基于机器学习的地层三压力钻前预测方法

付群超; 万肖箭; 陈沛; 刘挺; 刘宏坤; 李志

doi:10.11911/syztjs.2025062

基于机器学习的地层三压力钻前预测方法

付群超^1,,
万肖箭^2, ,,
陈沛³,
刘挺¹,
刘宏坤¹,
李志²

1.
中法渤海地质服务有限公司湛江分公司，广东湛江 524000
2.
南方海洋科学与工程广东省实验室（湛江），广东湛江 524000
3.
中海石油（中国）有限公司湛江分公司，广东湛江 524000

详细信息

作者简介:
付群超（1983—），男，河南南阳人，2009年毕业于中国石油大学（华东）石油工程专业，2020年获西南石油大学石油与天然气工程专业硕士学位，工程师，主要从事海洋石油录井随钻跟踪工作。E-mail：fuqc@cfbgc.com

通讯作者:
万肖箭，wanxiaojian@zjblab.com

中图分类号: TE242
计量
- 文章访问数: 24
- HTML全文浏览量: 3
- PDF下载量: 11
出版历程
- 收稿日期: 2024-02-09
- 修回日期: 2025-05-10
- 录用日期: 2025-06-01
- 网络出版日期: 2025-06-05
- 刊出日期: 2025-06-27

A Pre-Drilling Prediction Method of Formation Tri-Pressures Based on Machine Learning

1.
Zhanjiang Branch, China France Bohai Geoservices Company Limited, Zhanjiang, Guangdong, 524000, China
2.
Southern Marine Science and Engineering Guangdong Laboratory(Zhanjiang), Zhanjiang, Guangdong, 524000, China
3.
Zhanjiang Branch, CNOOC China Limited, Zhanjiang, Guangdong, 524000, China;

摘要

摘要:
海洋深层、超深层油气勘探采用传统的地层三压力剖面计算模型时存在预测可信度低、计算效率不足等问题。为此，基于机器学习方法，以深度神经网络模型为主体架构，引入岩石力学物理模型作为物理约束条件，通过在损失函数中联合优化数据拟合误差与物理残差项，实现物理约束的数学嵌入，同时考虑地质、测井、录井及现场工况等多源数据，建立了物理−数据双驱动的地层三压力钻前预测模型，并采用均方根误差（RMSE）、平均绝对误差（MAE）和决定系数（R²）进行模型评估。该模型在北部湾涠西南凹陷乌石区块进行了现场应用，结果显示，地层三压力钻前预测结果与实际差异较小，模型具有较高的预测精度，预测结果符合现场应用效果。基于机器学习的地层三压力钻前预测方法可以为深层、超深层安全高效钻进提供技术支持。
- 测井 /
- 岩石力学 /
- 机器学习 /
- 地层三压力 /
- 钻前预测
Abstract:
In response to the issues of low prediction reliability and insufficient computational efficiency associated with conventional models in calculating formation tri-pressures for offshore deep and ultra-deep oil and gas exploration, a machine learning-based method was used. In the method, a deep neural network (DNN) model was used as the main structure, and rock mechanics physical models were integrated as physical constraints. By jointly optimizing data-fitting errors and physical residual terms in the loss function, the mathematical embedding of the physical constraints was achieved. Additionally, it l incorporated multi-source data, including geological, logging, mud logging , and field operational data, to establish a physics − data-driven formation tri-pressure prediction model. Model performance was evaluated using root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²). The model was applied in the Wushi Block of the South Weixi Depression in the Beibuwan Basin. The results demonstrate minimal discrepancies between predicted and actual formation tri-pressure values, indicating high prediction accuracy, with results consistent with actual application. This machine learning-based pre-drilling prediction method of formation tri-pressures can provide technical support for safe and efficient drilling in deep and ultra-deep formations.
- logging /
- rock mechanics /
- machine learning /
- formation tri-pressure /
- pre-drilling prediction

HTML全文

图 1 待预测井钻前数据集

Figure 1. Pre-drilling input data for the well to be predicted

下载: 全尺寸图片幻灯片

图 2 深度神经网络机器学习模型结构

Figure 2. Structure of machine learning-based DNN model

下载: 全尺寸图片幻灯片

图 3 损失函数随迭代次数的变化

Figure 3. Variation of loss function with iteration times

下载: 全尺寸图片幻灯片

图 4 邻井三压力曲线计算结果

Figure 4. Calculation results of tri-pressure curves for adjacent wells

下载: 全尺寸图片幻灯片

图 5 不同模型计算的地层三压力结果

Figure 5. Formation tri-pressure calculation results by various models

下载: 全尺寸图片幻灯片

图 6 W7井不同模型模拟破坏效果

Figure 6. Failure simulation of various models for Well W7

下载: 全尺寸图片幻灯片

表 1 模型评价结果

Table 1 Model evaluation results

数据集	RMSE/MPa	MAE/MPa	R²
训练集	2.1	1.6	0.93
验证集	2.5	1.9	0.89
测试集	2.7	2.1	0.86

下载: 导出CSV

表 2 W7井钻前建议钻井液密度与钻后实际钻井液密度对比

Table 2 Comparison of suggested drilling fluid density before drilling and actual drilling fluid density after drilling for Well W7

