From: Andrzejewski, Jan (jan_andrzejewski@peragroup.com)
Date: Fri Jun 12 1998 - 20:42:00 EEST

Several years ago I tried to encourage many UK FEA software vendors that
they could write a simple code to output STL format thus adding to their
software portfolio and increase market base.

The idea was that as they deal with analysing geometric data from many
different CAD vendors that they would have very robust IGES translators, a
host of meshing algorithms that would enable the user to manipulate higher
accuracy in selected areas.

What I found was a lethargy to tackle such a development. IGES translation
problems were many and their geometric methodology was different. ie the
simpler the geometry the better, removal of radii, chamfers, ribs and small
bosses were usually asked for to speed up processing.

I did however gain some assistance from one company the package was NISA. He
( forget his name) did understand the STL format requirements when describe
to him and we devised a simple set of output rules so that I could write a
FE mesh to STL transfer program.

1) Only use tri meshing over the whole of the object. ( Tri mesh is not
always the best meshing to use for the different analysis methods).

2) Delete as many nodes as possible on the planer surfaces other wise the
file size would be too large. ( The STL mesh on planer surfaces is divided
into two whilst automatic tri mesh will place many small triangular facets
as dictated by the user input values. ) This would be less of a problem for
me now as I have a much higher processing engine than then.

3) Not to use the packages standard ASCII output as this gives xyz co
ordinate values of each numbered node then each triangle is given the
related three node numbers. But to use a macro routine that is included in
the NISA package to reformat the ASCII output to give the following for each

facet 1
node 1 ( x value) (y value ) (z value )
node 2 ( x value) (y value ) (z value )
node 3 ( x value) (y value ) (z value )
facet 2
node 1 ( x value) (y value ) (z value )
node 2 ( x value) (y value ) (z value )
node 3 ( x value) (y value ) (z value )
facet 3
node 1 ( x value) (y value ) (z value )
node 2 ( x value) (y value ) (z value )
node 3 ( x value) (y value ) (z value )

etc. for each facet.

I then wrote a simple AWK program ( UNIX text reformatting program language
) that converted the above into an ASCII STL file.

So very simple to write that I did the same for STL to DXF and DXF to STL..

Even STL to Ideas macro language so that SDRC Ideas would read in faceted
models. But a simple sphere of around 500 facets took 2 days to read in
whilst the DXF into 3D Studio only took 10 minutes.

Anyway back to the FE issue.

I can understand that if you have a CAD package, you have already paid for a
FE module to that package, why should you have to pay a lot more for an STL
interface output.

The various FE packages have slight differences to there mesh data outputs
and I doubt that each one has the same macro language options to change the
output to the users needs.

The same meshing algorithm that is used for the STL output would not
necessarily give the best analytical results for the FEA computations.

Greg Morwood .UNV problem will be looked at on Monday I have seen an example
of format and it can not be simply text reformatted but the relationship of
the file can be explained for a program to be devised.

Must sign off now winding down for the weekend.

Jan Andrzejewski

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 - 22:45:56 EEST