In the simulations N spherical particles, with diameters
, (i = 1,...,N) are used. If not explicitly mentioned
we use monodisperse spheres of diameter 
mm.
The N particles are placed into a container
with different boundary conditions at the bottom and also different
system sizes. Starting from a regular
close-packed triangular arrangement with L particles
in the lowermost layer M=0 at the bottom, we model heaps of slope
or 
by adding 
or 
particles for layer M
respectively. The number of particles is thus 
or
with the number of layers 
or 
int[(L-1)/3]+1. The largest pile we simulate has
L=100 and thus 
.
The initial velocities and overlaps of the particles are set to zero if not explicitly given, gravity is slowly tuned from zero to the selected magnitude and the system is simulated until the kinetic energy is several orders of magnutide smaller than the potential energy, and the stresses no longer vary. The particles at the bottom layer M=0 are either fixed, or may slide horizontally and penetrate the bottom vertically. In the sliding case, only the outermost particles are fixed in horizontal direction by the sidewalls. For a schematic drawing of the four possible situations see Fig. 1(a). The possible configurations of a regular contact network are schematically drawn in Fig. 1(b).
Figure 1:
(a) Schematic drawing of a pile in a box with smooth, flat bottom (left),
and on a bumpy bottom (right), with 
.
The solid bar at the right indicates that the particles in row M=0 are
fixed, so that the first relevant row with mobile particles is
M = 1 with here 
.
(b) Schematic drawing of the typical contact network configurations
in a regular arrangement.