Materials Informatics for Computational and Machine Learning (ML)-Assisted Design: An Overview for Polycrystalline Metals and Mechanical Metamaterials

In this talk, we will present an overview of the computational methods developed by our group to design metals and metamaterials for enhanced mechanical performance by modeling them in terms of micro-scale (~10^-6 m) features. First, we present numerical approaches to quantify the crystallographic texture and grain topology of polycrystalline metals. Similarly, a shape descriptor approach is developed to model the topology of mechanical metamaterials. Following the development and integration of such computational characterization approaches into numerical homogenization schemes to obtain mechanical properties, design optimization problems are developed and solved to design underlying microstructures of polycrystalline metals and mechanical metamaterials for improved mechanical performance.<br/> <br/>Next, the talk will focus on the impact of manufacturing-related uncertainty arising from the imperfections and defects during the processing and fabrication of materials on reliability and mechanical performance. We will discuss how to develop design under uncertainty formulations to address forward and inverse design problems in order to improve the elasto-plastic properties of polycrystalline metals and mechanical metamaterials. Additional topics will cover the integration of Artificial Intelligence (AI)/Machine Learning (ML) techniques into physics-informed material models to accelerate the design of material systems processed with conventional and additive manufacturing techniques. We will demonstrate applications of forward and inverse design problems for polycrystalline metals using ML-driven design approaches. The AI/ML methods will also be used for mechanical metamaterials to identify their mechanical property spaces showing all possible values of the selected properties.