Assessment of Textural Heterogeneity Tensor Using 3D Micro-CT-Scan Images

Quantification of subsurface heterogeneity and anisotropy in complex carbonate reservoir is crucial for optimizing drilling/completion techniques for developing a reliable reservoir scale petrophysical models. Subsurface heterogeneity is governed by rock texture (i.e., spatial arrangement of rock components) that affect the petrophysical and mechanical properties. Recently, our research group introduced a depth-by- depth heterogeneity index that incorporate rock texture in heterogeneity assessment (Sahu et al., 2024). However, the proposed method does not account for the influence of direction in the assessment of heterogeneity. Directional assessment of heterogeneity can help in optimizing the drilling, completion, and reservoir management strategies. Therefore, objectives of this project are (a) extracting rock textural features using 3D micro-CT-scans images in three orthogonal directions, (b) deriving rock textural characteristic values using extracted rock textural features across multiple scales of investigation, and (c) defining heterogeneity tensor and heterogeneity index that enables quantifying 3D spatial variability of rock texture within the formation. First, we preprocess 3D micro-CT-scan images and extract textural features in three orthogonal directions (i.e., XY, XZ, and YZ) from 3D micro-CT-scans images using the gray-scale cooccurrence matrix (GLCM) algorithm. GLCM is applied within varying sliding windows across multiple rock slices in each direction, capturing spatial variations in depositional and geological attributes. Next, principal component analysis (PCA) is applied to the extracted textural features, generating rock textural characteristic values (RTCV) for each direction. RTCV capture the maximum spatial variation in rock texture within each direction. We then fit multimodal Gaussian functions to the RTCV using an automatic inversion method to quantify directional spatial rock texture variations. Statistical parameters of fitted gaussian are used to develop an analytical model for the heterogeneity tensor (HTT). We also compute the heterogeneity index (HTI) by performing PCA on textural features extracted from rock slices in three orthogonal directions across varying scales of investigation. The proposed HTT and HTI enable assessment of spatial variability of rock texture across the formation, incorporating different directions and scales of investigation. We successfully applied the proposed method to 3D synthetic rock sample and 3D micro-CT-scan images from a well drilled in a Brazilian pre-salt carbonate sequence. We identified four image-based rock classes corresponding to lacustrine carbonates having varying degrees of mud and spherulites. We extracted rock textural features from micro-CT-scan images that capture variations in compositional/petrophysical rock properties in three orthogonal directions (i.e., XY, YZ, and ZX). This enabled comparison of different rocks in terms of the degree of heterogeneity in rock texture, while capturing both spatial variations and differences across scales of investigation. The introduced HTT quantify variation within rock texture in different direction that directly influences fluid flow. The proposed HTT enables making better decision on production rates, overall efficiency of drilling, completion, and reservoir management strategies. The introduced HTI ranked rock classes based on local heterogeneity in rock texture that affects the variation of petrophysical properties. The novelties of this work include taking into account of direction in quantifying heterogeneity tensor using extracted rock textural features, reducing the uncertainty in heterogeneity assessment, and enabling in making decisions on optimizing drilling and production operations.