2. School of Civil Engineering, Southwest Jiaotong University, Chengdu 610031, China
Sand or silt behaves like liquid and loses strength and bearing capacity, i.e., liquefaction, during the intense shaking of an earthquake. In such cases, structures built on or in these soils may be damaged. For example, the railway tracks resting on liquefiable foundation can be damaged by lateral flow, embankment settlement, or failure of foundation when an earthquake strikes (Adalier K. et al., 1998; Jiang Guanlu et al., 2007). Once the damage occurs, it is time-consuming, costly and difficult to repair. Liquefaction countermeasures should be taken to eliminate or reduce the risk of damage to railway routes caused by earthquake-induced liquefaction.
Three types of methods are generally used as the liquefaction countermeasures, including drainage method, soil densification and solidification (Andrus R. D. et al., 1995). The drainage method, mainly referring to the sand drain method and gravel drain pile, can accelerate the dissipation of excess pore water pressures (Andrus R. D. et al., 1995). However, it is found that fine particle in soil may migrate and hence tend to block drainage channels (Chen Guoxing et al., 2015). Soil densification by such as sand compaction pile and dynamic compaction is generally considered a highly reliable measure against liquefaction (Andrus R. D. et al., 1995). It reduces the soil void space, thereby decreasing the volumetric change that would lead to liquefaction. It has been shown that soil densification is able to effectively reduce the extent of lateral spreading (Adalier K. et al., 1998). However, this method can produce objectionable levels of work vibration that would cause excessive level of disturbance to the lifelines (Andrus R. D. et al., 1995). Solidification such as cement deep mixing and jet grouting, is also considered a highly reliable remedial measure against liquefaction. It prevents soil particle movement and provides cohesive strength. It can provide both reinforcement and encapsulation of liquefiable silty and clayey sands. The method is likely to be more costly and slower than the soil densification method, however, it can provide a higher level of improvement (Mitchell J. K., 2008). In addition, this method produces low levels of vibration during installation (Andrus R. D. et al., 1995).
Cement-mixed piles is known as one of the more effective methods for stabilizing the natural earth beneath roads or railway embankments to control stability and settlements (Esmaeili M. et al., 2016). It is formed by cement mixed with the foundation soil using a deep mixing machine. Cement-mixed piles are capable of effectively reducing the liquefaction potential, and improving the shear strength and the bearing capacity of foundation soil (Andrus R. D. et al., 1995; Chen Guoxing et al., 2015). Therefore, the cement-mixed piles can resist settlement and horizontal displacement when an earthquake happens (Mitchell J. K., 2008; Shi Yongqiang, 2010; Kang Huiru, 2016). The cement mixing method as liquefaction treatment has been studied in a few projects in the 1990s (Babasaki R. et al., 1992; Suzuki Y., 1996), followed by increasing applications. These studies include a static load test and a Rayleigh wave test by Cheng Lilai et al. (1998), an experimental program by Le Kouby A. et al., (2010), and a loading test by Esmaeili M. et al., (2016). Theories on cement-mixed piles dealing with building foundation liquefaction were discussed with a shaking table test and was concluded that after the consolidation of cement-mixed piles, the superstatic pore water pressure and the maximum pore pressure ratio of liquefied soil between piles was reduced, and the arrival time obviously lags behind that of unreinforced soil (Zhao Yi, 2016).
A comprehensive literature review reveals that the model test of cement-mixed piles reinforcing liquefiable railway foundation in high seismic intensity region is rare. Shake tables are capable of reproducing the motion of the ground during an earthquake. The shaking table test can simulate the failure process well and study the change laws of dynamic characteristics (Lei Da et al., 2017). Soil seismic liquefaction process can reappear and the reaction of structures can be obtained quickly through monitoring. Shaking table tests have been used to research the mechanics and effects of cement fly-ash gravel columns, geogrid sheet, geosynthetic fiber, biogas method and stone columns dealing with liquefiable foundation or subgrade (Jiang Guanlu et al., 2007; Maheshwari B. K. et al., 2012; He J. et al., 2013; Qu Mengfei et al., 2016). Regarding a cross-section of the Yuxi-Mengzi(YM) railway in earthquake region of Ⅷ degrees seismic fortification intensity as the prototype, a shaking table test is conducted to study the seismic performance of cement-mixed pile solidifying liquefiable foundation.1 SITE CONDITIONS
The Yuxi-Mengzi (YM) railway, with a total length of 142km (from DK0+000 to DK142+000), in the southern Yunnan Province of China, is an important part of the eastern line of the Trans-Asian Railway. The total length of liquefiable soil (i.e., sand and silt) sections add up to 10km. The thickness of the liquefiable layer is generally more than 10m. More importantly, the areas that the YM railway goes through are high-risk seismic zones with seismic fortification intensity of Ⅶ-Ⅷ degrees, design basic seismic acceleration of 0.20g, and 87.3% of the railway track goes through zones with seismic fortification intensity of Ⅷ degrees.
