Evaluation of the power performance of wind turbines in wind farms

The International Electrotechnical Commission (IEC) describes the standard procedure to measure the power performance of a wind turbine. Manufacturers follow the IEC guidelines to assess the performance of their turbines, while wind farm operators conduct IEC-compliant power performance measurements to determine whether the turbines are performing properly. Since most operating wind turbines are in wind farms, the IEC standard describes the procedure to ensure that the measured power performance is not impacted by the neighbouring turbines. Most importantly, wind directions are restricted to make sure that both the tested turbine and the wind measurement equipment are not in wake.
When several wind turbines are clustered in a wind farm, upstream wake-free turbines might also be affected by flow disturbances caused by the other turbines. The IEC standard recommends to measure the wind speed at a minimum distance of two rotor diameter from the turbines to retrieve blockage-free wind speed measurements. However, wind farm blockage effects might still influence the flow field at that distance. Additionally, the turbine under test might be subject to blockage effects from the neighbouring turbines. This thesis evaluates the impact of blockage effects on power performance measurements and how to correct for them. Additionally, since numerous wind turbines operate under waked conditions for a substantial amount of time, it is investigated how to accurately measure the power performance of a waked wind turbine.This thesis aims to advance the methods for the evaluation of the power performance of a wind turbine in a wind farm. To this purpose, nacelle lidar measurements are retrieved in front of the turbine under test. Several nacelle lidar configurations are tested for both wind speed and turbulence measurements under both waked and wake-free conditions. Specifically, it is tested whether nacelle lidar measurements can be used to improve the accuracy of power performance measurements relatively to the current standard procedure.
The impact of blockage on power performance measurements is evaluated through both simulations and measurements. Reynolds-averaged Navier-Stokes (RANS) simulations are performed with both a row of five turbines and a large wind farm with 100 turbines. Results are compared with simulations of a single isolated turbine operating under the same freestream conditions. Additionally, measurements are analyzed from a test site consisting of a single row of five turbines. The numerical results show consistent power performance deviations between the wind farm and the isolated cases, with C_P variations up to 4%. The measurements show that the power output of the turbine on one side of the row changes with the wind direction due to blockage effects from the neighbouring turbines. Specifically, compared to wind directions perpendicular to the row, the power output varies of −1.8% and +1.8% when the turbine is the most upwind and downwind of the row, respectively.
A method is presented to correct for blockage effects on IEC-compliant power performance measurements. Two different approaches are presented to apply the correction: one based on numerical simulations and one based on short range nacelle lidar measurements. Both the approaches are tested numerically through RANS simulations, showing that they improve the evaluation of the power performance. Additionally, the method is used to correct power curves measured under waked conditions.
In addition to the correction method, lidar-based data-driven power curves are defined to evaluate the wind turbine power performance under waked conditions. Specifically, multivariate power curves are implemented as multivariable polynomial regressions, whose input variables are several wind speed and turbulence measurements obtained with nacelle lidars. A numerical framework is implemented to test the multivariate power curves under both waked and wake-free conditions using different nacelle lidar configurations. The same framework is also used to test nacelle lidar turbulence measurements. Furthermore, the numerical results are validated with lidar measurements from the field. Results show that the multivariate power curves are more accurate than the IEC standard power curve under both wake-free and waked conditions. The power output estimation improves when using nacelle lidar turbulence measurements in addition to wind speed measurements. Several nacelle lidar scanning configurations are tested through both simulations and measurements. When measuring turbulence under nearly homogeneous conditions, at least six beams are needed, including one beam with a different opening angle, to retrieve all the six Reynolds stresses. Additionally, the Reynolds stresses estimation improves by increasing the opening angle, while no substantial improvement is obtained by increasing the number of beams beyond six. The optimal lidar configuration to implement the multivariate power curves is site-specific. However, both numerical and experimental results show that a circular scanning configuration provides similar accuracy to the optimal configuration, as long as the scanning pattern has a diameter equal to around 0.9 the turbine rotor diameter.