井深/m	实际钻井液密度/(kg·L⁻¹）	建议钻井液密度/(kg·L⁻¹）	应用效果
1 900～3 165	1.31	1.25～1.35	钻井液密度合理
3 165～3 417	1.31～1.39	1.25～1.35	钻井液密度合理
3 417～3 860	1.39～1.55	1.35～1.50	发生漏失，实际钻井液密度过高
3 860～4 535	1.51～1.55	1.50～1.54	钻井液密度合理

下载: 导出CSV

参考文献(26)

[1]	李战奎,吴立伟,郭明宇,等. 渤中凹陷深层高压井地质工程一体化技术研究与应用[J]. 石油钻探技术,2024,52(2):194–201. doi: 10.11911/syztjs.2024031 LI Zhankui, WU Liwei, GUO Mingyu, et al. Research and application of integrated geological engineering technology for deep high-pressure wells in the Bozhong Sag[J]. Petroleum Drilling Techniques, 2024, 52(2): 194–201. doi: 10.11911/syztjs.2024031
[2]	李凡,李大奇,金军斌,等. 顺北油气田辉绿岩地层井壁稳定钻井液技术[J]. 石油钻探技术,2023,51(2):61–67. doi: 10.11911/syztjs.2022041 LI Fan, LI Daqi, JIN Junbin, et al. Drilling fluid technology for wellbore stability of the diabase formation in Shunbei Oil & Gas Field[J]. Petroleum Drilling Techniques, 2023, 51(2): 61–67. doi: 10.11911/syztjs.2022041
[3]	张明明,李大奇,范翔宇. 破碎地层井壁坍塌压力及井周失稳区域研究[J]. 断块油气田,2024,31(5):916–921. ZHANG Mingming, LI Daqi, FAN Xiangyu. Study on wellbore collapse pressure and instability area around wellbore in broken formation[J]. Fault-Block Oil & Gas Field, 2024, 31(5): 916–921.
[4]	KING HUBBERT M, WILLIS D G. Mechanics of hydraulic fracturing[J]. Transactions of the AIME, 1957, 210(1): 153–166. doi: 10.2118/686-G
[5]	金衍,陈勉,柳贡慧,等. 大位移井的井壁稳定力学分析[J]. 地质力学学报,1999,5(1):4–11. doi: 10.3969/j.issn.1006-6616.1999.01.002 JIN Yan, CHEN Mian, LIU Gonghui, et al. Wellbore stability analysis of extended reach wells[J]. Journal of Geomechanics, 1999, 5(1): 4–11. doi: 10.3969/j.issn.1006-6616.1999.01.002
[6]	卢运虎,陈勉,袁建波,等. 各向异性地层中斜井井壁失稳机理[J]. 石油学报,2013,34(3):563–568. doi: 10.7623/syxb201303022 LU Yunhu, CHEN Mian, YUAN Jianbo, et al. Borehole instability mechanism of a deviated well in anisotropic formations[J]. Acta Petrolei Sinica, 2013, 34(3): 563–568. doi: 10.7623/syxb201303022
[7]	徐声驰,刘锐,孟鑫,等. 基于井眼坍塌角度和坍塌深度预测模型的泥岩水平段井壁稳定性评价方法[J]. 石油钻采工艺,2023,45(2):136–142. XU Shengchi, LIU Rui, MENG Xin, et al. Wellbore stability evaluation of horizontal wellbore in mudstone: a method based on wellbore collapse angle and depth model[J]. Oil Drilling & Production Technology, 2023, 45(2): 136–142.
[8]	耿殿栋. 泥页岩储层岩石力学特性及多状态下井壁稳定性研究[D]. 荆州:长江大学,2024. GENG Diandong. Study on the rock mechanics characteristics of mudstone-shale reservoirs and wellbore stability under multiple conditions[D]. Jingzhou: Yangtze University, 2024.
[9]	曹文科,邓金根,蔚宝华,等. 基于多孔介质热弹性理论的井壁诱导缝成因[J]. 天然气工业,2017,37(6):79–85. doi: 10.3787/j.issn.1000-0976.2017.06.011 CAO Wenke, DENG Jingen, YU Baohua, et al. Genesis of induced fractures on borehole walls based on the thermo-poroelasticity theory[J]. Natural Gas Industry, 2017, 37(6): 79–85. doi: 10.3787/j.issn.1000-0976.2017.06.011
[10]	韦世明,夏阳,陈勉,等. 超深层碳酸盐岩储层孔隙弹性动力学起裂规律[J]. 石油钻采工艺,2020,42(2):127–132. WEI Shiming, XIA Yang, CHEN Mian, et al. The fracture initiation laws of ultradeep carbonate reservoirs based on poroelastic dynamics[J]. Oil Drilling & Production Technology, 2020, 42(2): 127–132.
[11]	刘海龙,许杰,谢涛,等. 渤海中深层井壁稳定流固耦合研究[J]. 石油机械,2019,47(4):1–7. LIU Hailong, XU Jie, XIE Tao, et al. Study on fluid-solid coupling for borehole wall stability of medium-deep wellbore in Bohai Sea[J]. China Petroleum Machinery, 2019, 47(4): 1–7.
[12]	李高仁,史亚红,夏宏泉,等. 基于Mogi-Coulomb强度准则的井壁稳定性力学分析新方法[J]. 中国安全生产科学技术,2018,14(10):70–75. LI Gaoren, SHI Yahong, XIA Hongquan, et al. New method for mechanical analysis of wellbore stability based on Mogi-Coulomb strength criterion[J]. Journal of Safety Science and Technology, 2018, 14(10): 70–75.
[13]	陈颖杰,邓传光,马天寿. 井壁失稳风险的可靠度理论评价方法[J]. 天然气工业,2019,39(11):97–104. doi: 10.3787/j.issn.1000-0976.2019.11.013 CHEN Yingjie, DENG Chuanguang, MA Tianshou. A risk assessment method of wellbore instability based on the reliability theory[J]. Natural Gas Industry, 2019, 39(11): 97–104. doi: 10.3787/j.issn.1000-0976.2019.11.013
