I have a few questions about the correct selection of a turbulence model.

the flow is in the headspace of a tank exposed to ambient conditions. the headspace as a pressure outlet at the top. working fluid is air and since characteristic length is about 2 m, even for temperature difference of 5 K, rayleigh number goes in the transition region around 10e8 to 10e9

Is K omega model the only option for me?

Can I use K epsilon for this case?

Will K epsilon with RNG be a better choice?

If I want to use std wall functions for K epsilon, I have to keep the cell thickness at the wall pretty high, that makes the whole grid too coarse -- so I am clueless there.

If I go for K epsilon with enhanced wall, i know i need to maintain y+ around 1, but i have seen it go as high as 3. In one case, it went up to 10 -- IS that bad? will that give wrong results? Do I need to step back and rerun that time part with refined mesh? \

Please shed light on whatever questions you can.

thank you, CO2

Hi, About the enhanced wall treatment, you may need to follow the instruction given by Fluent, i.e. maintain y+ around 1. you may need to refine.