岩性油气藏 ›› 2026, Vol. 38 ›› Issue (5): 191–200.doi: 10.12108/yxyqc.20260518

• 石油工程与油气田开发 • 上一篇    

鄂尔多斯盆地神府区块二叠系太原组煤层气产能预测方法

李新泽1(), 吴浩1, 晁嘉豪2, 李娜1(), 王苏冉1, 白玉湖1, 冯汝勇1, 程加陆3   

  1. 1 中海石油(中国)有限公司北京研究中心 勘探开发研究院北京 100028
    2 中联煤层气有限责任公司 勘探开发研究院北京 100016
    3 中国石油大学(北京) 石油工程学院北京 102249
  • 收稿日期:2025-09-12 修回日期:2025-11-25 出版日期:2026-09-01 发布日期:2026-09-04
  • 第一作者:李新泽(1996—),男,硕士,工程师。主要从事非常规油气田开发、油藏工程、渗流方面的研究工作。地址:(100028)北京市朝阳区太阳宫南街6号院。Email:lixz@cnooc.com.cn
  • 通信作者: 李娜
  • 基金资助:
    中国海洋石油有限公司综合科研项目“煤储层精细描述与产能主控因素研究”(2021OT-XNY09);中海石油(中国)有限公司“十四五”重大科技项目“临兴-神府深层煤层气赋存机理及与致密气协同开发关键技术研究”(KJGG2024-1007)

Prediction method for coalbed methane deliverability of Permian Taiyuan Formation in Shenfu block, Ordos Basin

LI Xinze1(), WU Hao1, CHAO Jiahao2, LI Na1(), WANG Suran1, BAI Yuhu1, FENG Ruyong1, CHENG Jialu3   

  1. 1 Exploration and Development Research InstituteBeijing Research Center of CNOOC (China) Ltd., Beijing 100028, China
    2 Research Institute of Exploration and DevelopmentChina United Coalbed Methane Co., Ltd., Beijing 100016, China
    3 College of Petroleum EngineeringChina University of Petroleum (Beijing), Beijing 102249, China
  • Received:2025-09-12 Revised:2025-11-25 Online:2026-09-01 Published:2026-09-04
  • Contact: LI Na E-mail:lixz@cnooc.com.cn;lina28@cnooc.com.cn

摘要:

深层煤层气在开采过程中,单井产量差异大,产能预测较困难。以鄂尔多斯盆地神府区块二叠系太原组煤层气实际生产资料为基础,按照“数据净化-规律挖掘-模型构建”的研究路径,开展了产能评价与预测方法研究。研究结果表明:①鄂尔多斯盆地神府区块生产井存在“高产-递减”、“上产-稳产-递减”和“排水-上产-稳产-递减”3种典型产能类型,其中“排水-上产-稳产-递减”是该区深层煤层气井的主要产能类型。②利用产能主控因素动态筛选方法,融合渗流理论中参数对产能的单调性约束机制,构建了兼顾数据规律与渗流机理的产能预测模型。③产能主控因素呈现动态演化特征,初期产能主要受地质因素控制,稳产期产能受工程因素影响显著;融合物理约束的随机森林预测模型的预测精度达86%,较传统多元线性经验公式的预测精度提高了64.5%;根据模型预测结果提出的工程优化策略可提升区块单井产量近2倍。

关键词: 煤层气, 机器学习, 数据挖掘, 随机森林预测模型, 产能预测, 太原组, 二叠系, 神府区块, 鄂尔多斯盆地

Abstract:

