Simulation-based multi-objective optimisation of marine propellers using selective surrogate ensemble (SBD-SSE)

This study presents a simulation-based design method, SBD-SSE, that leverages a selective surrogate ensemble (SSE) for optimising marine propeller designs. By generating multiple surrogate models through bootstrap aggregating, the method adaptively selects a subset of models to evaluate key performance indicators (KPIs) during the optimisation process, improving the search direction. The approach is applied to the hydrodynamic shape optimisation of the KP505 propeller, where high-fidelity data from Computational Fluid Dynamics (CFD) is used. The performance of the traditional SBD method and the proposed SBD-SSE method are compared. Numerical experiments demonstrate that SBD-SSE performs competitively and offers advantages in accuracy. Statistical validation with the Wilcoxon signed-rank test confirms the effectiveness of incorporating SSE. Additionally, time efficiency analyses show that SBD-SSE is an effective and practical method for marine propeller optimisation.