guys, I got your points! But I will need some time to play an make some models to test what you have described.
Mhh I'm structural engineer. Buildings have self weight in z-direction with a fixed acceleration, they have loads and fixes, all well defined ... I never had model like you described but it totally make sense to me.
bernd
Why are Neumann BCs mandatory?
Moderator: bernd
Forum rules
and Helpful information for the FEM forum
and Helpful information for the FEM forum
Re: Why are Neumann BCs mandatory?
Just to get it the rigth way. From my point of view the only case which is not handled by the implementation above is no constraint fix, no constraint force or pressure or self weight, just constrained displacement and all is free and no imposed displacement. Thus Calculix does not return an error but results are zero. (I'm structural engineer. This would be a building flying around it spacebernd wrote:done. Would be coold if you guys could give it a try.bernd wrote:we need to check if really a strain and not a fixed was applied (the displacement constraint could be both). ...
I need to adapt the pre checks in this regard,
branch:
https://github.com/berndhahnebach/FreeC ... constraint
commit:
https://github.com/berndhahnebach/FreeC ... 012bcabb87
YEAH it is the russian MIR space station ). It is what Sgrogan described. But may be Jeremy is on the right track and we should just let the user decide an delete the force displacement pre check at all. Ahh could you check the file attached, it calculates without error in ccx but the results are different from what I would expect.
bernd
- Attachments
-
- wrong_move.fcstd
- (19.23 KiB) Downloaded 32 times
Re: Why are Neumann BCs mandatory?
Usually an object is not flying that's true. But in a lot of engineering situations the Free situation is what they want to know. just because it don't have any interactions.
The worst case is that the first 6 eigen modes are around 0 Hertz.
The worst case is that the first 6 eigen modes are around 0 Hertz.
Re: Why are Neumann BCs mandatory?
removed the load test at all, see https://github.com/berndhahnebach/FreeC ... 11d57f7819
https://github.com/FreeCAD/FreeCAD/pull/425
https://github.com/FreeCAD/FreeCAD/pull/425
Re: Why are Neumann BCs mandatory?
Without _any_ Dirichlet BC, the solver will fail. There is no way to solve the floating MIR. If you set one Dirichlet BC (no matter if it is homogeneous or not) but the solid may still rotate as a rigid body, then depending on how the solver handles singular matrices, it can work or fail. If you insert the appropriate boundary conditions in order to remove all rotations, the matrix should become non-singular and have an unique solution.bernd wrote:Just to get it the rigth way. From my point of view the only case which is not handled by the implementation above is no constraint fix, no constraint force or pressure or self weight, just constrained displacement and all is free and no imposed displacement. Thus Calculix does not return an error but results are zero. (I'm structural engineer. This would be a building flying around it spacebernd wrote:done. Would be coold if you guys could give it a try.bernd wrote:we need to check if really a strain and not a fixed was applied (the displacement constraint could be both). ...
I need to adapt the pre checks in this regard,
branch:
https://github.com/berndhahnebach/FreeC ... constraint
commit:
https://github.com/berndhahnebach/FreeC ... 012bcabb87
This is like my example above. The beam is allowed to rotate around the main axis. There are infinite solutions, but the solver manages to find one (good for CalculiX in this one). That is what happens with FIno https://www.seamplex.com/fino also.bernd wrote: Ahh could you check the file attached, it calculates without error in ccx but the results are different from what I would expect.
bernd
Bernd, I will try you branches and let you know.
--
jeremy
Re: Why are Neumann BCs mandatory?
jeremy it's in official master already.
Re: Why are Neumann BCs mandatory?
yes, I already found outbernd wrote:jeremy it's in official master already.
It works as expected. It is now my turn to add support for Fino for an alternative to CalculiX...
- Attachments
-
- works.png (311.52 KiB) Viewed 1492 times
Re: Why are Neumann BCs mandatory?
Yeah go for it! Have you seen the femsolverelmer branch on my gitgub. It may help to get started.kuroshivo wrote: Imit t is now my turn to add support for Fino for an alternative to CalculiX...