Prediction of lap shear strength of GNP and TiO2/epoxy nanocomposite adhesives

Abstract

In this study, graphene nanoplatelets (GNPs) and titanium dioxide nanofillers were added to epoxy resin P-5005 at five different weight percentages (wt%), viz., 1, 5, 10, 15, and 20 wt%. The tensile properties of the nanocomposites were experimentally tested following ASTM D638-14. Then, the above-mentioned nanocomposites were applied as adhesives for an overlap joint of two A5055 aluminum sheets. The apparent shear strength behavior of joints was tested following ASTM D1002-01. Moreover, experimentally obtained results were applied to train and test machine learning and deep learning models, i.e., adaptive neuro-fuzzy inference system, support vector machine, multiple linear regression, and artificial neural network (ANN). The peak tensile strength (TS) and joint failure load (FL) values were observed in epoxy/GNP samples. The ANN model exhibited the least error in predicting the TS and FL of the considered nanocomposites. The epoxy/GNP nanocomposites exhibited the highest TS of 28.49 MPa at 1 wt%, and the peak overlap joints exhibited an FL of 3.69 kN at 15 wt%.

Keywords

• graphene nanoplatelets
• titanium dioxide
• mechanical characteristics
• machine learning