Imperial College Consortium on Pore - Scale Modelling Yearly progress and final report January 3 rd 2011

Displacement experiments using the porous plate method were conducted on water-wet sandstones to measure the capillary trapping of oil by waterflooding as a function of its saturation after primary drainage. Three sandstone samples ranging in porosity from 12.2% to 22.1% were considered. Experiments on two samples were conducted at an elevated temperature and back-pressure of 343K and 9MPa respectively; experiments on the third sample were conducted at ambient conditions (292 to 297K and 0.06 to 0.17MPa). Residual oil saturation increases monotonically, but with a decreasing gradient, as initial saturation increases. The dependence of residual saturation on initial saturation is accurately predicted by a two-phase porenetwork simulator when a uniform distribution of intrinsic contact angles between 35° and 65° is assumed. The networks were extracted from X-ray microtomography images of small samples of the same rock as those used in the experiments. The laboratory measurements are also accurately described by trapping models proposed by Land (1968) and Spiteri et al. (2008). The residual saturations we measured were higher than in previous displacement experiments, suggesting, for example, that capillary trapping may be an effective way to store substantial quantities of carbon dioxide in aquifers. Introduction The trapping of non-wetting phase in a porous medium as discontinuous pore-scale droplets by capillary forces, or capillary trapping, has been studied extensively because of its importance to oil recovery and contaminant remediation. In these applications, the motivation is the extraction of the trapped phase. In contrast, in the context of geological carbon storage, the objective is to maximize trapping. Carbon dioxide (CO2) is injected into a target geological formation, forming a continuous plume. As the injected CO2 is driven upwards by buoyancy, ambient groundwater will flow into its wake to replace it. This replacement of CO2 by groundwater behind the rising CO2 plume can be regarded as a re-imbibition process. Accordingly, a portion of the migrating CO2 will be rendered immobile within the pores of the rock by capillary forces, and will no longer be at risk of leakage to the atmosphere provided that local conditions remain unchanged. Further, this process can be enhanced through injection of additional brine (Juanes et al. 2006; Qi et al. 2009). While many studies report laboratory measurements of capillary trapping in porous media (e.g., Abrams 1975; Agbalaka et al. 2009; Chatzis and Morrow 1984), literature that investigate the dependence of capillary trapping on initial non-wetting phase saturations are more limited. If we further restrict our consideration to experiments in which drainage and imbibition were achieved by displacement of phases within the porous medium, which is the process relevant to carbon storage, and to experiments on uniformly-wet porous media, existing data are limited to those listed in Table 1; a more extensive review of the literature is provided by Pentland et al. (2010). This paper complements these data with new oil/brine coreflooding experiments at flow rates representative of field conditions. Residual saturation data have traditionally been compared to a number of empirical trapping models. In this paper, our data will be compared to two such models. The first, which is the most widely used in the literature and commercial simulators, was proposed by Land (1968):       − − + = wc max nwr nwi nwi nwr 1 1 1 1 S S S S S , (1) Imperial College Consortium on Pore-Scale Modelling January 2011 5 where Snwi and Snwr are the non-wetting phase saturation after primary drainage and after secondary imbibition, respectively, max nwr S is the maximum Snwr, and Swc is the connate wetting phase saturation. The second, a quadratic relationship between initial and residual saturation, was proposed by Spiteri et al. (2008): 2 nwi nwi nwr S S S β α − = , (2) where α and β are fitting constants. Spiteri et al. (2008) report α and β that best-fit their pore-scale simulations of a Berea sandstone using macroscopic intrinsic contact angles ranging