Measurement and Correlation of Diffusion Coefficients for CO2 and Rich-Gas Applications

Abstract

Summary A novel in-situ method for measuring molecular diffusion coefficients of CO2 and other solvent gases in consolidated porous media at high pressure has been developed and is described. This technique is unique because visual observations and measurements of composition are not required. Experimental diffusion coefficients are reported for CO2 in decane up to 850 psia [5.86 MPa], for CO2 in 0.25 N NaCl brine up to 850 psia [5.86 MPa], and for ethane in decane up to 600 psia [4.14 MPa]. All tests were conducted in Berea cores saturated with liquid phase at 100°F [311 K]. Cores were oriented both vertically and horizontally to assess the effects of gravity-induced convection on the observed mass transfer. The experimental diffusion coefficients obtained from this study have also been correlated, together with literature data for methane, ethane, and propane, as a function of liquid viscosity and thermophysical properties of the diffusing gases. Introduction During both secondary and tertiary displacement of reservoir oils by CO2, the development of miscibility as a result of multicontact phase behavior strongly controls the ultimate recovery efficiency. On the microscopic (pore) scale, molecular diffusion is the mechanism by which intimate mixing of CO2 and oil occurs and on which the usual assumption of rapid local equilibrium is based for numerical simulation of CO2 flooding. Similarly, in rich-gas flooding, injection gases containing hydrocarbons of intermediate molecular weight develop miscibility with in-place oil by a multicontact condensing mechanism during which mass transfer as a result of diffusion plays an important role. Significant oil saturation may exist in dead-end pores or be trapped by water films in a porous medium during a CO2/water or rich-gas/water displacement when the medium is preferentially water-wet. This isolated oil saturation will remain largely unrecoverable unless the injection gas can effectively traverse the surrounding water barriers to contact and swell the trapped oil. At least one model1 exists that describes the pore-level diffusion processes occurring in gasfloods in the presence of water films. A major drawback to practical application of such a model is the lack of reliable diffusion coefficients for gas/oil/brine systems in porous media at high pressures. In addition, such diffusion data would also be valuable in modeling oil recovery from naturally fractured reservoirs. In this case, injection gas traveling through the fracture network may contact oil held within the rock matrix by a diffusional mass-transfer process. The literature contains data for diffusion of CO2 in liquid hydrocarbons at atmospheric pressure over a broad range of liquid viscosities2–5; the corresponding data for ethane in liquid hydrocarbons are much more limited.2 Data for CO2 in water at atmospheric pressure are abundant, dating back to the last century; however, for the purposes of this discussion, only the more recent data of Unver and Himmelblau6 and Thomas and Adams7 are considered. Few experimental studies of CO2 diffusion at high pressures have been reported. Grogan et al.8 measured diffusion coefficients for CO2 in pentane, decane, and hexadecane at 77°F [298 K] at pressures up to 820 psia [5.65 MPa]. Denoyelle and Bardon9 published CO2 diffusion coefficients for two stock-tank oils at reservoir conditions, but their results appear to have been dominated by convection. Studies reported by de Boer et al.10 concluded that observed CO2 diffusion rates in crude oil at reservoir conditions were consistent with calculated rates derived from diffusion coefficients measured at atmospheric pressure if there was no asphaltene precipitation. This result tends to support the work of Grogan et al.8 for simple binary CO2/oil systems. The lack of reliable high-pressure diffusion data is, at least in part, a consequence of the difficulty of making these measurements in the laboratory. Given sufficient data at reservoir conditions, it would be beneficial to determine whether or not a relationship between high-pressure and atmospheric-pressure data exist and, if so, on what physical parameters it depends. The situation is worse for diffusion of CO2 in water. To our knowledge, no diffusion data at reservoir conditions have been published. Grogan et al.8 stated that "the combination of low solubility of CO2 in the water phase compared to that in the oil phase and the thickness of the water blocking phase results in the controlling resistance to mass transfer being in the water phase." However, they did not report any actual measurements of the diffusion coefficient of CO2 in water at high pressure, but estimated a value from the Stokes-Einstein equation. In this paper, experimental molecular diffusion coefficients are reported for CO2 in decane up to 850 psia [5.86 MPa], for CO2 in 0.25 N NaCl brine up to 850 psia [5.86 MPa], and for ethane in decane up to 600 psia [4.14 MPa]. All tests were conducted in Berea cores saturated with the liquid phase at 100°F [311 K]. Cores were oriented both vertically and horizontally to assess the effects of gravity-induced convection on the observed mass transfer; visual observations and measurements of composition were not necessary. The data acquired during this study were combined with literature data, and empirical correlations were developed that relate the molecular diffusion coefficient to the liquid viscosity and thermophysical properties of the diffusing gases.