Forward Modeling Complex Diffusive Gradients and Matching Measured Bimodal Diffusive and Quasi-Equilibrium Fluid Distributions in Connected Reservoirs

Summary Given enough time after the onset of hydrocarbon charge, well-connected oil columns might be expected to exhibit equilibrated fluid gradients. However, a well-connected tilted sheet reservoir, deepwater Gulf of Mexico/America exhibits a bimodal compositional distribution where asphaltenes, solution gas-oil ratio (GOR), and methane carbon isotopes at the base of the column all exhibit a quasi-equilibrium gradient, whereas the GOR and methane carbon isotopes at the top of the column exhibit a diffusional gradient. In addition, the solution gas methane shows a large diffusive gradient of biogenic methane at the top of the oil column, whereas the base of the oil column shows solution gas methane that is 50% biogenic and 50% thermogenic. A proposed origin of this enigmatic oil column was recently proposed. Here, we explore this proposed origin with a compositional simulator. We simulated the history of reservoir charge over geologic time, investigated the origin of complex fluid compositional variations, and forward-modeled diffusive gradients at the top of the column and well-mixed fluids in the bottom half of the column, explaining present-day fluid realizations. With this capability, the model has predictive capabilities to examine fluid distributions in undeveloped sections of the reservoir. Oil samples were geochemically evaluated showing increasing biogenic gas toward the top of the oil column, while the lower half of the column remained largely invariant. Extreme variation of GOR was measured in the upper half of the oil column. Simulations were performed to evaluate the origin of complex fluid distributions. We modeled diffusional fluxes to evaluate the upper half of the column, implemented the cubic equation of state (EoS) to model the GOR gradient in the lower half, and fit asphaltene gradients with the Flory-Huggins-Zuo equation of state (FHZ EoS). Gas charged into the base of the oil column and the induced convection accounted for the well-mixed biogenic and thermogenic methane in the solution gas in the lower half of the oil column. Various diffusion models were tested to determine a reliable model for long geologic times. Modeling approaches were compared, including black-oil vs. compositional simulators, and we investigated parameters controlling diffusion timelines, such as dip angle and subsidence time. Simulations of gas charge at a point source into an undersaturated oil reservoir showed rapid gas dissolution into oil and excellent convective mixing. With excess gas charge, saturation pressure was attained and subsequent gas charge remained as a separate gas phase and migrated to the top, forming a gas cap. After the gas charge was completed, subsidence increased saturation pressure, thereby enabling gas diffusion into the oil column. Diffusion modeling with a compositional simulator allows gas components from the gas cap to diffuse down as well as light components in the oil to diffuse up. This has resulted in the elimination of the phase boundary (no meniscus), which depends on the initial volumetrics of the charge fluids, among other factors. In the present day, the upper half of the oil column (thus the whole oil column) is far from equilibrium; the dynamics of diffusion dominate. We found that compositional simulators are more accurate than black-oil simulators for our objective as they account for the diffusion of both gas components into the oil column and light-end oil components up into the gas cap.