[sonsonst]

Hi,

I am learning LES, and I see that it is important to estimate the modelled turbulent kinetic energy (k_sgs) and the resolved kinetic energy (k_res)

There is an equation to calculate k_res with the given outputs in fluent using the time average of the velocity fluctuations. One can also calculate with a little more effor using udf k_sgs.

Finally there is another output in fluent that is the turbulent kinetic energy (tke)

My question is that I am not sure if tke=k_res+k_sgs or there are other sources of tke that I am missing.

On the other hand I’ve seen variables in older versions fluent called subgrid eddy viscosity that I haven’t found un fluent 2021r2

[LuckyTran]

Your understanding is correct, you need to find kres + ksgs.

There is almost always going to be a subgrid eddy viscosity whenever you do LES because that is exactly what the subgrid model is. The only time you do not have a subgrid eddy viscosity, is if you are doing LES without a subgrid model. Sometimes you need to fetch the eddy viscosity in order to calculate ksgs using the subgrid model but that is its own problem. I am assuming of course that you somehow manage to correctly get ksgs from your udf.

There shouldn't be an option for tke unless you are doing a hybrid RANS/LES or DES. But if you are doing one of these, then the tke is entirely modeled and there is no resolved tke. If you in fact do see such a thing as a tke when you are doing LES, you should be very very careful indeed.

[sonsonst]

Originally Posted by LuckyTran

Your understanding is correct, you need to find kres + ksgs.

There is almost always going to be a subgrid eddy viscosity whenever you do LES because that is exactly what the subgrid model is. The only time you do not have a subgrid eddy viscosity, is if you are doing LES without a subgrid model. Sometimes you need to fetch the eddy viscosity in order to calculate ksgs using the subgrid model but that is its own problem. I am assuming of course that you somehow manage to correctly get ksgs from your udf.

There shouldn't be an option for tke unless you are doing a hybrid RANS/LES or DES. But if you are doing one of these, then the tke is entirely modeled and there is no resolved tke. If you in fact do see such a thing as a tke when you are doing LES, you should be very very careful indeed.

Thanks you very much for your quick answer LuckyTran.

As it usually happens to me in CFD the answers to my questions only bring more questions…

I am indeed running a Hybrid RANS/LES model (SBES) and , that must be why, as you said, I am having tke as an output as well. However I am a bit confused when you said that if I am using RANS/LES everything is modelled. There is no k_sgs in hybrid models?

I guess in a Hybrid RANS/LES tke=K_modelled+k_res where k_modelled=k_rans+k_sgs. That supports the idea of having an UDF to calculate k_sgs separately. I also think in the regions of the domain far from the walls ideally Tke=K_sgs+k_res