Numerical Analysis of Flow-Induced Vibrations for Energy Harvesting in Multi-Cylinder Arrays

Efficient hydrokinetic energy harvesting in low-velocity aquatic environments remains a significant challenge. This thesis presents a concept based on flow-induced vibrations (FIV), in which a fixed large upstream cylinder acts as a vortex generator to induce wake-induced vibrations (WIV) in tandem downstream cylinders. The core idea is to exploit WIV through geometric tuning of multi-cylinder arrays. High-fidelity computational fluid dynamics (CFD) simulations were used to analyze the influence of diameter ratios, inline spacing, reduced velocities, and damping ratios, revealing that specific conditions induce synchronized vortex shedding and amplify downstream vibrations, resulting in a total power conversion efficiency of up to 54\% across both downstream cylinders. Moreover, a machine learning (AutoML) framework was developed. This data-driven approach identifies optimal configurations that achieve total efficiencies of up to 59\%, while significantly reducing computational costs. The concept was further extended to side-by-side cylinder arrays, where appropriate lateral spacing is shown to reduce wake interference and maintain stable energy extraction across multiple rows. Overall, this thesis advances hydrokinetic energy harvesting by introducing a scalable design framework that uses passive upstream cylinders to synchronize vortex shedding, amplify vibrations, and enhance energy extraction. By integrating physics-based models and machine learning, it provides a reliable and efficient procedure for the design of robust, efficient, and adaptable multi-cylinder systems.