Inaccurate turbulent viscosity profile in case of flat plate
 

Hi all,

I'm simulating the well known flat plate turbulent boundary layer case. I have measurements (Hot Wire) for the velocity and RMS profile, with a vertical spacing of 0.1mm. I'm using the measurement points as 1D interpolated inlet profile, and I'm comparing the measurements at a downstream position with the CFD results.

The problem is, that in my case, the difference in turb. viscosity shows a high difference in the measured and simulated values. I'm using a fine mesh, yplus<1, Spalart-Almaras, k-e models, 2nd order central scheme, CFL=3, steady state settings. Inlet is velocity profile as mentioned above, outlet is average static pressure.

The domain is a channel, lower wall is NS type, and referred as the plate surface, upper wall is eulerian, because we have there laminar layer but we didn't measure there. On the lower wall, we used a rough paper to turbulize the BL, so the transition and the leading edge of the plate isn't simulated.

In my measurements, the nu_t increases in downstream direction, but in CFD, it seems to be like there is some kind of numerical damping.

Any suggestion is very welcome.

Best Regards,
Attila

Attachment 11116