基于机器学习的页岩气井井间干扰评价及预测

doi:10.13809/j.cnki.cn32-1825/te.2022.03.011

摘要/Abstract

摘要：

页岩气藏井间干扰严重制约气井生产,井间干扰程度评价与预测对页岩气高效开发具有重要意义。现有研究主要聚焦页岩气井间干扰现象、生产动态特征以及数值模拟参数优化等方面,但页岩气井间干扰程度定量评价及预测方面的研究较少,且参数体系不全,难以客观评价页岩气井间干扰程度。因此,采用机器学习方法综合考虑地质参数、压裂参数及生产参数,对A页岩气藏井间干扰程度进行评价及预测。先对初始数据进行数据处理,提高数据质量,然后基于处理后的数据,应用聚类分析及随机森林算法评价及预测Y页岩气井间干扰程度。结果表明：A页岩气藏中井间干扰程度低、中、高的井数占比分别为25.93 %、37.03 %、37.04 %,其中压裂因素对A页岩气藏井间干扰程度评价结果影响最大。调参后的页岩气井间干扰程度预测结果达到92.07 %,表明所建立的预测模型可应用于实际页岩气井间干扰程度预测,且模型精确度较高,为页岩气井井间干扰量化评价及预测提供了一种有效手段。

关键词: 页岩气, 井间干扰程度评价, 井间干扰程度预测, 机器学习, K-Means, 随机森林

Abstract:

Inter-well interference seriously affects the production of shale gas wells. The evaluation and prediction of well interference degree is of great significance to the efficient development of shale gas. But the existing research mainly focuses on the interference phenomenon between shale gas wells, production performance, and parameter optimization through numerical simulation. There are few studies on the quantitative evaluation and prediction of the interference degree between shale gas wells, and the selected parameters is incomplete, which makes it difficult to objectively evaluate the well interference between shale gas wells. Therefore, the machine learning method is used to comprehensively consider the geological parameters and fracturing parameters to evaluate and predict the degree of interference between wells in the shale gas reservoir. Firstly, the initial data are processed to improve the data quality. Then, based on the processed data, cluster analysis and random forest algorithm are used to evaluate and predict the interference degree of shale gas wells. The results show that the proportions of the wells with low, medium and high well interference in the shale gas reservoirs are 25.93 %, 37.03 % and 37.04 %, respectively. The fracturing factors show significant influence on the well interference degree in the shale gas reservoirs. After parameters optimization, the prediction results of well interference degree reaches 92.07 %, indicating that the developed prediction model can be applied to forecast the well interference degree in shale gas reservoirs.

Key words: shale gas, well interference evaluation, well interference prediction, machine learning, K-Means, random forest

中图分类号:

TE37

张庆,何封,何佑伟. 基于机器学习的页岩气井井间干扰评价及预测[J]. 油气藏评价与开发, 2022, 12(3): 487-495.

Qing ZHANG,Feng HE,Youwei HE. Well interference evaluation and prediction of shale gas wells based on machine learning[J]. Petroleum Reservoir Evaluation and Development, 2022, 12(3): 487-495.

