From 2D Parametric Study to 2.5D Scale-Resolving Simulation: A Comprehensive Study on VAWT Self-Starting Behaviour

The inability of VAWTs to reliably start up under low wind speeds remains a major challenge. Previous CFD studies have revealed three key limitations: the absence of a robust model for predicting accurate self-starting flow behavior, the limited development of passive flow control strategies to enhance the startup performance, and the inaccuracy of the predicted self-starting flow time in the 2D model. To address these challenges, this thesis is structured into three key research chapters.
First, a robust 2D self-starting model was developed using the Taguchi design of experiments. The analysis demonstrated that the temporal and spatial discretisation significantly influence the accurate prediction of the turbulent flow, particularly in the low TSR regime, while the turbine geometry and inlet turbulence intensity must be carefully controlled to ensure model stability.
Second, two passive flow control devices were examined: a J-shaped aerofoil and Gurney Flaps (GFs). At a low wind speed of 5 m/s, the 37.5%C J-shaped configuration showed the highest startup torque with the moderate turbine power performance, while attaching 1% h/C GFs at the trailing edge further enhanced the blade aerodynamics. Combining both techniques, the JGFin configuration achieved the greatest startup torque, whereas the JGFboth configuration provided a slight increase in startup torque but significantly improved the power coefficient at the TSR = 2.
Finally, a 2.5D SBES model was developed to evaluate its capability to predict the physical self-starting time of the turbine. The baseline 2D URANS, 2.5D URANS, and 2.5D SBES models were assessed, and the comparative results clearly demonstrated that the 2.5D SBES approach provides a substantial improvement in replicating the self-starting profile, particularly at the low to medium TSRs.
Overall, this thesis presents a validated CFD framework, introduces effective passive flow control strategies, and establishes an advanced simulation methodology for the accurate self-starting prediction.