Re: CAD software?

From: Michael Brindley, Chuck Kirschman (Clemson  University)
To: Michael Brindley
Cc: RP-ML
Date: Saturday, January 28, 1995
Subject: Re: CAD software?
> > Autocad leaves holes.  V12 is worse than V 13.  Meshing is
> > sub-optimal.
> 
> Ok, Chuck, I'll bite :) .  What do you mean by optimal meshing?
> The only definitions which I am aware of require the use of the
> fairly techinical phrases 'Voroni diagrams' and 'Delauney
> triangularization'.  These probably are not the best of subjects
> for this list.
> 
> For rapid prototyping purposes, optimal meshing probably means
> that the mesh closely or exactly represents the original object
> with the fewest number of facets (i.e., file size is 'small').
> 
Ah, thought I could sneak this one into the knowledge base.  

Meshing of a part is interesting.  There are what I consider to
be better and worse techniques.  For instance, consider a simple
sphere.  I have 2 CAD packages I use, and each provides a different
tesselation of the same 5 cm (2 inch) sphere.  1 sets up longitude
and latitude lines, and the sphere comes out looking sort of boxy,
since each face is 2 triangles (roughly square).  The poles
on this beast are some extremely thin triangles.  On the other
hand, the 2nd CAD package approximates the sphere with triangluar
faces, which provides a much more even mesh.  I much prefer the
results from the second, and it also provides for a better build
without the tiny trinagles at the poles.

So, given that there is no single meshing algorithm, consider the
multitude of possibilities for meshing a turbine blade.  The more
even the mesh, the better the final part, and the faster the slice
and build.  Some packages, including AutoCad mentioned above, do
not do a good job - they simply try to mesh pre-existing faces,
resulting in lots of long, thin triangles which are often ignored 
later in the process.  Other packages (ProE is a prime example)
mesh a curved surface in a way as to form a more even mesh.

Again, some people don't care about the results, and some use
so many triangles that it just bogs the process down (any triangle
less than 0.010" is less than a laser beam width) but seems to
produce a smoother part.  So, use this information only in the
context of what you are trying to do.

I'm not going to discuss meshing algorithms or any of the other
gizmos used in this process, but I'd be real interested to know
if anyone else out there spends as much time staring at individual
triangles as I do.

chuck