This article, written by Senior Technology Editor Dennis Denney, contains highlights of paper SPE 137998, ’Uncertainty Management in a Major CO2 EOR Project,’ by P.M. O'Dell, SPE, and Kirby C. Lindsey, SPE, Occidental Oil and Gas, prepared for the 2010 Abu Dhabi International Petroleum Exhibition & Conference, Abu Dhabi, UAE, 1-4 November. The paper has not been peer reviewed.
Occidental operates a large turbidite-sandstone reservoir in California that has progressed through primary depletion, limited gasflooding, and extensive waterflooding. Although a CO2-injection project could increase oil reserves significantly and extend productive field life, the project has significant technical and economic challenges. Subsurface uncertainties were evaluated, including the use of a two-stage design-of-experiments approach to choose optimal operator choices, regardless of subsurface uncertainties. Probabilistic production forecasts were prepared for project valuation.
Reservoir Model
A new geological model was assembled for this enhanced-oil-recovery (EOR) study. There are 14 correlated sands in the reservoir with limited vertical communication between them resulting from on-lapping of the sands and thinning of the intersand shales. Reservoir structure, a pay flag, pay porosity, and clay content were obtained from well-log data. A porosity/permeability transform that is based on significant core data and considers clay content are used to populate properties in the model.
Fluid modeling was based on extensive conventional and CO2 laboratory work conducted on a reservoir-oil sample collected in 2000. This sample was not an original-fluid sample, and to account for reservoir and historical variations, different reservoir-fluid compositions were created by varying the molar ratio of oil vs. solution gas in the sample. Simulated CO2-flood performances under the different compositions created in this manner were nearly indistinguishable from one another except for a small change in gas/oil ratio (GOR).
Relative permeabilities were modeled with Corey curves that were constructed on the basis of laboratory data. Because history matching had been performed on an earlier model, no history matching was performed here and the model was initialized at or slightly above residual-oil saturation to water (assuming fairly complete waterflood sweep in the subject sands).
Most of the existing wells are serviceable and will be used in any future development. The existing well stock and configuration are amenable to 18-acre five-spot patterns. Hypothetical full-interval-completion wells were used in the study rather than actual wells. Bottomhole injection-pressure and -temperature data were calculated on the basis of tubinghead pressure and temperature and variation of CO2 properties through the wellbore. Lift curves were generated with wellbore-modeling software and use of ranges of possible variations in water cut, producing GOR, tubinghead pressure, tubing sizes, oil-production rate, and nominal depth. Because the contemplated project will be a pressure-maintenance project, the composition of the reservoir fluid was assumed to be fixed for the purpose of calculating lift curves. Therefore, any change in the producing GOR would be the result of CO2 breakthrough. Output of the wellbore-modeling software had to be stitched together to account for the compositional change in the fluids being lifted.
