Summary In the Little Knife (ND) CO2 flood minitest, CO2 and water slugs were injected into a single injection well offset by three observation wells on a 5-acre [20 234 m2] circle. Cased hole logs and fluid samples were obtained periodically in these observation wells to monitor saturation and composition changes during the test. This paper compares saturation changes seen in the observation wells with those found in laboratory tests and analyzes differences in displacement mechanisms. Laboratory CO2 flood residual oil saturations during water-alternating-gas (WAG) floods in native-state Little Knife cores were low (1 to 3% PV) and not dependent on water/CO2 ratio. A wettability shift from water-wet initially to non-water-wet following CO2 displacement may be an explanation for these low values. Observation well saturation data showed that these low oil saturations were achieved in places. Overall displacement in the field, however, was more complex than that seen in the laboratory. Three distinct mechanisms were identified: water flooding alone, water flooding accompanied by CO2 swelling of oil where CO2 was transported in solution, and simultaneous CO2/water displacement where a vapor phase was present. Introduction A problem with the CO2 miscible flood process is the inherently unfavorable mobility ratio and the resulting poor volumetric sweep in horizontal reservoirs. Injection of CO2 as slugs alternated with slugs of water (the WAG method) is the only procedure presently being used for controlling mobility.1,2 There are problems and uncertainties with this method, however. These relate tothe extent to which a high mobile water saturation (especially in previously water flooded reservoirs) shields oil from CO2, resulting in incomplete displacement even above the miscibility pressure, andhow CO2 volumetric sweep is affected by the WAG process (i.e., will CO2 and water flow together so that Caudle and Dyes3 type mobility control will be achieved or will they segregate by gravity or some other reason?). The Little Knife CO2 flood mini test provided a unique opportunity to examine the performance of a CO2 WAG flood at reservoir conditions in detail. This was a nonproducing field experiment in which a single injection well was offset by three observation wells on an approximately 5-acre [20 234 m2] circle. A description of the Little Knife field and the operation of the mini test has been given by Desch et al.4 and, in more detail, in a final report to the U.S. DOE.5 The target for the mini test was the base of Zone C and the top of Zone D of the Mission Canyon formation. The important effects, as it turned out, were seen in a highly porous and permeable 15- to 16-ft [4.6- to 4.9-m] interval in Zone D. Fig. 1 shows the pattern configuration and the location of the target interval. Table 1 gives pertinent properties of the Mission Canyon Zone D reservoir. This paper describes laboratory work carried out as a part of the mini test to define the effects of simultaneous water flow on CO2 flood displacement efficiency above the minimum miscibility pressure (MMP) in native-state Little Knife cores. Observation well saturation data obtained from periodic cased hole logs during the mini test then were analyzed for what can be learned about field displacement mechanisms. The extent to which perceptions derived from the laboratory and field measurements are in agreement is also discussed. Background Raimondi and Torcaso6 and Thomas et al.7 first showed that oil trapped at high water saturation may be inaccessible to a miscible flood. This result was amplified by Shelton and Schneider,8 Stalkup,9 and Griffith and Horne,10 among others. They showed thattrapped oil in water-wet systems can be contacted slowly, at a rate apparently related to diffusion of solvent through water and stagnant oil films; andtrapping (or inaccessibility) does not occur in oil-wet systems. Stalkup and Griffith and Horne reached the conclusion that propane displacement at field conditions (low rates, long contact times) can be effective despite trapping at high water saturation. This is consistent with the results of Raimondi and Torcaso, who used a nonane-ethylbenzene system, considering that water solubility and diffusion coefficient will be lower for this system than for propane. If diffusion through a water film is the limiting step, the rate at which a solvent can contact trapped oil will be nearly proportional to the product of the solubility of the solvent in water and its diffusion coefficient in water.