Wind Speed Simulation Using Double Fourier Transformation

Abstract

Understanding the behavior and characteristics of wind speed will help to efficiently harness the wind's power. By simulating wind speed behavior, wind turbines can be matched to each energy application to provide a renewable energy resource capable of supplementing utility power. Annual mean wind speed can be used to estimate potential energy production for wind power, but this does not give an indication of the periodic variation in wind speed. In areas of West Texas, wind speed varies periodically both diurnally and seasonally. As a result, it can be characterized by a two-dimensional Fourier transformation. This research used a double Fourier series developed by Harbaugh and Merriam (1968) applied to a two-dimensional wind surface grid. Hourly wind speed data were collected from a weather station in St. Lawrence, Texas, to develop a wind surface map. The model was calibrated using data from 1992 and validated using data from 1995 and 1996. The mean absolute error (MAE) for the 288-point wind speed grid was 0.226, 0.799, and 0.707 m/s for 1992, 1995, and 1996, respectively. These errors fall within 6%, 26%, and 22% of the absolute mean wind speed for 1992, 1995, and 1996, respectively. Finally, the probability distribution providing best fit to the residuals was determined. Most frequently, the residuals were best fit by the normal distribution.