Too many neighbor bins & DEM timestep selection

Submitted by xiaobaishu67 on Thu, 04/23/2015 - 00:57

Dear liggghts users,

I use hertz contact model to simulate the sandy soil and the young's modulus is 20e9 for sand. But the contact force is so large with this 20e9 modulus so that it keeps telling too many neighbor bins message. So I am thinking to use a very small timestep. Here comes to problem, if I use small timestep like 1e-8 or 1e-9, the simulation is time consuming. So what can I do to enhance simulation time and make the simulation no error by that I mean reasonable?
Hope anybody could guide me!!! Deeply appreciated!!

Best wishes
xiaobaishu

Forums:

Daniel Queteschiner | Wed, 04/29/2015 - 16:34

In most cases you can get away with simply using a smaller Young's modulus.

nikros | Thu, 01/28/2016 - 09:05

I have the same error "Too many neighbor bins". I want to simulate steel particles with Young Modulus of 200 GPa (200.e9). The problem of "Too many neighbor bins" disappeared only if i decrease the Young Modulus up to 2.e9. Most of metals have Young modulus of hundreds GPa (in si units).
But most of examples of LIGGGHTS have Young modulus of 5.e6. Does it means that LIGGGHTS cannot simulate metal particles?
Or the value 5.e6 is in "per square inch"? So which units of Young Modulus do i need to use if i change units from "si" to "cgs"?

NTT1508 | Fri, 01/29/2016 - 04:43

5e6 should be in SI unit. I suppose the reason for suggestion of using small Young Modulus is that it would not effect much on the final result (for hard particles contacts). You can try to test this assumption by running a simple case with small number of particles then compare the result.

nikros | Tue, 02/02/2016 - 05:44

Sorry, i'm still confused. The value 5e6 Pa (from the examples in the LIGGGHTS-PUBLIC-master/examples/LIGGGHTS/Tutorials_public/ directory) is for the material which is sorter than the softest rubber (the Young's Modulus for rubber is ~ 1e7-1e8 Pa). Are you sure that it is in Pa, bur not in "per square inch"? The value of a few Mega (5e6) would be very appropriate for wood or sandstones if it is in psi "per square inches" .
http://www.engineeringtoolbox.com/young-modulus-d_417.html
I need to use cgs units (because the size of my particles is ~tens of microns), the Young's modulus of steel in cgs units is 180e10. So i receive the "Too many neighbor bins" error even if i use only 10 particles.

Daniel Queteschiner | Tue, 02/02/2016 - 15:07

Young's modulus in SI units is Pa, and Ba in cgs units.
The value 5e6Pa in the LIGGGHTS example cases is of course much lower than the Young's modulus of materials you find in literature. The reason is that the Young's modulus has a heavy impact on the time step required to resolve a particle contact. The higher the Young's modulus the smaller the time step the longer your simulation. On the other hand, a smaller Young's modulus may not be critical to the behaviour of the system under investigation. In this case you can specify a smaller Young's modulus and reduce the computation time.

mtennant | Thu, 02/07/2019 - 23:24

Hi,

I have also recently had this error when running a hertz contact model and have attempted various suggestions that I have found on this discussion but without any success. Regardless of the radius or young modulus inputs. My script works when I am using 5 different particle shapes however when I increase this to 10 I find this error. I was wondering if there was any suggestions on how to get the script working.

Thank you

arnom's picture

arnom | Thu, 02/21/2019 - 11:24

What is the ratio between the different particle sizes? Also you can try and play around with the neigh_modify command that allows you to set the bin size. If your domain is too large and the bins too small, then you'll run into this issue.

DCS team member & LIGGGHTS(R) core developer

mtennant | Mon, 02/25/2019 - 18:18

Thanks for replying,

The ratio is split evenly with 0.2 for each of the 5 multisphere particle types. I managed to get the simulation to run however I have ran into the issue that the particles are lost while they are settling. LIGGGHTS displays a warning of the energy may not be correctly calculated so do you think this may be causing this issue?

Thanks