Unexpected species increase in discontinuous phase
 

Hi all,

I'm running the following simulation:

- stirred tank (MRF), k-e turbulence (dispersed), gravity enabled
- eulerian multiphase w/ constant bubble diameter, only drag, no mass-transfer.
- Boundary conditions: velocity inlet, constant velocity (gas only) and degassing outlet.

I first calculated the steady-state gas volume fraction in the reactor. This worked fine.

I now switched on species transport and use a 2-species mixture in both the gas and the liquid. The goal is eventually to model oxygen transport between the phases, but currently the mass transfer mechanism is disabled to see if the simulation with species would work. Here is what I did:

- Created gas species 'o2' and 'air'. Both species have the exact same properties; same viscosity, density and molar mass.
- Patched the phase_2 species fraction for oxygen to 0.23kg/kg in the entire domain
- switched boundary condition to 0.23kg/kg oxygen, kept all else the same
- Mixture density is calculated by vol weighed mixing law and should be the same as for pure components
- set simulation to transient with a small timestep (MRF still)

I expect that the volume fraction of phase 2 would remain the same, and the concentration of oxygen in phase 2 uniform over the domain. However, I see:
a decrease of phase 2 fraction, and a steady increase of oxygen concentration (after 12 impeller rotations the content is 0.3kg/kg on average with local ratios close to 1kg/kg)

When making a contour plot I get something like below; the concentration increase seems to start at the top and from there move downward. After a longer period of time, basically the oxygen concentration at the top degassing outlet is 1kg/kg. I am completely lost on why this happens, is there a backflow setting for the degassing boundary or so? If so, I cannot find it.

Does anyone have an idea on this?

http://i.imgur.com/W7nSnt7.png