Surrogate hydrodynamic modelling for wave energy system control co-design

Optimisation and control co-design, for wave energy systems, is receiving increased focus, in the drive to design commercially viable wave energy systems. A consistent bottleneck in wave energy converter (WEC) system optimisation is the need to iteratively calculate hydrodynamic forces, which is computationally time-consuming. One solution, to improve the computational efficiency of WEC optimisation loops, is through the use of surrogate models, which are computationally efficient approximations of the fundamental calculations provided by linear potential-flow boundary-element solvers. Surrogate models offer a computationally efficient framework for conducting iterative design optimisation provided sampling of the search space, when collating training data at the pre-computation stage, is well designed. Surrogate models can be used both to evaluate hydrodynamic response parameters and to calculate objective function values (e.g., captured energy or cost of energy), within design optimisation loops. This study explores both options and examines the computational time/fidelity trade-off involved in the use of surrogate models for WEC geometry optimisation. • Surrogate models emulate hydrodynamics for efficient WEC geometry control co-design. • A single surrogate linear hydrodynamics model is applied to three WEC CCD scenarios. • Careful design space sampling for training data reduces BEM pre-computation times. • Artificial neural network interpolation accuracy is similar to cubic polynomials.