Discrete Fracture Modeling approach for simulating coupled thermo-hydro-mechanical effects in fractured reservoirs

We present a new formulation for modeling the coupled flow, thermal, and mechanical effects in real fractured subsurface formations. Natural fractures are represented explicitly as surfaces in contact with fluid between them. We use a finite volume method to discretize mass and heat conservation equations and a Galerkin finite element method to discretize mechanical equilibrium equations. We use the Nitsche approach to enforce fracture-contact constraints. The proposed formulation allows us to accurately evaluate fracture conductivity and volume change due to stress changes induced by variations in fluid pressure and temperature. To solve the resulting set of non-linear equations, we present a new sequentially implicit solution strategy based on a fixed-stress scheme for fractures, in addition to the established fully implicit approach. Our formulation is implemented within an existing general purpose research simulation platform that can account for complex physics such as multi-phase, multi-component flows. We show the convergence behavior of the sequential-implicit scheme for the problems considered. We demonstrate the utility of the formulation by considering the process of injecting cold water into a real fractured carbonate reservoir and evaluating the impact of resulting changes in formation pressure, temperature, and acting stresses on well bottom hole pressure measurements at the well. The goal of this application is to analyze the anomalously declining well pressure behavior observed during an actual short-term water-injection pilot operation towards informing the decision-making and preparation for potential large-scale waterflooding. We first perform a sensitivity analysis using a 2D conceptual model to assess key factors controlling the coupled effects for this problem: fracture and matrix properties, initial stress state, pressure- and temperature-induced stress changes and the propensity of fractures to slip, as well as Biots coefficient. We then applied the framework on a 3D sector model of a fractured reservoir. We conclude that the observation was likely due to the substantial stress reduction caused by pressure and temperature changes, leading to slip events and a significant increase in fracture network conductivity near the injector. Thermal effects have less of an impact during the relatively brief pilot injection in this case but are expected to become important for long-term waterflooding. The application to a real waterflooding problem demonstrates the usefulness of the proposed thermo-hydromechanical framework for practical studies of real fractured reservoirs.