A new method to determine the segmentation of pore structure and permeability prediction of loess based on fractal analysis

In previous research concerned on mercury intrusion porosimetry (MIP), the fractal curves of six fractal models tended to have segmentation characteristics. However, most of these models are rarely applied to loess pores and often have inconsistent results. Therefore, this paper used a combination of techniques to study the loess pore structure and permeability properties, including routine tests, MIP analysis, and scanning electron microscope (SEM) image analysis. The results indicate that different models have various goodness of fit and fractal dimensions. Among them, the Zhang-Li model had the best fractal features. However, there are certain similar segmentation diameters between different fractal curves, suggesting that the segmentation diameters are determined by the pore structure, rather than by the fractal models. By using a new modeling detection method, the paper determined that the segmentation diameters are centrally distributed around two diameters: d1 (7.08 mu m) and d2 (0.035 mu m). The pore structure is divided into three regions based on these two diameters. Furthermore, the paper proposes a new parameter lambda, the relative content for pores with diameters larger than d2 (0.035 mu m), which is used to correct effective porosity and permeability prediction. The traditional K-C equation is only 68.41% of the goodness of fit in predicted and measured values. After correcting the porosity, it rises to 77.99%, while the correcting prediction equation based on the pore-size distribution function is 80.46%. These results prove that our new segmentation method for loess pore structure and effective porosity correction is reasonable for permeability estimation.

Automated summary

This study investigated the pore structure and permeability of loess using various techniques, including routine tests, MIP analysis, and SEM image analysis. The results showed that different fractal models had different goodness of fit and fractal dimensions, with the Zhang-Li model performing the best. A new method for detecting segmentation diameter based on the pore structure was proposed, which divided the pore structure into three regions. A new parameter lambda was also introduced to correct the effective porosity and permeability prediction. The corrected prediction equation based on the pore-size distribution function showed the highest goodness of fit.