Dynamic Characterization of the AWT-26 Turbine for Variable Speed Operation

Document Type



Mechanical Engineering

Report Number



National Renewable Energy Laboratory

Publication Date


First Page


Last Page




The Advanced Wind Turbines, Inc. AWT-26 Prototype #3 at the National Wind Technology Center (NWTC) was modified for variable speed operation. This program was in support of emerging technologies in variable speed power generation. Electronic Power Conditioning Inc. (EPC) of Corvallis, Oregon manufactured the new variable speed generation system (VSGS) for the turbine. The VSGS consisted of a doubly-fed generator and a Unipolar Series Resonant Converter (USRC). The AWT-26 is a downwind, teetered, two-blade machine that operates at a constant 57.3-rpm rotor speed. The modification to variable speed operation allowed a rotor speed range of 32 rpm to 62 rpm. The turbine was not originally designed for variable speed operation and the potential for operating at a resonant condition existed. The purpose of the experiment described in this report was to measure turbine vibration response at discrete rotor speeds within the variable speed range and identify any potential resonant conditions. The converter variable speed algorithm could then be designed to avoid these resonant conditions.


The turbine was instrumented with four accelerometers. Two accelerometers were positioned to measure tower motion; the other two were placed in the nacelle to measure nacelle motion. The VSGS was then operated in constant speed mode at discrete rotor speeds over the variable speed range. Accelerometer spectra were acquired over a 10-minute averaging period with the turbine held at constant speed. Wind speed and direction data were also acquired to compare the spectra at different rotor speeds. Harmonic peaks were plotted for each accelerometer over the rotor speed range. Natural frequencies were also plotted to determine crossings with the rotor harmonic frequencies. Rising harmonic peaks at these crossings were indications of possible resonant conditions.


The reduced accelerometer data showed potential resonant conditions at 32 rpm, 38-40 rpm, 48 rpm, and 52 rpm. It was determined that the variable speed algorithm be designed to avoid 32 rpm, 39 rpm, and 48 rpm within a ± 1 rpm range. The crossing at 52 rpm was not as severe as the other critical speeds and avoiding this speed would significantly impact the variable speed operation. Another conclusion from the experiment was that the tower accelerometer data were not as useful as the nacelle accelerometer data. For similar turbine designs, this could reduce the cost of like experiments and turbine vibration monitoring equipment.