Design optimization of inclined pile permeable breakwater based on a CFD-SVR-WO framework

Abstract

The design and optimization of breakwaters is a crucial yet challenging problem in port and coastal engineering, as evaluating their wave dissipation characteristics is usually associated with high experimental or computational expenses. To achieve rapid and comprehensive optimization of an inclined pile permeable (IPP) breakwater, this study proposed a surrogate model -based optimization framework integrating computational fluid dynamics (CFD), support vector regression (SVR), and walrus optimizer (WO). Firstly, a three-dimensional numerical wave tank was established to evaluate the wave dissipation capacity of the breakwater. The accuracy of the model was validated through model tests. Then, a surrogate model was trained using SVR based on the data obtained from 103 high-fidelity CFD simulations. The hydrodynamic properties of the IPP breakwater under different wave periods were analyzed, showing that the IPP breakwater has better wave dissipation capability for short-period waves. Finally, an optimized design was given by applying WO under constraints of minimum permeability at 0.22 and maximum volume per unit length at 36 m². The proposed CFD-SVR-WO framework demonstrated a significant reduction in computational cost, validating its feasibility in handling complex optimizations for engineering structures.