Influence of alkaline and sodium bicarbonate treatments on the machinability of sisal jute epoxy hybrid biocomposites using RSM and ANN modeling

To overcome the critical challenge of machining-induced structural degradation that restricts the industrial adoption of eco-friendly materials, this study aims to improve biocomposite performance by enhancing machinability and overall functionality, thereby broadening their applicability across multiple industries. By using natural fibers and environmentally friendly treatments, the research advances the development of sustainable materials in the composite sector. Crucially, this work introduces a novel comparative evaluation of the drilling machinability of hybrid sisal-jute/epoxy biocomposites treated with alkaline (NaOH) and eco-friendly sodium bicarbonate (NaHCO 3 ) solutions was systematically investigated. Composites were fabricated with various stacking sequences of unidirectional sisal and bidirectional jute fibers, yielding three composite types: sisal jute/jute sisal (SJJS)/epoxy, SJJS-NaOH/epoxy, and SJJS-NaHCO 3 /epoxy. The experimental protocol evaluated three primary factors at three levels: drill diameter ( d , 4, 7, 10 mm), spindle speed ( N , 750, 1500, 3000 rev/min), and feed rate ( f , 50, 100, 200 mm/min). The core methodological novelty lies in the concurrent deployment and rigorous comparison of Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) to predict the delamination factor ( F d ) coupled with advanced genetic algorithm (GA) optimization tailored for chemically treated hybrid structures. Results indicate that NaHCO 3 treatment reduced delamination by up to 17.67% compared with untreated fibers. Specifically, the absolute delamination factor decreased from F d = 1.5542 for the raw composite to F d = 1.2795 for the treated specimens, providing a baseline for high-quality drilling. Optimization using an ANN/genetic algorithm and an RSM/desirability function identified optimal drilling conditions as d = 10 mm, N = 3000 rev/min, and f = 50 mm/min. In comparison to the RSM model, the ANN architecture of F d1 , F d2 , and F d3 demonstrated higher predictive accuracy in capturing complex non-linear structural behaviors, as evidenced by a lower mean squared error ( MSE = 0.1123, 0.1102, and 0.0653) and a higher coefficient of determination ( R 2 = 0.9918, 0.9912, and 0.9932). These outcomes underscore the potential of treated SJJS/epoxy biocomposites as sustainable alternatives for high-performance applications in the automotive and eco-construction sectors.