Description
The Savonius rotor, a prevalent type of Vertical Axis Wind Turbine (VAWT) for urban and small-scale applications, is recognized for its structural simplicity and self-starting capability [1]. However, optimizing dual-rotor configurations to mitigate wake interaction effects remains a crucial challenge in advancing wind energy technologies [2]. This study adopts a data-driven framework to analyze and improve the aerodynamic efficiency of dual Savonius rotors integrated with a deflector system. A Design of Experiment (DOE) approach was employed to systematically explore the impact of geometric parameters on rotor performance. An Artificial Neural Network (ANN) surrogate model, trained on the DOE dataset, predicted the average power coefficient (CP) with a high degree of accuracy (R2=0.995), significantlyreducing the reliance on computationally expensive CFD simulations. Optimization was performed using a Single-Objective Genetic Algorithm (SOGA), yielding an optimal deflector configuration with bleed jets that enhanced the average CP by 84.8% compared to the baseline dual-rotor setup without a deflector at a Tip Speed Ratio (TSR) of 0.8. This research underscores the potential of integrating machine learning with CFD in performance optimization, offering a scalable and efficient framework for advancing VAWT designs tailored for urban wind energy applications.
| Period | 4 Apr 2025 |
|---|---|
| Event title | EuroMech Colloquium on Data-Driven Fluid Dynamics and 2nd ERCOFTAC Workshop on Machine Learning for Fluid Dynamics |
| Event type | Conference |
| Conference number | 2 |
| Location | London, United KingdomShow on map |