Architecture-Specific Performance of Silica and MWCNT Modified Hybrid Epoxy Laminates Under Mechanical Loading and Fracture.
Bashir Muhammad Tariq MT, Naseem Khawaja Atif KA, Khan Atif A, Sadiq Muhammad Rehan MR et al.
Hybrid fiber-reinforced epoxy laminates integrate macroscale fiber architecture with nanoscale matrix modification. The effects of nanofillers cannot be isolated from laminate architecture when both variables are altered simultaneously. This study independently evaluated silica-modified Kevlar/basalt/alkali-resistant glass (K/B/ARG), silica-modified Kevlar/carbon/alkali-resistant glass (K/C/ARG), and multi-walled carbon nanotube (MWCNT)-modified Kevlar/glass/carbon (K/G/C) laminates. Each laminate family incorporated 0, 1.5, 3.0, 4.5, or 6.0 wt.% nanofiller. Mechanical performance was assessed using tensile, flexural, interlaminar-shear, quasi-static puncture, punch-shear, impact, and energy-absorption tests. Scanning electron microscopy (SEM) was employed to examine fracture morphology, and selected K/C/ARG laminates underwent thermogravimetric analysis (TGA/DTG). For each architecture, 4.5 wt.% nanofiller yielded the highest mean values for most measured responses. At this concentration, K/B/ARG achieved 167.93 ± 5.71 MPa tensile strength, 112.09 ± 3.70 MPa puncture strength, and 1598.1 ± 73.5 J/m Izod impact strength. K/C/ARG achieved 171.95 ± 5.50 MPa tensile strength, 7710 ± 285.3 N puncture load, and 1813.6 ± 79.8 J/m Izod impact strength. K/G/C achieved 215.58 ± 7.76 MPa tensile strength, 28.49 ± 1.20 MPa interlaminar-shear strength, 110.80 ± 5.54 MPa punch-shear strength, and 15.45 ± 0.77 J net energy absorption. Several matrix-sensitive responses declined at 6.0 wt.%, while K/G/C flexural strength remained nearly constant. SEM revealed greater matrix retention and more uniform damage distribution at intermediate nanofiller loading but clustered weak regions at 6.0 wt.%. Single-run TGA indicated a higher principal DTG temperature and greater residue for the 4.5 wt.% K/C/ARG formulation; however, replication is necessary. The convolutional neural network (CNN) analysis was constrained by the limited number of formulation levels, target leakage, and the absence of independent validation. Consequently, 4.5 wt.% is identified as the best-performing tested concentration within each architecture, rather than a universal optimum. Additional durability, fire, and structural-scale testing is required before these laminates can be considered for building design, resilient infrastructure, or sustainable construction.