# How to refine the mesh for near wall zone ?

Hi all,

I'm just new here to use salome to generate some mesh. I'm trying a very simple 2D rectangle & 3D parallelepipe simulation. The mesh type will be quadrangle. Now I need to refine the mesh for a near wall zone (parallel with OX along OY for 2D), that's to say, the closer to OX, the more intensive the mesh will be. The idealy effect should be like the attachement. I've tried several operations, but I couldn't reach it. Would anyone tell me how to realise it ? Thank you very much.

Hi

To get this do the following (supposing you have already managed to get a quadrangle mesh):

- in Geom module, do
**New Entity > Explode**to retrieve a 'vertical' (as in the picture) edge. - in Mesh module, create a sub-mesh on this vertical edge. In a contextual menu of your mesh
**Create Sub-mesh**.

Geometry: vertical edge

Algorithm: Wire discretization,

Hypothesis: Number of segments with scale factor = 3. (Or other hypotheses: Arithmetic 1D, Start and end length... )

Add. Hypothesis: Propagation Of 1D hyp ... - Compute

St.Michael

Hello St. Michael,

It works in the 2D model. I'll try to realize it in the same way for 3D model. Thank you very much.

Regards,

Wei

Hi St, Michael,

When I do the examination of the mesh in code saturne, it tells me that there are some warnings such as "1 elements of type quad4 are mis-oriented or highly warped", "The MED mesh contains 730 elements of type seg2 which are ignored by the Preprocessor." and "the mesh does not contain volume elements. We can go no further."

I've checked online, and someone sais that the mesh should have a thickness. Would please show me how to deal with it ?

Thanks very much.

Wei

p, li { white-space: pre-wrap; }

Previously Saint Michael wrote:

Hi

To get this do the following (supposing you have already managed to get a quadrangle mesh):

- in Geom module, do
New Entity > Explodeto retrieve a 'vertical' (as in the picture) edge.- in Mesh module, create a sub-mesh on this vertical edge. In a contextual menu of your mesh
Create Sub-mesh.

Geometry: vertical edge

Algorithm: Wire discretization,

Hypothesis: Number of segments with scale factor = 3. (Or other hypotheses: Arithmetic 1D, Start and end length... )

Add. Hypothesis: Propagation Of 1D hyp ...- Compute
St.Michael

Hi Wei

Can you share your study, either as a hdf file or as a python dump?

To make hdf file fit to 100K limit try to remove all elements using Clear Mesh (http://docs.salome-platform.org/latest/gui/SMESH/removing_nodes_and_elements_page.html#clear_mesh_anchor)

St.Michael

Hi, St. Michael,

Here is the study. Thank you very much.

Regards,

Wei

Previously Saint Michael wrote:

Hi Wei

Can you share your study, either as a hdf file or as a python dump?

To make hdf file fit to 100K limit try to remove all elements using Clear Mesh (http://docs.salome-platform.org/latest/gui/SMESH/removing_nodes_and_elements_page.html#clear_mesh_anchor)

St.Michael

"1 elements of type quad4 are mis-oriented or highly warped"

I checked Min Angle, it is 90 degrees, so no problems here.

"The MED mesh contains 730 elements of type seg2 which are ignored by the Preprocessor."

I think you can ignore this warning

"the mesh does not contain volume elements. We can go no further."

Your mesh is actually 2D. You can make it 3D in one step using Extrusion http://docs.salome-platform.org/latest/gui/SMESH/extrusion_page.html

St.Michael

Hi St. Michael,

I've draw several meshes to try to simulate a steady flow on a smooth solid wall with k-w SST, (the inlet is u=y^(1/7) when y<1; else u=1; v=w=0, upperboundary condition is sliding smooth wall to make a farfield, rou =1 kg/m3, p=101325 Pa, miu=3.33e-5, k=0.000833, omega=0.1318), but the result never goes well. I've checked the boundary conditions and other parameters, the result didn't turn better. It seems that something is still not correct in the mesh. Could you please help me to find out the problem ? Thank you very much. (Note: there is one thing strange when I draw the mesh. As you said, I retrived 4 vertical edge and created sub-meshes on the vertical edges. However, if I use the same hypothesis to generate the sub-mesh, the sections on x=0 and x=150 are reversed. So I have to use different hypothesis to make the horizon mesh lines parallel to OX. (segment 2 and segment 5 in the mesh file)).

Best regards,

Wei

Previously Saint Michael wrote:

"1 elements of type quad4 are mis-oriented or highly warped"

I checked Min Angle, it is 90 degrees, so no problems here.

"The MED mesh contains 730 elements of type seg2 which are ignored by the Preprocessor."

I think you can ignore this warning

"the mesh does not contain volume elements. We can go no further."

Your mesh is actually 2D. You can make it 3D in one step using Extrusion http://docs.salome-platform.org/latest/gui/SMESH/extrusion_page.html

St.Michael

Here are the other 3 meshes.

Previously Saint Michael wrote:

"1 elements of type quad4 are mis-oriented or highly warped"

I checked Min Angle, it is 90 degrees, so no problems here.

"The MED mesh contains 730 elements of type seg2 which are ignored by the Preprocessor."

I think you can ignore this warning

"the mesh does not contain volume elements. We can go no further."

Your mesh is actually 2D. You can make it 3D in one step using Extrusion http://docs.salome-platform.org/latest/gui/SMESH/extrusion_page.html

St.Michael

Hi

It seems that something is still not correct in the mesh. Could you please help me to find out the problem ? Thank you very much.

I see no errors in your meshes. The problem seems to be somewhere else.

(Note: there is one thing strange when I draw the mesh. As you said, I retrived 4 vertical edge and created sub-meshes on the vertical edges. However, if I use the same hypothesis to generate the sub-mesh, the sections on x=0 and x=150 are reversed. So I have to use different hypothesis to make the horizon mesh lines parallel to OX. (segment 2 and segment 5 in the mesh file)).

This effect is caused by different orientation of geom edges. Another way to cope with it than you found is to specify 'Reversed Edges' when you define 'Number of segments' hypothesis with non-uniform distribution.

St.Michael

Is there a way to do the same but without a quadrangle mesh? I mean to refine mesh for near wall zones with triangles for example?

Previously Saint Michael wrote:

Hi

To get this do the following (supposing you have already managed to get a quadrangle mesh):

- in Geom module, do
New Entity > Explodeto retrieve a 'vertical' (as in the picture) edge.- in Mesh module, create a sub-mesh on this vertical edge. In a contextual menu of your mesh
Create Sub-mesh.

Geometry: vertical edge

Algorithm: Wire discretization,

Hypothesis: Number of segments with scale factor = 3. (Or other hypotheses: Arithmetic 1D, Start and end length... )

Add. Hypothesis: Propagation Of 1D hyp ...- Compute
St.Michael