Predictive Simulation of Mechanical Properties and Iron-Carbon Phase Diagram Region for Steel Using Materials Informatics

This study presents a materials informatics framework for predicting the mechanical properties of medium carbon steels and linking these predictions with phase stability information from the Fe–C phase diagram. A dataset comprising 413 steel samples within the composition ranges specified for medium carbon steels was analyzed using three machine learning models: Random Forest, Artificial Neural Network (ANN), and Extreme Gradient Boosting (XGBoost). Model performance was evaluated using standard regression metrics, with XGBoost demonstrating the best predictive accuracy (R² ≈ 0.97) for tensile strength and hardness prediction. To enhance interpretability, SHAP (Shapley Additive Explanations) analysis was employed to quantify the influence of carbon content, alloying elements, and heat treatment processing parameters on predicted properties. The results show that carbon and key alloying elements significantly influence strength through mechanisms consistent with classical steel metallurgy, including solid-solution strengthening and carbide formation. The predictive results were further contextualized using Fe–C phase diagram region visualization to relate composition to expected phase stability. In addition, a streamlit-based interactive tool was developed to enable real-time prediction of mechanical properties and visualization of phase diagram regions for candidate steel compositions. The proposed framework demonstrates the potential of integrating machine learning with metallurgical principles and phase diagram analysis to support data-driven alloy design, enabling more efficient prediction and optimization of mechanical properties in alloyed steel systems.