Could you please clarify if the simulation is transient or steady state? Your ddt scheme is steady state, but then you are limiting the step size to 0.0005? Or are you perhaps using some pseudo-transient approach? Your omega (I'm assuming that refers to the turbulence specific dissipation rate) is also huge, on the scale of several 10^6 Hz, are you sure that is to be expected? As for k, and the shear stress, the difference in the plots you provided is quite small, relatively speaking. Your turbulence kinetic energy is tiny, whether its 5e-06 or 8e-06 should generally not cause a massive difference in the final result. Have you also checked the mesh for any major problems? If not, try checkMesh -allTopology -allGeometry (or the equivalent command in your openFOAM version). I've never encountered such a problem before, but my gut feeling is that there is something wrong with the underlying computation (possibly schemes, some schemes like Gauss linear are known to cause oscillations in regions with steep gradients) or the mesh itself.