I have set the config options:
retrieve_data_sharing_ref_parts to 'yes'
and
retrieve_merge_ref_parts to 'yes'
But I still get these errors when I perform a modelcheck regen:
Parent assem <copy geometry=" assembly=" name="> for feat (id <feature id=">) in part <component name="> not in session.
Warning: External ref. for feature/component not found, using old placement.
Warning: CopyGeom reference not found. Using old placement
A former employee surface modeled these components and used secondary "copy geometry assemblies" to share geometry from one part to another. The copy geometry assembly only has the parent and child parts in it. The guy said it was better to do it this way so that if the "parent" part failed, the real assembly wouldn't fall apart. (not sure I buy that) but now the problem is that people are working on the components without having the CG assemblies in session. Regeneration is skipping over the "missing references" and using the last known placement. When the CG assembly is brought in, the component fails because the "updated" copy geometry feature references don't play nice with the rest of teh component features that are dependant on the CG references.
I'm annoyed because the whole point of the CG assemblies was to make sure all the components share the same reference geometry. (though I don't see why any particular part should be made the parent! we have advanced assembly, so we should properly convey shared geometry from the top down)
I suppose at this point, instead of just ignoring the bad cg assemblies, I am no better off if I just make the copy geometry features independent.
I would like to completely get away from these CG assemblies all together. Does this make sense?
Not to mention that all these complicated, surfaced parts were modeled at relative accuracy of .0012 and sometimes fail to regenerate when assembled into larger assemblies. any advice? (I suggest change the relative accuracy to something like .0004 and fix everything that breaks).
any advice about surface modeling best practices, good techniques to share geometry between components, how to document model structure so that it can be handled/revised by more than one "expert", how to encourage properly fixing a model instead of a continually patching a model, etc will be welcomed.
Sincerely
Paul Bock