Would someone check my attached file for me to help me determine why the weight is greater when the surface is thickened inward compared to when the surface is thickened outward? (It has several layers which contain the original surface that can be thickened, if the normals point outward; the resulting large, single solid; 14 surfaces from joined facets with normals pointing inward; 14 solids resulting from those.)

I attempted to thicken a large surface. When normals are facing inward, ACIS collapses, the screen turns whited for 15 seconds or so, and there then is an error message about there being no solution...

I tried various tools all to no avail (except one) in an attempt to make the surface become something that ACIS could do anything with.

When I change the direction of the surface to outward, I successfully thickened the surface. I made it SAE1020 steel, and then i wrote down the weights values.

I then broke the large surface down to smaller surfaces. Longitudinally, I joined them with the result being 14 surfaces

I thickened each surface join group. Some which were larger were just too much for ACIS to handle, and I had to convert those surfaces into facets again and regroup, thus resulting in 14 final surfaces.

I then thickened each of the 14 and made them SAE1020 steel. I checked the weights.


There is a discrepancy. Here are the weights:

kg 22481.0348 for the surface thickened outward
kg 25955.2096 for the surface thickened inward

lbs 49562.1979 for the surface thickened outward
lbs 57221.4422 for the surface thickened inward.

Here are the weights differences:

kg 3477.1748
lbs 7659.2443

NOTE: I expected that by facing the normals inward that there would be some raggedy edges overlapping, which I'd have to systematically look for and figure out how to trim off at the seams or areas where they meet. I don't imagine that the overlapping solids somehow increases the overall weight since by thickening INWARD there would be less material if ACIS could deal with this in the first place.

Conceptually, in my mind, ACIS or ViaCAD would look at the overall shape. If it is planar or even self-intersecting, it could evaluate ignoring the self-intersections and compare them to an object thickened outward. Ideally, it would do a before and after comparison of the successful thicken operation, draw imaginary boundaries for each operation, and then extrapolate what the case would be for the same thickness if the geometry is thickened inward with very few self-intersections.

I am finding it difficult that ACIS is having a problem with this. The overlaps themselves should be non-significant given the object being thickened. I'd imagine a fraction of the total surface area being involved, and all of it very close to the vertices of the major curve areas.

I am thinking that even another programmatic backup ACIS could have deployed to spare the users agony would be to internally break the surface down, thicken them, then evaluate and adjust the seams or matching points. Programmatically, I imagine this would be seconds worth of work. For a human, it amounts to about 30 minutes of tedious, avoidable work if anticipated and programmed for.

Thanks for any input.

Hehehe... two things come to mind:

--laser & eyeball vs scalpel and camera

-- scalpel and eyeball vs laser and camera


I'm pretty sure I did not double up any surfaces as I color coded and zoomed and selected discrete distances and areas of surfaces. At no time did VCP throw error messages at me about operations, and the logicursor did not offer me multiple surfaces. I guess it's possible VCP created extra surfaces on changing from mesh, but that should happen to both resulting solids, I suspect.


I hope the next upgrade of Spatial's magic code can cope with this. I used to decompose the mesh to facets, then stitch all those, but it is tedious. Sometimes, there are gaps and weird angles, necessitating lots of inspection to troubleshoot. Seems there needs to be the CAD equivalent of rubber tennis shoe on plaster/mould. Can't CAD companies use light reflectance, or digital/mathematical distance checking and offer the user various degrees/methods of substituting a usable surface? I guess I'm asking too much, or somebody's got a patent on it somewhere...

Ah, I think in the past we've even asked about obtaining a reverse copy by putting the troublesome surface into a solid, doing a boolean operation, then replacing the surface with a new, clean, native VCP surface. Too bad we cannot use a mesh to cut something or measure to/from it.