Whom it may concern,
I am a retired 75 years old engineer and bumped into PrePoMax by chance. I started to play with it and I am quite impressed. I am using it only as a recreational hobby. It is weird, isn’t?
I rebuilt the “Dvizna_plosca.STEP” sample in a hybrid model, leaving the center hub as a solid and change the ribs in shells. I followed the instructions in the YouTube tutorial solid-shell with the simple cantilever model. In my model there are a number of contacts now. The program did not find them automatically, so I created them by hand, again following the tutorial instructions. Checking the model in the Analysis tab delivered errors with negative items in the Jacobian tensor. The original solid model without contacts runs fine.
I would welcome a detailed tutorial with solid-shell hybrids, because this is a daily bread-and-butter application.
Upon getting a contact address I would gladly transmit my PrePoMax file for examination or subject for a tutorial.
I also would be glad to see a detailed gear flank contact tutorial. Gears are another weird recreational hobbies of me. Sorry to ask for so much, despite XMas is over!
Have a Good New Year!
Can’t you share it (possibly the pmx file) here ? Not necessarily directly but using some hosting website like WeTransfer, Google Drive or Dropbox.
Dear Jakub,
I am positively surprised and grateful for such a fast response!
Now, here is the faulty model: Plate.pmx. The file is of modest size.
In the meantime I deleted the contacts and recreated the model using
constrains. It runs without any complaints. The results look quite
logical but I have to compare them to the solid model. But one hitch:
the reaction force at the fixed bottom of the hub are definitely wrong.
I am attaching this file as well: Plate_1.pmx.
Just for clarification, I am not completely a greenhorn in FEA but
certainly a greenhorn PrePoMax user.
Once again, thanks for your kind response.
Sincerely yours,
DI Sandor Palvoelgyi
Frankenberg 151
A-8200 Frankenberg
Plate.pmx (3.42 MB)
Plate_1.pmx (3.42 MB)
Tie constraint may work as well. I also changed the load and support a bit:
Plate mod.pmx (694.2 KB)
i do quick looking the model, it seems edge shell of Rib and Hub surface are not fully connected i.e half only. Perimeter of hub surfaces may need to define at whole and act as master.
also, the Top surface with end Hub elevation seems to need checking
This model works flawlessly. Thanks!
I realize now that one tie constrain can have many individual stitch areas.
I have to understand the sense of rigid body constrains, however.
The reaction force is also correct.
I think a tutorial with a bit more complex structures may be benefitial
for beginners.
Nevertheless, it is fun to use this software. I am happy that bumped
into it just by chance.
Something more: so about 40 years ago I was fortunate enough to attend
lectures by Prof. Zienkiewicz. No need to introduce him to you.
He was not exactly a fan of the colored contour plots, especially the
von Mises stress ones. He said that those are only for the marketing
people and for managers. He advised us to allways look into the
principal stress values and in addition look for the orientation of
these stresses using vectors.
The vectors should not be scaled by the magnitude of the stress (you
would get a messy forrest) but selectable length and colored as the
contour plots would be.
It is useful if the vectors are not everywhere but in selected areas.
The contour plots could be helpful to define node sets for these
vectors. The reason is self exclaiming: at nodes where both principal
stresses are tensile and big are highly subject of failure. Also sudden
changes of the directions of the vectors need attention. I have an old
book from Prof. Zienkiewicz in German tongue but there is almost nothing
about results interpretation.
May be you could introduce this feature to future releases. If I am
correct, Abacus has this output option.
I scanned a picture out of the book, you see the thermally induced
stress vectors. They are scaled, which was in the black-and-white era
neccessary.
Have a nice day!
Sandor
Symbol/vector plots would indeed be helpful. They are available in ParaView: Post-Processing of FEM Results with Paraview - FreeCAD Documentation
Here’s an existing FR for this: Visualize the directions of the principal stresses
Thank you, synt! I got a corrected model from FEAnalyst with exactly same fixes and the model works fine and the reaction force is correct, too. I am grateful for any tutorials and hope one day a reference book will be published for this fine program.
Does it matter some cells’s normal are pointing to opposite direction ?
Would be good to fix it. It’s better to keep normals consistent. PrePoMax can display (and change) them too for selected faces.