Data-Driven Approaches to Understanding the Flow-Induced Vibrations of Elliptical Cylinders

This thesis characterises, predicts, and subsequently controls the flow-induced vibration (FIV) of a freely oscillating thin elliptical cylinder, whose hydrodynamically unstable nature has potential applications for renewable energy harvesting from water flows. The work shows that a thin elliptical cylinder can achieve a factor-of-ten improvement in power compared to its circular cylinder counterpart, with vibrations highly susceptible to the cylinder’s orientation angle. It is also found that machine learning algorithms can effectively predict the FIV behaviour and identify the control strategies that maximise the vibration.