[mullenc525]

I've been studying heat and water vapor exchangers for some time. I understand there is a close relationship between heat transfer and mass transfer, so initially sought to understand heat transfer in the device. My geometries consist of channels on the order of 1mm rectangles with 10-20mm length, with periodic inlet/oulet.

Initially I studied one species (air) with an inward heat flux at one wall and a balancing energy sink over the entire domain. The results were as expected.

I am having trouble attempting to accomplish the analogue with species transfer instead of heat transfer. I have entered air and water vapor properties at the temperature and pressure of interest, and initialized the domain with the volume fraction appropriate for my device. One wall has a water vapor source, and the entire domain has a source with a negative constant to balance mass over the periodic domain. The negative source is extracted with velocities u,v,w, as I don't want it to affect the momentum of the air.

The rms residuals for species volume fraction are on the order of 0.02-0.04 after 60 iterations with a 10x timescale factor in a simple geometry (rectangular duct with appropriate inlet profile). Basically it is not converging, and of course the solution is nothing like expected.

Under fluid pair models I'm asked to enter information for interphase transfer, and drag coefficient. I don't know what this means or if its appropriate, it sounds to me like information for a 2-phase model.

Not sure what other details may be appropriate: steady state, laminar, both fluids continuous.

Does anybody have any experience with diffusion/advection of one gas in another?

edit - I just noticed another problem; the water vapor does not have the same velocity as the fluid average, as it is diffusing perpendicular to the stream. That makes it difficult to choose what velocity to remove it at without impacting the momentum of the other air.

[mullenc525]

bump - surely somebody must have done a problem with water vapor diffusing into air?

[ghorrocks]

Are you looking purely at diffusion of vapour or does the vapour change physical properties much (ie condense or evaporate, or even just have a significant pressure/temperature change)?

[ghorrocks]

Simple advection/diffusion can be done very easily with an additional variable. Don't use multiphase unless you need to (and you know how to use it).

[mullenc525]

Hi Glenn,

Purely diffusion of vapor, no change in state, pressure or temperature. I noticed another thread where someone was looking at ink diffusion into water, and the consensus was to use additional variables. However, a caveat mentioned was that it was only appropriate if you wanted to make the assumption the ink properties were the same as water and mixing ratios low.

As we know, water vapor and air are substantially different, and I am working with mixing ratios of about 0.24 (not insignificant).

Are additional variables appropriate?

[ghorrocks]

You can make the properties of the fluid a function of the amount of water vapour to some extent. This should work for a limited range of mass fractions, linked to viscosity, density.

Rather than a multiphase approach you might consider a single phase but multi-species approach using mass fractions. If you can't get the range of properties you need you should consider this approach as it will be able to handle pure air to pure vapour and everything in between.

[mullenc525]

Thanks Glenn,

I've gone the multi component route instead of using an additional variable; it's amazing how easy it is when you know what you are looking for!

I've selected water vapor to follow the transport equation with air using the constraint equation. Simply because I know the diffusivity of water in air and this is what somebody else did. Is this appropriate, and how does one select which to use in general?

Curtis

[ghorrocks]

I am no expert on multi-component models but I don't think it matters which one you set up as the constraint. I would try it either way and see what the difference is.