Wei Wei
Research Subjects
Research Interests
Wei Wei has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
Xiaomin Zhu
3 shared publicationsChen Zhu
2 shared publicationsMingwei He
2 shared publicationsMingwei Wang
2 shared publicationsMingxuan Tan
Institute of Marine Geology, College of Oceanography, Hohai University, Nanjing, 210098, China
2 shared publications- Pan Luo
Saudi Aramco, EXPEC Adv Res Ctr, Dhahran 31311, Saudi Arabia
2 shared publications
Latest Publications
- 16 September 2026
- Published by Elsevier in Gas Science and Engineering
- Vol. 156
Bathymetric control on sub-seabed CO2 storage in the hydrate stability zone: Implications for capacity, buoyancy, and long-term stability
- 16 September 2026
- Published by Elsevier in Gas Science and Engineering
- Vol. 156
Bathymetric control on sub-seabed CO2 storage in the hydrate stability zone: Implications for capacity, buoyancy, and long-term stability
- 1 September 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (35) , 355005
A quadratic neural network for multiphase flow prediction in CO 2 sequestration
Geological carbon sequestration (GCS) is widely recognized as a key strategy for mitigating global climate change by reducing atmospheric CO 2 concentrations. Deep learning methods, such as long short-term memory (LSTM) networks and multilayer perceptron (MLP) models, have been widely applied to multiphase flow prediction in CO 2 sequestration. However, most existing deep learning approaches struggle to capture complex nonlinear dynamics, making it challenging to achieve both high prediction accuracy and computational efficiency in large-scale scenarios. To address this limitation, a quadratic multi-layer perceptron (Quadratic-MLP) is proposed by introducing a quadratic polynomial input–output mapping at the neuron level, which enhances nonlinear representation capability while maintaining a compact network architecture. The proposed Quadratic-MLP is applied to multiphase flow prediction in CO 2 sequestration using 27 numerical simulation cases generated from three heterogeneous geological models, with a spatial resolution of 25 × 25 × 3 over 72 time steps. Results demonstrate that the Quadratic-MLP outperforms conventional LSTM and MLP models. Compared with the conventional MLP model, the proposed Quadratic-MLP reduces the mean squared error by more than 60% for all predicted variables, including pressure, saturation, and water production rate. Its prediction accuracy also surpasses that of deeper and wider MLP models with more trainable parameters, demonstrating improved parameter efficiency. Furthermore, the capability of the proposed Quadratic-MLP to capture temporal evolution trends beyond the training period highlights its potential for large-scale reservoir simulation and rapid prediction applications.
- 1 September 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (35) , 355005
A quadratic neural network for multiphase flow prediction in CO 2 sequestration
Geological carbon sequestration (GCS) is widely recognized as a key strategy for mitigating global climate change by reducing atmospheric CO 2 concentrations. Deep learning methods, such as long short-term memory (LSTM) networks and multilayer perceptron (MLP) models, have been widely applied to multiphase flow prediction in CO 2 sequestration. However, most existing deep learning approaches struggle to capture complex nonlinear dynamics, making it challenging to achieve both high prediction accuracy and computational efficiency in large-scale scenarios. To address this limitation, a quadratic multi-layer perceptron (Quadratic-MLP) is proposed by introducing a quadratic polynomial input–output mapping at the neuron level, which enhances nonlinear representation capability while maintaining a compact network architecture. The proposed Quadratic-MLP is applied to multiphase flow prediction in CO 2 sequestration using 27 numerical simulation cases generated from three heterogeneous geological models, with a spatial resolution of 25 × 25 × 3 over 72 time steps. Results demonstrate that the Quadratic-MLP outperforms conventional LSTM and MLP models. Compared with the conventional MLP model, the proposed Quadratic-MLP reduces the mean squared error by more than 60% for all predicted variables, including pressure, saturation, and water production rate. Its prediction accuracy also surpasses that of deeper and wider MLP models with more trainable parameters, demonstrating improved parameter efficiency. Furthermore, the capability of the proposed Quadratic-MLP to capture temporal evolution trends beyond the training period highlights its potential for large-scale reservoir simulation and rapid prediction applications.
