Applicability Evaluation of Wind Turbine Wake Models

Abstract

[Introduction] The wake effect of wind turbines is an important cause of energy loss in wind farms. The wake study of wind turbines is beneficial to optimization of the turbine arrangement and improvement of the economic efficiency of wind farms. [Method] This paper presented a comparative study of the wake wind speed decay and turbulence intensity prediction of eight common wake models, respectively. To ensure the rationality of the evaluation, the quantitative analysis was carried out based on wind farm measurements and wind tunnel experimental data, and the range of comparison data was limited to 3 to 10 times the diameter downstream the turbine. [Result] The analysis results show that, for the prediction of wake wind speeds, the two-dimensional model fits the actual wake wind speed distribution structure better than the one-dimensional model, in which Jensen-Guass has better wake width prediction ability, while the 2D-k-Jensen wake center wind speed prediction has higher accuracy and adaptability to multiple conditions, with the maximum mean deviation and standard deviation of 8.7% and 5.5%, respectively, which are both applicable to the prediction of turbine wake wind speed. Jensen model has the best prediction ability of wake center wind speed in one-dimensional model, and the mean deviation of some conditions is less than 10%. While Park model is better in predicting the horizontal distribution of wake wind speed. In Case 2, the prediction performance is comparable to that of two-dimensional model, and the mean deviation of each condition is less than 10%, so it is more suitable for wake speed prediction than the former. For the prediction of turbulence intensity, the Ishihara model shows a clear advantage in the prediction of turbulent structure, with mean deviations below 10%, but the prediction of turbulence intensity at the center of the wake is poor, which is not conducive to the prediction of the turbulence intensity at locations downstream the turbine. Among the remaining models, Frandsen and Jensen-Guass models are relatively good at prediction of low ambient turbulence intensities. However, there is an opposite trend between the two for high ambient turbulence intensities: the Frandsen model has higher prediction accuracy and is suitable for turbine turbulence intensity prediction, whereas the prediction result of Jensen-Guass is much larger than the experimental value, and is unstable. The prediction accuracy of wake wind speed for all models is greatly improved at high upstream wind speeds, and the increase in ambient turbulence intensity contributes to the prediction accuracy of the wake wind speed and turbulence intensity for all models, except for the Jensen-Guass model. [Conclusion] The predicted values of wake wind speed for the Jensen-Guass model and the 2D-k-Jensen model coincides better with the measured data, while the prediction performance of the Frandsen model turbulence intensity is better, so they can be used as the reference wake models for the optimization of wind turbine placement and wake control analysis for offshore wind farms.

Keywords

• wind turbines
• wake models
• wind speed attenuation
• turbulence intensity
• measured data
• comparison and research