Processed 3D Tomographic Images of Cyclic CO2 and Brine injections into Bentheimer Sandstone

This data arises from a series of core-flooding experiments intended to inform knowedge of carbon dioxide (CO2) sequestration in geologic reservoirs.

Four cycles of supercritical CO2 ("scCO2") and 1.0 M KI brine injections were undertaken in a Bentheimer sandtsone core under reservoir-relevant temperature and pressure conditions (40 deg C, 1250PSI/8.6 MPa). 3D X-ray tomograms were acquired at the endpoint of every injection process. Scanning was undertaken using the in-house developed high-pressure core-flooding apparatus at the ANU CTLab at a resolution of 3.74978 micron/voxel edge.

The raw grayscale tomographic data volumes for all "initial" (post-CO2 injection) and "residual" (post-brine injection) scans were first registered to the dry tomographic data set to allow direct and precise comparison of phase distributions in each data set. Data volumes were cropped to a central region, and grayscale intensity variation (due to beam hardening and attenuation distortion due to high-density flow diffusers at the core ends) was corrected in radial and vertical directions. The dry scan was segmented into three phases (pore, solid grains, and clays) using converging active contours "CAC") segmentation routines. Four-phase segmentation (scCO2, brine, solid grains, and clays) was accomplished by overlaying the phase-labeled dry data set on all the registered initial and residual data volumes, and performing a final additional CAC routine to distinguish between scCO2 and brine, only on the regions labeled "pore" in the dry data set. Speckle noise removal was accomplished by removing scCO2-labeled clusters smaller than 268 voxels (equivalent to a sphere with a radius of 15.0 micron), and also by removing any isolated inclusions from within the scCO2 phase. The final data volumes used for quantitative analysis were cylinders with a diameter of 8.14 mm and a height of 27.3 mm, equivalent to a physical volume of 1,420 cubic mm (or 2.69 X 10^10) voxels); and had an image-based porosity of 21.3%.

Data is segmented and in .nc format (https://www.unidata.ucar.edu/software/netcdf/).
For "DRY" data set: 0=void/pore space; 1=Sandstone grains; 2=clays.
For initial ("SI#") and residual (SR#) data: 0=scCO2; 1=brine; 2=sandstone grains; 3=clays

Full experimental and analytical details are included in:
A.L. Herring, R.T. Sun, C. Armstrong, Z. Li, J.E. McClure, M. Saadatfar, Evolution of Bentheimer Sandstone Wettability During Cyclic scCO2-Brine Injections, Water Resour. Res. (2021).

本数据集源自一系列岩心驱替实验，旨在为地质储层中二氧化碳（CO₂）封存的相关认知提供依据。

实验在本特海姆砂岩（Bentheimer sandstone）岩心中开展了4轮超临界CO₂（"scCO₂"）与1.0 M碘化钾盐水的注入循环，实验条件匹配储层环境：温度40℃，压力1250 PSI（即8.6 MPa）。每次注入流程结束后，均采集三维X射线断层扫描图像。扫描工作依托澳大利亚国立大学CT实验室（ANU CTLab）自研的高压岩心驱替装置完成，体素分辨率为3.74978微米/体素边长。

所有“初始态”（CO₂注入后）与“残余态”（盐水注入后）扫描得到的原始灰度断层数据体，首先被配准至干岩心断层数据集，以实现各数据集内相分布的直接且精准对比。随后将数据体裁剪至中心区域，并校正径向与垂直方向上的灰度强度偏差——该偏差源于岩心端部高密度流动扩散器造成的束硬化与衰减畸变。

采用收敛主动轮廓（converging active contours, CAC）分割算法，将干扫描图像分割为三相：孔隙空间、固体颗粒与黏土。后续通过将带相标签的干数据集叠加至所有配准后的初始态与残余态数据体，并仅针对干数据集中标记为“孔隙”的区域执行额外的CAC分割流程，完成四相分割（超临界CO₂、盐水、固体颗粒与黏土），以区分超临界CO₂与盐水相。

斑点噪声去除通过两种方式实现：一是移除尺寸小于268个体素（等效于半径15.0微米的球体）的超临界CO₂标记簇；二是移除超临界CO₂相内的所有孤立夹杂物。用于定量分析的最终数据体为圆柱体，直径8.14 mm，高度27.3 mm，对应物理体积1420立方毫米（或2.69×10¹⁰个体素），基于图像计算得到的孔隙度为21.3%。

数据集已完成相分割，存储为.netCDF格式（https://www.unidata.ucar.edu/software/netcdf/）。其中干数据集的标签映射规则为：0=孔隙空间；1=砂岩颗粒；2=黏土。初始态（SI#）与残余态（SR#）数据集的标签映射规则为：0=超临界CO₂；1=盐水；2=砂岩颗粒；3=黏土。

完整的实验与分析细节参见下述文献：A.L. Herring, R.T. Sun, C. Armstrong, Z. Li, J.E. McClure, M. Saadatfar, 《循环超临界CO₂-盐水注入过程中本特海姆砂岩润湿性的演化》，Water Resour. Res. (2021).