Combined Deep Learning and Optimization for Hydrogen-Solubility Prediction in Aqueous Systems Appropriate for Underground Hydrogen Storage Reservoirs DOI
Promise O. Longe, Shadfar Davoodi, Mohammad Mehrad

et al.

Energy & Fuels, Journal Year: 2024, Volume and Issue: 38(22), P. 22031 - 22049

Published: Oct. 30, 2024

The widespread use of fossil fuels drives greenhouse gas emissions, prompting the need for cleaner energy alternatives like hydrogen. Underground hydrogen storage (UHS) is a promising solution, but measureing (H2) solubility in brine complex and costly. Machine learning can provide accurate reliable predictions H2 by analyzing diverse input variables, surpassing traditional methods. This advancement crucial improving UHS, making it viable component sustainable infrastructure. Given its importance, this study utilized convolutional neural network (CNN) long–short-term memory (LSTM) deep algorithms combination with growth optimization (GO) gray wolf (GWO) to predict solubility. A total 1078 data points were collected from laboratory results, including variables temperature (T), pressure (P), salinity (S), salt type (ST). After removing 97 points, which identified as outliers duplicates, remaining 981 divided into training testing sets using best separation ratio selected based on sensitivity analysis. Standalone hybrid forms then applied develop predictive models optimized control parameters both algorithms. Among developed models, CNN-GO has lowest root-mean-square error (RMSE, train: 0.00006 mole fraction test: 0.00021 fraction) compared other standalone models. application scoring regression characteristic (REC) curve analysis showed that model generated prediction performance. Shapley additive explanation indicated P was most important factor influencing solubility, followed S, T, ST, order. Partial dependency revealed ability capture nonlinear relationships between features target variable.

Language: Английский

Technical challenges and opportunities of hydrogen storage: A comprehensive review on different types of underground storage DOI
Guangyao Leng, Wei Yan, Zhangxin Chen

et al.

Journal of Energy Storage, Journal Year: 2025, Volume and Issue: 114, P. 115900 - 115900

Published: Feb. 21, 2025

Language: Английский

Citations

1
Enhanced Data-Driven Machine Learning Models for Predicting Total Organic Carbon in Marine–Continental Transitional Shale Reservoirs DOI Open Access
Sizhong Peng, Congjun Feng, Zhen Qiu

et al.

Sustainability, Journal Year: 2025, Volume and Issue: 17(5), P. 2048 - 2048

Published: Feb. 27, 2025

Natural gas, as a sustainable and cleaner energy source, still holds crucial position in the transition stage. In shale gas exploration, total organic carbon (TOC) content plays role, with log data proving beneficial predicting reservoirs. However, complex coal-bearing layers like marine–continental transitional Shanxi Formation, traditional prediction methods exhibit significant errors. Therefore, this study proposes an advanced, cost- time-saving deep learning approach to predict TOC shale. Five well records from area were used evaluate five machine models: K-Nearest Neighbors (KNNs), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), Extreme (XGB), Deep Neural Network (DNN). The predictive results compared conventional for accurate predictions. Through K-fold cross-validation, ML models showed superior accuracy over models, DNN model displaying lowest root mean square error (RMSE) absolute (MAE). To enhance accuracy, δR was integrated new parameter into models. Comparative analysis revealed that improved DNN-R reduced MAE RMSE by 57.1% 70.6%, respectively, on training set, 59.5% 72.5%, test original model. Williams plot permutation importance confirmed reliability effectiveness of enhanced indicate potential technology valuable tool parameters, especially reservoirs lacking sufficient core samples relying solely basic well-logging data, signifying its effective assessment development.

Language: Английский

Citations

1
Computational intelligence investigations on evaluation of salicylic acid solubility in various solvents at different temperatures DOI Creative Commons
Adel Alhowyan, Wael A. Mahdi, Ahmad J. Obaidullah

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Feb. 28, 2025

This research shows the utilization of various tree-based machine learning algorithms with a specific focus on predicting Salicylic acid solubility values in 13 solvents. We employed four distinct models: cubist regression, gradient boosting (GB), extreme (XGB), and extra trees (ET) for correlation drug to pressure, temperature, solvent composition. The dataset was preprocessed using Standard Scaler standardize it, ensuring each feature has mean zero standard deviation one, followed by outlier detection Cook's distance. Hyperparameter optimization made Differential Evolution (DE) method improved performance models. Monte Carlo Cross-Valuation used evaluation Measures including R2 score, Root Mean Squared Error (RMSE), Absolute (MAE) helped measure their performance. With an value 0.996, Extra Trees model displayed remarkable accuracy consistency, so showing better than other study emphasizes resilience ensemble methods capturing intricate data patterns effectiveness regression tasks application pharmaceutical manufacturing.

