Modeling CO2 Injection in Fractured Reservoirs Using Single Matrix Block Systems

Abstract

In this thesis, CO2 injection in matrix/fracture systems has been studied using a finely-gridded compositional simulator representing a single matrix block. Three laboratory experiments were modeled to investigate whether CO2 injection in a fracture-matrix system could be simulated using commercial simulators that include basic fluid flow physics, phase behavior, and molecular diffusion.

The first experiment was performed by Karimaie (2007) using an equilibrium, saturated gas-oil fluid system (C1-n-C7) at 220 bar and 85 oC. Because no recovery was expected from non-equilibrium thermodynamic mass transfer, reported recovery stemmed only from Darcy displacement driven by gravity and capillary forces. When the oil production stopped from the equilibrium gas displacement, a second injection period with pure CO2 followed.

The numerical modeling was conducted using a compositional reservoir simulator (SENSOR) without diffusion. The 2-dimensional r-z model used fine grids for the core matrix and surrounding fracture. Automated history matching was used to determine parameters which were not accurately known (fracture permeability, fracture and matrix porosity, and separator conditions), using surface volumetric oil production rates reported experimentally. The final model match was relatively unique with a high degree of confidence in final model parameters. The oil recovery improved significantly with CO2 injection.

Our model indicated that the recovery mechanism in the Karimaie experiment was dominated, for both equilibrium gas and CO2 injection, by top-to-bottom Darcy displacement caused by low conductivity in the artificial fracture; little impact of capillary-gravity displacement was found. Changes in CO2 injection rate had a significant impact on recovery performance. This experiment was also modeled using ECL300, with the same production performance as SENSOR for the set of history-match parameters determined without diffusion. When molecular diffusion was used in ECL300, results were nearly identical with those found without diffusion.

Two other experiments were performed by Darvish (2007) at a higher temperature and pressure (130 oC and 300 bara) using a similar chalk and live reservoir oil. A similar modeling approach to that described above was also used for these experiments. In both experiments, the matching process based on reported oil production data gave a high degree of confidence in the model. The reported experimental mass fractions of produced-stream components were also matched well.

Our modeling study indicates that gravity drainage affects the displacement process, but that mass transfer – including vaporization, condensation and molecular diffusion – also impact the recovery performance of CO2 injection in the Darvish experiments. The CO2 injection rate and initial water saturation were investigated by comparing the two Darvish experiments.

Our studies from all of the Karimaie and Darvish experiments show a strong influence of the surface separator temperature on surface oil production, and this is an important consideration in designing and interpreting laboratory production data consistently.

Once the laboratory recovery mechanisms had been successfully modeled, predictive numerical simulation studies were conducted on field-scale matrix/fractured systems, albeit mostly for single matrix blocks surrounded by a fracture. The effects of several key parameters on recovery production performance were studied in detail for field-scale systems: matrix permeability, matrix block size, matrix-matrix capillary continuity (stacked blocks), and the use of mixtures containing CO2 and hydrocarbon gas.

The field-scale results were affected by gridding, so grid was refined to the degree necessary to achieve a more-or-less converged solution – i.e. recovery production performance didn’t change with further refinement.

We studied the effect of molecular diffusion on oil recovery by CO2 injection in laboratory experiments and field-scale systems. Because the fluid systems considered had complex phase behavior and a wide range of conditions from strongly immiscible to near-miscible, the diffusion driving potential used was total component potential including chemical and gravity effects; concentrationdriven diffusion did not represent the more-complex non-equilibrium CO2 injection processes observed in the laboratory tests.

A key result of this study was that diffusion can have an important effect on oil recovery, and that this effect varies with matrix block size and CO2 injection rate. We have shown that diffusion has a dominant effect on the recovery mechanism in experimental tests, except at very low rates of CO2 injection (and equilibrium hydrocarbon gas injection). For the field-scale matrix/fracture systems, diffusion can have a significant effect on the rate of recovery, with the effect becoming noticeable for low reservoir pressures and/or matrix block sizes less than ~40 ft.