Page 1 of 1

How to link separate mesh parts?

Posted: Thu May 15, 2014 5:19 am
by fomeh
Hello
Several mesh parts can be joined and coupled together by using an interface boundary domain. I'm confused how to do this in Code_Saturne. I found that it handles hanging nodes very well however i was not able to find how to a good approach to link domains.

Re: How to link separate mesh parts?

Posted: Thu May 15, 2014 11:16 pm
by Yvan Fournier
Hello,

Under the GUI, Calculation Environment / Meshes selection / Face joining.

In the user and theory doc, search foir "join" in the pdf's.

In Code_Saturne, joining (and periodicity, which is a extension) is not handled as a boundary condtion, but as a full mesh preprocessing step. joined faces become interior faces.

Regards,

Yvan

Re: How to link separate mesh parts?

Posted: Sun May 18, 2014 5:13 pm
by fomeh
Thanks Yvan for your prompt reply. It works 8-)

Re: How to link separate mesh parts?

Posted: Thu May 22, 2014 4:59 pm
by fomeh
Yvan I have another problem. I have merged two meshed parts along a curved surface. I have attaced a pic of my early results. As you see, there are some blank spots at the postprocessing stage. It is strange beacause in other planes at the same area they do not exist. Any idea what is the problem? is it a postprocessing bug or a merging problem?

Re: How to link separate mesh parts?

Posted: Sat May 24, 2014 3:06 pm
by Jacques Fontaine
Hello,

This is probably a merging problem. You can open the JOINING.case to see the "joined faces".
Regards,

Re: How to link separate mesh parts?

Posted: Tue May 27, 2014 10:25 pm
by fomeh
Hello Jacques,

In fact when i import it in tecplot, these blank spots disappear! It really seems that there is a problem related to Paraview. I've checked joining.case it has been correctly selected.

Re: How to link separate mesh parts?

Posted: Tue May 27, 2014 11:52 pm
by Yvan Fournier
Hello,

ParaView still seems to have some bugs displaying general polyhedra (which are produced by Code_Saturne's mesh joining when faces are initially non-conforming), but joining in curved sections can be tricky (the theory documentation goes into more details about this, and the best practice guidelines on this website should provide recommendations),

In any case, visualizing the joining (and possible the boundary mesh), checking everything is joined correctly is always recommended, but some checking methods are more "robust" than others. As such, checking the joined faces parts or the boundary parts only require handling polygons, whose display is usually less buggy. Another solution I am not too fond of is to activate splitting of polygons and/or polyhedra in the writer options. I would love getting rid of that option and the corresponding code, but that would require that most visualization tools coold handle polygons and polyhedra well enough, which is unfortunately not yet the case...

Regards,

Yvan