A simulation model for fracture reactivation accounting for thermo-hydro-mechanical forces with applications to geothermal systems

Abstract:

Fracture networks are the main conduits for fluid flow in many subsurface applications, hereunder production of geothermal energy. Injection and extraction of fluids may alter the stress on fracture surfaces to a state where the available frictional forces are insufficient to withstand tangential forces, causing the fracture to slide and deform. As the deformation results in enhanced fracture permeability, the deformation is sometimes intentionally triggered to stimulate wells, however, the sliding is also associated with undesired effects such as induced seismicity.

From a modeling perspective, the stress at the fracture walls is a manifestation of mechanical, hydraulic and thermal forces in the host rock, combined with fluid forces inside the fracture. Fracture deformation can be considered a frictional contact problem which is localized to the fracture network. Together, this calls for a model that can represent different governing equations in host rock and fracture network. Moreover, the governing processes are tightly and non-linearly coupled, thus the simulation model should be based on robust numerical methods.

Our simulation tool, the open-source software PorePy, is based on discrete fracture matrix (DFM) principles where major fractures are explicitly represented in the computational mesh. The model prescribes conservation of mass, energy and momentum, while fracture deformation is handled by techniques from contact mechanics. We present the simulation tool, with emphasis on coupling of processes between host rock and fracture network.

We validate our simulation framework by simulations of thermo-hydro-mechanical fracture deformation, and go on to consider in low-pressure stimulation and long-term cooling of a synthetic 3d geothermal reservoir. Furthermore, we apply our simulation framework to stimulation of a geothermal well at the Reykjanes peninsula, Iceland, and compare simulation results with seismic observations.