Application of the Frozen and Unfrozen Soil model to modelling effects of freeze-thaw on low-volume roads

Freezing and thawing of soils are of the utmost importance in the design of infrastructure in seasonally cold regions. Cycles of freezing-thawing can severely affect the performance of thinly paved low-volume roads, causing frost heave during the winter and thaw weakening in the spring. Thus, a designer must have an understanding of freezing effects on infrastructure and adequate design tools to deal with them. Freezing of soils entails several coupled processes, necessitating advanced analysis tools such as numerical modelling. This paper applies the Frozen and Unfrozen Soil material model for the finite element program Plaxis 2D into a case study of freezing and thawing induced deformation of a low-volume road on dry-crust clay subgrade in Vesilahti, Finland supported by a literature review and a laboratory study. The Vesilahti test site monitoring results of frost heave, temperature and groundwater level were used in this study to simulate the frost heave of a pavement with shallow structural layers. In addition, soil sampling and laboratory tests were performed to determine the Vesilahti dry-crust clay material properties at the Aalto University. These were combined to simulate the frost heave and thaw settlement measured on the site and to study the feasibility of the material model in replicating actual measured frost action. The road deformation at the measurement times was replicated qualitatively well and the model performed mostly according to expectations. However, the multitude of parameters, complexity of the coupled phenomena modelled and the black box nature of the model implementation in Plaxis make the evaluation of calculation errors tough and time-consuming. Further research on the model could include the implementation of various user-friendly improvements in Plaxis, development of cryogenic suction-and temperature-controlled laboratory testing for the model parameter determination and better utilization of the model's predictive capabilities by studying continuous frost heave measurements.