[AndyCF]

Hi!
I am modelling flow in a circular tank (diameter 14 m, depth 4 m) using Fluent. Rotating flow is achieved by supplying water through orifices (each 10 cm in diameter) tangentially to the tank wall. In total, 10 jets, spaced vertically 30 cm from each other, each with inlet velocity ~2 m/s. The outlet is in the bottom of the tank, in the middle.

Have used standard k-eps, sst k-omega and RSM. All gave reasonable velocity profiles across the tank (~0.2-0.3 m/s in average), comparable to experimental data found in the literature (the authors used ADCP measurements). However, the predicted turbulence parameters (from profiles across the tank far from the inlets) are too low: TKE ~1e-5 (CFD) vs 1e-3 (experimental), epsilon ~1e-6 vs 1e-4, I~2% vs 10%. RSM gives the lowest TKE (~1e-6), but the best velocity match.
Changing boundary turbulence (e.g., from I=5% to I=20%) did not affect much the overall turbulence after some distance from the inlets (looks like most of the turbulence in the tank is produced in layers). The mesh was refined several times, and the results were consistent.
Would like to ask:
1) If my predicted turbulence levels (TKE ~1e-5, epsilon ~1e-6, I~2%) are reasonable compared to yours from similar simulations (mixing tanks, rotating flow etc.)? If not, what can be wrong (taking into account that the velocity profiles look meaningful)?
2) Any ideas what could cause so high turbulence in the real experiment? Could it be any free-surface effects (in the CFD the free surface was modelled as a mirror plane)?
3) Finally, for those who have experience with ADCP, are the experimental results reasonable?
Would appreciate answers to any of these questions.