Uneven frozen heave can cause some damage to the canal lining structure, particularly in the frozen soil in the north cold region of China. The frost-heaving damage of canal lining is closely related to the lining-frozen soil contact action. Some relations of contact consist of the contact surface, such as the freezing constraint, and friction slip. The lining stress can also depend directly on the separate conditions. However, the current simulation of the lining-frozen soil contact is mostly focused on the steady state. It is still lacking in the experimental verification of the mechanical parts. Taking the integral U-shaped canal frost heaving damage experiment in the Hetao irrigation area as the prototype, this research aims to construct the canal frost heave model with/without the contact relationship. Finite element (FE) software ABAQUS was selected to simulate the frost-heaving damage process of the model. Two models were evaluated to compare the experimentally monitored data from the subsoil temperature field, layered frost heave displacement, and soil stress. A systematic analysis was made to determine the stress state and damage mechanism of the canal lining. The experimental and simulation were integrated to clarify the influence of lining-frozen soil separation and sliding process on the frost-heaving damage of the canal. The results show that the simulation considering the contact model was more consistent with the field test conditions, compared with no contact. There were the basically same distributions of temperature field between the experiment and simulation with/without contact. Specifically, the simulated freezing depths of the shady and sunny slopes increased by 2.44% and 1.68%, respectively, whereas, the freezing depth of the canal bottom decreased by 4.26%, compared with the experimental. Small errors were found between the experimental and simulated subsoil and lining normal frost heave displacement. The consideration of contact posed a weak limiting effect on the canal slope subsoil freezing. The average relative errors were 7.1% and 13.9% between the simulated and experimental data with/without contact, respectively. Moreover, the simulated stress peak at the side slope reached 3.3 times of the contact model. The voids were generated between the canal bottom lining and the foundation soil in the experimental and the considered contact model. By contrast, there were continuously increasing tensile stresses in the simulations without considering contact. The separation and sliding process between the lining-frozen soil contact surface were visualized dynamically to combine the considered contact model with the experiment. The lining at the bottom of the canal was the first to separate from the subsoil after the onset of frost heave, due to the inhomogeneity of the subsoil frost heave. There was no transferred contact stress. After that, relative sliding occurred between the lining and the frozen soil at the side slope. As such, the originally restrained lining stress was released. More importantly, the frost heave displacement of the lining was still increasing in the process of stress release. The sliding process changed the stress development trend from the "force increase" to the "force release". The peak values of normal stress were reduced by 903% and 164% on the upper and lower surfaces of the lining under the releasing force of the simulated slip in the considering contact model, compared with the simulation without considering contact. Meanwhile, the peak value of tangential force was reduced by 248% on the lower surface. All peak stresses occurred on the slightly shady slope at the bottom of the canal, where the frost-heaving damage was more likely to occur. The finding can provide a strong reference for the design and structural optimization of the frost heave prevention of canals in cold regions.