Summary Accurately determining rock mechanical parameters in tight sandstone reservoirs is often challenged by limited core availability and heterogeneous lithologies. We present a novel data-driven approach for predicting the microscopic Young’s modulus of drill cuttings by integrating nanoindentation experiments with machine learning techniques based on multisource microstructural features. A comprehensive data set was established using nanoindentation testing, thin-section imaging, micro-scale computed tomography (micro-CT) analysis, and X-ray diffraction (XRD) measurements from cuttings collected in the Tarim Basin. To capture the complex nonlinear relationships between microstructure and elasticity, a decision tree (DT)–based regression model was developed using features such as mineral content, grain size distribution, pore geometry, and cementation. The model achieved high predictive accuracy, demonstrating robustness across different wells. With this study, we provide an efficient, cost-effective framework for acquiring reservoir mechanical parameters from cuttings data, supporting improved reservoir evaluation, completion design, and stimulation planning in data-scarce tight formations.
Huang et al. (Mon,) studied this question.