Long-term floor heave analysis of loess tunnel based on modified Green-Ampt infiltration model

Floor heave is a common water-induced issue of loess tunnels. However, the mechanism of tunnel floor heave induced by repeated rainfall or irrigation remains poorly understood. This study analyzes moisture variation at the tunnel base and invert deformation through a 15-year survey of an operational loess tunnel. An improved Green-Ampt model was developed to capture the effects of interannual repeated infiltration on water content, permeability, and matric suction at the tunnel base. The model was validated using soil column infiltration tests. The calculated wetting front depth was incorporated into the tunnel numerical simulation to analyze the development of floor heave and assess its several influencing factors. Results show the base average water content at the tunnel base increases exponentially with operational years. After 15 years, over 50% of the monitored sections reach saturation, leading to significant invert uplift. Based on the plastic limit, saturation water content, and liquid limit of loess, floor heave is categorized into four levels with deformation thresholds at 10, 25, and 50 mm. The modified Green-Ampt model is suitable for repeated infiltration conditions and offers practical guidance. Neglecting hydraulic deterioration due to repeated infiltration may lead to significant prediction errors. Numerical simulations reveal that the inverted arch uplift height, width, and height-to-width ratio increase linearly with the rise in the number of infiltration events. The uplift deformation follows Gaussian distribution in both cross and longitudinal sections, forming a Λ-shaped floor heave. The influence of infiltration duration, water content, horizontal infiltration, asymmetric infiltration, burial depth, and the number of infiltration events enhanced successively on floor heave. This model facilitates the prediction of floor heave under dynamic hydrological conditions during long-term tunnel operation.