资源环境科技发展态势分析平台

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 vector field is achieved with a Gaussian mixture model leading to spatially compact rock‐types capturing the layering of the analysed 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 characterisation 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.