residual convergence with turbulence modeling

eddybreakup

I was wondering exactly why the residuals of turbulence do not convergence a few orders of magnitude when I am using k-omega or some other eddy viscosity model. The solution is pretty converged, physical quantities are correct and they are not changing. This solution is incompressible, but I've noticed it also in compressible flows. I'm running incompressible flows, 3d, perfect gas.

Thanks!