图/表 11

表1

图1

表2

图 2

图3

图 4

表3

图5

图6

表4

图7

参考文献 26

[1]	董大忠, 施振生, 管全中, 等. 四川盆地五峰组—龙马溪组页岩气勘探进展、挑战与前景[J]. 天然气工业, 2018, 38(4):67-76.
	DONG Dazhong, SHI Zhensheng, GUAN Quanzhong, et al. Progress, challenges and prospects of shale gas exploration in the Wufeng-Longmaxi reservoirs in the Sichuan Basin[J]. Natural Gas Industry, 2018, 38(4): 67-76.
[2]	胡凯. 川西南威远地区五峰—龙马溪组页岩储层特征及“甜点”分布规律研究[J]. 非常规油气, 2021, 8(5):34-44.
	HU Kai. Reservoir and sweet pot distribution characteristics of shale gas in Wufeng-Longmaxi Formation, southwest of Sichuan Basin[J]. Unconventional Oil & Gas, 2021, 8(5):34-44.
[3]	辛翠平, 白慧芳, 张磊, 等. 不同裂缝形态页岩气多级压裂水平井产能预测模型应用研究[J]. 非常规油气, 2020, 7(3):65-71.
	XIN Cuiping, BAI Huifang, ZHANG Lei, et al. Application study of multistage fracturing horizontal well production forecasting models for shale gas with different fracture forms[J]. Unconventional Oil & Gas, 2020, 7(3): 65-71.
[4]	GUO X Y, WU K, KILLOUGH J, et al. Understanding the mechanism of interwell fracturing interference with reservoir/geomechanics/fracturing modeling in eagle ford shale[J]. SPE Reservoir Evaluation & Engineering, 2019, 22(3):842-860.
[5]	郭旭洋, 金衍, 黄雷, 等. 页岩油气藏水平井井间干扰研究现状和讨论[J]. 石油钻采工艺, 2021, 43(3):348-367.
	GUO Xuyang, JIN Yan, HUANG Lei, et al. Review and discussion of the study of interwell interference in shale oil and shale gas reservoirs[J]. Oil Drilling & Production Technology, 2021, 43(3): 348-367.
[6]	陈京元, 位云生, 王军磊, 等. 页岩气井间干扰分析及井距优化[J]. 天然气地球科学, 2021, 32(7):931-940.
	CHEN Jingyuan, WEI Yunsheng, WANG Junlei, et al. Interwell-production interference and well spacing optimization in shale gas reservoir[J]. Natural Gas Geoscience, 2021, 32(7): 931-940.
[7]	曾庆磊, 庄茁, 柳占立, 等. 页岩水力压裂中多簇裂缝扩展的全耦合模拟[J]. 计算力学学报, 2016, 33(4):643-648.
	ZENG Qinglei, ZHUANG Zhuo, LIU Zhanli, et al. Fully coupled simulation of multi-cluster fracture propagation in shale hydraulic fracturing[J]. Chinese Journal of Computational Mechanics, 2016, 33(4): 643-648.
[8]	SARDINHA C, PETR C, LEHMANN J, et al. Determining interwell connectivity and reservoir complexity through frac pressure hits and production interference analysis[C]// Paper SPE-171628-MS presented at the SPE/CSUR Unconventional Resources Conference-Canada, Calgary, Alberta, Canada, September 2014.
[9]	SULLIVAN M, ZANGANEH B, SPRINGER A, et al. Post-fracture pressure decay: A novel (and free) stage-level assessment method[C]// Paper URTEC-2019-970-MS presented at the SPE/AAPG/SEG Unconventional Resources Technology Conference, Denver, Colorado, USA, July 2019.
[10]	KUMAR A, SETH P, SHRIVASTAVA K, et al. Well interference diagnosis through integrated analysis of tracer and pressure interference tests[C]// Paper URTEC-2901827-MS presented at the SPE/AAPG/SEG Unconventional Resources Technology Conference, Houston, Texas, USA, July 2018.
[11]	JI L J, SEN V, MIN K S, et al. Numerical simulation of DFITs within a coupled reservoir flow and geomechanical simulator-insights into completions optimization[C]// Paper SPE-194352-MS presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, The Woodlands, Texas, USA, February 2019.
[12]	王磊, 杨春和, 郭印同, 等. 基于室内水力压裂试验的水平井起裂模式研究[J]. 岩石力学与工程学报, 2015, 34(S2):3624-3632.
	WANG Lei, YANG Chunhe, GUO Yintong, et al. Investigation on fracture initiation modes of horizontal wells based on laboratory hydraulic fracturing test[J]. Chinese Journal of Rock Mechanics and Engineering, 2015, 34(S2): 3624-3632.
[13]	王军磊, 贾爱林, 位云生, 等. 基于多井模型的压裂参数—开发井距系统优化[J]. 石油勘探与开发, 2019, 46(5):981-992.
	WANG Junlei, JIA Ailin, WEI Yunsheng, et al. Optimization work-flow for stimulation-well spacing designs in a multiwell pad[J]. Petroleum Exploration and Development, 2019, 46(5): 981-992.
[14]	雍锐, 常程, 张德良, 等. 地质—工程—经济一体化页岩气水平井井距优化——以国家级页岩气开发示范区宁209井区为例[J]. 天然气工业, 2020, 40(7):42-48.
	YONG Rui, CHANG Cheng, ZHANG Deliang, et al. Optimization of shale-gas horizontal well spacing based on geology-engineering-economy integration: A case study of Well Block Ning 209 in the National Shale Gas Development Demonstration Area[J]. Natural Gas Industry, 2020, 40(7): 42-48.
[15]	HE Y W, GUO J C, TANG Y, et al. Interwell fracturing interference evaluation of multi-well pads in shale gas reservoirs: A case study in WY Basin[C]// Paper SPE-201694-MS presented at the SPE Annual Technical Conference and Exhibition, Virtual, October 2020.
[16]	位云生, 王军磊, 齐亚东, 等. 页岩气井网井距优化[J]. 天然气工业, 2018, 38(4):129-137.
	WEI Yunsheng, WANG Junlei, QI Yadong, et al. Optimization of shale gas well pattern and spacing[J]. Natural Gas Industry, 2018, 38(4): 129-137.
[17]	李维, 代锋, 左星. 存在井间干扰的页岩气井精细控压技术应用[J]. 钻采工艺, 2019, 42(5):103-105.
	LI Wei, DAI Feng, ZUO Xing. Application of fine pressure control technology in shale gas wells with inter-well interference[J]. Drilling & Production Technology, 2019, 42(5): 103-105.
[18]	钟思存, 何嘉, 赵素惠, 等. 基于概率模拟方法的页岩气井间干扰影响研究[C]// 第31届全国天然气学术年会(2019). 合肥: 中国石油学会天然气专业委员会, 2019:6.
	ZHONG Sicun, HE Jia, ZHAO Suhui, et al. Research on well interference of shale gas based on probability simulation[C]// The 31st National Natural Gas Academic Annual Conference (2019). Hefei: Institute of Petroleum and Gas Professional Committee of China, 2019: 6.
[19]	严子铭, 王涛, 柳占立, 等. 基于机器学习的页岩气采收率预测方法[J]. 固体力学学报, 2021, 42(3):221-232.
	YAN Ziming, WANG Tao, LIU Zhanli, et al. Machine-learning-based prediction methods on shale gas recovery[J]. Chinese Journal of Solid Mechanics, 2021, 42(3): 221-232.
[20]	钱辰, 杨少春, 许子君. 基于机器学习的页岩气“甜点”评价及其应用综述[C]// 油气田勘探与开发国际会议论文集. 西安: 西安石油大学, 2019:11.
	QIAN Chen, YANG Shaochun, XU Zijun. Review of evaluation of shale gas sweet spots and its application based on machine learning[C]// International Field Exploration and Development Conference. Xi’an: Xi’an Shiyou University, 2019:11.
[21]	李菊花, 陈晨, 肖佳林. 基于随机森林算法的页岩气多段压裂井产量预测[J]. 长江大学学报(自然科学版), 2020, 17(4):34-38.
	LI Juhua, CHEN Chen, XIAO Jialin. Yield production of shale gas multi-stage fracturing wells based on random forest algorithm[J]. Journal of Yangtze University (Natural Science Edition), 2020, 17(4): 34-38.
[22]	孙艺涵. 基于机器学习的页岩有机质含量预测方法研究[D]. 北京: 中国石油大学(北京), 2019.
	SUN Yihan. Research on prediction method of total organic carbon in shale based on machine learning[D]. Beijing:China University of Petroleum(Beijing), 2019.
[23]	周小金, 杨洪志, 范宇, 等. 川南页岩气水平井井间干扰影响因素分析[J]. 中国石油勘探, 2021, 26(2):103-112.
	ZHOU Xiaojin, YANG Hongzhi, FAN Yu, et al. Analysis of factors affecting frac hits in horizontal shale gas wells in the southern Sichuan Basin[J]. China Petroleum Exploration, 2021, 26(2): 103-112.
[24]	黄红良. 多井条件下页岩气水平井产能影响因素研究[D]. 北京: 中国石油大学(北京), 2017.
	HUANG Hongliang. Study on influence factors of hydraulic fractured horizontal well productivity in shale gas reservoir under multi-well condition[D]. Beijing: China University of Petroleum(Beijing), 2017.
[25]	苏晓眉, 张涛, 李玉飞, 等. 基于K-Means聚类算法的沉砂卡钻预测方法研究[J]. 钻采工艺, 2021, 44(3):5-9.
	SU Xiaomei, ZHANG Tao, LI Yufei, et al. Research on the sticking prediction method based on K-Means clustering algorithm[J]. Drilling & Production Technology, 2021, 44(3): 5-9.
[26]	王光宇, 宋建国, 徐飞, 等. 不平衡样本集随机森林岩性预测方法[J]. 石油地球物理勘探, 2021, 56(4):679-687.
	WANG Gangyu, SONG Jianguo, XU Fei, et al. Random forest lithology prediction method based on unbalanced sample set[J]. Oil Geophysical Prospecting, 2021, 56(4): 679-687.