from θ = 20° to 160°. Source phases porous media porosity permeability [m] symbol in Fig. 1 Kleppe et al. 1997 air/oil artificial consolidated medium 0.43 4.7×10 grey — Al Mansoori et al. 2010 air/brine sand pack 0.37 ± 0.002 (320±3)×10 green + Pentland et al. 2010 oil/brine sand pack 0.37 ± 0.002 (320±3)×10 green × Pickell et al. 1966 mercury/air; oil/water Dalton sandstone 0.286 4.17×10 grey ○ 0.248 1.90×10 grey □ 0.291 4.14×10 grey ◊ 0.286 2.55×10 grey ∆ 0.28 2.07×10 grey + Table 1: Studies that report residual saturation as a function of initial non-wetting phase saturation. Only studies that achieved drainage and imbibition by displacement are listed. Displacement experiments were conducted to measure the capillary trapping of oil by waterflooding as a function of its saturation after primary drainage in three water-wet sandstone cores. The results are compared with the two empirical models given above, and with simulations of the same sequence of displacements using the two-phase flow pore-network model developed by Valvatne and Blunt (2004) with networks previously extracted from X-ray microtomography images of samples from the same source rock as the cores. Experimental Procedure Three sandstone samples with porosities 0.221 (Berea), 0.122 (Clashach), and 0.156 (Stainton) were considered. Basic rock properties are summarized in Table 2. All three rocks are considered to be strongly water-wet. All cores were 3.8cm (1.5in.) in diameter and 7.5 to 7.7cm (3.0in.) in length. Experiments were conducted at ambient conditions in the Stainton core and at elevated temperature and pressure (ETP) in Clashach and Berea cores. In the ambient-condition runs, the temperature varied between 292 and 297K (19 to 24°C) and a backpressure of 0.06 to 0.17MPa was applied. The ETP experiments were conducted inside an air bath at 343K (70°C) with a back-pressure of 9MPa. The non-wetting phase was n-decane in the ETP experiments and n-octane in the ambient-condition experiments. This use of different non-wetting liquids is justified, as the estimated interfacial tensions of noctane/water and n-decane/water at the relevant experimental conditions only differ by 7% (Table 2). The wetting phase was aqueous 5wt.%-sodium chloride, 1wt.%-potassium chloride synthetic brine. In each ambientcondition run, the density of the fluids was measured at 293.2K using a digital density-meter (Anton Paar DMA48). The average density of brine and n-octane across these runs were 1042kg m-3 and 702.7kg m-3, respectively. The density of the brine and n-decane were not measured at ETP, but are expected to be approximately 1030kg m-3 (Rogers et al. 1982) and 700kg m-3 (Banipal et al. 1991; Lee and Ellington 1965). Imperial College Consortium on Pore-Scale Modelling January 2011 6 Berea Clashach Stainton Effective porosity (± standard error) 0.221 ± 0.001 0.122 ± 0.003 0.156 ± 0.001 Brine permeability [m] 5×10 8×10 3.8×10 Wetting phase 5wt.% NaCl, 1wt.% KCl aqueous solution Non-wetting phase n-decane n-octane Interfacial tension [mN m] 48.3 51.64±0.04 Wetting phase viscosity [Pa s] 4.554×10 1.0903×10 Non-wetting phase viscosity [Pa s] 5.47×10 5.08×10 Table 2: Petrophysical properties of the core samples and fluid properties used in pore-network simulations. a Vinogradov et al. 2010. b Zeppieri et al. 2001; linearly extrapolated to 343K. c Zeppieri et al. 2001; 293.2 ± 0.1K. d Kestin, Khalia, and Correia 1981; 5.8wt.% NaCl aqueous solution at 10.0MPa, 343.2K. e Kestin, Khalia, and Correia 1981; 5.8wt.% NaCl aqueous solution at 100kPa, 293.2K. f Lee and Ellington 1965; 9480kPa, 344K. g CRC Handbook of Chemistry and Physics; 298K. Coreflooding experiments. All experiments were conducted in custom-made horizontal Hassler-type core holders (radial confining pressure). A high-precision syringe pump (Teledyne ISCO 1000D or 500D) was used for all wetting and non-wetting fluid displacements. The capillary number was maintained at or below 3×10 during all displacement steps. All cores were cleaned before use by standard Soxhlet extraction with a mixture of methanol and toluene, oven-dried, and weighed. Initial water saturation. The test core was saturated with de-gassed brine at experimental conditions inside the core holder. A