Uncertainty analysis of carbon sequestration in an active CO2-EOR field

Enhanced Oil Recovery (EOR) is perhaps the most feasible option for geologic CO2 sequestration (GCS). However, the typical large extent of uncertainty in reservoir properties is a major obstacle to effective risk assessment of GCS. The primary objective of this study was to quantify uncertainties in key reservoir parameters for an active, commercial-scale CO2-EOR field. We selected the Morrow formation within the active Farnsworth Unit (FWU) EOR field in Texas for this case study. Critical for this study are historical and real-time CO2 injection/production data as well as fundamental hydrologic and geologic characterization data from injection wells and three dedicated characterization/observation wells. We designed and applied a response surface methodology (RSM) integrated with Monte Carlo simulations to evaluate and quantify uncertainty. Previous sensitivity studies identified critical uncertain parameters including reservoir permeability, anisotropy ratio of permeability (kv/kh), water-alternating-gas (WAG) time ratio, and initial oil saturation. Cumulative oil production, net CO2 storage, net water stored (difference between the injection water and produced water), and reservoir pressure at the injection well were the primary dependent variables used to evaluate uncertainties of CO2 storage associated with oil production and potential risk of reservoir pressure build-up. A 3-D static reservoir model was constructed based on the geology of the Farnsworth EOR site, serving as the basis for all multiple-realization reservoir simulations. After performing stepwise regression analyses, a series of response surface models of the dependent variables at each time step were constructed and validated using appropriate goodness-of-fit measures. Given the range of uncertainties in the independent variables, cumulative distribution functions (CDFs) and uncertainty bounds (5th and 95th percentiles) of output responses were estimated based on regression equations and Monte Carlo sampling. Forecasted cumulative oil production and net CO2 storage varied from 54,696 bbl, and 22,784 t, respectively, at the 5th percentile to 203,989 bbl, and 39,525 t at the 95th percentile after 5 years. These results suggest that a significant proportion of forecasted output response uncertainty, including forecasted storage capacity, is propagated from parameter uncertainties. For this case study, response surface results suggest that maximum cumulative oil production could be achieved with permeability in a specific range (10.0-31.6 mD, which is close to the mean value of the actual strata). The pressure near the injection well exceeded 40 MPa, and 54 MPa at the 95th percentile after 1 and 5 years, respectively. The reservoir pressure fracturing threshold is just under 37 MPa in the FWU, indicating a significant risk of caprock fracturing in the low permeability zones due to pressure build-up.