Lawrence Berkeley National Laboratory

A new TOUGH code, TOGA (Transport of Oil, Gas, Aqueous), is used to model enhanced oil recovery (EOR) from a South Texas field where a combination of water, CO2, and WAG (water-alternating-gas) injection wells are located in the down-dip region of a fault block, with oil and natural gas production occurring from wells located up-dip. The reservoir is composed of interlayered sands and shales at a depth of about 2000 m. TOGA considers three fluid phases (oil, gas, and aqueous), and multiple components. Oil field data including compositional analysis of stock-tank oil, gas-oil ratio, and gas gravity are used to determine the composition of oil and gas phases. Initial conditions for aqueous saturation are determined from a depth to surface resistivity survey conducted prior to the onset of CO2 injection and Archie’s Law. Simulation results after several years of EOR operations are then converted back to resistivity maps and compared to images created by time-lapse depth-to-surface resistivity surveys. These surveys show a strong increase in resistivity, interpreted as the replacement of aqueous phase by injected CO2. In contrast, preliminary TOGA results show that injected CO2 displaced more oil than water, suggesting that the three-phase relative permeability curves used to describe the interference of fluid phases need to be modified. Limited information on the composition of produced fluids provides another data set for comparison with model results. Model variations including using different relative permeability curve parameters and WAG cycle durations yielded improved model matches to the field data. The practical simplifications needed to simulate an actual operating oil field and the limited amount of field operation data available preclude using the present model to get an accurate history match or to make detailed predictions. Rather, the modeling is intended to illustrate the general trends occurring during CO2-EOR and to indicate where more detailed information is needed.