During the development of deep coalbed methane, significant variability exists in single-well production, which makes it difficult to predict the deliverability. Based on actual production data of deep coalbed methane from Permian Taiyuan Formation in Shenfu block of Ordos Basin, a research path of “data purification-pattern mining-model construction” was adopted to conduct deliverability evaluation and prediction methods. The results show that: (1) Three typical deliverability patterns are identified in Shenfu block of Ordos Basin: “high production-decline”, “production ramp-up-stable production-decline”, and “dewatering-production rampup-stable production-decline”, with “dewatering-production rampup-stable production-decline” being the dominant type for deep coalbed methane wells in this area. (2) Utilizing the dynamic screening method of deliverability controlling factors and integrating the monotonicity constraint mechanism of parameters on deliverability in seepage theory, a deliverability prediction model that considers both data patterns and seepage mechanisms was constructed. (3) The main controlling factors of deliverability exhibit dynamic evolution characteristics. Initial productivity is primarily controlled by geological factors, while productivity during the stable production period is significantly affected by engineering factors. The physical-constrained Random Forest prediction model achieves a prediction accuracy of 86%, which is 64.5% higher than that of conventional multivariate linear empirical formulas. Engineering optimization strategies proposed based on the model results have effectively increased the average single-well production in the block by nearly two times.

Key words: coalbed methane, machine learning, data mining, random forest prediction model, deliverability prediction, Taiyuan Formation, Permian, Shenfu block, Ordos Basin

中图分类号: 

  • TE319

图1

鄂尔多斯盆地神府区块位置(a)及二叠系太原组岩性地层综合柱状图(b)"

图2

鄂尔多斯盆地神府区块典型井二叠系太原组煤层气生产曲线"

表1

鄂尔多斯盆地神府区块二叠系太原组煤层气直井/定向井特征参数"

地质参数 范围/平均值 工程参数 范围/平均值 排采参数 范围/平均值
煤层深度/m 1 533.00 ~ 2 638.50
2 129.40
射孔厚度/m 0.6 ~ 16.4
4.9
泵径/(10-3 m) 32.00 ~ 70.00
44.60
煤层垂厚/m 4.50 ~ 27.90
13.40
射开程度/% 7.2 ~ 104.2
37.2
下泵深度/m 1 296.90 ~ 2 552.50
1 989.90
夹层垂厚/m 0 ~ 33.80
3.20
实际净液量/m3 1 193.1 ~ 6 556.9
2 471.6
冲程/m 2.40 ~ 3.95
3.20
含气量/(m3·t-1 7.80 ~ 22.12
14.14
实际砂量/m3 119.8 ~ 901.2
317.5
冲次 0.60 ~ 4.56
1.79
煤岩密度/(10-3 kg·m-3 1.27 ~ 1.65
1.42
净液强度(m3·m-1 15.5 ~ 560.0
164.8
见气时井底流压/MPa 10.66 ~ 25.87
19.93
孔隙度/% 2.91 ~ 5.54
4.23
加砂强度(m3·m-1 1.8 ~ 78.9
20.9
见气时间/d 0 ~ 132.00
26.70
渗透率/mD 0.01 ~ 0.38
0.10
返排率/% 11.3 ~ 120.2
27.7
见气前排采速度/
(10-3 MPa·d-1)
11.00 ~ 331.00
120.00
储层压力/MPa 10.10 ~ 26.40
19.10
砂比/% 4.1 ~ 22.9
11.7
地质强度指标GSI 43.85 ~ 77.52
63.40
破裂压力/MPa 26.1 ~ 36.0
30.4
储隔层应力差/MPa 0.60 ~ 7.60
2.90
混砂液量/m3 0 ~ 4 927.6
1 624.9
水平最小主应力/MPa 13.40 ~ 47.70
32.40
施工排量/m3 7.5 ~ 22.0
17.5
抗压强度/MPa 3.00 ~ 145.40
39.00
前置液量/m3 210.5 ~ 869.5
435.6
弹性模量/(103 MPa) 0 ~ 19.90
6.52
携砂液量/m3 961.3 ~ 3 864.3
1 787.1
泊松比 0.14 ~ 0.44
0.33
顶替液量/m3 20.1 ~ 132.6
43.4

图3

鄂尔多斯盆地神府区块二叠系太原组煤层气井产能特征参数选取流程"

图4

鄂尔多斯盆地神府区块二叠系太原组深层煤层气井产能影响因素相关性(据文献[28]修改) 注:实线箭头为直接影响,虚线箭头为间接影响。"

图5

鄂尔多斯盆地神府区块二叠系太原组煤层气产能影响因素间VIF特征相关性热力图"

