Hi Dear Friends,

Is the K-Epsilon model a subgrid model?

Thanks.

Hi,

I think the turbulent viscosity computed using the k-eps model cannot be considered as a subgrid viscosity because it doesn't vanish even if all the turbulent scales are resolved .

best regards

The standard k-eps model was developed for time-averaged equations. The term sub-grid models is normally used in the context of space-filtered simulations or Large eddy simulations.

Now, people have used sub-grid k-eps models too. Where they solve additional equations for the sub-grid 'k' and sub-grid 'eps'. This is useful for some reasons like 'control of back-scatter of energy' maybe.

chidu...

Hi,

there are subgrid scale models proper for 2D simulations (2D-LES)??

The 2D Les concept has been used as a cheap and dirty alternative to doing a full 3D simulation. I don't know what it means in a flow situation that actually involves three-dimensional turbulence. I think in such cases (mostly in the industry) people just use the same sub-grid scale models as they have for their 3D simulation. The best way to look at literature.

Of course, in principle, the LES concept can be applied to a problem in any number of dimensions.

chidu...

subgrid model concepts has been developed in LES that is essentially spatial filtering of turbulent flow-field. LES does not removed the 3d fluctuation of turbulent. Thus performing LES (using 2D subgrid model) would be pointless as the simulation is definitely innaccurate. (See study by Michael Brauer, e.g. his paper in ECCOMAS 2000)

Currently there is some attempt to make LES cheap by combining it with RANS concept, e.g. Detached Eddy Simulation of Spallart. Within this framework, "2D subgrid scale" has been developed. More information on this can be found in onera and cerfacs website.

Cheers

Backscatter is possible in the context of eddy viscosity models (like the K-e model) only is the eddy viscosity coefficient C_mu is negative (since K and e are by definition positive). Therefore, with fixed positive C_mu, there can not be any backscatter (in 2 or 3 dimensions).

Yeah, but is there a contradiction here? If you fix C_mu to be positive then of course there will be no back-scatter. I never said that one has to fix C_mu to be positive. In the case that you allow it go negative, using a sub-grid k-eps model allows some flexibility!

In many cases anyways the fact that C_mu goes negative is a wierdness of the model rather than physical back-scatter. Personally I don't believe LES models capture back-scatter well. I was just making a point.

Rereading your email, are you talking about RANS models? I was talking about LES!!

chidu...

I was referring to LES models as well.

And of course, I agree with your opinion that physical backscatter, despite what many LES researchers (some well regarded) say, has nothing to do with C_mu going negative. I can state the reasons behind my belief if some one wants to refute them. Perhaps you could do the same if any one shows interest.

more references please