This scientific research investigates the multi-objective drilling optimization of a novel flax/kenaf/epoxy biocomposites reinforced with abrasive magnesium hydroxide (Mg(OH) 2 ) fillers, employing a comparative statistical and machine learning framework to bridge the gap between predictive modeling and physical tool-wear mechanisms. A Face-Centered Central Composite Design (FCCCD) evaluated the effects of spindle speed, feed rate, and tool material (HSS, M35, M42) on delamination, surface roughness, and circularity error. Models were developed using Response Surface Methodology (RSM) and a Bayesian-regularized Artificial Neural Network (ANN). The results quantitatively identify feed rate as the dominant parameter, explaining over 86.5% of the variance in surface roughness. Furthermore, the M42 cobalt-alloyed drill successfully resisted abrasive filler adhesion, reducing maximum delamination by nearly 30% compared to baseline HSS tools. From a modeling perspective, the ANN demonstrated superior capability in capturing non-linear dynamic machining behavior, outperforming RSM prediction accuracy by 12.3% for circularity error and 5.7% for delamination. Desirability-based multi-objective optimization established a highly robust industrial process window (1296 rpm, 0.20 mm/rev, M42 tool) that simultaneously minimized all defects. This experimentally validated framework provides actionable guidelines for the high-quality, sustainable machining of mineral-filled hybrid biocomposites in structural automotive and aerospace assemblies.
Kalphana et al. (Sun,) studied this question.