[buidu]

Dear all,
I want to test the feasibility of different turbulence models on the simulation of convective heat loss from sphere in fluent. The free stream were set into different velocities with Re ranging from 3e3 to 3e5.
When I used kist the Y+ ranged from0.2 to 0.6, I do not know if it is right to do so. Because I read some threads on the forum that to use kwsst y+should be ~1. Other turbulence models also have their own Y+ requirements.

Then 1)should I make different meshes based on specific Re and turbulence model? 2)if I adopt enhanced wall treatment(ewt), does it mean I can use ewt for all turbulence model just follow the requirement of ewt.3) could you please give me general requirements for y+ when using different turbulence models. Thank you very much!

[LuckyTran]

Although Y+ ~ 1 is recommend, as long as you can afford it, there is nothing wrong with using Y+ < 1 (it's even better).

Especially if you are trying to simulate a heat transfer problem, it is very important to have a low Y+ because at high Y+, the boundary condition switches to an empirical correlation instead of law of the wall.

Generally I produce a mesh that satisfies the Y+ ~ 1 at the highest Reynolds number, knowing that at lower Reynolds number the Y+ will be even smaller so that I do not vary both the flow (Reynolds number) and grid at the same time.

The omega equation is meant to be integrated all the way to the wall, so the Y+ requirement for omega based models is generally Y+ ~ 1. Wall function approaches, can handle Y+ >> 1 and so they do not have this "requirement" however they can also handle low Y+ grids. Well, even for omega based models wall functions can still be used, so....

[buidu]

Thank you for your answer. Then if it means that I can use enhanced wall function/treatment in any turbulence models unless the Y plus is less than 1?

Besides, do you know the low reynolds modelling? What is the difference between LRM and enhanced wall treatment? I think both of them require a lower Y plus. Thank you very muh!

[LuckyTran]

Normally in two-layer models, the y+ must be either everywhere low (~1) or high (>30) and excessive error is incurred if any parts of the grid are in the intermediate region. The enhanced wall treatment blends the epsilon and turbulent viscosity and allows you to use a grid that may contain low, med, and high y+ values.

In principle, with EWT and a grid that is y+ ~ 1 everywhere, the enhanced wall treatment approach should be identical to the low Reynolds number approach, but there are some subtle differences. One of the key differences is the turbulent viscosity in the near wall region is calculated from the Wolfstein model when enhanced wall treatment is used. However, the purpose of EWT is to permit you to use grids where y+ may not be 1 everywhere.

[buidu]

now I understand. Thank you very much!