Optimizing Curved Bluff Bodies for Galloping Micro-Power Generators via Machine Learning

The emergence of flow micro-power generation has reignited the interest in researching galloping instability aiming to determine the shape of the bluff body that is most prone to galloping. In this work, we employ computational fluid dynamics in conjunction with machine learning to optimize the shape of the bluff body for energy harvesting in both steady-state and transient performance. We investigate a continuum shape that has straight frontal and dorsal faces with varying lengths, and side faces described by surfaces of different curvatures. The optimization study reveals that a curved-trapezoidal bluff body with the highest side surface curvature and the frontal-to-dorsal ratio is the perfect shape for steady flow conditions. On the other hand, a square profile with the highest side surface curvature is the ideal choice for highly fluctuating flow conditions because of its shortest rise time.