To systematically study the improvement of liquefiable soil by cement-mixed piles, the cross-section DK25+800 of the YM railway (Fig. 1) was selected as the prototype for the shaking table test. The underlying soils of the section are mainly silty clay (1.84m thick), silt (1.88m thick), soft soil (2.63m thick) and silty-fine sand. The center height of the railway embankment is 6.11m. The stuffing is graded gravel. Cement-mixed piles are designed to deal with liquefiable soil. Cement-mixed piles and soil between piles form composite foundation. The gravel cushion has a height of 0.6m, which is laid on the top of the ground. The piles are installed in a triangular pattern (Fig. 2). The diameter of these piles is 0.5m, with the length of nearly 7m and a 1.2m center-to-center spacing.
The test was conducted on the shaking table in the Road and Railway Engineering Lab of Southwest Jiaotong University, China. The size of the platform of the shaking table is 4m×2m, with a carrying capacity of 20t, and a horizontal maximum acceleration of 1.20g. In addition, the horizontal maximum displacement is ±400mm, and the loading frequency is 0.1-30.0Hz.2.2 Model Box
A rigid model box with a dimension of 3.5m long, 1.5m wide, and 1.2m high was applied in the test. The box is made of 10mm thick plexiglass and the external wall is fixed with hollow square steel welded with channel steel. 50mm thick foam board was laid on the inner sides of end-walls which are perpendicular to the vibration direction to reduce the boundary effect of the rigid model box. Vaseline was applied to the inner wall parallel to the vibration direction to reduce the friction between the model soil and model box.2.3 Monitoring Sensors
To meet the objectives and requirements, the monitoring sensors applied were as follows: (1) Pore-water pressure sensors: the measuring range is 100kPa and the resolution ≤ 0.12%F·S; (2) displacement sensors: the measuring range is 5mm and basic error ≤ ±1με.3 LIQUEFACTION TEST 3.1 Similarity Ratio
Being limited to the conditions of the shaking table and the model box, a geometric scale of 1: 10 was adopted. Similarity ratios of other physical quantities were calculated based on the dimensional analysis of the Buckingham π theorem (Buckingham E., 1914). According to similar constitutive relations and materials of the model and the prototype, and the unchanged gravity acceleration during the test, the similarity ratios of other parameters were calculated (Table 1).
The effect and mechanism of cement-mixed piles reinforcing muddy soft soil were studied maturely, so the foundation model was simplified. Silty clay, silt and soft soil were merged into a layer of silt 6.5m thick, so the foundation model was made up of 650mm thick silt and 350mm thick silty-fine sand according to the geometric scale of 1: 10. The grading curves of the soils are presented in Fig. 3. The densities of the silt and silty-fine sand were both 1.73g/cm3, and the water contents were 2%. The thickness of the reinforced foundation model was 700mm.
Six PVC pipes 100mm in diameter were erected at the four corners and the center of end-walls of the model box to be used for water filling at saturation and drainage at the end of the test. The foundation model and piles were made as follows: 100mm of silty-fine sand for each layer was filled in layer upon layer until 300mm thick. The pile positions were set on the silty-sand layer, and PVC pipes of 50mm in diameter used for make piles were embedded. The remaining 50mm of silty-fine sand and 200mm of silt were filled to secure the PVC pipes and tamped with small tools such as a hammer. After all the PVC pipes were pre-embedded, silt was filled to the design height. While filling, the monitoring sensors were embedded in the corresponding positions of the model (see Section 3.3). The cement-mixed piles were made of cement mortar (mixture ratio of early strength agent: cement: sand=1: 10: 110) by grouting embedded PVC pipes. The PVC pipe was pulled up slowly and the pipe wall was tapped when grouting. All piles are maintained for seven days.
