Mechanical properties of 3D-Printed aluminum, nickel, and titanium using a hybrid machine learning and computational mechanics approach

Abstract

This study presents a hybrid computational framework designed to accurately predict the mechanical properties of essential 3D printing materials, namely, Aluminum (Al), Titanium (Ti), and Nickel (Ni). By integrating first-principles simulations via the CASTEP code—grounded in Density Functional Theory (DFT)—with machine learning techniques, specifically Ridge regression, the approach aims to enhance prediction accuracy while minimizing computational costs. The analysis focuses on key elastic properties, including Bulk Modulus, Young’s Modulus, and Shear Modulus. Initial simulations using CASTEP provide benchmark mechanical values, which are subsequently used to train and validate the Ridge regression model. The results reveal outstanding predictive accuracy, with R2 values surpassing 0.999 across all properties and minimal mean squared errors. A close correlation between DFT-derived and AI-predicted values confirms the robustness of the approach. This methodology significantly reduces reliance on physical experimentation and heavy simulations, making it a powerful tool for material design and optimization. Moreover, the findings emphasize Aluminum’s potential for lightweight structures, Titanium’s superior stiffness suited for biomedical and aerospace applications, and Nickel’s strong resistance to compression, making it ideal for demanding industrial settings. Such insights contribute to faster and more efficient materials selection and customization in additive manufacturing.