Investigating the characteristics of CH4 adsorption in coal reservoirs based on interpretable machine learning

Authors

• CHEN Ruibin, School of Earth Resources, China University of Geosciences, Wuhan 430074, China; Key Laboratory of Tectonics and Petroleum Resources, China University of Geosciences, Ministry of Education, Wuhan 430074, ChinaFollow
• ZHOU Sandong, School of Earth Resources, China University of Geosciences, Wuhan 430074, China; Key Laboratory of Tectonics and Petroleum Resources, China University of Geosciences, Ministry of Education, Wuhan 430074, ChinaFollow
• ZHANG Weixin, School of Earth Resources, China University of Geosciences, Wuhan 430074, China; Key Laboratory of Tectonics and Petroleum Resources, China University of Geosciences, Ministry of Education, Wuhan 430074, China
• YAN Detian, School of Earth Resources, China University of Geosciences, Wuhan 430074, China
• WANG Hua, School of Earth Resources, China University of Geosciences, Wuhan 430074, China

Abstract

Objective Accurately predicting methane (CH₄) adsorption capacity in coal reservoirs is critical to the efficient exploration and exploitation of coalbed methane (CBM). Existing studies largely rely on experimental data from a small number of samples and theoretical adsorption models such as Langmuir. However, these studies exhibit time-consuming and laborious processes, as well as conclusions with limited generalizability, posing challenges to their widespread applications across different basins. Methods Based on 1150 sets of experimental data on CH₄ adsorption in coals, this study investigated five machine learning models, namely artificial neural network (ANN), support vector machine (SVM), K-nearest neighbor (KNN), random forest (RF), and gradient boosting decision tree (GBDT). Then, the SHapley Additive exPlanations (SHAP) algorithm was combined to achieve accurate prediction and interpretation of CH₄ adsorption in coal reservoirs. Results and Conclusion The stacked ensemble model, combined with the SHAP algorithm, reveals that the CH₄ adsorption capacity in coals increases with increasing pressure, maximum vitrinite reflectance (R_max), and mass fraction of fixed carbon and decreases with increases in the temperature and the mass fractions of water, ashes, and volatile constituents. Among these factors, R_max represents the dominant factor controlling CH₄ adsorption (contribution ratio: 32.7%), while high pressure, low temperature, and high coal rank create the most favorable conditions for CH₄ adsorption. Based on 2879 sets of data obtained through data expansion using the stacked ensemble model, this study established a multivariate nonlinear formula for calculating the CH₄ adsorption capacity in coal reservoirs. This equation was verified using 126 sets of new data from coals of varying ranks in the Ordos and Qinshui basins, yielding a coefficient of determination (R²) of 0.857. The total CH₄ adsorption capacity in coal reservoirs across various basins in multiple countries was assessed. The results reveal that the Ordos Basin of China exhibits an average total capacity of 31.87 × 10¹² m³ in coal reservoirs at burial depths of less than 5000 m, establishing it as the basin with the greatest resource potential among its various counterparts in countries such as China, the United States, and Australia. Overall, this study enables the accurate prediction of CH₄ adsorption capacity in coals, even under limited experimental data or incomplete coal quality data, thereby providing theoretical and technical support for CBM reserve evaluation and the selection of optimal CBM sweet spots.

Keywords

coalbed methane (CBM), CH4 adsorption, machine learning, SHapley Additive exPlanations (SHAP), multivariate nonlinear formula, resource evaluation

DOI

10.12363/issn.1001-1986.26.04.0263

Recommended Citation

CHEN Ruibin, ZHOU Sandong, ZHANG Weixin, et al. (2026) "Investigating the characteristics of CH4 adsorption in coal reservoirs based on interpretable machine learning," Coal Geology & Exploration: Vol. 54: Iss. 9, Article 10.
DOI: 10.12363/issn.1001-1986.26.04.0263
Available at: https://cge.researchcommons.org/journal/vol54/iss9/10

Reference

[1] 秦勇. 中国深部煤层气地质研究进展[J]. 石油学报,2023,44(11):1791−1811. QIN Yong. Progress on geological research of deep coalbed methane in China[J]. Acta Petrolei Sinica,2023,44(11):1791−1811.

