Malformatted vtk data from parallel run

Clearly defined bug reports and their fixes
Post Reply
Martin_
Posts: 11
Joined: 08 Feb 2010, 16:27

Malformatted vtk data from parallel run

Post by Martin_ »

Dear Elmer developers,

I have a simple simulation of flow in a tube (axisymmetric case). I would like to save the output as *.vtu files. However, I'm not able to visualize them. After failing to display the results in both ParaView and Visit and looking at http://www.elmerfem.org/wiki/index.php/ParaView , I checked whether the data in the output files are correct. Scalar variables (pressure in my case) seem to be fine, but for vector variables (velocity field, mesh displacement etc.), I see this:

Code: Select all

        <DataArray type="Float32" Name="velocity" NumberOfComponents="3" format="ascii">
   0.000000E+000   0.419648E-0010.0    0.199380E+000  -0.190733E-0220.0 0.0 0.0 0.0    0.000000E+000   0.419706E-0010.0    0.000000E+000   0.419664E-0010.0    0.000000E+000   0.419774E-0010.0    0.000000E+000   0.419686E-0010.0    0.000000E+000   0.419654E-0010.0 
... 
        </DataArray>
It looks like Elmer doesn't print spaces between the second and third component of the vectors - it 'glues' numbers together. ParaView says there are not enough values to display the velocity field. I fixed the 'velocity' data array manually. After that, it complained about the displacement field (which follows after 'velocity' data array).
Visit was able to read the first block of the mesh (I tried with 2 and with 3 partitions of my mesh), but doesn't load the following blocks ( it doesn't actually read any variables from the second or third file).

I was wondering if you could be so kind and point me to the place in source code where I could fix the output routine. I'm using sources from svn repository and updated to the latest revision.

Best regards,

Martin Vymazal
raback
Site Admin
Posts: 4838
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Malformatted vtk data from parallel run

Post by raback »

Hi Martin

Thanx for taking the effort to find the bug. There indeed was a problem between the 2nd and 3rd component. The 3rd was automatically written to be "0.0 " but the blank space was written on wrong side. In code this may be found at about line 3650 of ResultOutputSolve.src. I already put an update to the svn.

-Peter
Martin_
Posts: 11
Joined: 08 Feb 2010, 16:27

Re: Malformatted vtk data from parallel run

Post by Martin_ »

Dear Peter,

thank you for the quick reply and for the fix. It's very nice to be supported by such responsive developer team :D

Best regards,

Martin Vymazal
Post Reply