Reconstructing Petrophysical Borehole Images: Their Potential for Evaluating Permeability Distribution in Heterogeneous Formations

Abstract

Abstract Borehole electrical images have very high vertical and azimuthal resolution, and thus have proved to be of paramount importance over the past 10 years in assessing formation layering and internal organization. Standard openhole log measurements (e.g., density or photoelectric factor) have also experienced improvements in accuracy and vertical resolution. These improvements are the result of more compact integrated tools and the systematic use of accelerometer-based speed corrections and inclinometry devices to measure the orientation of sensors within the borehole. Recent advances in nuclear magnetic resonance (NMR) logging have also cast a new light on petrophysical evaluation, especially on continuous permeability logs. The technique proposed in this paper combines information from borehole images and advanced petrophysical logs in a two-step approach. In the first step, a transform or relation is established between two oriented borehole measurements. For example, the transform is computed between azimuthal resistivity from an imaging tool and a high-resolution density measurement, conditioned on the readings of nondirectional volumetric measurements such as neutron porosity, gamma ray and image-derived textural indicators. This transform, which is in most cases inherently nonlinear, is not explicitly determined. Rather, it is approximated by an artificial neural network, which is trained over key representative facies zones. As a quality control step, the procedure uses the resistivity and other measurements to compute a synthetic density curve. This is compared to the actual density log, and error bars are attached to the reconstruction process. In the second step, the same transform is then applied to each available azimuthal resistivity channel (column in the electrical image) to reconstruct a quantitative density image on the borehole wall from these reconstructed density channels. The same approach provides a method to interpret a permeability curve from NMR data. The recorded orientation of the CMR* Combinable Magnetic Resonance tool skid enables a transform to be established between NMR-derived permeability and azimuthal resistivity, azimuthal density and nondirectional volumetric measurements (neutron porosity, gamma ray and textural indicators). Reconstructed permeability images can then be obtained by applying the transform to the full suite of image channel data. The proposed technique is illustrated with examples from clastic and carbonate formations. It provides a powerful means to better understand permeability variations in heterogeneous formations.