Flow Visualization for CO2/Crude-Oil Displacements

Results of visual observations of high-pressure CO2 floods are reported. The displacements were performed in two-dimensional (2D) pore networks etched in glass plates. Results of secondary and tertiary first-contact miscible displacements and secondary and tertiary multiple-contact miscible displacements are compared. Three displacements with no water present were performed in each of three pore networks:displacement of a refined oil by the same oil dyed a different color;displacement of a refined oil by CO2 (first-contact miscible); anddisplacement of a crude oil at a pressure above the minimum miscibility pressure. In addition, three tertiary displacements were performed in the same pore networks;displacement of the refined oil by water, followed by displacement by the same refined oil dyed to distinguish it from the original oil;tertiary displacement of the refined oil by CO2; andtertiary displacement of crude oil by CO2. In addition, recovery of oil from dead-end pores, with and without water barriers shielding the oil, was investigated. Visual observations of pore-level displacement events indicate that CO2 displaced oil much more efficiently in both first-contact and multiple-contact miscible displacements when water was absent. In tertiary displacements of a refined oil, CO2 effectively displaced the oil it contacted, but high water saturations restricted access of CO2 to the oil. The low viscosity of CO2 aggravated effects of high water saturations because the CO2 did not displace water efficiently. CO2 did, however, contact trapped oil by diffusing through water to reach, to swell, and to reconnect isolated droplets. Finally, CO2 displaced crude oil more efficiently than it did the refined oil in tertiary displacements. Differences in wetting behavior between the refined and crude oils appear to account for the different flow behavior. Introduction If high-pressure CO2 displaces oil in a one-dimensional (1D), uniform porous medium (in which the effects of viscous fingering are necessarily absent), the displacement efficiency is controlled by the phase behavior of the CO2/crude-oil mixtures. The conventional description of the effects of phase behavior was given by Hutchinson and Braun1 for vaporizing gas drives and was extended to CO2 systems by Rathmell et al.2 In a rigorous mathematical treatment of the flow of three-component mixtures. Helfferich3 proved that the displacement will develop miscibility if the oil composition lies outside the region of tie-line extensions on a ternary diagram. Helfferich's analysis was for 1D flows in which fluids are mixed well locally, and the effects of dispersion are absent. Sigmund et al.,4 Gardner et al.,5 and Orr et al.6 showed that results of slim-tube displacements, which are nearly 1D and come close to eliminating the effects of viscous instability, can be predicted quantitatively by 1D process simulations based on independent measurements of the phase behavior and fluid properties of the CO2/crude-oil mixtures. Thus there is good experimental confirmation that the simple theory of the effects of phase behavior on displacement performance describes accurately the behavior of flow in an ideal displacement, such as a slim tube. In a CO2 flood in reservoir rock, however, a variety of other factors will influence process performance. Because the viscosity CO2 is much lower than that of most oils, viscous instability will limit the sweep efficiency of the injected CO2. In addition, Gardner and Ypma7 predicted, based on 2D simulations of the growth of a viscous finger, that an interaction between viscous instability and phase behavior would lead to higher residual oil saturation in regions penetrated by a viscous finger. Pore-structure heterogeneity may also influence displacement efficiency. Spence and Watkins8 found that residual oil saturations after CO2 waterfloods increased as the heterogeneity of the core increased. Several investigators have reported that high water saturations can alter mixing between oil and injected solvent. Raimondi and Torcaso9 found, in displacements in Berea sandstone cores, that significant fractions of the oil phase could not be contacted by injected solvent when the water saturation was high. Thomas et al.10 reported that a portion of the nonwetting phase can exist in "dendritic" pores whose shapes were determined by the surrounding wetting phase. They argued that material in the dendritic pores mixed with fluid in the flowing fraction only by diffusion. Stalkup11 and Shelton and Schneider12 also investigated effects of mobile water saturations in miscible displacements. Stalkup found that the flowing fraction decreased as the water saturation increased. Shelton and Schneider reported that the presence of a second mobile phase slowed recovery of either phase, but the nonwetting phase was affected more strongly. In their tests, all of the wetting phase was recovered by a miscible displacement, but significant amounts of nonwetting phase remained unrecovered.