Investigating moisture resistance of castor oil cortex fiber-reinforced epoxy composites through chemical treatments: Experimental characterization and prediction by machine learning

Abstract

The susceptibility of natural fiber composites to moisture-induced degradation poses a significant challenge for their application in humid and marine environments. This study investigates the effectiveness of chemical treatments in mitigating this issue for castor oil cortex fiber-reinforced epoxy composites. Fibers were treated with NaOH (1-10%) and KMnO₄ (0.1-0.5%) solutions, and unidirectional composites with a 40% fiber volume fraction were fabricated. Both untreated and treated samples were subjected to moisture absorption studies through immersion in distilled water and seawater for up to 60 days, followed by tensile, flexural, and impact testing. The 5% NaOH treatment (NaOH-2) proved most effective, enhancing the dry tensile, flexural, and impact strengths by 28%, 34%, and 115%, respectively, and demonstrating superior property retention after aging. A machine learning framework was developed to predict mechanical properties based on processing and exposure parameters. Gradient Boosting excelled in predicting flexural strength (Test R² = 0.98), Random Forest was optimal for impact strength (Test R² = 0.99), and Linear Regression performed best for tensile strength (Test R² = 0.92). The analysis identified immersion time as the most critical factor influencing property degradation. This integrated experimental and computational approach provides a robust tool for the design and lifetime prediction of durable natural fiber composites.

Graphical Abstract