[kjetil]

Hello,
I am trying to simulate some kind of separator where boundary inlet consists of mostly air, but also dispersed water droplets. For the moment I keep a homogenous temperature fixed at about 70C, though the original idea was to also look into the cooling and phase change. The water VF at inlet is ~0.1%.

My question: the idea is obviously to separate the to fluids, but is this at all possible when using the Eulerian approach? I am fully able to activate both Free surface flow and Wall adhesion, adding a contact angle. But so far I'm having a hard time getting the simulation to work, even when initializing from a partly converged single phase. Could the two pressure outlets be causing problems also? They are set to same relative pressure.

I have been looking at the CFX free flow tutorial, where they seem to focus a bit on the initial part (ramping up pressures and so on). But in my case there will really never be a significant hydrostatic pressure, as the water should exit through the bottom drain.

[ghorrocks]

At such a low VF you could model this as either Langrangian or Eularian particles. But you appear to be modelling this as a free surface simulation - that does not appear feasible. I suspect you will need eularian particles.

It is common to have convergence problems with simulations like this. This FAQ has some tips: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

But in this case it is quite likely there is no steady state answer and you will need to do a transient run.

Finally: The normal device to remove particles from an air stream is a cyclonic separator. You will find it probably is a more effective separator, if you are looking at separator design.

[kjetil]

Thanks Glenn,
seems my main question slipped out, which was about combining Eulerian and Free surface flow. Do you say that it is not possible? Or can Free surface be combined with Lagrangian? I was hoping CFX would model droplets hitting the walls, forming streams and finding its way out.

I fully agree with your separator design suggestion. This simulation is sort of to determine why it's not working, and this device's main purpose I'm told is a silencer. But apparently it does not fullfil either of its tasks :P

[ghorrocks]

CFX has a model for eularian particles hitting walls and forming wall films. Have a look in the documentation for details on this model. But I do not think it can be easily coupled to providing fluid to a free surface model.