Thicken surf vs extension normal to surface

[MIKENEUTRON]

I am currently building carbon fiber window
check and assy. fixtures. the nuts and bolts to these
fixtures is basically triming a surf (thru curve mesh)
typ. to a spline that represents edge of glass.
At this point I thiken surf. by say 50.8 mm
sounds easy right. My problem is this one of my custermers
do not use solid models . they build side walls
by free form featue extension normal to face.
The fixs can be fairly large say 6foot by 3foot
I am allowed +-.13 mm deviation on gage.
ther models seem to average .04_.05 different from
mine. (half of my total tol.)
Dev is random not top edge or bottom
and typ only 1 or 2 spots are different from my
model to there model
and say 3 out of 4 models are close enough
I had one out by.079 to theres?

What I need to know is why there is any difference
at all? we both build from same data set.
I have set dev to .005mm for trim and thicken opps
the reason i thicken is 90 return wall is req.
I also have had corners come out concave rather than
stright (90 to sheet body) this I suspect is
due to the small corners that I some times deal
with . I have about 20 years cad exp but only
2+ on ug. even worse i am basically self tought
on ug. I have been threw the cast files (seam very basic)
our sys is lagit but maint is out of date.
(before I started to work here)
we currently use nx2. Perhaps I dont fully understand
thicken vs extension. The closer our model are the more
tollerence I have to build my fixtures. The reason for
second data set is custermers second sorce cert.
This is supposed to get rid of any modling errors.
I even tried offsetting isg (in side of glass surf)
and projecting edge of glass curve (outer spine)
normal to face I offsetand manually building side walls
ten sewing together
these walls were slightly diff. from the other 2
(slight variation to other ways of building)
Also i am not sure if tollerence or dev is angular?

any help would be great

Mike Tucek cad man vision mach inc.

 

[Penderwyd]

Mike,

You are on the correct track. Most of the time Thicken Sheet is the best solution for models like yours. I believe it still is in your case. The excessive deviation is most likely due to the default modeling tolerance. When creating surfaces to datasets provided with such tight tolerances, you should reduce your model tolerances when generating your initial sheet body, then play with the specific tolerances of your thicken. I always start with the smallest tolerance the model will let me make and relax as necessary. Also check your surface deviation prior to thickening. You may actually need to remove specific data points to avoid wrinkles.

Basically, most of the time, tighter modeling tolerances and few actual definition curves and points will generate the better sheet. Better sheets produce better solids.

I hope this helps you.