Learning Forum

[JamesWright]

I recently did a simulation of a conical diffuser back-to-back using a SBES model. The first was without the velocity fluctuations at the inlet, the second was with "Spectral Synthesizer" used as the velocity fluctuation algorithm.

Relevant contour plots for the simulation are posted here. Contours are of a plane going down the middle. Flow is going right to left. Right most vertical surface is the inlet, everything else is a wall BC except for the top, bottom, and left-most walls of the plenum, which are outlets. The RMSE plots are obviously taken over a range, the SBES shielding is taken at a single time step (it's not included by default

As you can see, there are some very significant problems that plagued the SBES shielding function. Without the velocity fluctuations, the shielding function stuck close to the wall, just like it should have. With "Spectral Synthesis" though, it's now takes up the entire entrance tube and diffuser, and even has a "jet" that extends into the middle of the plenum.

Any idea what's causing this to happen?

[paguado]

Hi,

It seems that due to the additional turbulence introduced by the spectral synthesizer the resulting turbulent length scales are too small to be resolved in LES mode and therefore the shielding function is triggering the RANS mode across the pipe. I would recommend to decrease the turbulence level at the inlet to see if this behavior disappears. Also, I would extend the inlet further upstream to minimize the effect of the inlet BC on the region of interest. On the other hand, it seems that the shear layers generated with the Spectral Synthesizer enable are more physical (as long as the inlet velocity is high enough). Did you ensure that the sampling was performed once the solution was statistically steady?