a b s t r a c tA significant amount of theoretical, numerical and observational work has been published focused on various aspects of capillary trapping in CO 2 storage since the IPCC Special Report on Carbon Dioxide Capture and Storage (2005). This research has placed capillary trapping in a central role in nearly every aspect of the geologic storage of CO 2 . Capillary, or residual, trapping -where CO 2 is rendered immobile in the pore space as disconnected ganglia, surrounded by brine in a storage aquifer -is controlled by fluid and interfacial physics at the size scale of rock pores. These processes have been observed at the pore scale in situ using X-ray microtomography at reservoir conditions. A large database of conventional centimetre core scale observations for flow modelling are now available for a range of rock types and reservoir conditions. These along with the pore scale observations confirm that trapped saturations will be at least 10% and more typically 30% of the pore volume of the rock, stable against subsequent displacement by brine and characteristic of water-wet systems. Capillary trapping is pervasive over the extent of a migrating CO 2 plume and both theoretical and numerical investigations have demonstrated the first order impacts of capillary trapping on plume migration, immobilisation and CO 2 storage security. Engineering strategies to maximise capillary trapping have been proposed that make use of injection schemes that maximise sweep or enhance imbibition. National assessments of CO 2 storage capacity now incorporate modelling of residual trapping where it can account for up to 95% of the storage resource. Field scale observations of capillary trapping have confirmed the formation and stability of residually trapped CO 2 at masses up to 10,000 tons and over time scales of years. Significant outstanding uncertainties include the impact of heterogeneity on capillary immobilisation and capillary trapping in mixed-wet systems. Overall capillary trapping is well constrained by laboratory and field scale observations, effectively modelled in theoretical and numerical models and significantly enhances storage integrity, both increasing storage capacity and limiting the rate and extent of plume migration.

We provide a comprehensive experimental study of steady state, drainage relative permeability curves with CO 2 -brine and N 2 -deionized water, on a single Bentheimer sandstone core with a simple two-layer heterogeneity. We demonstrate that, if measured in the viscous limit, relative permeability is invariant with changing reservoir conditions, and is consistent with the continuum-scale multiphase flow theory for water wet systems. Furthermore, we show that under capillary limited conditions, the CO 2 -brine system is very sensitive to heterogeneity in capillary pressure, and by performing core floods under capillary limited conditions, we produce effective relative permeability curves that are flow rate and fluid parameter dependent. We suggest that the major uncertainty in past observations of CO 2 -brine relative permeability curves is due to the interaction of CO 2 flow with pore space heterogeneity under capillary limited conditions and is not due to the effects of changing reservoir conditions. We show that the appropriate conditions for measuring intrinsic or effective relative permeability curves can be selected simply by scaling the driving force for flow by a quantification of capillary heterogeneity. Measuring one or two effective curves on a core with capillary heterogeneity that is representative of the reservoir will be sufficient for reservoir simulation.

The current conceptual picture of steady-state multiphase Darcy flow in porous media is that the fluid phases organize into separate flow pathways with stable interfaces. Here we demonstrate a previously unobserved type of steady-state flow behavior, which we term "dynamic connectivity," using fast pore-scale X-ray imaging. We image the flow of N 2 and brine through a permeable sandstone at subsurface reservoir conditions, and low capillary numbers, and at constant fluid saturation. At any instant, the network of pores filled with the nonwetting phase is not necessarily connected. Flow occurs along pathways that periodically reconnect, like cars controlled by traffic lights. This behavior is consistent with an energy balance, where some of the energy of the injected fluids is sporadically converted to create new interfaces.steady state | pore-scale imaging | immiscible two-phase flow | dynamic connectivity | geologic CO 2 storage T he definition of relative permeability is based on a conceptual model where each phase occupies its own static, connected fraction of the pore space (1-3). Observations of two-phase flow in beadpacks and micromodels at low flow rates show that the wetting-and nonwetting-phase fluids flow through their own network of separate stable pathways. Increasing the saturation of one phase increases the number of channels occupied by one fluid and decreases the number occupied by the other. The nonwetting phase generally occupies the large pores, whereas the wetting phase occupies the small pores and indentations in the surface of the solid (4-7).When the balance of capillary and viscous forces is modified such that viscous forces begin to dominate at the pore scale, the nonwetting phase can be pushed out of the pore space and flow can occur through the advection of disconnected ganglia (8, 9). The onset of ganglion motion is well understood, and this flow behavior has been observed during steady-state flow in model pore spaces (1-3, 7, 10-13).The key observation during slow steady-state flow, typical of most displacements in natural settings, such as during hydrocarbon recovery or CO2 storage, is that the pathway, once established, remains stable (2, 4-7). Flow patterns for different wettabilities have also been quantified in micromodel studies (14). Again, once flow pathways were established, they were not observed to change. However, there are no studies thus far that have investigated this flow behavior in the pore space of natural rocks.Developments in micro X-ray computed tomography (X-ray CT) mean that pore-scale phenomena can now be observed in rocks at reservoir conditions in both laboratory-based scanners and synchrotron beamlines (8,9,(15)(16)(17). This technological advance has allowed, for instance, observations of residual trapping (18) and wetting behavior (19).Fast synchrotron tomography allows the acquisition of images in under 60 s. Many unsteady-state, dynamic, pore-scale processes have been imaged, including Haines jumps (15), capillary pressure changes during reservoir cond...

In this work, we analyze the characterization of drainage multiphase flow properties on heterogeneous rock cores using a rich experimental data set and mm-m scale numerical simulations. Along with routine multiphase flow properties, 3-D submeter scale capillary pressure heterogeneity is characterized by combining experimental observations and numerical calibration, resulting in a 3-D numerical model of the rock core. The uniqueness and predictive capability of the numerical models are evaluated by accurately predicting the experimentally measured relative permeability of N 2 -DI water and CO 2 -brine systems in two distinct sandstone rock cores across multiple fractional flow regimes and total flow rates. The numerical models are used to derive equivalent relative permeabilities, which are upscaled functions incorporating the effects of submeter scale capillary pressure. The functions are obtained across capillary numbers which span four orders of magnitude, representative of the range of flow regimes that occur in subsurface CO 2 injection. Removal of experimental boundary artifacts allows the derivation of equivalent functions which are characteristic of the continuous subsurface. We also demonstrate how heterogeneities can be reorientated and restructured to efficiently estimate flow properties in rock orientations differing from the original core sample. This analysis shows how combined experimental and numerical characterization of rock samples can be used to derive equivalent flow properties from heterogeneous rocks.