Summary This paper presents a novel technique to determine multicomponent diffusion coefficients for carbon dioxide (CO2) injection in a North Sea chalk field (NSCF) in Norway at reservoir conditions. The constant-volume-diffusion (CVD) method is used, consisting of an oil-saturated-chalk core in contact with an overlying free space, which is filled with the CO2. The experimental data are matched with an equation-of-state (EOS) -based compositional model. Transport by diffusion controls the dynamics of the constant-volume system and, together with phase equilibria, allows a consistent estimation of diffusion coefficients needed to describe the observed changes in system pressure. We conduct two experiments at reservoir conditions: One uses a core plug saturated with live oil and the other with stock-tank oil (STO). Once the experiments are completed, EOS-based compositional simulation is performed to match the experimental data by use of the oil- and gas-diffusion coefficients as history-matching parameters. The modeling work is conducted with a commercial reservoir simulator by use of a 2D radial-grid model to describe the experimental setup. The experiment uses an outcrop chalk core mounted in a vertically oriented core holder. The chalk is shorter than the core holder, thus resulting in an overlying void space. The system is initially saturated with oil at reservoir conditions. CO2 is then injected from the top, forming an overlying CO2 chamber and displacing oil toward the bottom of the core holder. Once CO2 fills the overlying bulk space, the system is isolated with no further injection or production. The CO2 and oil reach and remain in equilibrium locally at the gas/oil interface throughout the test, beginning and maintaining the diffusion mechanism. Diffusion of CO2 into the oil results in a decreasing pressure, which is the main history-matching parameter. The multicomponent diffusion coefficients are found to match the model pressure/time prediction to the experimental data. This suggests the modeling work flow incorporates a representative EOS model and the main transport dynamics controlled by diffusion are being treated properly. Proper simulation of CO2 injection in fractured-chalk reservoirs requires the ability to model multicomponent diffusion accurately. The proposed CVD method provides such modeling capabilities. Our modeling and experimental work indicate the novelty of the CVD method to determine the diffusion coefficients of a system where diffusion is the dominant displacement mechanism. The fact that the oil is contained within a low-permeability-chalk sample reduces density-driven convection that could result because of nonmonotonic oil-density changes as CO2 dissolves into the oil.

To reduce CPU time in compositional reservoir simulations, a minimum number of components should be used in the equation of state (EOS) to describe the fluid phase and volumetric behavior. A "detailed" EOS model often contains from 20 to 40 components, with the first 10 components representing pure compounds H 2 S, CO 2 , N 2 , C 1 , C 2 , C 3 , i-C 4 , n-C 4 , i-C 5 , and n-C 5 . The remaining components represent a split of the heavier C 6ϩ material in single-carbon-number (SCN) fractions such as C 6 , C 7 , C 8 and C 9 , or groups of SCN fractions such as C 10 -C 12 , C 13 -C 19 , C 20 -C 29 , and C 30ϩ . Occasionally the light aromatics BTX (benzene, toluene, and xylene isomers) are also kept as separate components for process modeling. Today's typical laboratory compositional analysis provides 50-60 components, including isomers with carbon numbers 6 to 10, SCN fractions out to C 35 and a residual C 36ϩ . This is in contrast to the 11-12 components (through C 7ϩ ) reported in most commercial laboratory reports pre-1980.A "pseudoized" EOS model might contain only 6-9 lumped components -e.g. lumping "similar" components such N 2 ϩC 1 , i-C 4 ϩn-C 4 ϩi-C 5 ϩn-C 5 , and some 3-5 C 6ϩ fractions. The selection of which components to lump together is difficult because of the huge number of possible combinations. This paper describes a systematic, automated method 1 to search a vast number of feasible pseudoized EOS models based on an initial, detailed EOS model.The obvious application of pseudoized EOS models is compositional reservoir simulation, where run time is an important issue and fewer components may be important. The method we present is based on (1) quantifying the "quality of match" between a pseudoized EOS model and the detailed EOS model from which it is derived, and (2) systematically testing all plausible lumping combinations. The method allows for a set of constraints to be imposed on the lumping of components, such as (1) not lumping certain non-hydrocarbons (e.g. CO 2 ), (2) forcing the first plus fraction to contain a specific carbon-number component (e.g. C 6 ), and (3) the last component in the original EOS not being lumped with other heavy fractions (e.g. C 30ϩ ).The proposed pseudoization procedure is comprehensive, and founded in the ability of an EOS with fewer components to describe a wide range of phase and volumetric properties covering all of the relevant pressure-temperature-composition (p-T-z) space expected for a given reservoir development. The litmus 1 The software used is EOS program PhazeComp and model automation platform Pipe-It. test of quality is how well the pseudoized EOS compares with the detailed EOS model from which it is derived, an EOS that accurately describes all key measured laboratory PVT data. The method proposed will find an optimal pseudoized EOS model to describe all PVT data that are relevant to a particular reservoir development -e.g. depletion performance, immiscible and miscible gas injection, compositional variation, and surface processing.