minimum of five pore volumes of brine were injected through the core against back pressure. In ambient-condition runs, the core was subsequently removed from the Hassler cell, weighed, and then reinserted into the cell. The Hassler cell, containing the brine-saturated core, was weighed. The pore volume was determined from the increase in mass of the core in ambient-condition runs, and from the increase in mass of the Hassler cell in ETP runs. Primary drainage. The porous plate method, in which a water-wet disk placed immediately downstream of the core retains the non-wetting phase inside the core, was used during primary drainage. With this method, equilibrium corresponds to uniform pressure distribution across the core in each of the phases, with the difference in the equilibrium pressure of the two phases corresponding to the capillary pressure. Capillary end effects are eliminated, and phase saturations are assumed to be uniform across the length of the core. The non-wetting phase was injected into the core against a back-pressure using one of two methods: at constant pressure or at constant flow rate. The former was used for all experiments with Stainton, for drainage at a capillary pressure Pc > 1000kPa in Berea, and for drainage at 50kPa < Pc < 500kPa in Clashach. Here, equilibrium was considered achieved when brine production ceased and the volumetric rate of oil injection by the pump reached a constant (the leakage rate). Capillary pressure, and hence initial saturation, was varied by changing the injection pressure. For all other runs at ETP, a pre-determined volume of oil was injected at a constant flow rate corresponding to capillary numbers between 1.6×10 and 3.3×10. Once the volume necessary to achieve the target Snwi was injected, the system was allowed to equilibrate, and capillary pressure was measured. The duration of primary drainage ranged from 23 to 50 hours for Berea, 22 to 70 hours for Clashach, and 153 to 182 hours for Stainton. Snwi was determined from the volume of the injected non-wetting phase, the decrease in the mass of the Hassler cell containing the core from its brine-saturated state prior to oil injection and, for ambient-condition runs additionally, the volume and the mass of the effluent brine. Waterflooding. Subsequently, the core was flooded with brine in the opposite direction as the oil injection. A minimum of five pore volumes of brine were injected in each ETP run; a minimum of ten pore volumes were injected in the ambient-condition runs. Residual saturation was determined from the decrease in the mass of the Hassler cell containing the core from its brine-saturated state for both ambient-condition and ETP runs and, in addition, from the increase in mass of the Hassler cell during secondary imbibitions for ETP runs and from the decrease in the mass of the core relative to its dry state for ambient condition runs. Imperial College Consortium on Pore-Scale Modelling January 2011 7 Pore-network simulations. A corresponding set of simulations of primary drainage and waterflooding were performed using the two-phase flow pore-network simulator developed by Valvatne and Blunt (2004). Input parameters were matched with experimental conditions where possible. We used networks previously extracted (Table 3), using Dong and Blunt (2009)’s algorithm, from X-ray microtomography scans of samples from the same source as the cores in which the laboratory experiments were conducted (Fig. 1). The voxel resolution of the scans was 5.789μm × 5.789μm × 5.789μm for Berea and Clashach and 17.578μm × 17.578μm × 17.578μm for Stainton. Interfacial tension, brine viscosity, and oil viscosity were matched with experimental conditions, as listed in Table 2. A receding contact angle of 0° was assumed everywhere in the domain to describe a strongly water-wet state during primary drainage. For waterflooding, two values for the macroscopic intrinsic contact angle were considered: θ = 15° and 50°. The local intrinsic contact angle was randomly assigned, with equal probability, a value in the range θ ± 15°. Saturations and relative permeabilities were computed in a sequence of increasing and decreasing capillary pressures for drainage and waterflooding, respectively.