The size of the embankment model was determined by the similarity ratio. The gravel cushion was modeled as a 60mm thick layer of uniform coarse sand, which has a particle size of 2mm. The effect of the embankment on the liquefaction characteristics of the foundation soil is mostly due to embankment weight, so the type of stuffing is not particularly important. Cement stabilized soil with a cement ratio of 3% was used as the stuffing of the embankment model, filling 150mm for each layer and tamping. After the embankment was filled to the designed height, the surface was leveled and the slope was trimmed according to 1: 1.5. The finished model is shown in Fig. 4.
The foundation model was saturated after the embankment model formed. Water was injected through the pre-embedded PVC pipes at the four corners (Fig. 4) and the center of end-walls of the model box and the water head was kept 2m high. The change of water level inside the foundation was observed through the plexiglass.Water was injected until the water level approached the foundation's surface. After the water level in the PVC pipes was stabilized, water was injected until the water level in the pipes was flush with the foundation surface. Thus, the saturated foundation model was obtained.3.3 Layout of Monitoring Sensors
Fig. 5 shows the model and the layout of monitoring sensors. The response of the model was monitored by twelve pore-water pressure sensors(P1-P12), six vertical displacement meters(D1-D6) and three horizontal displacement meters(D7-D9). Some sensors were set in one side of the model because of the symmetry.
The selection of vibration wave (frequency of vibration, vibration wave and main frequency) is one of the key factors in determining the reliability of simulation test results of the shaking table test. Theoretically, a better method is to decompose the seismic wave measured by the actual earthquake into irregular waves of each frequency band according to the energy, maximum acceleration and similarity, and then superimpose the waves as the seismic wave required by the research. However, whether artificial seismic waves or actual seismic waves are used, it will bring many difficulties to the test operation and analysis of test results. Therefore, in order to analyze the problem more clearly, the sine wave was used for vibration loading in this test. It was input with three frequencies of 1Hz, 2Hz and 3Hz. Maximum horizontal acceleration increased from 0.05g to 0.60g at a 0.05g step distance (Table 2). The loading direction was along the embankment section and the duration of each loading was 10s.
The foundation and subgrade had no obvious macroscopic deformation as loading increased from 0.05g to 0.25g at 1Hz and 2Hz. After loading 0.30g at 2Hz, the significant subsidence outside the treatment zone totaled 20mm, and lateral cracks and vertical faults appeared along the edge of peripheral cement-mixed piles. After the loading of 0.40g at 2Hz, non-penetrating longitudinal cracks 2mm wide and 200mm long formed, 100mm below the center of the right embankment slope. The cracks developed until they were 10mm wide and were still non-penetrating. After 0.55g at 2Hz, loading was applied. The relative position between the ink lines of the embankment slope section and model box side wall indicated a 10mm left translation of the whole embankment. During the loading of 0.60g sine-wave at 3Hz, the model deformed massively and was destroyed.
Regarding the center cross-section of the model as the boundary, one half foundation was dug out and the other half was retained completely to observe the deformation characteristics of the piles and soil in center cross-section. Some piles were broken (Fig. 6). Most of the faults in the piles were located at the position 1/3-1/2 pile length from pile top. The top part of the pile deviates to the left (Fig. 6(a)). The piles near the left edge of the foundation were broken, and the displacement of the pile tops was larger. However, most piles from the center of the left slope to the right slope were upright. As shown in Fig. 6(b), there are several holes in the soil between piles and obvious traces of liquid flow, under the embankment center and 200mm below the foundation surface. The piles in this position have faults but are still upright.
Fig. 7 and Fig. 8 show the settlements and horizontal displacements of the ground surface. The curves start from 0.05g at 2Hz. No signals were collected from the D4 vertical displacement meter because of unknown reasons. The displacements of some points were beyond the range of the displacement meters after loading. In the subsequent loading process, the displacement meters had no signal, so the data was measured by the steel ruler, e.g. D1, D7 and D8. Accumulated settlement distribution of ground surface is calculated based on Fig. 7, as shown in Fig. 9.
