[steve20024]

I'm doing VAWT simulation with CFX.
I want to reproduce the same results of a paper doing simulation and experiment to a 3-blade VAWT.
I found that if I empty the center of the rotate domain, I'll get a much better result than set a whole rotate domain.
After doing that, I get a predicted torque higher than experiment data, I think it's because the rotate domain I set is too small.But It didn't show a better result.
Can anyone doing VAWT simulation share some experience making rotate domain size? Or is there any paper mentioned about this? I can't find one...

Attachments are the mesh.(3-blade NACA0022 VAWT)
1. is the first rotate domain I tried, the center is set as rotating too, but the predicted torque is much less than experiment
2. is the second I tried, the center black area is set as station, and it predicted a little higher torque than experiment