非常规储层脆性研究进展及展望  

Progress and prospects of brittleness research in unconventional reservoirs

作　　者：刘震[1,2] 张军华[1,2] 于正军[3] 任瑞军[3] 孙有壮 LIU Zhen;ZHANG Junhua;YU Zhengjun;REN Ruijun;SUN Youzhuang(National Key Laboratory of Deep Oil and Gas,China University of Petroleum(East China),Qingdao,Shandong 266580,China;School of Geosciences,China University of Petroleum(East China),Qingdao,Shandong 266580,China;Geophysical Research Institute,Shengli Oilfield Company,SINOPEC,Dongying,Shandong 257022,China)

机构地区：[1]深层油气全国重点实验室(中国石油大学(华东)),山东青岛266580 [2]中国石油大学(华东)地球科学与技术学院,山东青岛266580 [3]中国石化胜利油田分公司物探研究院,山东东营257022

出　　处：《石油地球物理勘探》2023年第6期1499-1507,共9页Oil Geophysical Prospecting

基　　金：国家自然科学基金项目“含油气盆地低级序断裂的发育规律及其与高级序断裂的成因联系——以渤海湾盆地济阳坳陷为例”(42072169);中国石化股份公司先导项目“准东地区二叠系非常规甜点预测技术研究”(P21007)联合资助。

摘　　要：简要论述了脆性的国内外研究现状和影响因素,在此基础上总结了7种最主要的脆性指数表达式,阐述了各自的物理含义。将储层脆性的预测方法分为基于传统方法的地震脆性反演方法与基于深度学习技术的脆性预测方法,评价了两种方法的优缺点:传统方法应用广泛、算法相对成熟、稳定可靠但需要大量先验知识,在复杂地质条件下效果受限;深度学习方法可以适应复杂地质条件,无需先验信息,原理和操作简单,但是训练过程需要大量的计算资源和时间。介绍了贝叶斯AVAZ脆性直接反演、基于测井导数和波动率属性的机器学习脆性预测、基于CNN-LSTM模型的混合神经网络脆性预测等新技术。最后,展望了脆性预测技术的发展方向。This paper briefly reviews the current domestic and international research status and influencing factors related to brittleness.On this basis,we summarize seven main expressions for brittleness indices and expound upon their respective physical interpretations.Brittleness prediction methods for reservoirs are categorized into seismic brittleness inversion based on conventional approaches and brittleness prediction based on deep learning techniques.Meanwhile,the advantages and disadvantages of the two methods are evaluated.Conventional methods are characterized by wide applications,relatively mature algorithms,stability and reliability,and requirements for substantial prior knowledge with limited effects in complex geological conditions.Conversely,deep learning methods feature adaptability to intricate geological conditions,no need for prior information,and straightforward principles and procedures.However,the training process demands significant computational resources and time.Additionally,we introduce new technologies such as direct inversion of Bayesian AVAZ brittleness,machine learning brittleness prediction based on well-logging derivatives and volatility attributes,and hybrid neural network brittleness prediction by employing CNN-LSTM models.Finally,this study provides an outlook on the future development of brittleness prediction techniques.

关 键 词：非常规储层 脆性指数 杨氏模量 泊松比 叠前反演 机器学习 

分 类 号：P631[天文地球—地质矿产勘探]