Effects of airfoil's polar data in the stall region on the estimation of darrieus wind turbine performance

Interest in vertical-Axis wind turbines (VAWTs) is experiencing a renaissance after most major research projects came to a standstill in the mid 90's, in favour of conventional horizontal-Axis turbines (HAWTs). Nowadays, the inherent advantages of the VAWT concept, especially in the Darrieus configuration, may outweigh their disadvantages in specific applications, like the urban context or floating platforms. To enable these concepts further, efficient, accurate, and robust aerodynamic prediction tools and design guidelines are needed for VAWTs, for which low-order simulation methods have not reached yet a maturity comparable to that of the Blade Element Momentum Theory for HAWTs' applications. The two computationally efficient methods that are presently capable of capturing the unsteady aerodynamics of Darrieus turbines are the Double Multiple Streamtubes (DMS) Theory, based on momentum balances, and the Lifting Line Theory (LLT) coupled to a free vortex wake model. Both methods make use of tabulated lift and drag coefficients to compute the blade forces. Since the incidence angles range experienced by a VAWT blade is much wider than that of a HAWT blade, the accuracy of polars in describing the stall region and the transition towards the "thin plate like" behaviour has a large effect on simulation results. This paper will demonstrate the importance of stall and post-stall data handling in the performance estimation of Darrieus VAWTs. Using validated CFD simulations as a baseline, comparisons are provided for a blade in VAWT-like motion based on a DMS and a LLT code employing three sets of post-stall data obtained from a wind tunnel campaign, XFoil predictions extrapolated with the Viterna-Corrigan model and a combination of them. The polar extrapolation influence on quasi-steady operating conditions is shown and azimuthal variations of thrust and torque are compared for exemplary tip-speed ratios (TSRs). In addition, the major relevance of a proper dynamic stall model into both simulation methods is highlighted and discussed