Benchmark dataset for pore-scale CO2-water interaction
Data files
Mar 25, 2025 version files 209.88 GB
-
het_1_flipped.zip
18.47 GB
-
het_1.zip
18.98 GB
-
het_2_flipped.zip
24.92 GB
-
het_2.zip
25.74 GB
-
het_3_flipped.zip
22.82 GB
-
het_3.zip
23.53 GB
-
het_4_flipped.zip
20.27 GB
-
het_4.zip
20.73 GB
-
het_5_flipped.zip
16.64 GB
-
het_5.zip
17.80 GB
-
README.md
4.40 KB
Abstract
Accurately capturing the complex interaction between CO2 and water in porous media at the pore scale is essential for various geoscience applications, including carbon capture and storage (CCS). We introduce a comprehensive dataset generated from high-fidelity numerical simulations to capture the intricate interaction between CO2 and water at the pore scale. The dataset consists of 624 2D samples, each of size 512x512 with a resolution of 35 μm, covering 100 time steps under a constant CO2 injection rate. It includes various levels of heterogeneity, represented by different grain sizes with random variation in spacing, offering a robust testbed for developing predictive models. This dataset provides high-resolution temporal and spatial information crucial for benchmarking machine learning models.
https://doi.org/10.5061/dryad.jm63xsjn5
Dataset Summary:
The dataset presented consists of 624 high-resolution, two-dimensional samples, each with a size of 512 × 512 pixels and a spatial resolution of 35 µm. It captures the dynamic interaction between CO2 and water in porous media over 100 time steps, with a constant CO2 injection rate. The dataset includes varying levels of geological heterogeneity, represented by different grain sizes and random spacing, simulating realistic conditions in porous materials. This dataset is designed to facilitate the development and benchmarking of machine learning models for predicting CO2 behavior during multiphase flow in porous media. It provides both temporal and spatial granularity, making it a valuable resource for advancing machine learning applications in geosciences.
Description of the data and file structure
.zip Files
Description: The dataset contains 5 different levels of heterogenities (variations of grain sizes across the spatial domain), where the hetrogenitiy increases from level 1 to level 5. Each folder is named as het_{heterogenity level}. The word “flipped” in the file name indicates that the simulation was performed on a flipped version of the physical domain. Each zip archive has the following structure:
het_1.zip/
│── example1.hdf5
│── example2.hdf5
│── example1_csv_files/
│ ├── poroPerm.csv
│ ├── relperm.csv
│── example2_csv_files/
│ ├── poroPerm.csv
│ ├── relperm.csv
.
.
.
.hdf5 Files
Description: Each HDF5 file has the simulation of CO2, the velocities, pressure, and saturation. Each file has the following keys:
| Key | Size | Description | Units |
|---|---|---|---|
| Ux | 512x512 | x-component of flow velocity | m/s |
| Uy | 512x512 | y-component of flow velocity | m/s |
| alpha_water | 100x512x512 | Water saturation field | Dimensionless (0 to 1) |
| img | 512x512 | Physical domain | Dimensionless (Binary) |
| p | 100x512x512 | Pressure field | Pa (Pascal) |
| pc | 100x512x512 | Capillary pressure field | Pa (Pascal) |
.csv Files
Description: The .csv files contains information about the porosity, permeability as in the following table:
| File | Description |
|---|---|
| poroPerm.csv | Time, porosity, permeability (m2), the characteristic pore length L, the Reynolds number Re, and the Darcy velocity UD at the beginning of the simulation before any CO2 is injected into the model. |
| relperm.csv | Porosity, permeability (m2), and the capillary number of each phase (Ca1 for water and Ca2 for CO2) at the beginning of the simulation. The saturation of water Sw, the relative permeability of water krw, and the relative permeability of CO2 kwo are shown for each output timestep. |
Code/software
The input files used to simulate CO2 flow is built using GeoChemFoam (J. Maes and H. P. Menke, 2022) and is available at https://github.com/alhasan-abdellatif/Parallel-Simulation-of-CO2. The code is written in Python 3.11.9 and the list of the requirements is shown in the readme file. GeoChemFoam can be downloaded from http://github.com/GeoChemFoam and has been validated against experimental data (B. Zhao et al. 2019).