Data-driven modelling of wave–structure interaction for a moored floating structure

We present a data-driven nonlinear model for predicting the motions and loads of a moored floating structure in waves, which is a challenging problem in offshore hydrodynamics that requires coupled computation of nonlinear wave–structure interactions and mooring dynamics. A high-fidelity viscous-flow solver, based on solving the Navier–Stokes equations for the free surface flow and implicitly coupled resolution of rigid body nonlinear motions and mooring dynamics, is employed to generate the ground truth. The data-driven model is trained using wave elevations as inputs and computed hydrodynamic motions and loads as outputs. We assess the effectiveness of the data-driven framework for modelling wave–structure interactions in both regular and irregular waves, demonstrating accurate predictions of hydrodynamic motions and loads in nonlinear waves with varying wavelengths and steepness. Leveraging the long short-term memory network, our surrogate model offers substantial computational savings for complex physical models and has the potential to create a digital twin of real offshore structures, ensuring operability and safety in various nonlinear sea states.