Prediction of Tight Sandstone Reservoir Permeability Using NMR Logging Data Based on Deep Learning Algorithms: Wubao-Jiaxian Exploration Area Benxi Formation
1 College of Geology and Environment, Xi’an University of Science and Technology, Xi’an 710054, China
2 Key Laboratory of Coal Resources Exploration and Comprehensive Utilization, Ministry of Natural Resources, Xi’an 710021, China
Show less
文章历史+
Received
Published
Revised
2025-07-21
2026-07-28
2026-02-06
Online Date
2026-08-14
PDF (10220K)
摘要
致密砂岩储层孔隙结构复杂、非均质性强,呈现低孔低渗且渗透率变化大的特征,开展致密砂岩储层渗透率的测井定量预测,是关乎致密砂岩储层高效开发的关键一环。核磁共振测井因能提供与岩石成分无关的孔隙度信息而成为有效的渗透率预测手段,但现有的SDR模型(Schlumberger-Doll research)和Coates两种核磁共振(nuclear magnetic resonance,NMR)测井渗透率求解模型存在一定局限性。为此,基于Coates模型的关键参数,以鄂尔多斯盆地吴堡-佳县勘探区四口参数井作为数据集,构建了深度神经网络模型(deep neural network,DNN)和长短期记忆网络模型(long short-term memory,LSTM)核磁共振测井进行渗透率预测。实验表明,DNN-NMR模型在静态参数拟合中表现良好(R2=0.872),较Coates模型准确率提高14%;LSTM-NMR模型表现最优(R2=0.916)。研究结果为致密砂岩储层渗透率的核磁共振测井定量预测提供了高精度、强泛化的解决方案。
Abstract
Tight sandstone reservoirs are characterized by complex pore structures, strong heterogeneity, low porosity, low permeability, and highly variable permeability. Accurate well-log-based permeability prediction is therefore critical for their efficient development. Nuclear magnetic resonance (NMR) logging, which provides porosity information independent of rock composition, serves as a valuable tool for this purpose. However, conventional NMR-derived models such as SDR(Schlumberger-Doll research) and Coates have inherent limitations. Therefore, based on the key parameters of the Coates model, using four parameter wells in the Wuguo-Jixian exploration area of the Ordos Basin as the data set, a deep neural network model (DNN) and a long short-term memory network model(LSTM) were constructed for permeability prediction using nuclear magnetic resonance logging. Experimental results demonstrate that the DNN-NMR model achieves good performance in static parameter fitting (R2 = 0.872), representing a 14% accuracy improvement over the Coates model. The LSTM-NMR model performs best, achieving the highest accuracy (R2 = 0.916). The research results provide a high-precision and robust solution for the quantitative NMR-log-based prediction of permeability in tight sandstone reservoirs.
YANG Dan, LI Xin-hu, DING Jia-meng, LIU Xiao-mei, ZHAO Kai-yuan.
Prediction of Tight Sandstone Reservoir Permeability Using NMR Logging Data Based on Deep Learning Algorithms: Wubao-Jiaxian Exploration Area Benxi Formation[J].
