- Email: [email protected]
Seismic performance and numerical simulation of earth-fill dam with geosynthetic clay liner in shaking table test
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Geotextiles and Geomembranes journal homepage: www.elsevier.com/locate/geotexmem
Seismic performance and numerical simulation of earth-fill dam with geosynthetic clay liner in shaking table test Kyungbeom Jeonga,∗, Satoru Shibuyaa, Toshinori Kawabatab, Yutaka Sawadab, Hiroshi Nakazawac a
Graduate School of Engineering, Kobe University, 1-1, Rokkodai, Nada, Kobe-shi, 657-8501, Japan Graduate School of Agricultural Science, Kobe University, 1-1, Rokkodai, Nada, Kobe-shi, 657-8501, Japan c National Research Institute for Earth Science and Disaster Resilience (NIED), 3-1,Tennodai, Tsukuba-shi, 305-0006, Japan b
A R T I C LE I N FO
A B S T R A C T
Keywords: Earth-fill dam Clay liner Shaking table test Numerical analysis Aseismicity
In this paper, shaking table tests were carried out on both a small-scale and a full-scale earth-fill dams with geosynthetic clay liners to examine their seismic performance. The behavior of these fully instrumented earth-fill dams when subjected to seismic loading was also simulated by numerical analysis. Firstly, in the small-scale shaking table test, no failure was observed along the geosynthetic clay liner when the earth-fill dam was subjected to seismic motion. Numerical analysis confirmed that the behavior of the model earth-fill dam was unaffected by the geosynthetic clay liner. Secondly, a comparative shaking table test was carried out on full-scale earth-fill dams, one with a sloping core zone and another with a geosynthetic clay liner. Both model dams showed similar acceleration response and deformation behavior. It should be mentioned that the acceleration response increased gradually toward the top of the dam, and the deformation, after shaking, was relatively large near the foot of the slope. These observations were successfully simulated by the numerical analysis.
1. Introduction There are approximately 200,000 small earth-fill dams (i.e., reservoirs) in Japan. However, about 70% of these dams were constructed over 150 years ago, before modern design standards and compaction techniques were established (Ministry of Agriculture, Forestry and Fisheries, 2017). Furthermore, there are many small earth-fill dams where the fill material and construction method are unknown. Since the seismic performance of these dams cannot be assumed, there are concerns that they have been or will be damaged by seismic activity. Accordingly, it is considered necessary to prevent water leakage and improve seismic performance for these old earth-fill dams which have insufficient earthquake resistance. A large number of small earth-fill dams have been severely damaged during previous earthquakes in Japan. Almost all those damaged were constructed before the implementation of modern seismic design standards (Ministry of Agriculture, Forestry and Fisheries, 2009). Tani (1996, 2000) investigated the features of dams damaged in past earthquakes, including the 1995 Hyogo-ken Nanbu Earthquake, and classified the types of damage and the damage rate. The survey results indicated that large earth-fill dams with heights above 15 m constructed using current dam designs exhibit high earthquake resistance ∗
and are sufficiently safe (Tani, 2000). In 2011, the Tohoku Earthquake occurred. Though its epicenter was in the Pacific Ocean, the earthquake severely damaged numerous small earth-fill dams in Japan (Fukushima Prefecture, 2011). In Fukushima prefecture, a total of 745 out of 3730 earth-fill dams were damaged by sliding failure of upstream or downstream slopes, crest settlement, and cracks to the longitudinal section at the crown (Hori et al., 2012). Among them, three dams collapsed completely. In particular, the 18.5 m high Fujinuma earth-fill dam, which was completed in 1949 to supply irrigation water, collapsed, resulting in casualties (see in Fig. 1). Several researchers have investigated the condition and features of each damaged dam (Hori et al., 2012; Tanaka et al., 2012; Mohri et al., 2014). None were well-compacted or waterproof. It was reported that the Fujinuma dam had a low degree of compaction, which brought about a deterioration of undrained shear strength under high water levels by applying stronger cyclic loadings (Tanaka et al., 2012). One of the lessons of the 2011 earthquake was the importance of fill material compaction in the design of earth structures (Tatsuoka et al., 2017). Japan expects a strong earthquake to occur along the Nankai Trough within the next 30 years. Improving the earthquake resistance of earth-fill dams is paramount. To this end, the Ministry of Agriculture and the Waters Ministry undertake aseismicity inspections of earth-fill
Corresponding author. E-mail address: [email protected] (K. Jeong).
