Experimental study and simulation for unfrozen water and compressive strength of frozen soil based on artificial freezing technology

Abstract

Artificial ground freezing is widely used in various projects, such as mine, bridge, and subway tunnel construction and emergency repair, and has achieved good engineering results. The freezing strength and freezing effect based on artificial freezing technology have become a hot research topic. However, the relationship between the freezing strength and unfrozen water content under extremely low-temperature conditions has not been sufficiently studied. In view of this, a series of laboratory experiments were carried out. The unfrozen water characteristics and compressive strength of soil at ultralow temperatures from 0 °C to −80 °C were studied by nuclear magnetic resonance (NMR). The applicability of a genetic algorithm-back propagation (GA-BP) prediction model to unfrozen water content was discussed. The results show that there is a very small amount of unfrozen water present at −80 °C, with a content of approximately 0.1%. The GA-BP network prediction model can be used to predict the unfrozen water content of soil at very low temperatures. The compressive strength of frozen soils is significantly affected by temperature and unfrozen water content. The compressive strength of frozen soil is directly proportional to the absolute value of the soil temperature. The unfrozen water content and compressive strength follow a power function. As the temperature decreases, the stress–strain relationship of frozen soil is of the elasto-plastic type. Interestingly, soil at temperatures down to −80 °C show significant brittle failure. This paper supplements the basic theory of extremely-low-temperature frozen soil and quantitatively determines that the unfrozen water content affects the frozen soil strength. These conclusions are highly significant and can provide a theoretical reference for practical engineering applications, especially artificial freezing engineering.