摘要
酒泉盆地青西坳陷青南凹陷柳沟庄—窟窿山构造下沟组储层岩性主要为低孔、低渗砂砾岩类和泥云岩类,岩石矿物成分复杂、泥质含量高、黄铁矿富集、裂缝类型及组合形式复杂,属典型复杂岩性裂缝—孔隙型储层。在这类复杂岩性裂缝—孔隙型储层中,自然伽马等测井曲线不能很好指示地层中的泥质含量,常规测井资料难以准确识别地层的岩石矿物成分,单条测井曲线与岩心孔隙度之间的关联度低,采用常规的孔隙度测井计算方法存在明显的缺陷,孔隙度计算精度远远不能满足储层评价和储量计算要求。文章利用岩心分析数据和测井信息等资料,采用3层BP神经网络进行学习训练,得到砂砾岩岩类和泥云岩岩类的孔隙度计算模型。利用该模型计算储层孔隙度,其结果与岩心分析孔隙度比较,平均误差小于1.5%,能满足储量计算要求。在实际应用中见到良好效果,孔隙度计算精度明显得到提高。
The Xiagou Formation reservoir rocks in Liugo uzhuang-Qionglongshan structure in Qingnan Seg of Qingxi Depression in Jiuquan B ain are mainly comprised of the low porosity and low permeability glutenites and shaly dolostones, and the reservoir is regarded as the typical complex litholo gical fractured-porous one because of complicated rock mineral composition, hi gh shale content, enriched pyrites and complicated fracture type and combinatio n shape. In such kind of complex lithological fractured-porous reservoir, the shale content in formation couldn’t be well indicated by natural gamma-ray log and so on, it was difficult to identify accurately the rock mineral compositio n by conventional log data, and the associability between single log and core p orosity was low, so that there existed evident shortcomings in the calculation methods of adopting the conventional porosity logs and the porosity calculation accuracy was far from meeting the needs of reservoir evaluation and reserve esti mation. The porosity calculation model of the glutenites and shaly dolostones wa s set up through learning and training by use of 3-layer BP neural networks. Th e average error of the porosities calculated by the model was less than 1.5% as compared with the core analysis porosities, which can meet the needs of reserve estimation. The porosity calculation accuracy was greatly raised because of app lying this model.
出处
《天然气工业》
EI
CAS
CSCD
北大核心
2005年第5期29-30,36,共3页
Natural Gas Industry
