Elle examples:

Layered ice grain growth
I set up a model in Elle with a horizontal grain size layering, I ran the experiment for 6000 time steps. Except for the time=6000 stage we can still see some effect of the initial grain size variation. Figure 1 shows in black just those areas unswept by grain boundaries during the entire experiment. The inital grain size layering is still apparent as a size variation in these cores.
Mark Jessell

Localisation of deformation
These simulations couple a grain size and strain dependant viscous rheology with grain size reduction and grain growth processes. It was found that significant strain localization occurred only when the grain size dependence of the viscosity was non-linear, and was greatly enhanced by the activity of the grain size modifying processes. The intensity and location of the zone of strain localization varied spatially and temporally, with the result that the finite strain state showed a much broader, and hence less intense, zone of localized deformation than the instantaneous state.
Mark Jessell

Melt Pocket Angle=180
Evolution of a single melt pocket situated ata triple junction, with a melt-crystal wetting angle of 180°
Mark Jessell

Melt Pocket Angle=0
Evolution of a single melt pocket situated at a triple junction, with a melt-crystal wetting angle of 0°
Mark Jessell

Melt Pocket Angle=120
Evolution of a single melt pocket situated ata triple junction, with a melt-crystal wetting angle of 120°
Mark Jessell

Ostwald Ripening
Simulation of Ostwald Ripening. Ostwald ripening is the process by which larger particles (or, for emulsions, droplets) grow at the expense of smaller ones due to the higher surface energy and hence solubility of the smaller particles and to molecular diffusion through the second phase. In this example the green phase is also undergoing grain growth.
Mark Jessell

Grain boundary migration microstructures
Grain growth simulations using the microstructure simulation system Elle have been performed inmaterials with a pre-existing grain shape foliation. As might be expected, the foliation is destroyed by the end of the experiment, and grain areas have increased by a factor of seven. The area of material swept by the migrating grain boundaries was monitored, and it was found that at every stage, virtually all of the grains which survived the grain growth process contain one and only one core of unswept material. Remarkably these remnant unswept cores preserve a useable record of the initial grain size and the orientation of the grain shape foliation. This work suggests that, even for samples where no equivalent protolith can be found, it may be possible to see past a grain growth episode to estimate the original grain shape and grain size of the rock, and perhaps even reconstruct the grain boundary kinematics. In addition the identi?cation of unswept cores has the potential to help unravel the evolution of grain boundary chemistry in rocks during metamorphism. From:The preservation potential of microstructures during static grain growth
M. W. JESSELL, O. KOSTENKO AND B. JAMTVEIT. J. metamorphic Geol., 2003, 21, 481-491

Crystallisation from a melt
In this experiment the equilibrium melt level is decreased by a small amount each time step, which simulates crystallisation from the melt. The surface energies are anisotropic so the crystals tend to develop lath shaped crystals as they grow, until they touch each other, where the anisotropy of surface energy is significantly lowered, and irrational interfaces result.
Mark Jessell

Triple junction trails
These trails might be expected if triple junctions contain a different chemistry, or if partitioning is different at triple junctions. White lines are grain boundaries, purple lines track successive positions of triple junctions. Note the rough bilateral symmetry of triple junction trails that develops when a grain disappears.
Mark Jessell

Porhyroblast Rotation?
This model shows the effects that "mica" caps can have on the kinematic behavior of a rigid object (a porphyroblast in this case). This model runs to a shear strain of 2.0. In a homogeneous, viscous material undergoing simple shear, a rigid object with circular cross section should rotate 57 degrees relative to the instantaneous stretching axes after a shear strain of 2.0. By assigning different competency contrasts between the matrix and mica caps (the porphyroblasts is assumed to be rigid), we can achieve rotations between zero and 57 degrees relative to the ISA. This movie shows zero degrees of rotation! This was done with Elle. Mica caps are continually generated around rigid objects during deformation through dissolution of soluable phases like quartz and feldspar, with or without growth of new mica. An important aspect of this movie is that I included manual "dissolution" to concentrate mica in the cap. If this is not done, the mica flows past the porphyroblast and the cap diminishes.
Scott Johnson

Grain boundary sweeping
Development of grain boundary migration growth bands. Development of remnant unswept cores during grain growth, after 0, 500, 1000 and 1500 time steps. At each stage of the evolution, areas never swept by grain boundaries are left green, and areas swept at least once by a migrating boundary are coloured white. Grain boundaries are shown as black lines. At every stage of the evolution the vast majority of grains possess a core of unswept material. Some of these cores at each stage are in contact with the grain boundary network, whereas others are completely isolated from it.
Mark Jessell

Cyclic grain growth bands
Oscillatory zoning predicted by cyclical grain boundary chemistry variations during grain growth. Note preferential widening of bands around triple junctions. Figure 1 Pattern after 200 time steps, pattern after 10,000 time steps. Note repeated truncations of the patterns.
Mark Jessell

Defect energy driven gbm growth bands
Zoning predicted for grain boundary migration driven by defect energy contrasts. Oscillatory zoning produced by cyclical grain boundary chemistry variations. Note relatively uniform band widths. Figure 1 Distribution of defect energy levels in initial model (red high, blue low).
Mark Jessell

Exchange Reaction
Zoning predicted for garnet/biotite Fe exchange reactions. In this experiment both grain boundary and lattice concentrations are shown with the same look up table, (but with different scaling). The grain boundary concentrations show the gradual loss of Fe from the Biotite, and the lattice concentrations show the preferential enrichment at garnet grain boundaries, but also the preferential enrichment around the perimeter of the cluster, and finally the preferential enrichment where the biotite grains are nearest. Zoning patterns are roughly concentric with respect to current grain boundaries. Note that this model has cyclic limits, so that the biotite grains are also ?near? below and to the right of the cluster. Figure 1 Starting model configuration. Garnet grains are red, biotite grains are yellow, matrix grains are green, grain boundaries are white, and blue dots show the location of unconnected nodes that store lattice chemistry information.
Mark Jessell

Sub-grain growth
Numerical simulation of type II subgrain growth ? anisotropic surface energy. Representation of a combination of the 3 Euler angles in different colour. Grain boundaries with a lattice misorientation angle above 15° are shown in black. Each picture represents 100 model timesteps.
From: S. Piazolo, M. W. Jessell, D. J. Prior, P.D. Bons (submitted J. Microscopy) The integration of experimental in-situ EBSD observations and numerical simulations: A novel technique of microstructural process analysis.

Grain boundary diffusion
Simulation of grain boundary diffusion away from the boundaries of a single grain with a high concentration of a trace element.
Mark Jessell

Dislocation density driven grain boundary migration
In this experiment, the red grains have high dislocation densities and the blue grains have low values. This is the major driving force for grain boundary migration in this example.
Mark Jessell

Grain Growth
Elle simulation of Normal Grain Growth, driven by the reduction of grain boundary energy.
Mark Jessell

Exaggerated Grain Growth
Grain growth dominated by high surface energy anisotropy, which prevents most boundaries from moving. Grains whose boundaries which can still move eventually dominate the microstructure. Note that even though the starting microstructure is a foam texture, many of the triple junctions that develop, even in the apparently stable finer grained material, do NOT evolve to 120°, which would be the equilibrium angle if there was no surface energy anisotropy.
Mark Jessell

Crystallisation from a melt
Crystallisation from a single seed. The anisotropy of surface energy eventually controls the external crystal form.
Mark Jessell

Surface Energy Reduction of a Snowflake
Simulation evolution of grain shape of a single snowflake as a result of reduction of surface energy, but no change in overall area.