Carbon dioxide flooding under miscible conditions is being developed as a major process for enhanced oil recovery. This paper presents results of research studies to increase our understanding of the multiple-contact miscible displacement mechanism for CO2 flooding. Carbon dioxide displacements of three synthetic oils of increasing complexity (increasing number of hydrocarbon components) are described. The paper concentrates on results of laboratory flow studies, but uses results of phase-equilibria and numerical studies to support the conclusions.Results from studies with synthetic oils show that at least two multiple-contact miscible mechanisms, vaporization and condensation, can be identified and that the phase-equilibria data can be used as a basis for describing the mechanism. The phase-equilibria change with varying reservoir conditions, and the flow studies show that the miscible mechanism depends on the phase-equilibria behavior. Qualitative predictions with mathematical models support our conclusions.Phase-equilibria data with naturally occurring oils suggest the two mechanisms (vaporization and condensation) are relevant to CO2 displacements at reservoir conditions and are a basis for specifying the controlling mechanisms. Introduction Miscible-displacement processes, which rely on multiple contacts of injected gas and reservoir oil to develop an in-situ solvent, generally have been recognized by the petroleum industry as an important enhanced oil-recovery method. More recently, CO2 flooding has advanced to the position (in the U.S.) of being the most economically attractive of the multiple-contact miscibility (MCM) processes. Several projects have been or are currently being conducted either to study or use CO2 as an enhanced oil-recovery method. It has been demonstrated convincingly by Holm and others that CO2 can recover oil from laboratory systems and therefore from the swept zone of petroleum reservoirs using miscible displacement. However, several contradictions seem to exist in published results.. These authors attempt to establish the mechanism(s) through which CO2 and oil form a miscible solvent in situ. (The solvent thus produced is capable of performing as though the two fluids were miscible when performing as though the two fluids were miscible when injected.) In addition, little experimental work has been published to provide support for the mechanisms of multiple-contact miscibility, as originally discussed by Hutchinson and Braun.One can reasonably assume that the miscible CO2 process will be related directly to phase equilibria process will be related directly to phase equilibria because it involves intimate contact of gases and liquids. However, no data have been published to indicate that the mechanism for miscibility development may differ for varying phase-equilibria conditions.This paper presents the results of both flow and phase-equilibria studies performed to determine the phase-equilibria studies performed to determine the mechanism(s) of CO2 multiple-contact miscibility. These flow studies used CO2 to displace three multicomponent hydrocarbon mixtures under first-contact miscible, multiple-contact miscible, and immiscible conditions. Results are presented to support the vaporization mechanism as described by Hutchinson and Braun, and also to show that more than one mechanism is possible with CO2 displacements. The reason for the latter is found in the results of phase-equilibria studies. SPEJ P. 242

Mixing oil with high-ethane-content hydrocarbon gases or CO2 can produce multiple liquid phases and an asphaltic precipitate in low-temperature reservoirs. The residual saturation that occurs in a reservoir displacement is not significant from a recovery standpoint, but may produce three-phase relative permeability effects that reduce injectivity produce three-phase relative permeability effects that reduce injectivity and, thus, oil recovery rate during alternate gas-water injection. Introduction Multiple-contact miscible gas flooding has the potential for economically recovering a significant amount of incremental oil over that recoverable by conventional waterflooding. Under appropriate reservoir pressure and temperature conditions, rich gas or carbon dioxide will miscibly displace the oil it contacts. Water is injected alternately with the gas to decrease the mobility ratio and, thus, improve the sweep. One such project using rich gas in a West Texas reservoir has been described previously. However, unanticipated reduction in water injectivity occurred after injection of the first batch of rich gas, and indications were that the problem existed in depth within the reservoir and was not just a wellbore problem. Harvey et al. present the results of field studies that led to the work presented in this paper. Native-state reservoir core tests showed that the injectivity reduction could be caused by a residual oil saturation of a few percent of pore volume behind the rich-gas bank. A small amount of residual oil resulted in unusually high trapped gas saturation and a resulting decrease in water relative permeability. permeability. A small residual oil saturation and, hence, a lowered water injectivity, could arise in the field if oil was bypassed because of small-scale reservoir rock heterogeneities or if the rich gas did not completely displace all the oil contacted. Factors that produce incomplete displacement of contacted oil are presented. PVT cell data are presented that show the occurrence of PVT cell data are presented that show the occurrence of two liquid phases and an asphaltic precipitate when high-ethane-content rich gas and reservoir oil are mixed at reservoir conditions. Core tests are described showing that the formation of multiple phases results in a residual hydrocarbon saturation during displacement. Similar but less detailed studies of a CO2-oil system are also presented. Other authors, observed multiple liquid-phase presented. Other authors, observed multiple liquid-phase phenomena with some miscible gases and West Texas phenomena with some miscible gases and West Texas oils, but did not evaluate the effect on oil displacement. Windowed Cell Tests Procedure Procedure The general procedure used was to place the driving gas into a variable-volume windowed cells that was held at the reservoir temperature. The driving gases used were rich gas and CO2. A known amount of recombined reservoir oil was combined with the driving gas in the cell. A phase distribution test starting at a high pressure (usually 4,000 to 8,000 psi) was performed on this oil-driving gas mixture. This test consisted of a constant-composition expansion, during which the total volume of the cell and the distribution of the phases in the cell were measured at a known pressure. Upon completion of the phase distribution test, additional oil was introduced into the cell and another phase distribution test was performed. These tests, which were designed to simulate mixtures in the driving gas-rich region of all possible oil-driving gas mixtures, were terminated when sufficient oil had been added to dissolve all the driving gas in the cell. Any mixtures containing additional oil would have bubblepoint pressures that decrease regularly down to the saturation pressure of the recombined oil. JPT P. 1171

