A numerical study of mineral alteration and self-sealing efficiency of a caprock for CO₂ geological storage

Abstract

Geochemical interactions of brine–rock–gas have a significant impact on the stability and integrity of the caprock for long-term CO₂ geological storage. Invasion of CO₂ into the caprock from the storage reservoir by (1) molecular diffusion of dissolved CO₂, (2) CO₂-water two-phase flow after capillary breakthrough, and (3) CO₂ flow through existing open fractures may alter the mineralogy, porosity, and mechanical strength of the caprock due to the mineral dissolution or precipitation. This determines the self-enhancement or self-sealing efficiency of the caprock. In this paper, two types of caprock, a clay-rich shale and a mudstone, are considered for the modeling analyses of the self-sealing and self-enhancement phenomena. The clay-rich shale taken from the Jianghan Basin of China is used as the base-case model. The results are compared with a mudstone caprock which is compositionally very different than the clay-rich shale. We focus on mineral alterations induced by the invasion of CO₂, feedback on medium properties such as porosity, and the self-sealing efficiency of the caprock. A number of sensitivity simulations are performed using the multiphase reactive transport code TOUGHREACT to identify the major minerals that have an impact on the caprock’s self-sealing efficiency. Our model results indicate that under the same hydrogeological conditions, the mudstone is more suitable to be used as a caprock. The sealing distances are barely different in the two types of caprock, both being about 0.6 m far from the interface between the reservoir and caprock. However, the times of occurrence of sealing are considerably different. For the mudstone model, the self-sealing occurs at the beginning of simulation, while for the clay-rich shale model, the porosity begins to decline only after 100 years. At the bottom of the clay-rich shale column, the porosity declines to 0.034, while that of mudstone declines to 0.02. The sensitive minerals in the clay-rich shale model are calcite, magnesite, and smectite-Ca. Anhydrite and illite provide Ca²⁺ and Mg²⁺ to the sensitive minerals for their precipitation. The mudstone model simulation is divided into three stages. There are different governing minerals in different stages, and the effect of the reservoir formation water on the alteration of sensitive minerals is significant.

Appendix

Precipitation and dissolution of minerals are kinetically controlled. The kinetic laws used here are derived from transition state theory [19]. Effective reaction rates can be expressed through the following general equation:

$$ {\text{rate}} = A_{m} k_{m} \left\{ {1 - \left( \frac{Q}{K} \right)^{\mu } } \right\}^{\eta } $$

where A _m is the specific surface area, k _m is the kinetic rate constant, Q is the ion activity product, K is the equilibrium constant for the specific mineral–water reaction, and μ and η are two constants which depend on experimental data; they are usually but not always taken equal to 1.

Kinetic rates may depend on the pH and also on concentrations of non-basis species. The temperature dependence of the reaction rate constant is expressed via an Arrhenius equation:

$$ k_{m} = k_{25} \exp \left[ {\frac{{ - E_{a} }}{R}\left( {\frac{1}{T} - \frac{1}{298.15}} \right)} \right] $$

where E _a is the activation energy, k ₂₅ is the rate constant at 25 °C, R is gas constant, and T is the absolute temperature.

The TOUGHREACT code used in this work can model effects of supersaturation and nucleation phenomena by suppressing precipitation of a mineral up to a given, positive saturation index (SI) value. However, due to lack of reliable data, we did not invoke a “supersaturation window,” instead, minerals will start to precipitate as soon as SI > 0. The initial effective surface area of minerals not present at the start of a simulation is assigned as equivalent to those spheres with radius 10⁻⁵ m [35].