Scanned data, singluar mapping

Hi,

Two questions, the second resulting from the first.

Thanks

1. Has anyone/can anyone share their experience of scanned component data - Real reverse engineering?

Any links, contacts, papers, methodologies,,,

This is new-ish territory for us; previous experiences have been with simple componenents which are easliy remodelled, a process that generally resolves all issues.

We thought we would investigate 'proper' reverse engineering data so had a moderately interesting steel casting scanned and delivered as 'watertight' surfaces in a number of forms. NURBS, .stp, .igs. to begin to understand what the real issues are including that nothing is either cylindrical or flat (even though it's machined) and so shafts and holes (contact) become interesting problems ... even a simple 'flat' plate bolted interface is an interesting contact problem.

.igs and .stp are non-starters. Thousands of surfaces when we get them to load.

NURBS - Slow to load/spin (lowest quality shading) and Mechanica (Creo) mesher continuously gives the message 'singlular mapping'. Chasing one down, appling mesh controls leads to the next. Sometimes and assembly will mesh, but with a simple change of component (all other components are native pro/e) the assembly will produce more meshing errors.

Contacts disappear, the definition of the contacts are in the model and can be editied but those analyses that run fail insufficiently constrained with no contact measures reported.

Pro/e crashes regularly at the most obscure moments. msengine terminates regularly with no error logs.

I conclude that there is a lot we don't understand and that this is not limited to the pro/e end of things; it must include the scanning and methodologies employed in the processing of the scanned data.

2. Singlular mapping element - what is this and what causes this? I have avoided the question so far by knowing adding autogem controls and/or modelling over the top usually makes the problem go away (we have to be practical).