天府气田须家河组致密砂岩气藏“砂中找储”识别技术  

Identification technology of"finding reservoir in sand"in tight sandstone gas reservoir of Xujiahe Formation,Tianfu gas field

作　　者：熊文浩 张冲[1,2] 陶云贺 周新波 韩华洋 张文艺 XIONG Wenhao;ZHANG Chong;TAO Yunhe;ZHOU Xinbo;HAN Huayang;ZHANG Wenyi(Key Laboratory of Exploration Technologies for Oil and Gas Resources(Yangtze University),Ministry of Education,Wuhan 430100,China;Geophysics and Petroleum Resources College,Yangtze University,Wuhan 430100,China;Southwest Branch,CNPC Logging Co.,Ltd.,Chongqing 400021,China;Yingmai Oil and Gas Production Management Area,Tarim Oilfield Company,PetroChina,Korla 841000,China)

机构地区：[1]油气资源与勘探技术教育部重点实验室(长江大学),湖北武汉430100 [2]长江大学地球物理与石油资源学院,湖北武汉430100 [3]中国石油集团测井有限公司西南分公司,重庆400021 [4]中石油塔里木油田公司英买采油气管理区,新疆库尔勒841000

出　　处：《断块油气田》2025年第1期118-125,共8页Fault-Block Oil & Gas Field

基　　金：国家自然科学基金项目“致密气储层岩石导电机理研究及饱和度评价”(41404084);国家自然科学基金青年项目“各向异性精细储层的井间电磁法HIE-FDTD正演”(42204127);中国石油重大科技专项“西南油气田天然气上产300亿立方米关键技术研究与应用”(2016E-0606)。

摘　　要：天府气田须家河组致密砂岩具有岩性复杂、砂体致密、储集物性差、非均质性强等特点,缺乏一套系统的储层识别方法。由岩性识别的曲线重构法计算的裂缝综合参数、粒度表征参数,与声波时差测井参数一起作为储层识别的3个特征参数,利用Stacking集成策略,融合随机森林、支持向量机、多层感知器等3个子算法模型进行超参数寻优,结合岩心及测试资料提取特征参数,建立储层识别的二分类模型——“砂中找储”模型。此模型在验证集中的识别准确率达到94.91%。该模型在目标区块验证井的识别结果与试油结论及生产动态资料吻合,证明了该模型在天府气田须家河组的适用性,为同类型储层的快速识别和有效性评价提供了技术参考。The tight sandstone reservoir of Xujiahe Formation in Tianfu gas field is characterized by complex lithology,tight sand body,poor reservoir physical property and strong heterogeneity,which lacks a systematic reservoir identification method.In this paper,the fracture synthesis parameters and particle size characterization parameters calculated by the curve reconstruction method of lithology identification,together with the AC logging parameters,are used as three characteristic parameters for reservoir identification.The integration strategy of Stacking is used to integrate three sub-algorithm models,namely,random forest,support vector machine,and multi-layer perceptron for hyperparameter optimization.Based on the characteristic parameters extracted from the core and test data,the two-classification model of reservoir identification,the"finding reservoir in sand"model is established,and recognition accuracy of this model in verification set reaches 94.91%.The identification results of the validated well in the target block are consistent with the results of oil testing and production dynamic data,which proves the applicability of the model in Xujiahe Formation,Tianfu gas field,and provides a technical reference for the rapid identification and effectiveness evaluation of the same type of reservoirs.

关 键 词：致密砂岩气藏 Stacking集成算法 储层识别 须家河组 

分 类 号：TE132.1[石油与天然气工程—油气勘探]