Accuracy of wind power estimates

Several aspects of power estimation techniques for wind energy conversion systems are studied. The sampling rate at which data are collected, ranging from once every 2 minutes to once every 3 hours, does not appear to significantly affect the average power for recording periods of one season. Increased averaging times produce small underestimates (less than 10 percent) of available power. The Rayleigh and Weibull distributions both give poor estimates of power for low mean wind speed situations, with the former being significantly worse. At higher wind speeds, both give good estimates, and the Rayleigh distribution is considerably simpler in form and application. A height extrapolation scheme for Weibull parameters is also investigated. Results are satisfactory for power estimates of ensembles of machines, but the scatter of values about the mean makes the method inappropriate for individual cases.