Long post warning!
We are in the midst of upgrading to SW2011 SP3 (from SW2009 SP5.1) and during this process I ran a Simulation study intended as a check to make sure we would be getting consistent results (this model had already been used to compare SW2009 SP2.1, 3.1, 4.1 and 5.1 so seemed the logical choice for my reference to check 2011 SP3).
In its simplest form, the part used in this analysis is a
Ø4.6" x 42" cylinder with a number of machined features, supported at one end and subjected to a 0.7° bend (using a remote displacement) at the other. Unfortunately, the results generated were spectacularly out of line with what had been seen previously. Here is the result from SW2009:
And here is the result from SW2011:
As I'm sure you can appreciate, quite the difference!
As an attempt to solve this issue, I decided to go back to basics and analyse about as a simple a part as is possible - a
Ø2" x 20" solid bar. This bar had a fixed restraint applied at one end, and a 0.45° bend at the other. For the first test, the bar was bent in the positive direction around the Z-axis (assuming that SW2011 follows the right-hand-rule convention for bending used in SW2009). When viewed from the right, a deflection upwards would be the expected result, however this was in fact the result generated (note - the material applied was the stock SolidWorks "Alloy Steel"):
While not only being in the wrong direction, this analysis also yielded the same tropical-fish-like pattern along the part:
I then proceeded to try and bend the part around the Y-axis - if this was perpendicular to the Z-axis bend (as it should be), I felt I would at least have something consistent to work from when trying to find the root of the problem. However, this bend around the Y-axis generated the following deflection:
ie. nothing like perpendicular to the Z-axis! And again with the unexpected patterning along the body.
As another check, I changed the materials to one of our company-defined property sets. The bend direction should be fixed regardless of the properties of the material, however the following picture shows the bend in what is theoretically around the Y-axis:
Once again, this is not deflecting in the correct direction, however more concerningly it is not even consistent with the earlier Y-axis defection!
Finally, I split the model in half and applied a symmetry restraint to see if this would make a difference. Surprisingly enough, it did:
The stress plot here is exactly as you would expect for this specimen - maximum stress along the edges that see compression and tension, and minimal stresses along the neutral axis. So it would appear that somewhere in the system the bend feature is working, but only if what you're analysing is symmetrical (which our products rarely are) and is only any good if your worst-case bend direction is along the axis of symmetry (which our products never are!).
So, a few questions:
a) Has anyone ever come across this problem? And if so, how did you overcome it?
b) Can anyone replicate these results on the simple Alloy Steel rod? And if so, would you be able to post results and computer specifications (particularly the Service Pack version being used and the Operating System).
For what it's worth, these analyses were conducted on a HP Z600 with the following specifications:
- Intel Xeon E5540 CPU x2 (2.53GHz clock speed, total of 8 processing cores)
- 12Gb RAM
- NVidia Quadro FX3800 graphics card (with latest driver)
- Windows XP Pro x64 SP2
- SolidWorks Premium 2011 x64 Edition SP3
Apologies for the huge post, just wanted to convey the situation clearly as right now, we can't really do any analysis work!