I am trying to set up a (non-mixing) multiphase problem where AIR (or and low viscosity fluid) runs over WATER (or any higher viscosity fluid). The domain is any long channel (long in the x dir, dir of flow). What I'm trying to examine is the movement of water due soley to the shear developed at the interface between the water and the air stream above. (Imagine ripples in a pond due to air current.) I've divided the height of the channel into two blocks, set the initial distribution of the two phases using USRINT routine, set an inlet patch on the air (at low x end), a press patch on the water (at the low x end) and one pressure patch at the high x end. When I run this I inevitably get significant mixing of the two phases.

Any thoughts or help would be greatly appreciated. Also, since I imagine this as the inverse of the free-surface flow problem, if anyone with that kind of experience could contact me I'd appreciate it.

Thanks, Brett Towler / CFX4

Hi Brett

You have choosen the two fluid model, which basicly operates with the assumption that the two fluids each occupies a volume fraction of a box. A continuity and momentum equation is formulated for each phase. If you want to describe the effect of wind on the surface, you need to transfer this into an inter-phase friction which works through the momentum equation. I'm not an expert on wind/surface interactions, but from my previous knowledge, this might be difficult (generation of waves,boundary layer dynamics, varying roughnes, pressure gradients and other microscale effects). However, an empirical expression coupling shear stress and wind velocity are known from oceanographic sciences.

There is a possibility in CFX to choose the homogenous model, where only one set of continuity and momentum equations are solved. This model has a surface sharpening algorithm, which ensures a sharp surface. I'm not sure that the choosen algorithm is the best possible - considering your specific problem, but it will generate a clear interphase without dispersion of one phase into the other.

The last possibility I can think of is the Volume Of Fluid model (VOF), which approximates the free surface quite well. Still, you need to formulate an good boundary condition for the fluid in order to simulate your problem.To my knowledge, VOF is not an option in CFX (yet ??).

Regards

Michael

Thanks, Michael. I just figure out why my model was not running correctly.

As a piece of advice to all others using CFX4 to model free-surface flows -- make sure your USRBCS is doing what you think it is! In my case I was not correctly specifying hydrostatic pressure at a PRESS boundary.

Brett