[jayorayo]

Hi, I'm trying to model the transfer from O2 from air to water within a bioreactor using the species mass transfer option in available within the Fluent Eulerian model.
Can the water and air phases be solved initially, disabled, and then switch on the species equations to solve the mass transfer part of the analysis or are the full interactions between the phases and species required?
I was thinking that the mass transfer mechanism will take place at a much faster rate than the convection/diffusion components of the main flows so there would be no need to have them fully coupled.
Any thoughts on the subject would be much appreciated.
Thanks!!

[CeesH]

That depends very much on the details you are exactly interested in, and the level of rigor.

Conceptually "freezing the flow" gives you a valid flowfield for a given geometry, but it may be there are transients in the flow (macro-oscillations by the rotor, or by the bubble plume) that affect mixing. The flow dynamics may affect distribution of your dissolved oxygen, if there are for example regions in your reactor which bubbles cannot reach. So freezing the flow may give you a quick answer in a possible distribution of your dissolved oxygen, but it may omit important dynamics - you will need to estimate how strong their influence is to comment on the validity of the results with the frozen flowfield. (my experience: stirred-dominated reactors, freezing is fine. Bubble columns it's not.)

[jayorayo]

Thanks for your reply CeesH, its a stirred reactor so for the moment I'll keep the flow field frozen and see if that provides a representative distribution of the bubbles.

Further to this point, would I be correct in saying the 1st Order discretization scheme for the volume fraction won't predict the bubble distribution in the tank accurately and the QUICK scheme is necessary? I found the QUICK method more unstable than the 1st order method so I had been using the 1st Order method for my initial models.

[CeesH]

Stability of volume fraction is generally terrible, yes.
In the end, what matters is that the numerical diffusion is not going to significantly affect the system dynamics. If your system has a sharp bubble plume while other parts of the domain are free of bubbles, you may need quick or so to capture the steep gradient in volume fraction. If the bubbles are well dispersed, I don't think 1st order is terrible - the gradients are quite mild already, so numerical diffusion can't mix up much more.

[jayorayo]

Thanks CeesH