Simulation of Granular Material with a Hard Sphere Model

We are simulating granular material as hard spheres and calculate the trajectories of the particles using an event-driven molecular dynamics algorithm. Goal of this research is to understand the behaviour of the granular material under different gravitational forces (zero-g or micro gravity, 1g and hyper gravity).

Parabolic flight experiments

Some results of an experiment of granular material under 0g on a parabolic flight are shown here. The first flight was a student parabolic flight campaign in September 2002 from the european space agency (ESA).

The videos are avi-files encoded with an mpeg4 video codec. To view the videos under Windows, please download the DivX-codec (, under Linux you can use the MPlayer ( The numbers specify the parabola number (video size between 2-3MB):

  • Sand(red)-Wood(10mm,white) in cylinder (filled to 1/2): 23
    "Sand surrounds the wooden spheres leaving a halo"

  • Sand(red)-Stones(green) in rectangular-box (filled to 1/2): 28
    "Sand surrounds the stones leaving a halo"

  • Glas(2mm,white)-Wood(8mm,blue) in cylinder (filled to 1/2): 33, 34
    "Rotating compact structure of wooden speheres surrounded by glas spehere"

  • Glas(2mm,white)-Wood(8mm,blue) in rectangular-box (filled to 1/3): 43, 44, 45, 46
    "Compact structure of wooden spheres in the center of the box surrounded by glas spheres"

  • Glas(2mm,red)-Wood(10mm,white) in half-cylinder (filled to 1/3): 47, 48, 49, 50, 51
    "Compact structure of wooden spheres in the center of the half cylinder surrounded by glas spheres"

  • Glas(2mm,red)-Wood(10mm,white) in rectangular-box (filled to 3/4): 57, 60
    "Formation of a quadratic lattice by the wooden spheres with a small gap between the spheres"

    1g and 0g: 2s, 3s, 4s, 9s, 13s

Some results of our simulations are shown here.


Testing the (reverse) brazil nut effect with the following spheres (large/small): 4:1 diameter, 1:2 density. We should see a reverse brazil nut effect (large particle end up at the bottom of the box), but in our simulations, we see that the outcome depends on the starting position of the large particle:
large particle starts at 5% height (360kB)
large particle starts at 15% height (350kB)
...and a simulation with 10 large particles and 400 small ones (3MB).
You see in the last simulation, that not all large particles end up either at the top or bottom of the box. So there is no clear (reverse) brazil nut effect!


  • What will happen when granular material is shaken under zero-g ?
    Lets first see, what happens when shaken horizontally...

  • How does the 2-body-interaction potential for the large particles look like, if we assume, that the small particles act as media for the large ones ?


The simulation is written in C and is available under the GNU General Public License.
Download the version of 11.3.2004:

  • A very good list of references related to granular material
  • A description of an algorithm for an event-driven molecular dynamics simulation
  • Granmat - a zero-g experiment from students in Regensburg, flown on a ESA parabolic flight campaign