yade-dev team mailing list archive

[Klaus Thoeni <klaus.thoeni@xxxxxxxxx>]

Hi Bruno

> > Sorry for late reply, but here my thoughts. I think a separation at
> > material level would be better, i.e. FrictMat inherits ViscElastMat which
> > inherits something like  ViscElCapillarMat. I think it would make sense
> > because not all visco-elastic materials have capillar forces. What do you
> > think?
> 
> I think this is also what Anton had in mind. Base types should be free
> of capillary things, which are very specific.

So you agree that the parameters should already be separated at material 
level, correct?

> > In the case of CapillaryPhys none of the parameters is stored in FrictMat.
> 
> And it is good that way, isn't it? FrictMat has the contact parameters,
> CapillaryPhys inherits them and add the capillary parameters.
> It should be the same for ViscEl, as you suggest.
> 
> > There seams to be a bit of inconsistency. Same with kn and kt. Should the
> > material have these parameters or the Ip2 functor? IMO, stiffness is
> > related to a specific contact and should therefore be in the Ip2.
> 
> I don't get your point here. Ip2's are not supposed to have parameters
> at all, they convert material data to interaction data (e.g. they
> convert Material::Young to IPhys::kn).
> I know many Ip2 still have a few physical parameters but it is only lazy
> implementation, we should avoid that. Ip2_Frict and Ip2_CohFrict, for
> instance have no parameters at all.

The point is that MatchMaker only work for Ip2 functors and to me they really 
make sense for some specific cases (especially if you have many different 
materials). And as I said before, contact stiffness is a property of the 
contact and not the material. We generally calculate it as a harmonic average 
of the Youngs' modulus of the two materials in contact but in some cases you 
might want to define it explicitly. In this cases I think it should be defined 
in the Ip2 functor, or not?

Klaus