Artificial ground freezing has been applied to many geotechnical projects as temporary and permanent ground stabilization. This paper presents numerical formulations implemented in SMAP-T2, which is specially designed for geotechnical applications of heat transfer problems, along with a new efficient method to compute average heat capacity at freezing point. As one of the practical applications, a frozen backfilled egress shaft is presented in this paper. The aim of this example is to evaluate time requirements for freezing and maintaining the frozen shaft, to evaluate freezing front migrations and possible insulation schemes, and to assess the relative advantages of various stemming materials such as pure water and combinations of soil/water mixtures.

The seasonally frozen ground has been focused on as research topics such as the frost heaving under the asphalt road rather than the permafrost ground due to the latitudinal location of Korea. However, the recent construction of the second Korean Antarctic research station, the Jangbogo station and the participation on the development of the natural gas pipeline in Russia arouse the research interests on the behavior of the permafrost ground. At the design process of the geotechnical structures on the permafrost ground, the evaluation of the mechanical characteristics of frozen soil is very important. In the laboratory tests, it is crucial to understand the effects of testing conditions such as loading rate and temperature on the mechanical behavior of frozen soils. Therefore, unconfined compression tests are performed to evaluate the loading rate effects on the mechanical characteristics of frozen granular soils. Based on the experimental results, the loading rate affects both the strength and deformation characteristics of the frozen soils. The ASTM standards on the loading rate of the frozen sands are fast enough for the strength evaluations but not for the stiffness evaluations. Furthermore, the loading rate effects are evaluated on the bearing capacity and deformation of pile installed on the permafrost via numerical analyses with experimental results.

Rate sensitivity is an important characteristic of geomaterials for both saturated and unsaturated soils. However, many constitutive models for unsaturated soil have been constructed within the framework of the rate independent theory. The present study addresses an elasto-viscoplastic constitutive model which considers the effect of suction for unsaturated clayey soil and a soil-water-air three-phase coupled analysis using the elasto-viscoplastic model. The proposed constitutive model adopts the average skeleton stress for the effective stress from the viewpoint of the mixture theory. Hence, it has become possible to construct a model for unsaturated soil starting with a model for saturated soil by substituting the average skeleton stress for the effective stress and introducing the suction effect into the constitutive model. Furthermore, the collapse behavior, which is brought about by a decrease in suction, is described by the shrinkage of the overconsolidation boundary surface, the static yield surface, and the viscoplastic potential surface. A numerical analysis for multiphase materials is conducted within the framework of a continuum mechanics approach through the use of the theory of porous media. The theory is a generalization of Biot's two-phase mixture theory for saturated soil. A soil-water-air three-phase coupled finite element method is developed in the present study using the governing equations for multiphase soil based on the non-linear finite deformation theory. The average skeleton stress is defined as the difference between the total stress and the average pressure of the two fluids and is used in the proposed elasto-viscoplastic constitutive model. A van Genuchten (1980) type of equation is employed as the constitutive equation between the liquid saturation and the suction pressure. Numerical simulations of unexhausted-undrained compression with different strain rates are conducted under plane strain conditions, and the applicability of the proposed method is evaluated with respect to strain localization and the effect of suction.

Predicting the frost depth of soils in pavement design is critical to the sustainability of the pavement because of its mechanical vulnerability to frozen-thawed soil. The reliable prediction of frost depth can be challenging due to the high uncertainty of frost depth and the unavailability of geotechnical properties needed to use the available empirical- and analytical-based equations in literature. Therefore, this study proposed a new framework to predict the frost depth of soil below the pavement using eight machine learning (ML) algorithms (five single ML algorithms and three ensemble learning algorithms) without geotechnical properties. Among eight ML models, the hyperparameter-tuned gradient boosting model showed the best performance with the coefficient of determination (R²) = 0.919. Furthermore, it was also shown that the developed ML model can be utilized in the prediction of several levels of frost depth and assessing the sensitivity of pavement-related predictors for predicting the frost depth of soils.