Open Andres-MG opened 8 months ago
The way 2D cases are handled is very confusing I think. These three attributes: https://github.com/loganoz/horses3d/blob/41111c597d0f88a22d067d586e025df629688482/Solver/src/libs/mesh/HexMesh.f90#L72-L74 are set during the pre-processing stage to a value that depends on the keywords in the control file and the mesh itself. It seems that this is done in this function: https://github.com/loganoz/horses3d/blob/41111c597d0f88a22d067d586e025df629688482/Solver/src/libs/mesh/HexMesh.f90#L1619 but it seems to override the choice of the user in the control file.
I will not merge #158 until this is reviewed, as the characteristic length of the elements is now wrong in 2D cases, and methods depending on it (LES, shock capturing) will provide different results.
I have pinned this issue so we don't forget.
The way 2D cases are handled is very confusing I think. These three attributes: https://github.com/loganoz/horses3d/blob/41111c597d0f88a22d067d586e025df629688482/Solver/src/libs/mesh/HexMesh.f90#L72-L74 are set during the pre-processing stage to a value that depends on the keywords in the control file and the mesh itself. It seems that this is done in this function: https://github.com/loganoz/horses3d/blob/41111c597d0f88a22d067d586e025df629688482/Solver/src/libs/mesh/HexMesh.f90#L1619 but it seems to override the choice of the user in the control file.
I will not merge #158 until this is reviewed, as the characteristic length of the elements is now wrong in 2D cases, and methods depending on it (LES, shock capturing) will provide different results.