多点地质统计学建模的发展趋势

从算法研究、训练图像处理和实际应用三个方面详细解剖了国内外多点地质统计学的发展历程,在此基础上,分析了多点地质统计学主流的几种算法的核心原理、适用范围及优缺点,以此来对储层建模的发展趋势作出展望。目前,多点地质统计学虽是随机建模的一种前沿研究热点,但由于其尚未成熟,仍需对建模算法进行研究。为此,在前人研究的基础上,重点分析了多点地质统计学的发展趋势：合理处理训练图像;合理利用软信息;选择合适的相似性方法;选择合适的标准化方法;合理利用平稳性;算法间的耦合;选择合适的过滤器;拓展缝洞型碳酸盐岩模拟。最后,提出多点地质统计学在储层建模方面,应从增加储层的模拟区域、提高模拟精度、扩大储层相的模拟范围和提高计算机模拟效率等方面进行改进。

THE DEVELOPMENT AND PROSPECT OF MULTIPLE-POINT GEOSTATISTICS MODELING

Starting with algorithm designing,training image,and practical application,the authors analyzed multiple-point statistics research trends both in China and abroad.On such a basis,the core principles of the main four MPS algorithms,their applicable ranges and advantages as well as disadvantages were analyzed,so as to forecast the trend of reservoir modeling.Multiple-point geostatistics is an international forefront research tool in stochastic modeling;nevertheless,as the algorithm is not yet mature,it should be further improved.On the basis of previous researches,the existing problems of the multiple-point geostatistics that need to be modified and the direction of the processing of training images are proposed,such as suitable processing of training image,choice of similarity methods,choice of standardized data,smoothing,coupling among algorithms,and expansion of the simulation extent for fractured-vuggy carbonate reservoir.In order to improve the usage of multiple-point geostatistics for reservoir stochastic modeling,we should spare no efforts to increase the simulation area,improve simulation accuracy,expand the scope of simulation of reservoir types,save simulation time and improve simulation efficiency.