Integrating SEM image-based hydraulic radius with fluid intrusion data for improved permeability prediction in heterogeneous limestone

This study presents an exploratory workflow for permeability prediction in heterogeneous limestone by integrating high-pressure mercury intrusion (HPMI), low-temperature N 2 adsorption, and SEM image analysis within the Ergun-Wu framework. Instead of relying only on an arithmetic-average geometric descriptor, the method introduces the most probable hydraulic radius ( R mp ) extracted from SEM images as the characteristic radius for permeability estimation. Ten limestone samples from the Taiyuan Formation in the Fan’gezhuang Coal Mine were analyzed to examine the relationships among pore structure, hydraulic radius, and permeability. The results show abundant micropore-related specific surface area, with pore area more concentrated in the micropore range than in the mesopore range, and the samples display a typical low-porosity and low-permeability character. For the present sample set ( n = 10), permeability predictions based on R mp show much better agreement with measured values than predictions based on the arithmetic-average hydraulic radius, yielding R 2 = 0.939, RMSE = 7.65 × 10 –7 μm 2 , MAE = 6.10 × 10 –7 μm 2 , and MAPE = 15.8%. These results indicate that the most probable hydraulic radius provides a more representative image-derived parameter for permeability estimation in heterogeneous limestone. The present study should be regarded as an exploratory validation on a limited dataset, and broader validation with larger sample sets is still required.