图6

单调性知识嵌入机器学习算法原理"

图7

鄂尔多斯盆地神府区块二叠系太原组煤层气动态产能指标地质参数(a)、工程参数(b)重要性评估 注:Q30. 见气后30 d的平均产气量,m3/d;Q90. 稳产期峰值90 d的平均产气量,m3/d。"

图8

鄂尔多斯盆地神府区块二叠系太原组煤层气5种产能预测算法决定系数(a)及均方根误差(b)对比"

图9

鄂尔多斯盆地神府区块6口新部署定向井二叠系太原组煤层气预测产量与实际产量对比 注:图中的百分数均为预测产量与实际产量的相对误差,红色点代表井的产量数据,蓝色线为拟合曲线。"

表2

鄂尔多斯盆地神府区块SF-82井组生产井压裂参数及产气量(数据来源于文献[24])"

井名 加液量/m3 加砂量/m3 排量/
(m3·min-1)
90 d峰值
产气量(m3·d-1)
SF-82-1D 3 156 300.4 15 2 500
SF-82-2D 2 589 263.0 15 2 300
SF-82-3D 3 786 504.0 20 4 600
SF-82-4D 3 668 500.0 20 5 400
[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] 申有义, 王凯峰, 唐书恒, 等. 沁水盆地榆社—武乡区块二叠系煤系页岩储层地质建模及“甜点”预测[J]. 岩性油气藏, 2024, 36(4):98-108.
SHEN Youyi, WANG Kaifeng, TANG Shuheng, et al. Geological modeling and “sweet spot” prediction of Permian coal measures shale reservoirs in Yushe-Wuxiang block,Qinshui Basin[J]. Lithologic Reservoirs, 2024, 36(4):98-108.
[3] 徐长贵, 季洪泉, 王存武, 等. 鄂尔多斯盆地东缘临兴-神府区块深部煤层气富集规律与勘探对策[J]. 煤田地质与勘探, 2024, 52(8):1-11.
XU Changgui, JI Hongquan, WANG Cunwu, et al. Enrichment patterns and exploration countermeasures of deep coalbed methane in the Linxing-Shenfu block on the eastern margin of the Ordos Basin[J]. Coal Geology & Exploration, 2024, 52(8):1-11.
[4] 戴小河, 雷肖霄.我国发现千亿方深煤层气田, 储量或超1100亿立方米[N]. 新华每日电讯,2023-10-24(3).
DAI Xiaohe, LEI Xiaoxiao. China discovers hundred-billion-cubic-meter deep coalbed methane field with reserves exceeding 110 billion cubic meters[N]. Xinhua Daily Telegraph,2023-10-24(3).
[5] 安琦, 杨帆, 杨睿月, 等. 鄂尔多斯盆地神府区块深部煤层气体积压裂实践与认识[J]. 煤炭学报, 2024, 49(5):2376-2393.
AN Qi, YANG Fan, YANG Ruiyue, et al. Practice and understanding of deep coalbed methane massive hydraulic fracturing in Shenfu Block,Ordos Basin[J]. Journal of China Coal Society, 2024, 49(5):2376-2393.
[6] 朱光辉, 季洪泉, 米洪刚, 等. 神府深部煤层气大气田的发现与启示[J]. 煤田地质与勘探, 2024, 52(8):12-21.
ZHU Guanghui, JI Hongquan, MI Honggang, et al. Discovery of a large gas field of deep coalbed methane in the Shenfu block and its implications[J]. Coal Geology & Exploration, 2024, 52(8):12-21.
[7] 周德华, 陈刚, 陈贞龙, 等. 中国深层煤层气勘探开发进展、关键评价参数与前景展望[J]. 天然气工业, 2022, 42(6):43-51.
ZHOU Dehua, CHEN Gang, CHEN Zhenlong, et al. Exploration and development progress,key evaluation parameters and prospect of deep CBM in China[J]. Natural Gas Industry, 2022, 42(6):43-51.
[8] 刘建忠, 朱光辉, 刘彦成, 等. 鄂尔多斯盆地东缘深部煤层气勘探突破及未来面临的挑战与对策:以临兴—神府区块为例[J]. 石油学报, 2023, 44(11):1827-1839.