Language: Английский

Citations

1
Modeling the thermal transport properties of hydrogen and its mixtures with greenhouse gas impurities: A data-driven machine learning approach DOI
Hung Vo Thanh, Mohammad Rahimi, Suparit Tangparitkul

et al.

International Journal of Hydrogen Energy, Journal Year: 2024, Volume and Issue: 83, P. 1 - 12

Published: Aug. 8, 2024

Language: Английский

Citations

8
Advanced machine learning-based modeling of interfacial tension in the crude oil-brine-diethyl ether system: Insights into the effects of temperature and salinity DOI
Amir Mohammadi Khanghah, Mahsa Parhizgar Keradeh, Alireza Keshavarz

et al.

Journal of Molecular Liquids, Journal Year: 2024, Volume and Issue: 404, P. 124861 - 124861

Published: April 29, 2024

Language: Английский

Citations

7
Predicting CO2 and H2 Solubility in Pure Water and Various Aqueous Systems: Implication for CO2–EOR, Carbon Capture and Sequestration, Natural Hydrogen Production and Underground Hydrogen Storage DOI Creative Commons
Promise O. Longe, David Kwaku Danso,

Gideon Gyamfi

et al.

Energies, Journal Year: 2024, Volume and Issue: 17(22), P. 5723 - 5723

Published: Nov. 15, 2024

The growing energy demand and the need for climate mitigation strategies have spurred interest in application of CO2–enhanced oil recovery (CO2–EOR) carbon capture, utilization, storage (CCUS). Furthermore, natural hydrogen (H2) production underground (UHS) geological media emerged as promising technologies cleaner achieving net–zero emissions. However, selecting a suitable medium is complex, it depends on physicochemical petrophysical characteristics host rock. Solubility key factor affecting above–mentioned processes, critical to understand phase distribution estimating trapping capacities. This paper conducts succinct review predictive techniques present novel simple non–iterative models swift reliable prediction solubility behaviors CO2–brine H2–brine systems under varying conditions pressure, temperature, salinity (T–P–m salts), which are crucial many energy–related applications. proposed predict CO2 + H2O brine containing mixed salts various single salt (Na+, K+, Ca2+, Mg2+, Cl−, SO42−) typical (273.15–523.15 K, 0–71 MPa), well H2 NaCl (273.15–630 0–101 MPa). validated against experimental data, with average absolute errors pure water ranging between 8.19 8.80% 4.03 9.91%, respectively. These results demonstrate that can accurately over wide range while remaining computationally efficient compared traditional models. Importantly, reproduce abrupt variations composition during transitions account influence different ions solubility. capture salting–out (SO) gas types consistent previous studies. simplified presented this study offer significant advantages conventional approaches, including computational efficiency accuracy across conditions. explicit, derivative–continuous nature these eliminates iterative algorithms, making them integration into large–scale multiphase flow simulations. work contributes field by offering tools modeling subsurface environmental–related applications, facilitating their transition aimed at reducing

Language: Английский

Citations

7
Estimation the pH of CO2-saturated NaCl solutions using gene expression programming: Implications for CO2 sequestration DOI

Mohammad Rasool Dehghani,

Parmida Seraj Ebrahimi,

Moein Kafi

et al.

Results in Engineering, Journal Year: 2025, Volume and Issue: 25, P. 104047 - 104047

Published: Jan. 25, 2025

Language: Английский

Citations

0
Artificial intelligence in geoenergy: bridging petroleum engineering and future-oriented applications DOI Creative Commons
Sungil Kim, Tea-Woo Kim, Suryeom Jo

et al.

Journal of Petroleum Exploration and Production Technology, Journal Year: 2025, Volume and Issue: 15(2)

Published: Feb. 1, 2025

Language: Английский

Citations

0
Data-Driven Machine Learning Models for Predicting Deliverability of Underground Natural Gas Storage in Aquifer and Depleted Reservoirs DOI

Altaf Hussain,

Peng‐Zhi Pan, Javid Hussain

et al.

Energy, Journal Year: 2025, Volume and Issue: unknown, P. 134974 - 134974

Published: Feb. 1, 2025

Language: Английский

Citations

0
Data-driven strategy for contact angle prediction in underground hydrogen storage using machine learning DOI Creative Commons

Mehdi Nassabeh,

Zhenjiang You, Alireza Keshavarz

et al.

Journal of Energy Storage, Journal Year: 2025, Volume and Issue: 114, P. 115908 - 115908

Published: Feb. 22, 2025

Language: Английский

Citations

0