Large Scale Simulations of Grain Boundary Motion in Polycrystals

  • Selim Esedo
  • Published 2010

Abstract

Many common materials, such as most metals and ceramics, are polycrystalline: They are composed of tiny crystallites often called grains that are di erentiated from their neighbors by di ering orientation. The grain structure of a polycrystalline material a ects its physical properties, such as fracture strength and conductivity. Accordingly, simulating how the network of grains evolve under manufacturing processes such as annealing (heat treatment) is of great interest. For example, annealing leads to coarsening of the grain network, whereby certain grains grow at the expense of others, leaving fewer and thus larger grains on average. Many numerical techniques, including Monte Carlo, front tracking, and phase eld methods, have been used in simulations of this important phenomenon. Some of these methods, such as front tracking, can be very accurate in 2D but awkward in 3D simulations due to the variety of topological changes that inevitably take place during the coarsening process. In my talk at the SIAM Materials Science meeting in May, drawing on recent joint work with Matt Elsey and Peter Smereka [4, 5], I described new, level set based algorithms that have allowed us to carry out fully resolved 3D simulations with very large numbers of grains on modest hardware. Mathematically, we can represent a polycrystalline material as a partitioning of the volume D it occupies into connected, pairwise disjoint regions (grains) Σ1, . . . ,ΣN :

3 Figures and Tables

Cite this paper

@inproceedings{Esedo2010LargeSS, title={Large Scale Simulations of Grain Boundary Motion in Polycrystals}, author={Selim Esedo}, year={2010} }