Wall function mesh vs y+=1 mesh gives similar results
 

Hi Everyone,

I've been doing some 3D simulations in Star-ccm+ of external flow around an aerofoil style shape. I originally used a boundary layer mesh with a wall y+~1, but was curious to try a wall model style mesh. So I increased the wall y+, and it's now in a range of 10-40 .

The flow is modelled as steady and incompressible, with a Re=2E6. The turbulence model is Realizable K-Epsilon and I use the all y+ two-layer model for the boundary layer in both cases.

So I have two questions regarding these simulations:
1. The two different boundary layer mesh approaches give very similar answers, despite large areas of separation. Should I be worried by this?

2. The residuals converge nicely for a large number of iterations, but then start to oscillate. This happens even though the lift and drag forces have levelled out. I have a feeling that the residual behaviour is because the flow is actually unsteady, especially given the large degree of separation. Does this seem a fair assumption? The drag force does show very slow, low magnitude oscillation around a fixed value.

I'd appreciate your thoughts on either of the above, as I don't have a lot of experience of RANS. Thanks.