Fig. 7 and Fig. 9 indicate that the vertical displacements are mainly settlement. However, the toe of the slope uplifts slightly, because the foundation soil is laterally extruded by the embankment settlement and is constrained by the lateral soil. The settlements of all points of the embankment surface are relatively uniform, and the maximum accumulated settlement is 1mm after 0.20g at 2Hz and 5mm after 0.60g at 3Hz, respectively. In contrast, settlement of unreinforced foundation changes dramatically after 0.30g, which is up to 34.33mm. It indicates that cement-mixed piles can effectively reduce the foundation settlement and non-uniform settlement under seismic loading.
As shown in Fig. 8, before loading 0.45g at 2Hz, the surface horizontal displacements are small and almost constant to the right. After loading 0.45g at 2Hz, surface horizontal displacements except the right slope toe turned to the left and changed rapidly with increasing acceleration. Eventually, the horizontal displacement of the left slope toe is the largest, about 25mm, which is the same with that shown in Fig. 6(a).4.3 Pore-water Pressure Ratio
It is intuitive that pore-water pressure ratio is used as the basis to identify soil liquefaction in this study. According to the effective stress analysis method, the measured super static pore-water pressure is converted to the pore-water pressure ratio u/σ0, where u is the super static pore-water pressure and σ0 is initial effective overburden stress. Peak pore-water pressure ratio curves are shown in Fig. 10. The curves start with the data of loading 0.05g at 2Hz. Pore-water pressure sensor P8 was damaged during the test. It is widely believed that liquefaction would occur when u/σ0 is larger than 1. Previous studies (Ishihara K., 1993; Singh S., 1996; Zhang Rongxiang et al., 1997; Kammerer A.M. et al., 2004; Wu Jiaer et al., 2004), however, show that foundation liquefaction generally occurs when u/σ0 is much lower than 1, especially in silty sands or sandy silts containing some amount of fines. When saturated silty soil is subjected to vibration, the total pressure previously borne by the soil body is gradually replaced by pore water pressure, under the condition of no drainage. However, due to the existence of clay particles, water permeability is not as good as that of pure sand soil, so that the higher pore water pressure formed in some parts of the soil cannot be immediately transferred to the nearby parts of lower pore water pressure. Therefore, even if the effective stress at a certain point is completely lost and the pore water pressure rises to be equal to the confining pressure, some parts of the whole section of the soil mass are still partially liquefied, and the shear strength of the soil mass is not completely lost and the deformation does not increase indefinitely. On the other hand, due to the existence of clay particles, the soil mass has a certain cohesive force, which may be reduced in the process of vibration, but will not be completely lost. Liu Qian et al.(2007) examined cyclic triaxial tests and resonance vibration tests and proposed a pore-water pressure ratio approaching 0.68 for the silt and approaching 0.87 for the silty sand when liquefaction failure occurs. u/σ0=0.68 for the silt and u/σ0=0.87 for the silty-fine sand are chosen as the initial liquefaction criterions in this paper.
Silty-fine sand is used as the bearing stratum, where P3, P6, P9 and P12 are set (Fig. 5). The maximum u/σ0 of P3, P6, P9 and P12 is just 0.2 during the entire test (Fig. 10). The reason is that the silty-fine sand is buried deeply and its permeability is better. Silty-fine sand as the bearing stratum is safe because the u/σ0 is far less than the initial liquefaction criterion in this study.
Upper and middle silt outside the slope toe, where P1, P4 and P5 are set, is liquefied when acceleration is over 0.30g at 2Hz. It is consistent with the increase of settlement and the widening of the cracks. Liquefaction reduces the shear strength of foundation. As a result, the piles outside the slope toe are sheared in the upper middle (Fig. 6(a)). u/σ0 of P7 and P9-P12 are less than 0.4, so the foundation soil under the shoulders and center of the subgrade is not liquefied. Cement-mixed piles have good influence on liquefaction mitigation of foundation soil under the embankment, although they cannot completely eliminate the liquefaction of the upper middle silt outside the slope toe. The effective stress within the scope of foundation under embankment is increased by the embankment weight, thus it bears the excess pore-water pressure caused by seismic load. The effect of cement-mixed piles acts on anti-liquefaction of foundation outside the embankment is limited. Therefore, reinforcement measures should be set more than 2m-3m outside the slope toe, or loading berms of a certain width should be used to ensure the stability of embankment.
