Results for case Channel Flow

Code: Code_Saturne

Version: 2.0-rc1

Authors: J. Uribe

Method and Numerical Options

The Reynolds number simulated is %$Re_{\tau}=921$% based on the channel mid-height %$\delta$% and the friction velocity %$u_{\tau}$%. The calculations are periodic in the stream and span wise directions. The friction velocity is imposed by setting a source term in the stream wise velocity equation which drives the flow. %$\frac{\partial p}{\partial x}=-\frac{\rho u_{\tau}^2}{\delta}$%.

The set up has been done using the GUI.

  • RANS: Steady solver for all cases. Default options. Turbulence models: k-eps, SST, phi-f, SSG

  • LES: Unsteady solver. Centred scheme. Subgrid models: Smagorisnky, Dynamic and WALE

In addition, averages calculation have been done for LES simulations as well as restarting with averages.

Models

k-eps, SST, phi-f, SSG, LES-Smag, LES-Dynamic, LES-WALE

Mesh

Three 1D meshes (in NOPO format, from the Simail meshing tool) have been used for the RANS study:
  • The high Re mesh (canal_HRE.des) has 46 cells and a non-dimensional wall distance y+=20 on the first cell.
  • The low Re mesh (canal_BRE.des) has 100 cells and y+=1, refined near the walls.
  • The fine low re mesh (canal_BREF.des) has 1842 cells uniformly distributed and y+=0.5.

For the LES computation the 3D mesh has 356352 cells and a %$y^+=20$%

Description of the results files

The results files are in ASCII format. The data is organised in columns as:
y+, U+, k+, eps+, nut+

There is a file for each calculation and the name of the file gives the mesh and the model, e.g. HRE_KEPS_profiles.dat contains the results for k-eps models on the High Reynolds mesh.

The usproj.f90 subroutine is used to compute non-dimensional profiles.

For the LES results, the files are called "averages_XXX.dat" where XXX is the LES model used. The files have been produced by averaging in time and in space. The files contain the following columns:

y/%$\delta$%,<U+>,<V+>,<W+>,<uu+>,<vv+>,<ww+>,<uv+>,<uw+>,<vw+>,<p+>,<du/dy+>

Boundary conditions

For RANS:

For the high Re meshes the standard wall function is used. For the low Re meshes scalable wall function is used for k-eps and SSG models (IDEUCH=2)

For LES:

Wall functions are used at the walls (%$y^+ = 20$%). Periodicity in the stream and span wise directions

Reference Publications

RANS RESULTS

The results are in good agreement with the previous version (see figure 1). In figures 2 and 3 the results for velocity and kinetic energy with the high Reynolds mesh are shown. All models use the standard wall function. The velocity profiles are close to experimental values. The levels of kinetic energy are also correct except near the wall where the peak is not captured by the wall function.

In figures 4 and 5, the results with the use of the scalable wall function are shown. They are similar to ones obtained with the standard wall function except for the SSG model which slightly underestimate the velocity. As expected, the additional points below %$y^+=11$% do not imply the resolution of the viscous sublayer.

Figures 6 and 7 show the results with the models that resolved the flow down to the wall. The velocity profiles are slightly underestimated by the %$\varphi-f$% and SSG models and overestimated by the SST and %$k-\varepsilon$%. The peak in the kinetic energy is only reproduced by the %$\varphi-f$% model.

U_v1.4_2rc1.png

Figure 1. Velocity profiles comparisons between version 2.rc1 and 1.4 with the high Reynolds mesh.

U_hre.png

Figure 2. Velocity profiles for the high Reynolds mesh

k_hre.png

Figure 3. Kinetic energy profiles for the high Reynolds mesh.

U_bre.png

Figure 4. Velocity profiles for the low Reynolds mesh.

k_bre.png

Figure 5. Kinetic energy profiles for the low Reynolds mesh.

U_bref.png

Figure 6. Velocity profiles for the fine low Reynolds mesh.

k_bref.png

Figure 7. Kinetic energy profiles for the fine low Reynolds mesh.

