Investigation of the Impact of Alloying Elements on the Mechanical Properties of Superalloys Using Explainable Artificial Intelligence

OpenAlex İngilizce

This study investigates the impact of alloying elements on the mechanical properties of Ni-Cr-Fe-based superalloys using a computational materials science approach. Fifty different compositions of superalloys, commonly known as Inconel and Incoloy, were modeled using the JMatPro software. The mechanical properties, including 0.2% proof stress (MPa), fracture stress (MPa), and Young’s modulus (GPa), were simulated across a temperature range from 540°C to 920°C at 20°C intervals for each alloy. The simulation generated a comprehensive dataset comprising 1000 rows. This dataset was then utilized to train an explainable artificial intelligence (XAI) model, leveraging advanced techniques such as SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model-agnostic Explanations), and Partial Dependence Plots (PDP). The dataset was analyzed using an XAI-based regression model employing the XGBoost algorithm. The interpretability graphs were analyzed to evaluate the individual contributions of each alloying element to the mechanical properties over the entire temperature range. The findings provide detailed insights into the positive or negative effects of alloying elements, enabling a better understanding of their role in optimizing superalloy performance under various thermal conditions. This work highlights the potential of integrating computational materials modeling and explainable AI to advance the design and development of high-performance materials.