Speaker
Description
Core CT imaging is a fundamental tool for fracture identification, quantitative pore-structure characterization, and the estimation of reservoir and engineering parameters. However, its application to multiscale reservoir storage space characterization remains challenging due to limitations in image resolution, contrast, and scale heterogeneity. Here, we develop a Smart Core workflow for the intelligent identification of multiscale pore-throat-fracture systems and the prediction of reservoir storage space parameters from core CT images. The workflow integrates convolutional neural networks and Transformer architectures to enable multiscale feature learning and the unified representation of macroscopic fractures and microscopic pore-throat structures, substantially improving the detection of weak fractures and complex pore networks. To overcome intrinsic resolution constraints, a Transformer-based super-resolution reconstruction strategy is employed to enhance microfractures and fine-scale pore structures, thereby increasing the resolvability and quantitative fidelity of multiscale storage space characterization. Building on these advances, geometric and statistical descriptors of the pore-throat-fracture system are extracted and linked to reservoir petrophysical properties and mechanical responses, enabling the prediction of key parameters such as permeability. The proposed approach significantly extends the capability of core CT imaging for multiscale reservoir characterization and provides a robust data-driven basis for refined reservoir evaluation and engineering decision-making.
| Country | China |
|---|---|
| Acceptance of the Terms & Conditions | Click here to agree |
