ABSTRACT This study investigates the benefits of using J-shaped blades in a dual-row Darrieus wind turbine (DDWT), combining two innovative ideas whose simultaneous impact on the self-starting capability of Darrieus turbines has not been explored. The performance of the turbine was simulated using Computational Fluid Dynamics (CFD) and optimized using both the Taguchi method and a combined Machine Learning and Genetic Algorithm (ML-GA) technique. Three main operating parameters, including the tip speed ratio (λ), radial ratio (δ), and angular distance (ϕ) of dual-row turbines, were investigated across four levels using an L25 orthogonal array design. According to the Taguchi analysis, the impact of the parameters on the power output was ranked in the order of λ > δ > ϕ. After comparison, the optimized turbine predicted by the ML-GA technique was chosen due to the superior accuracy of the ML-GA approach compared to the Taguchi method. The optimal configuration of a DDWT with J-shaped blades was predicted at λ = 2.15, δ = 1.39, and ϕ = 75.36, with a Cp of 0.49 based on CFD simulations. Compared to the single-row wind turbine, the proposed turbine showed a significant increase of the Cp at low tip speed ratios resulting in a better self-starting performance.
Read full abstract