So guys, I try to do a comparative study using RANS, DES, and LES turbulence models in the geometry image below.

The boundary conditions are inlet, outlet and the rest of edges I set as wall. I dont use symmetry. Where a dotted line I set up as a wall.

First for the RANS approach I just solve a transient flow with SST k-omega model.

For the DES approach, first I solve the initial flow with SST k-omega as steady state with some iterations, after I set the viscous model to DES and calculate the transient flow.

For the LES approach, in the same way as use in DES approach, I solve the initial flow with SST k-omega as steady state and after I set the viscous model to LES. But it do not work. I try to switchon using the code ( (repsetvar 'les-2d? #t) ) and dont work! So, I dont know what's the problem...

I know the LES model is suitable for 3D and can cause non-physical 2D results. Even because, even if the flow in which the mean velocity and pressure vary in only one or two spatial dimensions, turbulent fluctuations always have three-dimensional spatial characteristics.

What's can be the problem? The LES is not available to ANSYS Fluent 17.2 or need some other setup?

Somebody help me? Thanks guys!