Different solutions for Rotating Reference Frame
 

Hi,
I am sort of a beginner in Fluent and ICEM, but I am using those software to model flows in a reactor.
The main model is a 3D one, but due to strange results I am now using only a 2D model to understand where the problems are coming from (basically a cylinder with a shaft).
The problem I have comes from the velocity observed near the outer wall of the cylinder.
I observe 2 cases:
1) rotating reference frame with velocity of 100 rpm, rotating shaft with 0 relative velocity and rotating walls with 0 absolute velocity
2) stationary frame and walls with rotating shaft at 100 rpm.

In the case 2, I have a good behaviour: the velocity decreases until 0 at the wall. However, in the case 1, it decreases then increases near the walls and is 0 at the wall.
But this increase does not make sense and the results should be the same in both case.

Am I missing something with the rotating reference frame, because in this case it is ok to use a stationary frame but for the 3D model with impeller, I will very likely need to use the rotating one.

I attach the velocities in both cases so that you may understand better my problem.

Thank you for your help.

Does anyone have already seen the same problem?
It does not make sense to me, but I really do not understand why I have this difference in the solution.

Thank you very much!