The Deep Crust Magmatic Refinery, Part 1: A Coupled Thermodynamic and Two-phase Flow Model

Metamorphic and magmatic processes occurring in the deep crust ultimately control the chemical and physical characteristic of the continental crust. A complex interplay between magma intrusion, crystallization, and reaction with the pre-existing crust provide a wide range of differentiated magma and cumulates (and / or restites) that will feed the upper crustal levels with evolved melt while constructing the lower crust. With growing evidence from field and experimental studies, it becomes clearer that crystallization and melting processes are non-exclusive but should be considered together. Incoming H2O bearing mantle melts will start to fractionate to a certain extent, forming cumulates but also releasing heat and H2O to the intruded host-rock allowing it to melt in saturated conditions. The end-result of such dynamic system is a function of the amount and composition of melt input, and extent of reaction with the host which is itself dependent on the migration mode of the melts. To assess the dynamics of this deep magmatic system we developed a new 2-D two-phase flow code using finite volume method. Our formulation takes into account: (i) melt flow through a viscous porous matrix with temperature- and melt-content dependent host-rock viscosity, (ii) heat transfer, assuming local thermal equilibrium between solid and liquid, (iii) thermodynamic modelling of stable phases, (iv) injection of fractionated melt from crystallizing basalt at the Moho and (v) chemical advection of both the solid and liquid compositions. Here we present the core of our modelling approach, especially the petrological implementation. We show in details that our thermodynamic model can reproduce well both the sub- and supra solidus phase relationship and composition of the host-rock. We apply our method to an idealized amphibolite lower crust that is affected by a magmatic event represented by the intrusion of a wet mantle melt into the crust at Moho depth. The models [see Bouilhol et al. associated abstract for results] allow calculating the different proportion of phases present in the system through time.