Kratos: [PFEM Fluid] Calculation runs but not printing results

[jginternational]

Pfem fluid application in windows.

The calculation output info seem to be correct but no results or mesh is written
Result files are created but empty

[AlejandroCornejo]

AAAh! I know that... I created an issue some time ago about this. It seems that in Windows, if you have two instances of GiD it does not work properly

[AlejandroCornejo]

you can check it by changing the PfemGiD with the standard GiD output and it works fine

[philbucher]

@jginternational can you try with #5653?

[jginternational]

@AlejandroCornejo What do you mean with "2 instances of GiD" ? GiDPost to write results or GiD the Gui?

[philbucher]

@jginternational I think #5653 and #4337 explain the problem and the proposed solution

[maceligueta]

More related to #3637, I suspect.

El 30 oct 2019, a las 12:11, Javi Gárate [e-mail] escribió:


So @philbucher I understand that the GiDIO class should be some kind of Singleton. I'll try something with that PR this afternoon

—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub, or unsubscribe.

[AlejandroCornejo]

More related to #3637, I suspect.
…
El 30 oct 2019, a las 12:11, Javi Gárate @.*> escribió:  So @philbucher I understand that the GiDIO class should be some kind of Singleton. I'll try something with that PR this afternoon — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub, or unsubscribe.

I think that is someting related to what @maceligueta says

[AlejandroCornejo]

@jginternational If you modify the file applications\PfemFluidDynamicsApplication\python_scripts\pfem_fluid_gid_output_process.py as: self.body_io = GidIO( the postprocess works :)

[jginternational]

@AlejandroCornejo I've tested by changing that file into:

        if self.body_output or self.node_output:
            self.body_io = GidIO( self.volume_file_name,
                                    self.post_mode,
                                    self.multifile_flag,
                                    self.write_deformed_mesh,
                                    self.write_conditions)

        if self.skin_output or self.num_planes > 0:
            self.cut_io = GidIO(self.cut_file_name,
                                self.post_mode,
                                self.multifile_flag,
                                self.write_deformed_mesh,
                                WriteConditionsFlag.WriteConditionsOnly) # Cuts are conditions, so we always print conditions in the cut ModelPart

And it works in Windows

So, we need to think about this solution for the release (Since it's the only one working both in windows and linux (Hope on mac also))

[philbucher]

@jginternational have you tried with #5653?
If this works then we could push for it to be merged soon in order to be included in the release

[maceligueta]

We did an override of GiDIO because it was not sufficient for the interest of PFEM post-process, I don't remember the exact reasons and I don't have time to check them today, but I will do on monday. @AFranci requested the changes, so he should remember as well.

[maceligueta]

@AFranci just told me that the flying nodes are not in a separate layer. The PFEM GidIO puts them in a separate layer, so they can be turned off.

[AFranci]

Hi all,
First of all, thank all of you for working on this!
Those changes were done to allow the visualization of the PFEM 'flying' nodes. These nodes were put in a different group to differentiate them from the rest of the (inner) nodes. I attach two videos to make it more clear:
GidIO + with nodes plotted

GidIO + without nodes

PFEMGidIO + flying nodes

So, it is clear that the PFEMgidIO allows a better post-process, however, being this change not so crucial and while we don't find a solution for the windows problem, I think we should use the core GiDIO for this release version.

[jginternational]

@philbucher I've tested the branch you pointed by merging master into that one, leaving the 'pfem_fluid_gid_output_process.py' script as it was originally and it did not compile for windows. I had to remove the KRATOS_API from the GidIO class

class KRATOS_API(KRATOS_CORE) GiDIOBase : public IO {

    protected:
    static int msLiveInstances;
};

template<class TGaussPointContainer = GidGaussPointsContainer, class TMeshContainer = GidMeshContainer>
class GidIO : public GiDIOBase

After removing the KratosAPI, it compiled, but the output was still empty:

[jginternational]

OK!

With this branch it's now working.
KratosMultiphysics/Kratos#5653

Ready to merge @maceligueta