Investigation on the Mechanical Behavior of Date Palm Fibers Reinforced Composites: Predictive Modelling Using Artificial Neural Networks (ANNs)

Abstract

This paper aims to strengthen composites by treated and untreated date palm fibers (PDF), with sodium hydroxide (NaOH), for light applications. With 75% cellulose content and a density of 1.2 g/cm3, the palm fibers were exposed to a preparatory treatment with 1.5% NaOH for 24 h prior to integration into a polyester. Four polyester samples comprising 30% of palm fiber were manufactured. Additionally, the palm fiber interface was evaluated using scanning electron microscopy (SEM) and optical microscopy. The specimens underwent mechanical testing and it shows that tensile (18% increase in stress and 1.2% increase in Young’s modulus) and flexural properties (20% increase in strength and 10% increase in Young’s modulus) of treated composites as compared with untreated fibers. A MATLAB-based Artificial Neural Network (ANN) model was applied to estimate stress and strain at break as well as the Young’s modulus, based on three input characteristics: section, sample length, and chemical treatment. It was obtained that the polyester reinforced by NaOH-treated palm fibers increased the mechanical characteristics relative to the untreated fibers. The coefficient of determination R2 in the ANN models is 0.87. These results suggest that the ANN model is a useful tool for predicting mechanical properties.