General issues with Code Saturne

[Jean-Marc Blanquies]

Hi,
 
I'm new with Code-Saturne and Salomé. I only made two simple cases in 2D : flow inside a circular pipe and the flow around a backward facing step. Results were quite good so i decided to go on with Salomé/saturne in my CFD project.
 
The main goal is to find the distance between the top of a vertical plate and the place where the flow isn't modificated by the plate.
 
I decided to construct a simple 2D geometry described after in attachments.
 
I meshed my geometry using the 1D algo Nb Segments and the ScaleFactor option for having a first cell with a correct y+.
 
Then, i have imported my med mesh in saturne and check my hypothesys.
 
I used K omega SST with an inlet velocity of 33,33m/s.
 
But, in the end, after the processing my velocity is constant and equal to 33,33m/s in all my volume.
 
Here are my question :

For the geometry :
- Is there a problem with my geometry ?
- Is there a problem with my symetry condition on the bottom of the geometry ?
- Is there an other way to made the plate inside my air flow ?
 
For the mesh :
- Am i correct in considering a first cell at 0,006m of distance with the plate ? (Inlet velocity of 33,33m/s and a y+ near 500 for a 1 layer k omega sst)
- How does the Nb Segment work ?

I assume that as the Space factor is the ratio of the last and the first cell and as the nb of segment correspond to the n number in the Geometric progression (Un=U0*k^n).
In this way k=SpaceFactor^(1/Nb Segment)
and U0 is defined as S*[1-SpaceFactor^(1/Nb Segment)]/(1-SpaceFactor),
 
In my work and assuming this things i determined SpaceFactor and Nb segment to have a first cell on my wall near 0,006m and a last cell near 50 times bigger.
- Do you know a better way for meshing ? In fact should i manage a 1D meshing by choising the first cell dimension and the ratio of the adjacent cell and this cell instead of the ratio of the last and first cell ?
 
For my calculation :
- Is the choice of k omega a good choice in a such example ?
- Should i consider the density as constant or should i manage other assumption ?
- How can i stop a calculation and restart it ? (In the option that i want to look after the convergence before launching a longer calculation)
 
Thanks you very much for your help.
 
Best regards.
 
Jean-Marc Blanquies

[Jean-Marc Blanquies]

I'm sorry i forget to give you my med files and my xml file.

[Yvan Fournier]

Hello,

The problem is that the flat plate you tried to mesh is not represented as a boundary:

You assigned groups to the plate, but the matching faces appear as interior faces (see the preprocessor log).

To have boundary faces with a thin wall, matching faces on each side of the wall must have at least one vertex which is different (they may be geometrically identical, but have a different vertex number). If you merged the 2 faces under Salome, you transformed a boundary to interior faces, albeit with an associated group (which culd be useful for postprocessing: Code_Saturne can't determine that this is a boundary simply because a group is associated, as that group could be useful for interior faces).

So basically, your geometry is a simple rectangle. If you have very high y+ values, the profile will be almost flat, so this probably explains your result.

To solve the issue, you need to ensure that the plate either as thickness or at least uses non-shared vertices on either side (except for the topmost vertex). You should postprocess the mesh with Code_Saturnes mesh checking functionnality before running the calculation. Using transparency or a cut through the mesh, you should see the plate. Otherwise, your mesh is still incorrect.

Best regards,

  Yvan