As shown in Fig. 10(a), liquefaction is observed in the upper middle silt outside the composite foundation (P1 and P2) when acceleration meets 0.30g. Then the growth range of u/σ0 has a tendency of decrease. u/σ0 of P1 decreases at 0.60g and 3Hz. u/σ0 of P2 decreases at 0.40g and 2Hz and increases later, and then decreases again. Liquefaction rearranges soil structure. The soil is consolidated and its compactness increases. Dilatancy of soil occurs, so the excess pore-water pressure doesn't rise but lowers. The reduction of u/σ0 at 200mm outside the slope toe is more significant than others, because the positions are closer to the model box, of which the walls act as vertical drainages. Boundary effect makes the pore-water pressure dissipation faster.5 CONCLUSIONS
(1) The shaking table test revealed that silty-fine sand, as the bearing stratum, was stable during the loading process. The pore-water pressure ratio was far less than the initial liquefaction criterion. Silty-fine sand used as the bearing stratum was safe.
(2) The upper middle silt of the foundation was stable before seismic horizontal acceleration increased to 0.30g at 2Hz. In the process of acceleration increasing from 0.30g at 2Hz to 0.60g at 3Hz, the upper middle silt outside the slope toe was partly liquefied. The foundation soil under the shoulders and center of subgrade was far from the initial liquefaction criterion during the test. The results showed that cement-mixed piles have good influence on liquefaction mitigation of foundation soil under embankment, although they cannot completely eliminate the liquefaction of the upper middle silt outside the slope toe. Therefore, reinforcement measures should be more than 2m-3m outside the slope toe, or loading berms of a certain width should be used to ensure the stability of embankment.
(3) The YM railway is in the area of Ⅷ degrees seismic fortification intensity and 0.20g design basic seismic acceleration. It was revealed experimentally that composite foundation was not liquefied and the cement-mixed piles could effectively reduce the foundation settlement and differential settlement at seismic loading of 0.20g at 2Hz. The design ensures the seismic performance of the subgrade, and satisfies the requirements of railway seismic design. It can provide experience for design of cement-mixed piles reinforcing liquefiable railway foundation in areas of Ⅷ degrees seismic fortification intensity. Cement-mixed piles can be used for liquefiable foundation with higher requirements of foundation strength.
|Adalier K., Elgamal A.W., Martin G.R. Foundation liquefaction countermeasures for earth embankments[J]. Journal of Geotechnical and Geoenvironmental Engineering, 1998, 124(6): 500–517. DOI:10.1061/(ASCE)1090-0241(1998)124:6(500).|
|Andrus R.D., Chung R.M. Ground Improvement Techniques for Liquefaction Remediation Near Existing Lifelines[R]. Gaithersburg: National Institute of Standards and Technology, 1995.|
|Babasaki R., Suzuki K., Suzuki Y. Centrifuge tests on improved ground for liquefaction. In: Proceedings of the 10th World Conference on Earthquake Engineering[C]. Rotterdam, Netherlands: A A Balkema, 1992. 1461-1464.|
|Buckingham E.. On physically similar systems; illustrations of the use of dimensional equations[J]. Physical Review, 1914, 4(4): 345–376. DOI:10.1103/PhysRev.4.345.|
|Chen Guoxing, Gu Xiaofeng, Chang Xiangdong, Li Xiaojun, Zhou Guoliang. Review and implication of successful ground improvement cases about mitigating soil liquefaction induced by 8 strong earthquakes from 1989 to 2011[J]. Rock and Soil Mechanics, 2015, 36(4): 1102–1118 (in Chinese with English abstract).|
