Energy & Fuels最新文献

{"title":"Impact of Polymer Ripening Degree on Fluidity Control and Enhanced Oil Recovery in Heterogeneous Reservoirs","authors":"Baofeng Hou,&nbsp;Yali Zhang,&nbsp;Shuo Zhao,&nbsp;Yansong Bai,&nbsp;Lifeng Chen,&nbsp;Jingyao Zhao,&nbsp;Zhiyuan Tang,&nbsp;Yu Chen,&nbsp;Wei Xu","doi":"10.1021/acs.energyfuels.6c04162","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c04162","url":null,"abstract":"<p>The performance and evaluation methods of polymers withdifferentripening degrees in enhanced oil recovery were investigated in thispaper, which can provide guidance for reducing the viscosity lossduring polymer injection in the oilfield. The results show that theindoor complete ripening time of the polymer was set as 50 min. Thestorage modulus and loss modulus of polymer systems with 60% and 80%ripening degrees are higher than those with a 100% ripening degree,indicating that incompletely ripened polymer systems exhibit superiorviscoelasticity. For the fully ripened systems, polymer molecularchains unfold completely, while incompletely ripened systems restrictsolvent penetration and retain unhydrated micellar particles, whichdirectly affect the seepage resistance and oil flooding recoveries.Capillary shear can further promote the ripening of the polymer unripenedmicelles and increase the apparent viscosity of the polymer solution.Compared with the polymer with a 100% ripening degree, higher resistancecoefficients and residual resistance coefficients can be establishedby the polymer systems with 60% and 80% ripening degrees. In heterogeneousporous media, incompletely ripened polymer systems exhibit an excellentplugging and flooding performance owing to the residual undissolvedmicellar particles, which enhance the fluid uptake in low-permeabilitylayers to improve oil recoveries. As validated by quantitative particle-sizeand image analyses, whether it is a macrodisplacement or microdisplacementexperiment, when the incompletely ripened polymer is used, the effectiveviscosity of the flooding fluid can be improved and the areal sweepefficiency can be increased to enhance the oil recovery. Thus, incompletelyripened polymers show promising potential for oil flooding.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24215–24227"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330386","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Thickeners for Anhydrous CO2 Fracturing Fluids: From Molecular Design to Field Applications","authors":"Ying Zhang,&nbsp;Kang Liu,&nbsp;Fei Ding,&nbsp;Weichu Yu,&nbsp;Mingwei Zhao,&nbsp;Qing You,&nbsp;Yongbiao Hua","doi":"10.1021/acs.energyfuels.6c03289","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c03289","url":null,"abstract":"<p>Conventional hydraulic fracturing consumes massive amountsof waterand causes severe formation damage. Anhydrous CO₂ fracturingfluids offer a promising low-carbon alternative. Nevertheless, thelow intrinsic viscosity of CO₂ severely restricts proppanttransport and complex fracture propagation, hindering large-scalefield application. Over the past three decades, research on CO₂ thickeners has evolved from empirical screening to mechanism-guidedmolecular design, with the thickening paradigm shifting from simplepolymer chain entanglement to dynamic supramolecular self-assemblynetworks. However, systematic and critical reviews focusing on theirdevelopment, key challenges, and paradigm limitations are rare. Thispaper presents a comprehensive review of CO₂ thickeners,covering their molecular design, synthesis approaches, performancecontrol, thickening mechanisms, and field applications. We summarizetheir evolution from fluorinated and siloxane-based polymers to hydrocarbonmodifiers, supramolecular small molecules, and surfactant systems,and clarify the intrinsic thickening mechanisms of each category.Field case studies show that although siloxane-based and fluorinatedsystems exhibit application potential, they are still constrainedby strong cosolvent dependence or environmental persistence risks.To address these bottlenecks, this review proposes prospective developmentdirections, including smart-responsive materials, bioinspired greenchemistries, and AI-assisted inverse molecular design. This reviewis expected to provide a theoretical reference and practical guidancefor the development of high-efficiency, eco-friendly, and field-scalableCO₂ fracturing technologies.