We present results studying the enhanced-oil-recovery (EOR) potential for carbon dioxide (CO 2 ) injection in the naturally fractured Haft Kel field, Iran, on the basis of detailed compositional simulations of a homogeneous single matrix block surrounded by fractures. Oil recoveries from CO 2 injection in this idealized model approach 90% for reservoir pressures of 1,400 psia and higher (i.e., at and above current reservoir pressure of 1,500-1,800 psia). It is expected that heterogeneity will reduce recovery on the field scale. This compares with 15-25% recoveries reported for gas-cap expansion and/or injection of hydrocarbon (HC) gas.Fundamentally different recovery mechanisms develop above and below 2,000 psia, the pressure at which CO 2 density equals the reservoir-oil density. At lower pressures, CO 2 is less dense than reservoir oil and traditional gas/oil gravity segregation results, with a highly efficient process driven by gravity, compositional effects, and interfacial-tension (IFT) gradients that cause capillary-induced oil flow. At pressures greater than 2,000 psia, CO 2 density is greater than reservoir-oil density, resulting in an unusual gravity-drainage mechanism whereby CO 2 enters the bottom of a matrix block and pushes oil out the sides and top of the matrix block.The effect of several key parameters has been studied in detail-matrix permeability, matrix-block size, matrix/matrix capillary continuity (stacked blocks), and the use of mixtures of CO 2 and HC gas. One of the key results is how the rate of recovery differs for combined injection of HC gas and CO 2 , and how it varies for CO 2 injection for different model parameters.EOR results are affected by grid sensitivity. Grid effects have been quantified and compared for different model parameters. Final EOR assessment is made using models in which sufficient grid refinement is used to minimize grid sensitivity.

No abstract

This paper presents a modeling study of CO2 injection in a chalk core based on experimental data reported by Karimaie (2007). The experiment consisted of a vertically-oriented 19.6 cm long chalk outcrop core initially saturated with reservoir synthetic oil consisting of C1 and n-C7 at a temperature of 85°C and pressure of 220 bar. After saturating the core with the oil mixture by displacement a small "fracture" volume surrounding the core was created by heating the solid Wood's metal which originally filled the volume between the core and core holder. Gas injection was first conducted using an equilibrium C1-n-C7 gas at 220 bar, resulting in no recovery by (thermodynamic) mass transfer but only from immiscible displacement with Darcy flow driven by induced pressure gradients and a minor impact of gravity-capillary equilibrium. Once oil production ceased in this first displacement, a second period with pure CO2 gas injection followed. Our modeling was conducted with a compositional reservoir simulator. The 2-dimensional r-z model used fine grids for the core matrix and surrounding fracture. Automated history matching is used to match experimental data. The match to reported production data gave a high degree of confidence in the model. Oil recovery improved significantly by CO2 injection. Our model study indicates that the recovery mechanism in the Karimaie experiment was dominated by Darcy displacement because of a low conductivity in the surrounding fracture. Another observation made in our study was the strong influence of surface separator temperature on surface oil production. Finally, gas injection rate changes had a significant impact on recovery performance for CO2 injection. Gravity-capillary recovery mechanism was of minor importance in the Karimaie experiments. Introduction CO2 injection has been considered as potentially enhancing oil recovery from naturally fractured reservoirs. Alavian and Whitson (2005)study the IOR potential for CO2 injection in the naturally-fractured Haft Kel field, Iran, based on detailed compositional simulations of the matrix-fracture system. Obviously, it would be useful to experimentally investigate the efficiency of gas injection in naturally fractured reservoirs, followed by CO2 injection, before this procedure is applied to a reservoir. Few experiments are reported in the literature to study gravity drainage mechanism in CO2 injection in fractured reservoirs. Li et al (2000) perform CO2 injection after water flooding in a dead oil system. They study water imbibitions followed by CO2 gravity drainage experiment on artificially fractured cores. They report that CO2 gravity drainage could significantly increase oil recovery after water flooding. They found that CO2 gravity drainage declines as the rock permeability decreases and initial water saturation increases. Asghari and Torabi (2008) conduct miscible and immiscible CO2 gravity drainage experiments with dead oil (n-C10). They show miscible CO2 injection improves oil recovery, but they could not match laboratory experiment with a simulation model. Karimaie (2007) performed equilibrium gas injection followed by CO2 experiments on chalk and carbonate cores at reservoir conditions where cores were saturated with live synthetic oil. The experiments were designed to illustrate CO2 injection in a fractured reservoir. The matrix permeability essentially controls the rate of recovery, and the ratio between the matrix and fracture permeability determines if viscous displacement (Darcy flow by pressure gradients) is governing the displacement or not. The pressure gradients over the fractures are negligible for high permeability fractures where most injected gas flows through the fracture space and the main production mechanism from matrix is gravity drainage. Darcy flow might play a role when fracture permeability is low. Fracture permeability should be adequately high to eliminate viscous displacement in the core, because it is difficult to achieve efficient recoveries by viscous dominated mechanism in low permeability fractured reservoirs. Uncertainties and error sources are analyzed to understand and simulate the Karimaie experiment.