Horizontal Axis Wind Turbine Aerodynamics: Three-Dimensional, Unsteady, and Separated Flow Influences

Surface pressure data from the National Renewable Energy Laboratory’s “Unsteady Aerodynamics Experiment” were analyzed to characterize the impact of threedimensionality, unsteadiness, and flow separation effects observed to occur on downwind horizontal axis wind turbines (HAWT). Surface pressure and strain gage data were collected from two rectangular planform blades with S809 airfoil crosssections, one flat and one twisted. Both blades were characterized by the maximum leading edge suction pressure and by the azimuth, velocity, and yaw at which it occurred. The occurrence of dynamic stall at all but the inboard station (30% span) shows good quantitative agreement with the theoretical limits on inflow velocity and yaw that should yield dynamic stall events. A full three-dimensional characterization of the surface pressure topographies combined with flow visualization data from surface mounted tufts offer key insights into the three-dimensional processes involved in the unsteady separation process and may help to explain the discrepancies observed with force measurements at 30% span. The results suggest that quasi-static separation and dynamic stall analysis methods relying on purely two-dimensional flow characterizations may not be capable of simulating the complex three-dimensional flows observed with these data. INTRODUCTION Wind turbine aerodynamic loads routinely exhibit startling spatial and temporal complexities, driven by the combined influences of three-dimensionality, unsteadiness, and dynamic separation. Three-dimensional cross flows as well as quasi-steady and unsteady separation events arise from both the stochastic nature of the inflow and the components of the turbine architecture. Rapid changes in wind speed or direction dynamically alter the local angle of attack along the span. Both angle of attack and dynamic pressure change radially along the span, and are further effected by variable speed or pitch operation. Because of these many influences, accurate and reliable prediction of wind turbine aerodynamics remains a unique challenge for today’s computational modelers. The overall contribution of dynamic stall to wind turbine structural loads continues to undergo opinion shifts in the research community. Several years ago, the effects of dynamic stall were deemed inconsequential as interest centered around predicting power performance. This conclusion was not surprising since the rapid transient loads produced by a single dynamic stall event occur over a small percentage of the total turbine rotation cycle. Opinions changed when it was shown that some type of dynamic stall model was necessary to adequately predict both peak and fatigue loads. Currently, the impulsive loading introduced via dynamic stall aerodynamic models can be shown to amplify, damp, or have little effect on the resulting structural loads depending on a particular time series, turbulent inflow condition, or turbine architecture. As such, the ability of current aerodynamic models to adequately predict quasi-steady, three-dimensional, post-stall performance remains in question, let alone three-dimensional dynamic stall performance in an extremely stochastic inflow environment. Most wind turbine structural design codes rely on Blade Element Momentum Theory (BEMT) to simulate blade aerodynamic performance. This algorithm allows the aerodynamics model to run quickly, enabling practical design trade-off analyses. The use of empirical two-dimensional wind