are not imposed. The lower boundary of is good visible in the vector plot. Small scale turbulent fluctuations in
the gap between the intersecting baffle plates (core baffle gap) is lo- the baffle jet and the main assembly flow are visible in the in-
cated 0.012 m downstream of the assembly inlet plane. Fig. 14c shows stantaneous velocity contours. A zone of very low velocities in the z-x
the applied boundary conditions. The 5 × 5 rods model-assembly in- plane (high velocity in y-direction) separates the baffle jet from the
tends to represent the corner rods in a 17 × 17 rods corner assembly axial flow region. This separation is good visible in the time averaged
(see Fig. 1). Therefore, front- and bottom face of the core baffle are values of the velocity as well as in the vector plot.

Fig. 13. Spectra of the force coefficients lift and drag: Re = 5010 in (a) and Re = 10300 in (b).

40
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 14. Description of the assembly model: Calculation domain with the precursor periodic box in (a), section of the assembly model in (b) and boundary and
thermal-hydraulic conditions in (c).

Fig. 15. Time averaged velocity magnitude and velocity vectors projected onto a horizontal plane normal to the z-axis in the periodic box: total view in (a) and zoom
to the corner rod in (b).

41
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 16. Velocities in a vertical plane located in parallel to the bottom face of the core baffle: Magnitude of time averaged values in (a), instantaneous values in (b)
and instantaneous velocity vectors in (c).

Fig. 17. Time averaged velocity magnitude and velocity vectors in horizontal planes normal to z-axis at two distances from the lower boundary of the baffle gap: At
z = 0.035 m in (a) and at z = 0.16 m in (b).

42
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 17 presents the time averaged velocity magnitude as well as 6. Conclusion


velocity vectors for a horizontal plane, oriented normal to the axial flow
direction (z-axis). Two axial distances from the lower edge of the core The risk of baffle jetting exists in pressurized water reactors, which
baffle gap are selected. Lines in Fig. 16a mark the location of the two are designed with a counter flow configuration in the core bypass re-
planes. It is seen from Fig. 17a, at z = 0.035 m (0.023 m above the gion. Baffle jetting can affect fuel assemblies that are located in the
lower boundary of the baffle gap), that the baffle jet penetrates hor- periphery of the core, adjacent to the core baffle (corner assemblies). If
izontally about 0.012 m into the assembly. A significant recirculation a gap between intersecting baffle plates becomes enlarged, the pressure
zone with swirling flow and low velocities is being established near rod drop across the baffle plate creates a water jet impinging on neigh-
n°1. In Fig. 17b, at z = 0.16 m (0.148 m above the lower boundary of bouring fuel rods. In a former CEA experimental program, baffle jetting
the baffle gap), the jet has traversed horizontally through the entire was found to induce early, large vibratory amplitudes of fuel rods at
assembly. Significant cross flow only exists in the baffle jet and does not low turbulence kinetic energy levels and thus failure from fretting wear.
exist in other regions of the assembly. The wall bounded jet spreads in Predicting baffle jetting induced vibrations of fuel rods is still a
crosswise direction with increasing distance from the jet inlet nozzle. As challenge in the field of fluid structure interaction. The present work
the gap between the core baffle plates is very thin (0.001 m), the jet aims to extend the understanding of the unsteady flow in rod bundles to
does not impact on rod n°1 and rod n°2. The baffle jet finally impacts on situations where baffle jetting occurs. The study is thus dealing with
rod n°3. predicting the complicated flow field around a rod bundle in conditions
It is expected that the swirling flow in the recirculation zone near mimicking the real life nuclear reactor geometry with baffle jetting.
rod n°1 and the direct impact of the jet on rod n°3 will significantly Only rigid rods are treated.
influence the pressure forces acting on these rods. Hence, the local force As a first step, the employed numerical model and the LES method
coefficients lift and drag were calculated for two axial distances from are evaluated on the example of a well-documented test case related to
the lower boundary of the baffle gap: z = 0.023 m for rod n°1 and fuel rods in cross flow: the flow around a circular cylinder at
z = 0.148 m for rod n°3. The method to calculate the force coefficients Re = 3900. The proposed LES method reproduces in good accordance
is described in chapter 4. For both rods, the mean jet velocity is takes as high quality measurements in the wake of the rod: velocity mean values
reference velocity Uc. The time history of the force coefficients is shown and Reynolds stresses. In the successive step, the validated modelling
in Fig. 18; on the left side for the corner rod (rod n°1) and on the right method is applied to the CEA PANACHET experiment. This analysis
side for the third rod in jet direction (rod n°3). For rod n°1, Fig. 18a helped to understand better the flow field of a corner baffle jet and
shows a dominant fluctuating frequency of the force coefficients, which further validated the modelling approach. Comparisons of calculated
cannot be observed in Fig. 18b for rod n°3. Higher force coefficients are and measured velocity profiles of the baffle jet showed a good agree-
obtained for rod n°1 compared to those of rod n°3 since rod n°1 is lo- ment. The lack of published data on baffle jets was stressed out.
cated in a zone of important horizontal pressure gradients. However, The flow field in the core baffle corner is finally analysed for rea-
rod n°3 shows higher fluctuation amplitudes. It is interesting to note listic reactor conditions. A plain jet is injected in parallel to a baffle wall
that the dominant force coefficient changes from drag for rod n°1 to lift by a baffle gap of 1 mm width into the axial flow of a corner assembly.
for rod n°3. The previously validated LES method is applied. The analysis of the
The corresponding spectral force coefficients are calculated as de- flow field showed that:
scribed in chapter 4. They are shown in Fig. 19. A dominant frequency
of about 80 Hz is obtained for rod n°1. Rod n°3 exhibits globally higher • Swirling flow exists close to the corner rod. The calculated force
spectral force coefficients, however, without showing a dominant fre- coefficients lift and drag of this rod show a dominant fluctuation
quency. frequency of about 80 Hz for the boundary conditions employed.
The distinguished frequency detected for rod n°1 and the direct • The baffle jet impacts on the third rod in jet direction (passing the
impact of the baffle jet on rod n°3 might explain the rod failures that first two rods without impact). The force coefficients of the third rod
have been detected on these rods due to flow induced early instabilities do not show a dominant oscillation frequency but show increased
(Tekatlian et al., 1999). oscillation amplitudes.

