code_saturne User's Forum

Hello,

I have a LES simulation to run. For the inlet, I plan to use the "mapped inlet" function in code_saturne version 6. I'm testing the LES inlet using an isothermal pipe with a length of 10D.

The mesh is generated by extrusion of a 2D plane. I first used k-omega SST to generate a preliminary velocity field. Then the LES is restarted from the result of SST. However, I found the velocity profiles of LES seemed not correct. The boundary layer is thicker compared to the SST result, and the peak velocity in LES is much higher than that in SST. A plot of the velocity profiles are attached. (Vel-SST is the result of SST, Vel-iwallf5 is the LES result with default iwallf=2, Vel-iwallf0 is LES with iwallf = 0).

I haven't figured out why the LES result is so different from the RANS result, and the LES seems to be quite laminar-like although the Reynolds number is around 8600 using the bulk velocity.

Could anyone help have a look at this case to help me out?

Best regards,
Xiaoxue

Hello,

The xml file for LES and additional files for RANS are attached here.

Xiaoxue

Hello,

The difference in profile could be related to differences in using wall laws or wall resolved values, but I would expect something closer.

How far is the inlet mapped compared to the geometry (i.e. how many cells downstream ?) You want to be far enough from the inlet, but not too close to some outlet's influence.

Also, mapped inlet might not works os well in LES as in RANS, as the interpolation aspects might degrade some conservation properties (our main LES specialist recommends mapping to a conforming sections, and doing so at faces instead of cell centers).

Best regards,

Yvan