Now, I understand more what you want for the simulation.
There is no gravity in the previous version.
I just added gravity (can be changed with slider) and also draw Y[sub]average[/sub] for three different particles.
The difference between those three curves (Y[sub]average[/sub])are due to statistical error (because number of particles are not big enough).

More options are added.
1. Gravity g can be changed with slider.
2. Collision between particles can be turn off with check box.
3. The top boundary can be turn on/off.

The separation process will be faster if you un-check [u]closetop[/u] check box.
The above simulation has been updated with new version.