Stochastic Model Of The Wind Environment PDF Download

The economic feasibility of installing a wind energy conversion system at a given site depends on an accurate assessment of the wind characteristics at that site. To accomplish this, data records must be analyzed effectively. This report utilizes statistical methods and probability models to determine optimal evaluation procedures for survey data. Several years of hourly records from five typical Midwest sites, one high tower Midwest site, and one Rocky Mountain site provide a data base to develop the models and procedures. The minimum duration of survey is analyzed as a function of desired confidence level and acceptable tolerance range. Alternate survey procedures are discussed in terms of total survey duration and number of recorded values. Single year velocity records appear acceptable for velocity statistics, but are marginal for wind power estimates unless additional information is available. Persistence of the wind is measured in terms of velocity run duration, and a simple but apparently general model is developed for the probability of run duration. This model, which is calibrated to the average velocity, is based on the co linearity of probability versus run duration after each has been logarithmically transformed. Autocorrelation and cross correlation functions of both wind velocity and wind power are analyzed. The observed histograms for velocities exhibit a reasonable fit to both the Chi -2 and Weibull distributions, and the wind power appears to agree with distributions derived from these.