vtkCGNSReader failed to read mesh_NACA0012_inv.su2

[zhang-qiang-github]

Describe the bug

I want to use vtkCGNSReader to read a su2 file. But a bug is reported:

cgio_file_open: ADF 16: Internal error: Memory boundary tag bad.

To Reproduce

import vtkmodules.all as vtk
reader = vtk.vtkCGNSReader()
reader.SetFileName('mesh_NACA0012_inv.su2`)
reader.Update()

Desktop (please complete the following information):
vtk: 9.1.0

Question

Is su2 a standard cgns file?

[pcarruscag]

No, it is not.

[bigfooted]

if you want to read an SU2 mesh in python or paraview, you can use meshio: https://pypi.org/project/meshio/

[zhang-qiang-github]

@bigfooted , Thanks for your reply. But, I find there are some bugs using meshio to read an SU2 mesh.

The example is:

mesh = meshio.read('TestCases\Tutorials\compressible_flow\ActuatorDisk_VariableLoad\propeller_variable_load.su2')
The reported bug is:

Traceback (most recent call last):
File "D:\Anaconda\envs\work\lib\site-packages\meshio_helpers.py", line 103, in _read_file
return reader_mapfile_format
File "D:\Anaconda\envs\work\lib\site-packages\meshio\su2_su2.py", line 48, in read
mesh = read_buffer(f)
File "D:\Anaconda\envs\work\lib\site-packages\meshio\su2_su2.py", line 102, in read_buffer
).reshape(num_verts, dim + extra_columns)
ValueError: cannot reshape array of size 259228 into shape (51846,5)

[bigfooted]

this specific su2 file is not in standard su2 format: it has additional information in it that meshio cannot handle (not sure why, I don't know this test case). You have to use a different case or modify the su2 file according to the standard su file format if you want to read it with meshio

[zhang-qiang-github]

@pcarruscag If I have a mask of vessel, how can I convert it into SU2 by vtk?

[bigfooted]

I do not know what you mean by 'mask of vessel', or converting it into su2 by vtk.
Can you tell us what your final goal is that you want to achieve?

[zhang-qiang-github]

@bigfooted , I cannot upload figure in this reply. Please have a look at the following reply.

[zhang-qiang-github]

The mask means a 3D matrix, and the pixel inside the vessel is 1, that outside the vessel is 0. I can extract the surface of vessel, for example:

The blood will flow through the vessel, and my final goal is to obtain the pressure under the flowing blood.

So, I wonder how to convert the 3D mask or surface into SU2.

[bigfooted]

if you have the point and connectivity information available from the data, you can just write it to a file and that's the entire su2 file format:
https://su2code.github.io/docs/Mesh-File/

[zhang-qiang-github]

But, I have the connectivity information for only the surface. I have no information for the point inside the surface.

So, do you know any tool to generate the connectivity information for the point inside the surface?

[bigfooted]

If you want to actually use the internal points, you need a tool that can do a 3D constrained Delaunay triangulation. Most Delaunay triangulations are 2D, due to theoretical limitations of Delaunay in 3D. Paraview has some capabilities, but it needs a lot of work imo to use the surfaces as constraints. you can maybe use cgal, and there are some matlab tools. Frankly, I think it is easier to just discard the internal points and recreate the mesh from the surface.

[zhang-qiang-github]

Can SU2 accept the surface of 3D object to calculate the pressure?

[bigfooted]

I don't understand what you want to do. The pressure on the surface is always part of the CFD solution and you can get surface information either directly in SU2 or by a postprocessing tool like paraview.