To whom it may concern,

 

I've been experiencing a consistent, unusual phenomenon in the simulations I've run using the VOF-to-DPM solver. I would like to know if it is normal and possibly how to deal with it. What I've noticed is the following:

When I use the VOF-to-DPM solver for gas liquid flows (eg. liquid jet in crossflow, air-blast atomizer) initially everything runs fine until the transition mechanism kicks in. In other words, when liquid ligaments start being converted to the discrete phase. Velocity hotspots will show up in small clusters of cells at seemingly random locations. A closer inspection seems to indicate that these locations are where ligaments were converted into the discrete phase. I've deduced this based on the presence of refined cells and DPM parcels at these locations. For example, in simulating an air-blast atomizer with a gas flow of 48 m/s and a liquid flow of 2m/s - spots with velocity magnitudes of 150 m/s will appear in very small volumes at select locations in the domain (in the primary atomization zone). 

Is it possible that this is due to the transition mechanism? I imagine that when the liquid part of a cell is filled with gas (to conserve volume), the mass seen by the continuum phase in that cell suddenly drops. In order for the system to conserve momentum, the velocity then shoots up. With the right settings, I've managed to keep the simulation stable. The problem is that these high velocity spots limit my time-step size prohibitively (due to the CFL condition). 

Any feedback is appreciated. Thanks for your time.