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24077–24098"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330298","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mechanistic Insights into Chemical-Assisted Steam Injection: Pore-Scale Evidence for More Efficient and Lower-Carbon Heavy Oil Recovery","authors":"Jingjing Huang,&nbsp;Tayfun Babadagli","doi":"10.1021/acs.energyfuels.6c04650","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c04650","url":null,"abstract":"<p>Steam-assistedgravity drainage (SAGD) is the primary method for <em>in situ</em> extraction of heavy oil and bitumen in Canada. However,it (as well as other steam injection methods) suffers from low efficiency,high water and energy consumption, and significant environmental impact.Adding chemicals to steam offers a potential solution by enhancingoil recovery while reducing carbon emissions through lower steam usage.Our previous studies have shown that deep eutectic solvents (DES)and thermally stable nonionic surfactants improve steam chamber sweepefficiency and microscopic displacement efficiency, respectively.However, their mechanisms and effects in porous media, especiallyat the pore scale, remain unclear. This study explores the microdisplacementmechanisms of these chemical additives. A porous medium model wascreated by packing glass beads between two plexiglass plates, whichwas then saturated with heavy oil at a slow injection rate. The researchfocused on steam chamber growth dynamics, physicochemical changesat the oil-steam-water-rock interface, and their impact on oil drainage.By comparing oil recovery efficiency, energy efficiency, and carbonintensity of SAGD with and without chemical additives, the effectsof DES (formulated in our laboratories and named as DES 11) and Novelfroth(a commercial nonionic surfactant) on steam injection efficiency wereevaluated. The results demonstrate that DES 11 and Novelfroth 190enhance the recovery rate by 31.1% and 37%, respectively, while reducingcarbon emission intensity to one-third to one-half of conventionalSAGD. However, their underlying mechanisms differ. Also, the introductionof these chemical additives significantly altered the state of residualoil compared to the base case. Ultimately, the improvement in oilrecovery per unit of steam injected in the chemical cases was quantifiedand linked to a reduction in greenhouse gas emissions during steaminjection.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24440–24451"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330307","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mode-Aware Production Forecasting of Coalbed Methane Wells via Unsupervised Parameter-Image Clustering and Similarity-Based GAF Correction","authors":"Jiyuan Zhang,&nbsp;Jianhao Liu,&nbsp;Xia Yan,&nbsp;Juankai Bai,&nbsp;Qihong Feng,&nbsp;Sen Wang","doi":"10.1021/acs.energyfuels.6c04232","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c04232","url":null,"abstract":"<p>Accurate and efficient production forecasting is criticalfor coalbedmethane (CBM) well development and management. However, conventionalengineering methods rely heavily on physical assumptions and parametercalibration, whereas direct data-driven models typically establisha global input–output mapping without explicitly distinguishingdifferent production patterns or exploiting similarities among samples,thereby limiting their forecasting accuracy and generalizability.To address these limitations, this study proposes a structured forecastingframework integrating parameter-image representation, unsupervisedclustering, lightweight classification, and similarity-based correction.Geological and development parameters are first converted into grayscaleimages, and spectral clustering combined with K-means++ is used toidentify production-related modes. A MobileNetV2-based classifieris then trained to determine the mode category of new samples. Withinthe identified category, the most similar sample is retrieved, andits production sequences are encoded as Gramian Angular Field (GAF)images. Difference-image-guided posterior correction is further appliedbefore reconstructing the corrected images into production time series.Results based on numerical simulation samples show that the proposedframework achieves high accuracy for both gas and water productionforecasting, with <em>R</em>² values of 0.99 and0.99 for daily gas and daily water production, respectively. Comparedwith representative machine-learning and deep-learning models, theproposed method demonstrates superior accuracy and efficiency, indicatingits potential for rapid CBM production forecasting.