CFD modeling of varying complexity for aerodynamic analysis of H-vertical axis wind turbines

Computational fluid mechanics (CFD) is considered as an efficient approach for studying aerodynamic characteristics of vertical axis wind turbines (VAWTs). Currently, 2D Unsteady Reynolds-Averaged Naviere Stokes (URANS) is widely applied, although previous researches revealed its limit accuracy in the aerodynamic analysis. This paper investigates the accuracy and feasibility of various CFD modeling techniques, namely 2D URANS, 2.5D URANS, 2.5D large eddy simulations (LES), 3D URANS and 3D LES, in the aerodynamic study of VAWTs through a comparison with the wind tunnel results. Compared with the URANS method, the LES approach can provide more accurate prediction on the aerodynamic performance for VAWTs operating at the dynamic stall. The significant improved simulation results by 2.5D LES imply that the neglect of tip vortices may not be the major mechanism causing the over prediction in 2D and 2.5D URANS. 2.5D LES can be regarded as a promising and efficient approach to investigate the aerodynamic behaviors of VAWTs, considering the compromise between the accuracy and computational cost among 2.5D LES, 3D LES and 3D URANS. Furthermore, considering the huge amount of time consumed by CFD simulations, a hybrid meta-model is therefore proposed to predict the power coefficient of VAWTs. The prediction results show that the accuracy of the hybrid meta-model satisfies the requirements, and the calculation time is also reduced.