The Savonius rotor is a widely favoured type of Vertical-Axis Wind Turbine (VAWT) for small-scale and urban applications, owing to its simple design and self-starting capability. Harnessing wind energy through VAWTs to achieve more efficient and cost-effective power generation is an attractive prospect. However, the negative torques on the returning blades of these turbines lead to poor power generation efficiency. To address this, an innovative cylindrical deflector with bleed jets is proposed to enhance the performance of dual Savonius rotors. These bleed jets are released through narrow slots extending from the front stagnation point to regions near the top and bottom separation points, creating flow perturbations. The interaction of the bleed jets with the boundary layer modifies the wake shear layer around the deflector. Optimizing the aerodynamic performance of VAWT wind farms, particularly by addressing wake interactions, is essential to meet future energy demands. Additionally, the aerodynamic noise produced by VAWTs poses a challenge, especially when they are installed near residential areas. This research uniquely investigates how the interaction of wakes from adjacent rotors, coupled with a deflector, influences both the aerodynamic performance and noise levels of dual Savonius rotors. In the initial phase of the study, a Design of Experiment (DoE) methodology, based on the Taguchi method and Analysis of Variance (ANOVA), is utilized to optimize five design variables with three levels each for the deflector system. The aim is to enhance the power output of dual rotors. Two-dimensional Unsteady Reynolds-Averaged Navier-Stokes (URANS) simulations reveal that the distance between adjacent rotors (W) is the most critical factor, contributing 75% to performance improvement. The maximum average power coefficient (CP) of dual rotors equipped with the optimized deflector is 0.318, representing a 33.1% increase compared to the peak CP of a rotor without a deflector at a Tip Speed Ratio (TSR) of 0.8. Furthermore, the optimized deflector design improved the average CP of the dual rotors by 17.4% at TSR values of 1.2, compared to configuration without a deflector. In the final phase, three-dimensional Large Eddy Simulation (LES) is employed to capture detailed turbulent wind flow interactions with the turbines. An advanced optimization framework integrating Machine Learning (ML) with Computational Fluid Dynamics (CFD) is developed to enhance rotor performance. A single-objective approach focuses on maximizing power efficiency, while a multi-objective optimization method aims to maximize the average power coefficient and minimize noise generation. This framework incorporates wake interactions and an innovative deflector system to improve both aerodynamic and acoustic performance. The influence of geometric parameters on aerodynamic and aeroacoustics characteristics is also analysed using a DoE approach to build a comprehensive database. Subsequently, the CFD model is replaced with an Artificial Neural Network (ANN) model to predict rotor performance. A Multi-Objective Genetic Algorithm (MOGA) is applied to optimize the aerodynamic and acoustic characteristics of the rotors. Optimal design parameters are identified from the Pareto front using the TOPSIS decision-making method. The ANN model exhibits high accuracy, with coefficients of determination (R2 of 0.99 for average CP and 0.97 for Overall Sound Pressure Level (OSPL) predictions. The multi-objective optimization reveals an optimal deflector design with bleed jets, achieving a 59.2% improvement in average CP and a 5.2% reduction in OSPL compared to the one of the rotors in dual rotor configuration at a TSR of 0.8.
| Date of Award | 25 Jul 2025 |
|---|
| Original language | English |
|---|
| Supervisor | Rakesh Mishra (Main Supervisor) |
|---|