The study investigates the impact of water absorption on local materials and natural resources-based composites of jute/epoxy (J/epoxy) and kenaf/epoxy (K/epoxy) using Artificial Neural Networks (ANN) and Response Surface Methodology (RSM). The research focused on short fiber laminates with fiber lengths of 4 mm, 8 mm, and 12 mm to evaluate water uptake behavior. It was observed that kenaf-based composites reached saturation after 25 days of immersion. The kinetic characteristics of moisture absorption were analyzed and compared with predicted outcomes from the developed models. Among the samples, kenaf composites reinforced with 4 mm fibers showed reduced shrinkage compared to those with 8 mm and 12 mm fibers. Water absorption rates were recorded at 0.88 % for JL5 and 2.71 % for KL5, which were significantly lower than those of JL10, JL15, KL10, and KL15, which exhibited absorption rates of 1.13 %, 3.75 %, 1.74 %, and 4.66 %, respectively. Optimal performance, with a desirability value of 1.000, was achieved after 482 h of immersion using fibers of 11.73 mm length. ANN models demonstrated superior predictive performance compared to RSM, with RSM showing higher prediction errors. ANN's high accuracy in modeling water absorption behavior offers a valuable tool for reducing the need for extensive laboratory testing. The findings support the use of jute and kenaf fiber composites as eco-friendly materials and provide engineers with a cost-effective method for estimating swelling behavior without relying solely on physical experimentation.
Arunachalam et al. (Fri,) studied this question.