Research On The Power Enhancement Of Large-Scale Wind Turbines With Winglets

The tip effect of the horizontal axis wind turbine blade,a finite-length component,gives rise to a swirling motion that is concentrated into the known helical tip vortices,causing corresponding induced drag,hence decreasing the torque of the blade.Winglets,as a kind of power enhancement devices,can spread out the effect of the wing tip,resulting in reduction of the downwash and thereby the induced drag on the blade.In order to study the effect of the winglet on the blade aerodynamic characteristics deeply,the following aspects were studied in this paper.The computations of the aerodynamic characteristics of the NREL Phase ? and NH1500 blade were performed with the CFD and FVW methods.The calculation results were in good agreement with the wind tunnel test results of the model,indicating the reliability of the CFD and FVW method.The performance of the NREL 5MW blade was also computed and compared with the FAST results to verify the calculation accuracy,and then the tip vortex lines were analyzed to study the aerodynamic loss effect caused by the tip vortex.The geometrical parameters in consideration included twist angle,dihedral angle and sweep angle.Orthogonal tables were established for the parameters above in order to reduce the overall computational costs.Aerodynamic performance of 16 configurations with wind speed of 5 m/s,8 m/s,11.4 m/s were simulated with CFD.Range and variance analysis were carried out to study the influence of the parameter variation.Wind sensitivity tests of the typical configurations were also performed.The flow state around the blade were analyzed and compared with the original blade to find out the work mechanism of the winglet thoroughly.Machine learning prediction models including multivariate regression models(linear,secondorder,third-order)and neural network model(back propagation and radical basis function)were established based on the CFD calculation data.Current results showed that the prediction ability of the multivariate second-order regression model and the neural network model were better.Free vortex wake model was developed by arranging lift surface model in the winglet part to simulate the tipvane effect.The results conformed well to the prediction results above,which provided mutual accuracy verification.