[g7sznAc4]

Hello,
I'd like to run a 2D simulation of a vertical-axis wind turbine in Fluent in order to examine the forces on the blades as it spins. I'm fairly experienced with CFD in general, but this is my first time using Fluent or running incompressible CFD.
I've spent a lot of time so far trying to get a solution with little luck - forces are significantly off what is predicted by Xfoil and there are clearly unphysical features in the turbine wake. I'd really appreciate any insight that anyone could offer. I will try to provide as much detail as possible so forgive me for a lengthy post.

Geometry:
Very low solidity vertical axis wind turbine with four blades.
Airfoil chord = 7.5mm, turbine diameter = 0.5m
Turbine rotates at 36.4 rad/s

Grid:
Generated in Pointwise. 310,000 nodes, quad-dominant mesh (Figure 1). ~310 points on airfoil surface (Figure 2)
Yplus is generally below 1, but as high as 3.
Grid is refined in turbine wake.
I've also tried other grids with as little as 100,000 nodes and no wake refinement, made up only of triangles.
Sliding mesh interface between static freestream and rotating turbine core grid. Mesh motion appears to be working correctly.

Boundary conditions:
Fluid: carbon dioxide. Density = "incompressible-ideal-gas"
Operating pressure: 9,210,000 Pa.
Velocity-inlet: Vx = 1.4 m/s, Turbulent Intensity = 1%, Turbulent viscosity ratio = 1.
Pressure-outlet: Gauge pressure = 0.
Symmetry boundary conditions on the sides.
I've also tried the "outflow" BC at the exit, or the "pressure-far-field" BC on the sides, which did not fix the issues I've been having.
SST Turbulence model

Solver settings:
Pressure-based.
Scheme: SIMPLE. PISO seems to be recommended for transient flows but quickly diverges.
Discretizations: Second-order for all.
Time step: corresponds with a 1deg rotation of the turbine.
Subiterations: I've been starting with 10-20. Last night I ran with 150 subiterations and I noticed the residuals drop quite a bit (Figure 3) - but this seems like an excessive number of subiterations.
Time steps: I've run for as many as 20 turbine revolutions and the flowfield still doesn't look reasonable.

Results:
Velocity magnitude plots show large features in the turbine wake (Figure 4). Velocity or pressure plots zoomed in on the airfoils appear to show reasonable blade loading. However, the downstream half of the revolution passes through the upstream wake which has a huge impact on loading - not sure if this wake is physical (Figure 5).

Turbulent viscosity ratio is very high in the wake downstream of the turbine, all the way through the exit. I'd show a picture but I've reached the maximum number of attachments.

I've played around a lot with different grids, numerical settings, and boundary conditions. I have gotten a "converged" solution once with periodic forces, but they were nowhere close to Xfoil's predictions.


Thanks in advance for any help.