code_saturne User's Forum

Hi,
 
I have a problem during the chek mesh when I use one periodic boundary.
 
My mesh is created using gambit and neutral solver.
 
the chek mesh without periodic boundary is ok and when I ask for one periodicity, I obtain the following message:
 
----------------------------------------------------------
Processing periodicity 1 ----------------------------   Periodicity type: pure translation     Translation parameters       - translation vector:      0.00000e+00  2.00000e-02  0.00000e+00   Number of elements before joining     Number of cells                   :  2286953     Number of faces                   :  4694114     Number of vertices                :   441844     Number of boundary faces          :   240416     Number of selected boundary faces :   240416   Joining parameter values:   - Fraction of the length of the shortest edge     connected to a vertex under which 2 edges     intersect and 2 vertices are merged:                1.0000000000E-01   - Value of the dot product of the unit normals of     2 faces above which these faces are considered     to be coplanar:                                     8.0000000000E-01   Search for edge intersections:     Checked possibilities             :    3268465     Intersections found               :    1446671   Subdivision of edges:     Initial number of vertices        :     562052     Initial number of edges           :    3209629     Number of vertices after division :    2292356     Number of edges after division    :    3967189 Warning ======= The intersection merging tolerance as been reduced for 136 edges so as to avoid crossing the local intersection tolerance through transitivity. -> Eventually reduce tolerance for this joining   Merging vertices:     Initial number of vertices        :    2292356     Number of vertices after merging  :     842944     Number of modified vertices       :     470007   Merging of topologically identical edges:     Initial number of edges           :    3967189     Number of edges after merging     :    3778935   Removal of degenerate edges if present      :     Initial number of edges           :    3778935     Number of edges after processing  :    3778918   Removal of degenerate corners if present :     Number of simplified faces        :       6626 Warning ======= Reconstruction problem for 1881 face(s) during joining (159 open cycles, 1702 edges traversed twice, 20 faces split into more than ECS_RC_MAX_FAC_DEC = 100).
  ------------------------------------------------------------------------------------------------------------
 
My geometry is a kind of tube (more complexe).
The periodicity is for the faces on the top and the bottom boundaries (y direction)
 
I don't know where I can reduce the tolerance for the joining?
 
Thank you very much
 
SYlvain

Hello,

You may try to reduce the tolerance, but first, you should look at check_mesh's postprocessing output instead of simply the log file. You should have 2 EnSight case files (or 2 MED or CGNS files), one for mesh quality, one for preprocessor output. If you load the PREPROCESSOR.case with ParaView and use the "Extract Blocks" filter for example (or import the MED file under SALOME), you should be able to see which faces were not joined correctly, and this may help you determine what to try next...

Best regards,
 
Yvan

Hello and thank you for your answer.
In fact, I did something wrong on the definition of the periodicty and I found this error thank to paraview...
 
Now, I have another problem.
 
The chech mesh is ok with the periodic boundary and the simulation run but all the velocitires are equal to zero...
Do you have a tutorial with the version 2 of code-saturne that consider periodic boundary?
I have one periodicity in the Y direction, there is one big face decomposed in many little faces in order to "link" the mesh face by face in gambit.
 
I am not sure to well understand how I have to enter in the GUI of code-saturne which faces are periodic with which other...
 
 
Thank you
 
Sylvain
 
 
 
 

Hello,

Using check_mesh and ParaView, (with the "extract blocks" filter in ParaView), you will be able to see which faces were made periodic, so you may check that you defined Periodicity correctly.

do not believe there is a tutorial with periodicity, but basically, with Periodicity of translation, periodic faces become true interior faces (values are synchronized using the same ghost cell mechanism as for parallelism, with an added translation where necessary).

f your boundary was previously on the faces you made periodic, you may not have an inlet our outlet anymore. In this case, you will need a velocity source term to have nonzero velocity.

Best regards,

Yvan

Hi

my boundary was on the faces I made periodic, and I do not have
an inlet neither outlet.

I initilaze the volume with a velocity of 1 m/s. Is it considered has velocity
source term?

If not, where do I have to add this velocity (or mass flow rate...) term?
 
thank you for all your quick answers
 
Sylvain

Hello,

You can add a velocity source term in any part of the volume using the ustsns user subroutine. I have not run this type of calculation myself, but using a "feedback" system in which you increase the source term if velocity is under the prescribed value and decrease it if it is too high is probably the easiest way to obtain the correct boundary.

Without a source term, the initial velocity will probably dampen and go to zero.

Best regards,

  Yvan Fournier

Hello,

Another example, if you want to simulate a plane chanel flow with periodic boundaries, you have to compute the balance between the pressure force and the friction force, leading to the source term you want to add.

So, in ustsns.f90 you will add an explicit source term (without forgetting to multiply by the volume of the cell) like this: Here, the pressure gradient is balanced by the friction forces, leading to this expression depending on the density, the friction velocity and the half-height of the channel (this is related to the "wet perimeter")

David

Hi all,
 
The first problem about the use of periodicity with the .neu file comes from the boundary layers on the periodic faces.
At the end, I meshed the edges with refinement near the wall then the faces. Don't forget to link the faces that will be periodics.
In pratice, you need only to mesh on face and the other will be meshed automatically.
 
About the subroutine ustsns, this problem is solved here
https://code-saturne.info/products/code-saturne/forums/general-usage/10359587
 
Sylvain