井号		总含气量（m³/t）	压裂段数		压裂簇数	改造体积（10⁴m³）		水平段长（m）		压裂段长（m）		入地液量（m³）		入地砂量（t）
1		4.04	28		77	3 847.69		1 500		1 686		54 411.00		2 442.40
2		3.50	12		38	1 974.63		1 786		747		22 100.12		2 815.57
3		5.51	15		41	4 801.09		1 500		1 087		29 988.88		2 790.85
4		3.90	22		64	4 760.67		1 800		1 689		38 072.96		2 784.36
5		6.22	33		99	8 188.04		2 200		2 154		65 063.00		3 674.25
6		4.93	30		93	5 848.50		2 000		1 972		54 017.06		3 328.43
7		4.85	31		99	5 041.82		2 000		1 974		53 143.52		3 302.30
8		4.53	31		93	5 947.01		2 000		1 970		52 443.90		2 860.20
井号	渗透率（10^-6μm²）		孔隙度	平均累产气量（m³/d）			最小水平主应力（MPa）		脆性矿物含量（%）		黏土矿物含量（%）		井间干扰影响程度（%）
1	0.28		6.09	73 553.62			71.10		64.75		19.82		84
2	0.30		6.11	36 020.87			68.94		69.52		15.85		85
3	0.28		5.60	31 608.30			71.50		55.42		10.47		100
4	0.29		5.70	128 536.00			68.10		59.80		20.60		79
5	0.29		5.83	262 288.20			70.36		74.52		22.25		91
6	0.29		6.03	350 960.10			67.70		68.01		20.55		47
7	0.48		6.20	221 543.30			69.65		71.14		18.40		89
8	0.29		6.08	30 792.40			69.45		72.63		19.70		54

