yade-dev team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> Hi V., if I correctly understood, the shear force (the incremental
> one) behaves as you describe, but the TOTAL shear force (the one that
> we store in the simulation) is really expressed in the global system
> (right?).

(let's suppose that https://bugs.launchpad.net/yade/+bug/493102 is
solved for the rest of this discussion)

shearForce (the total, current one) is defined for interaction (not for
particle), in global coordinates. Now, let's say shearForce is
(-1,-1,-1), and that we have 2 bodies, A and B, which are assigned to
id1 and id2. It means that you apply (-1,-1,-1) to A (id1) and
-(-1,-1,-1)=(1,1,1) to B (id2).

If you swap the particles (A=id2, B=id1), then shearForce will be
computed in the inverse sense:

See e.g. ScGeom::updateShear (or updateShearForce), where it is clear:
normal is already inverse (always from id1 to id2), then everything is
symmetric for id1 and id2 (therefore the same if you swap id1/id2);
relativeVelocity will be inverted, so will be shearVelocity,
shearDisplacement and shearIncrement / shearForce increment.

(This will happen similarly in Dem3DofGeom, but it is less obvious to
follow in the code)

With swapped particled (A=id2, B=id1), therefore, shearForce will be
(1,1,1). But we apply now (1,1,1) to B (which is now id1) and (-1,-1,-1)
to A (which is now id2).

So you see real forces on bodies are the same.

> Moreover if my external force on the particle is in one direction, I
> would always expect an opposite sign for the total shear force (we
> oscillates around the same value but the sign should always be
> opposite to it). 
Sorry, I am not clear as to what you mean here.

HTH, Vaclav