2. Key Laboratory of HighSpeed Railway Engineering, Ministry of Education, Chengdu 610031, China
With the construction of highspeed railway in China, a large number of bridge foundations are inevitably built on steep slopes, especially in highintensity seismic areas. Therefore, how to carry out reasonable seismic study on slope for such projects is particularly important.
Beena K. S. et al. conducted a series of model tests to study the problem of laterally loaded pile on sloping ground (Beena K. S. et al., 2018). Based on the results, the bending moment variation along the embedment depth of pile, obtained from strain, using strain gauges pasted along the depth of the pile were quantitatively studied. Kobayashi H. et al. conducted the hybrid vibration experiment on seismic behavior of bridgesoil system, and examined the applicability of hybrid vibration experiment to study seismic response of bridgesoil system (Kobayashi H. et al., 2002). Based on shaking table model experimental studies, Lei Da et al. revealed that the filtering effect existed in soil and was affected by the dynamic constraint conditions. The amplitude is strengthened around natural frequency and weakened on other frequency band in Fourier spectrum (Lei Da et al., 2017). The recorded data and observations of the centrifuge models showed that the PGA amplification coefficient, the acceleration response spectra and the settlements at slope crest were decreased to some extent when the lower bedrock deposit slope was reinforced with the stabilizing piles (Sun Zhiliang et al., 2017).
Because of the accidental factors of earthquake, the shaking table model test is an effective method to study the failure characteristics of the slope and the dynamic response of structure (Lai Jie et al., 2015; Ye Hailin et al., 2012; Wu Dong et al., 2014). Based on the research project of ChengduLanzhou railway project, a shaking table model test of antislide pile protecting bridge foundation on slope was designed. Under the effect of Wenchuan earthquake wave, the dynamic response of antislide pile and bridge foundation structure and the failure mode of landslide were studied.
1 DESIGN OF SHAKING TABLE MODEL TESTThe model test was carried out on a unidirectional electrohydraulic shaking table in the road and railway engineering laboratory of Southwest Jiaotong University, which has a carrying capacity of 25 tons. A rigid steel box with inner space of 3.7m×1.5m×2.1m was adopted in the test. In order to reduce the boundary effects (Goodings D. J. et al., 1996; Zhou Zhong et al., 2010), the inner surface of the test box was covered with a 5cmthick layer of polyethylene foam board.
1.1 Design of Similarity RelationsAccording to the dimensional analysis theory (Yang Junjie, 2005), geometric dimension l, cohesion c, internal friction angle φ, severe γ, Poisson ratio ν, gravity acceleration g, time t, frequency ω, displacement s, strain ε, stress σ, and seismic acceleration a are taken as the main similarity indexes. The similarity criterions are listed as follow:
$\begin{array}{l} \frac{{{C_c}}}{{{C_l}{C_\gamma }}} = 1\;\;\;\;{C_\varphi } = 1\;\;\;\;\frac{{{C_E}}}{{{C_l}{C_\gamma }}} = 1\;\;\;{C_\mu } = 1\;\;\;\;\frac{{{C_{vs}}}}{{{C_l}{C_\omega }}} = 1\;\;\;\;\frac{{{C_g}}}{{{C_l}C_\omega ^2}} = 1\;\;\;\;\frac{{{C_a}}}{{{C_l}C_\omega ^2}} = 1\\ {C_\omega }{C_{td}} = 1\;\;\;\;{C_s} = {C_l}\;\;\;{C_\theta } = 1\;\;\;\;{C_\varepsilon } = 1\;\;\;\;\;\frac{{{C_\sigma }}}{{{C_l}{C_\omega }}} = 1\;\;\;\;\;\;\;\;\frac{{{C_v}}}{{{C_l}{C_\omega }}} = 1\;\;\;\;\frac{{{C_\alpha }}}{{{C_l}C_\omega ^2}} = 1 \end{array}$ 
Since the dimension, weight and acceleration are the main controlling factors in similarity relations, the similar constants of physical quantities can be derived based on the Buckingham π theorem and are tabulated in Table 1.
The model test was based on the slop of Zhengjiangguan bridge in the ChengduLanzhou railway. As depicted in Fig. 1, the bedrock is carbonic slate and the landslide is gravel soil. The bridge foundation is located at the foot of the slope and is reinforced by the rear antislide pile. According to the similarity theory, a simplified reducedscale model of 1:40 was made in Fig. 2. The size of antislide pile is 10cm×8cm×100cm. The piles of bridge foundation with dimensions of 100cmhigh and 3cmindiameter were arranged in three rows. Both the antislide pile and the bridge foundation were made of micro concrete precast and maintained the same reinforcement ratio as the original model.
