side-to-side tower top force

Hello everybody,

For my master thesis I’ve developed a Matlab model of an onshore wind turbine, now I’m doing comparison between my model and FAST V8.16. I found a error in side-to-side tower top force (YawBrFyp) see figure.
I’ve made a comparison using Test 18 with the following changes:
-all DOF set to “FALSE”, except drivetrain and generator DOF;
-0 deg shaft tilt;
-0 deg yaw error;
-0 wind power law exponent;
-unsteady wind with: U-mean component of 6.16 m/s, V-mean component of 0 m/s, W-mean component of 0 m/s.
In my model I have considered only the drag force to the nacelle:
Fyd= 0.5CdDLV^2*rho
-Cd = 1 (drag coefficient);
-D = 5m (nacelle diameter);
-L = 10 m (nacelle length);
-V (V-component of the wind)
-rho = 1.2 kg/m^3 (air density).
I used BEM theory to model blade loads. Since I have three blades, I supposed that the Fy contribtion of each blade go to elide each other.
Are there phisical phenomena that I overlooked that justify the error between my model and FAST?

Thank you,

Dear Riccardo,

It may help to plot the statistical distribution (histogram) of these time series, as well as the power-spectral density (PSD). From what I can tell, both FAST and your model have zero mean, but the standard deviation and tails of the distribution are larger from FAST. It is hard for me see from the time series how different the frequency content is between the models, but I would expect from FAST to see strong peaks at 3P, 6P, 9P etc., where “P” is per revolution; hopefully your model also captures these peaks?

I would guess the differences between FAST and your model are related to differences in how the aerodynamic loads and/or spatial variation of turbulence across the rotor are treated, but this is just a guess knowing nothing about your model.

Best regards,

Dear Jason,

Thanks for your prompt reply. As you suggested I plotted the power spectral density shown in figure.
As you can see from the figure, Fast presents the peaks you were talking about, while my model doesn’t present peaks.
I guess that the reason of these differences is that FAST take into account some forces in Y direction derived from the 3-bladed rotor revolution, while in my model I supposed nothing about this force for the reasons I wrote to you earlier.
If so, what effect I have neglected with my supposition?

Thank you for your help, best regards.


Dear Riccardo,

I don’t know any details about your model, so, I can’t really comment on what is missing.

But the 3P, 6P, 9P, etc. peaks from OpenFAST are the result of the aerodynamic loads being calculated locally to each node along each blade as the blade passes through spatially varying turbulence, i.e., the so-called rotational sampling of turbulence. Does your model account for that?

Best regards,

Dear Jonkman,

Thank you for your reply. In my model I don’t take into account any side-to-side Force when I calculate the load at each node of each blade, beacause I neglected that the wind varies spatially due the turbolence, which entails an imbalance of the loads between the blades. In fact I take into account anly the force in X direction and the tangential loads of each blade to calculate the Rotor torque. Now I understand the reason of the difference between FAST and my model.

Thank you for your help,
Best regards.