[kjetil]

Dear CFX folks,

From time to time I come across the problem to look at pollution and air intakes, having to use buoyancy. Normally I get around this by using incompressible gas/air. This time I'm looking at a small amount of H2S gas into air.

First I thought I might still be able to get an "incompressible mixture" by using the incompressible versions of these two gases. Turns out the variable mixture is still treated as an ideal gas, at least the Domain/Fluid properties setup is as if it was ideal gas (defining ref density, not temperature).

I then tried to create a formula for the outlet pressure, using a completely steady (no flow) setup of the domain. This is inspired by the Ansys solution, which I didn't quite get to work (attached).


Short story: still using a proper initialization, nothing produces realistic results. Have anybody come across a proper method to get such a setup to work?


- as the release of H2S is very small, would it be possible to run the air first, and a sort of one-way coupled for the polluting gas, as it is for particles?
- would it be possible to have the gases be treated as fully incompressible if run as a "faux" liquid [mixture]?

[Gert-Jan]

In a given flow field (set solve fluids = f), why not add the H2S as a passive scalar. This should be possible if the amount of H2S be very low, not?
And, since it is very low, why not also ignore buoyancy?

[ghorrocks]

The simplest way to model this situation is with the H2S modelled as a convection/diffusion additional variable. This approach will allow you to continue modelling air as a simple incompressible fluid.

Modelling large columns of air as compressible gas is challenging as obtaining a stable atmosphere is not simple. If you want to do this I recommend doing some background reading in meteorology textbooks on things like potential temperature and lapse rates.

[kjetil]

Thanks, will it be possible to include the H2S density difference when using the additional variable method, you think? Approximately 16% difference between the gases.

[ghorrocks]

If you want to model the buoyancy effects of H2S things are more complex. Some ideas:
* You could continue to use the additional variable/passive scalar approach but you would need to develop a momentum source term for the buoyancy effects. You would have to look into this but I suspect it is possible.
* The most physically complete model for this is a multicomponent mixture (to model two mixed gases), but buoyancy effects on this model are not fully implemented. It might work if you use a compressible gas model - but no promises. I am pretty sure it will not work if you use incompressible gasses.

[kjetil]

Had a look at your first suggestion about momentum sources, but seems that is not available for boundaries at the moment. Only a "total source" is possible, with the units "kg/s".

[ghorrocks]

It would be a volumetric source term, not a boundary source term.