LIU Jianzhong, ZHU Guanghui, L1U Yancheng, et al. Breakthrough,future challenges and countermeasures of deep coalbed methane in the eastern margin of Ordos Basin:A case study of Linxing-Shenfu block[J]. Acta Petrolei Sinica, 2023, 44(11):1827-1839.
[9] 邓英尔, 黄润秋, 郭大浩, 等. 煤层气产量的影响因素及不稳定渗流产量预测[J]. 天然气工业, 2005, 25(1):117-119.
DENG Ying’er, HUANG Runqiu, GUO Dahao, et al. Affecting factors of coal-bed gas production and production prediction of unstable percolation[J]. Natural Gas Industry, 2005, 25(1):117-119.
[10] 张海茹, 李昊. 煤层气峰值产量拟合及产量动态预测方法研究[J]. 岩性油气藏, 2013, 25(4):116-118.
ZHANG Hairu, LI Hao. Study on coalbed methane peak production fitting and production forecast by different dynamic analysis methods[J]. Lithologic Reservoirs, 2013, 25(4):116-118.
[11] 吕玉民, 汤达祯, 李治平, 等. 煤层气井动态产能拟合与预测模型[J]. 煤炭学报, 2011, 36(9):1481-1485.
LYU Yumin, TANG Dazhen, LI Zhiping, et al. Fitting and predicting models for coalbed methane wells dynamic productivity[J]. Journal of China Coal Society, 2011, 36(9):1481-1485.
[12] 吴晓东, 王国强, 李安启, 等. 煤层气井产能预测研究[J]. 天然气工业, 2004, 24(8):82-84.
WU Xiaodong, WANG Guoqiang, LI Anqi, et al. Productivity prediction of coal-bed gas wells[J]. Natural Gas Industry, 2004, 24(8):82-84.
[13] 未志杰, 康晓东, 刘玉洋, 等. 煤层气藏全流固耦合数学模型[J]. 岩性油气藏, 2019, 31(2):151-158.
WEI Zhijie, KANG Xiaodong, LIU Yuyang, et al. A fully coupled fluid flow and geomechanics model for coalbed methane reservoir[J]. Lithologic Reservoirs, 2019, 31(2):151-158.
[14] 未志杰, 康晓东. 煤层气藏强化采收全流固耦合模型[J]. 岩性油气藏, 2021, 33(5):181-188.
WEI Zhijie, KANG Xiaodong. A fully coupled fluid flow and geomechanics model for enhanced coalbed methane recovery[J]. Lithologic Reservoirs, 2021, 33(5):181-188.
[15] 龚斌, 王虹雅, 王红娜, 等. 基于大数据分析算法的深部煤层气地质—工程一体化智能决策技术[J]. 石油学报, 2023, 44(11):1949-1958.
GONG Bin, WANG Hongya, WANG Hongna, et al. Integrated intelligent decision-making technology for deep coalbed methane geology and engineering based on big data analysis algorithms[J]. Acta Petrolei Sinica, 2023, 44(11):1949-1958.
[16] 董维强, 孟召平, 沈振, 等. 基于循环神经网络的煤层气井产气量预测方法研究[J]. 煤炭科学技术, 2021, 49(9):176-183.
DONG Weiqiang, MENG Zhaoping, SHEN Zhen, et al. Research on coalbed methane well gas production forecast method based on cyclic neural network[J]. Coal Science and Techno-logy, 2021, 49(9):176-183.
