地质统计学储层建模发展述评

两点地质统计学储层建模通过变差函数确定随机变量的相关关系，解决了地质变量的空间线性插值问题，借助在不同方向上定义的独立变差函数，这种插值算法还能表征地质变量的各向异性。多点地质统计学（MPS）模拟主要有基于像素的模拟和基于样式的模拟等两大类算法，根据对样式分类与模拟过程的差异，又可将基于样式的模拟算法分为直接模拟算法、基于过滤器的模拟算法和基于距离的模拟算法。从地质统计学的算法演化角度出发，总结了两点地质统计学和MPS算法及其适用性；以图像识别为基础， MPS在识别沉积体中薄的非渗透夹层、集成生产数据、模拟井间储层连续性及裂缝分形特征模拟等方面拓展了两点地质统计学的研究领域。

A Review of Geostatistical Reservoir Modeling Development

Two-point geostatistical reservoir modeling uses a variogram to confirm the correlation between random variables, solving the issue of spatial linear interpolation of geological variables. With an independent variation function defined for different directions, this type of interpolation algo-rithm can even characterize the anisotropy of geological variables. Multiplepoint geostatistics simulation (MPS) uses two major types of algorithms:pixel-based simulation and pattern-based simulation. Based on the differences of the pattern classification and simulation process, we can further classify the pattern-based simulation algorithms into direct, filter-based, and distance-based algorithms. Based on geostatistical algorithm evolution, we summarized the two-point geo-statistics and MPS algorithms and their application. Based on image identification, the MPS expands the two-point geostatistics research field, identifying thin and impermeable interbeds in sedimentary bodies, integrating production data, simulating the continuities of interwell reservoirs, and simulating the fractal characteristics of fractures.