Hi everyone,
I am now working on the simulation of Janssen effect with Liggghts. The command I used in the input script is
"stress/atom" , no keywords is attached after this command. (The keywords attached can be ke,pair, bond,angle, dihedral, improper, kspace, fix, or virial ).
A packing is simulated in a cylinder with 20 0000 granules.
The result is that, after posprocessing the data of sigmazz, I drew the graph of sigmazz vs the height. The trend of the curve is right, which shows the Janssen effect. The postprocessing contains two steps: first, sum all the simazz value in a certain interval height, delta h; second, devided the sum by the value (sectional area of the cylinder multiplied by the delta h).
Are the command and keywords used here right? what do the command bond, dihedral, improper mean here? I really doubt the process and the result aquired by this method.
Please help me~~~
I would like to discuss with you about the simulations of granular flow and packing.
Yours Goddy
2015.04.18
| Attachment | Size |
|---|---|
 the curve of sigmazz vs height | 497.57 KB |
richti83 | Mon, 04/20/2015 - 08:36
other method
I did a similar investigation with another method: I devided the bounding mesh into small stripes over hight and used mesh/surface/stress for every stripe, than dumped pressure to vtk using dump mesh/vtk and used a Paraview programmable filter to sum over two triangles in the same area.
The results are quite interesting and showing a big divergence for different rolling friction models.
See this graph:
http://www.richtisoft.de/transfer/uncompressed.png
http://www.richtisoft.de/transfer/voidfraction_and_pressure_over_height_... (F over t)
(note: typeC is my personal implemention of a rolling friction model)
The results are not statisticly satisfied yet !
ckloss | Mon, 05/11/2015 - 12:54
Hi Goody and Christian,
Hi Goody and Christian,
alternatively you can also use fix ave/euler!
best wishes
Christoph