[Starcatcher]

Hi folks,

the models inside of cells are called subscale models. That means that the length scale of an les cell is called subscale. What is the analogy to that for the much bigger RANS cells?
It's not "fine scale" beacuse that stands for the very smallest turbulent structures in the real flow.

Thanks
Andrew

[LuckyTran]

You can think of it in terms of filtering or a low-pass filter. But a subtle difference in LES and RANS is what exactly is being modeled.

LES is spatially filtered and you solve for the low spatial frequency components and end up needing to model the influence of high spatial frequency components (the subgrid model).

RANS/URANS can be interpreted as a temporal filtering. You solve for the low temporal frequency: the time-average in the case of steady RANS, and the slowly-varying components in the case of URANS. The influence of the high temporal frequency components, is your turbulence model where we most commonly use a Boussinesq eddy viscosity approach.


That is, subscale in the LES sense refers to sub-spatial-scales and in the RANS sense temporal scales. In LES, the filtering is usually linked to the the grid and hence we call these the the subgrid scales. In RANS, the temporal filtering doesn't really have anything to do with the grid. That's why we don't give it any fancy name.

[Starcatcher]

cool thanks. What is with URANS? The time filtering there is qualitatively not different from the one in e.g. VLES. Can I then also use the term "subscale"?

[FMDenaro]

Originally Posted by Starcatcher

Hi folks,

the models inside of cells are called subscale models. That means that the length scale of an les cell is called subscale. What is the analogy to that for the much bigger RANS cells?
It's not "fine scale" beacuse that stands for the very smallest turbulent structures in the real flow.

Thanks
Andrew

There is no analogy between the concept of "subgrid scale model" used in LES and the turbulence model used in RANS. In LES the spatial filtering mimics a formal separation in the energy spectrum into a resolved and unresolved part. Only this latter is the object of the SGS model that acts on the smallest resolved scales. In RANS there is not such a separation, the entire spectrum is under statistical averaging and all the components are affected by the action of the turbulence model.

[LuckyTran]

Originally Posted by Starcatcher

Can I then also use the term "subscale"?

I wouldn't recommend it, unless you want to upset your peers. There's better names already in use (e.g. turbulence model).


I posed an explanation for the analogy purely for educational & philosophical purposes.