Cross-Sectional Streamlines for an In-compressible Flow Inside Pipe
 

Hello Everyone,
I have simulated an incompressible flow inside a pipe using a steady-state RANS model. Inlet boundary conditions are inlet velocity and zero gradient pressure, and at the outlet, I have zero-gradient boundary conditions.

Cross-sectional streamlines exactly at the inlet are outward from the centre toward the wall, and slightly after inlet, they are changing direction, and then became inward from the wall toward the centre. This phenomenon appears multiple times along the length of the pipe. (Picture is attached)

Does anyone know why this happens?

How do you compute these streamlines? In a plane-section of a 3D flow you have disregarded the third velocity components?

Yes. In a 3D flow, I have three velocity components (Ux, Uy, Uz). Streamlines show the planar velocity components, which is (Uy, Uz), in this case.
(Ux, the streamwise velocity is disregarded)

Ok, what you see are not exactly the streamlines as would be in a real 2d case.
However, if you prescribed as inlet an inviscid velocity profile the flow in the section next to the inflow one should show a small radial velocity from the wall towards the center. But if you see that the radial components changes sign that appears as a nunerical oscillation.
Check the computed variables not the streamiline to assess the solution is free from spurious oscillations.