| Abstract Scope |
A principal aspect of modern computational materials analysis is quantifying process-structure relationships due to thermomechanical processing. In his storied career, Prof. Anthony Rollett has made seminal contributions in all aspects of this field, including methodologies for process simulation and analysis techniques for such simulations’ results. With the proliferation of machine learning, new tools have arisen that provide further approaches for bridging the gap between manufacturing process and microstructural features. In this presentation, we highlight how early advances in extreme value statistics, pioneered by Tony, provided a robust toolset for analyzing microstructure in the context of processing and performance as predicted from simulation. We showcase how these techniques informed further developments utilizing unsupervised machine learning to zone simulated process and structure information, producing quantitative “maps” between these two legs of the materials tetrahedron. Finally, we discuss the utility of such maps in driving inverse materials design: using design inputs for microstructure and performance to yield parameters for thermomechanical processing. These examples are presented in the context of both classical and novel thermomechanical processes: thermoelasticity in coatings, bulk metallic forgings, and metallic additive manufacturing, all fields to which Tony has provided foundational contributions. |