|Abstract: ||The need for better quality aerofoil data, extending to incidences well beyond stall, for the modelling of vertical-axis wind turbine (VAWT) start-up is established through blade element-momentum modelling. The model is used to show that differences between existing post-stall data are large enough to impact on turbine performance. The differences between the existing aerofoil data are found to result primarily from inconsistencies introduced by wind tunnel blockage and the potential flow-derived formulae used to correct for it.
A blockage tolerant wind tunnel test section has been constructed and calibrated for aerofoil testing. The tunnel has two semi-permeable walls made up of an array of transverse aerofoil-shaped slats. It produces free-air equivalent data without the need for blockage corrections. The ratio of slat to open area in the permeable walls that best minimises blockage has been obtained through testing of five different-sized NACA 0015 aerofoils.
In free-air, results for the five aerofoils would be identical. The open area ratio that produces the most consistent results for the five aerofoils, based on a standard deviation analysis of the results, is therefore judged to be the best. The aerofoils are tested in a solid-walled wind tunnel, and the data processed using a selection of blockage corrections. The corrections are also judged using a standard deviation analysis.
The tolerant tunnel, configured with the best open area ratio, outperforms the best corrections. Comparisons are made between results from the tolerant and solid-walled tunnels for the smallest (least blocked) aerofoil, with the latter corrected for blockage. Results are equal to within experimental error.
Three additional aerofoils (a symmetrical NACA 0018 and two cambered versions of it) are tested in the tunnel. Results are used in a numerical study of virtual camber effects on VAWT blades. The effect is found to be significant in turbines with large blade chord to turbine radius ratios. Assessments of Reynolds number effects between 20,000 and 300,000 and camber, for attached and detached blade flows, on forces are presented for all four profiles. A critical Reynolds number is established, above which a laminar separation bubble is able to form on the aerofoil's suction surface. Lift generation before stall improves greatly at supercritical Reynolds numbers.
VAWT start-up is modelled using a blade-element momentum method and the new experimental results. A conclusion is reached on the causes of characteristic VAWT start-up behaviour. The turbines enter an idling phase and either get stuck in it due to the presence of a "dead band" of negative torque production at low tip-speed ratios, or exit it into a rapid acceleration to final highest speeds. The behaviour results from blade Reynolds number effects, specifically the large jump in aerofoil performance at the critical Reynolds number. When turbines successfully self-start, blades are able to operate in local flows with supercritical Reynolds number for a sufficient portion of their rotation to slowly accelerate through the idling phase. When they encounter a "dead band" this is not the case.|