What's the physical meaning of characteristic velocity?

Submitted by jerry_1988 on Wed, 03/23/2011 - 19:55

Hi,

I'm a new comer to use LIGGGHTS. I can see that the characteristic velocity is a parameter used in the contact model of LIGGGHTS. What's its physical meaning and how to set it?

Thanks!

Jerry

Forums:

ckloss's picture

ckloss | Wed, 03/23/2011 - 20:41

I would think of it as a "linearization point" for the linear Hooke contact model. See the doc pair_gran.html to see how it influences the calculation of the contact stiffness.

Christoph

jerry_1988 | Thu, 03/24/2011 - 21:21

Thank you very much. But how to select its value. For example if I want to simulate one kind of sand, how can I know the characteristic velocity of this sand?

Thanks!

Jerry

ckloss's picture

ckloss | Fri, 03/25/2011 - 08:12

I would not say it's a material property, I would rather choose it as characteristic velocity that is to be expected in the simulation.

Christoph

J_Wang | Thu, 04/26/2012 - 01:50

Hi Christoph,

Are there any references for the calculation of a contact stiffness kn in hooke contact model in LIGGGHTS. In many other DEM softwares, the contact stiffness for hooke linear model is set by hand directly. May you give me the references that how the young's modulus, possion's ratio and velocity effect the contact stiffness? That will give us a better understanding of the contact model in LIGGGHTS for all the users.

Thank you.

J.Wang

Yansan | Sat, 04/28/2012 - 21:24

As far as i know the references can be found in the documentation of liggghts.

marketos | Fri, 03/25/2011 - 13:18

Hi,

I think Jerry is making a very good point here. Sometimes I prefer specifying the stiffness values directly, eg kn, kt, gamman, gammat. This is what was originally done in LAMMPS.

Would it be possible for the user to select whether he wants to specify these directly or not? In my LIGGGHTS version I have just added a new .h and a new .cpp file to allow me to do just this. Sometimes it does not make 100% sense to link stiffnesses to material properties, as the contact behaviour can be extremely complex, with plastic deformation and other phenomena giving you a response that is much softer than what might be expected from just elastic deformation of eg quartz.

George

ckloss's picture

ckloss | Fri, 03/25/2011 - 15:24

Yes I agree. It would be easy to take the LAMMPS models, by taking those classes, deriving them from PairGranHookeHistory with the need to define just init_substyle() and deriveContactModelParams(). This would let the user define kn,kt etc by means of fixes of type property/global as it is done for Y,nu etc in the current LIGGGHTS models.
Same with the wall styles. This would allow the user to use the full LIGGGHTS functionality (meshes etc)

Christoph