https://doi.org/10.1016/j.geotexmem.2019.11.006
0266-1144/ © 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: Kyungbeom Jeong, et al., Geotextiles and Geomembranes, https://doi.org/10.1016/j.geotexmem.2019.11.006
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
leakage was observed in either dams. Therefore, earth-fill dams with geosynthetic clay liners are not considered to display inferior seismic performance compared to dams with sloping core zones (Sawada et al., 2018). Another full-scale shaking table test was carried out on earth-fill dams with geosynthetic clay liners, comparing the installation shapes: one was installed straight, the other was installed in a staircase shape with overlapping joints (Sawada et al., 2019a,b). Based on the previous research, this paper first examined the sliding-type failure and seismic behavior of a small earth-fill dam with a geosynthetic clay liner in a small-scale shaking table test, as well as by numerical analysis. Secondly, a comparative shaking table test were carried out on a full-scale earth-fill dam with a sloping core zone and another with a geosynthetic clay liner in order to examine their seismic performance. In addition, the behavior of these fully instrumented earth-fill dams was simulated by numerical analysis.
Fig. 1. Collapsed Fujinuma earth fill dam after earthquake (Hori et al., 2012).
dams nationwide (Suzuki, 2015). Boring surveys, sounding tests, and laboratory tests on fill materials are performed to evaluate the sliding failure stability, residual displacement, and liquefaction risk. If deemed necessary, remedial works are carried out to improve waterproofing and earthquake resistance. A widely-used method for improving small earth-fill dams is to build a sloping core zone with low-permeability clay on the upstream side of the dam (Ministry of Agriculture, Forestry, and Fisheries in Japan, 2015). Recently, because of a shortage of low-permeability clay and logistics making bringing a large amount of soil to a site impractical (Oda et al., 2015), geosynthetic clay liners, which boast outstanding waterproofing abilities, have been used to prevent water leakage and improve dams’ earthquake resistance (see in Fig. 2). However, geotechnical engineers are concerned about the potential for a sliding-type failure at the interface between the soft clay liner and the fill material. It is therefore necessary to research the seismic performance of small earth-fill dams with geosynthetic clay liners. Sasaki et al. (2015) carried out direct shear tests under low confining pressures to find out the frictional interaction between a geosynthetic clay liner and the typical soil in an embankment. The shear strength of the geosynthetic clay liner and the interface between the woven or non-woven geotextile of the geosynthetic clay liner and soil were evaluated, respectively. Other researchers have undertaken shaking table tests (Koyama et al., 2014; Jeong et al., 2016a). Their research did not corroborate the concerns about a sliding-type failure at the interface between the liner and the fill material. However, because the shaking table tests were qualitative studies on small-scale models, the results were not conclusive. Thus, full-scale shaking table tests were carried out on 3 m high earth-fill dams with either a sloping core zone or a geosynthetic clay liner to examine the seismic behavior of both (Sawada et al., 2016; Oda et al., 2016) with prior numerical analysis (Jeong et al., 2016b). Large longitudinal cracks had developed at the crest of the dam with the geosynthetic clay liner. Nonetheless, after subjection to seismic, the amount of residual settlement in each dam was similar and no water
2. Small-scale shaking table test and numerical simulation 2.1. Shaking table test The sliding-type failure of a small-sized earth-fill dam model with a geosynthetic clay liner was examined in a small-scale shaking table test. In this test, a Perspex container (length = 1600 mm, height = 1000 mm, width = 800 mm) with a steel frame was used to observe the shape and location of the failure surface. Well-graded soil from an earth-fill dam with a mean diameter D50 = 0.35 mm, mixed with poorly graded clean sand with D50 = 0.3 mm was employed in the test. The D50 of the mixed soil was 1.7 mm. The model fill of 400 mm in height was prepared by compacting the soil at every layer of 100 mm in thickness to the degree of compaction, Dc, ranging from 80% to 85%. The moisture content of the fill material was approximately 4%. Fig. 3 shows the model tests performed in this study. Case 1 refers to the test without the geosynthetic clay liner. Case 2 shows the fill with the geosynthetic clay liner installed. Fig. 4 shows the structure of the geosynthetic clay liner used in this study. A wave frequency of 5 Hz was employed to mimic the input motion of the full-scale shaking table. A sine wave horizontal acceleration was applied in one direction for a period of 8 s. The maximum acceleration was gradually increased to 6 m/s2, 10 m/s2, and 12 m/s2 until the slope failed. As seen in Fig. 5, no failure was observed in either case when the maximum acceleration of 6 m/s2 was applied. When the maximum acceleration of 10 m/s2 was applied, a shallow circular failure developed over the central portion of the fill slope in Case 1 (the case without the geosynthetic clay liner). A localized failure was observed at the foot of the slope in Case 2 (the case with the geosynthetic clay liner). When the maximum acceleration was increased to 12 m/s2, the failure at the foot of the slope in Case 2 gradually expanded and a circular failure eventually occurred above the geosynthetic layer. However, no slip was observed at the interface between the geosynthetic clay liner and the fill underneath.
