An artificial intelligence approach to enhance blade element momentum theory performance for horizontal axis wind turbine application

Abstract

So far, the Blade Element Momentum (BEM) theory remains the most widely used method for predicting aerodynamic performance of horizontal axis wind turbines (HAWTs) owing to its simplicity. The BEM theory is mainly based on airfoils data for wide range of conditions (airfoil shape, angles of attack (AOAs)). These data are usually collected in wind tunnel experiments for stationary airfoils at low AOAs. However, a rotating wind turbine, has higher AOAs. The motivation behind this work is to improve the classical BEM method by determining the airfoil performance coefficients where little or no experimental data exists such as at high AOAs, new airfoils shape and for low Reynolds numbers. For this purpose, an artificial intelligence approach, namely Artificial Neural Networks (ANNs) is proposed for predicting the airfoils lift and drag coefficients. Firstly, the optimum number of layers as well as the optimum neurons number for training input-output data have beenselected numerically. Afterwards, the results of the proposed BEM-ANN method have beencompared with available experimental results in order to investigate its validity.Good agreementswereobtainedbetween numerical predictions and experimental results.