Analysis of wake flow models performance in an operating wind farm

Abstract

An open-source framework has been developed to design and analyze wind farms. This framework can describe the operation of each wind turbine and their interaction with each other through their wakes. The framework must be specified with site topography, wind turbine locations, and roughness length. It must also incorporate hourly or ten-minute meteorological time series, including wind speed, direction, and temperature at two heights. As a result, a production time series is obtained for each wind turbine with the same time interval as the meteorological data. The flow in the wake produced by each wind turbine is described using different analytical models. In this work, the framework is used to analyze the performance of several wake models and compare it with the commercial code WAsP. Actual production data from an operating wind farm located in the northern region of Uruguay is considered for the study. The results show that the correlation between actual and calculated production was slightly lower than 0.6 for each wake flow model considered in the analysis. The empirical probability density function of the difference between actual and calculated power had a mean and mode value of zero but a standard deviation close to 29% of the wind farm power. The calculated annual production of each wind turbine was approximated to the actual production using any of the wake models considered, as well as with WAsP. Although, the best performance was obtained using the open-source framework and the wake model proposed by Jensen.