[14]	韩正波,刘厚彬,张靖涛,等. 深层脆性页岩力学性能及井壁稳定性研究[J]. 特种油气藏,2020,27(5):167–174. doi: 10.3969/j.issn.1006-6535.2020.05.026 HAN Zhengbo, LIU Houbin, ZHANG Jingtao, et al. Research on the mechanical properties and borehole stability of deep brittle shale[J]. Special Oil & Gas Reservoirs, 2020, 27(5): 167–174. doi: 10.3969/j.issn.1006-6535.2020.05.026
[15]	马天寿,张东洋,杨赟,等. 基于机器学习模型的斜井坍塌压力预测方法[J]. 天然气工业,2023,43(9):119–131. doi: 10.3787/j.issn.1000-0976.2023.09.012 MA Tianshou, ZHANG Dongyang, YANG Yun, et al. Machine learning model based collapse pressure prediction method for inclined wells[J]. Natural Gas Industry, 2023, 43(9): 119–131. doi: 10.3787/j.issn.1000-0976.2023.09.012
[16]	彭超,吴立伟,李战奎,等. 多元录井参数随钻定量监测方法提高生烃超压地层压力监测效率[J]. 石油钻采工艺,2024,46(5):549–561. PENG Chao, WU Liwei, LI Zhankui, et al. Multi dimensional logging parameter quantitative monitoring method while drilling improves the efficiency of pressure monitoring in hydrocarbon generating over-pressure formations[J]. Oil Drilling & Production Technology, 2024, 46(5): 549–561.
[17]	马天寿,向国富,石榆帆,等. 基于双向长短期记忆神经网络的水平地应力预测方法[J]. 石油科学通报,2022,7(4):487–504. MA Tianshou, XIANG Guofu, SHI Yufan, et al. Horizontal in-situ stress prediction method based on the bidirectional long short-term memory neural network[J]. Petroleum Science Bulletin, 2022, 7(4): 487–504.
[18]	赵军,李勇,文晓峰,等. 基于斑马算法优化支持向量回归机模型预测页岩地层压力[J]. 岩性油气藏,2024,36(6):12–22. doi: 10.12108/yxyqc.20240602 ZHAO Jun, LI Yong, WEN Xiaofeng, et al. Prediction of shale formation pore pressure based on Zebra Optimization Algorithm-optimized support vector regression[J]. Lithologic Reservoirs, 2024, 36(6): 12–22. doi: 10.12108/yxyqc.20240602
[19]	曾义金,李大奇,陈曾伟,等. 基于自然语言处理与大数据分析的漏失分析与诊断[J]. 石油钻探技术,2023,51(6):1–11. doi: 10.11911/syztjs.2023069 ZENG Yijin, LI Daqi, CHEN Zengwei, et al. Loss analysis and diagnosis based on natural language processing and big data analysis[J]. Petroleum Drilling Techniques, 2023, 51(6): 1–11. doi: 10.11911/syztjs.2023069
[20]	AHMED S A, MAHMOUD A A, ELKATATNY S, et al. Prediction of pore and fracture pressures using support vector machine[R]. IPTC 19523, 2019.
[21]	贾利春,李柱正,陈丽萍. 基于有效应力的裂缝性碳酸盐岩地层孔隙压力预测[J]. 钻采工艺,2023,46(5):93–99. doi: 10.3969/J.ISSN.1006-768X.2023.05.15 JIA Lichun, LI Zhuzheng, CHEN Liping. Pore pressure prediction of fractured carbonate formation in central Sichuan based on effective stress principle[J]. Drilling & Production Technology, 2023, 46(5): 93–99. doi: 10.3969/J.ISSN.1006-768X.2023.05.15
[22]	EATON B A. The effect of overburden stress on geopressure prediction from well logs[J]. Journal of Petroleum Technology, 1972, 24(8): 929–934. doi: 10.2118/3719-PA
[23]	邓金根,陈峥嵘,耿亚楠,等. 页岩储层地应力预测模型的建立和求解[J]. 中国石油大学学报(自然科学版),2013,37(6):59–64. DENG Jingen, CHEN Zhengrong, GENG Yanan, et al. Prediction model for in-situ formation stress in shale reservoirs[J]. Journal of China University of Petroleum(Edition of Natural Science), 2013, 37(6): 59–64.
[24]	安康. 超深层碳酸盐岩力学实验及井壁坍塌压力研究[D]. 北京:中国地质大学(北京),2021. AN Kang. Mechanical experiment of ultra-deep carbonate rock and research on collapse pressure of well. [D]. Beijing: China University of Geosciences(Beijing), 2021.
[25]	蔡文军,李中,殷志明,等. 基于地质工程一体化的裂缝性地层漏失压力预测[J]. 断块油气田,2024,31(4):676–683. CAI Wenjun, LI Zhong, YIN Zhiming, et al. Prediction of leakage pressure in fractured formation based on integration of geology and engineering[J]. Fault-Block Oil & Gas Field, 2024, 31(4): 676–683.
[26]	幸雪松,袁俊亮,李忠慧,等. 南海深水高温高压条件下地层破裂压力的确定[J]. 石油钻探技术,2023,51(6):18–24. doi: 10.11911/syztjs.2023052 XING Xuesong, YUAN Junliang, LI Zhonghui, et al. Determination of formation fracture pressure under high temperature and high pressure in deep water of the South China Sea[J]. Petroleum Drilling Techniques, 2023, 51(6): 18–24. doi: 10.11911/syztjs.2023052

