Computer-assisted tomography (CAT) is used to obtain cross-sectional images of Berea sandstone cores during oil displacement experiments. Local oil saturation averaged over an area of about 0.03 × 0.03 in. [0.8 × 0.8 mm] square is computed as a function of spatial position and time. A series of CAT scan images displaying the time evolution of the fluid distribution at one cross section are shown to illustrate the formation of viscous ringers. Introduction CAT 1–2 is a method that uses computerized mathematical algorithms to reconstruct tomographic image of an object. The image reconstruction is based on multiple X-ray measurements made around the object's periphery. This technique has been used in the present research to obtain oil saturation distribution information during immiscible oil displacement in Berea sandstone cores. The objective is to investigate various problems involved in oil recovery processes, including,heterogeneity of the porous structure,surface interactions between oil. the displacing fluid, and the reservoir rock formation, andthe viscosity ratio between the two fluids. The flow phenomenon is very complex. and previous experimental methods have offered insufficient information for the understanding of oil recovery processes. The CAT scan image acquisition is rapid: thus, it yields directly local oil saturations over a cross section as a function of spatial position and time. Dynamic fluid distribution profiles then can be used to analyze the effectiveness of various oil recovery strategies. Experiment An unmodified second-generation CAT scan apparatus (DeltaScan-50 CT scanner by Ohio-Nuclear) is used to obtain oil distribution histories during immiscible oil displacement experiments in oil-bearing Berea sandstone cores. In spite of the dense silica materials. CAT scan has been used successfully to observe fluid flow in sandstone cores. The porous-media models used in this laboratory are cylindrical Berea sandstone cores (5 cm in diameter and 25 cm in length). The permeability of the core is 300 aid, and the porosity is about 20%. The core initially was evacuated and filled completely with oil. A displacing fluid of 1 M KI solution was injected into the core at a rate of 10 mL/hr [10 cm3/h] (superficial velocity is approximately 24 in./D 160 cm/d]). This core was placed in the CAT scanner, and cross-sectional images were taken at different axial locations and different times during the displacement experiment. The computer unit computes local X-ray attenuation coefficients over the scanning, cross section for picture elements of 0.8 by 0.8 mm [0.03 × 0.03 in.] per square. The thickness of the element is approximately equal to the width of the X-ray beam, which is about 1 cm [0.4 in.]. These average X-ray attenuation coefficients result from linear combinations of the silica rock formation and the oil/KI solution mixtures that occupy the pore spaces. Therefore, the oil saturation distribution over a cross section can be computed from local X-ray attenuation data for each CAT scan image. Results and Discussion A typical CAT scan image showing the oil saturation distribution at a certain axial position is illustrated in Fig. 1. Darker regions indicate water-rich area where most of the oil has been displaced, and lighter regions indicate oil-rich area. From Fig. 1, the spatial distribution of oil and water can be observed. The ordered fluid saturation changes seem to indicate sonic periodicities coincident with the presence of bedding planes existing in Berea sandstone cores. If sequential scannings are taken at different axial positions at a given time, the structure of the water "fingers" can be reconstructed. Figs. 2A, 2B, and 2C plot the oil saturation distributions at 5 cm [2 in.] from the injection point after 3, 4.5, and 15 PV of the displacing fluid (1 M KI) have been injected into the sandstone core. The top of the diagram indicates 100% oil, and the bottom of the diagram indicates 100% water (1 M KI). These CAT scan images have dramatically represented the invasion of water into the oil region and the displacement of oil from a specific cross section as a function of time. Time derivatives of oil saturations also are computed from these image data, which yield rate of changes of local oil saturations SPEJ P. 53^

Range3 223.15 0.4622 -0.0308 2.0 48.7-292.3 248.15 0.8968 -0.0273 1.9 55.7-391.2 273.15 1.1763 -0.0241 0.4 25.0-393.8 298.15 1.4513 -0.0221 0.9 0.1-353.5 0.1-392.4 3%. 15 1.7746 -0.0206 0.9 Values obtained by expressing h (l@D/rn*.~-~) as a linear function of (P/MPa). Lowest experimental pressure ( M R ) for which D ir prrdicted within eiprrmrntal uncertainty by Eq I , and highest erprirnental pressure. *Mean deviation (%)of experimental poinb from values calculated from equatiom to the lines d bpst fit.temperatures are plotted in Figure 1. The variation of D with P for n-hexane is typical normal liquid behavior. The slopes of the curves decrease considerably with increasing pressure, particularly at elevated temperatures. Clearly the pressure dependence of D is so strong that if data were available up to only 200 MPa, for example, it would be difficult to reliably extrapolate the curves to much higher pressures. The data at each temperature could no doubt be fitted to functions such as polynomials in P for extrapolation purposes, but (depending on the order of the polynomial) a considerable number of data points may be required, so that polynomials have limited value as predictive functions. Fits of the n-hexane data to Eq. 1 are shown in Figure 2, which demonstrates that apart from the low pressure (<50 MPa) region at low temperatures, the plots are linear up to the highest pressures attained in the measurements. The best-fit parameters of Eq. 1 are listed in Table 1. The estimated accuracy of the experimental data is f2.5%, so that for each temperature the mean deviation is smaller than the uncertainty in individual data points. Since Eq. 1 has only two adjustable parameters, as few as two experimental diffusion coefficients at pressures above say 50 MPa are adequate to make good estimates of D at other pressures. The general applicability of Eq. 1 suggests that it can serve also as a useful criterion of normal liquid behavior. The relationship does not hold for tracer and self-diffusion coefficients of water.However, the variation of D with P for water is abnormal in that with increasing pressure D first increases, then subsequently decreases (Woolf, 1975; Woolf and Harris, 1980). Work in progress on tracer diffusion of methanol has revealed a similar anomaly at low temperature. At higher temperature the diffusion of methanol in water approaches normal liquid behavior.