To verify that all blades are being used in the power calculation (where Power=Torque*Angular Velocity), look at the region used in the torque calculation to ensure it represents all three blades.

If all blades are being used in the torque calculation, then things to check are:

-Boundary conditions (i.e. inlet wind velocity, angular velocity)

-Model convergence

If the above checks out, I would look at velocity vector plots from 10%, 50%, and 90% blade span in CFD-Post. An easy way to do this is create isosurfaces of radius at the appropriate locations and plot vectors on these surfaces. Do the resulting velocity triangles at blade leading edge/trailing edge agree with your design calculations? Is the leading edge flow incidence as expected? Is there flow separation on the blades that's resulting in the reduced power output?