Fig. 2. Construction of geosynthetic clay liner.
Fig. 3. A small-scale shaking table test with geosynthetic clay liner. 2
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Table 1 Material properties.
Model γt (kN/m3) E (kN/m2) ν c (kN/m2) φ (°)
Fig. 4. Structure of geosynthetic clay liner employed (refer to Sasaki et al., 2015).
earth-fill dam
GCL
Mohr-Coulomb 18.5 5000 0.34 0.5 38
Mohr-Coulomb 16.0 15,000 0.35 15 28
2.2. Numerical analysis In order to examine the seismic behavior and the shape of the failure surface of the earth-fill dams with and without the geosynthetic clay liner, 2D-FEM eigenvalue analysis, dynamic analysis and pseudo-static limit equilibrium analysis were carried out on the two cases (see Fig. 3). The material properties used in this analysis are shown in Table 1. First, in the eigenvalue analysis for finding the natural frequency of the earth-fill dams with and without the geosynthetic clay liner, a static analysis was performed during the first phase with a static force acting laterally on top of the crest (Plaxis manual, 2017). Then, free vibration analysis was carried out over a 2 s period after reducing the static force by assuming the property of the test fill being the linear elastic. As a result, the time history of horizontal displacement at the top of the slope was obtained. Fig. 6 shows a time history of horizontal displacement during the free vibration analysis, suggesting that the natural frequency of both cases was approximately 20 Hz. It was manifested that no difference was observed in terms of the natural frequency for these two cases, with and without the geosynthetic clay liner. Second, the time history seismic response analysis was performed in order to examine the response characteristics of the test fills using PLAXIS2D. In this analysis, a sine wave (frequency, f = 5 Hz) horizontal acceleration was input at the base boundary over a period of 8 s, which was similar to the shaking table test. The maximum response acceleration, together with the amplification ratio (maximum response acceleration/input acceleration) of horizontal acceleration at the top as well as the foot of slope is summarized in Table 2. This numerical simulation verified that the geosynthetic clay liner does not affect the seismic response characteristics. The effects of amplification were insignificant in both cases. Lastly, a pseudo-static limit equilibrium analysis was performed along the circular slip by following the modified Fellenius's slice method to compare failure shapes and the factor of safety by using the COSTANA program (Jeong et al., 2016a). The geosynthetic clay liner was modeled as a solid element. In this paper, the interface between the soil and the geosynthetic clay liner was not considered. The parameters of the Mohr-Coulomb model for the geosynthetic clay liner were obtained from the direct shear test and used for the pseudo-static and FEM analysis (see Sasaki et al., 2015). A parametric study in which the seismic coefficient (kh) varied 0.15, 0.3, 0.6, and 1.0 was performed. The values of 0.6 and 1.0 correspond to the maximum horizontal acceleration of 6 m/s2 and 10 m/s2 of input motion in the test, respectively. Table 3 shows the results, suggesting no differences between the
Fig. 6. Time history of horizontal displacement at the top of slope.
