Stress Question

[ickby]

Hello,

I just played a bit with the FEM workbench again, and the results I get seem odd. For example this analysis:

femtest.png (100.88 KiB) Viewed 1376 times

The minimal stress is larger than the maximal one, and the average does not even lie between min and max. Furthermore the visualization seems to be not correct, there need to be a non-zero stress throughout the cylinder, and not only that localized. Maybe that is a scaling problem, that the differences are that large between force-action point and the rest of the cylinder that everything else looks like zero?

Annother question: why is only the von misses stress stored, and not all 6 relevant values for the stress tensor? They are calculated by calculix, so it seems to be a loss of information to throw them away.

femtest.fcstd

(286.32 KiB) Downloaded 40 times

[PrzemoF]

Thanks for testing FEM wb!
1. fandaL just found error in our Von Mises formula [1]
2. You have force pulling a single vertex - that's something that cannot happen in the real world and it causes the hot spot problem. The hot spot problem happens when max stress is very high compared to the rest of values. Everything else looks like zero, because of the stress of the hot spot that is used to scale the colours. It's not yet accesible from GUI, but you can control hot points from python [2]
Open your file, activate analysis, purge results and run:

Code: Select all

>>> import FemTools
>>> analysis=FemTools.FemTools()
>>> analysis.run()
>>> analysis.load_results()
>>> analysis.show_result('Sabs', 50)

FreeCAD_vm_stress_with_limit.png (144.62 KiB) Viewed 1359 times

3. Correct, we only store Von Mises stress, but it should be very easy to extend ccxFrdReader to include the rest, but I don't consider it as a loss of information. CalculiX is producing a lot of data and we use only part of it, especially with frequency analysis. We're working to extend it as we need it.

[1] viewtopic.php?f=18&t=10882#p105639
[2] http://opencyclingcomputer.eu/FreeCAD/S ... ae2154ba82

[ickby]

Thanks for the feedback, with the updated formula and with avoiding the hotspot the result is way better.