Mesh CGNS coordinates, dimensions and error

[BertR]

Dear,

I import CGNS meshes into code_saturne. The CGNS meshes are made with ANSYS meshing. I wonder which dimensions I should use in the meshing application? Apparantly, the coordinate positions in code_saturne are automatically in the dimensions chosen for the mesh?

If I for instance import a mesh that is made in [mm] with a max size of 1500 mm and I perform a "check mesh" the max coordinate extent is 1500. Hence, it takes over the [mm] convention. Is this a risk for the boundary conditions? I.e. a flux should be given in W/m^2. However, the coordinates now are in mm so can this form a problem?

Then, if I try to import a mesh that is made in [m] with a max size of 1.5 [m] I get the following error:
> index 1, label 'GridCoordinates_t' not a child of 'Zone_t'

What does the latter mean?

Kind regards,

Bert

[Yvan Fournier]

Hello,

You can scale the mesh either using some CGNS tools (in the standard CGNS distribution), or using cs_user_mesh_modify (see example) in cs_user_mesh.c.

I see no reason why scaling the mesh under ANSYS meshing should lead to an error, but I have seen metadata referencing non-existing data CGNS files output by ICEM/ANSYS meshing before, so that might be a bug on their side.

Regards,

Yvan

[BertR]

Dear,

Thank you for your quick reply. The difference with the [m] and [mm] in the ansys meshing was caused by a mistake on my side. In the ANSYS meshing tool it is important to give a physics and solver preference (CDF/Fluent) in order to not have midside nodes and to have cell-centered elements. That's why the error occured. Now I have succesfully loaded a mesh in [m] dimensions and the maximum mesh coordinated of the "check mesh" is 1.5 m as required. I hope now the boundary condition given in W/m^2 gives a correct influence.

However, I still have another question:
In ANSYS meshing in the export options it is possible to choose HDF5 e.g. v3.1, binary (or ASCII) or one can chose to have ADF.
From the user guide it is not clear to me whether there is a preference: binary vs ASCII, ADF vs HDF5, v2 or v3?
http://code-saturne.org/cms/sites/defau ... 0/user.pdf

Kind regards,

Bert

[Yvan Fournier]

Hello,

I would start with HDF5/V3.0, but all should work (with ANSYS meshing 12 and 13, there were issues due to ANSYS lying in the file about the CGNS version, but things are fixed with ANSYS meshing 14 and above).

Regards,

Yvan