Study on dynamic characteristics and dynamic constitutive relationship of inclined layered soil in western Sichuan under freeze-thaw cycles

Frequent earthquakes and significant freeze-thaw cycles during winter in western Sichuan introduce complexity into the dynamic characteristics of sloping layered soil slopes, thereby posing a threat to the safety of infrastructure. Existing studies predominantly focus on horizontal homogeneous soils, yet there is a paucity of research regarding inclined bedded soils that are commonly encountered in practical engineering scenarios. This study investigates the Egang Highway inclined layered slope located along National Highway 317 as the research subject. Through systematic laboratory dynamic triaxial tests, the effects of different sequences (GCM, GMC, MGC, CGM), confining pressures (ranging from 100 to 300 kPa), and freeze-thaw cycles (0 to 8 cycles) on the dynamic characteristics of soil are analyzed. The results indicate the following: (1) The sequence exerts a substantial influence on the skeleton curve by altering the efficiency of soil energy conduction. An increase in confining pressure shifts the curve toward higher stress levels. Dynamic stress exhibits a rapid decline during the initial freeze-thaw cycles and stabilizes after four cycles. (2) The dynamic elastic modulus ( E _d ) demonstrates nonlinear decay with respect to dynamic strain. Higher confining pressures enhance E _d , whereas freeze-thaw cycles reduce E _d by damaging the pore structure. The damping ratio increases with an increasing number of freeze-thaw cycles. (3) Based on the Hardin-Drnevich model and the power law function, a dynamic elastic modulus attenuation model incorporating sequence, confining pressure, and freeze-thaw cycles was developed. The fitting parameter R ² > 0.85 confirms the model's applicability. These findings provide a theoretical foundation for analyzing the dynamic response and disaster prevention of slope engineering in the high-cold earthquake-prone region of western Sichuan. However, further optimization of the model with additional variables remains necessary.