Professor Raúl A. Radovitzky

Multiscale constitutive modeling of engineered materials

Professor Radovitzky’s research in this area addresses the critical need to develop descriptions of material behavior directly rooted in first principles of micromechanics. The overall strategy is to create the necessary mathematical frameworks, multiscale constitutive models and numerical algorithms to achieve this objective.

The research involves the development of tools providing a better understanding of the mechanical behavior of materials and a rational approach for material design.  The overall approach is aimed at building hierarchical models of complex material behavior with a minimum of empiricism and uncertainty.  Accordingly, the relevant unit processes are identified at each length scale in the entire hierarchy of material behavior. The unit processes at one scale represent averages of unit processes operating at the immediately lower length scale. For systems where these relations are well defined, the modeling effort reduces to the analysis of each unit mechanism in turn and the computation of averages, eventually leading to a full description of the macroscopic behavior of the material.

This inductive process stops at the atomic scale, where fundamental theories describing atomic bonds apply.  This paradigm has been applied to the modeling and simulation of the dynamic response of polycrystalline materials. During the past four years, Professor Radovitzky has been involved with Caltech, Rutgers University and LLNL in the Advanced Simulation and Computing (ASC) DOE program, an element of our nation’s Stockpile Stewardship Program.

This effort is multiscale in character and has led to the first full multiscale model of the dynamic response of tantalum for shock physics applications from quantum mechanics to continuum. The research has developed the modeling capability to bridge the meso- and macroscopic scales and establish the connection between microstructure and effective behavior. Part of this effort has gone into experimental validation of the models showing that the multiscale models not only reproduce the effective response but also capture local details of the deformation and grain interactions.

A large-scale simulation facility for investigating the dynamic response of polycrystals, scalable to thousands of processors, has been developed as part of this project. This has enabled simulations of specific problems in solid mechanics with a high level of spatial resolution. This facility also allows investigation of the sources of grain-level surface roughening in plastically deformed aluminum sheets.

Professor Radovitzky has also investigated the anomalous grain-size dependence of strength and ductility of nanocrystalline materials. His research has shown that the inverse Hall-Petch relation between the yield stress and the grain size predicted by molecular dynamics simulations and observed in experiments is confirmed by a continuum analysis; and that the discrepancy in the absolute values of strength between experiments and atomistic results may be partially due to loading rate effects.

The large-scale simulation capability developed has enabled ascertaining the role of grain interaction in determining effective response in a variety of important industrial scenarios. This has led to work with Alcoa, the main world provider of aluminum for the aerospace industry. Alcoa has provided high-purity Aluminum oligocrystals for exploring this new way of validating the high-fidelity simulations with attention to local details of grain interactions and has expressed interest in using the modeling and simulation capability.

Another collaboration with Tenaris, the world leader in the production of seamless steel pipes for the oil industry, applies this modeling and simulation paradigm to describe the onset of void formation in the Mannesmann piercing process.