Core Digitization: A Journey to Digital Transformation of Unconventional Mudstone Reservoir

Digital innovation is transforming core analysis workflows in unconventional mudstone reservoirs by integrating high-resolution datasets with machine learning. This study aims to enhance reservoir characterization by employing core digitization to generate high-resolution datasets, improve core-to-log correlations, automate analytical processes, and establish a digital core repository for future applications. A non-destructive core-scanning method was utilized to acquire a depth-co-registered, multiparameter dataset at centimeter-scale resolution. The dataset includes gamma ray (GR), density (Den), P-wave velocity (Vp), magnetic susceptibility (MS), and X-ray fluorescence (XRF) measurements. The analysis of this dataset reveals significant vertical heterogeneity in lithofacies, with quantifiable variations in mineralogical composition (clay-carbonate-silica fractions), organic geochemical parameters (TOC, HI, thermal maturity), petrophysical attributes (porosity, fluid saturation, permeability), and geomechanical properties. By integrating high-resolution core data with machine learning-aided lithofacies classification, this study enables the systematic identification of depositional cyclicity that was previously unresolved through manual core descriptions and beyond well-log resolution. This data-driven approach minimizes subjective bias in sample description and selection while facilitating multivariate statistical integration with downhole petrophysical data. The implementation of this method improves workflow efficiency by 40-60% compared to conventional techniques, reducing manual processing time and associated costs while preserving core samples for future research. The enhanced resolution and predictive capabilities of this approach contribute to more accurate hydrocarbon reserve estimation, improved identification of sweet-spot zones, and reduced development risks in complex mudstone plays. As industry advances toward data-driven reservoir characterization, integrating high-resolution digital datasets with machine learning represents a transformational shift in optimizing unconventional resource development and maximizing hydrocarbon recovery.