This work has investigated the roles of effective stress induced by changes in fluid pressure, temperature and chemistry in contributing to the evolution of permeability and induced seismicity in geothermal reservoirs. This work has developed continuum models [1] to represent the progress or seismicity during both stimulation [2] and production [3]. These methods have been used to resolve anomalous observations of induced seismicity at the Newberry Volcano demonstration project [4] through the application of modeling and experimentation. Later work then focuses on the occurrence of late stage seismicity induced by thermal stresses [5] including the codifying of the timing and severity of such responses [6]. Furthermore, mechanistic linkages between observed seismicity and the evolution of permeability have been developed using data from the Newberry project [7] and benchmarked against field injection experiments. Finally, discontinuum models [8] incorporating the roles of discrete fracture networks have been applied to represent stimulation and then thermal recovery for new arrangements of geothermal wells incorporating the development of flow manifolds [9] in order to increase thermal output and longevity in EGS systems.

Stress and permeability variations around a wellbore and in the reservoir are of much interest in petroleum and geothermal reservoir development. Water injection causes significant changes in pore pressure, temperature, and stress in hot reservoirs, changing rock permeability. In this work, two- and three-dimensional finite element methods were developed to simulate coupled reservoirs with damage mechanics and stress-dependent permeability. The model considers the influence of fluid flow, temperature, and solute transport in rock deformation and models nonlinear behavior with continuum damage mechanics and stress-dependent permeability. Numerical modeling was applied to analyze wellbore stability in swelling shale with two- and three-dimensional damage/fracture propagation around a wellbore and injection-induced microseismic events. The finite element method (FEM) was used to solve the displacement, pore pressure, temperature, and solute concentration problems. Solute mass transport between drilling fluid and shale formation was considered to study salinity effects. Results show that shear and tensile failure can occur around a wellbore in certain drilling conditions where the mud pressure lies between the reservoir pore pressure and fracture gradient. The fully coupled thermo-poro-mechanical FEM simulation was used to model damage/fracture propagation and microseismic events caused by fluid injection. These studies considered wellbore geometry in small-scale modeling and point-source injection, assuming singularity fluid flux for large-scale simulation. Damage mechanics was applied to capture the effects of crack initiation, microvoid growth, and fracture propagation. The induced microseismic events were modeled in heterogeneous geological media, assuming the Weibull distribution functions for modulus and permeability. The results of this study indicate that fluid injection causes the effective stress to relax in the damage phase and to concentrate at the interface between the damage phase and the intact rock. Furthermore, induced-stress and far-field stress influence damage propagation. Cold water injection causes the tensile stress and affects the initial fracture and fracture propagation, but fracture initiation pressure and far-field stress are critical to create a damage/fracture plane, which is normal to the minimum far-field stress direction following well stimulation. Microseismic events propagate at both well scale and reservoir-scale simulation; the cloud shape of a microseismic event is affected by permeability anisotropy and far-field stress, and deviatoric horizontal far-field stress especially contributes to the localization of the microseismic cloud.

A review was made of major reservoirs of the Corps of Engineers with regard to their experiences with induced seismicity. The Corps has 24 large reservoirs, i.e. those with dams 60.96 m (200 ft) in height or greater, and with volumes of 1,233,500,000 cu m (1,000,000 acre-ft) or greater. Thirteen reservoirs were instrumented for monitoring microearthquakes. They range from large to small and are located in aseismic to highly seismic areas. Only one Corps reservoir, Clark Hill in Georgia-South Carolina, has experienced a felt earthquake, but the 21-year interval between impoundment and the occurrence of the earthquake is not typical of reservoir-induced earthquakes and a direct relationship appears improbable.

Mots-clés de l'auteur: Induced seismicity ; stress management ; hydraulic fracturing ; reservoir stimulation ; Enhanced Geothermal Systems ; state of stress ; stress preconditioning ; reservoir depletion ; carbon storage ; fluid injection.

This research describes the importance of a hydro-geomechanical coupling in the geologic sub-surface environment from fluid injection at geothermal plants, large-scale geological CO2 sequestration for climate mitigation, enhanced oil recovery, and hydraulic fracturing during wells construction in the oil and gas industries. A sequential computational code is developed to capture the multiphysics interaction behavior by linking a flow simulation code TOUGH2 and a geomechanics modeling code PyLith. Numerical formulation of each code is discussed to demonstrate their modeling capabilities. The computational framework involves sequential coupling, and solution of two sub-problems- fluid flow through fractured and porous media and reservoir geomechanics. For each time step of flow calculation, pressure field is passed to the geomechanics code to compute effective stress field and fault slips. A simplified permeability model is implemented in the code that accounts for the permeability of porous and saturated rocks subject to confining stresses. The accuracy of the TOUGH-PyLith coupled simulator is tested by simulating Terzaghi's 1D consolidation problem. The modeling capability of coupled poroelasticity is validated by benchmarking it against Mandel's problem. The code is used to simulate both quasi-static and dynamic earthquake nucleation and slip distribution on a fault from the combined effect of far field tectonic loading and fluid injection by using an appropriate fault constitutive friction model. Results from the quasi-static induced earthquake simulations show a delayed response in earthquake nucleation. This is attributed to the increased total stress in the domain and not accounting for pressure on the fault. However, this issue is resolved in the final chapter in simulating a single event earthquake dynamic rupture. Simulation results show that fluid pressure has a positive effect on slip nucleation and subsequent crack propagation. This is confirmed by running a sensitivity analysis that shows an increase in injection well distance results in delayed slip nucleation and rupture propagation on the fault.