Streamlines at the boundary of rotating and stationary domains are malformed

Question

I am simulating a turbine and have divided the mesh into domains: the piping before and after the blade and the turbine blades. The domain containing the blades rotates at 2000 rpm while the other domains are stationary.

The mesh is structure and the mesh (between the domains) is conformal.

The turbine I am simulating is an axial-flow Wells turbine. The solution is steady state. This turbine has 8 blades (360 / 8 = 45° angle between blades). Due to the circumferential symmetry, only one-eighth of the annulus has been computed, imposing periodic boundary conditions in the tangential direction. In my model, the computational domain has been restricted in the axial direction to four chord lengths upstream and eight chord lengths downstream of the blade. Blades are symmetry (Naca00XX).

The computational domain is as follows:

According the following figures,

I want to obtain streamlines as in the following image from Insight analysis of biplane Wells turbine performance (Fig. 7):

Shaaban, S., & Hafiz, A. A. (2012). Effect of duct geometry on Wells turbine performance. Energy Conversion and Management, 61, 51-58.

Answer 1

Your stationary domain can't incorporate the time/spatial dependency that the turbine blades should induce in them.

Suppose there's a region of high velocity flow that trails a turbine blade by 10 degrees. When that high velocity flow crosses into the stationary domain, it should cross at a location that is 10 degrees trailing the blade, but that location is always moving in a stationary reference frame.

The only time you can have a stationary and moving domain share a boundary is if you expect all of the free variables to be uniform in the direction of relative movement.

I would recommend making your entire domain rotating and applying a fixed velocity to the walls corresponding to zero velocity in the non-rotating frame.

Far from the blades the flow should become uniform across the circumferential direction, at that point if you wanted to have a boundary to a stationary domain you could. However, you would still get kinks in your streamlines at the border, as if the air is not rotating in the stationary reference frame the streamlines would be aligned with the turbine axis whereas in the rotating reference frame they would form a helix, so there would be a sharp transition between the two.