Optimization of drilling parameters to minimize delamination in CNT-filled GFRP composites using machine learning

Composites are most commonly fastened in assemblies by drilling. The current investigation examines the effect of the drilling factors on the quality of the drilled holes. The holes were drilled on epoxy resin polymer composites reinforced using glass fibers with Carbon nano tube (CNT) as fillers. Hand-layup was done to fabricate the composites. The laminated composites were produced with 0 %, 1 %, and 1.5 % of CNT fillers. Operating parameters such as spindle speeds-1000 rpm, 2000 rpm, and 3000 rpm, feed rates- 50 mm/min, 100 mm/min and 150 mm/min were used during the experiments. Torque (T) and thrust force (F) were measured using a digital drilling machine with a dynamometer. A machine learning based multi-output random forest regression model with hyper parameter tuning was used to predict the T, F, and delamination factor (F d ). The algorithm showed that the most important parameter that influenced delamination was speed (s) followed by the feed rate (f) and filler content respectively. Further, it predicted the thrust force and F d with ±5% accuracy and T with ±10% accuracy. The best combination of speed, feed, filler which would result in a minimized F d was arrived at with the help of a Bayesian optimization.