Accuracy Computational Simulation of the Savonius Wind Turbine Aerodynamic Performance: Effect of RANS Turbulence Models

Abstract This study investigates the impact of Reynolds-averaged Navier–Stokes (RANS) turbulence models on the accurate computational simulation of the aerodynamic performance of the Savonius wind turbine. Five different models, including the Spalart–Allmaras (SA), realizable k–ε (RKE), shear stress transport (SST) k–ω, generalized k–ω (GEKO), and transition SST (TSST) models, are considered. The insight analysis utilizes high-fidelity two-dimensional (2D) incompressible unsteady RANS simulations for the two-blade turbine type with two configurations, TB1 and TB2, under various working conditions. The evaluation is based on quantitatively comparing the numerical aerodynamic parameters with the available experimental data and a deep analysis of the flow mechanism-induced turbine. The numerical results, through the comparison with an inviscid simulation, indicate a high sensitivity of the computed torque and power coefficients to the viscous model. All turbulence models fail to accurately predict the turbine's aerodynamic parameters at low speeds, where the tip speed ratio (TSR) is below 0.4. The SA and k–ω based turbulence models are effective in reproducing turbine performance in the optimal regime, where the TSR ranges from approximately 0.5 to 0.8. Additionally, the RKE model provides satisfactory predictions compared to experimental data, extending its accuracy to the high-speed regime across all tested configurations. Finally, the RKE model is the recommended turbulence model for computational fluid dynamics (CFD) simulations of Savonius wind turbines.