Machine Learning-Assisted Prediction of Thermophysical Properties of Nickel-Base Alloys over a Temperature Range




Mondal, Sudeepta
Menon, Nandana
Ray, Asok
Basak, Amrita

Journal Title

Journal ISSN

Volume Title



Calculation of thermophysical properties of complex metallic alloys as functions of composition and temperature is essential to design new alloy systems that are suitable for advanced manufacturing processes such as additive manufacturing. Once the properties are obtained, they are typically integrated with a meso-scale simulation framework to understand the impact of composition on different properties. While the forward problem is straight forward, the inverse problem necessitates the integration of the thermodynamic and meso-scale modeling with an optimization framework. The usage of machine learning (ML) tools is, therefore, deemed to be conducive to the development of a digital twin framework for both thermodynamic as well as meso-scale modeling. This paper implements a Gaussian Process (GP) framework to predict thermophysical properties (e.g., bulk density, solidus/liquidus temperatures) of a nickel-base metallic alloy system, nickel-chromium-aluminum (Ni-Cr-Al), over a temperature range. The results show that the proposed GP-based framework is conducive to predicting thermophysical properties with good accuracy and, thus, can be implemented as a surrogate in the digital twin development of additive manufacturing processes.


LCSH Subject Headings