Seminar Pauline Chassonnery

Thermo-Hydro-Mechanical models in faulted porous media play an important role in many applications in geosciences. Fault networks act as corridors for fluid flows and associated mass and energy transfers and they have a preeminent role on stress (re)distribution interacting with these transfers. Numerical modelling is a key tool to better understand, assess and control these processes. They couple the flow and energy transport along the faults and in the surrounding porous rock, the rock mechanical deformation, and the mechanical behaviour of the faults related to contact-mechanics.

The ComPASS project (https://gitlab.com/compass) aims to provide a generic, open-source, user-friendly platform to conduct high-performance simulations of multicomponent multiphase non-isothermal flows in faulted porous media. The main objective is to create a quantitative subsurface model to better understand phenomenon such as fault reactivation, with applications in geoscience fields like geothermal energy, CO2 underground storage, etc. It is developed in collaboration between the French National Geological Service (BRGM), the University Côte d’Azur (UniCA) and the French National Agency for radioactive waste management (ANDRA).

In this talk, we will focus on the simulation of contact-mechanics in a faulted porous medium. First, we will present a discretization of the elastic deformation problem using the Virtual Element Method (VEM), which is a natural extension of the P1 Finite Element Method (FEM) to polyhedral meshes. Then we will discuss the implementation of Coulomb-type contact mechanics across faults, using face-wise constant unknowns for contact surface tractions at the matrix-fault interfaces. This choice allows to cope with fault networks including corners, tips and intersections and leads to contact conditions local to each fault face, that can be solved with efficient semi-smooth Newton non-linear solvers. Its requires a stabilisation which is achieved by enrichment of the displacement field with an additional “virtual” bubble unknown at one side of each fault face. Lastly, we will consider the fully coupled thermo-hydro mechanical problem. Because this problem results in a very large and usually badly conditioned linear system, we will address it through a possibly accelerated iterative coupling algorithm.

These results highlight the necessity of explicitly considering fracture traces in the shallow subsurface for effective gas monitoring, and this improved understanding is expected to enable more realistic predictions of gas behavior and support successful environmental management.