地震孔隙度反演技术在川西砂岩储层中的应用与比较
|
摘要
孔隙度是描述储层品质的一个重要参数,对生产开发和储量估计具有重要的意义。岩石孔隙度和地震波速度之间存在着内在联系,利用地震和测井数据反演井间孔隙度值是储层预测的主要内容之一。文章探讨了单属性和多属性多元线性回归法、神经网络法、几种克里金法、随机模拟法等多种孔隙度反演方法的基本原理、特点和适应条件,并分析对比了孔隙度预测方法在川西BMM构造陆相碎屑岩储层的应用效果。结果表明,在川西地区岩性变化快、砂岩储层非均质性强、孔隙度值变化大的地质条件下,多属性法优于单属性法,数据驱动的克里金法和神经网络法预测效果又优于线性回归法,数据驱动与地质模型约束相结合的随机模拟方法获得了最佳的预测效果,与实际的地质情况非常吻合。
Porosity is one of the important parameters for describing reservoir quality and is of great significance to production, development and reserves estimation. Interal relationship exists between porosity and seismic-wave velocity, and inter-well porosity inversion with seismic and logging data is one of the key elements of reservoir prediction. This paper discusses the concepts, characteristics and application conditions of several porosity inversion methods, such as linear regression of single attribute and multivariate linear regression of multiple attributes, neural network, several Kriging methods and stochastic modeling, and compares the application effects of these methods to the terrestrial detrital reservoirs in BMM structure, western Sichuan basin. Under the specific geological conditions of rapid lithological change and strong heterogeneity and large porosity variation of the sandstone reservoirs in the study area, the multiple attribute method is better than the single attribute method, while data-driven Kriging and neural network methods are better than the linear regression method. Prediction results of stochastic modeling method that integrates data-driven method with constraints of geological model are the best and well coincide with the real geologic situations.
Porosity is one of the important parameters for describing reservoir quality and is of great significance to production, development and reserves estimation. Interal relationship exists between porosity and seismic-wave velocity, and inter-well porosity inversion with seismic and logging data is one of the key elements of reservoir prediction. This paper discusses the concepts, characteristics and application conditions of several porosity inversion methods, such as linear regression of single attribute and multivariate linear regression of multiple attributes, neural network, several Kriging methods and stochastic modeling, and compares the application effects of these methods to the terrestrial detrital reservoirs in BMM structure, western Sichuan basin. Under the specific geological conditions of rapid lithological change and strong heterogeneity and large porosity variation of the sandstone reservoirs in the study area, the multiple attribute method is better than the single attribute method, while data-driven Kriging and neural network methods are better than the linear regression method. Prediction results of stochastic modeling method that integrates data-driven method with constraints of geological model are the best and well coincide with the real geologic situations.
引文
[1]王家华,等.油气藏储层随机模拟[M].北京:石油工业出版社,2001.
[2]印兴耀,等.储层建模中地质统计学整合地震数据的方法研究进展[J].石油地球物理勘探,2002,37(4):423-430.
[3]吴大奎,等.地震、测井资料联合反演的神经网络算法研究[J].天然气工业,2004,24(3):55-57.
[4]李跃刚,等.川西白马庙气田蓬莱镇组气藏地震储层预测技术研究[J].天然气工业,2002,22(1):18-22.
[5]DANIEL P HAMPSON,JAMES S SCHUELKE,JOHN A QUIREIN.Use of multiattribute transforms to predict log properties fromseismic data[J].Geophysics,2001,66(1):220-236.
[2]印兴耀,等.储层建模中地质统计学整合地震数据的方法研究进展[J].石油地球物理勘探,2002,37(4):423-430.
[3]吴大奎,等.地震、测井资料联合反演的神经网络算法研究[J].天然气工业,2004,24(3):55-57.
[4]李跃刚,等.川西白马庙气田蓬莱镇组气藏地震储层预测技术研究[J].天然气工业,2002,22(1):18-22.
[5]DANIEL P HAMPSON,JAMES S SCHUELKE,JOHN A QUIREIN.Use of multiattribute transforms to predict log properties fromseismic data[J].Geophysics,2001,66(1):220-236.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心