储层随机建模研究进展

THE PROGRESS OF RESERVOIR STOCHASTIC MODELING

按照模拟网格不同储层随机建模技术可以分为基于目标的方法和基于象元的方法2类。介绍了基于目标方法中的布尔模拟方法及其改进,这些改进包括井间和井上模拟时采取不同的概率分布、模拟多个岩相单元、模拟具有波动顶底的目标、综合地震数据、采取G ibbs分布的广义示性点过程模拟目标之间相互影响以及采用线模型模拟河道的目标层次建模(F luvsim)。论述了基于象元方法中的新方法,指出多点地质统计学比传统的两点统计学具有明显优势,分析了多点地质统计学存在的问题及其改进方向(平稳性问题处理、目标体连续性处理、数据样板选择、综合地震信息方面)。最后评述了储层随机建模的发展方向,指出在目标之间相互影响函数的确定、获取反应实际储层结构特征参数、先验概率求取以及随机模拟结果评价方面有待进一步研究。

Reservoir stochastic modeling is a new method for characterizing the heteroigeneties integrating multi-disciplines knowledge. The mehod can divided into two categories,the object based method and pixelbased methods. The boolean algorithm belongs to object-based method and its improvements are intro duced ,including the differences of the probability distribution on the wells and inter-wells,modeling multi ple rocks,modeling the undulating objects,integrating seismic information,modeling object which is pro- jected by multiple wells, modeling the interaction of different object with Gibbs distribution and using hierarchical object based method（Fluvsim） which is based on line model. The article also intoduces the multiple point statistics which belongs to pixel-based methods. The modeling result shows the multiple-point statistics is more reasonable compareing to the traditional two-point statistics. The multiple-point statistics also get some progresses in dealing with stationary problem ,the continuity of the object ,the choice of data template and integration of seismic data. Finally the author point out that the future work will focus on three aspects,that is the function which depicts the interation of different objects ,the priori probability, the extraction of modeling parameters and the evaluation of realization.