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24291–24309"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dimensionless Scaling of First-Cycle Huff-n-Puff Hydrocarbon Recovery and CO2 Storage in Shale Reservoirs","authors":"Qian Zhang,&nbsp;Ming Ma,&nbsp;Hamid Emami-Meybodi","doi":"10.1021/acs.energyfuels.6c04290","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c04290","url":null,"abstract":"<p>Cyclic CO₂ injection is the key componentof carboncapture, utilization, and storage (CCUS) that enhances hydrocarbonrecovery while sequestering CO₂, contributing to globalgreenhouse gas mitigation. However, its optimization remains challengingbecause the process is governed by numerous interacting variableswith nonlinear and competing effects, limiting the transferabilityof optimized parameters across reservoir systems. To address this,for the first time, this study applies dimensional analysis to identifythe key dimensionless groups and establish physically interpretablescaling relations governing CO₂ injection performance,enabling optimization with substantially reduced computational cost.A species-based mass transport model incorporating inhomogeneous fluiddistribution induced by fluid–solid interactions is used tosimulate cyclic CO₂ injection and generate a comprehensivedata set for dimensional analysis. Based on the Vaschy–BuckinghamPi theorem, dimensional analysis reveals key dimensionless groupsgoverning hydrocarbon recovery (RF) and CO₂ storage (EF),including response-specific Fourier numbers (Fo_R, Fo_S), diffusivity ratios (<em>R</em>_DR, <em>R</em>_DS), fluid–solid friction modulus (φ_R, φ_S), together with time ratio (<em>R_T</em>), normalized drawdown pressure (<em>P</em>_D), and relative fluid–solid interaction coefficient(χ). The RF scaling relation identifies a critical point, i.e.,ψ_RC = 0.0015, separating mixing-dominated and mixing-limitedregimes, providing a calibrated transition criterion for identifyingdiminishing recovery gains. The critical point ψ_SC = 0.0025 in the EF scaling relation marks the transition betweenretention-growth and retention-limited regimes, beyond which additionalgains in postcycle CO₂ retention diminish. This study providespractical guidance for cyclic CO₂ injection in ultratightreservoirs, supporting more effective CO₂ utilization forhydrocarbon recovery and improved CO₂ storage performance.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24310–24327"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330385","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Correction to “Rare Earth-Doped MoS2 for Supercapacitor Application”","authors":"M. Isacfranklin,&nbsp;L. Esther Maria Princy,&nbsp;Yuvakkumar Rathinam,&nbsp;L. Kungumadevi,&nbsp;G. Ravi,&nbsp;Abdullah G. Al-Sehemi,&nbsp;Dhayalan Velauthapillai","doi":"10.1021/acs.energyfuels.6c05148","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c05148","url":null,"abstract":"","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24464–24464"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330318","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CO2 Enhanced Hydrocarbon Recovery in Organic-Rich Shales: Insights from 500 MHz 1H NMR","authors":"Jin-Hong Chen,&nbsp;J. David Broyles,&nbsp;Mohammed Boudjatit","doi":"10.1021/acs.energyfuels.6c03407","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c03407","url":null,"abstract":"<p>CO₂ huff-and-puff hasemerged as a promising approachfor enhanced oil recovery in unconventional source rock shales, yetthe underlying recovery mechanisms remain difficult to isolate andquantify. Here, high-field ¹H nuclear magnetic resonance(NMR) spectroscopy was used to investigate <em>in situ</em> CO₂–shale interactions in a preserved, intactslim source rock plug. CO₂ was stepwise loaded from 0 to6000 psi and unloaded back to 0 psi at 40 °C. At each pressurestep, 500 MHz ¹H spectra were acquired and deconvolutedinto two Lorentzian peaks representing pore fluid and one Gaussianpeak representing kerogen. Pressure-dependent changes in the integralsand line widths of these peaks were used to investigate hydrocarbondepletion from the rock sample and CO₂-induced changesin molecular mobility of kerogen and pore fluids. In control experiments,helium