Learning Forum

[miratip]

I am trying to run Fluent on my school's supercomputer with multiple cores. When I do this, however, the results are somewhat erroneous. Essentially, there are certain grid points where the solution is zeroed out (shown in the figure). I am assuming these locations are where the domain is split up between the cores. This issue does not occur when I run only using one core.

Anyone seen this before or have any idea how to fix this?

Thanks.

 

 

[Rob]

Are the results showing the problem in Fluent too? Just trying to narrow down if it's the data or whatever you're exporting & reading into. 

[miratip]

Thanks for your response. I will spend some time trying to figure out how to assess my results in Fluent and get back to you.

[Amine Ben Hadj Ali]

Probably issue with the export. Try to assess the results inside Fluent first.

[miratip]

Thanks for your response. I am not using the Fluent GUI so I'll have to spend some time figuring out how to assess my results within Fluent. 

Is there a reason why the export would work fine with one core but have issues with multiple cores?

[Amine Ben Hadj Ali]

Bugs are always possible even in the post-processing module you are using. Check the results in Fluent first and let's know if you see issues there too. If yes then it has nothing to do with export!

[miratip]

Ok, the results look fine within Fluent so it seems the problem is with the export. Any ideas what could be going on?

[Rob]

Which export option are you using? Check the export limitations, and also check how the post processing tool handles duplicate points. 

[miratip]

Here is the command I use to set up the export:

/file/transient-export/ascii result_file () x-velocity y-velocity pressure phase-water-vof () no no export1 time-step 10 time-step

How would I check the export limitations? Also for post-processing I am just using a script that I wrote. I don't think there are duplicate points being exported because the number of nodes written out matches the number of nodes in my mesh.

[Amine Ben Hadj Ali]

Your Export Format? ASCII?

[miratip]

Yes, ASCII. Would a different format work better?

[Amine Ben Hadj Ali]

If the values from ASCII Exports do make a sense then I will suspect the post-processor of these values given the root values from Ansys Fluent are correct. Have you tried with another format your tool does sustain?

[miratip]

The values from the ASCII exports don't make sense; the solution is zeroed out at certain points resulting in the weird structures seen in the image in my initial post. This is the crux of my issue.

[Amine Ben Hadj Ali]

Can you reproduce the problem on a dummy example so that we are able to do the same on our side? Is your case 2D planar?