Artificial neural network correction for downwind sail simulations based on experimental results

Abstract

A method for reducing errors of downwind sail simulations is presented. This method can be used to improve the results of any simulation making use of Computational Fluid Dynamics (CFD) models and is particularly useful when simplified CFD models are considered, as in these cases the error of the results achieved can be large. In the present approach, in order to reduce the error the results are corrected by means of the application of neural network based approximations. This correction is carried out by using the results of wind tunnel experiments on downwind sails and by comparing them to numerical ones. To ensure than the sail shape used in both the CFD and the wind tunnel analysis are the same, the shapes achieved under real test conditions have been captured by means of a laser scanner.