PREDICTION OF DELAMINATION FACTOR IN DRILLING GLASS FIBER REINFORCED EPOXY PLASTICS USING NEURAL NETWORKS

Drilling of holes in fiber reinforced plastics (FRPs) becomes almost unavoidable in order to facilitate joining of parts. The drilling induced damage in FRPs is an area of paramount concern as the delaminated holes act as areas of stress and lead to reduced life and efficiency of parts. The present research initiative is to study the delamination produced in drilling of unidirectional and [(0/90)/0]s glass fiber reinforced epoxy laminates (GFREP). A Carbide Jodrill of two different diameters has been used at three different levels of speeds and feed rates. A predictive model based upon artificial neural networks (ANN) has been developed to predict delamination factor. The results reveal that artificial neural networks can be successfully applied to predict delamination at a given speed and feed for a particular GFREP laminate. In normal cases the predicted values are in close agreement with the experimental values. The mean percentage error in training and test data sets is found to be 1.1 % and 2.25%.