The integration of computational thermodynamics and geodynamics is critical for a wide range of problems from mantle convection with consistent phase changes, to open system reactive transport during fluid/magma migration. These problems are strongly coupled, non-linear, involve large numbers of variables, and remain some of the outstanding computational challenges in solid-earth science. To address these challenges requires flexible modeling software that allows the user to control the degree of complexity in both thermodynamic and geodynamic models.

Advances in computational software over the last decade have provided a range of numerical libraries that allow the application of symbolic computational mathematics to geodynamics.  These libraries turn simple problem descriptions into problem-specific, high-performance code.  Importantly for highly coupled problems, the use of symbolic calculus allows the automatic calculation of derivatives for use in the nonlinear solver.  ThermoCodeGen, part of the larger ENKI project for thermodynamic modeling, applies these principles to thermodynamic problems, automatically generating the derivatives of free energy models to supply consistent thermodynamic parameters for use in geodynamic models.  I will present results applying ThermoCodeGen to mantle convection problems, demonstrating a range of complexities in both the geodynamic problem formulation and the degree of thermodynamic consistency.