Thermo-Hydro-Mechanical Impacts of Carbon Dioxide (CO2) Injection in Deep Saline Aquifers

Abstract

[EN] Coupled thermo-hydro-mechanical (THM) effects related to geologic carbon storage should be understood and quantified in order to convince the public that carbon dioxide (CO2) injection is safe. This Thesis aims to improve such understanding by developing methods to: evaluate the CO2 plume geometry and fluid pressure evolution; define a field test to characterize the maximum sustainable injection pressure and the hydromechanical (HM) properties of the aquifer and the caprock; and propose an energy efficient injection concept that improves the caprock mechanical stability in most geological settings due to thermo-mechanical effects. First, we investigate numerically and analytically the effect of CO2 density and viscosity variability on the position of the interface between the CO2-rich phase and the formation brine. We introduce a correction to account for this variability in current analytical solutions. We find that the error in the interface position caused by neglecting CO2 compressibility is relatively small when viscous forces dominate. However, it can become significant when gravity forces dominate, which is likely to occur at late times and/or far from the injection well. Second, we develop a semianalytical solution for the CO2 plume geometry and fluid pressure evolution, accounting for CO2 compressibility and buoyancy effects in the injection well. We formulate the problem in terms of a CO2 potential that facilitates solution in horizontal layers, in which we discretize the aquifer. We find that when a prescribed CO2 mass flow rate is injected, CO2 advances initially through the top portion of the aquifer. As CO2 pressure builds up, CO2 advances not only laterally, but also vertically downwards. However, the CO2 plume does not necessarily occupy the whole thickness of the aquifer. Both CO2 plume position and fluid pressure compare well with numerical simulations. Third, we study potential failure mechanisms, which could lead to CO2 leakage, in an axysimmetric horizontal aquifercaprock system, using a viscoplastic approach. Simulations illustrate that, depending on boundary conditions, the least favorable situation may occur at the beginning of injection. However, in the presence of low-permeability boundaries, fluid pressure continues to rise in the whole aquifer, which may compromise the caprock integrity in the long-term. Next, we propose a HM characterization test to estimate the HM properties of the aquifer and caprock at the field scale. We obtain curves for overpressure and vertical displacement as a function of the volumetric strain term obtained from a dimensional analysis of the HM equations. We can then estimate the values of the Young¿s modulus and the Poisson ratio of the aquifer and the caprock by introducing field measurements in these plots. Results indicate that induced microseismicity is more likely to occur in the aquifer than in the caprock. The onset of microseismicity in the caprock can be used to define the maximum sustainable injection pressure to ensure a safe permanent CO2 storage. Finally, we analyze the thermodynamic evolution of CO2 and the THM response of the formation and the caprock to liquid (cold) CO2 injection. We find that injecting CO2 in liquid state is energetically more efficient than in supercritical state because liquid CO2 is denser than supercritical CO2. Thus, the pressure required at the wellhead is much lower for liquid than for gas or supercritical injection. In fact, the overpressure required at the aquifer is also smaller because a smaller fluid volume is displaced. The temperature decrease close to the injection well induces a stress reduction due to thermal contraction of the media. This can lead to shear slip of pre-existing fractures in the aquifer for large temperature contrasts in stiff rocks, which could enhance injectivity. In contrast, the mechanical stability of the caprock is improved in stress regimes where the maximum principal stress is the vertical.