C2+C1 succeeds but C1+C2 fails in geomtool

The following file contains geometry composed of a sphere, a torus, and two cylinders inside a block.

combineC1andC2.fea



The two cylinders overlap.

When combining them by addition using "C1 + C2":



geomtool fails with:



Complete text:
geomtoolCrash.txt

However reversing the operands, "C2 + C1" yields the expected result.



I assume this is a bug, unless there is something amiss with my construction (or my thinking :-)

Kind regards,
Randal

No surprise to you I'm sure,  but build 20.08.221 solves this issue for me. Thanks for the quick turnaround!

Kind regards,
Randal

Thank you for the detailed issue report. Geometry objects which have this issue will unfortunately have to be re-generated with the fixed engine This issue will primarily effect geometry objects featuring axis alignment, such as cylinders, in configurations that might be difficult to tesselate due to floating point tolerances (touching edges/faces with sharp angles). It should hopefully work more robustly now though, but please do let me now if there are further issues.

Precise Simulation wrote

Thank you for the detailed issue report.

No problem. I'll keep reporting 'em if you'll keep fixing 'em :-)

Precise Simulation wrote

 Geometry objects which have this issue will unfortunately have to be re-generated with the fixed engine

This is very important to know, please continue to let me know this when it applies to an update.

Precise Simulation wrote

 It should hopefully work more robustly now though, but please do let me now if there are further issues.

I will.

Kind regards,
Randal

Just to clarify, you don't need to remake your geometry if it can be successfully used to generate a mesh. After you have a mesh to compute on, a geometry is not really needed anymore.

Understood.  Thanks for the clarification. As it turns out, this particular geometry was a ‘toy’ used to test the geometry split scripting and so not a real FEA problem.
-Randal