Knowledge-Based Convolutional Neural Network for the Simulation and Prediction of Two-Phase Darcy Flows
2404.03240
0
0
Abstract
Physics-informed neural networks (PINNs) have gained significant prominence as a powerful tool in the field of scientific computing and simulations. Their ability to seamlessly integrate physical principles into deep learning architectures has revolutionized the approaches to solving complex problems in physics and engineering. However, a persistent challenge faced by mainstream PINNs lies in their handling of discontinuous input data, leading to inaccuracies in predictions. This study addresses these challenges by incorporating the discretized forms of the governing equations into the PINN framework. We propose to combine the power of neural networks with the dynamics imposed by the discretized differential equations. By discretizing the governing equations, the PINN learns to account for the discontinuities and accurately capture the underlying relationships between inputs and outputs, improving the accuracy compared to traditional interpolation techniques. Moreover, by leveraging the power of neural networks, the computational cost associated with numerical simulations is substantially reduced. We evaluate our model on a large-scale dataset for the prediction of pressure and saturation fields demonstrating high accuracies compared to non-physically aware models.
Get summaries of the top AI research delivered straight to your inbox:
Overview
- The paper presents a "knowledge-based convolutional neural network" (KBCNN) model for simulating and predicting two-phase Darcy flows in porous media.
- The KBCNN approach aims to incorporate prior domain knowledge about the physical processes governing two-phase flow behavior into the neural network architecture and training.
- Key applications include modeling subsurface flow in oil/gas reservoirs, aquifers, and other porous media systems.
Plain English Explanation
The paper describes a new type of artificial intelligence (AI) model called a "knowledge-based convolutional neural network" (KBCNN) that can be used to simulate and predict the flow of two different fluids, like oil and water, through porous materials like rock or soil. This is an important problem in fields like oil and gas, hydrology, and environmental science.
Typical AI models trained on data alone can struggle to accurately capture the underlying physics governing these complex multiphase flow processes. The KBCNN approach aims to address this by incorporating relevant domain knowledge about the physical laws and principles that govern two-phase flow behavior directly into the neural network's architecture and training process.
The key idea is to design the neural network in a way that "knows" about things like fluid properties, rock permeability, and the equations describing flow through porous media. This allows the model to better learn the patterns in the data and make more accurate predictions, compared to a standard "black box" neural network.
The researchers demonstrate the KBCNN model on several benchmark problems in two-phase Darcy flow, showing improved performance over traditional simulation approaches. This suggests the potential of leveraging domain knowledge to enhance the capabilities of AI systems for complex physical modeling tasks.
Technical Explanation
The paper introduces a "knowledge-based convolutional neural network" (KBCNN) architecture for simulating and predicting two-phase Darcy flows in porous media. The core innovation is the integration of relevant domain knowledge about the underlying physical processes into the neural network design and training.
Specifically, the KBCNN incorporates known mathematical relationships governing multiphase flow, such as Darcy's law and capillary pressure-saturation functions, directly into the neural network layers and loss function. This allows the model to better capture the complex nonlinear interactions between fluid properties, rock characteristics, and flow dynamics.
The proposed KBCNN model consists of convolutional layers that operate on discretized representations of the porous media domain, combined with fully-connected layers that encode the physics-based constraints. The network is trained end-to-end on simulation data generated from traditional numerical solvers.
Experiments on canonical two-phase flow benchmarks demonstrate the KBCNN's ability to outperform standard data-driven neural network approaches in terms of predictive accuracy, particularly for challenging high-contrast permeability fields. The results suggest the benefits of integrating domain knowledge into neural network architectures for complex physical modeling problems.
Critical Analysis
The paper presents a compelling approach to enhancing the capabilities of neural networks for simulating multiphase flows in porous media. By explicitly incorporating relevant physical principles into the model architecture and training, the KBCNN framework aims to address some of the key limitations of purely data-driven deep learning models in this domain.
That said, the authors acknowledge several limitations and caveats to their work. First, the current KBCNN implementation is limited to 2D flow scenarios, and extending it to 3D would require significant additional research. Additionally, the models were only tested on relatively simple benchmark cases, and their performance on more realistic, complex geological models remains to be seen.
Another potential concern is the reliance on having accurate constitutive relationships (e.g., relative permeability functions) as model inputs. In practice, these can be difficult to determine precisely, and errors or uncertainties in specifying these inputs could impact the KBCNN's predictive capabilities.
Further research is also needed to fully understand the tradeoffs between the benefits of incorporating domain knowledge and the added complexity and potential for overfitting that this brings. Careful regularization and model selection strategies may be required to strike the right balance.
Overall, the KBCNN represents an interesting and promising direction for enhancing physics-informed machine learning models. However, additional validation and exploration of the approach's limitations and failure modes will be important to assess its true potential for real-world applications in reservoir engineering and other porous media flow problems.
Conclusion
The paper introduces a novel "knowledge-based convolutional neural network" (KBCNN) architecture that aims to leverage domain knowledge about two-phase Darcy flow processes to improve the accuracy and robustness of neural network-based simulations and predictions in porous media. By directly incorporating relevant physical principles into the model design, the KBCNN framework shows promising results on benchmark problems compared to standard data-driven approaches.
