Data-Driven Prediction of Mechanical Properties in Support of Rapid Certification of Additively Manufactured Alloys

Predicting the mechanical properties of additively manufactured parts is often a tedious process, requiring integration of multiple stand-alone and expensive simulations. Furthermore, as properties are highly location-dependent due to repeated heating and cooling cycles, the properties prediction models must be run for all locations, thus compounding the computational expense, before the part-level performance can be analyzed for certification. This work has proposed a rapid prediction framework that replaces the physics-based mechanistic models with Gaussian process (GP) metamodels, a type of machine learning model for limited data and statistical inference, that can predict the varying properties in an entire part in a fraction of the time. The framework was demonstrated with the prediction of the tensile yield strength of Ferrium® PH48S maraging stainless steel fabricated by additive manufacturing processes. An impressive agreement was found between the metamodels and the mechanistic models, while the computation was sped up dramatically from hours of physics-based simulations to less than a second with GP metamodels. This method can be extended to predict various materials properties in different alloy systems whose process-structure-property-performance interrelationships are linked by mechanistic models, which can then be replaced by GP metamodels, providing fast and accurate estimations with uncertainty quantification. This approach is powerful for rapid identification of spatially-variant properties within a part where compositional and processing parameter variations are present, and can support part certification by providing a fast interface between materials models with part-level thermal and performance simulations.