Recreationall TRIMIX

[dfreeman]

Mike,

This is a great observation. What is happening is that there is less turbulence from the gas entering the cylinder the slower you add the gas, reduciong the mixing. This observation answers a question that I have had for a while. With no mixer in the scuba tank, what is the controlling form of mixing when blending gases? The choices are: free convection from temperature difference, diffusion, free convection from density difference, or flow turbulence. All of these have an effect, which is the big one, especially working with the supercritical conditions at scuba tank pressures where gases act more like liquids.

In your experiment, you are adding the heavy gas first and then the light gas, this will minimize free convection from density difference. You are adding the gases slowly, therefore it is more difficult to set up a thermal convection cell since there is not a lot of temperature difference from the inside to the tank surface. What is left is diffusion and turbulence. Gases usually diffuse rapidly, but at these pressures they act more like a liquid, slowing diffusion. In your experiemnt, the flow is low and there is not much turbulence for mixing.

The results is that the gases layer on you and only mix after filling by diffusion and this diffusion is slowed by density difference. The geometry of the cylinder enhances this effect.

There are bar drinks where you can layer different density liquors in a glass if you add the liquors in the right density order and very carefully so that you do not intorduce turbulence. Give the glass a shake, the drink rapidly mixes due to the turbulence. The shape of the glasses used are tall and thin, like a scuba cylinder.

Tip you tank over and rolling it does the same thing. Now there is enough turbulence to mix the gases and away you go.

Your observations answer several scuba gas mixing questions that I have been curious about but have not bothered to calculate (the physical data is hard to find at these pressures and I know that I will probably have to deal with Bessel functions in the end based on past experience).

Keep the information coming.

Doug