cases. In concurrence with the results from the shaking table test, no sliding was observed at the boundary of the ground and the geosynthetic clay liner; the circular sliding surface formed well above the geosynthetic clay liner (Fig. 7). 2.3. Discussion In summary, no failure was observed along the geosynthetic clay liner, implying that it did not act as a weak layer in the shaking table test. This was confirmed by the numerical analysis. Furthermore, the geosynthetic clay liner did not affect the seismic behavior of the earthfill dam. However, since these results were obtained from small-scale models without ground water conditions, it is necessary to carry out a full-scale shaking table test and numerical simulation. Also, since compacting the soil on the soft geosynthetic clay liner was difficult, it may be surmised that the soil adjacent to the geosynthetic liner was not well-compacted, which may cause the progressive type of failure seen in Case 2 during the shaking table test. 3. Full-scale shaking table test and numerical simulation 3.1. Outline Shaking table testing using full-scale earth-fill dams, one with a sloping core zone and one with a geosynthetic clay liner was carried out using a huge three-dimensional shaking table at the Hyogo Earthquake Engineering Research Center. Numerical simulation was also performed in order to examine the seismicity of the two earth-fill dams. In this study, two steel containers were placed on the shaking table to test both model fills simultaneously.
Fig. 5. Results of shaking table test; (a) Case1 without the GCL (b) Case2 with the GCL. 3
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Table 2 Maximum response acceleration and amplification ratio. Input acceleration
Maximum response Acceleration (m/s2)
Case1 Case2
Amplification ratio
Case1 Case2
Lateral seismic coefficient
Case1 Case2
kh = 0.3
kh = 0.6
kh = 1.0
1.54 1.54
1.19 1.19
0.76 0.77
0.46 0.46
10 m/s2
top of slope toe of slope top of slope toe of slope
6.12 5.81 6.13 5.81
9.91 9.68 10.17 9.68
top of slope toe of slope top of slope toe of slope
1.02 0.97 1.02 0.97
0.99 0.97 1.02 0.97
time history analysis was performed. In this analysis, as seen in Fig. 10, two types of sine wave (frequency, f = 5 Hz) of maximum acceleration of 1.77 m/s2 and 4.71 m/s2 (hereafter referred to as level 1 earthquake motion and level 2 earthquake motion, respectively), which were used in the tests, were applied at the base boundary for 12 s by simulating the dynamic motion applied in the shaking table test. The sloping core zone and the random soil were both modeled as a solid element by means of a hardening soil model with small-strain stiffness (Table 4). The geosynthetic clay liner was modeled using a Mohr-Coulomb model (Table 1). In this paper, the interface properties between the geosynthetic clay liner and the fill material are not properly considered. The strength parameters for both the random soil and the sloping core zone were obtained by a triaxial compression test (see Sawada et al., 2016).
Table 3 Factor of safety.
kh = 0.15
6 m/s2
3.1.1. Test model Fig. 8 shows the full-scale 3 m high model fills. The water level upstream was maintained at 0.5 m below the fill top. The model fills were compacted to over 95%. Cohesive soil was used for the sloping core zone. The random soil mixed clean sand with the cohesive soil was employed in the shaking table test. Fig. 9 shows the grain size distribution curves of the materials used. In these tests, two types of sine waves (frequency, f = 5 Hz) of the maximum acceleration of 1.77 m/s2 and 4.71 m/s2 were applied, which correspond to level 1 and level 2 earthquake motions, respectively (see Fig. 10). A wave frequency of 5 Hz was chosen in order to avoid the sloshing frequency (1.25 Hz) of water in the reservoir and also to avoid the natural frequency (3 Hz) of the model dams in the full-scale shaking table test (see Sawada et al., 2016). The full-scale shaking table test has been described in detail by Sawada et al. (2016).
