Distinguishing Carbonate Reservoir Pore Facies with Nuclear Magnetic Resonance Measurements

Abstract

Characterization of carbonate rocks may involve identifying the important pore types which are present. In the past, this task has required detailed petrographic analysis of many core samples. Here, we describe a method which uses nuclear magnetic resonance (NMR) measurements to reduce the amount of petrographic analysis needed for porosity typing of carbonate reservoir rocks. For a rock sample which has been measured with NMR, our method decomposes the log(T2) spectrum into at most three Gaussian-shaped components and gives a set of nine parameters. Two characteristic quantities having geological significance are extracted from the nine parameters. Values of the two quantities are compared with a reference set, established from samples having both NMR and petrographic evaluations of porosity types. We use a Bayesian approach to the classification of the dominant porosity type. Tests of our method on 103 samples show a correct prediction in 60 to 90 percent of the samples. The lower success rate was obtained for samples with five porosity types from three fields; the higher success rate obtained with samples with three porosity types from one well. The use of geologically significant quantities extracted from the decomposition gives comparable success rate to those obtained using a standard, non-geological approach such as canonical variates.