[liu_csu]

Dear all,
I am working on the multiphase flow simulation using the two-fluid model.

My physical model is a rectangular bubble column with 100 mm height, 100 mm width and only 12 mm depth (see attachment, please note the container is very thin, which is comparable to the bubble diameter in the container). At the center on the bottom plane there is a single needle to produce the single bubble chains. The measured equivalent bubble diameters vary from 6 mm to 10 mm with different gas flow rates. In the numerical model, gas is considered to be mono-dispersed with constant diameter which is derived from the experiment, liquid is considered to be continuous. The drag, lift, virtual mass, wall lubrication and the turbulent dispersion force were included in my simulation. The inlet condition is set to be mass flow inlet and the outlet is degassing. The wall is considered to be no slip for the liquid and free slip for the gas. The SST k-w turbulence model is adopted for the liquid phase and a transient simulation is selected.

Now the problem is, when I use the course mesh (4mm*4mm*4mm, just 3 cells along the depth direction, which can be considered to be two dimensional simulation in my opinion), the results seem to be good and the meandering phenomenon of the bubble plume can be well captured comparing to the experimental data. However, when the mesh is refined (After I tested, I found the results were very sensitive to the mesh size along the depth direction, even if 5 cells are used along the depth direction), the gas void fraction will be concentrated in the center along the depth direction (cannot be dispersed along the depth direction, the more fine mesh the more concentrated gas void fraction along the depth direction) and the bubble plume will stand still and the plume oscillation will be damped after some time evolution. I guess there should be something wrong due to the very thin container and the interracial forces which control the gas void fraction along the depth direction cannot be well predicted by the model. I was struggling on this topic for a long time and trying a lot of ways to solve the problem but all did not work and now I have no idea how to figure out the problem.

Any help would be greatly appreciated. Thanks all
Liu

[sfn]

first of all check your mesh quality
also you may try other turbulence models with more enhanced near-wall treatment
that's all what came to my mind