The distinguished fluctuation frequency detected for rod n°1 and the
direct impact of the baffle jet on rod n°3 might explain the rod failures

Fig. 18. Time history of the force coefficients: Rod n°1 at z = 0.035 m in (a) and rod n°3 at z = 0.16 m in (b).

43
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

Fig. 19. Spectra of the force coefficient lift and drag: Rod n°1 at z = 0.035 m in (a) and rod n°3 at z = 0.16 m in (b).

that have been detected for these rods. Berlin-Heidelberg.


Fortin, T., 2006. Une méthode Eléments Finis à décomposition L2 d’ordre élevé motivé
par la simulation d’écoulement diphasique bas Mach. PhD thesis. l'Université de
Acknowledgement Paris VI.
Fujita, K., 2009. Simulation analysis using CFD on vibration behaviours of circular cy-
This work was granted access to the HPC resources of TGCC and linders subjected to free jets through narrow gaps in the vicinity of walls. WIT Trans.
Built Environ. 105, 85–95.
CINES under the allocation A0032A07571 made by GENCI. Fujita, K., Ito, T., Khono, N., 1990. Experimental study on the vibration of circular cy-
linders subjected to cross flow jetted from a narrow gap. J. Fluids Struct. 4 (1),
Appendix A. Supplementary data 99–124.
Gsell, S., Bourguet, R., Braza, M., 2017. Vortex-induced vibrations of a cylinder in planar
shear flow. J. Fluid Mech. 825, 353–384.
Supplementary data to this article can be found online at https:// Gsell, S., Bourguet, R., Braza, M., 2018. Three-dimensional flow past a fixed or freely
doi.org/10.1016/j.pnucene.2019.02.006. vibrating cylinder in the early turbulent regime. Phys. Rev. Fluid. 3 (1), 013902.
Harlow, F., Welch, J., 1965. Numerical calculation of time-dependent viscous in-
compressible flow of fluid with free surface. Phys. Fluids 8 (12), 2182–2189.
References Hirt, C.V., Nichols, B.D., Romero, N.C., 1975. SOLA - A Numerical Solution Algorithm for
Transient Flow. Los Alamos National Lab. Report LA-5852.
Afgan, I., Kahil, Y., Benhamadouche, S., Sagaut, P., 2011. Large eddy simulation of the IAEA-TECDOC-1119, 1999. Assessment and Management of Ageing of Major Nuclear
flow around single and two side-by-side cylinders at subcritical Reynolds numbers. Power Plant Components Important to Safety: PWR Vessel Internals. IAEA.
Phys. Fluids 23, 075–101. Jacques, Y., 1999. Etude de la stabilité de crayons combustibles REP soumis à un
Angeli, P.-E., Bieder, U., Fauchet, G., 2015. Overview of the Trio_U code: main features, V écoulement transverse. PhD thesis. Ecole Central de Lyon.
&V procedures and typical applications to engineering. In: 16th International Topical Jacques, Y., Barbier, D., Boulanger, P., 1997. Flow induced vibrations for PWR fuel rods
Meeting on Nuclear Reactor Thermal Hydraulics, NURETH-16, Chicago, USA. under local jets : resonance amplitude in the stability range. In: Transactions of the
Angeli, P.-E., Puscas, M.-A., Fauchet, G., Cartalade, A., 2017. “FVCA8 benchmark for the 14th International Conference on Structural Mechanics, Reactor Technology (SMiRT
Stokes and Navier–Stokes equations with the TrioCFD code—benchmark session”. In: 14), Lyon, France, August 17-22.
Finite Volumes for Complex Applications VIII - Methods and Theoretical Aspects. Kravchenko, A., Moin, P., 2000. Numerical studies of flow over a circular cylinder at
Springer-Verlag, New York, pp. 181–202. Re=3900. Phys. Fluids 12 (2), 403–417.
Axisa, F., 2001. Vibrations sous écoulements". Modélisation des systèmes mécaniques Kuzmin, D., Turek, S., 2002. Flux correction tools for finite elements. J. Comput. Phys.
(Tome 4). HERMES Science Publications. 175, 525–558.
Bernard, P., Messainguiral-Bruynooghe, C., Cloué, J., Puyal, C., Vincent, C., Baeyens, R., Lourenco, L.M., Shih, C., 1994. Characteristics or the plane turbulent near wake of a
