Summary. Although CO 2 injectivi- ty should be significantly greater than brine injectivity because CO 2 has a much lower viscosity than brine, this behavior is not always seen, as shown in a Denver Unit field test. This paper examines features that cause differences in CO 2 injectivity with a model that uses simple nondisper- sive flow with a series of constant- composition slugs to approximate the analytical solution (normally a se- quence of shocks and tails) in a se- quence of noncommunicating layers. Because of its simplicity, this model identifies the primary features that re- sult in the different observed CO 2 in- jectivities more clearly than the finite-difference model. This paper shows that the qualita- tive differences between Cedar Creek anticline corefloods and field behavior result solely from differ- ences in geometry. That is, a single set of centrifuge-measured, quasina- tive-state, secondary-drainage rela- . tive permeabilities can be used to predict both laboratory and field be- havior. Primary factors that contribute to the differences between the two field tests are fluid/rock properties, ef- fective well bore radius (or skin), and heterogeneity in the layering. 226 Comparison of Laboratory- and Field-Observed CO 2 Tertiary Injectivity Peter G. Christman, SPE, Shell Western E&P Inc., and Sheldon B. Gorell, SPE, Shell Development Co. Introduction In the past decade, Shell Western E&P Inc. has conducted two tertiary CO 2 pilots in carbonate reservoirs. The first was in the Denver Unit (Wasson field). It experienced CO 2 injectivity that was surprisingly lower than the preflood brine injectivity. CO 2 in- jectivity in the second pilot at the South Pine Unit (Cedar Creek anticline) was significant- ly higher than brine injectivity. Ironically, this behavior was also surprising because of both the Denver Unit experience and the lab- oratory experiments conducted on South Pine core material before CO 2 injection. The Denver Unit pilot was history- matched with a simulator featuring perme- ability reduction factors similar to models used by Claridge, 1 Todd et al. ,2 and Chase and Todd. 3 Without the permeability-re- duction factors, this model predicted CO 2 injectivity greater than brine injectivity. The majority of the Denver Unit tertiary CO 2 corefloods also displayed CO 2 injectivities greater than the brine injectivity. Fig. 1 shows the normalized injectivity of 20 ter- tiary CO 2 corefloods conducted in Denver Unit core. (Details about these experiments are found in Ref. 4.) Only two ofthese core- floods displayed low CO 2 injectivities con- sistent with the pilot behavior. The simulator curve plotted in Fig. 1 represents a linear finite-difference simulation with pilot-tuned reduction factors of a coreflood. Patel et al. 4 showed that mixed wettability, and the resulting relative-permeability hysteresis, was the critical property in the only two corefloods that displayed low CO 2 injec- tivity. Before CO 2 injection at the South Pine pilot, tertiary corefloods were conducted to study CO 2 injectivity. Frozen core materi- al, taken from the pilot injection well, was mounted in coreholders, and every effort was made to preserve the native-wettability state. As Fig. 2 shows, these experiments displayed lower CO 2 injectivities than brine, in qualitative agreement with the mixed-wettability Denver Unit counterparts. When the South Pine pilot CO 2 injectivity was much higher than the brine preflood in- jectivity, it seemed to be completely at odds Copyright 1990 Society of Petroleum Engineers with coreflood results and the Denver Unit pilot (assumed to be a good analog). To understand fully the differences in the CO 2 injectivity observed here, it is neces- sary to examine both the fluidlrock proper- ties and the flow behavior in the geometries involved in the different tests. In this paper, we compare the mixed-wettability Wasson relative-permeability curves with a set of quasinative-state, secondary-drainage rela- tive permeabilities that were measured in Cedar Creek core material. Then we discuss a model that considers the flow of constant- composition slugs in a noncommunicating layered system to show that the rather sur- prising COrinjectivity behavior is caused primarily by differences in flow geometry and tertiary oil-bank properties. Throughout this paper, properties that are representative of the field (such as API gravity and rock type) are referred to by their field designation (Wasson or Cedar Creek). Results specific to the pilot or in- jectivity test are identified by the appropri- ate operating unit name (Denver Unit or South Pine Unit). Fluid Mobilities Clearly, any study of injectivity must begin with relative permeability because it directly affects the fluid-bank mobilities. Patel et ai. 4 showed that mixed wettability and the resulting relative-permeability hysteresis ex- plained the low CO 2 injectivity observed at the Denver Unit (Wasson field) pilot. The imbibition (waterflood) and secondary- drainage (C0 2 flood) curves for San Andres rock in the Wasson field are present- ed in Fig. 3. Similar curves were measured by Schneider and Owens 5 on native-state San Andres rock by a steady-state method. The major characteristic of the hysteresis in Fig. 3 is the increase in the immobile water saturation from 15% on imbibition to 25 % on secondary drainage. This increase occurs because the water is displaced by secondary drainage from oil-wet pores in a mixed-wettability medium; i.e., part of the water, the nonwetting phase, is trapped by the wetting oleic phase. We have directly observed this behavior in 2D mixed- wettability glass models on secondary drainage. In his extensive literature survey, February 1990 • JPT