A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems
From MaRDI portal
Publication:776737
DOI10.1016/J.JCP.2020.109456zbMATH Open1436.76058arXiv1908.05823OpenAlexW3016309349MaRDI QIDQ776737FDOQ776737
Authors: Meng Tang, L. J. Durlofsky, Yimin Liu 
Publication date: 13 July 2020
Published in: Journal of Computational Physics (Search for Journal in Brave)
Abstract: A deep-learning-based surrogate model is developed and applied for predicting dynamic subsurface flow in channelized geological models. The surrogate model is based on deep convolutional and recurrent neural network architectures, specifically a residual U-Net and a convolutional long short term memory recurrent network. Training samples entail global pressure and saturation maps, at a series of time steps, generated by simulating oil-water flow in many (1500 in our case) realizations of a 2D channelized system. After training, the `recurrent R-U-Net' surrogate model is shown to be capable of accurately predicting dynamic pressure and saturation maps and well rates (e.g., time-varying oil and water rates at production wells) for new geological realizations. Assessments demonstrating high surrogate-model accuracy are presented for an individual geological realization and for an ensemble of 500 test geomodels. The surrogate model is then used for the challenging problem of data assimilation (history matching) in a channelized system. For this study, posterior reservoir models are generated using the randomized maximum likelihood method, with the permeability field represented using the recently developed CNN-PCA parameterization. The flow responses required during the data assimilation procedure are provided by the recurrent R-U-Net. The overall approach is shown to lead to substantial reduction in prediction uncertainty. High-fidelity numerical simulation results for the posterior geomodels (generated by the surrogate-based data assimilation procedure) are shown to be in essential agreement with the recurrent R-U-Net predictions. The accuracy and dramatic speedup provided by the surrogate model suggest that it may eventually enable the application of more formal posterior sampling methods in realistic problems.
Full work available at URL: https://arxiv.org/abs/1908.05823
Recommendations
- Deep-learning-based surrogate flow modeling and geological parameterization for data assimilation in 3D subsurface flow
- Efficient uncertainty quantification for dynamic subsurface flow with surrogate by theory-guided neural network
- Deep global model reduction learning in porous media flow simulation
- A data-driven surrogate to image-based flow simulations in porous media
- Convolutional neural networks (CNN) for feature-based model calibration under uncertain geologic scenarios
Cites Work
- Title not available (Why is that?)
- Mesh Adaptive Direct Search Algorithms for Constrained Optimization
- Development and application of reduced-order modeling procedures for subsurface flow simulation
- Modeling fluid flow in oil reservoirs
- Reduced-order optimal control of water flooding using proper orthogonal decomposition
- Data assimilation and uncertainty assessment for complex geological models using a new PCA-based parameterization
- A deep-learning-based geological parameterization for history matching complex models
- Gaussian processes for history-matching: application to an unconventional gas reservoir
- Bayesian deep convolutional encoder-decoder networks for surrogate modeling and uncertainty quantification
- Non-intrusive subdomain POD-TPWL for reservoir history matching
Cited In (32)
- Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
- Efficient and data-driven prediction of water breakthrough in subsurface systems using deep long short-term memory machine learning
- Theory-guided auto-encoder for surrogate construction and inverse modeling
- Bayesian reduced-order deep learning surrogate model for dynamic systems described by partial differential equations
- Data-space inversion using a recurrent autoencoder for time-series parameterization
- Multi-domain encoder-decoder neural networks for latent data assimilation in dynamical systems
- Mesh-Conv: convolution operator with mesh resolution independence for flow field modeling
- Bayesian multiscale deep generative model for the solution of high-dimensional inverse problems
- Efficient uncertainty quantification for dynamic subsurface flow with surrogate by theory-guided neural network
- Deep-learning-based surrogate flow modeling and geological parameterization for data assimilation in 3D subsurface flow
- Use of multifidelity training data and transfer learning for efficient construction of subsurface flow surrogate models
- Stochastic pix2vid: a new spatiotemporal deep learning method for image-to-video synthesis in geologic \(\mathrm{CO_2}\) storage prediction
- Weak form theory-guided neural network (TgNN-wf) for deep learning of subsurface single- and two-phase flow
- Transformers for modeling physical systems
- Surrogate and inverse modeling for two-phase flow in porous media via theory-guided convolutional neural network
- A deep learning based reduced order modeling for stochastic underground flow problems
- A gradient-based deep neural network model for simulating multiphase flow in porous media
- Convolutional -- recurrent neural network proxy for robust optimization and closed-loop reservoir management
- Solute transport prediction in heterogeneous porous media using random walks and machine learning
- Surrogate optimization of deep neural networks for groundwater predictions
- Physics-informed machine learning method with space-time Karhunen-Loève expansions for forward and inverse partial differential equations
- Efficient deep data assimilation with sparse observations and time-varying sensors
- Blending machine learning and sequential data assimilation over latent spaces for surrogate modeling of Boussinesq systems
- Accelerated reactive transport simulations in heterogeneous porous media using Reaktoro and Firedrake
- A nonintrusive hybrid neural-physics modeling of incomplete dynamical systems: Lorenz equations
- Variational data assimilation and its decoupled iterative numerical algorithms for Stokes-Darcy model
- A data-driven surrogate to image-based flow simulations in porous media
- Deep reinforcement learning for optimal well control in subsurface systems with uncertain geology
- Gradient-boosted spatiotemporal neural network for simulating underground hydrogen storage in aquifers
- Physics-constrained deep learning forecasting: an application with capacitance resistive model
- Production well placement and history matching by hyperparametric optimization and machine learning
- An encoder-decoder deep surrogate for reverse time migration in seismic imaging under uncertainty
Uses Software
This page was built for publication: A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q776737)
