I am using FAST in a code comparison in which one of the initial load cases, intended to compare aerodynamics, uses a steady wind, free rotation and a rigid structure. The turbine has a 5-deg axis tilt, but gravity is set to zero, so I expect to see variation in blade root forces only from changing angle of attack at 1p. But the results from my FAST model exhibit large oscillations in Fx & My, much greater than anticipated. I estimated fluctuations to be 1% - 2% for Fx, but I am seeing more like 20%. I should also say that the rotor speed gets quite high (42 rpm for 50m rotor radius), but is very close to what the other codes calculate. Also, mean values of blade root loads are close to the results from the other codes.
My Aerodyn settings are:
StallMod = STEADY
UseCm = USE_CM
InflModel = EQUIL
IndModel = WAKE
AToler = .005
TwrShad = 0.0
DTAero = .01
In the FAST input I set pitch and VS control to zero and TimGenOn to a time equal to my simulation time, and only GenDOF and CompAero are set to TRUE.
Does anyone have suggestions?
I’m assuming your steady wind file is free of vertical or horizontal wind shear?
I don’t have a gut feeling for how much fluctuation in load would be seen by a 5-deg shaft tilt, but in addition to the direct (geometric) angle-of-attack variation, there is also a skewed-wake effect caused by the tilt that will induce a sinusoidal variation in the axial induction.
What code(s) are you comparing FAST too?
Thanks for the response. That’s right, no wind shear. The other codes are Bladed and our software, OneWind. These results show a 1p Fx pp-amplitude around 1%. Also, I built a spreadsheet that calculates blade element forces for a blade moving through 3:00 and at 9:00, for a rough estimate of the force fluctuation, which gave about 2%.
I tried to attach an image file with comparison plots.
Fx_root.ps (2.44 MB)
I’m assuming in your plots that the FAST solution is the blue-ish color, while Bladed and OneWind are orange and green. The FAST solution seems to have some sort or numerical problem. Have you confirmed that this is a converged solution, meaning that reducing the time steps (in both FAST and AeroDyn) results in the same solution? A smaller time step is required when the rotor spins faster due to increased centrifugal stiffness, etc.