施引文献(13)

期刊类型引用(8)

1.	陈允禄，李玮健，冯伟风. 净浆搅拌转速对混凝土性能的影响. 建材世界. 2024(01): 30-33 . 百度学术
2.	蒋祥光，许明标. 纳米氧化铝对油井水泥基复合材料力学性能的影响. 当代化工. 2023(05): 1252-1255 . 百度学术
3.	唐凯，陈小荣，李志宏. 表层套管固井低剪切速率对水泥浆性能的影响——以厄瓜多尔安第斯区块为例. 石油地质与工程. 2023(06): 85-89+96 . 百度学术
4.	宋建建，许明标，王晓亮，张敏，胡顺，杜佳琪. 胶乳粉固井水泥浆体系研究与应用. 油田化学. 2021(03): 406-411 . 百度学术
5.	徐力群，张兴国，王银东，金也，丁辉，刘增，刘开强，郭小阳. 低返速固井对油井水泥浆性能的影响. 钻井液与完井液. 2019(01): 70-76 . 百度学术
6.	寇云鹏，齐兆军，宋泽普，杜加法，杨纪光. 全尾砂高浓度充填料浆流变特性试验研究. 矿业研究与开发. 2018(12): 32-35 . 百度学术
7.	李建山. 泾河油田水平井固井难点与对策研究. 石油钻探技术. 2017(06): 19-23 . 本站查看
8.	刘振通，党冬红，和建勇，王莹，宋元洪，郭文猛，张玉鹏，王洪峰. 委内瑞拉低压高渗漏地层小间隙尾管固井技术. 钻井液与完井液. 2017(06): 83-88 . 百度学术