Installing CS 4.2.1 and Syrthes 4.3.0 on Xubuntu 15.10

All questions about installation
Forum rules
Please read the forum usage recommendations before posting.
karlv

Re: Installing CS 4.2.1 and Syrthes 4.3.0 on Xubuntu 15.10

Post by karlv »

Hi jb_l

I have been following this post as I'm in the same boat as you. Have you managed to couple the two codes? I can also run both saturne and syrtes in serial and parallel but I am not able to run a coupled simulation. I believe it has something to do with my syrthes installation.

If I run a syrthes serial simulation I can view all the results time steps in paraview. If I run a parallel syrthes simulation I can only view the final result (although still correct). It does not seem that syrthes is rebuilding all the results. My guess is that this is causing my coupled run to fail as there is no intermediate result to send to code_saturne.

Do you maybe have a similar issue?

Many Thanks
Karl
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: Installing CS 4.2.1 and Syrthes 4.3.0 on Xubuntu 15.10

Post by Yvan Fournier »

Hello,

What type of error do you have running the coupled simulation ? Dis you install Syrthes with the same MPI library as Code_Saturne, and Code_Saturne's PLE library ?

Did you do the post-install stage in Code_Saturne to tell it where is Syrthes ?

In any case, your postprocessing issue has no relation to the couping issue, as the codes communicate through MPI, not through postprocessing files which would be very expensive to read/write at every time step...

Regards,

Yvan
karlv

Re: Installing CS 4.2.1 and Syrthes 4.3.0 on Xubuntu 15.10

Post by karlv »

Hi Yvan

I have attached the 2 log files from the coupled run. From the listing file it seems that code_saturne can not map the result to the Syrthes boundary? I assume 3d fluid to 2D solid is possible.

Not sure what you can make out from the logs. It looks like everything else is ok.

Regards,
Karl
Attachments
syrthes.log
(7.34 KiB) Downloaded 327 times
listing.log
(31.7 KiB) Downloaded 328 times
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: Installing CS 4.2.1 and Syrthes 4.3.0 on Xubuntu 15.10

Post by Yvan Fournier »

Hello,

Did you require the 2D projection in the Code_Saturne side ?

Also, did you start from the example XML file, or recreate if from scratch using the tutorial ?

I seem to recall some other user on this forum had similar issues a few months ago, and discovered that the tutorial text itself is incomplete and missing some references for faces that need to be coupled (we need to update it, but have lacked the time to do it). Starting from the XML file, even though less interesting from a "training" point of view, should help.

You can also try to use visualization of boundary postprocessing output (on both codes, though I am less familiar with that from Syrthes) to find which references should be coupled.

Once this works, you'll probably have a crash at the end of the computation, as the Syrthes team has not yet released a public version with a fix for a simple bug (for which you will find the fix on this forum checking older posts relative to this tutorial)... (though I didn't check in the last weeks, and mentioned the issue to them, so they might have posted a fix since).

Regards,

Yvan
karlv

Re: Installing CS 4.2.1 and Syrthes 4.3.0 on Xubuntu 15.10

Post by karlv »

Success!

I build a 3D model (coupling walls of a house to the volume inside) and got it working. See attached picture.

As far as the 2D model.. I did build it from scratch but after opening the Syrthes example (solid-coupling.syd) after you mentioned it I saw what might have been the error. In the tutorial the boundary under conjugate heat transfer only shows a 1 (instead of 1,2, etc.). And on the Saturne side the boundary is actually called 1. So I thought you specify the fluid boundary number in both Syrthes and Saturne. But now I realize that you specify Syrthes coupling boundary in Syrthes and Saturne coupling boundary in Saturne.

Now if I can just get Syrthes to write out intermediate results when running in parallel. Currently I have to run the Syrthes part of the simulation in serial and the Saturne part in parallel. Otherwise I only get one Syrthes result.

Thanks Karl
Attachments
result.png
Post Reply