摘要
珠江口盆地X油田是典型的高泥质疏松砂岩油田,储层主要发育三角洲前缘水下分流河道~河口坝沉积,沉积时期水动力弱,粒度细,砂岩储层疏松且泥质含量高。油田面临储层数量多且单层薄;砂岩疏松引起测井扩径导致声波时差和密度等测井曲线测量失真;砂泥叠置,常规波阻抗反演不能区分出有效储层等难题。目前油田开发阶段急需对储层空间展布精细刻画,而常规反演方法难以做到高精度储层预测。针对这些难点,这里分“三步法”逐一攻克并形成了一套完整的针对高泥质疏松砂岩薄储层的识别技术系列:①遗传化神经网络技术重构得到新的测井曲线,解决了测井扩径导致的声波时差和密度曲线测量失真问题;②对比优选的储层特征敏感参数能很好区分储层与非储层;③相控分频反演基于地震波形实现地震相约束,分频反演实现高分辨率反演。实际应用证实,该技术在X油田调整井优化实施及开发调整方案井网设计等方面取得了较好的应用效果,解决了该油田3 m~5 m薄层精确描述的难题,形成了一套完整的针对高泥质疏松砂岩薄储层预测的技术流程。该技术对同类油藏表征和油田高效开发具有参考意义。
The X oil field in the Pearl River Delta Basin is characterized as a typical high-argillaceous and unconsolidated sandstone.The reservoir mainly develops the underwater distributary channel-mouth bar deposits in the Delta Front.During the sedimentary period,the hydrodynamic force is weak and the grain size is fine.Loose sandstone reservoir and high shale content are observed in this field.The oil fields are faced with the following difficulties:1)the reservoir is numerous thin single-layers,and the logging curve measurement distortion such as acoustic wave and density is caused by the well logging enlargement caused by the loose sandstone;2)the sand-mud stack and the wave impedance inversion can not distinguish the effective reservoir.At present,it is urgent to describe the distribution of reservoir space in the stage of oilfield development,but it is difficult to predict reservoir with high precision by conventional inversion method.In view of these difficulties,we reconstruct and restore the real logging curve by neural network technology,constructs the new reservoir sensitive parameters and based on the seismic waveform phased frequency division high resolution inversion in this paper.Thus,a complete set of identification techniques for thin reservoirs of high argillaceous unconsolidated sandstone is formed.The results show that:(1)the technique of reconstruction of logging curves by neural network can solve the problem of logging curve measurement inaccuracy caused by logging enlargement with loose sandstone;(2)The reservoir characteristic sensitive parameter Siméon Denis Poisson's ratio can distinguish the reservoir and the non-reservoir well;(3)waveform facies controlled frequency division inversion realizes the seismic phase restraint based on the seismic waveform,and frequency division inversion realizes the high resolution inversion.The practical application proves that the technology has achieved good application results in optimizing the implementation of adjustment wells and well pattern design of development adjustment program in x oilfield.It also solved the difficult problem of fine description of thin reservoir of 3M~5m in this oilfield,and then a complete set of technical process for prediction of thin reservoirs in high argillaceous unconsolidated sandstone has been formed.This technology has reference significance for the characterization of similar reservoirs and high efficiency development of oil fields.
作者
代玲
万钧
罗泽
DAI Ling;WAN Jun;LUO Ze(CNOOC China Limited,Shenzhen Branch,Shenzhen 518000,China)
出处
《物探化探计算技术》
CAS
2022年第1期1-8,共8页
Computing Techniques For Geophysical and Geochemical Exploration
基金
中海石油有限公司项目(CCL2021SZPS0294)。
关键词
测井扩径
砂泥叠置
神经网络
泊松比
波形相控
分频反演
log enlargement
sand-mud overlay
neural network
siméon denis poisson ratio
waveform control
frequency division inversion
