Hia, I'm after a good combination of visual effect and real science.  I think i know what will happen in reality and I'm trying to form a model that will show what happens with a set of principles and constraints that can be argued from  scientific first principles.  The effect of gravity will not be the same on all particles as the force downwards will depend on the mass. the acceleration downwards might be uniform but there is also the buoyancy effect from the displacement of the other atoms as one falls due to gravity. The lighter gasses will rise to the top and the heavier gasses will sink to the bottom, relative to the mass of the main constituent of the volume. Setting the gravitational effect on gas A as uniformly up and down (i.e. no effect) was in my mind a way of simplifying the process to make the model appear more like the real situation.  This should stop the appearance of a vacuum at the top and a dense cloud of gas at the bottom.  The relative effects would be multiplied by the gravitational acceleration.

In effect the particle A should have no relative gravitational effect, B should have a positive effect in the upwards direction and C should have a positive effect in the down wards direction.

A balloon of helium has the potential to lift the amount of weight that is equal to the difference in mass of the volume of Helium and the mass of air that is displaced. There is therefore an upward force provided by each molecule that is equal to it's buoyancy or mass difference  times the gravitational constant.

To explain this better 24 litres of helium would displace 24 litres of Air  so the mass of the 24 litres would be 4gramms rather than 29gramms. There is therefore the potential to lift 25 grams  so the overall force upwards is 0.025 x 9.8 N. mathematically this equates to  (Mb-Ma/1000)*9.8 (negative = upwards, positive = downwards) 

In the case of an individual molecule it would be (Mb-Ma)x1.66054 x 10E-27 X 9.8 N. again this would be negative for upwards and positive for downwards.

We can easily expand the area so that there is more space between the molecules if that helps the reality of the gas.

In a big enough system the gasses should separate out completely given enough time with only a small overlap  in the "mixing region". 

If I can get it to give the results I need with a reasonable level of scientifically sound assumptions then I'd like to use it in a lecture and a scientific paper.  (Naturally you will be acknowledged.)

Sorry if it seems a bit confused above.

C