1985. Jet de cloisonnement: phénomènes vibratoires possibles et effet neutroniques circular cylinder, a particle image velocimetry study. In: Beaudan, P., Moin, P. (Eds.),
sur les détecteurs cœur. Prog. Nucl. Energy 15, 261–272. Numerical Experiments on the Flow Past a Circular at a Sub-critical Reynolds
Bieder, U., Costa, D., 2018. Analysis of the heat transfer in a 4x4 rod bundle. Ann. Nucl. Number”. Report Number TF62, Thermosciences Division, Department of Mechanical
Energy 115, 352–366. Engineering. Stanford University, pp. 1–44.
Bieder, U., Rodio, G., 2019. Large Eddy Simulation of the injection of cold ECC water into Lysenko, D.A., Ertesvåg, I.S., Rian, K.E., 2012. Large-eddy simulation of the flow over a
the cold leg of a pressurized water reactor. Nucl. Eng. Des. 341, 186–197. circular cylinder at Reynolds number 3900 using the OpenFOAM toolbox. Flow,
Bieder, U., Falk, F., Fauchet, G., 2014. LES analysis of the flow in a simplified PWR Turbul. Combust. 89, 491–518.
assembly with mixing grid. Prog. Nucl. Energy 75, 15–24. Nicoud, F., Ducros, F., 1999. Subgrid-scale stress modelling based on the square of the
Bieder, U., Falk, F., Fauchet, G., 2015. CFD analysis of the flow in the near wake of a velocity gradient. Flow, Turbul. Combust. 62 (3), 183–200.
generic PWR mixing grid. Ann. Nucl. Energy 82, 169–178. Norberg, C., 1987. Effects of Reynolds Number and a Low- Intensity Free-Stream
Boulanger, P., Fardeau, P., Rigaudeau, J., 1997. FIV of PWR fuel rods under local jets: Turbulence on the Flow Around Circular Cylinder. Publication No. 87/2.
instability threshold and influence of jet configuration. In: Transactions of the 14th Department of Applied Thermodynamics and Fluid Mechanics, Chalmers University
International Conference on Structural Mechanics, Reactor Technology (SMiRT 14), of Technology, Sweden.
Lyon, France, August 17-22. OCTAVE https://www.gnu.org/software/octave/.
Breuer, M., 1998. Large eddy simulation of the subcritical flow past a circular cylinder: Ong, L., Wallace, J., 1996. The velocity field of the turbulent very near wake of a circular
numerical and modelling aspects. Int. J. Numer. Methods Fluids 28, 1281–1302. cylinder. Exp. Fluid 20, 441–453.
De La Llave Plata, M., Lamballais, E., Couaillier, V., May 2017. A discontinuous Galerkin Parnaudeau, P., Carlier, J., Heitz, D., Lamballais, E., 2008. Experimental and numerical
variational multiscale approach to LES of turbulent flows. In: ERCOFTAC Workshop studies of the flow over a circular cylinder at Reynolds number 3900. Phys. Fluids 20,
Direct and Large-Eddy Simulation 11 (DLES11), (Pise, Italy). 085101.
Drumond, J., et al., 2018. Industrial Applications of Framatom's State-Of-The-Art CFD Pope, S.B., 2000. Turbulent Flows. Cambridge University Press, Cambridge.
Methods to Nuclear Reactor Savety Analysis. TOPFUEL conference, Prague, Czech Sagaut, P., 2001. Large eddy simulation for incompressible flows: an introduction. In:
Republic. Scientific Computation Series. Springer-Verlag, Berlin.
Ducros, F., Bieder, U., Cioni, O., Fortin, T., Fournier, B., Fauchet, G., Quéméré, P., 2010. Seki, K., Kuwabara, S., Tanimura, K., Matsumoto, S., Toba, M., 1986. A study on fuel rod
Verification and validation considerations regarding the qualification of numerical vibration induced by baffle jet flow. Nucl. Technol. 74 (1), 27–37.
schemes for LES for dilution problems. Nucl. Eng. Des. 240, 2123–2130. Singh, S.P., Mittal, S., 2004. Flow past a cylinder: shear layer instability and drag crisis.
Ferziger, J., Peric, M., 1997. Computational Methods for Fluid Dynamics. Springer Verlag, Int. J. Numer. Methods Fluids 47 (75–98).
Smagorinsky, J., 1963. General circulation experiments with the primitive equations: 1.