[17] 孔鹏, 李忠城, 杨晨曦, 等. 基于PCA-NB方法的煤层气井压后产量预测研究[J]. 特种油气藏, 2019, 26(6):108-112.
KONG Peng, LI Zhongcheng, YANG Chenxi, et al. Production forecast of fractured coalbed methane wells based on PCA-NB Method[J]. Special Oil & Gas Reservoirs, 2019, 26(6):108-112.
[18] 李媛, 郭大立, 康芸玮. 融合注意力机制的煤层气产量动态预测[J]. 科学技术与工程, 2023, 23(2):550-557.
LI Yuan, GUO Dali, KANG Yunwei. Dynamic prediction of coalbed methane production with attention mechanisms[J]. Science Technology and Engineering, 2023, 23(2):550-557.
[19] 冯玉龙, 周林元, 王乾, 等. 基于多层次模糊数学法的煤层气井产能综合评价模型[J]. 煤田地质与勘探, 2021, 49(2):125-132.
FENG Yulong, ZHOU Linyuan, WANG Qian, et al. Multi-level fuzzy mathematics-based comprehensive evaluation model of CBM well productivity[J]. Coal Geology & Exploration, 2021, 49(2):125-132.
[20] GUO Zixi, ZHAO Jinzhou, YOU Zhenjing, et al. Prediction of coalbed methane production based on deep learning[J]. Energy, 2021, 230:120847.
[21] 宋宣毅, 刘月田, 马晶, 等. 基于灰狼算法优化的支持向量机产能预测[J]. 岩性油气藏, 2020, 32(2):134-140.
SONG Xuanyi, LIU Yuetian, MA Jing, et al. Productivity forecast based on support vector machine optimized by grey wolf optimizer[J]. Lithologic Reservoirs, 2020, 32(2):134-140.
[22] DU Shuyi, WANG Meizhu, YANG Jiaosheng, et al. An enhanced prediction framework for coalbed methane production incorporating deep learning and transfer learning[J]. Energy, 2023, 282:128877.
[23] YANG Ruiyue, QIN Xiaozhou, LIU Wei, et al. A physics-constrained data-driven workflow for predicting coalbed methane well production using artificial neural network[J]. SPE Journal, 2022, 27(3):1531-1552.
[24] 孙立春, 刘佳, 李娜, 等. 鄂尔多斯盆地神府区块深部煤层气井产量主控因素及合理压裂规模优化[J]. 石油实验地质, 2025, 47(1):43-53.
SUN Lichun, LIU Jia, LI Na, et al. Main controlling factors of production and reasonable fracturing scale optimization of deep coalbed methane wells in Shenfu block,Ordos Basin[J]. Petroleum Geology & Experiment, 2025, 47(1):43-53.
[25] 李启晖, 任大忠, 甯波, 等. 鄂尔多斯盆地神木地区侏罗系延安组煤层微观孔隙结构特征[J]. 岩性油气藏, 2024, 36(2):76-88.
LI Qihui, REN Dazhong, NING Bo, et al. Micro-pore structure characteristics of coal seams of Jurassic Yan’an Formation in Shenmu area,Ordos Basin[J]. Lithologic Reservoirs, 2024, 36(2):76-88.
[26] ZHU Jie, ZHAO Yuhao, HU Qiujia, et al. Coalbed methane production model based on random forests optimized by a genetic algorithm[J]. ACS Omega, 2022, 7(15):13083-13094.