Science Technology and Engineering, 2026, 26(21): 9099-9108 DOI:10.12404/j.issn.1671-1815.2504966
渗透率预测是量化评价储层渗流能力、指导油气开发方案制定及提高采收率的关键地质参数。利用常规测井曲线进行渗透率预测的方法包括线性和非线性方法,主要通过建立多种测井曲线,分析孔隙度与渗透率的一元、多元关系,以及采用神经网络、深度学习等非线性计算方法,并取得了一定的应用效果[5-7],但在致密砂岩等孔渗关系复杂地层中回归相关性显著降低,且依赖人工特征筛选,难以处理高维非线性数据[8-9]。核磁共振(nuclear magnetic resonanceNMR)测井因其能提供与岩石矿物骨架成分无关的孔隙度信息,且T2弛豫时间与岩石内部孔隙大小成正比,成为致密砂岩等非常规储层渗透率预测的重要手段[10-11]。SDR(Schlumberger-Doll research)和Coates作为核磁共振测井渗透率求解的经典模型,在实际生产中应用广泛。但这两种模型也存在一定局限性。SDR模型在水层计算结果较好,油水层预测误差较大,主要适用于中高孔渗储层,且对核磁共振测井数据的质量要求较高[12-14];Coates模型是一种基于岩心分析数据和实验室渗透率测试结果得出的经验关系式,在孔隙结构复杂或储层条件特殊时误差增大,需根据地质条件频繁调整参数[15],增加了应用难度。
针对致密砂岩储层利用核磁共振测井进行渗透率预测时,传统方法存在的孔隙结构表征不足与时序依赖性缺失问题,现构建一种多尺度特征融合的LSTM-NMR渗透率预测模型。模型的输入特征基于Coates模型参数,融合NMR参数自由流体孔隙度(free-flowing fluid porosity,FFI)、束缚水孔隙度(bound water porosity,BVI)[23]与常规测井曲线声波时差AC(反映岩石骨架-流体响应)、自然伽马GR(表示泥质含量与岩性)、补偿中子CNL(刻画含氢指数与流体性质)构建多尺度特征集。该设计突破了传统方法依赖单一数据源的局限,通过有机融合微观核磁共振(NMR)孔隙结构参数与宏观常规测井物理响应,实现微观孔隙结构与宏观物理响应的协同表征。在此基础上,引入LSTM网络捕捉测井数据的深度时序依赖性。通过对比实验验证模型的预测精度。以期为致密砂岩储层的渗透率预测提供新的思路和解决方案。
HuangDaojun, LiXinhu, LiuYan, et al. Reservoir characteristics and favorable interval optimization of tight sandstone in the Benxi Formation, central-eastern Ordos Basin[J]. Journal of Xi’an University of Science and Technology, 2023, 43(1): 109-118.
LuPing, WangHaochen, GaoChunyun, et al. Research progress on permeability prediction technology for tight sandstone reservoirs[J]. Progress in Geophysics, 2022, 37(6): 2428-2438.
ZhuangChunhui, LiYajun, ZhangHuan, et al. The law of oil-water two-phase seepage and the calculation method of relative permeability in nanoscale pores[J]. Science Technology and Engineering, 2025, 25(11): 4505-4514.
XiongFansheng, GanLideng, SunWeitao, et al. Permeability characterization of fracture-pore mediareservoirs and analysis of its influencing factors[J]. Chinese Journal of Geophysics, 2021, 64(1): 279-288.
WangZhiguo, WangXiongxiong, ChangYongfeng, et al. Characteristics of water lock damage in tight sandstone reservoirs based on NMR technology: a case study of the taking the Sulige tight sandstone reservoir[J]. Science Technology and Engineering, 2025, 25(25): 10653-10662.
ZhaoJun, HeXuan, LaiQiang, et al. Permeability prediction model for carbonate rocks based on digital core[J]. Progress in Geophysics, 2024, 39(6): 2207-2218.
ZhangPeng, FanYunfeng. Permeability logging model-building of low porosity and low permeability reservoirs of in Ordos Basin[J]. Science Technology and Engineering, 2017, 17(17): 172-177.
YanJianping, ZhangFan, WangMin, et al. Classification and application of rock-electric parameters for low-permeability sandstone based on NMR experiments: a case study of the Es4 member in the southern slope of Dongying Sag[J]. Chinese Journal of Geophysics, 2019, 62(7): 2748-2758.
[17]
LiangZ, LiX, SunA, et al. Nuclear magnetic resonance logging-based permeability classification modeling for tight sandstone reservoirs[J]. Processes, 2024, 12(7): 1309.
[18]
XiaoL, LiuX, ZouC, et al. Comparative study of models for predicting permeability fromnuclear magnetic resonance (NMR) logs in two Chinese tight sandstone reservoirs[J]. Acta Geophysica, 2013, 62(1): 116-141.
LiuZongbin, LiChao, LuYan, et al. Fluid occurrence characteristics and permeability evaluation of low-permeability sandstone based on pore structure characterization[J]. Journal of Jilin University (Earth Science Edition), 2024, 54(4): 1124-1136.
