CO₂ EOR by Diffusive Mixing in Fractured Reservoirs

Miscible CO₂ flooding as an enhanced oil recovery (EOR) technique is currently employed in large parts of the U.S., and has a great potential. We here report on CO₂ injection in heterogeneous rock samples with fractures to study recoverable oil by diffusive mixing. In laboratory tests, CO₂ is injected through fractured core samples to displace oil. The injected CO₂ will follow the path of least resistance in the fracture, but the majority of oil is located in the oil-saturated matrix block adjacent to the CO₂-filled fracture. With a high fracture transmissibility the oil is produced mainly by diffusion because the viscous forces are limited due to the high transmissibility, of the fracture and the low viscosity of the injected CO₂. With decreasing fracture transmissibility viscous forces became more dominant at early times. Gravity is not believed to be of major significance due to the similar density of CO₂ and oil at the experimental conditions.

Oil recovery by CO₂-oil diffusion in fractured core samples was visualized with nuclear magnetic resonance (MRI) and X-ray computed tomography (CT), both with and without residual water present. Oil recoveries above 90% OOIP were observed in several tests during injection of several pore volumes CO₂. The presence of residual water reduced the rate of production by reducing the pore space and changed the tortuosity and CO₂ flow paths. The development of local CO₂ and oil saturations confirmed that diffusion was the main oil-recovery mechanism at high fracture transmissibility, where oil was produced symmetrically from each side of the fracture along the length, without signs of viscous displacement. It was also revealed that diffusion produced oil from the inlet and outlet face of the core sample, which could not be captured from production measurements alone.

An effective diffusion coefficient was found using numerical simulation (CMG GEM) to reproduce the experimentally measured development in local CO₂ and oil saturation. The validated numerical model was then used to perform a sensitivity analysis of important parameters, such as, sample size and porosity. Numerical simulations gave an effective molecular diffusion coefficient of m₂/s in the studied chalk samples and indicated a large degree of sensitivity to system size and tortuosity.