velocity profile in a simple cylindrical flow tube
 

Hello,
I'm wondering about the velocity profile of a narrow (2cm radius) and long (70cm) flow tube when enabling different turbulence models.
Here I have an example of such simulations with the cross-section profile taken after 50cm flow length. The variables are normalized with respect to the tube diameter (therefore xaxis 0-1) and on the y-axis with respect to the plug-flow velocity. The profile expected at "infinity" is the fully developed hagen poisoille 2(1-r^2) (therefore 0-2 on yaxis).
Now the solid curve is a semi analytical approach for the velocity profile which fits the laminar solution of ANSYS quite well.
However how can - when enabling the turbulence model - vary the profile so much?
The first data points are done with k-omega SST, the second k-omega standard low-reynolds, and the third also k-omega standard low-reynold turned off.
compressibility effects are turned on, vicious heating as well, production kato-launder and limiter are turned on as well (even though the flow should be incompressible I calculated everything as being compressible, though it shouldnt make a difference).
The flow is about 20standard litre per minute at 100hPa and 293K and I expect it to be laminar.
http://fs5.directupload.net/images/161230/j3hsmz9v.png