[21]
KenyonW E, DayP I, StraleyC, et al. A three-part study of NMR longitudinal relaxation properties of water-saturated sandstones[J]. SPE Formation Evaluation, 1988, 3(3): 622-636.
FanZhuoying, HouJiagen, XingDonghui, et al. Nuclear magnetic resonance experiments and logging applications for low permeability reservoirs[J]. Journal of China University of Petroleum (Edition of Natural Science), 2019, 43(1): 53-59.
[24]
ChengY, ChenS, EidM. Determination of permeability from NMR T1/T2 ratio in carbonates[C]// SPWLA Annual Logging Symposium. Oklahoma City: SPWLA, 2017: SPWLA-2017-RR.
[25]
ValentinM B, BomC R, CompanA L M, et al. Estimation of permeability and effective porosity logs using deep autoencoders in borehole image logs from the brazilian presalt carbonate[J]. Journal of Petroleum Science and Engineering, 2018, 170: 315-330.
[26]
XuR, MaL, SongB, et al. Geohazard risk assessment model based on deep neural network[J]. Results in Engineering, 2026, 29: 109029.
ZengRui, ShaoYanlin, HuangYu, et al. Intelligent identification of logging lithology based on deep learning[J]. Science Technology and Engineering, 2025, 25(30): 12804-12812.
MaLongfei, XiaoHanmin, TaoJingwei, et al. Research and application of lithology classification based on deep learning[J]. Science Technology and Engineering, 2022, 22(7): 2609-2617.
[31]
FangZ, BaJ, CarcioneM J, et al. Permeability prediction using logging data from tight reservoirs based on deep neural networks[J]. Journal of Applied Geophysics, 2024, 229: 105501.
HeShuzhou, ChengChao, ChenLi, et al. Reservoir parameter prediction method for fine-grained sedimentary rocks under the dual-driven of physical and numerical modeling[J]. Chinese Journal of Geophysics, 2025, 68(7): 2712-2735.
GaoMing, LiHu, LiuXinjin, et al. Prediction method of groundwater in karst strata based on distance-attribute hybrid clustering combined with ConvLSTM model[J]. Science Technology and Engineering, 2025, 25(20): 8424-8434.
ChenXuefei, XinXiankang, YuGaoming, et al. Logging curve prediction based on modal decomposition fused with machine learning[J]. Science Technology and Engineering, 2025, 25(17): 7132-7141.
LiPengfei, LiPengju, ZhangQiang, et al. A reservoir permeability prediction method based on small sample data[J]. Computer Technology and Development, 2024, 34(7): 199-206.
ZhaiCheng, SunYong, FanYiren, et al. Application and prospect of low-field nuclear magnetic resonance technology in the accurate characterization of coal pore structure[J]. Journal of China Coal Society, 2022, 47(2): 828-848.
[42]
MasoudifarM, MahsuliM, TacirogluE. Deep learning-based stochastic ground motion modeling using generative adversarial and convolutional neural networks[J]. Soil Dynamics and Earthquake Engineering, 2025, 194: 109306.
[43]
NezhadG A, NiriE M. Enhancing water saturation predictions from conventional well logs in a carbonate gas reservoir with a hybrid CNN-LSTM model[J]. Journal of Petroleum Exploration and Production Technology, 2025, 15(5): 89.
ZhangFan. Study on occurrence state of clay-bound water and nuclear magnetic resonance saturation evaluation method[D]. Beijing: China University of Petroleum (Beijing), 2023.
LiNing, WangKewen, WuHongliang, et al. Permeability logging evaluation: current status and development directions[J]. Petroleum Science Bulletin, 2023, 8(4): 432-444.
WangKun, ZhouHangyu, LaiJie, et al. Application of nuclear magnetic resonance technology in petrophysics and pore structure characterization[J]. Chinese Journal of Scientific Instrument, 2020, 41(2): 101-114.
LiXinhu, DingJiameng, ChengGuangyi, et al. The comprehensive evaluation method research of gas bearing property in tight sandstone of carboniferous Benxi Formation using well logging[J]. Science Technology and Engineering, 2024, 24(13): 5364-5371.