validation of results

[bnrg]

Hi,
I'm a fairly new user of Pro/Mech but an experienced design engineer, and I am slowly working on my confidence level of the results that I obtain. I design things that mostly use plates so I compared the results of some round and square plates in mechanica to roark. A little disappointed so far, but maybe I am expecting too much?

Best results for a round plate (uniform normal pressure load)was to use a solid, compressed into a shell. Stress compared well (~4%) but displacement was off by about 12%.
Best results for a square plate (same loading method) was again a solid compressed to shell but in this case the deflection was right on with stress off by 15%.

Results using 'thin' protrusions was poor. Note that I varied how constraints were applied (edge & surface) and no real difference was noted. Square plate width/thk=40, round plate dia/thk=12.5. I applied light loads to ensure staying within elastic range and used 'standard' mat'l properties.

Are these results typical? I realize that these are only approximations and that integrated mode (all that I have) maybe makes the approximation even more approximate, but these usefulness of these results seem a bit questionable to me. I have spoken with some engineers who do not even question the results. Maybe I am too old or conservative to do that, but I would appreciate feedback as to others results.

Thanks in advance for your help,
Bob

 

[RocketRonnie]

The way you apply the loads and constraints are fundamental to the "accuracy" of the results.
You can install the validation models supplied on the Mechanica CD which do exactly what you are trying - validates simple analyses against theory. Maybe try these to get confidence, and examples of how these models were loaded and constrained.