Reproducible material for A Wasserstein GAN with gradient penalty for 3D porous media generation.
Corrales M., Izzatullah M., Hoteit H., and Ravasi M.
Submitted to Second EAGE Subsurface Intelligence Workshop, 28-31 October 2022, Manama, Bahrain
This repository is organized as follows:
- 📂 checkpoints: folder containing the trained generator every two epochs for RockGAN and CRockGAN.
- 📂 data: folder containing data and instructions on how to retrieve the data.
- 📂 figures: folder containing the 3D figures of the results obtained.
- 📂 notebooks: set of jupyter notebooks reproducing the experiments in the paper (REV, Training for RockGAN and CRockGAN, and metrics by epochs).
- 📂 rockgan: package of the project.
The following notebooks are provided:
- 📙
01_Representative_Elementary_Volume_REV.ipynb: notebook performing Representative Elementary Volume for porosity and permeability to determine sub-volume size for data augmentation. - 📙
02_Training_RockGAN.ipynb: notebook performing Training of RockGAN. - 📙
03_Training_CRockGAN.ipynb: notebook performing Training of CRockGAN. - 📙
04_Results_Minkowski_by_epochs.ipynb: notebook performing comparison results of the Minkowski functionales by epochs (RockGAN VS CRockGAN). - 📙
05_Results_TwoPointStat_by_epochs.ipynb: notebook performing comparison results of the Two-point statistics by epochs (RockGAN VS CRockGAN). - 📙
06_Results_Permeability_by_epochs.ipynb: notebook performing comparison results of permeability by epochs (RockGAN VS CRockGAN). - 📙
07_Results_3D_Visualization.ipynb: notebook performing 3D visualization of the samples after training.
To ensure reproducibility of the results, we suggest using the environment.yml file when creating an environment.
Simply run:
./install_env.sh
It will take some time, if at the end you see the word Done! on your terminal you are ready to go. After that you can simply install your package:
pip install .
or in developer mode:
pip install -e .
Remember to always activate the environment by typing:
conda activate rockgan
Disclaimer: All experiments have been carried on a Intel(R) Xeon(R) CPU @ 2.10GHz equipped with a single NVIDIA GEForce RTX 3090 GPU. Different environment configurations may be required for different combinations of workstation and GPU.
- Mosser, L., Dubrule, O., & Blunt, M. J. (2017). Reconstruction of three-dimensional porous media using generative adversarial neural networks [https://github.com/LukasMosser/PorousMediaGan]
- Gostick J, Khan ZA, Tranter TG, Kok MDR, Agnaou M, Sadeghi MA, Jervis R. PoreSpy: A Python Toolkit for Quantitative Analysis of Porous Media Images. Journal of Open Source Software, 2019. doi:10.21105/joss.01296 [https://github.com/PMEAL/porespy]
- Gostick et al. "OpenPNM: a pore network modeling package." Computing in Science & Engineering 18, no. 4 (2016): 60-74. doi:10.1109/MCSE.2016.49 [https://github.com/PMEAL/OpenPNM]
- Arnout M.P. Boelens, and Hamdi A. Tchelepi, QuantImPy: Minkowski functionals and functions with Python, SoftwareX, Volume 16, 2021, 100823, ISSN 2352-7110, doi: 10.1016/j.softx.2021.100823 [https://github.com/boeleman/quantimpy]
@article{eage:/content/papers/10.3997/2214-4609.2022616005,
author = "Corrales, M. and Izzatullah, M. and Hoteit, H. and Ravasi, M.",
title = "A Wasserstein GAN with Gradient Penalty for 3D Porous Media Generation.",
journal= "",
year = "2022",
volume = "2022",
number = "1",
pages = "1-5",
doi = "https://doi.org/10.3997/2214-4609.2022616005",
url = "https://www.earthdoc.org/content/papers/10.3997/2214-4609.2022616005",
publisher = "European Association of Geoscientists & Engineers",
issn = "2214-4609",
type = "",
}