injections up to 2500 psi produced no measurable changes inthe NMR spectra. In contrast, CO₂ injections caused pronounced,pressure-dependent changes in intensities and line widths of ¹H NMR signals, revealing mixing with pore fluids and penetrationinto the kerogen matrix. Both pore fluid and kerogen spectra changedmarkedly as CO₂ entered the supercritical regime above1000 psi at 40 °C, indicating a distinct transition in molecularmobility. Above 4000 psi, substantial light hydrocarbon evaporationfrom the plug left heavier, less-mobile components in the pores, resultingin broader NMR peaks. Analysis of the complete loading–unloadingcycle indicated that approximately 24.5% of the initial pore fluidwas produced, dominated by late-stage high-pressure CO₂ loading. These results demonstrate that high-field ¹HNMR can quantify hydrocarbon recovery while resolving molecular-scaleCO₂ interactions with different components in source rockshales that are not easily accessible by conventional mass-balancemethods.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24250–24257"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330355","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep Learning-Based Recognition and Quantitative Characterization of CO2 Hydrate Growth in Microfluidic Porous Media","authors":"Chenyi Zhang,&nbsp;Chaozheng Ma,&nbsp;Tingting Luo,&nbsp;Tao Han,&nbsp;Weihao Yang,&nbsp;Yanghui Li","doi":"10.1021/acs.energyfuels.6c03648","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c03648","url":null,"abstract":"<p>The growth and morphological evolution of CO₂ hydratesin marine sediments are recognized as highly significant for hydrate-basedCO₂ sequestration (HBCS). However, it remains challengingfor their pore-scale processes to be directly characterized via macroscopicparameters. To ensure that issues such as weak boundaries, matrixtexture interference, and internal pores in microfluidic CO₂ hydrate images are addressed, an automated recognition and quantitativeanalysis method based on deep learning semantic segmentation was developedin this study. An image data set was compiled from microfluidic experiments,and four models (U-Net, Res-U-Net, Attention U-Net, and DeepLabv3+)were subsequently trained and compared. The results demonstrated thatthe hydrate regions were effectively identified by all four models.Among these, optimal performance was exhibited by the Attention U-Netmodel, with hydrate intersection over union (IoU) and Dice coefficientsbeing recorded at 0.951 and 0.975, respectively. Furthermore, theoptimal model was applied to continuous image sequences; consequently,the dynamic tracking of the CO₂ hydrate growth processwas achieved, and parameters, including hydrate saturation, internalpore ratio, and growth rate, were extracted. It was revealed thatinternal pores in the saline system are predominantly distributedwithin the hydrate envelope and that the inhibitory effect of thesaline solution on hydrate growth is strengthened as the ion concentrationis increased. At NaCl concentrations of 1, 3, and 5 wt %, the averagegrowth rates were reduced by approximately 10.6%, 33.1%, and 70.4%,respectively. Among the various 3 wt % saline solutions tested, thelowest growth rate was observed in the NaCl system, whereas relativelyhigh areal growth rates were maintained by the KCl, MgCl₂, and CaCl₂ systems. Thus, a reproducible pixel-levelrecognition and quantitative characterization approach for the microfluidicCO₂ hydrate growth process is provided by our proposedmethod.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24407–24420"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330317","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Molecular Design Pathways of Phenanthroline-Based Interwell Water Tracers for Oilfield Applications","authors":"Bora Yoon,&nbsp;Sehoon Chang,&nbsp;Hsieh Chen","doi":"10.1021/acs.energyfuels.6c03103","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c03103","url":null,"abstract":"<p>Interwell water tracer tests are critical for characterizingreservoirconnectivity and flow behavior during waterflood and enhanced oilrecovery operations. Existing tracers, however, are often limitedby analytical complexity and challenges in scaling to multiwell deployments.This work presents a molecular design strategy for phenanthroline-based,lanthanide-sensitized interwell tracers by systematically examininghow backbone functionalization