Reconstruction of three-dimensional porous media from a single two-dimensional image using three-step sampling

Area of Science:

• Geoscience
• Image Analysis
• Computational Modeling

Background:

• Reconstructing 3D porous media from 2D images is crucial for understanding material properties.
• Maintaining continuity and variability between reconstructed layers is essential for accurate representation.
• Existing methods struggle to precisely control these inter-layer characteristics.

Purpose of the Study:

• To develop an original sampling method for accurate 3D porous medium reconstruction.
• To ensure continuity and variability consistency with the training image (TI).
• To validate the method's efficacy using a Berea sandstone sample.

Main Methods:

• Proposed a "three-step sampling" method for 3D porous medium reconstruction.
• Sampling points extracted from template centers and edge areas using a two-point correlation function.
• Continuity and variability controlled across adjacent layers during sampling.

Main Results:

• The three-step sampling method successfully reconstructed 3D porous media.
• Reconstructed results closely matched the original sample's statistical characteristics.
• Key properties like porosity, pore/throat size, and relative permeability were accurately reproduced.

Conclusions:

• The proposed three-step sampling method effectively reconstructs 3D porous media from 2D images.
• This technique ensures high fidelity in replicating the statistical properties of the original medium.
• The method offers a reliable approach for digital rock physics and material science applications.