Chris,
First, how much RAM and SWAP space do you have?
I experimented with hollow lattice structures like "Swiss Cheese" as a
way of reducing the mass of large objects 7 or 8 years ago. I also ran
into the problem of processing these large datasets and abandoned the
idea after limited success.
I did, however, work out a way of processing large data files with
limited RAM and CPU resources both on UG and in SLA.
By dividing the CAD model into smaller pieces you can minimize the
problem you are having. Just cut both the object and the sphere-set
with the same planes and do your Boolean intersect, the resulting
parts should match up just fine when you build them together.
If the STL data is too complex to process, then you can keep the STL
objects separate and build them one at a time. Back in the early days
of SLA, the disk drives were too small to hold the massive build files
I needed to work with so I actually built them as one "slab" on top of
another in successive builds. You may or may not be able to do that on
your SLS machine, I've never tried.
With SLA, the process worked like this:
1.) Add registration objects (tall, thin cylinders) at the extreme
limits of machine build envelope in CAD model. They must match the
height of the object. (This will keep the successive builds aligned
properly)
2.) Position part, orient, and cut entire model and registration
cylinders into reasonable thickness slabs. (You may have to
experiment to determine what "reasonable" is.)
3.) Prepare each "slab" as a separate build but add a pause (no drain)
command to the end of the each build.
4.) Build slab #1 which will pause at the end of the last layer, build
"slab" #2 on top by instructing it to start on the first layer of
that "slab" which will then adhere to the previous slab.
5.) Repeat until entire model is completed.
I also use Unigraphics, and I would love to see this model data
myself.
Sincerely,
Larry Blasch
System Administrator for Engineering Services
OPW Fueling Components Voice: (513) 870-3356
P.O. Box 405003 Fax: (513) 870-3338
Cincinnati, OH 45240-5003 USA
*********************************************************************
* "When all else fails, Manipulate the data... Albert Einstein" *
*********************************************************************
______________________________ Reply Separator
_________________________________
Subject: CAD problems and Swiss Cheese
Author: "Chris Sutcliffe" <c.j.sutcliffe@liverpool.ac.uk> at internet
Date: 10/1/98 6:03 AM
Can anyone help
We are doing some research at Liverpool which involves the generation
of = a plastic structure made up from small interlinked balls. We
have a CAD = solid model of the geometry but we need to turn the
geometry into the = inverse of Swiss Cheese!!!! and subsequently
manufacture it on out SLS = machine. The route I have used so far is
as follows
1. Create the geometry on UG
2. Create an array of interlinked spheres
3. Intersect the two solid models
Unfortunately because there are so many balls my machine (SG octane) =
keeps killing the application which is unfair to say the least. Can =
anyone help,does anyone know of a structure like Quick Cast for example =
which does not have the outer skin and can be automatically generated?
Thanks
Dr. Chris. Sutcliffe
Product Innovation and Development Centre
Department of Engineering
3 Brownlow St.
Liverpool L69 3GL
(0151) 794 8022 Direct
(0151) 794 8021 Secretary
01893 209 556 Bleeper
EMail c.j.sutcliffe@liv.ac.uk
For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
For more information about the rp-ml, see http://ltk.hut.fi/rp-ml/
This archive was generated by hypermail 2.1.2 : Tue Jun 05 2001 - 22:46:50 EEST