Shape Optimization of a Horizontal Axis Tidal Turbine (HATT) Blade Using Neural Networks for Response Surface Methodology

Abstract

Ocean energy is gaining popularity with tidal energy being the most mature technology. Presently, most of the tidal energy research is being focused on tidal stream turbines because it is cheaper and less environmentally invasive. However, this technology is currently considered highly site-specific due to the current tidal turbines being designed for high flow velocity areas. Countries in the tropical region (e.g., the Philippines) are characterized as having low-velocity tidal currents. To efficiently exploit the energy from the ocean tides for relatively low velocity flows, a site-specific Horizontal Axis Tidal Turbine (HATT) blade must be designed and optimized. In this study, a novel robust design optimization framework is used to optimize the blade. This study uses Artificial Neural Networks (ANN) for response surface methodology to generate a surrogate model while using Particle Swarm Optimization (PSO) as the meta-heuristic algorithm to find the optimized blade. The results indicate that the optimized blade increases the turbine's maximum power coefficient and Annual Energy Production (AEP) while limiting the bending stress and cavitation number. The maximum power coefficient of the blade is increased by 30.30%. The maximum bending stress of the blade is decreased by 6.62 % while remaining cavitation-free. One key feature of this method is its robustness as it can be applied to any site located within the design space of interest. Overall, this could be viable for a more computationally efficient method to optimize the performance of HATTs for any set of conditions.