I am currently working on a Journal that performs hundreds or sometimes even thousands of mass properties calculations. I am seeing an interesting trend in my results. It seems that the more calulations I do during the run time of the Journal the less accurate the overall mass properties is. If I do less calculations the results are more accurate. Either way I am having an issue getting accurate volumes, and as a result other mass properties as well. These calculations include using the trim bodies command to trim a model a long the Z axis in small increments (typically in mm) and get mass properties for each slice.
Is there a perhaps a margin of error for each mass properties calculation that is compounded over the run time?
Note there is a customer default (Analysis -> Weight management) that controls calculation accuracy
There is also a preference (File -> Preferences -> Measurement -> ...)
You might want to investigate these.
But you could also have found a bug - I would report this to GTAC.
Production: NX10.0.3.5 MP5 + patch/TC11.2
Apparently I've turned into a gearhead
I agree with Ken. This sounds like it might be a bug. There are parameters that control the accuracy of mass properties calculations, but accuracy should not be degrading all by itself.
Just out of (intellectual) interest...
Are you saying that
running your code with with slice = 10 produces a different answer to running your code with slice = 1000
running your code a number of time (on the same volume with same number of slices!) produces a "degradation" of the calculated volume?
The first one, the more mass properties calculations you do in one run time the less accurate your accumulative results are. Now this degradation usually results in decimal places out to the 3rd or 4th place (and so on) for properties like volume and mass. However, for moments of inertia and centroidal moments of inertia, this loss in accuracy has a bigger effect on the total results.