Quantitative Study On Three-dimensional Microstructure Of Loess

As a special bulk material,the microstructure of loess has been a hot topic of research in the field of loess because its special properties are often closely related to its microstructure.On the one hand,the most intuitive microstructural feature of loess is the pore structure,which as a water and gas transport channel,directly affects the seepage of loess and thus its hydrological properties;on the other hand,the size,morphology,type of cementation,arrangement and combination of the soil particles and colloids that make up the loess directly affect the mechanical and hydrological properties of the loess body.The skeletal particles and pores of loess,as two important parts of its microstructure,are inextricably related to each other.The study of the microstructure of loess from these two parts can help to further understand the hydro physical and mechanical properties of loess,and even reveal the mechanism of loess disaster from this perspective,so it has important theoretical and practical significance.In this paper,taking the western loess of Lvliang mountain area as an example,a series of research have been carried out on the microstructure of loess through the preparation of loess microstructure samples,3D image acquisition,processing and analysis,and the main results achieved are as follows.The preparation process of loess samples used in the gluing and grinding process and CT scanning process was systematically studied,and a perfect set of loess 3D image acquisition process was formed.The two types of representative technical solutions were also compared,and the emphasis of different solutions in microstructure study was discussed.Based on the acquired microscopic images,an operation procedure suitable for 3D image pre-processing of loess is determined by comparing the commonly used image processing algorithms.Based on this procedure,the separation of soil skeleton and pore space,extraction of clear boundary particles,and detection of connected pore space can be well accomplished.This procedure is not limited to a particular software platform but is also suitable for other digital image processing software.The equivalent diameter,sphericity,anisotropy and specific surface area of loess particles were statistically found: the sphericity of soil particles decreases with increasing equivalent diameter;the anisotropy of small particles has a large variation interval,while the anisotropy of large particles gradually becomes larger overall but the variation interval is not large,and the analysis suggests that this occlusion between large particles with long bias contributes to the strength of the soil body.Based on the physical model method,the specific surface area of the soil was predicted using volume and surface area data,and the predicted results were in good agreement with the results measured by laser particle size meter.The connected pores in the uniform dense zone were counted,and the pore and pore throat relationships of the connected pores were analyzed.The results showed that the connected pore volume of the six soil samples from the Lvliang Mountain area was above 99.6%,and only 0.1%-0.3% of the pores were isolated pores.Although the pore volumes differed to different degrees,the frequency distribution of the pore throats connected by the pores showed good statistical consistency,and all of them had a typical right-skewed distribution.From the perspective of connectivity,such highly connected core node pores will have a significant impact on the permeability of the soil,and the seepage channels of such core pores must be the dominant seepage channels in the soil,thus proving the existence of dominant seepage channels in the loess at the microscopic scale from the mathematical statistical point of view.The objective fact of the existence of dominant seepage channels in loess at the microscopic scale is thus proved from the mathematical statistics.