参数	Y₁	Y₂	Y₃	Y₄	Y₅	Y₆	Y₇	Y₈
平均累产气量	0.320 333	0.101 433	-0.332 990	-0.100 060	0.478 627	-0.050 270	0.037 667	0.424 910
压裂段数	0.416 403	-0.080 640	0.124 939	0.171 114	-0.110 600	-0.060 660	0.400 095	-0.017 620
压裂级数	0.440 225	-0.041 560	0.351 620	0.115 790	-0.277 470	-0.030 570	0.221 530	0.104 619
改造体积	0.178 355	-0.244 070	-0.127 860	0.073 144	0.357 113	-0.112 400	0.068 482	-0.548 110
水平段长	0.303 558	0.022 013	-0.562 230	-0.098 540	-0.323 940	0.393 068	-0.054 400	0.126 982
压裂段长	0.360 660	0.010 346	-0.035 770	0.138 368	0.047 098	-0.041 400	0.114 013	-0.306 540
入地液量	0.346 230	-0.327 860	0.342 088	-0.108 870	0.159 753	0.127 573	-0.223 240	0.236 313
入地砂量	0.105 936	-0.166 550	0.264 322	-0.236 370	0.329 727	0.285 875	-0.435 300	-0.174 210
渗透率	0.104 210	0.208 533	-0.005 220	-0.097 980	0.029 974	0.608 051	0.034 170	-0.030 270
孔隙度	0.131 329	0.788 948	0.290 244	-0.313 310	0.150 284	-0.102 290	0.091 473	-0.083 900
总含气量	0.145 061	0.322 258	-0.174 330	0.677 641	0.194 926	-0.020 280	-0.347 040	-0.104 310
最小水平主应力	-0.064 370	0.070 389	0.306 970	0.459 899	-0.086 270	0.114 290	-0.257 370	0.330 922
脆性矿物	0.223 720	0.101 532	-0.017 020	-0.169 200	-0.493 420	-0.128 880	-0.495 090	-0.329 260
黏土矿物	0.204 261	-0.032 390	-0.141 630	-0.195 640	-0.019 610	-0.561 910	-0.296 870	0.273 371

等级	Y₁	Y₂	Y₃	Y₄	Y₅	Y₆	Y₇	Y₈	井间干扰程度
等级低	0.585 049	0.218 316	-0.175 240	0.034 695	0.071 724	0.052 852	-0.004 221	-0.034 615	0.474 394
等级中	-0.644 967	-0.055 477	-0.055 778	0.016 812	0.055 988	0.035 766	-0.016 125	0.004 557	0.769 811
等级高	0.235 432	-0.097 344	0.178 446	-0.041 099	-0.106 195	-0.072 763	0.019 169	0.019 674	0.905 660

参数	最优值	参数	最优值
Random state	70	中间节点分枝所需最小样本数	6
树模型数量	41	分枝时考虑最大特征数	2