[2] 罗平亚,朱苏阳,李小刚. 中浅层煤层气地面建井开发提高采收率研究进展:机制、主控因素与主要技术[J]. 油气藏评价与开发,2026,16(5):949−962. LUO Pingya,ZHU Suyang,LI Xiaogang. Research progress on enhancing recovery of medium–shallow coalbed methane through surface well development: Mechanisms,main controlling factors,and main technologies[J]. Petroleum Reservoir Evaluation and Development,2026,16(5):949−962.

[3] WEI Qiang,CHEN Song,YI Wangfei,et al. Gas content,geochemical characteristics and implications of coalbed methane from the deep area of Qi'nan coalmine in Huaibei Coalfield[J]. Scientific Reports,2024,14:28334.

[4] 周立宏,闫霞,熊先钺,等. 深部煤层气超临界状态下赋存特征及分配规律[J]. 煤炭科学技术,2025,53(3):73−90. ZHOU Lihong,YAN Xia,XIONG Xianyue,et al. Characteristics of occurrence and distribution rule of deep coalbed methane in supercritical state[J]. Coal Science and Technology,2025,53(3):73−90.

[5] 石军太,徐凤银,曹运兴,等. 煤储层改造背景下深部煤层气藏物质平衡方程的建立与应用[J]. 煤炭科学技术,2025,53(3):200−217. SHI Juntai,XU Fengyin,CAO Yunxing,et al. Establishment and application of material balance equation for deep coalbed methane reservoirs based on coal formation stimulation background[J]. Coal Science and Technology,2025,53(3):200−217.

[6] 韩思杰,郭庆,桑树勋,等. 深部高阶煤层气赋存状态分配特征与游离气赋存下限孔径探析[J/OL]. 煤炭学报,2026:1−18 [2026-03-25]. https://link.cnki.net/urlid/11.2190.TD.20260325.1326.004. HAN Sijie,GUO Qing,SANG Shuxun,et al. The occurrence state of deep coalbed methane and the lower limit pore diameter for the existence of the free gas[J/OL]. Journal of China Coal Society,2026:1−18 [2026-03-25]. https://link.cnki.net/urlid/11.2190.TD.20260325.1326.004.

[7] 宋学梅,张琨,桑树勋,等. 基于单层吸附及吸附势理论的煤中CH4与CO2跨临界态吸附模型改进[J]. 煤田地质与勘探,2026,54(1):94−106. SONG Xuemei,ZHANG Kun,SANG Shuxun,et al. An improved model of trans–critical CH4 and CO2 adsorption in coals based on monolayer adsorption and adsorption potential theories[J]. Coal Geology & Exploration,2026,54(1):94−106.

[8] 王洁. 基于RF与GA–BP融合模型的深部煤层含气量预测[J]. 录井工程,2025,36(4):21−28. WANG Jie. Gas content prediction of deep coalseam based on RF and GA–BP fusion model[J]. Mud Logging Engineering,2025,36(4):21−28.

[9] 武守亚,梁卫国. 煤层吸附气密度模型及气体吸附行为再认识[J]. 煤炭学报,2025,50(2):1115−1126. WU Shouya,LIANG Weiguo. Adsorbed gas density model and re–understanding of gas adsorption behavior in coal seam[J]. Journal of China Coal Society,2025,50(2):1115−1126.

[10] 周三栋,魏佳成,刘大锰,等. 鄂尔多斯盆地东缘深部多相态煤层气含量预测模型[J]. 煤炭科学技术,2025,53(增刊2):234−248. ZHOU Sandong,WEI Jiacheng,LIU Dameng,et al. Multi–phase coalbed methane content model in deep strata of eastern margin of Ordos Basin[J]. Coal Science and Technology,2025,53(Sup.2):234−248.

