COGRA: Decoding the mechanics of metals by coarse-grained atomistics
An ERC Consolidator Project
"First principles" and "bottom-up" have become buzz words across scientific and engineering disciplines when it comes to the discovery, prediction and understanding of material properties and their link to processing and microstructure. Reality, however, teaches us that in the foreseeable future computational resources will be insufficient to apply predictive techniques such as quantum mechanics or atomistics to the technologically most relevant length and time scales - far above nanometers and nanoseconds. This project aims for nothing less but the seemingly impossible: the application of atomistic techniques to problems occurring over microns to millimeters and seconds to minutes. Instead of relying on computational power, this will be achieved by a combination of scale-bridging methodologies (involving the our nonlocal and meshless quasicontinuum techniques, concepts from particle methods, continuum and statistical mechanics) and computational science strategies in order to produce new theory and an open-source, computational toolset for long-term, large-scale simulations relying solely on atomistic input. Spatial upscaling, temporal upscaling as well as heat and mass transfer will be addressed. Enabled by the new scale-bridging capabilities, two representative, open challenges will be investigated: recrystallization in magnesium during thermo-mechanical processing and corrosion in iron by hydrogen embrittlement. Both are of enormous technological and economic importance but current techniques are insufficient to bridge the gap between the macroscopic mechanical performance, microstructural mechanisms and predictive atomic-scale simulations. The outcomes of this five-year research program will provide never-before techniques and numerical tools to catalyze a user community across science and technology. Although the focus is on metals, several of the proposed techniques are applicable to a significantly wider range of materials and applications.
Figure: this project is broken down into tasks that adress spatial coarse-graining, temporal upscaling, and the handling of multiple species in order to produce a new computational toolset as well as insight into the representative metallurgical problems of recrystallization and hydrogen embrittlement.
We develop a fully-nonlocal quasicontinuum (QC) formulation that is based on an updated-Lagrangian setting and improved meshless interpolation schemes for adaptive refinement and coarsening of discrete, atomistic ensembles. Unlike most prior related techniques, this will enable the simulation of large samples with on-the-fly adaptivity to tie full atomistic resolution to evolving regions of interest.
Figure: The QC methodology reduces the number of degrees of freedom by introducing representative atoms and suitable interpolation schemes to recover the full ensemble.
- I. Tembhekar, J. S. Amelang, L. Munk, D. M. Kochmann. Automatic adaptivity in the fully-nonlocal quasicontinuum method for coarse-grained atomistic simulations, Int. J. Numer. Meth. Engng. 110 (2017), 878–900.
- J. S. Amelang, G. N. Venturini, D. M. Kochmann. Summation rules for a fully-nonlocal energy-based quasicontinuum method, J. Mech. Phys. Solids 82 (2015), 378-413.
- J. S. Amelang, D. M. Kochmann. Surface effects in nanoscale structures investigated by a fully-nonlocal energy-based quasicontinuum method, Mech. Mater. 90 (2015), 166-184.
- D. M. Kochmann, G. N. Venturini. A meshless quasicontinuum method based on local maximum-entropy interpolation, Mod. Sim. Mat. Sci. Eng. 22 (2014), 034007.
- M. Espanol, D. M. Kochmann, S. Conti, M. Ortiz. A Γ-convergence analysis of the quasicontinumm method, Multiscale Model. Simul. 11 (2013), 766-794.
This project is sponsored by the European Research Council through the 2017 Consolidator Award COGRA.