Numerical simulations of microstructures using the Elle platform: A modern research and teaching tool

The last 20 years have seen a manifold increase in the application of numerical simulations in the earth sciences. This contribution aims to provide an overview of the possibilities of using numerical techniques, in particular the numerical simulation package Elle, as an aid in the training for microstructural interpretation in rocks. Three sets of experiments are described to illustrate the range of simulations currently possible, investigating the effects of grain growth, polyphase deformation and dynamic recrystallisation. Numerical simulations of static annealing of a pre-deformed natural sample show that Crystallographic Preferred Orientations can still be used for the interpretation of kinematic and deformation conditions, even after substantial postdeformational annealing. However, the grain network characteristics such as grain size, grain size distribution, boundary shapes and aspect ratios are rapidly altered during annealing, especially if the grains possessed highly contrasting internal strain energies. Experiments modelling two and three phase viscous deformation show that the rheology and microstructural evolution of a rock is largely determined by the linearity or non-linearity of viscous deformation; whereas the number of phases with differing viscosity is less important. Variations in strain, in the same mineral phases can be used to infer flow properties. The spatial distribution of phases significantly influences the rheology at a specific point in time. During dynamic recrystallization the rates of the competing processes of grain size reduction and increase can have a systematic influence on the evolution of grain characteristics. Relatively high rates of grain size increasing processes result in larger grain sizes, lower aspect ratios, stabilization of grain size at lower strain and less strain localization. Numerical simulations can help to answer field related questions and to identify the number of significant active processes. They are also extremely useful as teaching tools as processes can easily be turned on or off, parameters can be changed and boundary conditions altered. Fortunately, numerical simulations have now matured to a point where users that do not want to spend time in coding a model can still perform and analyse numerical simulations.