基于Markov链模型的储层岩相随机模拟

Stochastic reservoir lithofacies simulation based on Markov Chain model

在油气储层随机建模研究中,基于Markov链模型的方法是一类较受欢迎的技术,同时也是一类不成熟的技术.问题的症结之一在于侧向的转移概率矩阵很难求取.针对这种情况,作者在深入理解Walther相律的基础上,借鉴模拟退火算法的相应思路,提出了一种岩相模拟的新方法,该方法依据不同岩相的百分比进行随机模拟得到一幅初始图象,而后以按岩相组织剖面得到的垂向和侧向的岩相转移概率矩阵的相似性作为判别标准对图象进行扰动,直至得到满意的图象.二维模型试算结果表明了这种岩相随机模拟方法的可行性.

Methods based on Markov Chain model are newly appeared techniques popular with researchers at work on stochastic reservoir modeling. However, these methods are underdeveloped on the other hand. One of the bottlenecks is how to get transition probability matrixes on lateral directions. In order to overcome the obstacle, a fresh method combining Markov Chain model with simulated annealing algorithm is put forward here. After obtaining initial image, a series of disturbing are carried out until target image appears. The validity of the method is testified through a 2-D experiment.