Benchmarking FEniCS for mantle convection simulations

This paper evaluates the usability of the FEniCS Project for mantle convection simulations by numerical comparison to three established benchmarks. The benchmark problems all concern convection processes in an incompressible fluid induced by temperature or composition variations, and cover three cases: (i) steady-state convection with depth- and temperature-dependent viscosity, (ii) time-dependent convection with constant viscosity and internal heating, and (iii) a Rayleigh–Taylor instability. These problems are modeled by the Stokes equations for the fluid and advection–diffusion equations for the temperature and composition. The FEniCS Project provides a novel platform for the automated solution of differential equations by finite element methods. In particular, it offers a significant flexibility with regard to modeling and numerical discretization choices; we have here used a discontinuous Galerkin method for the numerical solution of the advection–diffusion equations. Our numerical results are in agreement with the benchmarks, and demonstrate the applicability of both the discontinuous Galerkin method and FEniCS for such applications.

► Effectiveness of utilizing the FEniCS Project for mantle convection simulations.
► The simulation results are in agreement with the published benchmarks.
► Advection–diffusion equations are solved by the discontinuous Galerkin (DG) scheme.
► DG scheme is an effective alternative for tracking compositional interfaces.