Geometry check present when creating a round feature

[tbraxton]

I have a round feature that is throwing a geometry check when modeling the same underlying geometry created differently. The round appears to be there and accurate visually.

I am looking for an explanation of the root cause of this error. This is a simplification of a much more complex model that includes the same geometry. The production model uses a mirror part geometry as only half the part is explicitly modeled. This is the only geom check in the model and I have not been able to get rid of it. I have tried datum reference surfaces for the mirrored boss but that did not resolve the error. Does anyone have any insight into why this error is generated?

1) Both models (Creo 4) use the same absolute accuracy and are included here for reference

2) Both models share geometry that is congruent

3) Only the model that mirrors the 180 degree revolve feature throws the error (mirr_part_geom_check.prt)

The error is:

Set cannot be constructed on the highlighted pair of surfaces.

Recommended actions:

     Try to create feature as "Unattached", and complete it using quilts.

[Tdaugherty]

Hi, 

It seems like Round4 is causing the issue. In this feature, you are selecting two surfaces, the outer surface of the Revolve1 and the perpendicular surface inside Revolve2. This is a fine way of doing things but it seems like it's getting confused when you mirror the geometry. Take a look at the the round in the non-mirror part. The surface references look correct. In the mirror, the same round feature shows the outer surface of revolve 1 on the other "side" of the part. Like I said, I'm not sure but it seems like mirroring is flipping your reference to the opposite "side" of the part. 

Delete your pattern and redefine round4 to the edge and then repattern. This seems to fix the issue. 

[tbraxton]

Did you delete all of the patterns or just the one for round 4? I just tested your suggestion in Creo 4  M110 and the model does not regenerate. The mirror fails when patterning round 4. I had tried this previously with the same result.

If I delete all of the patterns (not just round 4) and then recreate them and use the edge for round 4 then it does regen and the geom check is not thrown.

[Tdaugherty]

Sorry! I should have clarified the steps a little better. It's a bit confusing when you have a geometry dependent mirror. 

1. Open REV_360_TEST.prt
2. Delete the pattern of Round4
3. Redefine Round4 to the edge
4. Repattern Round4
5. Regen if you need to
6. Open MIRR_PART_GEOM_CHECK.prt
7. Delete the pattern of Round4
8. Redefine Round4 to the edge
9. Repattern Round4
10. Regen if you need to

[tbraxton]

Thanks for the clarification on what you were doing. I was able to quickly determine how to get it to fail regen even when following your steps. The rerouting of the reference of round 4 is sensitive to which instance of the pattern leader (ref pattern for round 4) is selected when following your steps.

I usually use the first instance of a parent pattern to define references of its children, in this case when I do that the regen of the mirror fails. If I follow the existing ref of the instance that is highlighted in the graphics window then your method works.

Something strange is happening with the mapping of references with the mirror feature present in the model prior to the creation of these patterns.

[Tdaugherty]

I agree. That's some odd behavior. I'm wondering if it's specific to cylindrical surfaces. Would have to test. 

Ty

[tbraxton]

The scourge of Pro/E since the first release, splitting circles into halves without robust reference handling when crossing the split. I was hopeful that datum reference features would provide a workaround for this when dealing with references and stability of children referencing cylinders. The first thing I tried for this issue was to define a datum intent surface as both halves of the cylinder but that did not work.

The fix in this simplified model is not working in the actual model with 330 features in front of this geometry. I can get around the geom check by not patterning the round and just explicitly selecting the edges so it is not a show stopper but I will open a call with support to address this.