The Subduction Framework Utilising Scientific computing (SubFUSc) - an open-source simulation tool for two-phase dynamics of subduction zones

Subduction zones (SZs) are the crossroads of plate tectonics, where volatile-rich sediments, basalts, and lithosphere exchange mass and energy with the shallow and deep mantle. Solids descend while liquids ascend; temperatures vary by 800K; melting occurs by hydrous flux and decompression; liquids range from hydrous to silicic; energy is transported in every direction. This physical and chemical complexity means that our understanding of SZ magmatism lags behind other tectonic settings. Progress requires a modelling framework that consistently couples essential physics and allows for hypothesis testing and interpretation of observations. To this end we introduce SubFUSc, open-source software under active development. SubFUSc is based on the two-phase physics of partially molten rocks. It simulates viscous mantle shear and compaction with porous melt segregation. The mechanics are coupled to conservation of energy and chemical-component mass. Melting is computed assuming local thermodynamic equilibrium, but the code is agnostic to the complexity of this calculation. We use parameterised phase diagrams with 3-4 (pseudo-)components. Discretization is by finite element, finite volume, and characteristics-based methods, matching the method with the demands of each physics component. The Stokes/Darcy mechanics are treated with Pk-Pk-1 mixed finite elements on a locally refined triangular mesh. Temperature is evolved on an overlapping mesh, but with a high-order basis to resolve sharp gradients and obviate numerical stabilization. Composition is discretized on a uniformly fine, Cartesian grid; transport is by conservative semi-Lagrangian method with monotone and bound-preserving reconstruction. The code is written in C; its only dependency is the PETSc. Abstraction layers within SubFUSc enable re-configuration for different two-phase flow problems. We demonstrate the code with results from two ongoing studies: the role of energy transport by magma in thermal structure and the pathways of carbon through SZs. For the former, we compare simulation results with simpler models [arxiv.org/abs/1701.02550] showing that thermal structure is significantly modified by magma transport.