What does the Connection Analysis tool actually measure and is it important?

[Pettersson]

The Connection Analysis tool allows me to check for how surfaces connect with each other. I can analyze a quilt and look at the edges between surfaces to see if they match up in position, tangency, curvature and acceleration. the last three are quite obvious. Wheich edges are tangential and which are not? But the first one is a bit strange. If I have a quilt, and no one-sided edges, and it can solidify, obviously it's a closed quilt. But I can still analyze it and see that it has edges with a position error more than zero. It may even have edges with a position error more than the accuracy of the part.

You'd think such a thing would cause problems, but it doesn't, necessarily. Creo still highlights the edge as two-sided and I may still solidify the quilt. I'm guessing Creo is checking the edge of one surface against the corresponding edge on the other and looks at the difference, and due to the inexact nature of the mathematics here, it's not going to be exactly the same.

The question is: Is this a problem? My gut tells me that a position error of more than the accuracy of the part is an issue that can cause problems. Logically it should cause problems, or at the least create a geometry Check.

For those who do surface modeling, is this something you use? Is it something you worry about? I've never really bothered with this tool, except for checking tangency and curvature errors, but I got some questions about it and wasn't sure how to respond. What are your takes on this? Is it something one should run regularly? Should the policy be to never have a position error that's bigger than the accuracy? Or is it a question of "If it can merge and solidify, it's fine"?

[tbraxton]

In the context of your question regarding the "position" errors. When you query geometry checks on one of these models are there any geom check warnings associated with the quilts/surfaces in question? If there are no geom checks then I would not be surprised that the quilt is "closed" in the eyes of the geometry kernel.

[Pettersson]

Yeah, no geometry checks. That's what I replied as well, that if there are no geometry checks it's probably fine, but if there are issues with solidification or something else, it may be worth using the Connection Analysis tool as a troubleshooter to find any problematic areas. But having a position error larger than the accuracy "feels" like it should be potentially problematic.

[tbraxton]

We do a fair amount of surfacing work and use this connection tool to confirm continuity (G0, G1, G2...). I don't recall ever seeing anyone use it to troubleshoot edge matching. We would query feature references to confirm edge matching and more organic surfaces we would be working with ISDX super features which are different than core Creo curve and surface features. We do commonly use Creo curves as parents to ISDX surfaces and it is generally quite robust if you know how manage the references. When ISDX was integrated into Pro/E they obviously developed methods to connect the two types of features.

Your question would probably have to go to PTC R&D to get a definitive answer.

Some thoughts on this (I am not an expert on the code or algorithms used ). There is "tolerance" on geometry congruence which is generally used in CAD to minimize regeneration time. If all curves and surfaces are numerically approximated and then evaluated locally as required based on the features and dependency. This requires dealing with "mismatch" and applying constraints (connections) if it is close enough.

[Pettersson]

That sounds like my understanding and use case for the tool, as well. But surely G0 continuity and "edge matching" is the same thing? If an edge is not G0 continuous, it means it's not quite "sewn together", which in theory should mean there's a hole, but in practice Creo manages anyway

Your note on tolerance and regeneration time is interesting. Perhaps G0 mismatch means Creo has to do some extra work to fix it up, increasing regeneration time..