Modelling a sphere with PATRAN
Thomas Schacht
does anyone know how to create and mesh a simple sphere in a cube with
PATRAN?
I want to simulate a void or second phase particle in a metal matrix but
didn't know that there are so many difficulties in creating the mesh
around the sphere.
Thanks a bunch for your help.
Regards
Thomas
Pook99
generators. Often, what you really WANT is to have the surface of the sphere
contain elements of similar size and shape. What you usually GET are tons of
sliver elements at the poles. The common mesh generator programs in use today,
such as PATRAN, FEMAP, etc. will give you the sliver elements, which is
probably not what you want. Hopefully they will include more helpful solutions
for this common problem in the future.
The only way I have found to generate reasonable elements INSIDE a sphere is to
break it up into six sections, corresponding to the six faces of a cube. Mesh
one section, and then copy the mesh to the other five sections. A nice
illustration of this procedure is given at www.truegrid.com/quality.html. (Note
that the meshing programs STILL do not make it easy to do THIS procedure, but
it can be done!)
You may be able to use this procedure for the EXTERIOR of a sphere. At least
what you will wind up with is decent looking elements on the entire surface of
the sphere.
Keith A. Honkala, Ph.D., P.E.
Engineering Consultant
Poo...@aol.com
In article <37808BCE...@tu-bs.de>, Thomas Schacht <t.sc...@tu-bs.de>
writes:
>Subject: Modelling a sphere with PATRAN
>From: Thomas Schacht <t.sc...@tu-bs.de>
>Date: Mon, 05 Jul 1999 12:41:19 +0200
Carrick Hill
Try meshing an interior cube with a regular quad mesh, so that
its vertices are almost touching the surface of the sphere. Touch
the sphere surface with tetrahedrons from the cube vertices and
fill in the gaps with a quads.
In the end, its always better to use an axisym circle (if at all
possible) with the same philosophy as above.
Carrick Hill
University of Auckland
NEW ZEALAND
Matthew H. Koebbe
Thank you for pointing out the nice illustration of a meshed sphere
provided by the folks who author TrueGrid. I'd like to further add
that that model takes only 6 user-understandable commands in TrueGrid.
I regularly show that demo (as I was shown when I started with the
company) and the well-meshed sphere takes less than 3 minutes to
generate. And notice that our technique produces the highest quality
elements at the surface and in the center. The "bad" angles
(plus or minus 45 degrees) occur just outside that inner cube --
and the user has control of exactly where they lie.
You do not have to separately mesh each surface in TrueGrid. We look
at the process as more akin to sculpture. Start with a multi-block
structure (lump of clay) and discard those regions that you don't
need. Finally, the molding process is achieved by projecting the
faces of the multi-block grid onto whichever surfaces (or combinations
of surfaces) is appropriate. I'm simplifying, for explanatory
purposes, but it really is a powerful technique for easily generating
high quality meshes (at least it is easy when the geometry is simple!)
Sincerely,
Matthew Koebbe
XYZ Scientific Applications, Inc. (authors of TrueGrid)
> >Hi there,
> >
> >does anyone know how to create and mesh a simple sphere in a cube with
> >PATRAN?
> >
> >I want to simulate a void or second phase particle in a metal matrix but
> >
> >didn't know that there are so many difficulties in creating the mesh
> >around the sphere.
> >
> >Thanks a bunch for your help.
> >
> >Regards
> >Thomas
> >
--
Matthew H. Koebbe
koe...@home.com
Brian Hogg
> Thank you for pointing out the nice illustration of a meshed sphere
> provided by the folks who author TrueGrid. I'd like to further add
> that that model takes only 6 user-understandable commands in TrueGrid.
> I regularly show that demo (as I was shown when I started with the
> company) and the well-meshed sphere takes less than 3 minutes to
> generate. And notice that our technique produces the highest quality
> elements at the surface and in the center. The "bad" angles
> (plus or minus 45 degrees) occur just outside that inner cube --
> and the user has control of exactly where they lie.
> You do not have to separately mesh each surface in TrueGrid. We look
> at the process as more akin to sculpture. Start with a multi-block
> structure (lump of clay) and discard those regions that you don't
> need. Finally, the molding process is achieved by projecting the
> faces of the multi-block grid onto whichever surfaces (or combinations
> of surfaces) is appropriate. I'm simplifying, for explanatory
> purposes, but it really is a powerful technique for easily generating
> high quality meshes (at least it is easy when the geometry is simple!)
> Sincerely,
> Matthew Koebbe
> XYZ Scientific Applications, Inc. (authors of TrueGrid)
I didn't get on with Truegrid for quite a while, being used to the
conventional meshers, then one day it just clicked. Its a very powerful
tool and very easy to create complex 3D shapes. Only small True grid session
files are needed to generate the structure so you don't need to save a
massive Patran model binary file or an incomprehensible Patran session file.
The structure can also be easily parameterised in the Truegrid session
file making it very easy to carry out design studies. Pity 2D is not
supported otherwise the other meshers could be discarded. The spherical hole
in a cube is one of the standard examples.
Brian.
--
Voyager Argonaut
RiscPC + Storm + StrongArm => Acorn Power
Convert to Pluto and Newshound
...theh...@argonet.co.uk
......www.argonet.co.uk/users/thehoggs/
Claude bryant
Patran has a very nice way to model what you want. You can write a vector function
from the side of the cube to the inner sphere portion. Patran can create the mesh
according to the function. You can produce some beautiful meshes this way. Be
sure to ask tech support for some examples of this method. Also, be sure to take
advantage of your symmetry if possible.
Regards,
cbryant