← Back to team overview

yade-users team mailing list archive

Re: [Question #683967]: GlobalStiffnessTimeStepper with Potential Blocks

 

Question #683967 on Yade changed:
https://answers.launchpad.net/yade/+question/683967

    Status: Open => Answered

Bruno Chareyre proposed the following answer:
Hi,

We got a similar situation some time ago with Hertzian models. Their
tangent stiffness (kn=dfn/dun)  depends on the normal force itself, and
it's actually not needed when calculating the contact force.

The solution was as follows, by far the simpler:
- reserve attributes "kn" and "ks" to reflect the derivatives dfn/dun and dft/dut (can be updated by the law functor) to be used by GSTS, and compute them just in order to determine stability condition.
- give other names to any other stiffness used when actually computing the contact forces.

I would suggest the same approach. This way GSTS will work as is with
every contact model, provided that kn, ks always have the same meaning.

If would start to introduce conditionals in GSTS to specialize it, OTOH,
it will soon be a big mess.

Bruno

-- 
You received this question notification because your team yade-users is
an answer contact for Yade.