Robust Ensemble Learning Models for Predicting Hydrogen Sulfide Solubility in Brine

Energy & Fuels, Journal Year: 2024, Volume and Issue: 38(21), P. 21174 - 21188

Published: Oct. 25, 2024

Hydrogen sulfide (H2S) sequestration in geological formations can be one of the promising techniques for reducing greenhouse gas emissions. Accurate predictions phase behavior and H2S solubility aqueous solution phases are vital to provide better accuracy designing, well planning, process injection optimizations. In this study, a vast number data sets pure water solutions NaCl have been collected. regard, three intelligent paradigms, including Categorical Boosting (CatBoost), Extra Trees, Light Gradient Machine, were implemented establishing accurate predictive paradigms brine. It was found that data-driven model achieved outstanding accuracy. Among suggested schemes, CatBoost outperformed other resulted more solubilities at wide range operating pressures, temperature, solvent salinities. context, yielded an overall root-mean-square error only 0.0218 performed than thermodynamic-based approach. Additionally, application SHapley Additive exPlanations Local Interpretable Model-Agnostic Explanations methods revealed excellent degree explainability interpretability newly proposed ensemble method modeling Lastly, help significantly dealing with tasks challenges related managing through also monitoring issues associated production from sour reservoirs, mainly corrosion controlling rise content produced gas.

Language: Английский

---

Mechanisms and Production Enhancement Effects of CO2/CH4 Mixed Gas Injection in Shale Oil

Energies, Journal Year: 2025, Volume and Issue: 18(1), P. 142 - 142

Published: Jan. 2, 2025

Shale oil, a critical unconventional energy resource, has received substantial attention in recent years. However, systematic research on developing shale oil using mixed gases remains limited, and the effects of various gas compositions crude rock properties, along with their potential for enhanced recovery, are not yet fully understood. This study utilizes PVT analysis, SEM, core flooding tests mixtures to elucidate interaction mechanisms among gas, rock, as well recovery efficiency different types. The results indicate that increasing mole fraction CH4 substantially raises saturation pressure, up 1.5 times its initial value. Pure CO2, by contrast, exhibits lowest rendering it suitable long-term pressurization strategies. CO2 shows exceptional efficacy reducing interfacial tension, though viscosity reduction exhibit minimal variation. Furthermore, markedly modifies pore structure through dissolution, porosity 2% enhancing permeability 61.63%. In both matrix fractured cores, rates achieved were 36.9% 58.6%, respectively, demonstrating improved production compared single-component gases. offers theoretical foundation novel insights into development.

Language: Английский

---

Advanced Smart Models for Predicting Interfacial Tension in Brine-Hydrogen/Cushion Gas Systems: Implication for Hydrogen Geo-Storage

Energy & Fuels, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 27, 2025

Language: Английский

---

Robust ensemble learning frameworks for predicting minimum miscibility pressure in pure nitrogen and gas mixtures containing nitrogen–crude oil systems: Insights from explainable artificial intelligence

The Canadian Journal of Chemical Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: May 4, 2025

Abstract Miscible gas injection techniques, such as nitrogen injection, are among the attractive enhanced oil recovery (EOR) techniques for improving factors in reservoirs. A key challenge implementing these is accurately determining minimum miscibility pressure (MMP). While laboratory experiments offer reliable results, they costly and time‐consuming, existing empirical correlations often have moderate accuracy, which limits their practical use. In this study, robust ensemble methods, namely light gradient boosting machine (LightGBM), extra trees (ET), categorical (CatBoost), were implemented modelling MMP pure mixtures containing nitrogen–crude systems. An extensive experimental database involving 164 data points was used to elaborate on predictive models. The findings revealed that proposed methods achieved outstanding accuracy training test datasets, with ET consistently outperforming other model provided most consistent predictions a total root mean square error (RMSE) of only 0.3197 MPa determination coefficient 0.9976. Additionally, exhibited very small RMSE values across broad range operational conditions. Furthermore, Shapley additive explanations (SHAP) method further validated interpretability model, allowing clear insights into impact input features. This study underlines significant potential learning enhance prediction systems, thereby aiding appropriate design kind EOR process supporting better decision‐making reservoir management.

Language: Английский

---

Modeling wax disappearance temperature using robust white-box machine learning

Fuel, Journal Year: 2024, Volume and Issue: 376, P. 132703 - 132703

Published: Aug. 5, 2024

Language: Английский

---

Data‐driven framework for predicting the sorption capacity of carbon dioxide and methane in tight reservoirs

Greenhouse Gases Science and Technology, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 17, 2024

Abstract As energy demand continues to rise and conventional fuel sources dwindle, there is growing emphasis on previously overlooked reservoirs, such as tight reservoirs. Shale coal formations have emerged highly attractive options due their substantial contributions global gas reserves. Enhanced shale recovery (ESGR) enhanced coalbed methane (ECBM) based injection are advanced techniques used increase the extraction of from formations. One key challenges associated with these methods accurately predicting sorption process its profile. This crucial because it affects how (CH 4 ) carbon dioxide (CO 2 stored released rock, significantly impacts evaluation content potential productivity Due high cost experimental procedures moderate accuracy existing predictive approaches, this study proposes various cheap consistent data‐driven schemes for CH CO in In regard, three intelligent models, including generalized regression neural network (GRNN), radial basis function (RBFNN), categorical boosting (CatBoost), were taught tested using more than 3800 real measurements To find automatically appropriate control parameters improve prediction ability, RBFNN CatBoost evolved grey wolf optimization (GWO). The obtained results exhibited encouraging capabilities suggested models. addition, was found that CatBoost‐GWO most accurate scheme total root mean square (RMSE) determination coefficient ( R 0.1229 0.9993 sorption, 0.0681 0.9970 respectively. Additionally, approach demonstrated physical validity by respecting tendencies respect operational parameters. Furthermore, model outperforms recently published machine learning approaches. Lastly, findings offer a significant contribution demonstrating can greatly ease estimating formations, thereby facilitating simulation other related process. © 2024 Society Chemical Industry John Wiley & Sons, Ltd.

Language: Английский

---