Abstract
This study investigates the development of methods in directed energy deposition (DED) process-structure-property (PSP) maps to enable rapid process-informed design of materials for high temperature and harsh environments. Data-driven and physics-based PSP approaches are investigated through experimentation with 316L stainless steel and Inconel 718, characterization, and analysis with analytical and numerical modeling. DED methods of laser-engineered net shaping (LENS) and wire-arc additive manufacturing are compared through an analysis of geometry, surface roughness, and process-parameter relationships. Application based performance and fabrication feasibility of heat exchangers (HX) are studied through fabrication of flow channels, measurement of channel surface properties, and numerical modeling and performance analysis of a parallel flow HX with the channel walls representative of the as measured surface roughness. Data-driven approach is investigated using machine learning (ML) methods with a comparison and analysis of destructive and non-destructive measurements of clad geometry. A ML framework is developed and used to predict fluctuations in processing and melting geometry. A physics-based approach is studied through development and experimental validation of a DED process model in Multiphysics Object-Oriented Simulation Environment (MOOSE) Application Library for Advanced Manufacturing UTilitiEs (MALAMUTE). Validation is performed through an investigation of laser and powder efficiency with ML models trained using a stochastic approach.