High curvature 90 degree bend using SST, eddy viscosity drop??
 

Hi

I am simulating a radial centrifugal compressor inlet with a decreasing in area (0.5 area reduction), 90 degree bend with SST. Exit results show high entropy generation not just in the boundary layer but parabolic-like entropy generation from 20% span to the shroud. It seems that the shear strain rate is high and computing the reynolds stresses, those are at a peak around 80% span. The eddy viscosity is highest in the boundary layer, decreases below laminar values just outside the boundary layer and then increases again near the 80% span region.

Also, if I change my inlet turbulence properties (reduce the length scale by 90%, change the intensity from 5 to 10%) I get a change in eddy viscosity and NO CHANGE in entropy, even though the velocities at all points in the channel look the same !!


My question is... (1) in high adverse pressure gradient flows, are these problems in the eddy viscosity expected?? Most papers only discuss the boundary layer region. What happens to k-epsilon (the free stream equations of SST) in the FREE STREAM when subjected to pressure gradient??

Note: I am not using curvature correction.

Also: I have read that the k-e model is not sensitive to inlet turbulence conditions....but if the eddy viscosity is changing.. why doesnt the entropy generated change?