44
U. Bieder and A. Rashkovan Progress in Nuclear Energy 114 (2019) 31–45

The basic experiment. Mon. Weather Rev. 91, 99–164. July 23.
Tanaka, H., Takahara, S., 1981. Fluid elastic vibration of tube array in cross-flow. J. United states nuclear regulatory commission office of inspection and enforcement, 2014.
Sound Vib. 77 (1), 19–37. Licensee Event Report No. 50-339/2014-002-00.
Tekatlian, A., Rigaudeau, J., Aubert, S., Jacques, Y., 1999. Flow induced instabilities of Welch, P., 1967. The use of fast Fourier transform for the estimation of power spectra: a
fuel rods bundles subjected to a plane jet cross flow: experimental and numerical method based on time averaging over short, modified periodograms. IEEE Trans.
analysis. Am. Soc. Mech. Eng. Press. Vessel. Pip. Div. (Publ.) PVP 389, 57–63. Audio Electroacoust. 15 (6), 70–73.
TrioCFD http://www-trio-u.cea.fr/. Zdravkovich, M.M., 1997. Flow Around Circular Cylinders, vol. 1 Oxford University Press,
United states nuclear regulatory commission office of inspection and enforcement, 1980. Oxford, New York, Tokyo.
Fuel Pin Damage Due to Water Jet from Baffle Plate Corner. IE Circular No. 80-17.

45

Vous aimerez peut-être aussi