|Cheng Lilai, Gao Yunfeng, Miao Xianwen, Duan Zonggui. Cement power deep jet mixing pile transacation of the composite foundation and its effect[J]. Henan Geology, 1998, 16(1): 45–52 (in Chinese with English abstract).|
|Esmaeili M., Khajehei H. Mechanical behavior of embankments overlying on loose subgrade stabilized by deep mixed columns[J]. Journal of Rock Mechanics and Geotechnical Engineering, 2016, 8(5): 651–659. DOI:10.1016/j.jrmge.2016.02.006.|
|He J., Chu J., Ivanov V. Mitigation of liquefaction of saturated sand using biogas[J]. Géotechnique, 2013, 63(4): 267–275. DOI:10.1680/geot.SIP13.P.004.|
|Ishihara K. Liquefaction and flow failure during earthquakes[J]. Géotechnique, 1993, 43(3): 351–451. DOI:10.1680/geot.1922.214.171.1241.|
|Jiang Guanlu, Liu Xianfeng, Zhang Jianwen, Zhao Ruyi. Shaking table test of composite foundation reinforcement of saturated silty soil for high speed railway[J]. Frontiers of Architecture and Civil Engineering in China, 2007, 1(3): 353–360. DOI:10.1007/s11709-007-0047-6.|
|Kammerer A.M., Seed R.B., Wu J., Riemer M.F., Pestana J.M. Pore pressure development in lique'able soils under bi-directional loading conditions. In: Proceedings of the 11th International Conference on Soil Dynamics and Earthquake Engineering[C]. Berkeley, USA: University of California, 2004. 697-704.|
|Kang Huiru. Application of cement mixing pile in the building foundation prevent liquefaction[J]. Water Sciences and Engineering Technology, 2016(1): 94–96 (in Chinese with English abstract).|
|Le Kouby A., Bourgeois E., Rocher-Lacoste F. Subgrade improvement method for existing railway lines-an experimental and numerical study[J]. Electronic Journal of Geotechnical Engineering, 2010, 15: 461–494.|
|Lei 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): 403–413.|
|Liu Qian, Zheng Xilai, Liu Hongjun, Wang Jigang, Dong Shuyun. Experimental studies on liquefaction behavior of silt in the Huanghe River delta[J]. World Earthquake Engineering, 2007, 23(2): 161–166 (in Chinese with English abstract).|
|Maheshwari B.K., Singh H.P., Saran S. Effects of reinforcement on liquefaction resistance of Solani sand[J]. Journal of Geotechnical and Geoenvironmental Engineering, 2012, 138(7): 831–840. DOI:10.1061/(ASCE)GT.1943-5606.0000645.|
|Mitchell J.K. Mitigation of liquefaction potential of silty sands. In: From Research to Practice in Geotechnical Engineering[C]. New Orleans, Louisiana, USA: American Society of Civil Engineers, 2008. 433-451.|
|Qu Mengfei, Xie Qiang, Cao Xinwen, Zhao Wen, He Jianjun, Jin Jiang. Model test of stone columns as liquefaction countermeasure in sandy soils[J]. Frontiers of Structural and Civil Engineering, 2016, 10(4): 481–487. DOI:10.1007/s11709-016-0355-9.|
|Shi Yongqiang. Research on Shake Table Test of Liquescent Foudation Consolidated by Cement-soil Mixing Pile[D]. Master's thesis. Chengdu: Southwest Jiaotong University, 2010 (in Chinese with English abstract).|
|Singh S. Liquefaction characteristics of silts[J]. Geotechnical & Geological Engineering, 1996, 14(1): 1–19.|
|Suzuki Y., Saito S., Onimaru S., Kimura T., Uchida A., Okumura R. Grid-shaped stabilized ground improved by deep cementmixing method against liquefaction for a building foundation[J]. Tsuchi-to-kiso, JGS, 1996, 44(3): 46–48.|
|Wu Jiaer, Kammerer A.M., Riemer M.F., Seed R.B., Pestana J.M. Laboratory study of liquefaction triggering criteria. In: Proceedings of the 13th World Conference on Earthquake Engineering[C]. Vancouver, Canada, 2004.|
|Zhang Rongxiang, Gu Baohe, Shi Zhaoji. Critical pore water pressure of foundation liquefaction elimination and seismic isolation[J]. Geotechnical Investigation & Surveying, 1997, 25(2): 5–7 (in Chinese).|
|Zhao Yi. Inquiry on cement-soil mixing pile processing liquefiable soil[J]. Shanxi Architecture, 2016, 42(26): 98–99 (in Chinese with English abstract).|