Design Optimization of a Fluidic Diode for a Wave Energy Converter via Artificial Intelligence-Based Technique

A pair of turbines can harness power from a wave energy Converter. Their performance is poor than individual turbines due to flow reversal. A fluidic diode (FD) which offers variable resistance to the flow, can be used to prevent flow reversal and improve the performance of these units. Its performance is given by diodicity (ratio of reverse to forward flow pressure drop). A higher diodicity enables it to prevent flow reversal better and improve the turbine unit’s overall efficiency. In this work, the geometrical shape of the FD is optimized to obtain higher diodicity. Six geometrical variables of the FD are varied to obtain sample points using the sampling technique, which is numerically investigated by solving steady-state Reynolds averaged Navier–Stokes (RANS) equations. These numerical results were fed into a neural network code that produced an optimal FD design. The optimum model showed a 36.5% improvement in diodicity at 0.35 m3/s. The fluid flowing through the optimized model experience higher resistance in the reverse direction because of the increased vortex strength than the base model. Among all the design variable considered, nozzle angle is a highly sensitive parameter in the optimization process. The optimum FD model enhanced the overall efficiency of the turbine unit by 13.3.