In this study, we investigate grain growth within pure olivine systems through numerical simulations. The level set (LS) approach within a finite element (FE) context enables modeling 3D microstructural evolutions such as grain growth. As this phenomenon is inherently a 3D mechanism, the comparison between 2D and 3D models shows differences despite the use of 2D/3D transformation tools. The 2D level set approach is then compared with 2D models of grain growth performed with the ELLE software 1. Both approaches give consistent results. Our results confirm the rapid annealing of fine grained structures in pure olivine aggregates at temperatures of 1473 and 1573K. Comparison with previously published experimental results yields an estimate of the activation energy at 171−180 kJ•mol-1 for olivine grain boundary mobility.

We present a full field framework based on the level-set approach, which enables to simulate grain growth in a multiphase material. Our formalism permits to take into account different types of second phases, which can be static or dynamic (i.e., evolving also by grain growth) and reproduce both transient (evolving relative grain sizes) and steady-state structures. We use previously published annealing experiments of porous olivine or olivine and enstatite mixtures to constrain the parameters of the full field model, and then analyze the results of a peridotite-like annealing simulation. The experimental grain growth kinetics is very well reproduced while the simulated microstructure morphologies show some differences with experimental ones. We then propose a mean field model calibrated thanks to the full field simulations, which allow us to predict the mean grain size evolution depending on the simplified peridotite composition (e.g., second phase mean grain sizes, fractions). Physical Processes and MethodsWhile peridotites are mostly composed of olivine (generally close to forsterite composition with Mg/(Mg+Fe) near 0.9), they display a large variability in terms of mineral composition, which depends