Assessing the possibility of CO2 leakage is one of the major challenges for geological carbon sequestration. Injected CO2 can react with wellbore cement, which can potentially change cement composition and transport properties. In this work, we develop a reactive transport model based on experimental observations to understand and predict the property evolution of cement in direct contact with CO2-saturated brine under diffusion-controlled conditions. The model reproduced the observed zones of portlandite depletion and calcite formation. Cement alteration is initially fast and slows down at later times. This work also quantified the role of initial cement properties, in particular the ratio of the initial portlandite content to porosity (defined here as φ), in determining the evolution of cement properties. Portlandite-rich cement with large φ values results in a localized “sharp” reactive diffusive front characterized by calcite precipitation, leading to significant porosity reduction, which eventually clogs the pore space and prevents further acid penetration. Severe degradation occurs at the cement–brine interface with large φ values. This alteration increases effective permeability by orders of magnitude for fluids that preferentially flow through the degraded zone. The significant porosity decrease in the calcite zone also leads to orders of magnitude decrease in effective permeability, where fluids flow through the low-permeability calcite zone. The developed reactive transport model provides a valuable tool to link cement–CO2 reactions with the evolution of porosity and permeability. It can be used to quantify and predict long-term wellbore cement behavior and can facilitate the risk assessment associated with geological CO2 sequestration.

Understanding long-term property evolution of cement fractures is essential for assessing well integrity during geological carbon sequestration (GCS). Cement fractures represent preferential leakage pathways in abandoned wells upon exposure to CO 2-rich fluid. Contrasting self-sealing and fracture opening behavior have been observed while a unifying framework is still missing. Here we developed a processbased reactive transport model that explicitly simulates flow and multi-component reactive transport in fractured cement by reproducing experimental observation of sharp flow rate reduction during exposure to carbonated water. The simulation shows similar reaction network as in diffusion-controlled systems without flow. That is, the CO 2-rich water accelerates the portlandite dissolution, releasing calcium that further reacted with carbonate to form calcite. The calibrated model was used for CO 2-flooding numerical experiments in 250 cement fractures with varying initial hydraulic aperture (b) and residence time (τ) defined as the ratio of fracture volume over flow rate. A long τ leads to slow replenishment of carbonated water, calcite precipitation, and self-sealing. The opposite occurs when τ is small with short fracture and fast flow rates. Simulation results indicate a critical residence time  cthe minimum τ required for selfsealing-divides the conditions that trigger the diverging opening and self-sealing behavior. The  c value depends on the initial aperture size through 42 9.8 10 0.254 c bb       . Among the 250 simulated fracture cases, significant changes in effective permeabilityself-healing or openingtypically occur within hours to a day, thus providing supporting argument for the extrapolation of short-term laboratory observation (hours to months) to long-term prediction at relevant GCS time scales (years to hundreds of years).

The implementation of acid gas cosequestration requires investigation of the potential for acid gas leakage along existing wellbores at sequestration sites. In this study, the interaction between pozzolan-amended wellbore cement (35 vol % pozzolan/65 vol % cement, hereafter referred to as 35:65 sample) and acid gas (e.g., a mixture of CO 2 and H 2 S) was simulated using the reactive transport code CrunchFlow. The model was applied to describe, interpret, and extrapolate scanning electron microscopy-backscattered electron (SEM-BSE) and scanning electron microscopy-energy dispersive spectroscopy (SEM-EDS) results on pozzolan-amended cement samples exposed to a 1 wt % NaCl solution saturated with an acid gas mixture of 21 mol % H 2 S and 79 mol % CO 2 under the temperature of 50 °C and pressure of 150 bar. Simulation outputs included calcite volume percentage, total Ca and S weight percentages in the solid phase, porosity, and effective permeability from the surface to the interior of pozzolan-amended wellbore cement. The model reproduced the observed calcite zone formed in the brine-cement interface region of the sample after 2.5 days of exposure. The model also predicted that the calcite layer became dense (calcite vol % in the layer reached 55%) after 90 days of exposure, consistent with the experimental observation. C−S−H was the primary Ca 2+ source to form the calcite layer, followed by C3S and Ca(OH) 2 . The main observed products of reaction between the 35:65 sample and H 2 S were pyrite and ettringite. Pyrite was primarily formed within 0.5 mm from the brine-cement interface; ettringite mainly formed within 1 mm from the interface. The model simulated these reactions that only the interface region (up to 2 mm distance from the surface) of the 35:65 sample became porous after 30 years of exposure. However, this narrow porous region could still serve as a migration pathway for acid gas, which was indicated by the increase in effective parallel permeability values determined from the simulation results. Those results show consistency with results of neat cement samples exposed under similar conditions. An increase in H 2 S content (in the range of 0 mol % to 40 mol %) results in more dissolution of Ca-bearing minerals in cement and more precipitation of calcite. Overall, this study indicates that an increase of porosity and permeability of pozzolan-amended wellbore cement at the cement interface with brine saturated with CO 2 and H 2 S can cause significant changes in effective permeability of the cement.

No abstract