Aerodynamic optimization of vertical axis wind turbine blades: A hybrid surrogate modeling strategy combining CFD and a modified BP neural network

Vertical axis wind turbines (VAWTs) have significant development prospects in urban buildings and low-wind-speed environments due to their simple geometry, low noise, operation under multi-directional wind, and other technological advantages. The operation efficiency of VAWT is an essential factor restricting its development, and blade design plays a key role in determining its operation efficiency. In this paper, a model-driven optimization design framework for the aerodynamic performance of vertical axis wind turbine blades is proposed. Computational fluid dynamics (CFD) and an improved back-propagation neural network (BPNN) are combined to establish the relationship between design variables and aerodynamic performance. Then, a DBO–BPNN surrogate model coupled with the non-dominated sorting genetic algorithm (NSGA) is employed to identify the optimal design scheme. Compared with the original VAWT blade, the lift-to-drag ratio of the optimized blade is increased by 7.52 units, the manufacturing material is saved by 8.15%, and the carbon emissions in the raw material stage are reduced by 91.079 kg (carbon fiber reinforced plastic) and 62.132 kg (glass fiber reinforced plastic), respectively, realizing the collaborative optimization of energy efficiency and environmental benefits.