governs photophysical performance,chemical robustness, analytical detectability, and reservoir-relevanttransport behavior. A focused library of phenanthroline derivativeswas synthesized and evaluated to decouple the roles of chelating andsolubilizing functionalities. Photophysical characterization showsthat while the phenanthroline antenna is broadly preserved acrossthe derivative set, efficient Eu³⁺ sensitization and extendedexcited-state lifetimes require incorporation of a 2,9-dicarboxylatecoordination motif. Sulfonation alone does not enable effective ligand-to-metalenergy transfer. In contrast, combining sulfonation with dicarboxylatechelation in 5-((3-sulfopropyl)amino)-1,10-phenanthroline-2,9-dicarboxylicacid (SA-PDCA) produces strong Eu³⁺ emission, favorableexcited-state stabilization, and improved aqueous compatibility. SA-PDCAis detected optically at ppb-level concentrations and is efficientlyrecovered from produced-water matrices via solid-phase extractionwith substantial retention of signal intensity. Hydrothermal stabilitytesting demonstrates that substituent identity and attachment modestrongly influence chemical robustness under high-temperature, high-salinityconditions. Coreflood experiments conducted under coinjection conditionsshowed that SA-PDCA exhibited transport behavior comparable to dipicolinicacid, with near-quantitative apparent recovery in carbonate rock.Collectively, these results establish SA-PDCA as a promising leadtracer candidate and define generalizable design principles for advancedreservoir surveillance applications.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24178–24186"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine Learning-Based Prediction of Shale Adsorption Capacity after Supercritical CO2 Soaking","authors":"Yiming Zhang,&nbsp;Wenbin Ma,&nbsp;Qiao Lyu,&nbsp;Linyuan Chen,&nbsp;Xiaoli Zeng","doi":"10.1021/acs.energyfuels.6c03880","DOIUrl":"https://doi.org/10.1021/acs.energyfuels.6c03880","url":null,"abstract":"<p>Supercritical CO₂ (scCO₂) soaking can alterthe pore structure, mineral composition, and surface properties ofshale, thereby affecting its adsorption capacity. Therefore, accurateprediction of shale adsorption capacity after scCO₂ soakingis of great significance for evaluating CO₂ geologicalsequestration potential and shale gas recovery performance. In thisstudy, a database of the amount of CO₂ adsorbed by shalewas collected, which contained 315 data points from 26 shale samplesbefore and after scCO₂ soaking. Eight input variables wereselected, including adsorption pressure, adsorption temperature, soakingpressure, soaking temperature, soaking time, quartz content, claycontent, and total organic carbon (TOC) content, with CO₂ adsorption capacity as the output variable. Four machine learningmodels, namely multilayer perceptron (MLP), support vector regression(SVR), extreme gradient boosting (XGBoost), and Light Gradient BoostingMachine (LightGBM), were constructed and optimized to predict theshale adsorption capacity. The results showed that XGBoost exhibitedthe best predictive performance among the four models, achieving an <em>R</em>² of 0.9772 on the test set. To address theinsufficient predictive performance of XGBoost under high soakingpressure conditions, an XGBoost–Langmuir model was developedby inversely determining the Langmuir adsorption parameters from XGBoostpredictions. Compared with the XGBoost and Langmuir models, the XGBoost-Langmuirmodel significantly improved prediction accuracy and reduced errorsunder high soaking pressures. Analyses of SHAP (SHapley Additive exPlanations)values and SHAP interaction values indicated that clay content exhibitedasymmetric contributions to model predictions, adsorption temperaturegenerally contributed negatively, and quartz content contributed positively,while the contribution of TOC to model predictions was associatedwith its interactions with pressure conditions. These results demonstratethat the proposed XGBoost-Langmuir model enhanced the reliabilityof predicting shale adsorption capacity after scCO₂ soakingand revealed the changes in adsorption-controlling factors inducedby scCO₂ treatment.</p>","PeriodicalId":35,"journal":{"name":"Energy & Fuels","volume":"40 44","pages":"24274–24290"},"PeriodicalIF":6.0,"publicationDate":"2026-11-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149330226","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}