timestep

Submitted by yangdm on Fri, 09/23/2011 - 11:09

Hi, all

How to choose an appropriate timestep in LIGGGHTS? In the examples, I presume the timestep, e.g. 0.00001, is not generally applicable.

Normally a ratio of the out-of-balance force over the average contact force is used to assess if the granular system is stable or not. I'm just wondering whether this is robust in LIGGGHTS or I have to take care of the timestep by myself in each simulation?

cheers

Forums:

Yansan | Fri, 09/23/2011 - 13:50

Hi,

I scale the timestep to the contact time of the particles. A common value is timestep=tc/50 where tc is the contact time.

The calculation of the contact time for a simple case can be found here and should give you a start:

S. Luding
Collisions & Contacts between two particles, in: Physics of dry granular Media, eds. H. J. Herrmann, J.-P. Hovi, and S. Luding

Greeting,

Yansan

ckloss's picture

ckloss | Sun, 09/25/2011 - 16:09

Hi Yang - as Mark said there are two time scale criteria implemented in LIGGGHTS - wave propagation time scale (after Rayleigh) and contact time scale (after Hertz). Depending on how accurately you want to model your system, the time step should not exceed a certain percentage of these time scales. I would say the 2% that Yansan suggests make a very accurate model. If you need some speed up to model a larger system, go up to 10% or so. The stability limit for LIGGGHTS is somewhere between 15% and 25%, depending on the system

Christoph

msbentley's picture

msbentley | Fri, 09/23/2011 - 16:23

You can use the fix check/timestep/gran command to compare your current timestep against the Rayleigh and Hertz timesteps (see the docs) and warn if it's too large. There is also a fix dt/reset command, which can change the timestep according to how far a particle moves in a timestep.

Cheers, Mark