Enhancing quasi-static mechanical properties with bio-inspired optimization and experimental assessment of composite t-joint

Abstract

Composite T-joints consist of the skin, stiffener and flanges made of carbon fiber/epoxy laminate, which are joined by adhesive bonding. This study presents a bio-inspired optimization method to improve the quasi-static mechanical performance of carbon fiber epoxy composite T-joints. Three configurations are investigated: Design A (baseline conventional stacking sequence [45°/0°/−45°/90°/90°/−45°/0°/45°]), Design B (bio-inspired optimized sequence [45°/0°/−45°/90°/45°/−45°/0°/90°] derived from tree branch-trunk morphology) and Design C (bio-inspired design incorporating a PVC foam core within the delta-fillet region). A hybrid structures combining Artificial Neural Network (ANN) and Multi-Objective Genetic Algorithm (MOGA) structure is developed to optimize the geometric parameters inspired by tree branch-trunk joints and the stiffener ply stacking sequence. Finite Element Analysis (FEA) identified stress distributions, and the ANN established high predictive accuracy with a mean absolute error of 0.311%. The optimized (Design B) [45°/0°/-45°/90°/45°/-45°/0°/90°] reduced tensile stress by 33.62% compared to Design A. Experimental validations showed that the bio-inspired design improved bending, tensile and compressive strength by 15.7%, 16.2% and 16.49%, respectively compared to the baseline. (Design C) presented even greater enhancements, with strength improvements of up to 24.9% over Design A and 7.44% over Design B under bending loading. FEA results additionally confirmed significant decreases in failure indices, up to 66.29% under compressive loadings for Design C. This work establishes an effective bio-inspired optimization methodology for designing high-performance, damage-resistant composite joints.