Laboratory Sandstone Micro-CT Volume Data for Porosity-Controllable Generation of Three-Dimensional Porous Geomaterials

Description

This dataset contains laboratory sandstone micro-CT volume data and associated metadata used in the study entitled “A Porosity-Controllable Generative Framework for Three-Dimensional Porous Geomaterials Based on Discrete Latent-Space Diffusion”. The micro-CT volume represents a binary sandstone digital core with a size of 800 × 800 × 800 voxels and a voxel resolution of 3.5 μm/voxel. In the binary volume, voxel value 0 denotes the solid matrix and voxel value 1 denotes the pore space. The dataset is provided to support the construction of training samples and the development of porosity-controllable three-dimensional porous geomaterial generation models. The associated metadata file describes the data format, voxel size, image dimensions, voxel-value convention, and basic usage information. The Fontainebleau sandstone digital rock samples used for independent validation in the associated manuscript were obtained from a previously reported Australian National University (ANU) digital rock dataset and are not redistributed in this repository.

Steps to reproduce

The dataset was obtained from laboratory micro-CT scanning of a sandstone sample followed by image segmentation to produce a binary pore–solid volume. The final volume file, Bei_800x800x800.raw, is a RAW binary file with dimensions of 800 × 800 × 800 voxels. The voxel resolution is 3.5 μm/voxel, corresponding to a physical size of 2.8 mm × 2.8 mm × 2.8 mm. The data type is uint8. In the binary volume, voxel value 0 represents the solid matrix, and voxel value 1 represents the pore space. The voxel-calculated porosity was obtained by dividing the number of pore-space voxels by the total number of voxels. For this volume, the number of pore-space voxels is 66,679,025 and the total number of voxels is 512,000,000, giving a voxel-calculated porosity of 0.130232, or 13.023%. The RAW file can be inspected and reproduced using the following Python workflow: import numpy as np data = np.fromfile("Bei_800x800x800.raw", dtype=np.uint8) data = data.reshape((800, 800, 800)) porosity = np.mean(data == 1) print("Shape:", data.shape) print("Data type:", data.dtype) print("Unique values and counts:", np.unique(data, return_counts=True)) print("Voxel-calculated porosity:", porosity)

Institutions

• China University of Petroleum, East China

  Shandong, Qingdao

Categories

Licence