Sudden trouble with simple solids...

Hello,

while designing a plastic part using a lot of surface
functionality (merging surfaces, then generating solids
using quilts) I suddenly run into problems making a
simple protrusion of a sketched rectangle - it either
does not show up with its correct length but up to the
next (solid) surface only or - even worse - it refuses
to be created at all.

There are no suppressed or unregenerable features atop
of it, about 200 features modeled yet and still some 100
to go.

How to find out what geometry makes the model go crazy?

BTW, ProE 2000i2 Build 370.

TIA

Walther

 

If you're using surfaces you really ought to be using absolute accuracy,
(about 0,02 is good)

If you aren't, set this and see if it fixes things.

 

: If you're using surfaces you really ought to be using absolute accuracy,
: (about 0,02 is good)
:
: If you aren't, set this and see if it fixes things.
:
Accuracy on surfaces is always a good thing to check, though it's hard to see how
the simple geometry of a solid feature would fail because of this. But, who knows
with Pro/e. Successful surface merges and a 'use quilt' solid should have been
enough of a test of integrity to assure no problems in the future. However, just
to be sure, try a couple things:
* First, make sure there are no 'geom checks'; and if you're unsure of their
influence, create a closed surface volume and make it all a solid. A 'water tight'
surface will become instantly solid; one with 'leaks' which confuse poor baby as
what is inside and what outside will balk at becoming a solid. Skins which are
used to create a shell are not so fussy; a skin can have holes in it.
* Try reordering features which is a crude way of checking feature parent/child
relations. Incestuous relations can cause endless grief. And check your working
directory for any files that end in CRC; these document circular references. At
least check 'Info>Parent/Child' on your errant feature to see if its relations are
somewhat promiscuous. Sad to say, you can be tainted by the company you keep.
Beware of using anything for a sketcher reference which is not parallel to the
direction of feature projection.
* Pay attention to some other basic surface questions (besides accuracy which can
be very big when very small patches are involved)
Did you have any trouble with merging surfaces, waiting for 30-60 seconds
and finally, resorting to 'Join' instead of 'Intersect'?
Any difficulties with surface tangency in boundary surfaces?
Either of these may indicate problems with surface integrity, thus possibly
unstable surfaces and later problems with regeneration. Recommend not creating
solid until the end of process as 'solidifying' tends to obscure underlying
surface problems.
* Use blind for depth or, if up to surface, create an offset surface so that
surfaces are 'dirty' (overhang, sloppy is better in surfacing). The difference is
that this type of surface merges/trims almost instantly as Pro/e has to trouble
figuring out surface intersection. It's really remarkable what a difference it
makes to merging having overlapping surfaces.

David Janes

 

Thank You very much, David, for Your detailed response. Thank You all.

I have tried some things until now and it seems You´re right:
some surface merges took very long to finish, I took out some candidates
one after another (all of them were regenerating fine) and finally found
one which caused the erratic behaviour of pro/E.

As I understand it computer programs to define complex 3D geometry do a
lot of numerical equation solving, and if some of it comes near a
non-converging solution it may hang... or not, but consume time then.

I often do extend all boundaries of surfaces before merging to ensure
that the resulting quilt does not leak as You described. In this case it
seems to have had some near-tangential areas of to-be-merged, overally
very small surfaces which caused lengthy calculation (sic) but finally
regenerated sucessfully.

With those successed features in the chain above I could only create
drafts and rounds, but no protrusion or cut anymore - pretty strange.

I learned that best practice is not to have geometry behind that was not
fully understood (by the designer) - ProE cannot heal any singularities.

Again: thanks a lot.

Walther

 

I had a horrible experience a few years ago when relative accuracy would
allow a feature to regenerate at one point in the model creation process,
but when it was reordered later it failed, as the actual accuracy it was
being created at was relative to the instantanous model size, and thus
changing. I use abs accuracy as default, relative just doesn't like big
parts with small features.

 

Most obviously that´s the point: big parts with small features...

So lucky me: ProE did regenerate "successfully" - normally it would take
a deep breath from my patience (and a big chunk of computer memory) and
then exit without saying goodbye (resp. generate a core-dump on unix
systems like the ones I use).

Kind Regards

Walther

 

At least Unix boxes would fall over repeatably, as opposed to the XP crash
lottery.