3.2. Results and discussions 3.2.1. Damage and residual deformation In the shaking table test, crest settlements were measured at nine points. Fig. 12 shows time histories of vertical deformation at the crest of the model fill obtained from the results of the test and the numerical simulation (+means settlement). When the level 1 earthquake motion was applied, a residual deformation of less than 1 mm developed at the crest and on the slope face in both cases. No water leakage or cracking was observed. In concurrence with the shaking table test results, a crest settlement of approximately 2.6 mm occurred in the fill with a sloping core zone. Another of approximately 4.2 mm occurred in the geosynthetic clay liner case. Based on these results, it can be inferred that there was no harmful deformation in either of the cases as a result of the level 1 earthquake motion. However, when the level 2 earthquake motion was applied, a couple of cracks developed at the crest and at the slope face in both cases. As seen in Fig. 10(a), in the earth-fill dam with a sloping core zone, small cracks –about 1 mm in width and 100 mm in depth–developed on both the upstream and downstream slope faces. As seen in Fig. 10(b), in the earth-fill dam with the geosynthetic clay liner, large longitudinal cracks –about 10 mm in width –developed at the crest. When the cross-section of the embankment was inspected after the shaking event, the cracks were found to have developed because of the geosynthetic clay liner
3.1.2. Numerical simulation 2D-FEM dynamic analysis using the program PLAXIS2D was carried out for the two cases, as shown in Fig. 6. In the analysis, the test models and the shaking table were both modeled as a solid element. The base of the shaking table was fixed horizontally and vertically, and the viscous boundaries were used to absorb the reflected wave of dynamic loading in the base as well as the side boundary of the model. As seen in Fig. 11, the ground water line was set with reference to the phreatic lines of both the model fills, which were estimated by measuring pore water pressure in the tests (Sawada et al., 2016). After setting the boundary conditions together with the ground water conditions, a plastic analysis was performed during the first phase to generate the initial stress conditions. In the next phase, the
Fig. 7. Circular sliding surface in the case of kh = 0.6; (a) Case1 without the GCL (b) Case2 with the GCL. 4
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Fig. 8. Cases of full-scale shaking table test (Sawada et al., 2016); (a) Case 1 with the SCZ (b) Case 2 with the GCL.
Table 4 Material properties.
Model Unsaturated unit weight, γt (kN/m3) Saturated unit weight, γsat (kN/m3) Secant stiffness in standard drained triaxial test, E50 (kN/m2) Unloading/reloading stiffness from drained triaxial test, Eur (kN/m2) Poisson's ratio for unloading-reloading, νur Reference shear modulus at very small strains (ϵ < 10–6), G0 (kN/m2) Threshold shear strain at which Gs = 0.722G0, γ0.7 Cohesion, c’ (kN/m2) Friction angle, φ’ (°)
random soil
core zone
HS-small 20.04 21.34 11,000
HS-small 20.86 21.24 11,000
33,000
33,000
0.2 40,000
0.2 40,000
0.2E-3 6.1 35.5
0.2E-3 38.4 33.2
Fig. 9. Grain size distribution curve (Sawada et al., 2016).
numerical analysis, the average residual crest settlement was 10.0 mm and 10.5 mm, respectively. Although there is some difference in the net displacement between these two cases, no appreciable difference in the deformed shape was observed (Fig. 14). In the numerical simulation, although the observed residual deformation was less than in the shaking table test, the crest settlement and the deformations near the foot of the slope on the upstream side were successfully simulated (Figs. 12(b), Fig. 14(b)). It is necessary to consider more details of strength and stiffness deterioration in saturated as well as unsaturated conditions by applying stronger cyclic loadings in order to get more accurate simulation.
3.2.2. Acceleration response Fig. 15 shows the amplification ratio of the horizontal acceleration response along the vertical, which were obtained from the results of the tests and the numerical simulations, respectively, noting that the height is dimensionless. As shown in Fig. 15, when the level 1 earthquake motion was applied, the response acceleration increased gradually toward the crest. The amplification of the horizontal response acceleration was approximately 1.3 at the crest in both the model fills, implying that the response was successfully simulated by numerical analysis (Fig. 15(b), Table 5). When the level 2 earthquake motion was applied, a large amplification was observed at a lower part of the model fills than in the shaking table test (Fig. 13). The amplification ratio at the crest was approximately 1.4 in the fill with a sloping core zone. In the fill with a geosynthetic clay liner, the ratio was approximately 2.3 before the development of cracks and 3.3 after. The response characteristics between the upstream side and the downstream side separated by the geosynthetic clay liner were noticeably different from each other (the phase difference of acceleration on the horizontal plane). Regarding
Fig. 10. Input sine waves acceleration data (Sawada et al., 2016).
