svFSI error, Failed to project faces

[Mattia Zanchi]

Greetings,
I'm attempting to follow the SV tutorial 14 ( FSI with ALE ), everything works correctly until the simulation run.
In particular, I'm getting this error :
"Failed to project faces between: <wall_interior> and <lumen_wall>
Note: The following floating-point exceptions are signalling: IEEE_DENORMAL "

Any suggest to solve this problem ?
I've read on the svForum that may be correlate to the nodeID of fluid and structural mesh but I can't find a solution tho.

Thanks
Regards

[David Parker]

Hi Mattia,

The Failed to project faces between: <wall_interior> and <lumen_wall> error means that there is not a 1-1 correspondence between the wall_interior and lumen_wall meshes.

1) If there is a GlobalNodeID array in the .vtp files then the wall_interior and lumen_wall meshes have node IDs that don't match.

2) If the is no GlobalNodeID array in the .vtp files then he wall_interior and lumen_wall meshes have nodal coordinates that don't match.

Where did you get the mesh for your simulation?

The meshes used in the example of a 3D ALE simulation here https://github.com/SimVascular/svFSI-Te ... 03-pipe_3D is good. Maybe give that a try.

Cheers,
Dave

[Mattia Zanchi]

Hi Dave,
thanks for the reply!
Now I understand the problem, as you suggest in my mesh there isn't a 1-1 correspondence from the global ID point of view between the 2 mesh.
The meshes I use for my FSI simulation is obtained following the SV tutorial 14 son youtube. Any suggest to modify the global IDs from one mesh in order to have a perfect correspondence and let the simulation run correctly ?

I've try to use the mesh here https://github.com/SimVascular/svFSI-Te ... 03-pipe_3D and everything goes well.

I'm working on SV 21.06.15 version on Ubuntu20.04.

Regards
Mattia

[David Parker]

Hi Mattia,

The latest svFSI code does not need the VTK GlobalNodeID data array, if it is not in the mesh files it will determine the node mappings using nodal coordinates. You could write a Python script using VTK to remove this array.

Cheers,
Dave

[Mattia Zanchi]

Hi Dave,
Thanks for your help. As you suggest I've try to remove the GlobalNodeID from both the mesh ( I've done that using Paraview ).
Now when I try to run the simulation the following error appears:

" Program received signal SIGSEGV: Segmentation fault - invalid memory reference. "
" /usr/local/sv/svFSI/2021-09-29/svfsi: line 56: 3495 Segmentation fault (core dumped) $SV_HOME/bin/svFSI $* "

How con I do to fix the issue ?
Now I'm working on SV21.09.29 version on Ubuntu20.04 with the last release of svFSI.


Regards
Mattia

[David Parker]

Hi Mattia,

Maybe there are other problems with your meshes. Please upload your projects svFSI directory someplace I can download it and I'll have a look.

Cheers,
Dave

[Mattia Zanchi]

Hi Dave,
The SV project is available here: " https://drive.google.com/drive/folders/ ... sp=sharing "

Thx for your patience
Regards
Mattia

[David Parker]

Hi Mattia,

It seems that GlobalNodeID is needed for .vtp files but not for .vtu files, sorry for misleading you on this!

I went through the FSI Tutorial and got the same error you did. The problem is that the node IDs don't match between the fluid and solid interface. The SV boundary layer meshing code is not computing the node IDs correctly. I will investigate.

Cheers,
Dave

[Mattia Zanchi]

hi Dave,
Let me know if you will find a solution.

Regards
Mattia

[Sara Zambon]

I have the exact same problem!

And also here: viewtopicPhpbb.php?f=188&t=10837&p=30045&start=0&view=

Hoping to hear some good news soon ..

Sara