This research highlights the potential benefits of integrating physical constraints and mathematical relationships into neural network architectures, rather than treating them as pure "black box" function approximators. If successful, such physics-informed machine learning models could lead to significant advances in the modeling and simulation of complex multiphase flow phenomena in a wide range of geophysical and engineering applications.
However, the current KBCNN implementation also has several limitations that require further study, including extension to 3D domains, handling of uncertain constitutive relationships, and strategies for balancing the benefits of domain knowledge integration with the risks of increased model complexity. Continued research in this direction could yield important insights into the synergies between physics-based and data-driven modeling approaches for simulating the behavior of complex physical systems.
This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!
Related Papers
Graph Convolutional Networks for Simulating Multi-phase Flow and Transport in Porous Media
Jiamin Jiang, Bo Guo
0
0
Numerical simulation of multi-phase fluid dynamics in porous media is critical for many energy and environmental applications in Earth's subsurface. Data-driven surrogate modeling provides computationally inexpensive alternatives to high-fidelity numerical simulators. While the commonly used convolutional neural networks (CNNs) are powerful in approximating partial differential equation solutions, it remains challenging for CNNs to handle irregular and unstructured simulation meshes. However, simulation models for Earth's subsurface often involve unstructured meshes with complex mesh geometries, which limits the application of CNNs. To address this challenge, we construct surrogate models based on Graph Convolutional Networks (GCNs) to approximate the spatial-temporal solutions of multi-phase flow and transport processes in porous media. We propose a new GCN architecture suited to the hyperbolic character of the coupled PDE system, to better capture transport dynamics. Results of 2D heterogeneous test cases show that our surrogates predict the evolutions of pressure and saturation states with high accuracy, and the predicted rollouts remain stable for multiple timesteps. Moreover, the GCN-based models generalize well to irregular domain geometries and unstructured meshes that are unseen in the training dataset.
4/16/2024
🧠
New!Data-Driven Physics-Informed Neural Networks: A Digital Twin Perspective
Sunwoong Yang, Hojin Kim, Yoonpyo Hong, Kwanjung Yee, Romit Maulik, Namwoo Kang
0
0
This study explores the potential of physics-informed neural networks (PINNs) for the realization of digital twins (DT) from various perspectives. First, various adaptive sampling approaches for collocation points are investigated to verify their effectiveness in the mesh-free framework of PINNs, which allows automated construction of virtual representation without manual mesh generation. Then, the overall performance of the data-driven PINNs (DD-PINNs) framework is examined, which can utilize the acquired datasets in DT scenarios. Its scalability to more general physics is validated within parametric Navier-Stokes equations, where PINNs do not need to be retrained as the Reynolds number varies. In addition, since datasets can be often collected from different fidelity/sparsity in practice, multi-fidelity DD-PINNs are also proposed and evaluated. They show remarkable prediction performance even in the extrapolation tasks, with $42sim62%$ improvement over the single-fidelity approach. Finally, the uncertainty quantification performance of multi-fidelity DD-PINNs is investigated by the ensemble method to verify their potential in DT, where an accurate measure of predictive uncertainty is critical. The DD-PINN frameworks explored in this study are found to be more suitable for DT scenarios than traditional PINNs from the above perspectives, bringing engineers one step closer to seamless DT realization.
5/20/2024
🧠
Bayesian Reasoning for Physics Informed Neural Networks
Krzysztof M. Graczyk, Kornel Witkowski
0
0
We present the application of the physics-informed neural network (PINN) approach in Bayesian formulation. We have adopted the Bayesian neural network framework to obtain posterior densities from Laplace approximation. For each model or fit, the evidence is computed, which is a measure that classifies the hypothesis. The optimal solution is the one with the highest value of evidence. We have proposed a modification of the Bayesian algorithm to obtain hyperparameters of the model. We have shown that within the Bayesian framework, one can obtain the relative weights between the boundary and equation contributions to the total loss. Presented method leads to predictions comparable to those obtained by sampling from the posterior distribution within the Hybrid Monte Carlo algorithm (HMC). We have solved heat, wave, and Burger's equations, and the results obtained are in agreement with the exact solutions, demonstrating the effectiveness of our approach. In Burger's equation problem, we have demonstrated that the framework can combine information from differential equations and potential measurements. All solutions are provided with uncertainties (induced by the model's parameter dependence) computed within the Bayesian framework.
4/30/2024
🧠
Conformalized Physics-Informed Neural Networks
Lena Podina, Mahdi Torabi Rad, Mohammad Kohandel
0
0
Physics-informed neural networks (PINNs) are an influential method of solving differential equations and estimating their parameters given data. However, since they make use of neural networks, they provide only a point estimate of differential equation parameters, as well as the solution at any given point, without any measure of uncertainty. Ensemble and Bayesian methods have been previously applied to quantify the uncertainty of PINNs, but these methods may require making strong assumptions on the data-generating process, and can be computationally expensive. Here, we introduce Conformalized PINNs (C-PINNs) that, without making any additional assumptions, utilize the framework of conformal prediction to quantify the uncertainty of PINNs by providing intervals that have finite-sample, distribution-free statistical validity.
5/15/2024