1. Upscaling of two-phase flow processes in CO 2 geological storage Orlando Silva (1), Insa Neuweiler 2), Marco Dentz (3,4), Jesús Carrera (3,4) and Maarten Saaltink (4) (1) Fundación Ciudad de la Energía (CIUDEN), Programa de Almacenamiento Geológico de CO 2, Ponferrada (León), España. (2) Institute for Fluid Mechanics and Environmental Physics in Civil Engineering, University of Hanover, Hanover, Germany. (3) GHS, Institute of Environmental Assessment and Water Research (IDÆA, CSIC), Barcelona, España. (4) GHS, Department of Geotechnical Engineering and Geosciences, Technical University of Catalonia (UPC-Barcelona Tech), Barcelona, España. INTRODUCTION Heterogeneity of physical and chemical properties is an essential feature of geological media that affects the major spreading and trapping mechanisms of CO 2. Heterogeneity acts on the multiphase flow properties of the CO 2 -brine system and can lead to (i) trapping of brine behind the CO 2 phase and (ii) increased spread of the CO 2 -brine interface. Increasing the CO 2 -brine contact area leads to increased CO 2 dissolution efficiency. The mixing of the resulting denser CO 2 -rich water and the reservoir water is influenced by dispersion and the interaction with spatial heterogeneity and buoyancy effects. While heterogeneity can lead to increased spreading and mixing of waters with different chemical compositions, chemical reaction rates in heterogeneous media can be much smaller than those obtained in laboratory settings. These effects, along with the influence of capillarity are in general disregarded in large-scale reservoir models used, e.g., in petroleum applications. However, the impact of these heterogeneity and capillarity related processes is critical for assessing the long-term behavior of the CO 2 geological storage. THE PROBLEM Current multiphase flow and reactive transport models do not take into account the impact of heterogeneity on front spreading and mass transfer between high and low permeability zones of the heterogeneous medium and the impact of physical heterogeneity and chemical heterogeneity on chemical reactions rates. Effective equations are available only for single phase reactive transport and under development for multiphase flow. In the present work we aim specifically at the upscaling of the two- phase flow dynamics. Figure 2. 2D two-phase flow through a heterogeneous storage aquifer. (a) Heterogeneity in hydraulic conductivity is represented by a Gaussian random field. (b) Numerical simulation: evolution of CO 2 -rich phase saturation in the aquifer. … 12 j N … Mobile zone: “background material” Immobile or less mobile zones (“inclusions”) FjFj FNFN F2F2 F1F1 Mass exchange Heterogeneous media Figure 1. Conceptual model used to approach multiphase flow in heterogeneous media. METHODOLOGY: EQUIVALENT MRMT MODEL We aim at the quantification of the impact of spatial heterogeneity on the large scale two-phase flow behavior in terms of effective two-phase flow equations. 1. Heterogeneity in the hydraulic conductivity of the storage aquifer is accounted through, e.g., a Gaussian random field (Figure 2a). 2. Numerical simulation (IMPES coded in MatLab) of 2D two-phase flow in heterogeneous media (Figure 2b). 3. An equivalent model of 1D two-phase flow in homogenous media with MRMT (Figure 1) and capillarity contrast between mobile and immobile zones is used to describe the 2D heterogeneous results (Figure 3 and 4). 4. Reproducibility for other flow rates with upscaling of mass transfer coefficients (Figure 5). Governing equations CONCLUSIONS  A simple 1D homogeneous model with MRMT and capillarity contrast at mobile-immobile interface is able to describe two-phase flow in heterogeneous media.  Macrodispersion in the mobile zone term still has to be included.  Extension to heterogeneous multiphase flow is straightforward.  The present methodology could contribute significantly to the quantification of the heterogeneity-induced uncertainty of the predicted large scale multiphase flow and transport behavior. However, it is still far from developed and tested. ACKNOWLEDGMENTS This work is funded by the “Fundación Ciudad de la Energía” (CIUDEN), and by the European Union through the “European Energy Programme for Recovery”. Disclaimer. The sole responsibility of this publication lies with the author. The European Union is not responsible for any use that may be made of the information contained therein. UPSCALING APPROACH The impact of heterogeneity on the two-phase flow dynamics can be quantified in the framework of a multi-continuum approach, which subdivides the heterogeneous medium into subunits with basically homogeneous properties. Between the subunits, the contrasts, i.e., the heterogeneity, can be high. Based on this map of the medium heterogeneity, the local scale flow equations (that are defined on a support scale for which we assume local equilibrium) are averaged using the theory of Volume Averaging and Homogenization Theory. This approach allows for the quantification of mass exchange between different regions of the medium and does accounts systematically for local scale non-equilibrium and thus for the complex flow dynamics in highly heterogeneous and fractured media. The mass exchange between mobile (background material) and immobile (inclusions) zones is taken into account by using a Multi-Rate Mass Transfer (MRMT) model. Memory function Figure 3. Comparison between the vertically averaged saturations in the heterogeneous media with those derived from the effective MRMT model. Figure 4. CO 2 -rich phase saturation profiles at t = 30 d for different CO 2 injection flow rates. Figure 4. Evolution of mean CO 2 -rich phase saturation at 10.3 m from the injection well. Inclusion Background material Mobile zone - VG Relative permeabilities Retention functions Mass transf. coeffs. upscaling Immobile - BC Inclusions Background material PcPc Boundary and initial conditions non-flow P c, Pa nS lr Mobile500000.70 Immobile400000.70 Memory function parameters slope t ini, d t end, d 1.01.5x10 -6 60 , kg/m 3 , cP CO 2 6350.018 Water9920.53 Simulation parameters