yade-users team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

>         not sure to understand your answer ;( What do you actually
>         mean with "interaction" in this regard? Are you talking about
>         the contact point? I would appreciate if you could say more.
>         When I will have got it I will add documentation (since now to
>         me it is still not clear).

Interaction is (usually) positioned in space by its contact point and
oriented by what we call normal (for spheres, unit vector from one
sphere's center to center of the other one). Imagine now that spheres
rotate around the contact point, without changing their mutual position;
obviously, this should not change shearForce taken relative to the
particles. However, shearForce is expressed in global coordinates,
therefore it must be updated so that it "follows" the motion of spheres
in global space.

Global & local coordinates:

If shearForce were expressed in interaction-local coordinates, it would
not have to be updated; but unfortunately, 2 points (centers) in space
only define direction (normal) and perpendicular plane (shear plane),
but do not define orientation of the plane itself. 

In some FEM codes, you have to define reference point for beam elements
(besides the two nodes that are connected by that element), so that it
is possible to define local coordinate system, fixing orientation of the
cross-section (shear plane, if you will).

Since DEM happens in large strains (any motion is possible), any
reference point could become aligned with the two centers, leading to
loss of uniqueness of the local system. Therefore, we need to use global
coordinates for shearForce.

HTH, Vaclav