[Gmsh] very thin layer -> huge number of elements

[Zenker, Dr. Matthias]

Hi,

I have played a bit with the example file since I am also looking for a means to make an anisotropic mesh, and I have some questions.
In the example file, the field is defined as follows

Field[2] = MathEvalAniso;
Field[2].m11 = "1./(0.1)^2";
Field[2].m12 = "0";
Field[2].m13 = "0";
Field[2].m22 = "y+1/(0.005)^2";
Field[2].m23 = "0";
Field[2].m33 = "1/(0.1)^2";
Background Field = 2;

1. Do I understand correctly that the components of the metric tensor in the MathEvalAniso are the *inverse* of the desired mesh width in x(m11), y(m22) and z (m33) direction?

2. Does this mean that the y in m22 has almost no effect, since the mesh width in y direction is 1/(y+40000)?

3. What if I want to stack several cubes, one above the other in y direction, and need a different mesh width in y direction for each of them, e.g. 0.1 for cube 1, 0.001 for cube 2 and 0.0001 for cube 3?

Thank you,

Matthias

> -----Ursprüngliche Nachricht-----
> Von: Christophe Geuzaine [mailto:cgeuzaine at ulg.ac.be]
> Gesendet: Dienstag, 9. Juli 2013 10:34
> An: Jan Martinek
> Cc: gmsh at geuz.org
> Betreff: Re: [Gmsh] very thin layer -> huge number of elements
>
>
> On 06 May 2013, at 07:32, Jan Martinek <jmartinek at cmi.cz> wrote:
>
> > Hello,
> >
> > I am trying to mesh a very thin curved layer (imagine a piece of cloth). It
> would be fine for me if the elements were thin, too. I think, GMSH attempts
> to make all the elements to be approximately regular tetrahedron and it
> results in huge number of elements.
> >
> > Is there any possibility how to tell gmsh to create elements, which may be
> rather flat?
> >
>
> Yes, using bamg/mmg3d. Here's a simple example:


_____________________________________________________________________
ERBE Elektromedizin GmbH
Firmensitz: 72072 Tuebingen
Geschaeftsfuehrer: Christian O. Erbe, Reiner Thede
Registergericht: Stuttgart HRB 380137