Multi-variate Process Models for Predicting Site-Specific Microstructure and Properties of Inconel 706 Forgings

High throughput measurements of structure from simulationsSimulation and microstructural analysis were developed and combined into an adaptable data-analytics processing–structure–property modeling framework to gain insight into the design envelope for forgingsForging of Inconel 706Inconel 706. We highlight how thermal profiles from finite element simulationsSimulation (DEFORM) can establish time–temperature boundary conditions for CALPHADCALPHAD predictions of the combined γ′ γ′′ precipitate distributionPrecipitate distributions in Inconel 706Inconel 706 (structure predictors). Experimental observations of these precipitate distributionsPrecipitate distributions allow for the tailoring of the CALPHADCALPHAD interfacial energy. In this manner, a 25 × reduction in the number of physical observations of γ′ and γ′′ distribution (78 to 4) results in site-specific processing–structure, and processing–property models of forged parts with 80% predictive power. Additionally, the gradient boost modeling provides an opportunity to interpret feature importance in structure–property models to provide engineers with design insights for future development efforts. The adaptable framework would enable future DEFORM and CALPHADCALPHAD simulationsSimulation to be added to the dataset so that engineers can interpolate within the existing experimental dataset.