Analyzing aggregated characteristics of distributed wind farms

Summary form only given. Wind generation is now a major power source in many countries especially in China. It is well known that dealing with large amounts of intermittent generation resources connected to an electric power grid is a difficult for power system planners and operators. Operating cost may increase significantly due to wind variability and uncertainty for systems with high wind penetration because of the increased reserve and other ancillary services. Although the generation capacity of a single wind turbine is highly uncertain, the grid operator would be more interested in the aggregated generation of all wind farms. The operating practice of some power grids show that geographically distributed wind farms tends to be much less volatile due to the strength variation of within a wind dynamic system. This effect of “uncertainty compensation” is favorable for improving wind power utilization and reducing the system reserve requirements. We focus on analyzing the aggregated stochastic characteristics of geographically distributed wind farm generation. A dynamic system based on the MM5 and dynamic downscaling technology for weather forecasting is established to describe the relationship between atmospheric and near-surface wind fields of the individual wind farms and to forecast the wind speed of each wind farm. A recursive algorithm based on the extended Kalman filter theory is developed to estimate the near-surface wind speed of individual wind farms based on their geographical locations, the wind dynamics in atmosphere, and the initial conditions of the wind fields. Then the probability distribution of aggregated generation of wind farms is estimated based on the finite Gaussian mixture distribution. Furthermore the temporal and spatial correlations among the individual wind farms are analyzed and the major factors affecting the probability density function of the aggregated generation are studied. The actual data of the NCEP/NCAR Reana- ysis datasets validates the assertion that the aggregated wind generation of distributed wind farms is far less volatile than that of a single wind farm.