FreeCAD Forum

Nice.
But with FEM you will always have to worry if you have made your fixation in a correct way.
Why not use something like rigid body simulation like... https://m.youtube.com/watch?v=nHVYYMG3QVY

HoWil wrote: ↑Mon Aug 17, 2020 6:58 pm Nice.
But with FEM you will always have to worry if you have made your fixation in a correct way.
...

Yes, thats the problem. My solution is constructed.
Because I know the analytical solution. A fake...
sorry.

In this case the axis of rotation is obvious. So you can manipulate the fixings/load in such a way, that the z-component is zero. That is the point of equilibrium, where the external load is in balance with the selfweight of the table among the rotation axis. Case a) and b). If I don’t know the analytical result a priori and ad a fixing more, the result is wrong, case c).
With an unknow axis of rotation (the “round” table) a rigid body approach wont work, in my opinion.
I used a beam model to show what I mean, hope you get the point…

Hi

Today I tested an algorithm for this in Elmer and it seems to work nicely:

1) Create and solve the unloaded case: Ax=b
2) Create and solve a controllable load case: Ay=f
3) Compute reaction forces resulting from 1) and 2)
4) Find scaling factor c such that sum of reaction forces at selected node vanishes
5) Update x:=x+c*y, b:=b+c*f and continue iteration, if wanted

As said, almost similar algorithm has been used in Elmer for temperature control. The algo is hidded under the hood. The use of reaction forces is something new, and also the fact that this can now be used for many other Elmer modules as well.

The control is done in one sweep for linear problems. Geometrically nonlinear problems may be possible but would requite some iteration. Parallel operation is not yet supported.

I'm not a FreeCAD expert but with the free text fields of the Elmer plugin I would guess that this could in the future be done also with FreeCAD. You just need rather fresh Elmer. The feature has been in https://github.com/ElmerCSC/elmerfem (devel branch) for 15 minutes

My test was even simpler than any on this thread. If somebody wants to try out this for real I can give some guidance.

-Peter

raback wrote: ↑Thu Aug 27, 2020 6:29 pm My test was even simpler than any on this thread. If somebody wants to try out this for real I can give some guidance.

Yes, please elaborate.

Kunda1 wrote: ↑Thu Aug 27, 2020 6:40 pm Yes, please elaborate.

Well, there is a consistency test for Elmer in:
https://github.com/ElmerCSC/elmerfem/bl ... p/case.sif

Make the basic case with FreeCAD using Elmer as the solver (or ElmerGUI, emacs,...). Then augment the command file with lines of style 61-65 and 88. I hope these could be added to the new free text fields of the relevant sections.

raback wrote: ↑Thu Aug 27, 2020 6:29 pm ...
3) Compute reaction forces resulting from 1) and 2)
...

Peter,
how to set the fixings of the structure?
Howil has given a very good (and very difficult) question.
mmhh...

All the test case solution all boundaries at the bottom may be fixed. The code finds the critical force needed to create a zero reaction force at the chosen node. So in this system one needs to guess what is the node where point of contact is first lost ("control node coordinates"). There is no easy way to study the case beyond losing the contact since the static analysis does not have a solution then. Hence including proper contact mechanics would be an overkill.

Good afternoon thschrader, raback, HoWil and above all to the moderator Kunda1 and the whole community!
It is really nice that the post, which started with some "controversy", is producing positivity and training!
This is the purpose of the technical forums to try to solve the most disparate situations even if they seem trivial.
Good continuation and experimentation, I am following the evolution with a lot of curiosity and passion.
Thank you!

raback wrote: ↑Thu Aug 27, 2020 6:29 pm I'm not a FreeCAD expert but with the free text fields of the Elmer plugin I would guess that this could in the future be done also with FreeCAD. You just need rather fresh Elmer. The feature has been in https://github.com/ElmerCSC/elmerfem (devel branch) for 15 minutes

My test was even simpler than any on this thread. If somebody wants to try out this for real I can give some guidance.

-Peter

Nice!
Would like to see my freetextinput branch to me merged.
Unfortunately, currently and for the complete September I am completely booked.
So, if anyone is interested in trying it out or extend it feel free to go ahead. https://github.com/HoWilgh/FreeCAD/tree ... etextinput
BR,
HoWil