Stochastic analysis of wind stream and turbine power

Stochastic analysis of a high-frequency wind data tape has been performed. The tape includes wind speed and direction as well as wind-turbine-generated power. In an attempt to correlate wind speed with turbine power, data were sampled every 2 s from a United States Department of Energy demonstration 200 kW wind turbine installation. Wind speeds were recorded from three heights on a meteorological tower and from the wind-driven generator. Auto-correlation and spectral density functions were found for both the wind speed and the turbine power. Spatial and temporal averaging was performed, and time-lagged spatial cross-correlations, cross-spectral density functions and coherence functions were computed. A time-lagging technique was used to translate meteorological tower data to the turbine. Nonstationarity in the mean and standard deviation were intvestigated. These analyses form the bases for data collection procedures for initial site evaluation and for full-scale machine power predictions.