  • usproj.f90: subroutine to calculate non-dimensional values and create output file

  • ustsns.f90: subroutine to impose the pressure gradient

LES RESULTS

The averaged profiles for velocity and stresses are shown in figures 8 to 12. All models yield similar results. They overestimate the velocity due to the coarse resolution of the mesh and the use of the wall function. The stresses are predicted reasonably well with the standard Smagorinsky model giving large levels of fluctuations in the wall normal direction (%$v'v'^+$%).

U_LES.png

Figure 8. Velocity profiles.

uu_LES.png

Figure 9. %$u'u'^+$% profiles.

vv_LES.png

Figure 11. %$v'v'^+$% profiles.

uv_LES.png

Figure 12. %$u'v'^+$% profiles.



Current Tags:
Saturne_Validation_2.01Add my vote for this tag create new tag
, view all tags
Topic attachments
I Attachment Action Size Date Who Comment
elsedat BREF_KEPS_INV_profiles.dat manage 154.7 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat BREF_KW_profiles.dat manage 154.7 K 2010-06-25 - 08:00 UnknownUser RANS results
elsedat BREF_SSG_INV_profiles.dat manage 154.7 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat BREF_V2F_profiles.dat manage 154.7 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat BRE_KEPS_INV_profiles.dat manage 8.4 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat BRE_KW_profiles.dat manage 8.4 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat BRE_SSG_INV_profiles.dat manage 8.4 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat HRE_KEPS_profiles.dat manage 3.9 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat HRE_KW_profiles.dat manage 3.9 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat HRE_SSG_profiles.dat manage 3.9 K 2010-06-22 - 15:26 UnknownUser RANS results
elsedat averages_dyn.dat manage 6.5 K 2010-09-06 - 10:26 UnknownUser LES results with Dynamic Smagorinsky model
elsedat averages_smag.dat manage 6.5 K 2010-09-06 - 10:25 UnknownUser LES results with Smagorinsky model
elsedat averages_wale.dat manage 6.5 K 2010-09-06 - 10:26 UnknownUser LES results with WALE model
xmlxml bre_keps_inv.xml manage 6.7 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bre_kw.xml manage 6.7 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bre_ssg_inv.xml manage 6.9 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bre_v2f.xml manage 6.8 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bref_keps_inv.xml manage 6.8 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bref_kw.xml manage 6.9 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bref_ssg_inv.xml manage 6.9 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml bref_v2f.xml manage 6.9 K 2010-06-25 - 09:31 UnknownUser xmf setup files
elsedes canal_BRE.des manage 20.3 K 2010-06-25 - 09:06 UnknownUser channel rans meshes
elsedes canal_BREF.des manage 367.3 K 2010-06-25 - 09:06 UnknownUser channel rans meshes
elsedes canal_BREF2.des manage 3670.0 K 2010-06-25 - 09:06 UnknownUser channel rans meshes
elsedes canal_HRE.des manage 9.6 K 2010-06-25 - 09:06 UnknownUser channel rans meshes
elsegz channel_les.tar.gz manage 293.4 K 2010-09-17 - 15:50 UnknownUser LES set up files
xmlxml hre_keps.xml manage 6.7 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml hre_kw.xml manage 6.7 K 2010-06-25 - 09:31 UnknownUser xmf setup files
xmlxml hre_ssg.xml manage 6.9 K 2010-06-25 - 09:31 UnknownUser xmf setup files
elsef90 usproj.f90 manage 30.6 K 2010-06-25 - 09:33 UnknownUser subroutine to calculate non-dimensional values and create output file
elsef90 ustsns.f90 manage 13.4 K 2010-06-25 - 09:35 UnknownUser subroutine to impose the pressure gradient
Topic revision: r115 - 2018-09-25 - 13:25:56 - ConstantinosKatsamis
 

Computational Fluid Dynamics and Turbulence Mechanics
@ the University of Manchester
Copyright & by the contributing authors. Unless noted otherwise, all material on this web site is the property of the contributing authors.