Unconditional sequential Gaussian simulation for 3-D flow in a heterogeneous core

Abstract

Three-dimensional (3-D) porosity distribution realizations from unconditional sequential Gaussian simulations were input into a 3-D miscible displacement simulator to model miscible flow through a heterogeneous carbonate core containing a residual phase. The displacement simulations using unconditional sequential Gaussian simulation generated data as input were compared to the displacement simulations using the actual experimental data as input in order to evaluate the effectiveness of unconditional sequential Gaussian simulation for these applications. Previous X-ray computer tomography experiments provided the porosity and residual saturation distribution parameters. The experimental porosity data were fit to spherical, exponential, Gaussian and power semivariogram models. The power model provided the best fit followed by the exponential model. Both the power and the exponential models were used in the fluid displacement simulations. Both models predicted similar effluent profiles with relatively little spread between realizations. Breakthrough occurred late in all runs using the simulated input data, but otherwise the effluent profiles were in excellent agreement with the results using the experimental data. The spread between realizations, however, was not large enough to include all the points calculated using the experimental data. Differences between results using experimental data and sequential Gaussian simulation generated data were accentuated in numerical displacement simulations where a single miscible pulse was injected.