Merge pore clusters: A novel method to construct pore networks and predict permeability from 2D rock images

Chenhui Wang*, Kejian Wu, Gilbert G. Scott, Ailin Jia

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)


We propose a novel method to predict fluid flow properties of reservoir rocks from 2D rock images by connecting pore clusters based on their intrinsic properties. Spatial proximity and connecting large clusters with priority are the two basic connecting principles of this method. A few methods from complex networks were used to analyze the topology and connectivity of the connected pore networks, resulting in more realistic and optimal pore networks from 2D images. A new topological descriptor is proposed and was found to perform well in quantifying the network topology by considering both the isolated pore cluster number and the Euler characteristic. The new method can predict permeability from the 2D rock images with a reasonable agreement compared to the reference cases for different rock types and complex pore structures.

Original languageEnglish
Article number104238
Number of pages12
JournalAdvances in Water Resources
Early online date8 Jun 2022
Publication statusPublished - 1 Aug 2022

Bibliographical note

Funding Information:
C.W. thanks the financial support from China Scholarship Council (CSC) for his Ph.D. study. The authors thank Anasuria Operating Company Limited (AOC) for making available the core plug samples. The authors thank John Still of University of Aberdeen Centre for Electron Microscopy, Analysis and Characterisation (ACEMAC) for the assistance in the acquisition of micro-CT and SEM images.

Publisher Copyright:
© 2022 Elsevier Ltd


  • 2D rock images
  • New topological descriptor
  • Optimal connection
  • Pore network modeling
  • Predict permeability


Dive into the research topics of 'Merge pore clusters: A novel method to construct pore networks and predict permeability from 2D rock images'. Together they form a unique fingerprint.

Cite this