← Back to team overview

yade-users team mailing list archive

Re: Some issues with Facet objects (python newbie inside)


> First, I tried to duplicate the gts-horse example in order to test
> another geometry (sphere5.gts from the same source), and the result is
> quite strange with some spheres going through the facets, even with a
> very small timestep.
Gravity is too high, making bodies going through. If you use
gravity=Vector3(0,0,-1e2), it will be just fine.

> Second, I generated a simple example in order to be familiar with the
> "python way" of scripting (testGravity_Box.py  and
> testGravity_Facet.py attached). Even if I am still confused with the
> different attributes to define in the dispatchers, simulations run,
> but, I get an error when I use Facet instead of Box which says:
> yade.InteractingFacet /home/sch50p/yade/pkg/common/DataClass/InteractingGeometry/InteractingFacet.cpp:32  postProcessAttributes: InteractingFacet has coincident vertices 2 (0 0 0) and 0 (0 0 0)!

this is from utils.facetBox; since the box has zero height, it is
creating degenerate facets, whence the "warning" (it actually makes the
simulation go nuts if there is interaction with such facet, IIRC).
(Anton, utils.facetBox could perhaps check for zero dimensions and not
create those walls, no?)

You can add just 1 facet instead of the (degenerate) box, specifying all
3 vertices explicitly:


Try to avoid using Boxes... I would like to remove them at some point
since Facets (and Walls) replace them so well.

> ##### Sphere assembly: cylindric sample
> ... pack.inAlignedBox...
(you are aware of that, right?)

> Do you have any suggestions? I saw that Wall objects can also be
> defined. Are they suitable? How is defined their AABB
> (InteractingWall2AABB)?

Since Walls are always aligned with axes, their AABB's have zero width
and the other 2 dimentsions are infinite (see Wall2AABB). OTOH, there is
not (yet) implementation of something like
Sphere2Wall4SpheresContactGeometry (planned
https://blueprints.launchpad.net/yade/+spec/walls-spheres-contact-geometry) only the Dem3Dof functor is written as of now.

HTH, Vaclav

Follow ups