Magic Numbers in Parametric Software

From my experience it appears that as an assembly gets bigger there
comes a time, (ie. a "magic" number of constraints) where the program
doesn't seem to be able to handle minor changes anymore. One of the
solutions for this is to stuff parts into sub assemblies. This
appears to drastically reduce, (usually) the number of constraints
that SW needs to calculate and this seems to resolve the problems.

little round off error and when these errors get too large the whole
model seems to shatter like a piece of glass.

Has anyone else ran into this? Are there any rules of thumbs for
this?

Also from my experience, it seems like SW is about 3 times more
tolerant then Inventor because it doesn't take very many parts at all
in Inventor before it turns to garbage. Where as SW seems to be able
to handle a fairly good size assembly befor problems start to be
created.

Thanks,

EdT

 

Inventor doesn't have as good a Beta testing program.

TOP

One trick that utilizes this is to put your BIG assembly into another
assembly, then mate whatever parts/subs to it and then drag those
parts/subs into the BIG assembly, save the big assembly and then just
shut down the dummy assembly.

 

From my experience it appears that as an assembly gets bigger there

I think everyone has seen something like it, but there is no magic
number. The mate solve time probably increases approximately
exponentially with the number of mates, with different increases for each
mate because of different situations for each mate. Stictly speaking, it
probably should blow up sooner, but the mate solver allows for a lot
redundancy in the constraints as long as each redundant constraint solves
the same geometrically. That is, more than one mate con constrain the
same degree of freedom. The solver has to figure out that this conflict
is actually ok and not blow up. When it does blow up, it usually has to
do with the solver algorithm not being able to figure out what to do with
all the redundancies.

The only way to combat this is to limit the number of mates and redundant
constraints. Use the subassemblies (they have other advantages too), and
try to constrain only the degrees of freedom that you need to. One other
thing to look out for is mating to less robust entities, such as edges.
Esprecially if there are in-context relations involved. That is a no-no,
and causes blow-ups and other entertaining behavior very quickly. Proper
skeleton modeling techniques also help. You can look that up on the
subscription site.

Why is it so complicated? Because SW gives you a lot of rope to hang
yourself with.

 

One trick that utilizes this is to put your BIG assembly into another

Doesn't that eventually end up in nightmware rebuild times for the BIG
assembly and any drawing views that show it?

 

Yup. I never tried dummy assembly with a drawing. Might be something
to try.

TOP

This would wreak havoc on PDM. Supposedly a detached drawing would
stop the rebuilding and solving too.