Multiscale pore structure characterization based on SEM images

Yuzhu Wang, Shuyu Sun

Research output: Contribution to journalArticlepeer-review

15 Scopus citations


The micropore structure’s permeability contribution to total permeability of the heterogeneous reservoir with multiscale pore structures is critical for reservoir evaluation but still not well understood. This paper proposes a multiscale pore structure characterization method based on high-resolution SEM images to quantitatively analyse the micropore structures’ content and their permeability contributions via six steps. First, the image-based rock typing is implemented to classify a multiscale pore structure into different rock types using the random forest algorithm. Second, the 3D model of the macropore structure and every micropore structure is reconstructed applying the MPS method. Third, the permeability of each reconstructed 3D micropore structure is calculated using LBM, and the corresponding permeability REV of this structure is estimated. Four, an upscaling process is carried out to divide the reconstructed 3D macropore structure into many cells whose length is determined by the maximum permeability REV of the micropore structures. Five, the permeability of every cell of the coarse grid is calculated by LBM except some cells that are randomly selected as micropore structures whose permeability is assigned directly according to their rock types. Finally, the permeability contribution of each micropore structure is evaluated by comparing the permeability calculated before and after assuming the target micropore structure is impermeable. The result shows that the permeability contribution of a micropore structure varies significantly according to its permeability, content, spatial distribution, and the permeability of the macropore structure.
Original languageEnglish (US)
Pages (from-to)119915
StatePublished - Dec 21 2020

Bibliographical note

KAUST Repository Item: Exported on 2020-12-28


Dive into the research topics of 'Multiscale pore structure characterization based on SEM images'. Together they form a unique fingerprint.

Cite this