Page 2 of 2

Re: Issue with PLE coupling between Code Saturne and Syrthes

Posted: Fri Oct 08, 2021 2:48 pm
by kenneth
Hello Yvan,

Thank you for the suggestion and help. Firstly, I should mention that I have finally been able to test the bug fix 0561b6d7 for the async + reverse coupling (https://github.com/code-saturne/code_sa ... bf397bd2e5.), and I can confirm that it fixed the earlier issue. Thank you and apologies for the delay, I didn't check my inbox and missed your earlier message.

Pertaining to the new issue, I had tried altering the starting value for tolerance and conditional to "allow_nonmatching". If I do set a high enough tolerance (around 5) with allow_nonmatching =false, it does locate with no missed cells/or warnings. However, when I check the location of the solid elements, it is not right. Below, I have attached the expected distribution from a python script and that from the present test (please note, presently I am using the area to check, as this is the easiest way to do so with our modified coupling).
From Python Script
From Python Script
From coupling
From coupling
It would seem with this approach we do not get the warnings any more, but the location is amiss. I have checked the distribution of the area from this test and the one where we activate_nonmatching(and the search extends to infinity), and they are the same. I presume there is still a locating issue, but perhaps buried by using a higher tolerance ? (although I don't quite get how this can be).

Kind regards,

Kenneth

Re: Issue with PLE coupling between Code Saturne and Syrthes

Posted: Sun Oct 10, 2021 10:32 am
by Yvan Fournier
Hello,

Yes, I guess we have a search issue, because in theory, you should get the same results.

I'll need to look into this more in detail. If you do have a smaller portion of the case in which we can analyze this, that would make debugging slightly easier. Otherwise, the case I already have is manageable.

I'll keep you informed.

Best regards,

Yvan

Re: Issue with PLE coupling between Code Saturne and Syrthes

Posted: Sun Nov 07, 2021 1:32 am
by Yvan Fournier
Hello Kenneth,

I found the issue, which is related to an overly aggressive optimization (from September 2019).

With the attached file, things should work as before. I'll place the fix on GitHub soon (but will probably
try to extract a separate Git repository for PLE first, so it may take a few days, so as to do things cleanly).

Best regards,

Yvan

Re: Issue with PLE coupling between Code Saturne and Syrthes

Posted: Mon Nov 08, 2021 11:16 am
by kenneth
Hello Yvan,

Thank you so much for the fix and time you have spent helping us on this. I will try with the file.

Kind regards,

Kenneth