基于灰色关联法和主成分分析法的致密砂岩储层流体性质测井解释评价方法  

Logging interpretation and evaluation method for fluid properties of tight sandstone reservoirs based on grey relational and principal component analysis

作　　者：屈凯旋 于春勇 彭力 孙皓 孙春辉 QU Kaixuan;YU Chunyong;PENG Li;SUN Hao;SUN Chunhui(Oil&Gas Cooperation and Development Company,BHDC,CNPC,Tianjin 300457,China;No.1 Mud Logging Company,BHDC,CNPC,Tianjin 300280,China;Mud Service Company,BHDC,CNPC,Tianjin 300280,China)

机构地区：[1]中国石油渤海钻探油气合作开发分公司 [2]中国石油渤海钻探第一录井公司 [3]中国石油渤海钻探泥浆技术服务分司

出　　处：《录井工程》2024年第3期89-97,共9页Mud Logging Engineering

基　　金：国家科技重大专项“上古生界海陆过渡相页岩气赋存机理与富集规律研究”(编号:2016ZX05034-001);中国石油渤海钻探工程公司重大工艺研究项目“苏里格西区构造-岩性复合气藏富集成因机理与储层流体性质识别技术研究”(编号:2023ZD19Y)联合资助。

摘　　要：苏里格气田致密砂岩气藏储层非均质性强,气水关系复杂,流体测井响应特征不明显,导致常规测井参数交会图法对储层流体性质的识别符合率较低,且容易引起误判,严重制约了气田的勘探开发。基于灰色关联法和主成分分析法,建立了一种致密砂岩储层流体性质测井解释新方法,并利用苏20区块西南部30口井112个试气层位资料对该方法的准确性进行了验证。首先利用灰色关联法优选出对储层流体性质变化反映最为敏感的声波时差、补偿密度、补偿中子、孔隙度和声波时差-补偿中子曲线幅度差(A_(1))共5种测井响应参数,在此基础上,开展主成分分析,构建了Y_(1)-Y_(5)共5个相互独立的主成分变量,并选取累计方差贡献率为86.99%的主成分Y_(1)和Y_(2)来综合反映多维测井变量的原始信息,最后,利用Fisher线性判别法,建立了气水定量解释模型。应用新方法结合多种测井敏感参数信息对储层流体性质进行识别,整体解释符合率为86.6%,气层解释符合率达到92.3%,相对常规交会图法均有明显提升。同时,新方法在研究区盒8段实际应用中效果显著,进一步表明新方法能够为研究区致密气藏的高效开发提供有效的技术支撑。Due to the significant heterogeneity and intricate gas-water relation of the tight sandstone gas reservoirs in Sulige Gas Field,there is a limited distinction in fluid logging response characteristics.Consequently,this leads to a low coincidence rate in identifying the reservoir fluid properties for the conventional logging parameters cross-plot interpretation methods,making it challenge to effectively identify gas layers.A new gas-water identification method for tight sandstone reservoirs is established based on grey relational and principal component analysis methods.112 gas test reservoirs from 30 wells in the southwestern part of Su 20 block are selected as analysis samples.First,the grey relational method is used to select 5 logging response parameters,which are the most sensitive to the change of reservoir fluid properties,namely interval transit time,compensated density,compensated neutron,porosity and the difference(A_(1))in amplitude between internal transit time and compensated neutron curves.On this basis,principal component variables from Y_(1)to Y_(5) independent from each other are constructed based on the principal component analysis method,and principal components Y_(1)and Y_(2),with a cumulative variance contribution rate of 86.99%,are selected to synthesize the raw information reflecting the multidimensional logging variables.Finally,Fisher linear discriminant method is used to establish a gas-water quantitative interpretation model.The new method recognizes the reservoir fluid properties by integrating the information of multiple logging sensitive parameters,the overall interpretation compliance rate is 86.6%,and the gas layer interpretation compliance rate reaches 92.3%,which is an obvious improvement in the recognition efficiency compared with the conventional crossplot chart method.At the same time,the new method is effective in practical application on Member 8 of Shihezi Formation,which further indicates that the new method can provide effective technical support for the efficient developm

关 键 词：灰色关联 主成分分析 测井曲线 致密砂岩储层 流体识别 

分 类 号：TE132.1[石油与天然气工程—油气勘探]