yade-users team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

> Saying that "it is better do define interaction force as it applies on
> id2", you mean, in the case of ScGeom:
> 
> Vector3r f = phys->normalForce + shearForce; (POSITIVE)
> ncb->forces.addForce(id1,-f); (NEGATIVE -> compression)
> ncb->forces.addForce(id2,f); (POSITIVE -> tension)
> ncb->forces.addTorque(id1,-c1x.Cross(f)); 
> ncb->forces.addTorque(id2,(c2x).Cross(f))
> 
> right?
Almost. f is a Vector3r, and it will be positive/negative with respect
to the normal: normal.dot(f) negative for compression and positive for
tension. (Geotechnicians have opposite convention, which might have been
the source of the inverse; but they are persuaded now ;-) ).
> So it is clear that the convention between ScGeom and Dem3DofGeom is
> different, but why? The normal vector in both cases is defined in the
> same way. Where is the difference that matters in terms of signs? It
> is just to know, at the moment I am using ScGeom but it is useful to
> understand such basic things and where they do come from. Actually I
> had a look at Dem3DofGeom but is quite hard to understand although it
> is probably better defined than ScGeom. It would be great to have the
> same convention because this could cause a bit of confusion at least
> at a first look.
As Luc already mentioned, interaction of 2 bodies defines direction, but
not sense. It is purely conventional which way it is, but I agree it
should be uniform. I will fix that next week, I hope.

Thanks for your quesions/remarks on this.

Cheers, Vaclav