[Nek5000-users] Representing Curved Side in any plane

[nek5000-users at lists.mcs.anl.gov]

Hi Michael, Markus,

There will of course be a threshold, as there are limits to the amount
of acceptable deformation.

I don't know how best to advise you, other than to start with something
as simple as possible, so you can gain some familiarity.

You should also note that you need to deform all shared internal
edges (typically 4, but could be more or less) in the same way if
your mesh is to be consistent.

Ideally, if you had a way to generate 20-noded brick elements
then these midside nodes would correspond to the 12 edge 
vertices (in some funny ordering).

If I have an idea of what you're after I might be able to help
you further.

Paul



On Thu, 8 Apr 2010, nek5000-users at lists.mcs.anl.gov wrote:

>
>
>
>
> Hi Paul,
>
>
>
> We tried a few simple geometries with the new midpoint feature and did have
>
> some success, however we are also experiencing a new learning curve with the
>
> feature and thought you could shine some light on it.
>
>
>
> We tried with only one element and have been successful.  We then tried with
>
> many elements, including all midpoints in the curved side section, and we
>
> received the "vanishing jacobian" error. Next, we tried removing from the curved
>
> side section; internal edges. So, any midpoints on edges that were "interior" to the
>
> geometry.  My best way to explain this would be midpoints which lie on edges that
>
> Â do not see a boundary (meaning the edge is shared by elements on all sides.)Â  This
>
> Â left us with only midpoints on the outer surface of the geometry, which we were able
>
> to deform quite nicely. However, we did reach a threshold at which point
>
> we got the "vanishing jacobian" again.
>
>
>
> So this brings up two questions:
>
>
>
> 1.   Is there no support for internal edges, meaning only midpoints on the surface
>
> of the geometry can be deformed? (If so, I think that is not an issue but good
>
> information to have).
>
>
>
> 2.  Is there a threshold distance from the midpoint to the undeformed edge for it
>
> to work?  Markus ran a case where the following worked:
>
>
>
> 10 Â 1 Â  0.75000 Â  Â  -0.090000 Â  Â  Â 0.750000 Â  Â  Â  0.00000 Â  Â  Â  0.00000Â Â Â Â Â Â Â  m
> Â  9 Â 3 Â  0.75000 Â  Â  -0.090000 Â  Â  Â 0.750000 Â  Â  Â  0.00000Â Â Â Â Â Â  0.00000 Â  Â Â Â Â  m
>
>
>
> But, below did not.     
>
> 10 Â 1 Â  0.75000 Â  Â  -0.100000 Â  Â  Â 0.750000 Â  Â  Â  0.00000 Â  Â  Â  0.00000Â Â Â Â Â Â  m
> Â  9 Â 3 Â  0.75000 Â  Â  -0.100000 Â  Â  Â 0.750000 Â  Â  Â  0.00000Â Â Â Â Â Â Â 0.00000 Â  Â Â Â  m
>
>
>
> Thanks for any insight you can provide with this and the new feature!
>
>
>
> Regards,
>
> Michael
>
> ----- Original Message -----
> From: nek5000-users at lists.mcs.anl.gov
> To: nek5000-users at lists.mcs.anl.gov
> Sent: Monday, April 5, 2010 10:17:21 PM GMT -06:00 US/Canada Central
> Subject: Re: [Nek5000-users] Representing Curved Side in any plane
>
>
>
>
> Hey Paul,
>
>
>
> Sounds great! And I think this is exactly what we needed! We will try it tomorrow.
>
>
>
> Thanks again,
>
> Michael
>
>
> ----- Original Message -----
> From: nek5000-users at lists.mcs.anl.gov
> To: nek5000-users at lists.mcs.anl.gov
> Sent: Monday, April 5, 2010 9:29:33 PM GMT -06:00 US/Canada Central
> Subject: Re: [Nek5000-users] Representing Curved Side in any plane
>
> Hi, I've just added the midside-node support to the current svn repo for nek.
>
>
>
> Sorry - I thought I had upgraded this in October but apparently had not committed to
>
> the repo then. Hopefully all should work -- I retested my original benchmark w/ this new
>
> version and it is functioning. Please let me know if this now works for you.
>
> You should be able to modify any one of the 12 edges.
>
>
>
> Paul
> _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users