Optimized Wear Behavior of LM13/ Zircon/Carbon Hybrid Composites: Experimental Analysis, Statistical Modeling, and Machine Learning Predictions

This Paper investigates the wear behaviour of Aluminum-alloy (LM13) specimens cast under varying chill levels (12%, 9%, 6%, 3%, and as-cast) and quantitatively evaluates their tribological response through experimental, regression modelling and statistical analysis. Wear tests conducted at multiple rotational speeds showed a consistent increase in wear rate with RPM. Polynomial regression produced the highest predictive accuracy (R² = 0.9524–1.000), outperforming linear, Random Forest, and SVR models. Error metrics remained extremely low for all chill levels (RMSE = 0.000019–0.000047 g; MAE = 0.000014–0.000044 g), confirming the stability of the datasets. Shapiro–Wilk results (p > 0.314) and QQ-plots validated the normality of residuals, supporting the reliability of the statistical models. Microstructural analysis revealed that higher chill levels generated finer dendritic arm spacing, reduced porosity, and enhanced phase distribution uniformity. These features contributed to lower wear rates and reduced sensitivity to RPM, as reflected by minimal slope values (≈ 1.0×10⁻⁶ for 12% chill). In contrast, lower chill and as-cast samples exhibited coarser dendrites and weaker matrix integrity, resulting in higher wear and increased rate sensitivity. Overall, the study establishes that chill-induced microstructural refinement significantly enhances wear resistance and stabilizes the wear–speed relationship, providing a strong quantitative basis for optimizing casting parameters in tribological applications.