Study on seismic dynamic failure of tunnel-soil-frame structure system through shaking table test and numerical simulation

To investigate the dynamic response characteristics of an underground structure–soil–aboveground structure interaction system under seismic loading, a shaking table test based on a metro section in Zhengzhou was conducted. A three-dimensional numerical model including both aboveground and underground structures was developed in ABAQUS. Four typical working conditions and three representative seismic motions were selected. Key response parameters such as acceleration, shear force, bending moment, and inter-story drift angle of the tunnel and frame structures were systematically analyzed. Results reveal strong dynamic coupling between the aboveground and underground structures. The integrated system significantly alters seismic wave propagation and energy distribution, showing a “near-field attenuation and far-field amplification” pattern, with an influence range up to 33.3 times the structural width. The aboveground frame dissipates part of the seismic energy near the site center, reducing peak acceleration by 2.75 % and delaying amplification at around 86 m and 92 m. Meanwhile, the tunnel causes non-uniform far-field amplification. With aboveground structures, tunnel acceleration becomes more spatially uneven, with a 5 % increase in peak response at the center section, indicating stronger localized effects due to wave path disturbance. In terms of internal forces, without the aboveground frame, the tunnel shows pronounced nonlinear responses, with peak shear force and bending moment reaching ±4 × 10⁴ N and ±1 × 10⁵ N m, revealing potential weak zones. With the frame present, although peak shear increases to ±1.2 × 10⁵ N, force distribution becomes smoother and more delayed, reflecting an “energy storage–release” effect that buffers seismic impact. Inter-story drift angles in the aboveground structure are also affected by the tunnel; the maximum drift increases from 0.021 % to 0.030 %, following a “larger at the bottom, smaller at the top” pattern, indicating reduced stiffness in lower stories and an enhanced soft-story effect. Under RSN32 input, resonance with low-order modes causes a significant increase in bottom-story response and more variability across different input motions, highlighting increased sensitivity to seismic frequency. Overall, the dynamic coupling in such systems profoundly affects seismic response in dense urban areas, and seismic design should account for these interactions to enhance structural resilience.