Unstable EOR Displacements and Their Prediction Using the Karhunen-Loeve (K-L) Decomposition

摘要

The paper describes a novel approach to model unstable fluid displacements in porous media. The approach is based on the Karhunen-Loeve (K-L) decomposition which is able to predict the fluid distributions of miscible displacements inside a porous medium. Several first-contact miscible displacement experiments, each with different fluid properties, were conducted, and the fluid distributions inside the porous media were mapped at various times using nuclear magnetic resonance imaging (NMRI). The K-L decomposition is described to identify the coherent spatial structures from spatio-temporal patterns arising from these experiments. It was found that these complex spatio-temporal bvehavior can be successfully described by few dominant eigenfunctions. The technique is based on the diagonalization of the covariance or two-point correlation matrix. The K-L decomposition provides information for the successful prediction of Enhanced Oil Recovery (EOR) processes.