D-Rex++: A new and improved tool to bridge microstructure evolution in mantle-scale geodynamics models

Olivine, the most ubiquitous mineral in the upper mantle, is believed to control the mantle’s rheological properties, and its evolution of crystallographic preferred orientation (CPO) is the primary cause of the mantle’s seismic anisotropy. However, tracking microstructural and textural evolution of olivine-rich mediums in a large-scale geodynamic model has proven to be challenging. The mean-field, kinematic modeling algorithm ‘D-Rex’ has been widely used to simulate CPO evolution in geodynamic models of mantle flow. Despite D-Rex being able to successfully simulate the evolution and formation of CPO, it lacks important microstructural properties. A key limitation of D-Rex has been its lack of dimensionalization and inability to predict a dimensional grain size evolution, which yields an unrealistic evolution of grain size, and over-predicting CPO strength when run to strains where steady-state is expected (>5).

Here, we present D-Rex++, a mean-field CPO evolution framework built upon D-Rex and embedded with the large-scale geodynamic model ASPECT, which enables the simulation of the co-evolution of CPO and the dimensionalized grain sizes of olivine aggregates deformed to high strains. Evolution of grain size results from the competition between nucleation of strain-free grains during dynamic recrystallization and subsequent recovery, driven either by strain energy gradients (strain-induced grain boundary migration, SIGBM) or by grain boundary curvature (grain coarsening). The model's prediction is influenced by the choice of two major free parameters: M_b, which represents the grain boundary mobility and controls the degree of grain size evolution during SIGBM, and Δ_rx, the rate of dynamic recrystallization.

To benchmark the free parameters, Δ_rx and M_b, we compared the results from simple shear-box models with existing data from laboratory shear experiments. We observe that increasing M_b increases the strength of the predicted pole figure and increases the range of the grain size distribution (GSD). The value of Δ_rx serves to decrease the strength of the CPO due to the influx of grains whose orientations are dispersed from their parent grain (i.e., recrystallized). Simulations were conducted to assess the model's ability to predict the impact of pre-existing fabric. In addition, models run under tectonic-scale strain rates were able to simulate the natural occurrence of CPO evolution in response to accumulated shear strain.

A key aspect of the new model is its ability to account for the evolution of the GSD in conjunction with the texture and deformation history. This enables the use of a composite diffusion–dislocation creep viscosity formulation with varying evolving grain size within ASPECT and coupling of microstructural evolution with large-scale geodynamic models. To conclude, the dimensional treatment of microstructural parameters provides a physically interpretable framework that enables systematic calibration and direct integration with the prediction of rheology and viscosity evolution in geodynamic models. D-Rex++ thus provides a pathway toward mantle convection models in which grain size and CPO can evolve consistently, and can be progressively grounded in realistic microstructure evolution.