Does standard k-epsilon predicts poor turbulent shear stress?
 

I am testing standard k-epsilon model for a 3D test case but its turbulent shear stress is e.g. -uv*1000=~600 while experimental data shows -uv*1000=~3. This applies to uu and vv too. Is this normal for standard k-epsilon model?

Standard k-epsilon model is however able to provide more accurate prediction of turbulent shear stress for 2D case

The standard k-epsilon model has several flaws, but not all of them are corrected by the other variants.

I don't have a lot of experience in using it (so I can't say if this is your case) but, one of them is indeed the so called "round-jet/plane-jet anomaly", in which turbulent 2d planar jets are, mostly, well predicted but not the 3d round ones. The realizable version, among others, corrects this.

Also, without proper corrections, most k-something models tend, in some circumstances, to overproduce k that, in turn, will give larger eddy viscosities.