Experimental Study of the Seismic Response of Tabriz Subway Tunnel in Dry Sand

A series of 1 g shaking table tests were performed to investigate the response of Tabriz subway tunnel, a circle-type tunnel embedded in dry sand, under sinusoidal excitations. In prototype, the subway tunnel with 9.2 m diameter and 0.35 m thickness was embedded in a soil layer. Two reduced-scale 1 g shaking table models, designated as FF and SF, were constructed in 1/45 scale. The FF was constructed to study the seismic response of the soil layer in free field condition, while the SF model includes a subway tunnel to study its seismic response during different excitations.
The shaking table of Tabriz University with a platform of 3m×2m and one-degree of freedom was used to induce the desired excitations to models. The table can carry up to 6 tones and can reach acceleration levels up to 1.5 g with peak displacements of ±100 mm. A laminar shear box was designed in Tabriz University that includes 20 aluminum frames with dimensions of 1320×814×860 mm (L×H×W). In order to reduce the friction between the layers and simulate the displacement of soil layers, ball bearings were used between two adjacent frames. In this box type, the lateral boundary effect on the seismic response of the soil layer is reduced.
The simulation laws for 1 g shaking table tests were utilized in the current study. Based on the simulation laws and the size of the laminar shear box, the prototype to model scale factor was considered to be 45. Therefore, the tunnel model was constructed by aluminum alloy with a diameter of 195.5 mm and thickness of 1.5 mm.
Uniform dry sand provided from Qomtapeh was used in this study. During the construction, the tunnel and all the embedded instruments were placed in the model. To avoid any interaction of the tunnel with the laminar shear box, the tunnel was selected shorter than the box width. Two PVC circular plates were placed at both the tunnel ends to avoid the sand entrance into the tunnel model. To simulate the effects of friction on the soil–tunnel interaction, the outside surface of tunnel was covered by sand particles using epoxy coating. For reaching the same target relative density (Dr=65%) during the construction of models, the bulk unit weight was controlled to be constant for all layers. Seven strain gauges were installed on the tunnel surface to monitor the behavior of the tunnel. Five accelerometers were placed in different levels of the model to record the acceleration in the soil. Besides, two LVDTs were placed on the top of the model to measure the soil surface settlement. A 32-channel dynamic data logger was used to record and transfer all the measured data to a personal computer.
Two types of excitation were applied to the models by shaking table: I) irregular waves with high frequency content and low amplitude to determine the natural frequency of the models, and II) harmonic waves with low frequency content and high amplitude to study the seismic response of the tunnel. Two peak ground accelerations of 0.35 g and 0.50 g with frequencies of 1, 3, 5 and 8 were applied to the models at this stage.
The recorded data highlighted significant aspects of the dynamic response for the above type of underground structures:
- The results show that the ground response of the free field model is different from the tunnel-soil model and the natural frequency of the free field is slightly larger than soil-tunnel model. This indicates the effect of the tunnel on the applied frequency to the system.
- The recorded horizontal accelerations at different levels indicate that accelerations are amplified towards the soil surface and the tunnel acts as an obstacle against the propagation of shear waves upward.
- According to the results, the dynamic response of circular tunnels can be split into two stages: transient stage and steady-state cycles. During the transient stage, which lasts for the first few cycles, the tunnel reaches a dynamic equilibrium configuration. The transient stage is followed by the steady-state cycles, during which the forces in the tunnel lining oscillate around a mean value.
- For all tests, bending moments and lining deformations increase by increasing in maximum base acceleration, but the location of the highest and the lowest amounts stays the same.
- According to the results, for A=0.35 g, maximum bending moment is constant or reduces a little by increasing frequency; however, for A=0.50 g, maximum bending moment reduces sharply by increasing of the loading frequency.
The results show that in the earthquakes with high PGA, the dynamic bending moments caused in the tunnel lining are larger than cracking moment that can lead to a serious damage to the lining in combination with other loads.