Machine Learning-Driven Surrogate Modeling and Quantum Metaheuristics for Gurney Flap Optimization in Darrieus Wind Turbines

The optimization of Darrieus wind turbine designs is critical to improving their aerodynamic performance and energy efficiency. The present study focuses to enhance the (C<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">P</inf> ) of the turbine by optimizing the design parameters of Gurney flaps attached to the trailing edges of the turbine rotors. Specifically, the flap angle and height are treated as the design variables. To explore the design space and generate random design points efficiently, Latin Hypercube Sampling (LHS) is employed. URANS simulations are carried out on these design points to provide high-fidelity data. A RBF surrogate model is then trained on this data, and is integrated with the Quantum-based Salp Swarm Algorithm (QSSA). The findings indicate a significant improvement in C<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">P</inf> , with the optimized Gurney flap configurations outperforming the smooth baseline case by approximately 158% and the initial Gurney flap setup by 188%. Further, the performance of QSSA is shown with another popular evolutionary algorithm— the Harris Hawks Algorithm. This approach demonstrates the potential of data-driven optimization techniques in the energy sector.