Surrogate models provide accurate solutions to problems in science and engineering at a fraction of the computational time. Our group has recently developed a technique for training neural networks using the finite element method (patent pending), so that surrogate models can be easily developed using existing finite element codes.
Funding: National Science Foundation
Most ductile metals (e.g., steels, aluminum alloys) fail when small voids nucleate and grow to coalescence. This project is focused on using molecular dynamics simulations and machine learning (in collaboration with Prof. Aziz Ezzat of ISE) to study the micromechanics of void nucleation at small particles. The goal is to use these simulations to develop a micromechanically informed theory for void nucleation and damage.
Funding: National Science Foundation
Many advanced steel and titanium alloys undergo a so-called martensitic phase transformation during deformation, giving these alloys exceptional strength and ductility. At present, the mechanisms underlying this process are poorly understand, making it difficult to predict mechanical response and develop better alloys. The goal of this project is to reveal the fundamental micromechanics which govern the nucleation and growth of martensitic phases in steel alloys.
Funding: Department of Energy – Office of Science
The mechanical behaviors of most metals are governed by the evolution of dislocation network structures. However, existing models for predicting deformation and damage in metals largely neglect the dislocation network. The goal of this project is to develop a new theory that captures evolution of the dislocation network in precipitation strengthened metals to provide a physics-based plasticity modeling framework.
Funding: Army Research Office, Naval Nuclear Laboratory
In heavily deformed metals, dislocations arrange themselves into intricate, cellular patterns (see TEM image to the right, courtesy D. L. Medlin). It has been a long-standing theoretical quandary why dislocations would form such ordered networks rather than random jumbles. The goal of this project is to couple discrete dislocation dynamics simulations with three-dimensional electron tomography to gain insight into what stabilizes the walls of cellular dislocation structures.
Funding: Sandia National Laboratories
Under fatigue loading, cracks often initiate at microstructural features, such as twin boundaries (as shown in the image to the right, courtesy C. W. San Marchi) and deformation-induced phases. The microstructural conditions which drive these initiation events are poorly understood, however. The goal of this project is to utilize multiscale materials modeling to better understand how defect interactions can drive crack initiation in hydrogen-affected metals.
Funding: Sandia National Laboratories
In collaboration with Prof. Jon Singer of MAE, we are simulating the formation of microporous nanocomposite emulsion thermosets (MINETs), a highly tunable class of microporous composites recently developed by Prof. Singer’s group. The composites are produced by mixing oil, epoxy, surfactant, and nanoparticles, and then rinsing out the oil phase after the epoxy has cured. Our goal is to understand the mechanisms and scaling behaviors underlying the formation of these microporous networks.
Graphene and other two-dimensional materials exhibit numerous interesting properties that are sensitive to the state of deformation. In this research, we are using molecular dynamics simulations to understand the fundamental mechanisms that govern plasticity in graphene during out-of-plane loading.
Ceramics are ideal structural materials in all ways except one: they are brittle and flaw intolerant. The goal of this research, in collaboration with the Riman Group, is to determine optimal designs for ceramic composites which enhance toughening to reduce brittleness.
