Improvement of Cross Section of Magatama Type VAWT Blade by Reinforcement Learning

Abstract

Magatama type blade configuration, which has been exclusively designed for vertical axis wind turbine (VAWT), is improved by Reinforcement Learning (RL). In this RL, VAWT blade configuration that generates larger aerodynamic force in direction of rotor rotation is developed. Power performances of the VAWT with the improved new Magatama type blades are tested by wind tunnel experiments. Results show the power from the VAWT rotor with the new Magatama blades is increased. Flow structure around the new Magatama blades in the VAWT rotor and aerodynamic forces generated by the blades are investigated by unsteady numerical simulation. Results show the new blades increase rotational force in downwind zone. In this downwind zone, direction of aerodynamic force is close to that of rotor rotation.