Foamed lightweight soil (FLS) subgrades in cold regions are highly susceptible to temperature-induced shrinkage cracking, compromising long-term durability. This study investigates the coupled effects of silty clay content, temperature cycling, and service temperature range on FLS shrinkage behavior and crack evolution through laboratory tests and finite–discrete-element method (FDEM) simulations. Experimental results show nonlinear accumulation of shrinkage strain under cyclic thermal loading, with 20% silty clay significantly enhancing stability. Analysis of variance identifies silty clay content as the dominant factor over temperature cycling frequency. FDEM simulations reveal strain localization and crack initiation under heterogeneous temperature fields, with cracks evolving in three stages: initiation, rapid propagation, and stabilization. Lower minimum service temperatures play a dominant role in accelerating crack development and structural deterioration. Integrating experiments and simulations clarifies how key material and environmental factors govern FLS thermal shrinkage, providing a preliminary reference for crack-resistant design and long-term reliability assessment of FLS subgrades in cold regions.
Honghuan Cui, Hongyan Guo, HuiZhen Wu et al.· Journal of cold regions engi...· 0 citations
To gain a deeper undestanding of the dynamic characteristics of foam lightweight soil (FLS) subgrades in cold regions, resonant column tests were conducted to investigate the effects of wet density, clay content, and freeze–thaw cycles on the dynamic properties of FLS. The results indicate that the relationship between the dynamic shear modulus ratio (
G
/
G
0
) and shear strain (
γ
) of FLS exhibits a consistent pattern:
G
/
G
0
decreases slowly at low strain levels, with the rate of attenuation increasing significantly once a certain strain threshold is reached. Furthermore, the decay rate of
G
/
G
0
with increasing
γ
gradually increases as the wet density of FLS decreases and the clay content increases. In this study, the
G
/
G
0
–
γ
relationship substantially conforms to the Martin–Davidenkov model, which can be used to predict the variation trend of
G
/
G
0
beyond the strain range of the resonant column tests, with the relevant formulas and parameter values ultimately provided. Based on the Hardin model, a multi-factor predictive model for the maximum shear modulus (
G
0
) was developed, incorporating wet density, clay content, and freeze–thaw cycles, with its validity verified using a BP neural network algorithm. The research findings contribute to the accurate prediction of the dynamic response of FLS subgrades under varying conditions.