An effective framework for wake predictions of tidal-current turbines

Abstract

Deployment of tidal array farms is the next stage of tidal energy development which concentrates on the capture of the enormous energy. Characterization into the wake of tidal turbine aids in determining the tidal farm layout and energy yields. This paper develops a framework that employs the Computational Fluid Dynamics (CFD) simulation and machine learning to analyze turbine wake with high accuracy and good efficiency. Multilayer perceptron neural network (MLP-NN) was introduced to establish the interrelation between the incoming flow conditions and the wake profiles. The Reynolds-averaged Navier-Stokes equations (RANS) are associated with a k − ε turbulence model to offer a number of datasets of wake profile for training, testing, and validation of the MLP-NN models. It was found that the MLP–NN–based model has achieved a considerably high level of accuracy by comparing it with empirical and numerical models. The reliability of the MLP-NN based model coupled with the wake combination RSS model to predict the resultant wake profile and power output of multiple turbines are also assessed. The techniques significantly enhance the efficiency and accuracy of wake predictions.