No abstract

The formation of stall cells over a NACA 0012 airfoil at a Reynolds number of one million has been investigated numerically, using unsteady Reynolds-averaged Navier-Stokes (URANS) and delayed detached-eddy simulation (DDES) approaches. The simulations are performed with a very wide computational domain (10 chord length) to minimize the influence of spanwise periodic boundary conditions. For the URANS simulations, four different spanwise mesh resolutions are tested to determine the minimum resolution required to capture the formation of stall cells. Both URANS and DDES results show a sudden decrease in lift and increase in drag between 16°and 17°angle of attack, accompanied by a significant change of separated flow patterns. Stall cell structures are observed clearly in the URANS solutions between 17°and 19°with a spanwise spacing of about 1.4 to 1.8 chord length, which agrees well with a theoretical prediction based on the slope of the lift curve in this angle-of-attack range. The DDES results show much more complex flow patterns over the airfoil at these high angles of attack, although the spectral analysis of wall shear stress suggests the existence of flow structures having a similar spanwise length scale to the stall cells.

A Double Multiple Streamtube model, a free-wake vortex model (both widely used for vertical axis wind turbine design) and RANS CFD simulations are used in this work to predict the performance of the 17m Vertical Axis Wind Turbine, field tested by Sandia National Laboratories. The three-dimensional, full scale calculations are compared with the experiments in terms of power coefficient, power and instantaneous turbine torque to assess the validity of each model. Additionally, the two aerodynamic models and RANS CFD are compared to each other in terms of thrust and lateral force. The two models and CFD agree well with the experiments at the turbine optimal tip speed ratio. However, away from the optimal tip speed ratio, the streamtube model significantly deviates from the experimental data and from the other numerical models. RANS CFD gives a good agreement with the experiments, slightly underestimating the power coefficient at every tip speed ratio tested. The vortex model proves to be a useful tool with a better accuracy than the streamtube model and a much lower computational cost compared to RANS CFD.

No abstract

No abstract