I would like to exchange opinions about the right way to schedule the pitch during startup of a variable speed gearless permanent magnet small wind turbine (20-50 kW).
My idea is the following:
THETA_startup = the angle of maximum aerodynamic torque for zero rpm (revolutions per minute), that is with static rotor
THETA_fine = the angle, around zero, of maximum CP (of optimum lambda)
set the torque to zero
bring the pitch to THETA_startup
schedule the pitch vs. rpm linearly between THETA_startup and THETA_fine, in a way that when the rpm reach the cut-in rpm, the angle is THETA_Fine.
when the rpm > rpm-cutin, set the torque gradually to the quadratic torque required with the classical control law.
There is not much published around this subject, at least I do not find it. Maybe it is proprietary infomation, but anyway I would be interested in opinions and or bibliography about this isssue, most important to us in order to create a “low-wind” cut-in turbine.
I use an idealized optimal pitch angle versus tip-speed-ratio curve. This curve is developed with performance analysis tools like WTPerf. It represents the highest Cp predicted for each TSR from 0 to the TSR for optimal Cp. I fit that with a 3rd order polynomial and follow it during startup. I filter the calculated TSR (from rotor speed and wind speed) with a first order filter with a 1-second time constant to filter out fluctuations in wind speed.
Unfortunately, the analysis tools are way out of their valid operating areas when predicting performance at low TSRs. Thus, a little experimentation in the field usually allows me to adjust this curve up or down just a bit. Fortunately, turbine startup performance is not very sensitive to getting this curve just right.
I had thought something similar, but rejected it as too complex. Now that I am not completely satisfied with the startup behavior, I will have to reconsider!
What do you think about the objection that the wind velocity that you feed in your TSR calculation is actually “wrong”, in the sense of 3D turbulence (not representative of the velocity on the whole rotor disc) and is also disturbed by the blades themselves (I assume you measure it through the nacelle anemometer)?
If I can ask, is your fitted cubic THETA(lambda) very far from my straight line?