Loess exhibits typical water sensitivity and dynamic vulnerability. The combination of rainfall and earthquakerelated issues presents a complex disaster process, posing a significant threat to the infrastructure in the loess region. A cyclic simple shear test was conducted on undisturbed loess under a constant vertical stress ranging from 50 to 300 kPa, comprising three stages(C-W-D): consolidation, pre -humidification, and cyclic loading. The deformation behavior under humidification and cyclic loading was analyzed. The microstructure evolution of loess during three stages was examined using scanning electron microscopy (SEM) and mercury intrusion porosimetry (MIP). Results indicated that: (1) Cyclic deformation increased with the rise in vertical stress sigma v , humidification parameters S w , and dynamic shear stress amplitude gamma d . The sensitivity of cyclic deformation to sigma v and gamma d gradually decreased as S w increases. (2) The total deformation in the three stages correlated positively with S w , sigma v , and gamma d . The proportion of humidification deformation and cyclic deformation in the total deformation was largely unaffected by sigma v , with cyclic deformation gradually dominating as gamma d increases. (3) The prehumidification stage promoted aggregates and the formation of numerous intergranular pores. Cyclic loading mainly leads to the change of pore structures, forming obvious seismic damage area. Based on the relationship between humidification deformation and cyclic deformation, a loess deformation prediction model was proposed, which can comprehensively consider S w , sigma v , and gamma d . This can provide a theoretical reference for earthquake disaster prediction in collapsible loess areas.

Humidity diffusion can stimulate soil deformation. However, evidence for triggering water-induced excessive floor deformation in underground structures remains elusive. This study investigated the issue by solving a case of a loess tunnel affected by floor heave. First, the study tracked the development of field damage during tunnel operation and examined the potential influence of base moistening on inverted arch uplift. Subsequently, a largescale field tunnel model test was conducted to analyze the moisture distribution, stress, and deformation development at the tunnel base during the humidification process. Finally, the mechanical properties of the base loess at the humidification stages were tested to assess the degradation caused by moistening. Results showed that the moisture distribution at the tunnel base changes from W- to U-shaped during long-term humidification. Moreover, after base humidification, the soil pressure at the arch foot initially decreases sharply and then increases, while the soil pressure at the inverted arch continues to increase. Furthermore, the lining at the arch foot shows an increase in compressive stress, while the inverted arch shows an increase in tensile stress. The differential settlement between the arch foot and inverted arch widens, transitioning to uplift deformation of the inverted arch and ultimately causing floor heave. Laboratory tests showed that the floor heave is primarily caused by the deterioration of the mechanical properties of the loess resulting from humidity diffusion in the tunnel foundation. A time-dependent floor heave model was established by combining the water content, shear strength, compressive strength, and compressibility of the tunnel-base loess, and its feasibility was verified. The model exhibited a sequential decrease in the influence of the internal friction angle, compressive strength, cohesion, and compression coefficient on the floor heave. The findings of this study are considerably important with respect to uncovering the mechanism of floor heave during the operation of loess tunnels and advancing the prediction of damage.