CO2 reactive transport simulations in an Italian deep saline aquifer

Abstract

In this study numerical simulations of reactive transport in an off-shore deep saline aquifer for the geological sequestration of carbon dioxide are presented and discussed. The main goals are to assess: i) the CO2 injection impact in the reservoir and ii) the cap-rock stability, both being strategic requisites for feasibility studies that are about to be started in Italy. The stratigraphic succession is characterized by a sedimentary succession: from Triassic anhydrites to Jurassic Tuscan calcareous units, up to Cretaceous calcarenites, belonging to the Liguride units, and Quaternary shallow marine sediments. Stratigraphic data from a deep well indicate that below an 1,800 m thick cap-rock, constituted by allochtonous marly calcarenites and clay marls, a regional deep saline aquifer is present. This aquifer, hosted in six Late Triassic to Early Jurassic formations, belonging to the Tuscan Nappe units, consists of porous limestone (mainly calcite) and marly limestone deposits at 1900-3100 m b.s.l. A common problem working with off-shore closed wells, where only the well-log information are available, is that to obtain reliable physico-chemical parameters (e.g. petrophysical and mineralogical) to be used for numerical simulations. Available site-specific data include only basic physical parameters such as temperature, pressure, and salinity of the formation waters. Bulk and modal mineralogical composition were obtained after sampling each formation in contiguous on-shore zones. Mineralogy was determined by X-Ray diffraction analysis coupled with Rietfield refinement. The latter was performed using Maud v2.2. The surface reactive area of minerals was assumed as geometric area of a truncated sphere calculated on the basis of Scanning Electronic Microscopy analysis. Porosity and permeability were inferred by the well-log data along with the use of boundary conditions such as surficial measurements and temperature profiles. The chemical composition of the aquifer pore water is unknown. As a consequence, this was calculated by batch modeling, assuming thermodynamic equilibrium between minerals and a NaCl (0.45 M) equivalent brine at reservoir conditions (up to 118 °C and 300 bars). The reconstructed dataset represented the base of numerical simulations to evaluate the potential geochemical impact of CO2 storage and to quantify water-gas-rock reactions. Three dimensional simulations were performed by the TOUGHREACT code via the implementation to the source code and the correction of the chemical parameters at the theoretical CO2 injection pressure. A re-interpretation of the available seismic reflection data was carried out to: i) define the 3D geometry, and ii) evaluate the volume of the geological structure potentially suitable for CO2 storage. In particular the main surfaces where physicochemical modeling was applied, i.e. the top and the bottom of the cap-rock units and the spill point surface, to better define the 3D geometry of the potential injection reservoir, were reconstructed. Reactive transport simulations were conducted under multiphase advection, aqueous diffusion, gas phase participation in multiphase fluid flow and geochemical reaction in non-isothermal conditions. Feedbacks between flow and geochemical processes were taken into account to evaluate changes in porosity and permeability as kinetic reactions were proceeding. Twenty years of CO2 injection at the rate of 1.5 Mt/year were simulated, whereas water-gas-rock interactions between CO2-rich brines and minerals over a period of 100 years were performed. Preliminary results suggest that injected CO2 can safely be retained in the reservoir by mineral trapping and that the cap-rock can be considered as efficient barrier.