Abstract:
Currently, as an effective technique to face the energy crisis, large-scale wind turbines in MW level is more and more widely used in the field of environment protection and green energy industry.In boundary layer wind field, random wind load can bring more obvious aerodynamic load fluctuation to long blades and high tower, which affects the structural stability of wind turbines.An accurate random wind speed model plays a significant role in the aerodynamic load calculations of wind turbines in the atmospheric boundary layer.Random wind speed can be decomposed into averaging wind speed and fluctuating wind speed.The averaging wind speed can be described by wind shear effect, and the fluctuating wind speed can be described by turbulence spectra.In the runtime environment of wind turbines, it's proper to use specific turbulence spectra to describe the fluctuating wind speed of inflow on the inlet because the entrance region of wind farm is the equal of a flat field.Compared with static buildings, the dynamic turbine rotor causes great disturbances to the wind speed.So the hybrid numerical simulation to combine the random wind speed generated by specific turbulence spectra on the inlet with turbulence models is more appropriate to the aerodynamic load calculations.In this paper, we verified the accuracy of random wind speed generation methods on the inlet of wind turbines.IEC61400-1 and many research papers recommend to use exponential law to describe the wind shear effect.The improved Von Karman spectrum corrected the flaws of Von Karman spectrum below about 150 m, which is more appropriate to wind farms.Based on the exponential law, 3D improved Von Karman spectrum and cross-spectrum model.We used the harmony superposition method and auto-regressive moving-average method to generate fluctuating wind speed of a 33 kW horizontal axis wind turbine.The hub height of this wind turbine was 15.4 m, and the rated wind speed was 11 m/s.The terrain roughness length was 0.03 m and the wind shear index was 0.2.The time step was 0.1 s and the upper rate limit was 2π rad/s.The length of the simulation was 100 s.The inlet was a rectangular area with 30 m × 30 m, and the grid step was 2 m.As a contrast, in this paper, we used the PSCAD to generate fluctuating wind speed with the same parameter.We compared these three methods, power spectrum, self-correlation and cross-correlation.The results showed that the power spectrum which was produced by HSM fit well with target power spectrum.The self-correlation of these three methods decreased as time increased, and the self-correlation of ARMA and PSCAD was better than HSM.The cross-correlation from the HSM method decreased as distance increased, which fit well with the measured results.Last, in this paper, we compared the turbulence intensity and got the three-dimensional distributions of velocity on the inlet.The turbulence intensity decreased as height increased, the difference value of turbulence intensity in the along-wind, crosswind, vertical wind was 8.96%, 6.98%, and 4.91%, respectively.This difference value would be more obvious on the large-scale wind turbines.