I am in the process of trying to perform a lifetime fatigue study using MLife on a very small blade (1.5 m) with a rotation rate of about 400 rpm. In order to do so, I have been trying to adapt discon.dll to change the pitch angle of the small blade to accommodate wind speeds from 2-30 mps. The program has been modified by multiplying the torque and power-related parameters (VS_MaxRat, VS_MaxTq, VS_Rgn2K, and VS_RtPwr) by 3 kW / 5000 kW. In order to accommodate the increased rotor speed, I changed the gearbox ratio to 3.0 to provide the same generator speed as the 5 MW wind turbine. Finally, I found the best results by changing PC_DT and VS_DT to 0.001 sec. and PC_MaxRat to 0.25 rad/sec.
The results were mixed. The controller worked very well at high wind speeds when the pitch was constantly changing. Here the pitch angle graph looked very similar to the wind speed graph. The problem occurred when the wind speed was both below and above the rated wind speed (i.e, ~12mps). When the wind speed increased from below the rated wind speed (pitch angle = 0) to above the rated wind speed, the pitch angle spiked from 0 to perhaps 6 degrees in a very short period of time. In most cases the pitch angle then went back to zero very fast. The result was a significant number of spikes in pitch angle. If however, the wind speed remained above the rated wind speed for a period of time, the pitch angle graph appeared to reflect the changes in wind speed.
The problem with the spikes occurred when using MLife. These spikes in pitch angle caused big changes in the in-plane and out-of-plane bending moments at the blade root. As a result, the predicted lifetime fatigue of the blade dramatically increased.
I tried a significant number of changes to the parameters (including reducing PC_MaxRate), however the values of PC_DT, VS_DT and PC_MaxRat provided above worked the best. I was wondering if anyone else tried to modify discon_dll to accommodate a different blade and had issues with spikes in pitch angle. Also, I was wondering if a rapid change in pitch angle increases the turbulence and negatively affects the generator torque. This might explain the rapid decrease in pitch angle (back to zero) causing a pitch angle spike.
I gather that you are modifying the parameters in the baseline controller DLL for the NREL 5-MW wind turbine. Due to the large size difference between your 3-kW turbine and the 5-MW NREL turbine, I would not recommend applying scaling relationships to derive the controller parameters. Instead, I would derive the controller parameters from the wind turbine properties. For example, VS_Rgn2K should be derived from the peak power coefficient, the associated tip-speed ratio, and rotor diameter. The proportional and integral gains in the blade-pitch controller can be derived from the sensitivity of aerodynamic power with respect to blade-pitch angle. The maximum pitch rate should be based on blade-pitch actuator considerations (0.25 rad/s is over 14 degree/s, which may be too fast for an actuator). See the documentation on the NREL 5-MW baseline controller for more information on how the parameters were derived for this turbine: nrel.gov/docs/fy09osti/38060.pdf.
The problem you are describing sounds related to the transition between below- (Region 2) and above- (Region 3) rated controller operation. The transition region is difficult to get right and highly impactful on turbine loading. I would first ensure that the torque controller functions well below rated and the pitch controller functions well above rated before tackling the transition region. The NREL 5-MW baseline controller improves the transition by saturating the integral term in the blade-pitch controller and by placing a conditional statement on the generator torque controller so that the torque is computed as if it where in Region 3–regardless of the generator speed-whenever the blade-pitch angle is above 1 degree.
I hope that helps.
Hello Dr. Jonkman,
Thanks so much for your response. It appears this is much more complicated than I thought it would be. After I sent the post, I considered modifying discon.dll to include a ramp in pitch angle rate during the transition from region 2 to region 3. For example, it may start at .05 rad/sec and ramp up to .15 rad/sec over a few seconds. To test this possibility, I again tried to reduce the PC_MaxRat from 0.25 rad/sec to 0.07 rad/sec. Results showed a reduction in the size of the spikes, however it now appears that a ramp in PC_max rate would lessen, but not solve the problem. The first time I tried lowering PC_MaxRat, I must have had a different value for another parameter that caused the very poor results.
In my post of Oct 30, 2019, I described the problems in trying to adapt the DLL controller designed for a 5-MW wind turbine to a 3 kW wind turbine. I could not resolve the problem so I increased the size of the wind turbine blades from 1.5 m to 4.5 m with a resulting power of 20 kW and rotation rate of 125 rpm. I used basically the same approach as described in my post of Oct 30, 2019. Results showed smooth control of the pitch angle without the spikes seen in the 1.5 m blade. Most likely, the initial problems were caused by the very rapid rotation rate of the 1.5 meter blades (400 rpm).