Unveiling Hidden Correlations in Bio-Sourced Fiber Composites: A Principal Component Analysis Approach for Mechanical Property Optimization

Natural fiber composites have attracted increasing attention in recent years due to their potential as sustainable alternatives to synthetic composites. These plant-based materials, such as hemp fibers, provide low density, biodegradability, and superior mechanical properties, making them ideal for various engineering applications. However, their potential has not been fully exploited. This study applies principal component analysis (PCA) to identify patterns in four fiber composites: epoxy (Epo), high-density polyethylene (HDPE), polyester (PE), and polyurethane (PU). PCA revealed significant trends, with a total variance of 75%, which increased to 96% upon subcategorization of composites. The inverse correlation between filler weight percentage (wt.%) and flexural modulus (EF) was observed, emphasizing the impact of matrix integrity and interfacial bonding on mechanical performance. Additionally, removing outliers improved data clustering and enhanced the representativity of PCA analysis. By refining the data set, the study demonstrated a direct correlation between tensile modulus (ET) and tensile strength (σT), reinforcing PCA’s effectiveness in extracting meaningful relationships between compositional and mechanical features of composites. To increase PCA variance, strategies such as outlier removal and data set splitting into subcategories were employed. Removing outliers increased total variance by nearly 5%, leading to higher cluster distribution. These findings not only provide deeper insights into the mechanical behavior of natural fiber composites but also establish a robust statistical framework for optimizing their composition and performance in structural applications. Ultimately, this work highlights the power of PCA as an analytical tool to enhance the understanding of composite material behavior, paving the way for improved material design and industrial applications.