> >
> > Exact, I haven't done any coarsening but I have a lot of h refinement
> levels.
> > The problem arise after some refinement steps.
> > I'm running an INS solver on the lid-driven cavity problem, the mesh
> > is generetad
> > using MeshTools::Generation::build_square ().
> > Unfortunately my application isn't a short sample code...
> >
>
>  It's not possible that you've refined to the point where the mesh has
>  torn, is it?  It used to be the case that around edge lengths of
>  10^-6, floating point tolerances would make it impossible for us to
>  keep the refined mesh correctly connected; AFAIK it's still the case
>  that that can happen around edge lengths of 10^-15.
>  ---


I see, this could be a good explanation.
In this case would I see some connectivity problems in the mesh or the
problem would be only at the dof level?
What AFAIK is?


>  That sounds ridiculous, but on the lid-driven cavity it's not impossible to 
> pick
>  an error estimator and refinement scheme that decide to refine only
>  in the corners over and over and over again.

Yes, this is true.

Lorenzo

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to