The siting of modern wind turbines to generate electricity requires accurate wind flow data over at least an annual weather cycle for optimum selection of both the wind turbine and its installed height, as well as for predicting the expected annual energy production at the site.
Since wind energy is a relatively new industry in North America, detailed annualized data are often not available, especially in rural areas, except for a single prediction of the average annual wind speed that is modeled by interpolating weather data from sensors that may be located several miles from the intended wind turbine site. This has led to several user disappointments when the actual energy obtained from a wind turbine was less than predicted.
In response, the present study focused on the development of a more realistic predictive model of power generation based on actual wind patterns at the site. Data on wind flow and energy output were obtained from an installation of three industrial-sized wind turbines at a local K-12 school. Minute-by-minute wind data were sorted and aggregated to provide the model input.
The empirical model developed predicted energy output within 4.1% of that observed over a period of 78 days; that is, an estimate of 13,875 kWh versus 14,470 kWh actually produced. A simpler model applicable to lower wind speeds (up to a “Moderate Breeze” of 7.5 m/s) was also developed that yielded a good energy output estimate of 11.5 kWh versus the full model's estimate of 11.7 kWh for an actual day's wind conditions.