U-FNO -- An enhanced Fourier neural operator-based deep-learning model for multiphase flow
Abstract
Numerical simulation of multiphase flow in porous media is essential for many geoscience applications. Machine learning models trained with numerical simulation data can provide a faster alternative to traditional simulators. Here we present U-FNO, a novel neural network architecture for solving multiphase flow problems with superior accuracy, speed, and data efficiency. U-FNO is designed based on the newly proposed Fourier neural operator (FNO), which has shown excellent performance in single-phase flows. We extend the FNO-based architecture to a highly complex CO₂-water multiphase problem with wide ranges of permeability and porosity heterogeneity, anisotropy, reservoir conditions, injection configurations, flow rates, and multiphase flow properties. The U-FNO architecture is more accurate in gas saturation and pressure buildup predictions than the original FNO and a state-of-the-art convolutional neural network (CNN) benchmark. Meanwhile, it has superior data utilization efficiency, requiring only a third of the training data to achieve the equivalent accuracy as CNN. U-FNO provides superior performance in highly heterogeneous geological formations and critically important applications such as gas saturation and pressure buildup "fronts" determination. The trained model can serve as a general-purpose alternative to routine numerical simulations of 2D-radial CO₂ injection problems with significant speed-ups than traditional simulators.
Additional Information
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) G. Wen and S. M. Benson gratefully acknowledges the supported by ExxonMobil through the Strategic Energy Alliance at Stanford University and the Stanford Center for Carbon Storage. Z. Li gratefully acknowledges the financial support from the Kortschak Scholars Program. A. Anandkumar is supported in part by Bren endowed chair, LwLL grants, Beyond Limits, Raytheon, Microsoft, Google, Adobe faculty fellowships, and DE Logi grant. The authors would like to acknowledge the reviewers and editors for the constructive comments. Code and data availability. The python code for U-FNO model architecture and the data set used in training is available at https://github.com/gegewen/ufno. Web application https://ccsnet.ai hosts the trained U-FNO models to provide real time predictions.Attached Files
Submitted - 2109.03697.pdf
Files
Name | Size | Download all |
---|---|---|
md5:9296126302486655496a69d21f2fb49b
|
5.7 MB | Preview Download |
Additional details
- Eprint ID
- 115614
- Resolver ID
- CaltechAUTHORS:20220714-224722475
- ExxonMobil
- Stanford University
- Kortschak Scholars Program
- Bren Professor of Computing and Mathematical Sciences
- Learning with Less Labels (LwLL)
- Beyond Limits
- Raytheon Company
- Microsoft Faculty Fellowship
- Google Faculty Research Award
- Adobe
- Caltech De Logi Fund
- Created
-
2022-07-15Created from EPrint's datestamp field
- Updated
-
2023-06-02Created from EPrint's last_modified field
- Caltech groups
- Center for Autonomous Systems and Technologies (CAST)