Probabilistic surrogate-based optimization of ship hull-propulsor design with bi-level infill sampling technique

Abstract

This work develops a new multi-level hull-propulsor optimization tool, systematically considering the full range of expected operating conditions. Different solvers with variable fidelities are used to evaluate hydrodynamic objective functions and constraints. The bi-fidelity surrogate model is applied to integrate the accuracy advantage of a medium-fidelity solver with the efficiency advantage of a low-fidelity solver. A bi-level technique of infill sampling integrated with a new version of multi-objective evolutionary algorithms is presented to enhance the effectiveness of the surrogate models. A ship velocity- and sea state-based joint probability density function is employed to take into account the real lifetime operational space. The well-known S175 containership, KP505 propeller and a MAN B&W marine engine are utilized as the initial hull-propulsor model. The final results show that the optimization framework is able to achieve some optimum hull-propulsor designs from the lifetime fuel consumption and energy efficiency design index points of view and to reduce the overall lifetime ownership cost significantly.