.png){kind=icon}
.png){kind=icon}
Return
Advanced algorithms for accurate mean pore size prediction in dolomite: Leveraging experimental petrophysical data
.png){kind=icon}
PRE
AI
M
V
A
D
A
K
G
S
Y
K
DOI：10.1016/j.pce.2026.104472.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Accurate estimation of pore structure in dolomite reservoirs is essential for predicting fluid flow and reservoir performance, yet conventional techniques such as MICP are costly, destructive, and limited in coverage. This study introduces an experimentally validated, data-driven framework for predicting mean pore size using routinely measured petrophysical parameters including porosity, permeability, cementation factor, and grain density. A suite of machine learning models was systematically optimized and benchmarked, revealing that Random Forest provides near-perfect predictive accuracy. Beyond model training, the workflow incorporates outlier detection, cross-validation, and SHAP-based interpretability to ensure physical consistency and generalizability. The proposed approach offers a non-destructive, high-resolution alternative to laboratory pore-size measurements and represents a significant advancement over existing empirical correlations by enabling reliable pore-scale characterization directly from core-analysis data.
Keywords:
Pore size
Data-driven
Core analysis
Petroleum
Dolomite
Journal
IF:
4.1
Papers:
3.3K
Citations:
6.9K
Organization
Cited Papers