Statistical Analysis of intra-farm microscale wind characteristics at selected Xcel wind farms

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 27 January 2011: 4:15 PM
Statistical Analysis of intra-farm microscale wind characteristics at selected Xcel wind farms
4C-2 (Washington State Convention Center)
Yuewei Liu, NCAR, Boulder, CO; and Y. Liu, W. Cheng, G. Wiener, B. Lambi, and B. Mahoney

Understanding the micro-scale flow features within and around wind farms can be used to improve real-time wind energy forecasts through optimization of data assimilation, model output post-processing, and model performance evaluation. In this study, wind-farm observations, including turbine nacelle wind and power measurements along with nearby meteorological tower observations from seven wind farms in Colorado, Texas and Minnesota are analyzed to study the statistical properties of the microscale flows in these wind farms.

This presentation describes three aspects of intra-farm wind properties. Firstly, we studied the dominant wind spatial-distribution patterns and their occurrence frequencies using the self-organizing maps (SOMs) clustering analysis technology based on turbine nacelle wind observations and met-tower data. The dominant intra-farm wind patterns are associated with the weather regimes. Secondly, the relationship between the farm-wide mean hub-height wind speed and the total farm power generation is computed, referred as to node-level-power-curve (NLPC). Tests with retrospective data indicate that reliable power forecasts can be achieved using NLPC when high-accuracy farm-wide mean hub-height wind speeds are available. Thirdly, we analyzed the wind power persistence property. It is found that the wind persistence (i.e. persistent forecast accuracy) highly depends on the weather regimes and wind farm locations. For example, at a wind farm in SW Colorado, two hours persistence forecasts can be very accurate in most days during winter months. However, for the summer season, strong ramp events occur frequently and the persistence forecast degrades quickly with time.