- Open Access
- 21 July 2026
- Published by Elsevier in SSRN Electronic Journal
Physically-Constrained Hybrid Machine Learning Framework for Predicting CH4 and CO2 Solubility in Water and NaCl Brines
Predicting CH4 and CO2 solubility in water and brines is important for evaluating dissolution trapping, gas partitioning, and compositional evolution in geological carbon storage, and related gas-production and reservoir-engineering applications. This study proposes a physically-constrained hybrid machine learning (ML) framework that combines baseline modeling with physical trend constraints from thermodynamic modeling data and residual learning from experimental data to predict gas solubility in reservoir conditions with pressure (1-2000 bar), temperature (298.15-550.15 K), salinity (mNaCl 0-6 mol/kg), and gas-composition. A unified database containing 10,106 records, including 3,625 experimental and 6,481 thermodynamic-model generated data points, was compiled for pure CH4, pure CO2, and mixed CH4–CO2 aqueous systems. The results show that the baseline model preserves smooth and physically reasonable solubility trends across broad parameter ranges, while residual correction substantially improves agreement with experimental observations. For CH4–H2O–(NaCl), the hybrid MLP model improved [[EQUATION]] from 0.75 to 0.98, and reduced RMSE from 0.35 to 0.09. For CO2–H2O–(NaCl), [[EQUATION]] increased from 0.90 to 0.99, and RMSE decreased from 0.55 to 0.12. For mixed CH4–CO2 systems, the framework reproduces physically credible composition-dependent dissolution trends, although broader validation is still needed because of limited experimental data. Across different regressors, the hybrid strategy showed strong performance, with the MLP implementation achieving the best overall balance between predictive accuracy and physically reasonable trend behavior in this study. The proposed framework offers a computationally efficient and physically consistent thermodynamic surrogate for gas–water–brine systems, supporting rapid screening, process simulation, and uncertainty analysis in carbon storage and subsurface energy systems.
- 21 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessPhysically-Constrained Hybrid Machine Learning Framework for Predicting CH4 and CO2 Solubility in Water and NaCl Brines
Predicting CH4 and CO2 solubility in water and brines is important for evaluating dissolution trapping, gas partitioning, and compositional evolution in geological carbon storage, and related gas-production and reservoir-engineering applications. This study proposes a physically-constrained hybrid machine learning (ML) framework that combines baseline modeling with physical trend constraints from thermodynamic modeling data and residual learning from experimental data to predict gas solubility in reservoir conditions with pressure (1-2000 bar), temperature (298.15-550.15 K), salinity (mNaCl 0-6 mol/kg), and gas-composition. A unified database containing 10,106 records, including 3,625 experimental and 6,481 thermodynamic-model generated data points, was compiled for pure CH4, pure CO2, and mixed CH4–CO2 aqueous systems. The results show that the baseline model preserves smooth and physically reasonable solubility trends across broad parameter ranges, while residual correction substantially improves agreement with experimental observations. For CH4–H2O–(NaCl), the hybrid MLP model improved [[EQUATION]] from 0.75 to 0.98, and reduced RMSE from 0.35 to 0.09. For CO2–H2O–(NaCl), [[EQUATION]] increased from 0.90 to 0.99, and RMSE decreased from 0.55 to 0.12. For mixed CH4–CO2 systems, the framework reproduces physically credible composition-dependent dissolution trends, although broader validation is still needed because of limited experimental data. Across different regressors, the hybrid strategy showed strong performance, with the MLP implementation achieving the best overall balance between predictive accuracy and physically reasonable trend behavior in this study. The proposed framework offers a computationally efficient and physically consistent thermodynamic surrogate for gas–water–brine systems, supporting rapid screening, process simulation, and uncertainty analysis in carbon storage and subsurface energy systems.
- 18 February 2025
- Published by Elsevier in Geoenergy Science and Engineering
- Vol. 249
Evaluating and predicting deliverability of natural gas storage sites using stacking machine learning models
- 18 February 2025
- Published by Elsevier in Geoenergy Science and Engineering
- Vol. 249
Evaluating and predicting deliverability of natural gas storage sites using stacking machine learning models
- 29 May 2024
- Published by American Chemical Society (ACS) in Energy & Fuels
- Vol. 38 (12) , 11119-11136
Advanced Machine Learning Models for CO2 and H2S Solubility in Water and NaCl Brine: Implications for Geoenergy Extraction and Carbon Storage
- 29 May 2024
- Published by American Chemical Society (ACS) in Energy & Fuels
- Vol. 38 (12) , 11119-11136
Advanced Machine Learning Models for CO2 and H2S Solubility in Water and NaCl Brine: Implications for Geoenergy Extraction and Carbon Storage
- 1 August 2018
- Published by Elsevier in Journal of Natural Gas Science and Engineering
- Vol. 56, 593-607
Distribution of mantle-derived CO2 gas reservoir and its relationship with basement faults in Songliao Basin, China
- 1 August 2018
- Published by Elsevier in Journal of Natural Gas Science and Engineering
- Vol. 56, 593-607
Distribution of mantle-derived CO2 gas reservoir and its relationship with basement faults in Songliao Basin, China