[27] LIU Feitony, TING Kaiming, ZHOU Zhihua. Isolation forest[R]. Pisa,2008 Eighth IEEE International Conference on Data Mining (ICDM), 2008.
[28] 闵超, 文国权, 李小刚, 等. 可解释机器学习在油气领域人工智能中的研究进展与应用展望[J]. 天然气工业, 2024, 44(9):114-126.
MIN Chao, WEN Guoquan, LI Xiaogang, et al. Research pro-gress and application prospect of interpretable machine learning in artificial intelligence of oil and gas industry[J]. Natural Gas Industry, 2024, 44(9):114-126.
[29] 吴辰文, 梁靖涵, 王伟, 等. 基于递归特征消除方法的随机森林算法[J]. 统计与决策, 2017, 33(21):60-63.
WU Chenwen, LIANG Jinghan, WANG Wei, et al. Random forest algorithm based on recursive feature elimination[J]. Statistics & Decision, 2017, 33(21):60-63.
[1] 黄力, 赵龙梅, 陈东, 赵浩阳, 张艺馨, 石石, 张稳, 季亮. 鄂尔多斯盆地宜川地区石炭系—二叠系煤岩气储层特征及勘探潜力[J]. 岩性油气藏, 2026, 38(5): 40-50.
[2] 张卫刚, 刘广林, 王勇, 宁凡, 马绍辉, 闫百泉. 鄂尔多斯盆地铁边城地区三叠系长7段夹层型页岩油储层特征及主控因素[J]. 岩性油气藏, 2026, 38(5): 94-103.
[3] 樊宇洁, 万永平, 蒲仁海, 折印楠, 崔宏俊, 高小平. 鄂尔多斯盆地延安—甘泉地区山23亚段大型河道带沉积特征及勘探潜力[J]. 岩性油气藏, 2026, 38(5): 125-135.
[4] 赵森, 陈胜, 李新豫, 杜文辉, 王秀姣, 杨昊, 李艳东, 代春萌. 基于混合随机森林的煤岩气储层孔隙度地震智能预测方法——以鄂尔多斯盆地米脂北地区石炭系本溪组8#煤为例[J]. 岩性油气藏, 2026, 38(5): 159-169.
[5] 王铮, 徐守成, 胡修权, 凌航, 涂文茂, 周月, 张小青, 孙文娜. 四川盆地元坝地区二叠系长兴组低幅生物礁沉积模式与精细刻画[J]. 岩性油气藏, 2026, 38(4): 12-22.
[6] 王敬国, 赵正威, 秦大鹏, 高磊, 白帅, 潘晓飞, 安庆, 肖富强. 三塘湖盆地侏罗系低煤阶煤层气成藏特征及勘探潜力[J]. 岩性油气藏, 2026, 38(4): 157-169.
[7] 黄巧, 罗顺社, 张亮, 黄礼, 郭康, 朱莉叶, 南军武. 塔西南柯克亚构造带二叠系棋盘组储层特征及主控因素[J]. 岩性油气藏, 2026, 38(4): 23-37.
[8] 白益远, 弓虎军, 王东旭, 李渊, 罗芬红, 王军军. 鄂尔多斯盆地延长探区氦气分布特征及控制因素[J]. 岩性油气藏, 2026, 38(3): 149-161.
[9] 聂明龙, 刘皓源, 王兆明, 赵书茂, 于太极. 哈萨克斯坦楚-萨雷苏盆地富氦气藏特征、富集条件及勘探启示[J]. 岩性油气藏, 2026, 38(3): 162-172.
[10] 任佳伟, 李莉, 王德玉, 白晓虎, 康博, 白宇恩, 白建文, 陈军斌. 页岩油藏水平井四维地应力演化及重复压裂时机优化方法——以鄂尔多斯盆地庆城油田三叠系长7段为例[J]. 岩性油气藏, 2026, 38(3): 190-200.
[11] 陈旋, 郝彬, 苟红光, 张晶, 徐雄飞, 杨占龙, 胡军, 何昌松. 吐哈盆地北缘博格达山前二叠系地层划分及其勘探意义[J]. 岩性油气藏, 2026, 38(3): 24-37.
[12] 刘若彤, 张大智, 隋立伟, 肖利梅, 田亚, 孙山, 彭丹丹, 李建智. 川中合川地区志留系龙马溪组页岩岩相测井智能识别方法[J]. 岩性油气藏, 2026, 38(3): 94-106.
[13] 薛博文, 张兆辉, 张皎生, 邹建栋, 张闻亭. 基于GWO-XGBoost模型的致密砂岩储层流体测井智能识别——以鄂尔多斯盆地洪德地区三叠系长8段为例[J]. 岩性油气藏, 2026, 38(2): 111-121.
[14] 杨占龙, 郝彬, 谭开俊, 张晶, 张丽萍, 廖建波, 李在光, 史江龙. 中国陆上中小盆地群发育特征与油气勘探方向[J]. 岩性油气藏, 2026, 38(2): 12-31.
[15] 周文娟, 蒲仁海, 卢子兴, 王康乐, 张鹏, 闻星宇, 王通, 关蕴文. 鄂尔多斯盆地中东部奥陶系马五段膏盐岩分布特征及主控因素[J]. 岩性油气藏, 2026, 38(2): 122-133.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!