解决地质非平稳性的新方法

在传统的储层随机建模中,获得具有代表性的统计数据需要平稳的训练图像,因此,建模中都基于平稳性假设。针对非平稳地质现象是普遍存在的这一事实,首先提出了四种不同的多点地质统计学算法来模拟非平稳现象,然后从方法和应用的角度具体阐述了用存在的模型作为训练图像结合核函数新的多点模拟方法。结果发现,该方法能很好地捕获真实空间地质体的空间差异性。同时,这个算法具有很重要的实践意义,解决了基于地质过程储层模型的井数据条件化的问题。

New method to solve non-stationarity of geology

In the traditional stochastic modeling, acquiring representable typical statistical data requires training images to be stationary, Hence the assumption in modelling is stationarity. According to the common existence of non-stationarity, in this paper, four different types of training images and their corresponding multiple-point simulation algorithms are reviewed. The multiple-point simulation method by existing model combined with kernel-function is discussed. The result shows that non-stationary models can capture well spatial variation＇ of space phenomena in real world. It has practical significance to solve the problem of conditioning geologic process-based reservoir models to well data.