一种基于常规测井资料计算碳酸盐岩储层裂缝孔隙度新方法  

A New Method for Calculating Fracture Porosity Based on Conventional Logging Data

作　　者：雷明[1,2] 陈涛 韩乾凤[1,2] 程木伟 高庚 沙雪梅 张亚军[1,2] Lei Ming;Chen Tao;Han Qianfeng;Cheng Muwei;Gao Geng;Sha Xuemei;Zhang Yajun(PetroChina Research Institute of Petroleum Exploration&Development-Northwest,Lanzhou 730020,China;Key Laboratory of Reservoir Description,CNPC,Lanzhou 730020,China;PetroChina Research Institute of Petroleum Exploration&Development,Beijing 100083,China;Exploration and Development Research Institute of Daqing Oilfield,Daqing 163712,China)

机构地区：[1]中国石油天然气股份有限公司勘探开发研究院西北分院,甘肃兰州730020 [2]中国石油天然气集团有限公司油藏描述重点实验室,甘肃兰州730020 [3]中国石油天然气股份有限公司勘探开发研究院,北京100083 [4]大庆油田有限责任公司勘探开发研究院,黑龙江大庆163712

出　　处：《地球科学》2023年第7期2678-2689,共12页Earth Science

基　　金：国家科技重大专项(No.2017ZX05030-003).

摘　　要：裂缝作为地下油气储集空间和油气运移的通道,是裂缝型储层研究的重要内容,裂缝孔隙度是裂缝型储层测井评价中的重要参数之一.虽然裂缝定性识别和描述的方法很多,但是用常规测井资料进行裂缝孔隙度的定量计算一直是储集层裂缝解释中的难题.以阿姆河盆地某气田上侏罗统卡洛夫‒牛津阶组台缘上斜坡相对高能滩相和丘滩复合体的裂缝‒孔隙型储层为例,提出一种成像测井解释的裂缝孔隙度数据约束条件下,基于神经网络算法的常规测井资料计算裂缝孔隙度新方法.针对研究区少量有成像测井资料的井,首先利用深浅双侧向电阻率资料,结合密度曲线数据和声波曲线数据,运用多种经典模型方法计算裂缝孔隙度;然后计算加权因子,将各种模型计算的裂缝孔隙度进行加权计算,利用成像测井资料计算出的精度较高的裂缝孔隙度作为约束,并对计算结果进行标定,完成有成像资料井的常规测井资料的最终裂缝孔隙度计算;最后,运用概率神经网络算法建立起计算的有成像测井资料的裂缝孔隙度与常规测井曲线之间的映射关系,外推计算无成像测井资料所有井的裂缝孔隙度,并利用交叉验证准则确定其最终预测误差.结果表明该方法计算的裂缝孔隙度与成像测井解释的裂缝孔隙度吻合好,对无成像测井资料的井横向外推计算后,根据目的层段实际井漏、生产动态资料分析、储层参数验证对比,与现场生产状况契合,间接证实了计算结果的可靠性,表明该方法是一种行之有效的方法.Fracture as reservoir space and migration channel of oil and gas,is an important part of the fracture reservoir study,so that,fracture porosity is the most important parameter in fracture reservoir logging evaluation.Although there are many qualitative identification and description methods,but how to use conventional log data to quantitatively calculate fracture porosity has always been a difficult problem in fracture reservoir interpretation.A new method for calculating fracture porosity based on conventional logging data of probabilistic neural network is proposed in this study,taking a reservoir of a gas field in the Amu Darya Basin as an example.The reservoir in Upper Jurassic Callovian-Oxfordian order group is located on the platform margin slope,with relatively high energy beach facies and high beach complex fracture-pore type.To calculate fracture porosity of the wells which have imaging well logging data,first,a variety of classic model methods are used to calculate fracture porosity with dual laterolog resistivity data,comprehensive with the acoustic data and density data.Then the weighted factor to weighted the fracture porosity is calculated by those kinds of models,and the weighted calculation result is calibrated using the accurate fracture porosity calculated by imaging logging data as a constraint to get the final fracture porosity curve.For the wells which do not have imaging well logging data,using probabilistic neural network algorithm of deep learning to establish the mapping relation between the calculated fracture porosity curve from imaging logging data and conventional well logging data,so that the fracture porosity curve of other wells can be calculated,and the calculation error can be determined by using cross validation criterion.The results show that the fracture porosity calculated by this new method is in good agreement with the fracture porosity interpreted by imaging logging.For the wells without imaging logging data,after the lateral extrapolation calculation,according to the actu

关 键 词：常规测井 成像测井 裂缝孔隙度 阿姆河盆地 概率神经网络. 

分 类 号：P631.8[天文地球—地质矿产勘探]