Principal Stress Plots

[ickby]

I will give it a try when I find the time Thomas. It is difficult to choose standard settings though. It takes quite a bit of playing with scale factors, colors, line thicknesses, etc. to get an acceptable result. That flexibility is available to the user in Paraview, but not in the FC VTK pipeline. So I fear that the plots will look clunky, driving people towards Paraview anyway.

You can expose exactly the same parameters to freecad pipeline as available in paraview, as both use VTK filters. You just need to figure out which filters you want to use, and than implementing those is very simple. Only the parameter GUI may need some experimenting... I think one guy already did it for the arrow plots... but can't find it anymore

[HarryvL]

Thanks @iclby. Do you mean this?:

https://forum.freecadweb.org/viewtopic.php?f=18&t=24637

I will have a look at @amstuff's repo.

PS: I can't find the option of "double-headed" arrows in Paraview. Do you know how to do that?

[HarryvL]

PS: I can't find the option of "double-headed" arrows in Paraview. Do you know how to do that?

@amstuff coded it himself in src/Mod/Fem/App/FemPostFilter.cpp by superimposing two "opposing" vectors, i.e. "v" and "-v"

[HarryvL]

Now combining the reinforcement ratio development (https://forum.freecadweb.org/viewtopic. ... 40#p235356) with the principal stress development.

Model_007_1.png (120.33 KiB) Viewed 16699 times

-

Principal_Stresses_007_1.png (280.62 KiB) Viewed 16699 times

The lines show the minimum principal stress (compression) in the concrete and the colours shows the reinforcement ratio in x-direction. Note that all principal stresses in the concrete are either negative (compressive) or zero. Any tension is assumed to be carried by the reinforcement.

[oldmachine]

Thank you

The pipeline of results in FC on Mac Sierra is worthless - text/values are not readable and not tweakable. Color was ok but, without Legend/Values it's useless.

I saw this post and simply downloaded ParaView and opened a .vtk file I exported from FC. Did not watch any vid's or read any 'How to do it'. Bingo! Intuition and one click resulted in success.

[HarryvL]

Perforated (10x100mm) steel strip under uni-axial tension

Perforated_Strip_Model.png (111.97 KiB) Viewed 16656 times

-

Perforated_Strip_Stresses.png (692.85 KiB) Viewed 16656 times

[HarryvL]

As a step towards plotting stress trajectories (see https://forum.freecadweb.org/viewtopic. ... 85#p237229) I managed to export principal stress directions to Paraview VTK.

To be continued...

Well, I finally got it to work, but the results are not as pleasing as I hoped. Here the example of a deep cantilever beam:

Cantilver_002_0.png (52.37 KiB) Viewed 16594 times

The contour lines for minor principal stress (compression) broadly agree with the theory (see https://forum.freecadweb.org/viewtopic. ... 85#p237229), but look very ragged.

Cantilver_002_1.png (39.26 KiB) Viewed 16594 times

And here in combination with the minor principal stress direction vectors:

Cantilver_002_2.png (107.34 KiB) Viewed 16594 times

which should point along the contour lines (which they approximately do) ...

... and finally in combination with the major principal direction vectors, which should be perpendicular to the contour lines:

Cantilver_002_3.png (134.54 KiB) Viewed 16594 times

The method I came up with involves solving a set of 1st order partial differential equations for the mesh domain using the finite element method and is therefore very computationally intensive. If it would ever be of general interest, it would require Cython or F2y to turn into something with acceptable
performance.

I spent a long time looking for the bug that caused the raggedness, but come to the (tentative) conclusion that there isn't one and that the lack of smoothness is (potentially) due to the fact that in some areas the minor principal stress is in the plane perpendicular to the plane of interest (this is a 3D analysis!), creating singularities in the surface used to create the contour lines.

For now I will move on the other things and open a new topic for that ...

[HarryvL]

Found a bug. Better, but still not right...

Cantilver_002_4.png (68.31 KiB) Viewed 16581 times

[HarryvL]

This is as good as it gets:

Cantilver_002_5.png (62.55 KiB) Viewed 16561 times

I conclude that the easiest way to visualise stress trajectories is by simply plotting principal stresses, like in my earlier posts.

[thschrader]

Harry,
amazing work!
I like the perforated plate (the stress "flow" around the holes...)
Thomas