Reservoir scale reactive-transport modeling of a buoyancy-controlled CO2 plume with impurities (SO2, NO2, O2)

Abstract

A demonstration project for the geological storage of CO 2 is currently being considered in the deep Precipice Sandstone formation of the Surat Basin, Queensland, Australia. Because of the presence of potential fresh water resources in this formation, a reservoir-scale two-dimensional reactive-transport model was developed to assess temporal and spatial changes in water quality imposed by co-injecting CO 2 with SO 2 , NO 2 , and O 2 at this location. The model shows that because the injection rate is low (60 tons/year), flow is buoyancy-dominated and under these conditions the predicted CO 2 flow pattern is quite sensitive to fine-scale heterogeneities and the resolution of the numerical mesh. The model also shows that SO 2 and NO 2 readily partition into the aqueous phase in close vicinity of their injection point, lowering pH somewhat beyond the acidification from CO 2 dissolution. Only O 2 under redox disequilibrium conditions is modeled to persist in the CO 2 plume away from the injection point, however at sub-ppm levels. This modeling effort demonstrates acidification near the wellbore due to the preferential stripping of gas impurities, and accumulation of CO 2 around a lithostratigraphic boundary above the target formation, where relatively rapid mineral dissolution (muscovite, chlorite and calcite) and precipitation (ankerite, kaolinite and chalcedony) occur.

A perspective on applied geochemistry in porous media: Reactive transport modeling of geochemical dynamics and the interplay with flow phenomena and physical alteration

“…Even a trace amount of impurities can affect the solubility of CO 2 and cause deviation of modeling results from the actual value. (Spycher et al 2019) revealed that when impurities like SO 2 and NO 2 are incorporated in the model, they readily partition into the aqueous phase close to the injection well and further lower the pH, compared with the case of CO2 acidification only. As for pH buffering capacity, a host rock rich in feldspars and clay minerals can hinder the drop of pH after CO 2 injection, which slows down the reaction between CO 2 and the host rock (or caprock).…”

Reactive Transport Modeling and Sensitivity Analysis of CO2–Rock–Brine Interactions at Ebeity Reservoir, West Kazakhstan

“…To obtain the unique environmental parameters of the storage sites, geochemical and hydrogeological modeling has been considered essential as it can provide comprehensive knowledge about long-term interactions, such as aluminosilicates' dissolution/precipitation [13], the behavior of minerals, their distributions and abundance, the effect of the relevant geological factors, and the impurity of injected CO 2 on its storage performance. Recently, the simulation of CO 2rock-brine interactions has been implemented in several research projects and studies [14][15][16][17][18][19]. PHREEQC v3 has been widely used for reactive transport modeling (RTM).…”

Comparison of geological models for the simulation of CO2 migration: a case study in Ordos, China

“…For example, Spycher et al performed simulations on detailed vertical 2D cross-sections that accounted for geological heterogeneities on a relatively small scale, while keeping the size of the numerical mesh to a manageable size to minimize computational time. 20 Moreover, Kumar et al simulated the injection of CO 2 into a heterogeneously layered carbonate formation for a period of 50 years using a geological model with different horizontal and vertical permeabilities, which employed an average value for each layer. 21 Xu et al performed six numerical simulations of a simplified 1D radial region surrounding the injection well, whereby flow and transport modeling in the geologic media was based on space discretization by means of integral finite differences.…”

“…24 Furthermore, Xie et al reported that the Chenjiacun site formation was homogenous with respect to permeability and porosity in any given layer, but heterogenous across the different layers. 25 In these previous studies, the simulated domain used radial grids, [20][21][22][23][24][25] and the majority did not focus on the interpolation methods used in the geological model. However, rectangular grids used in geological models are encrypted in the vertical direction.…”