The Ili River Valley in Xinjiang, China, is a typical seasonal frozen area where loess landslide disasters have become increasingly common during the freeze–thaw periods in recent years. This study analyzed the macroscopic mechanical strength and microstructure changes of the Ili loess under different freeze–thaw cycles (FTCs) through the post-freeze–thaw triaxial compression test on the unsaturated soil in laboratory. Apart from the scanning electron microscopy (SEM), and the nuclear magnetic resonance (NMR), the macro–micro correlation analysis and the cluster-principal component analysis were applied for the theoretical discussion. The results indicated that the cohesive force of the loess exhibits an initial decreases, followed by the increases, and eventually keep stable after various FTCs, while the internal friction angle showed the opposite developing trend before the final constant. Similar to the strong correlation between the cohesive force and the particle abundance, the internal friction angle is also closely related to the abundance and orientation fractal dimension of the loess particles. However, the principal component analysis results showed that cohesive force strongly correlates with the average maximum pore size and the pore size fractal dimension, for which the internal friction angle most strongly affected by the average maximum particle size. The possible reason is that the extracted principal components represent a class of microscopic parameters with the same or similar change trend, although there may be a certain offset between them. The mechanical deterioration of loess is attributed to the repeated frost heaving force and the migration potential caused by FTCs. The alterations of the microstructure accelerated the deterioration of the macroscopic mechanical properties of the loess, which further widens the understanding of the mechanism behind the deterioration of loess mechanical strength in the Ili River Valley under FTCs, and contributes to the prevention and management of the local landslide disasters.
Read full abstract