Numerical simulation of miscible EOR processes requires calculation of the phase equilibria that exist between solvent and oil over the entire solvent/oil composition range. One calculation approach is to tabulate K-values from experimental data and allow the simulator to access the table when necessary. This approach can lead to erroneous conclusions for CO2 miscible processes if there are insufficient tabulated phase equilibrium data to cover all possible compositions, in particular those cases where three phases exist at some point in the displacement. This paper presents a different approach to the problem. If a generalized equation of state (EOS) can match experimental data, then it can be used in a reservoir simulator to calculate the phase equilibria necessary for the prediction of fluid compositions, densities, and viscosities during a displacement process. Previous work has shown how a generalized Redlich-Kwong equation can be used to calculate typical hydrocarbon phase behavior relationships that exist in condensate and black oil reservoir systems. The equation parameters have been modified further for use in hydrocarbon/CO2 calculations over a wide range of CO2 concentrations. Appropriate mixing niles for description of the phase equilibria in CO2/hydrocarbon systems have been developed. Experimental binary vapor/liquid equilibrium (VLE) data have been used to evaluate the constants in the EOS for pure CO2 as well as interaction terms used within the mixing rules. Predictions of phase equilibria then have been made and compared with experimental data for a synthetic multicomponent hydrocarbon/CO2 system and a crude oil/CO2 system. Introduction Current interest in miscible EOR methods has led to the use of compositional reservoir simulators to understand and evaluate performance. An essential pan of such a simulation is a means of predicting the complex phase equilibria possible during EOR processes. However, because of computational time constraints, the numerical complexity of such a technique must be limited. In recent years, a number of relatively simple EOS have been developed and applied to hydrocarbon phase equilibria calculations. One such development is the generalized Redlich-Kwong EOS by Yarborough. This EOS has been applied to many reservoir fluid property calculations with excellent results because of its simplicity. For the simulation of CO2 miscible EOR processes, an EOS must be capable of predicting phase equilibria over a wide range of CO2 compositions. CO2/hydrocarbon mixtures can exhibit complex phase equilibria -e.g., liquid/liquid immiscibility, liquid/liquid/vapor equilibria, and asphaltene dropout. While it is highly unlikely that any simple EOS can provide accurate phase equilibria predictions for all these situations, the generalized Redlich-Kwong EOS referenced previously has been adapted to provide adequate phase equilibria predictions for CO2/hydrocarbon systems over a wide range of conditions. This adaptation involves the use of special parameters to describe pure CO2 above its critical temperature and of modified mixing rules to describe CO2/hydrocarbon mixture behavior. General functions are developed for the parameters involved to permit interpolation (or extrapolation) to other systems and conditions. Literature binary CO2/hydrocarbon VLE data were used to establish these parameters. The generalized EOS was tested through comparisons of predicted phase equilibria with experimental data for a number of binary and multicomponent CO2/hydrocarbon systems. The multicomponent systems include CO2 /synthetic oil and CO2/reservoir oil data presented in this paper. SPEJ P. 308^

Experimental vapor-liquid equilibrium phase composition data and K values are presented for 43 multicomponent mixtures containing nitrogen, methane, carbon dioxide, ethane, hydrogen sulfide, propane, n-pentane, n-heptane, toluene, and ndecane. Data were measured at temperatures from -50-250°F and for pressures from 100-4500 psia. An unusual physical attraction was observed to exist between hydrogen sulfide and toluene, and the magnitude of the attraction depended on temperature. Published K values for n-heptane and toluene, measured in binary systems with methane, were in error at temperatures below 0°F and pressures less than 1000 psia.

No abstract