Design and Optimization of Sustainable Green Composites for High-Performance Applications

Abstract

This study presents a data-driven framework for the design and optimization of next-generation sustainable green composites that are aimed at high-performance industrial applications. A hybrid dataset that comprises 180 experimental records of natural fiber-reinforced biopolymer composites was analyzed using Machine Learning (ML) algorithms, including Random Forest Regression (R² = 0.962), Artificial Neural Network (R² = 0.948), and Support Vector Regression (R² = 0.921). Feature importance analysis identified fiber volume fraction (38.5%), filler type (24.7%), and matrix viscosity (18.9%) as the most influential variables that govern tensile strength and biodegradability. Multi-objective optimization with the application of NSGA-II achieved a tensile strength of 127 MPa and biodegradability of 73%, which represent a 19.6% increase in mechanical performance and a 42% improvement in environmental compatibility when compared to conventional composites. Life-cycle assessment revealed significant sustainability advantages: embodied energy reduced by 33.8% (from 68 MJ/kg to 45 MJ/kg), carbon footprint lowered by 52% (from 2.5 kg CO₂-eq/kg to 1.2 kg CO₂-eq/kg), and end-of-life recyclability enhanced from 42% to 78%. Furthermore, the optimized composite achieved a processing temperature reduction of 21.4% and a 20.5% lower material cost. These results confirm that the integration of ML-driven prediction and optimization with green composite fabrication can accelerate sustainable materials development, reduce resource waste by up to 60%, and provide a replicable model for digital twin-assisted design. The proposed framework demonstrates clear potential for adoption in automotive, aerospace, and packaging sectors, where lightweight, recyclability, and environmental performance are critical.