Multi-Scale and Multi-Physics Quantitative Workflow for Integrated and Image-Assisted Rock Classification in Complex Carbonate Formations

Abstract

Diagenetic processes in carbonate formations result in rapid spatial variation in rock fabric (i.e., spatial distribution of rock components) and complex pore structure. Therefore, assessment of petrophysical properties, necessary for efficient exploitation of hydrocarbon-bearing carbonate reservoirs, can be challenging. Estimation of the aforementioned properties requires integration of multi-scale and multi-physics data that captures rock fabric at multiple scales. Image logs offer high-resolution information of the rock fabric and its variation. However, this information is typically employed qualitatively, especially in carbonate formations. The objectives of this paper include (a) quantifying the visual content of high-resolution image logs through computation of image-based rock-fabricrelated features, (b) integrating conventional well logs with image-based features and core measurements for rock classification. First, we conducted joint inversion of well logs to estimate volumetric concentration of minerals and total porosity. Then, we characterized the pore structure of the evaluated intervals fitting a multimodal Gaussian model to the available nuclear magnetic resonance (NMR) transverse relaxation (T2) distribution logs and core measurements. Next, we employed image analysis techniques to capture variations of rock fabric in the form of greyscale and textural features. We integrated compositional, pore structure, and image-based rock fabric related features for rock classification. Finally, we evaluated various strategies to select the adequate number of rock classes using the available data on the evaluated depth interval. We applied the proposed workflow to one well intersecting a complex carbonate formation. We demonstrated that the extracted image-based rock fabric features and the pore structure features from NMR T2 distribution captured the complex pore structure and rapid spatial variation of rock fabric observed in the evaluated depth interval. We honored spatial variation in rock fabric at pore- and log-scale by selecting the adequate number of rock classes using log and core NMR T2 distribution data, image-based features, and petrophysical properties of the evaluated depth interval. The number and location of obtained integrated rock classes honored rock fabric variation at both the coreand log-scale. A unique contribution of the proposed workflow is the quantification of the visual content of image logs and their integration with core-/log-scale measurements, improving rock classification and potentially formation evaluation efforts by accounting for rock fabric variation at different scales and honoring various physical measurements, which is not attainable through conventional rock typing techniques. This becomes specifically important in carbonate formations, with complex pore structure and rapid variation in rock fabric.