Wave energy converter "wave carpet" is an artificial flexible structure originally introduced and actively developed by the Theoretical and Applied Fluid Dynamics Laboratory (TAFLab) for nearly a decade. Wave carpet is submerged on shallow-water seabed and absorbs the energy of incoming surface waves. Several studies conducted by TAFLab members numerically and experimentally proved its concept and supported its technical feasibility.
Inheriting from the previous works, this thesis further extends the understanding of wave carpet by numerically investigating one of its possible configurations and heuristically optimizing its planar shape. The study concentrates on arbitrary-shaped wave carpet locally embedded on the seafloor. This type of wave carpet is simplified into a visco-elastic part of the surrounding rigid seabed. In order to prepare the simulation environment, we first solve the problem of wave-viscoelastic seabed interaction using the High-Order Spectral (HOS) method. As a result, the solution presents the propagation of periodic waves over the entire visco-elastic ocean bottom. Wave carpet is subsequently limited to a partial seafloor, and the simulation domain is repetitively wave-fed and marched over time by Runge-Kutta method. Eventually, the procedure results in a robust numerical model of a wave tank with a wave carpet of an arbitrary shape. The model is validated against several criteria to ensure its correctness. It is referred to as the HORSK model that resembles a wave tank.
The HOSRK wave tank is numerically stable, three-dimensional and possibly highly nonlinear. It serves as a computational environment simulating the real-world scenarios in which incoming waves propagate over a wave carpet and damp out. Through many experiments with the numerical wave tank, we find the optimal water deepness and the optimal restoring force and damping ratio, two main characteristics of the wave carpet, for the maximum energy absorption. We further investigate a wide range of linear and nonlinear elliptical wave carpets, and we find that toward the optimal point the difference between linear and nonlinear carpets in terms of energy absorption is subtle.
To optimize the wave carpet shape, we propose the Neural Network-based optimization method for linear optimization and the Genetic Algorithm for nonlinear optimization. The results show that in most wave cases the optimal-shaped carpets can absorb approximately twice more energy than the baseline circular shape, and the nonlinear optimal shapes are very similar to the linear optimal ones.
