Xiao-Hui Wu
Xiao-Hui Wu has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Alvaro L. G. A. Coutinho
Department of Civil Engineering, COPPE/Federal University of Rio de Janeiro, Rio de Janeiro, Brazil
1 shared publication - Amanda Camacho Novaes de Oliveira
Universidade Federal do Rio de Janeiro (UFRJ)
1 shared publication - Chen-Song Zhang
Academy of Mathematics and Systems Science, Beijing, China
1 shared publication - Dakshina Murthy Valiveti
ExonMobil Technology and Engineering Company Spring TX USA
1 shared publication - Ezequiel S. Santos
Department of Civil Engineering COPPE University of Rio de Janeiro Rio de Janeiro RJ
1 shared publication - Fernando Alves Rochinha
Federal University of Rio de Janeiro, Brazil
1 shared publication
Latest Publications
- Open Access
Hybrid DeepONet Surrogates for Multiphase Flow in Porous Media
The solution of partial differential equations (PDEs) plays a central role in numerous applications in science and engineering, particularly those involving multiphase flow in porous media. Complex, nonlinear systems govern these problems and are notoriously computationally intensive, especially in real-world applications and reservoirs. Recent advances in deep learning have spurred the development of data-driven surrogate models that approximate PDE solutions with reduced computational cost. Among these, Neural Operators such as Fourier Neural Operator (FNO) and Deep Operator Networks (DeepONet) have shown strong potential for learning parameter-to-solution mappings, enabling the generalization across families of PDEs. However, both methods face challenges when applied independently to complex porous media flows, including high memory requirements and difficulty handling the time dimension. To address these limitations, this work introduces hybrid neural operator surrogates based on DeepONet models that integrate Fourier Neural Operators, Multi-Layer Perceptrons (MLPs), and Kolmogorov-Arnold Networks (KANs) within their branch and trunk networks. The proposed framework decouples spatial and temporal learning tasks by splitting these structures into the branch and trunk networks, respectively. We evaluate these hybrid models on multiphase flow in porous media problems ranging in complexity from the steady 2D Darcy flow to the 2D and 3D problems belonging to the $10$th Comparative Solution Project from the Society of Petroleum Engineers. Results demonstrate that hybrid schemes achieve accurate surrogate modeling with significantly fewer parameters while maintaining strong predictive performance on large-scale reservoir simulations.
Open AccessHybrid DeepONet Surrogates for Multiphase Flow in Porous Media
The solution of partial differential equations (PDEs) plays a central role in numerous applications in science and engineering, particularly those involving multiphase flow in porous media. Complex, nonlinear systems govern these problems and are notoriously computationally intensive, especially in real-world applications and reservoirs. Recent advances in deep learning have spurred the development of data-driven surrogate models that approximate PDE solutions with reduced computational cost. Among these, Neural Operators such as Fourier Neural Operator (FNO) and Deep Operator Networks (DeepONet) have shown strong potential for learning parameter-to-solution mappings, enabling the generalization across families of PDEs. However, both methods face challenges when applied independently to complex porous media flows, including high memory requirements and difficulty handling the time dimension. To address these limitations, this work introduces hybrid neural operator surrogates based on DeepONet models that integrate Fourier Neural Operators, Multi-Layer Perceptrons (MLPs), and Kolmogorov-Arnold Networks (KANs) within their branch and trunk networks. The proposed framework decouples spatial and temporal learning tasks by splitting these structures into the branch and trunk networks, respectively. We evaluate these hybrid models on multiphase flow in porous media problems ranging in complexity from the steady 2D Darcy flow to the 2D and 3D problems belonging to the $10$th Comparative Solution Project from the Society of Petroleum Engineers. Results demonstrate that hybrid schemes achieve accurate surrogate modeling with significantly fewer parameters while maintaining strong predictive performance on large-scale reservoir simulations.
- 15 January 2014
- Published by Springer Nature in Computational Geosciences
- Vol. 18 (3-4) , 357-372
Grid adaptation for the Dirichlet–Neumann representation method and the multiscale mixed finite-element method
- 15 January 2014
- Published by Springer Nature in Computational Geosciences
- Vol. 18 (3-4) , 357-372
Grid adaptation for the Dirichlet–Neumann representation method and the multiscale mixed finite-element method
- Open Access
Combined Preconditioning with Applications in Reservoir Simulation
We develop a simple algorithmic framework to solve large-scale symmetric positive definite linear systems. At its core, the framework relies on two components: (1) a norm-convergent iterative method (i.e. smoother) and (2) a preconditioner. The resulting preconditioner, which we refer to as a combined preconditioner, is much more robust and efficient than the iterative method and preconditioner when used in Krylov subspace methods. We prove that the combined preconditioner is positive definite and show estimates on the condition number of the preconditioned system. We combine an algebraic multigrid method and an incomplete factorization preconditioner to test the proposed framework on problems in petroleum reservoir simulation. Our numerical experiments demonstrate noticeable speed-up when we compare our combined method with the standalone algebraic multigrid method or the incomplete factorization preconditioner.
Open AccessCombined Preconditioning with Applications in Reservoir Simulation
We develop a simple algorithmic framework to solve large-scale symmetric positive definite linear systems. At its core, the framework relies on two components: (1) a norm-convergent iterative method (i.e. smoother) and (2) a preconditioner. The resulting preconditioner, which we refer to as a combined preconditioner, is much more robust and efficient than the iterative method and preconditioner when used in Krylov subspace methods. We prove that the combined preconditioner is positive definite and show estimates on the condition number of the preconditioned system. We combine an algebraic multigrid method and an incomplete factorization preconditioner to test the proposed framework on problems in petroleum reservoir simulation. Our numerical experiments demonstrate noticeable speed-up when we compare our combined method with the standalone algebraic multigrid method or the incomplete factorization preconditioner.
