Assessment of Petrophysical and Textural Heterogeneity Using Core-Scale Images, Image Logs, and NMR Measurements

Abstract

Abstract Assessment of subsurface heterogeneity in carbonate reservoirs is critical for optimizing core sampling locations for reliable calibration of core- and reservoir-scale petrophysical models. The objectives of this paper are (a) extracting NMR-based pore-size distribution (PSD) parameters (i.e., parameters defining the best multi-modal Gaussian function to PSD) to derive pore-scale heterogeneity (HTIpore), (b) extracting image-based textural features from image logs that capture the spatial distribution of rock components to estimate rock textural heterogeneity (HTItextural), and (c) proposing a workflow that honors pore- to log-scale information to quantify depth-by-depth petrophysical heterogeneity index (HTI). To achieve these objectives, we quantify parameters describing PSD through fitting multimodal Gaussian functions to NMR T2 distribution data quantitatively via an automatic inversion method. We use extracted PSD parameters to compute depth-by-depth pore-scale characteristic features (i.e., pore-scale characteristic values and weight factors) to derive a depth-by-depth variability index. Then, we develop an analytical model using the calculated variability index values to assess depth-by-depth pore-scale heterogeneity. Next, extract textural features using image logs (i.e., acoustic image logs in this paper) by employing the gray-scale cooccurrence matrix (GLCM) algorithm at various moving window sizes throughout the images. This step enables quantifying rock texture across multiple scales. We conduct principal component analysis (PCA) on the extracted features to obtain rock textural characteristic values (RTCVs) that capture maximum spatial variance in rock texture. RTCVs are used to estimate depth-by-depth textural heterogeneity. Finally, we introduce a new analytical index for depth-by-depth petrophysical heterogeneity, called HTI, through integrating pore-scale and log-scale textural heterogeneities. We successfully applied the proposed method to a field dataset from a well drilled in a Brazilian pre-salt carbonate sequence. The extracted textural features from image logs were used to obtain image-based rock classes. We identified five image-based rock classes corresponding to lacustrine carbonates including varying degrees of mud and spherulites (spherulitic shrub stone, shrub stone with calcite inclusion, cemented packstone, cemented packstone with shrub fragments, and shrub stone). Results demonstrated variations in extracted well-log-scale rock textural and pore-scale characterizing features capturing variations in compositional/petrophysical rock properties within each rock class at multiple scales. This enabled the comparison of the degree of heterogeneity in the pore structure and the rock texture associated with different scales of investigation within the formation. We detected quadrimodal Gaussian distribution characterizing the pore-size distribution ranging from 0.03 to 850 µm. The introduced HTI ranked rock classes based on local heterogeneity and located depth interval having the highest heterogeneity within the rock class, which was in agreement with the textural content of the images. HTI results showed that rock classes corresponding to shrub stone having calcite inclusion and cemented packstone have the highest and lowest local heterogeneity, respectively. When making decisions on core sampling, more core samples were detected to be required to be analyzed from the depth interval with higher HTI. The proposed workflow integrates information from multiple measurement scales (from pore- to well-log-scale) to quantify depth-by-depth spatial heterogeneity. The method proposed in this paper enables honoring quantitative rock textural features and the impact of measurement scale in quantifying heterogeneity. It also potentially enables making real-time decisions on the optimum locations of core samples for laboratory measurements to enhance reliable petrophysical evaluation and reservoir characterization at optimum cost.