Re: [ff3d-users] Problem with elliptic hollow cylinder section

[Stephane Del Pino]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dear Tobias,
 
The computed solution is not reasonable at the middle height of the cylinder 
section: 

On jeudi 22 Avril 2004 21:44, Tobias Nähring wrote:
> but the output is rather erroneous. I have played
> already with other solver settings, such as bicg(...) and
> krylov(type=bicg) without any luck. The grid size is as large as
> possible on my computer with 512MB RAM. (If I enlarge the grid size
> then ff3d freezes during "Marching cube ...".
I do not think it is frozen but just that meshing time becomes enormous ...

> With the current grid size I get the error message:
>
>   .
>   warning : meshing object plane {
>   (0, 1, 0),2
>   }
>    a problem occured ...
>   Check that:
>   - the object is not to small compare to the background mesh
>   - the object intersects the background mesh
>
> I don't know how to handle that.

This means that a particular object mesh has no triangle. This can ocure
- - when objects are too small, or
- - outside of the hexahedra mesh
I think that in your case this comes from the box: boxes are meshed face by 
face, one of them may be outside the domain ...

> The ff3d computed solution of the problem are graphically represented
> in the figures:
>
> http://www.tn-home.de/Tobias/Soft/Fem3d/
>
> Is there any chance to get a better solution?
I hope so. I think you just have to fix the geometry problem.
If not, you can use the solver with an unstructured grid that you may generate 
using 'tetgen' or 'gmsh'.

Note also that for this kind of problem, the grid steps do not need to be as 
small as you chose them.

My question, concening your geometry is: why do you use those two planes and 
this box? I was just thinking that for your geometry you just needed the 
cylinders ...

> The pov file (named cable-ellipse-annulus-num.pov):
> //////////////////////////////////////////////////////////////////////
> // Model of a cable with ellipsoidal cross-section.
> // rx=1.0 radius of the inner conductor in x-direction
> // ry=0.5 radius of the inner conductor in y-direction
> // a=2.0 the inner conductor scaled by a gives the shielding
> // CableL=5 length of the cable.
> // Since we impose symmetric boundary conditions we need only
> // to model a quarter of the cable. (0 < x < a*rx, 0 < y < a*ry)
> intersection {
>   difference {
>     cylinder {
>       <0,0,0>,
>       <0,0,1>,
>       2.0 // a==2.0
>     }
>     cylinder{
>       <0,0,-0.01>,
>       <0,0,1.01>,
>       1.0
>     }
>     plane{
>       <0,1,0>, 0
>     }
>     plane{
>       <1,0,0>, 0
>     }
>   }
>   box{
>     <0,0,0>,
>     <2,2,1>
>   }
>   scale <1.0,0.5,5.0> // <rx,ry,CableL>
>   pigment { color rgb <1,0,0> }
> }


Best regards,
Stephane.

- -- 
Stéphane Del Pino          Tel: (33)144277170 - Fax: (33)144277200
Laboratoire Jacques-Louis Lions - Université Pierre et Marie Curie
- -+-
http://www.freefem.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFAiOMObZyhKhZwdc0RAsiPAJwM6uqj0kxyAYfb0/mPZRLFMRCbKQCdHL2J
WOeYdi45lpwvjbEeeVfOV+w=
=+lBT
-----END PGP SIGNATURE-----