Learning Pore-scale Multiphase Flow from 4D Velocimetry
arXiv:2603.12516v1 Announce Type: new Abstract: Multiphase flow in porous media underpins subsurface energy and environmental technologies, including geological CO$_2$ storage and underground hydrogen storage, yet pore-scale dynamics in realistic three-dimensional materials remain difficult to characterize and predict. Here we introduce a multimodal learning framework that infers multiphase pore-scale flow directly from time-resolved four-dimensional (4D) micro-velocimetry measurements. The model couples a graph network simulator for Lagrangian tracer-particle motion with a 3D U-Net for voxelized interface evolution. The imaged pore geometry serves as a boundary constraint to the flow velocity and the multiphase interface predictions, which are coupled and updated iteratively at each time step. Trained autoregressively on experimental sequences in capillary-dominated conditions ($Ca\approx10^{-6}$), the learned surrogate captures transient, nonlocal flow perturbations and abrupt inter
arXiv:2603.12516v1 Announce Type: new Abstract: Multiphase flow in porous media underpins subsurface energy and environmental technologies, including geological CO$_2$ storage and underground hydrogen storage, yet pore-scale dynamics in realistic three-dimensional materials remain difficult to characterize and predict. Here we introduce a multimodal learning framework that infers multiphase pore-scale flow directly from time-resolved four-dimensional (4D) micro-velocimetry measurements. The model couples a graph network simulator for Lagrangian tracer-particle motion with a 3D U-Net for voxelized interface evolution. The imaged pore geometry serves as a boundary constraint to the flow velocity and the multiphase interface predictions, which are coupled and updated iteratively at each time step. Trained autoregressively on experimental sequences in capillary-dominated conditions ($Ca\approx10^{-6}$), the learned surrogate captures transient, nonlocal flow perturbations and abrupt interface rearrangements (Haines jumps) over rollouts spanning seconds of physical time, while reducing hour-to-day--scale direct numerical simulations to seconds of inference. By providing rapid, experimentally informed predictions, the framework opens a route to ''digital experiments'' to replicate pore-scale physics observed in multiphase flow experiments, offering an efficient tool for exploring injection conditions and pore-geometry effects relevant to subsurface carbon and hydrogen storage.
Executive Summary
This article proposes a novel multimodal learning framework to infer pore-scale multiphase flow in realistic three-dimensional materials from time-resolved 4D micro-velocimetry measurements. The framework combines a graph network simulator and a 3D U-Net to capture flow velocity and multiphase interface predictions. The model is trained on experimental sequences in capillary-dominated conditions, reducing simulation time to seconds. This framework has the potential to replicate pore-scale physics and provide efficient predictions for subsurface energy and environmental technologies. The authors' approach may open new avenues for exploring injection conditions and pore-geometry effects relevant to subsurface carbon and hydrogen storage.
Key Points
- ▸ The proposed framework infers pore-scale multiphase flow from time-resolved 4D micro-velocimetry measurements.
- ▸ The framework combines a graph network simulator and a 3D U-Net for flow velocity and multiphase interface predictions.
- ▸ The model is trained on experimental sequences in capillary-dominated conditions, reducing simulation time to seconds.
Merits
Strength in reproducing pore-scale physics
The framework's ability to replicate pore-scale physics observed in multiphase flow experiments demonstrates its potential for accurately modeling complex flow dynamics.
Efficient predictions and simulations
The framework's ability to reduce simulation time from hours to seconds enables rapid exploration of injection conditions and pore-geometry effects.
Demerits
Limitation in generalizability
The framework's performance is demonstrated in capillary-dominated conditions, and its generalizability to other regimes remains uncertain.
Dependence on high-quality experimental data
The framework's accuracy relies on the quality of the experimental data used for training, which may be challenging to obtain in all cases.
Expert Commentary
The proposed framework demonstrates a significant advancement in the field of multiphase flow modeling, offering a novel approach to inferring pore-scale dynamics from time-resolved 4D micro-velocimetry measurements. The framework's efficiency and accuracy make it a valuable tool for exploring injection conditions and pore-geometry effects relevant to subsurface energy and environmental technologies. However, further research is needed to assess the framework's generalizability and robustness in different regimes. Furthermore, the framework's dependence on high-quality experimental data highlights the importance of investing in experimental research and data collection in this field.
Recommendations
- ✓ Future research should focus on assessing the framework's generalizability and robustness in different regimes and developing strategies to improve its performance in low-quality data scenarios.
- ✓ Investing in experimental research and data collection is essential to develop and train accurate machine learning models for complex flow dynamics.