yade-users team mailing list archive

[Jan Stránský <honzik.stransky@xxxxxxxxx>]

Hi Alexander,


> At the previous picture i showed "magnitude" of stress (I can select such
> a mode in Paraview, but frankly i don't understand yet how it represents
> 3x3 matrix tensor
> http://i11.pixs.ru/storage/5/9/2/pic1JPG_5715109_18008592.jpg :)
>

exactly, it is always good to know what you are plotting :-) I don't know
what magnitude for tensors is, therefore I don't use it :-) paraview
numbering is 0=xx, 1=xy, 2=xz, 3=yx, 4=yy, 5=yz, 6=zx, 7=zy, 8=zz. So for
the first idea, I would use component 0 (xx)


>
> Hm, I think the displacement of spheres belonging to KGHO  is constant
> because of the minimal value is set to 0.
> (http://i11.pixs.ru/storage/6/0/7/pic2JPG_9158907_18008607.jpg).


Again, here you plot magnitude, i.e. length of the vector. I would use 0
component (x direction)


> You are
> right, spheres of KGHO  fixed just along Y direction. Also it seems to
> be a simple uniaxial loading, but as i mentioned in future i will
> replace specimen with another plate which consists inclusions of
> different metal (mo). Now i need to calibrate test that's why using
> simple uniform plate with stiffer material.
>

ok. What is actually your motivation to use DEM for this example?


>
> Today i created simple tension test in Ansys with the steel plate, the
> following picture shows stress distribution.
> (http://i11.pixs.ru/storage/6/3/3/pic3JPG_3955405_18008633.jpg). I want
> to see something like this in Yade, but after Young's modulus and
> Poisson's ratio were changed, the results become even worst. The spheres
> under constant force simply rip off the model.
>

Probably it is because you use CpmMat, intended for simulating concrete
including damage. To remain in elastic range, add neverDamage=True to
CpmMat definition.


> (http://i11.pixs.ru/storage/6/4/2/pic5JPG_4026499_18008642.jpg , huh i
> found how to show coordinate axis)
>
> With you advise i'm proccessing  to cut my MWE example. Now i try to use
> VTKExporter to save data but _builtin__.my_stresses doesn't work. That's
> why i save just dicplacement, you can uncomment variant with
> __builtin__.my_stresses . Pleace could you test it and said what's wrong
> with _builtin__.my_stresses.
>

each component of stresses is tensor (Matrix3), you can't pass it to
float() function (Yade should give you TypeError). Just delete float and it
should work.

Hm, u right in picture
> http://i11.pixs.ru/storage/6/0/7/pic2JPG_9158907_18008607.jpg
> displacement of spheres belonging to KGHO  it's not real 0, but may be
> it's because of rotation


it is because you plot magnitude, taking into account also lateral
displacements (which are not blocked and therefore nonzero).

cheers
Jan