The problem of material parameters.

Hello,

[1] is a good starting point for discussion.
Also see similar discussions [2,3]

What would be the use of the same "kn" and "ks" for all contacts, regardless particle size?

> In Yade, we set ..

please be more specific, as it also depends on used materials and contact laws.
Are you refering to Law2_ScGeom_FrictPhys_CundallStrack?

> the modulus and poisson’s ratio directly.

please be more specific.
You set directly Young's modulus and Poisson's ratio for the Hertz-Mindlin contact laws.
You set "young" and "poisson" named parameters for Law2_ScGeom_FrictPhys_CundallStrack, but they "are not directly" Young's modulus and Poisson's ratio.

> kn = 2.*E*R1*E*R2/(E*R1+E*R2), Ks/Kn=P ### some people say that kn = E*R1*E*R2/(E*R1+E*R2)

see [1], there are other choices than 1 or 2 for the multiplier.
Basically it is (some constant) * (harmonic mean of stiffnesses)

> That’s to say the kn at different contact of the particles differs because the particle radius varies (different combination of R1 and R2).

Yes, and it is desired and expected behavior.
Imagine a two exactly touching spherical particles. You then push them together by given displacement "d". The particles are compressed with a certain repulsive force.
Now imagine the same situation, but particles has double radius and are pushed the same displacement "d".
In reality, you would expect higher repulsive force. Higher force is also result of the harmonic mean taking into account particle size.
With constant kn, you would have "unphysical" same repulsive force.

> However, we can set the value of kn and ks directly in PFC, which means the kn at different contact are the same.

I don't know PFC.
But it may be that PFC "kn" and "ks" are equal to Yade "young" and "poisson" and are actually particle parameters and not actual contact parameters?

> Why this difference in software development?

I call it "implementation choice".
There are some formulas, principles, ... that may be expressed and computed in different ways (although with the same result).
The developer then has to choose one.

In Yade terms, setting interaction kn and ks directly or to the same value is possible, but it is not "physical", that is why the harmonic mean approach is (mostly) used.

Cheers
Jan

[1] https://yade-dem.org/doc/formulation.html#normal-stiffness
[2] https://answers.launchpad.net/yade/+question/298001
[3] https://answers.launchpad.net/yade/+question/679638