Abstract
Optimizing vertical-axis wind turbine (VAWT) pairs is vital for addressing future energy needs, given that their power and aerodynamics are notably affected by wake interactions. This study investigates the self-starting behavior of adjacent rotors, using dynamic Computational Fluid Dynamics (CFD) start-up models with variable angular velocities. A Taguchi-based Design of Experiment (DoE) methodology, coupled with Analysis of Variance (ANOVA), is employed to optimize design parameters for rotor pairs, enhancing self-starting characteristics. Results show that the angle between adjacent rotors (factor D) has the strongest effect on the downstream rotor’s start-up time, while the number of blades (factor B) has the least influence. In the optimized layout, the downstream rotor (rotor 2) self-started faster than the upstream rotor (rotor 1) at a lower angular velocity of 57 rad/s, demonstrating that rotor 2 can initiate self-start at a lower tip speed ratio than rotor 1. However, the start-up time of rotor 1 is also influenced by rotor 2, which accelerates later than it would in a single-rotor scenario. In the optimized layout, the downstream rotor (rotor 2) self-started faster than the upstream rotor (rotor 1) at a lower angular velocity of 57 rad/s, demonstrating that rotor 2 can initiate self-start at a lower tip speed ratio than rotor 1. However, the start-up time of rotor 1 is also influenced by rotor 2, which accelerates later than it would in a single-rotor scenario. In this optimized configuration, the wake of rotor 1 shifts downward, recovers more rapidly, and reduces in size by approximately 50% compared to a single-rotor layout. This allows additional rotors to be positioned in the same designated zones more efficiently compared to a conventional single-rotor setup, as the faster wake recovery increases momentum in the downstream flow. Consequently, downstream turbines could be placed closer together without significant aerodynamic losses, enhancing the overall efficiency and layout of the wind farm.Also, this research uniquely explores the effects of wake interactions between adjacent Darrieus wind turbines on their aerodynamic performance and noise emissions, a critical consideration for optimizing wind farm design and operation in proximity to populated areas. Additionally, it examines the vortex interactions between rotor blades and analyzes the dynamic stall phenomenon, offering valuable insights into unsteady aerodynamic behavior. By employing Large-Eddy Simulation (LES), the study captures complex turbulence patterns and unsteady rotor interactions, providing detailed insights into both aerodynamic performance and aeroacoustic behavior. A multi-objective evolutionary algorithm, coupled with machine learning and CFD, is then used to optimize rotor designs for maximized power output and minimized noise, explicitly accounting for wake interactions between adjacent turbines. The study assesses the impact of physical and geometric parameters on rotor performance by generating a high-fidelity database using a Design of Experiments approach based on CFD simulations. This database is then used to train an Artificial Neural Network (ANN) as a surrogate model for rapid performance prediction, thereby reducing computational cost while retaining CFD-based accuracy. The nondominated Sorting Genetic Algorithm II (NSGA-II) refines aerodynamic and aeroacoustics attributes, with optimal design parameters identified using the linear programing technique for multidimensional analysis of preference (LINMAP). The LINMAP-optimized rotor demonstrates superior aerodynamic and aeroacoustic performance compared to the Point O rotor. Its wider blade spacing improves airflow and increases the torque coefficient (CT), while the Point O rotor is adversely affected by intensified vortex interactions. In the downwind region, the LINMAP rotor sustains positive CT values, whereas the Point O rotor experiences negative torque, highlighting the effectiveness of the optimized design in maintaining consistent performance across the rotor field. Furthermore, compared to the baseline Point O rotor, the LINMAP-optimized rotor produces stronger low-frequency noise, whereas the Point O rotor exhibits higher sound pressure levels above 100 Hz.
Finally, this study develops a hybrid machine-learning framework that couples CFD and artificial neural networks to quantify and optimize the performance of paired Darrieus turbines in clean-air and particle-laden environments. Targeting arid, semi-arid, and coastal sites where airborne sand accelerates blade degradation, the framework resolves the influence of sand-induced surface roughness on turbine aerodynamics and power output. Clean-air CFD results are used to determine baseline optimal design parameters, which are subsequently tested under sand-particle conditions (0.5–2 mm) using LES to evaluate aerodynamic losses and erosion. A Single-Objective Genetic Algorithm (SOGA) identifies the turbine configuration that maximizes the average power coefficient (CP). Results show that larger sand particles significantly reduce performance, with a 4.8% decrease in CP at a tip speed ratio of 2.24. Regression analysis reveals a strong correlation between particle size and blade erosion rate. The results show that leading-edge erosion reduces the pressure difference across the blade, slightly increases suction-side pressure at θ=0°, weakens the suction peak at θ=75°, and decreases flow attachment at θ=150°. Lift is significantly lowered in the upwind region, while delayed flow reattachment and weaker lift recovery occur in the downwind region. Cumulative erosion analysis indicates that blades exposed to 0.5 mm particles exceed the 2 mm erosion limit in under two years, while exposure to 2 mm particles leads to critical wear within months, far below the 15-year design life.
Although the erosion analysis represents a focused component of the overall study, it provides practical insights for turbine durability, maintenance planning, and deployment in harsh environments, complementing the aerodynamic and aeroacoustic optimization work and demonstrating the versatility of the hybrid data-driven framework.
| Date of Award | 9 Apr 2026 |
|---|---|
| Original language | English |
| Supervisor | Rakesh Mishra (Main Supervisor) & Frankie Jackson (Co-Supervisor) |