Automated Marine Propeller Optimal Design Combining Hydrodynamics Models and Neural Networks

In the present paper, a computationally efficient methodology to develop fast and reliable propeller selection procedures based on a fully automated optimization technique is described. To this aim, a comprehensive propeller hydrodynamics model is combined with performance prediction acceleration techniques based on Neural Networks. Under given operating conditions, screw characteristics and blade shape details are optimized around a baseline configuration via general-purpose numerical optimization software based on genetic algorithms and via a parametric model. Numerical applications concern the propulsion retrofitting of marine vessels. A off-design performance verification study is presented to evaluate the robustness of the identified optimal configurations.