Computational study of stalled wind turbine rotor performance
Simplification of the aerodynamic control of large horizontal axis wind turbines (HAWTs) has been identified as an important step towards improved reliability and reduced cost. At present the majority of large HMrrs use active control to regulate power and loads. A simpler strategy is to use the inherent stalling of the rotor blades in high winds to limit power and loads. Unfortunately the performance of stall regulated HAWTs 1S poorly understood; current performance models often fail to correctly predict peak power levels. The benefits of passive control of power and loads cannot be utilised because of this uncertainty. This study examines the possible reasons for the poor performance of current prediction techniques 1n high winds with the objective of fonmulating a new model. The available experimental evidence suggests that rotor stall is caused by turbulent separation at the rear of the blade aerofoil, growing in extent from the root in increasing wind. This 'picture' of the stalling HAW! rotor forms the basis of the approach. The new model consists of a prescribed vortex wake, first order panel method (extended to represent the viscous region of trailing edge separation) and three dimensional integral boundary layer directly coupled in an iterative scheme. A sensitivity study of rotor indicates that the most important performance to wake geometry factor is the rate at which the wake is convected downstream. However, it is found that stalled power levels are insensitive to wake geometry; the study concludes that the problem of poor prediction of high wind performance lies on the rotor blades. Before using the complete code to calculate the performance of a rotor it 1S first tuned for the aerofoils used on the blade. Aerofoil perfonmance characteristics measured in a wind tunnel are synthesised by the model. Ideally these characteristics should include measured pressure profiles below and above stall. Validation of the complete code against detailed measurements taken under controlled conditions on a three metre diameter machine indicates significant differences in the perfonmance of aerofoil sections on a wind turbine blade when compared to the same section when tested in a wind tunnel. Derived lift coefficients show a reduced lift curve slope and more gentle delayed stall. Similar results are found when the code is applied to two Danish stall regulated machines. These two machines although having very similar geometries and using the same family of aerofoils do however show differences in derived post stall drag. This is thought to be due to the different thickness distributions of the two rotors. The validation and applications of the new model show that it can accurately predict the peak power level of stall regulated machines.