Digital core analysis offers unprecedented insight into fluid transport mechanisms at the pore scale, yet the integration of this knowledge into standard workflows requires the consideration of larger sample volumes. For hierarchical materials, this can be achieved by appropriate classification schemes and associated upscaling of pore‐scale transport properties to continuum measures. This work introduces a robust upscaling approach based on regional Minkowski measures or Minkowski maps with or without an additional hydraulic attribute. Data clustering of this feature vector field is achieved with a Gaussian mixture model leading to spatially compact rock types capturing the layering of the analyzed laminated sandstone. Other regional attribute fields like pore size, throat size, and volume/surface ratio are analyzed using these partitions. This is followed by a characterization of their transport properties, namely, electrical conductivity and permeability, for which an efficient sampling scheme utilizing spectral methods is introduced. We compare the upscaled transport properties on the basis of homogenization to full‐scale direct pore‐scale simulations for a thinly laminated sandstone and demonstrate very good agreement and large computational speed‐up.

中文翻译：

---

使用Minkowski贴图和水力属性的层状砂岩孔隙尺度放大方法

数字岩心分析提供了对孔隙尺度流体传输机制的空前洞察力，但是将这些知识集成到标准工作流程中需要考虑更大的样品量。对于分层材料，这可以通过适当的分类方案和将孔尺度传输特性的相关升级提升为连续测量来实现。这项工作引入了基于区域Minkowski测度或Minkowski映射的健壮的放大方法，带有或不带有附加的水力属性。该特征向量场的数据聚类是通过高斯混合模型实现的，该模型导致空间紧凑的岩石类型捕获所分析的叠层砂岩的分层。使用这些分区可以分析其他区域属性字段，例如孔径，喉咙大小和体积/表面积比。其次是表征其传输特性，即电导率和磁导率，为此引入了利用频谱方法的有效采样方案。我们将基于均质化的高档输运性质与薄层砂岩的全尺寸直接孔隙尺寸模拟进行了比较，并显示出很好的一致性和较大的计算速度。