I am a student doing a master thesis on model predictive control of windfarms. The turbine I am using in my simulations is the DTU 10MW RWT. I want to create a fatigue load look-up table including Damage Equivalent Loads (DELs) for different windspeeds, yaw angles, turbulence intensities, etc. For this purpose, I am using OpenFAST v2.5.0 and TurbSim v2.00.07 to get timeseries of tower base bending moments and blade root bending moments. From these timeseries, I calculate the DEL (skipping the first 600 seconds of the simulations to exclude start-up transients).
In a one dimensional analysis of DEL as function of windspeed with a turbulence intensity of 3%, there is a huge peak at a windspeed of 6 m/s. This peak is a consequence of the extremely large amplitude of the bending moment fluctuations at this windspeed. I was wondering what could be the cause of the abnormal behaviour at this specific windspeed. I have attached the OpenFAST input files, a summary file of the TurbSim simulation and the plots of the DELs and timeseries.
Thank you in advance!
inputFiles_wsp6.zip (25.2 KB)
I’m not too familiar with this model or its controller, but are you perhaps operating in a resonance condition, e.g., with the rotor speed or its harmonic (e.g., 1P or 3P) overlapping with natural frequency of the tower (e.g., first fore-ft or side-side bending mode)?
Dear Dr. Jason Jonkman,
You are right, the first tower bending mode gives a natural frequency of 0.247 Hz. Whenever the rotor speed gets around 4.94 rpm (frequency of 0.0823 Hz) the third harmonic has a frequency of 0.247 Hz, which is equal to the natural frequency.
I am using the ROSCO controller for the DTU 10MW RWT. Shouldn’t this controller avoid a rotor speed that has harmonics overlapping with the natural frequency of the tower/blades? Maybe it is because of the turbulent inflow that the rotor speed needs to go through the danger zone multiple times? (I also attached the rotor speed timeseries to this reply)
Typically, a wind turbine system is designed so that the first tower-bending natural frequencies lies between 1P and 3P for the operational speed of rotor, so as to avoid resonant excitation of the tower. If this is not possible, a tower resonance avoidance strategy could be implemented, e.g., a notch filter around the tower natural frequency.
Again, I don’t know much about this turbine, but it sounds like the model you are using is not set up to avoid tower resonance, which could be fixed through a tower redesign or a change to the controller.
Dear Dr. Jason Jonkman,
Thank you for your valuable help!
Good Call, Jason. Aerospace numeric sims are intended to reveal unexpected aeroelastic dynamics.
In this case, the rotor disc as a whole may be partly stalled, promoting bluff-body oscillation as the primary resonance with the tower-sway harmonic. A clue is if the tower harmonic is not an exact multiple of blade phase-velocity, then sway is most likely bulk bluff-body rotor-disk interaction. A possible higher-order factor is vertical wind shear not damped by cyclic pitch.
As HAWT sizes run larger, and are no longer simply overbuilt, “flying” them becomes complex and critical. Modulating blade AoA and/or generator torque-loading can generally effect “notch filter” solution. Generative electronic speed control will increasingly be needed. Blades will have strongly damped aeroelastic structure, and may even be provided with active or tuned passive-elastic flaps, to shorten phase response.