A Review of Wind Turbine Polar Data and its Effect on Fatigue Loads Simulation Accuracy
To certify a Wind Turbine the standard processes set out by the GL guidelines and the IEC61400 demand a large number of simulations in order to justify the safe operation of the machine in all reasonably probable scenarios. The result of this rather demanding process is that the simulations rely on lower fidelity methods such as the Blade Element Momentum (BEM) method. The BEM method relies on a number of simplified inputs including the coefficient of lift and drag polar data (usually referred to as polars). These polars are usually either measured experimentally, generated using tools such as XFoil or, in some cases obtained using 2D CFD. It is typical to then modify these polars in order to make them suitable for aeroelastic simulations. Some of these modifications include 360° angle of attack extrapolation methods and polar modifications to account for 3D effects. Many of these modifications can be perceived to be a black art due to the manual selection of coefficients. The polars can misrepresent reality for many reasons, for example, inflow turbulence can affect measurements obtained in wind tunnels. Furthermore, on real wind turbine blades leading edge erosion can reduce performance. Simulated polars can even vary significantly due to the choice of turbulence models. Stack these effects on top of the uncertainties caused by yaw error, pitch error and dynamic stall and one can clearly see an operating environment hostile to accurate simulations. Colloquial evidence suggests that experienced designers would account for all of these sources of errors methodically, however, this is not reflected by the certification process. A review of experimental data and literature was performed to identify some of the inaccuracies in wind turbine polars. Significant variations were found between a range of 2D polar techniques and wind tunnel measurements. A sensitivity study was conducted using the aeroelastic simulation code FAST (National Renewable Energy Laboratory) with lift and drag polars sourced using different methods. The results were post-processed to give comparisons the rotor blade fatigue damage; variations in accumulated damages reached levels of 164%. This variation is not disastrous but is certainly enough to motivate a new approach for certifying the aerodynamic performance of wind turbines. Such an approach would simply see the source of polar data and all post-processing steps documented and included in the checks performed by certification bodies.