The landslide was simulated by coarse sand and gravel with a density of 2100kg/m^{3}, friction angle of 35° and moisture content of 8.7%. The bedrock was composed of silty clay, cement, coarse sand and gypsum, which were filled in layers (Liu Jingwen et al., 2014; Zheng Tong et al., 2016). In order to test the dynamic characteristics of the slope and structure, the horizontal acceleration sensors, earth pressure cells, strain gauges were placed in the central area of the slope, as shown in Figs. 3 and 4.
The effect of the peak acceleration on the seismic response of slope was studied by applying Wenchuan seismic wave, whose peak value was gradually increased to 0.1g (Fig. 5). With the aim of detecting the natural frequency of the model, the white Gaussian noise was loaded before formal test. The particular loading scheme is shown in Table 2.
The side facing the top of the slope is regarded as the backside while the side facing the foot is defined as the frontside. As is shown in Fig. 6, the peak value of the dynamic earth pressure from the backside of the antislide pile is distributed in a triangular shape, which increases constantly with the peak acceleration. The result can be observed that the maximum value appeares near the sliding surface, indicating that the progressive failure of the slope develops continually along the potential sliding surface.
The peak value of the dynamic earth pressure from the backside of the bridge pile is presented as a parabolicshaped distribution, as shown in Fig. 7. It can be seen that the dynamic earth pressure changes slightly above the sliding surface. When the acceleration reached 0.4g, the peak value of the soil pressure at the pile top increases rapidly, which demonstrates that the landslide thrust on antislide piles is simultaneously increased.
As shown in Fig. 8, the dynamic earth pressure above the sliding surface is distributed in an inverted triangular shape, while below that is in a triangular shape. Under the action of seismic waves, the dynamic response of the soil is enhanced and the surface amplification effect is thus generated on the front slope of the bridge. With the increase of vibration acceleration, this effect becomes increasingly acute.
The earthquake may trigger landslides and deformation of antislide piles. Fig. 9 shows the peak value of dynamic strain varying with vibration acceleration. It can be seen that the peak value of the dynamic strain of the loadbearing section is in a parabolic shape, and the maximum value is mainly distributed near the slip surface.Such observing data is consistent with the experimental results of the earth pressure distribution.
Fig. 10 illustrates the peak value of dynamic strain of bridge pile varying with the seismic wave. As can be observed, the maximum strain occurs at the top of the pile, among which the strain at the top of the rear pile is larger than that of the front pile. This is because with the increase of seismic acceleration, the residual sliding force transmitted by the antislide pile increases continuously, and the earth pressure acting on the bridge foundation increases, resulting in a significant increase in the dynamic strain at the top of the pile.
Based on the same seismic waves input, acceleration of slope has been monitored, and the acceleration amplification coefficient variation rules with height are summarized. By analyzing the horizontal acceleration of the surface(A0A3A4A5A9A12) and the inside(A0A6A7A8A9) of slope, it is found that the PGA amplification factor of the surface is greater than the inside, as shown in Fig. 11. A complex seismic wave field is formed due to the reflection of seismic waves on the empty face, the PGA amplification factor increases nearby in empty face, and the landslide of the slope is characterized by shallow sliding.
As shown in Fig. 12, the PGA amplification factor decreases with the increases of the input seismic acceleration peak. Due to the obvious nonlinear characteristics of deposit slope, the natural frequency of interaction system decreases and the damping ratio of soil increases with the increase of the shaking.
Under the action of Wenchuan earthquake wave, the soil is disturbed by repeated shear. When the peak acceleration of seismic wave ranges from 0.1g to 0.3g, the slope deformation is not obvious. As the 0.4g seismic wave is applied, a few tensile cracks begin to appear on the side slope (Fig. 13). When the peak value reaches 0.5g, the slope crack develops rapidly. Due to the reinforcement effect of the antislide piles, the sliding leads to excessive uplift deformation of the slope surface. At the same time, obvious tension cracks are produced near the bridge foundation, and a large amount of gravel accumulation appeared at the foot of the slope.
(1) Under the action of seismic load, the landslide deforms along the sliding surface, and the earth pressure of antislide pile and bridge pile increase with the seismic acceleration. The increase and amplitude of soil pressure of bridge piles are less than that of antislide piles.The result shows that antislide piles play an effective role in slope reinforcement.