[11] WANG Liang,LIU Minxuan,LI Jing,et al. Investigation of methane adsorption mechanism in low–rank bituminous coal considering internal water distribution: A molecular simulation and experimental study[J]. Chemical Engineering Journal,2024,495:153562.

[12] WENIGER P,KALKREUTH W,BUSCH A,et al. High–pressure methane and carbon dioxide sorption on coal and shale samples from the Paraná Basin,Brazil[J]. International Journal of Coal Geology,2010,84(3/4):190−205.

[13] 郑永旺,李超,吴剑,等. 沁水盆地北缘晋中区块深部煤层气富集规律及有利区优选[J]. 东北石油大学学报,2025,49(6):85−97. ZHENG Yongwang,LI Chao,WU Jian,et al. Enrichment law of deep coalbed methane and optimization of favorable areas in Jinzhong Block on the northern margin of Qinshui Basin[J]. Journal of Northeast Petroleum University,2025,49(6):85−97.

[14] FENG Qihong,ZHANG Jiyuan,ZHANG Xianmin,et al. The use of alternating conditional expectation to predict methane sorption capacity on coal[J]. International Journal of Coal Geology,2014,121:137−147.

[15] LI Wenshuo,LI Wei,BUSCH A,et al. Machine learning algorithm to predict methane adsorption capacity of coal[J]. Energy & Fuels,2024,38(24):23422−23432.

[16] 刘浪,袁三一,于越,等. 基于多尺度随机森林融合地质、测井和地震资料的煤层含气量预测[J]. 石油地球物理勘探,2026,61(2):283−293. LIU Lang,YUAN Sanyi,YU Yue,et al. The prediction of coalbed methane content based on multi–scale random forest integrating geological,logging,and seismic data[J]. Oil Geophysical Prospecting,2026,61(2):283−293.

[17] WANG Yongjun,VO THANH H,AL–MUDHAFAR W J,et al. Data driven based deep learning for optimizing carbon storage and methane adsorption in unconventional shale gas reservoirs[J]. Journal of Environmental Chemical Engineering,2025,13(3):116901.

[18] CHEN Ruibin,ZHOU Sandong,LIU Xinyu,et al. Application of Stacking machine learning model with SHapley Additive explanation (SHAP) in estimating the changes of mechanical properties in coals under CO2[J]. Fuel,2026,405:136461.

[19] DENG Jinchang,KANG Jianhong,ZHOU Fubao,et al. The adsorption heat of methane on coal: Comparison of theoretical and calorimetric heat and model of heat flow by microcalorimeter[J]. Fuel,2019,237:81−90.

[20] DUTTA P,BHOWMIK S,DAS S. Methane and carbon dioxide sorption on a set of coals from India[J]. International Journal of Coal Geology,2011,85(3/4):289−299.

[21] GENSTERBLUM Y,MERKEL A,BUSCH A,et al. High–pressure CH4 and CO2 sorption isotherms as a function of coal maturity and the influence of moisture[J]. International Journal of Coal Geology,2013,118:45−57.

[22] LI Song,TANG Dazhen,PAN Zhejun,et al. Evaluation of coalbed methane potential of different reservoirs in western Guizhou and eastern Yunnan,China[J]. Fuel,2015,139:257−267.

[23] LIU Dameng,ZOU Zhuo,CAI Yidong,et al. An updated study on CH4 isothermal adsorption and isosteric adsorption heat behaviors of variable rank coals[J]. Journal of Natural Gas Science and Engineering,2021,89:103899.

[24] MA Ruying,YAO Yanbin,WANG Meng,et al. CH4 and CO2 adsorption characteristics of low–rank coals containing water: An experimental and comparative study[J]. Natural Resources Research,2022,31(2):993−1009.

[25] SHAN Changan,ZHANG Tingshan,LIANG Xing,et al. Influence of chemical properties on CH4 adsorption capacity of anthracite derived from southern Sichuan Basin,China[J]. Marine and Petroleum Geology,2018,89:387−401.