- Open Access
- 1 January 2012
- Published by Begell House Inc. Publishers in International Journal for Uncertainty Quantification
- Vol. 2 (3) , 259-278
EFFECTIVE PARAMETRIZATION FOR RELIABLE RESERVOIR PERFORMANCE PREDICTIONS
The purpose of reservoir modeling and simulation is to predict reservoir performance for development and depletion planning. Despite decades of research, efficient and reliable reservoir performance predictions are still a challenge in practice. In this paper, we present an overview of reservoir modeling as it is commonly practiced today and the challenges it faces. More specifically, we focus on the challenges posed by the large amount of uncertainty inherent in the characterization of reservoirs that are heterogeneous at multiple scales. We discuss the practical implications of these challenges and recent developments toward addressing them. In particular, we examine the need for effective parametrization of geologic concepts and related recent advances in parametrization and parameter reduction techniques, including their advantages and limitations. Using numerical examples from two different depositional environments, we show that effective parametrization can be achieved by taking advantage of the geologic hierarchy underlying most geologic concepts and a general understanding of the impact of geologic features on fluid flow. Finally, we propose an approach to systematically derive fit-for-purpose parametrization for practical reservoir modeling problems.
- 1 January 2012
- Published by Begell House Inc. Publishers in International Journal for Uncertainty Quantification
- Vol. 2 (3) , 259-278
Open AccessEFFECTIVE PARAMETRIZATION FOR RELIABLE RESERVOIR PERFORMANCE PREDICTIONS
The purpose of reservoir modeling and simulation is to predict reservoir performance for development and depletion planning. Despite decades of research, efficient and reliable reservoir performance predictions are still a challenge in practice. In this paper, we present an overview of reservoir modeling as it is commonly practiced today and the challenges it faces. More specifically, we focus on the challenges posed by the large amount of uncertainty inherent in the characterization of reservoirs that are heterogeneous at multiple scales. We discuss the practical implications of these challenges and recent developments toward addressing them. In particular, we examine the need for effective parametrization of geologic concepts and related recent advances in parametrization and parameter reduction techniques, including their advantages and limitations. Using numerical examples from two different depositional environments, we show that effective parametrization can be achieved by taking advantage of the geologic hierarchy underlying most geologic concepts and a general understanding of the impact of geologic features on fluid flow. Finally, we propose an approach to systematically derive fit-for-purpose parametrization for practical reservoir modeling problems.
- Open Access
- 1 June 2008
- Published by SAGE Publications in Journal of Algorithms & Computational Technology
- Vol. 2 (2) , 223-248
Global Scale-up on Reservoir Models with Piecewise Constant Permeability Field
Global scale-up was first proposed in the 1980s, and its benefits are well-described in the literature. However, global scale-up has not been widely applied in practice due to significant technical challenges. In this paper, we analyze some of the theoretical and numerical difficulties and present practical resolutions. First, we derive the flux and energy formulations for scale-up by using a Multiscale Finite Element framework. These formulations can be applied to local, extended local and global scale-up. Then we present a new method to improve scale-up accuracy for geo-cellular models with piecewise constant permeability. On these models, large jumps in the permeability field lead to well-known singularities in the flow solutions, making them difficult to be calculated accurately. We show that inaccurate flow solutions can lead to large scale-up errors. To mitigate the effect of the singularities and improve the scale-up accuracy, we have developed a hybrid method for solving flows which utilizes the fact that the Continuous Galerkin (CG) finite elment method and the Mixed finite element method provide upper and lower bounds, respectively, for the effective permeability. The new method uses CG to compute the pressure solution followed by a weighted L2 projection of the numerical fluxes to enforce local mass conservation. Numerical examples are presented to demonstrate the effectiveness of the hybrid method.
- 1 June 2008
- Published by SAGE Publications in Journal of Algorithms & Computational Technology
- Vol. 2 (2) , 223-248
Open AccessGlobal Scale-up on Reservoir Models with Piecewise Constant Permeability Field
Global scale-up was first proposed in the 1980s, and its benefits are well-described in the literature. However, global scale-up has not been widely applied in practice due to significant technical challenges. In this paper, we analyze some of the theoretical and numerical difficulties and present practical resolutions. First, we derive the flux and energy formulations for scale-up by using a Multiscale Finite Element framework. These formulations can be applied to local, extended local and global scale-up. Then we present a new method to improve scale-up accuracy for geo-cellular models with piecewise constant permeability. On these models, large jumps in the permeability field lead to well-known singularities in the flow solutions, making them difficult to be calculated accurately. We show that inaccurate flow solutions can lead to large scale-up errors. To mitigate the effect of the singularities and improve the scale-up accuracy, we have developed a hybrid method for solving flows which utilizes the fact that the Continuous Galerkin (CG) finite elment method and the Mixed finite element method provide upper and lower bounds, respectively, for the effective permeability. The new method uses CG to compute the pressure solution followed by a weighted L2 projection of the numerical fluxes to enforce local mass conservation. Numerical examples are presented to demonstrate the effectiveness of the hybrid method.
- 4 January 2008
- Published by Springer Nature in Computational Geosciences
- Vol. 12 (1) , 29-45
Upscaled modeling of well singularity for simulating flow in heterogeneous formations
- 4 January 2008
- Published by Springer Nature in Computational Geosciences
- Vol. 12 (1) , 29-45
Upscaled modeling of well singularity for simulating flow in heterogeneous formations