(2) The peak value of dynamic strain of antislide pile is near the anchorage point, while that of bridge pile is at the top of the pile. To meet the requirement of the seismic design of structure, local reinforcement should be carried out according to the characteristics of dynamic response of structure to ensure safety.
(3) The PGA amplification factor of the surface is greater than the inside. And the natural frequency of interaction system decreases and the damping ratio of soil increases with the increase of the shaking, the PGA amplification factor decreases with the increases of the input seismic acceleration peak.
(4) The failure process shows that the deformations of upper and lower parts of landslide are different. The sliding mass which mainly exhibits shallow slide develops along the sliding surface under the earthquake action, resulting in a large number of tension cracks and excessive slope uplift. At the foot of the slope, a large amount of gravel soil is accumulated.
Beena K.S., Kavitha P.E., Narayanan K.P.Beena K.S., Kavitha P.E., Narayanan K.P. Model studies on laterally loaded piles in sloping clay bed[J]. Indian Geotechnical Journal, 2018, 48(1): 114124. DOI:10.1007/s4009801702484 
Goodings D.J., Gillette D.R.Goodings D.J., Gillette D.R. Model size effects in centrifuge models of granular slope instability[J]. Geotechnical Testing Journal, 1996, 19(3): 277285. DOI:10.1520/GTJ10353J 
Kobayashi H., Tamura K., Tanimoto S.Kobayashi H., Tamura K., Tanimoto S. Hybrid vibration experiments with a bridge foundation system model[J]. Soil Dynamics and Earthquake Engineering, 2002, 22(9/12): 11351141. 
Lai Jie, Zheng Yingren, Liu Yun, Li Anhong, Liu HongweiLai Jie, Zheng Yingren, Liu Yun, Li Anhong, Liu Hongwei. Shaking table test studies and numerical analysis of doublerow antislide piles under earthquake[J]. Journal of Central South University (Science and Technology), 2015, 46(11): 43074315 (in Chinese with English abstract). 
Lei Da, Qi Zhihui, Jiang Guanlu, Wang Zhimeng, Li AnhongLei Da, Qi Zhihui, Jiang Guanlu, Wang Zhimeng, Li Anhong. Shaking table test study on dynamic characteristics of bridge foundation reinforcement on slopes[J]. Earthquake Research in China, 2017, 31(3): 403413. 
Liu Jingwen, Huang Bo, Deng Hui, Ling Daosheng, Guo Xiaoguang, Gao MeibengLiu Jingwen, Huang Bo, Deng Hui, Ling Daosheng, Guo Xiaoguang, Gao Meibeng. Shaking table tests and throwing phenomenon of deposit slopes under earthquakes[J]. Chinese Journal of Geotechnical Engineering, 2014, 36(2): 307311 (in Chinese with English abstract). 
Sun Zhiliang, Kong Lingwei, Guo AiguoSun Zhiliang, Kong Lingwei, Guo Aiguo. Centrifuge modeling tests on seismic response of lower bedrock deposit slopes before and after reinforced with stabilizing piles[J]. Chinese Journal of Rock Mechanics and Engineering, 2017, 36(6): 14131423 (in Chinese with English abstract). 
Wu Dong, Gao Bo, Shen Yusheng, Zhou JiameiWu Dong, Gao Bo, Shen Yusheng, Zhou Jiamei. Shaking table test study of seismic dynamic response of tunnel entrance slope[J]. Rock and Soil Mechanics, 2014, 35(7): 19211928 (in Chinese with English abstract). 
Yang Junjie. Similarity Theory and Structural Model Test[M]. Wuhan: Wuhan University of Technology Press, 2005 (in Chinese).

Ye Hailin, Zheng Yingren, Li Anhong, Du XiuliYe Hailin, Zheng Yingren, Li Anhong, Du Xiuli. Shaking table tests on stabilizing piles of slopes under earthquakes[J]. Chinese Journal of Geotechnical Engineering, 2012, 34(2): 251257 (in Chinese with English abstract). 
Zheng Tong, Liu Hongshuai, Yuan Xiaoming, Qi WenhaoZheng Tong, Liu Hongshuai, Yuan Xiaoming, Qi Wenhao. Experimental study on seismic response of antislide piles with anchor cables by centrifugal shaking table[J]. Chinese Journal of Rock Mechanics and Engineering, 2016, 35(11): 22762286 (in Chinese with English abstract). 
Zhou Zhong, Liu Baochen, Yang Xiaoli. Model test study on dynamic responses of loess slope by dynamic compaction. In: Proceedings of the 2010 GeoShanghai International Conference. Shanghai, China: Geo Institute, 2010: 3849.