[26] TANG Zongqing,YANG Shengqiang,XU Guang,et al. Evolution law of adsorption and desorption characteristics of CH4 in coal masses during coalbed methane extraction[J]. Energy & Fuels,2018,32(10):10540−10548.

[27] WANG Linlin,LONG Zhengjiang,QU Zhenghui,et al. Effects of functional groups on supercritical CH4 adsorption and desorption characteristics of tectonically deformed coals[J]. Fuel,2022,325:124934.

[28] WENIGER P,FRANCŮ J,HEMZA P,et al. Investigations on the methane and carbon dioxide sorption capacity of coals from the SW Upper Silesian Coal Basin,Czech Republic[J]. International Journal of Coal Geology,2012,93:23−39.

[29] YAN Jiwei,MENG Zhaoping,LI Guoqing. Diffusion characteristics of methane in various rank coals and the control mechanism[J]. Fuel,2021,283:118959.

[30] ZHAO Pengtao,WANG Haiqiao,WANG Long,et al. Prediction of methane adsorption capacity in different rank coal at low temperature by the Polanyi–based isotherm model[J]. Energy Science & Engineering,2022,10(4):1384−1397.

[31] ZHAO Junlong,XU Hao,TANG Dazhen,et al. Coal seam porosity and fracture heterogeneity of macrolithotypes in the Hancheng Block,eastern margin,Ordos Basin,China[J]. International Journal of Coal Geology,2016,159:18−29.

[32] ZHAO Junlong,XU Hao,TANG Dazhen,et al. A comparative evaluation of coal specific surface area by CO2 and N2 adsorption and its influence on CH4 adsorption capacity at different pore sizes[J]. Fuel,2016,183:420−431.

[33] 张云骥,李倩,蔡益栋,等. 基于机器学习测井反演的宏观煤岩类型评价:以鄂尔多斯盆地佳县区块本溪组8号煤为例[J]. 现代地质,2025,39(5):1427−1436. ZHANG Yunji,LI Qian,CAI Yidong,et al. Evaluation of coal macrolithotypes based on machine learning logging inversion: A case study of the No. 8 coal seam in the Benxi Formation,Jiaxian Block,Ordos Basin[J]. Geoscience,2025,39(5):1427−1436.

[34] 郭广彬,田婷,周忠国,等. 多机器学习算法下沧源县滑坡地质灾害易发性评价[J]. 地理空间信息,2026,24(2):64−67. GUO Guangbin,TIAN Ting,ZHOU Zhongguo,et al. Landslide geological hazard vulnerability evaluation in Cangyuan County,Lincang City,Yunnan Province under multiple machine learning algorithms[J]. Geospatial Information,2026,24(2):64−67.

[35] 周斐然,尹子悦,孙晓敏,等. 融合数值模拟和机器学习的民勤盆地地下水潜力评价与主控因素识别[J]. 地学前缘,2026,33(1):511−522. ZHOU Feiran,YIN Ziyue,SUN Xiaomin,et al. Integrating numerical simulation and machine learning for identification of groundwater potential zone and its governing factors in the Minqin Basin,Northwest China[J]. Earth Science Frontiers,2026,33(1):511−522.

[36] 张恩威,孟庆涛,唐佰强,等. 基于机器学习的TOC测井预测方法:以松辽盆地南部青山口组一段为例[J]. 吉林大学学报(地球科学版),2026,56(3):1062−1075. ZHANG Enwei,MENG Qingtao,TANG Baiqiang,et al. Machine learning–based TOC logging prediction method: A case study of the first member of Qingshankou Formation in the southern part of Songliao Basin[J]. Journal of Jilin University (Earth Science Edition),2026,56(3):1062−1075.

[37] 尉瑞,宋鑫,廉振山,等. CH4,CO2和N2及其多元气体在不同煤阶中吸附特性实验研究[J]. 煤炭技术,2022,41(10):129−134. YU Rui,SONG Xin,LIAN Zhenshan,et al. Experimental study on adsorption properties of CH4,CO2 and N2 and their multivariate gases in different coal ranks[J]. Coal Technology,2022,41(10):129−134.

