Deep learning-based prediction of mechanical properties in MWCNT-modified epoxy adhesives and co-cured CFRP composites joints
N. Karthikeyan, Jesuarockiam Naveen, Anandhan Venugopal, P. Edwin Sudhagar, M. Rajesh
Abstract
Epoxy has been effectively used in structural applications as both an adhesive and a matrix. The present research investigates the tensile and flexural properties of an epoxy adhesive incorporating distinct weight proportions of multi-walled carbon nanotubes (MWCNTs) (0.25, 0.5, 0.75, 1.0, 1.25, and 1.5%) and their influence on the shear behaviour of CF (carbon fibre)/Polymer composite co-cured joints. The results demonstrate that the epoxy adhesive incorporated with MWCNT exhibits significantly improved tensile and flexural strengths. The tensile and flexural strengths of epoxy adhesive modified with 0.75 wt.% MWCNTs are 44.8 MPa and 76.95 MPa, which are 21% and 22% higher than the neat epoxy adhesive samples. Moreover, compared to the neat epoxy adhesive, the tensile and flexural moduli are 1470 MPa and 4070 MPa, which are 9% and 63% higher, respectively. Moreover, this research found that a CF/Polymer composite co-cured joint with 0.75 wt.% MWCNT-incorporated epoxy adhesive had the greatest shear strength of 14.7, which is 21% higher than the neat epoxy-based CF/Polymer composite co-cured joints, and 91% and 102% higher than the neat secondary and co-bonded joints, respectively. Furthermore, the material underwent FTIR and FESEM analyses, confirming excellent interfacial bonding behaviours and failure mechanisms. The ANOVA statistical tool was employed to evaluate the hypothesis based on the experimental results, which confirmed that the p-values were less than 0.05. Moreover, the artificial neural network accurately forecasts the mechanical properties of MWCNT-modified samples and CF/Polymer co-cured composite joints through a sophisticated deep-learning approach, with an error of less than 8%.