[38] 刘振建,蒲超,都喜东,等. 不同煤阶煤纳米孔隙结构及其对CO2/CH4选择性吸附的影响[J]. 煤质技术,2025,40(3):69−75. LIU Zhenjian,PU Chao,DU Xidong,et al. Nano–pore structure of different rank coals and its effect on selective adsorption of CO2 over CH4[J]. Coal Quality Technology,2025,40(3):69−75.

[39] 高昌靖. 煤储层中甲烷与二氧化碳竞争吸附及二氧化碳地质封存机理研究[D]. 北京: 中国地质大学(北京), 2024. GAO Changjing. Study on mechanism of CH4/CO2 competitive adsorption and CO2 geological sequestration in coal reservoir[D]. Beijing: China University of Geosciences (Beijing), 2024.

[40] 姜延航,白刚,周西华,等. 煤吸附CH4体积测试与分析[J]. 煤炭科学技术,2022,50(12):144−152. JIANG Yanhang,BAI Gang,ZHOU Xihua,et al. Test and analysis of coal adsorption volume of CH4[J]. Coal Science and Technology,2022,50(12):144−152.

[41] ZHANG Songhang,TANG Shuheng,LI Zhongcheng,et al. Effect of pore structure on competitive sorption and diffusion of mixed methane and carbon dioxide in anthracite,south Qinshui Basin,China[J]. International Journal of Coal Geology,2022,253:103956.

[42] CHEN Ruibin,ZHOU Sandong,ZHANG Weixin,et al. Evaluation of hydrogen adsorption capacity in coal by explainable machine learning: Implication for hydrogen geo–storage[J]. Fuel,2026,425:139381.

[43] 王琳琳,龙正江,朱冠宇. 中阶构造煤等温吸附/解吸特征及机理[J]. 沉积学报,2022,40(1):60−72. WANG Linlin,LONG Zhengjiang,ZHU Guanyu. Isothermal adsorption/desorption characteristics and mechanism of tectonically deformed medium–rank coals[J]. Acta Sedimentologica Sinica,2022,40(1):60−72.

[44] 陈建奇,胡维强,李洋冰,等. 鄂尔多斯盆地东缘临兴–神府区块深部煤储层孔隙结构对甲烷吸附影响研究[J]. 岩矿测试,2026,45(1):116−130. CHEN Jianqi,HU Weiqiang,LI Yangbing,et al. Influence of pore structure on methane adsorption in deep coal reservoirs of the Linxing–Shenfu Block,eastern margin of the Ordos Basin[J]. Rock and Mineral Analysis,2026,45(1):116−130.

[45] 李国欣,陈瑞银,温志新,等. 全球煤岩气资源潜力与分布[J]. 石油勘探与开发,2026,53(1):14−26. LI Guoxin,CHEN Ruiyin,WEN Zhixin,et al. Global coal–rock gas resource potential and distribution[J]. Petroleum Exploration and Development,2026,53(1):14−26.

[46] 丁蓉,曹毅民,赵培华,等. "十四五"(2021–2025年)煤层气资源评价挑战、结果及产业发展启示[J]. 石油学报,2026,47(2):330−338. DING Rong,CAO Yimin,ZHAO Peihua,et al. Challenges,results and industrial implications of coalbed methane resource evaluation during the 14th Five–Year Plan period (2021–2025)[J]. Acta Petrolei Sinica,2026,47(2):330−338.

[47] 李国欣,张君峰,赵群,等. 煤岩气资源评价方法与中国陆上有利区勘探潜力[J]. 石油勘探与开发,2025,52(6):1231−1245. LI Guoxin,ZHANG Junfeng,ZHAO Qun,et al. Evaluation methodology of coal–rock gas resources and exploration potential of favorable areas in onshore China[J]. Petroleum Exploration and Development,2025,52(6):1231−1245.