(see Oda et al., 2016). The crack at the crest developed at the interface of the geosynthetic clay liner and the fill materials. Nevertheless, it should be noted that no circular failure developed in either case or no water leakage was observed. After the level 2 earthquake motion was applied in the shaking table test, crest settlement and lateral deformation at the foot of the slope were observed (Figs. 10(a), Fig. 14(a)). These residual deformations were larger on the upstream side than the downstream side. In the shaking table test, an average crest settlement of 21.4 mm was observed in the fill with a sloping core zone. An average crest settlement of 16.7 mm was observed in the fill with the geosynthetic clay liner. In the
Fig. 11. Ground water line in the numerical analysis (Jeong et al., 2016b); (a) with the SCZ (b) with the GCL. 5
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Fig. 12. Time histories of crest settlements.; (a) results of shaking table test (refer to Sawada et al., 2018) (b) results of numerical simulation.
geosynthetic clay liner and the fill (Oda et al., 2016). In other words, the phase difference in the horizontal acceleration triggered cracks to develop at the interface between the soil and the geosynthetic clay liner. The upstream slope and the downstream slope, when separated
this observation in particular, the following two points could be considered; the difference in dynamic characteristic between the saturated soil on the upstream side and the unsaturated soil on the downstream, and the strength characteristic at the interface between the
Table 5 Amplification ratio of horizontal response acceleration at the crest. Water proof method
Level 1 earthquake motion
Level 2 earthquake motion
Shaking table test
sloping core zone geosynthetic clay liner
1.3 1.3
1.4 2.4 (before development of the crack) 3.3 (after development of the crack)
Numerical analysis
sloping core zone geosynthetic clay liner
1.3 1.4
1.2 1.3
6
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Fig. 13. Cracks observed on the surface of earth-fill dam (refer to Nakazawa et al., 2017; Oda et al., 2016); (a) with sloping core zone (b) with geosynthetic clay liner.
earth-fill dams with the geosynthetic clay liner. Despite concerns that the geosynthetic clay liner may act as a weak layer, no failure was observed along the geosynthetic clay liner when the small-scale earth-fill model was subjected to seismic motion. This finding was confirmed by the pseudo-static limit equilibrium analysis. Moreover, time history FEM analysis demonstrated that the geosynthetic clay liner did not affect seismic behaviors, namely, the natural frequency and response characteristics. In the full-scale shaking table test, when the maximum acceleration of 1.77 m/s2 (level 1 earthquake motion) was applied, no harmful deformation was observed in either the fills with a sloping core zone or a geosynthetic clay liner. There was no difference in seismic behaviors in either, such as the response characteristics or residual deformation. Numerical analyses verified these findings. When the maximum acceleration of 4.71 m/s2 (level 2 earthquake motion) was applied, some cracks developed in the fills; however, no
by the geosynthetic clay liner, moved opposite sides, which may generate tensile stress at the interface (Sawada et al., 2018). The acceleration response increased when the high-level earthquake motion was applied. It should be mentioned that the phase difference of acceleration on the horizontal plane, together with the effects of tension cracks on the response characteristic were not properly simulated by numerical analysis. It is necessary to consider more details of strength and stiffness deterioration in saturated as well as unsaturated conditions by applying greater amplitude of cyclic loading. 4. Conclusion Shaking table tests were carried out on small-scale and full-scale earth-fill dams with geosynthetic clay liners in order to evaluate their seismic performance. In addition, numerical analyses were carried out in order to examine the seismic behavior and the shape of failure of
Fig. 14. Residual deformation after shaking; (a) results of shaking table test (modified from Sawada et al., 2016) (b) results of numerical simulation. 7
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
References Fukushima Prefecture, 2011. Damage to the Public Facilities of the Ministry of Agriculture by the 2011 off the Pacific Coast of Tohoku Earthquake. (in Japanese). http://www.pref.fukushima.lg.jp/download/1/230427_aff_higai-02.pdf. Hori, T., Ueno, K., Matsushima, K., 2012. Damages of small earth dams induced by the 2011 off the pacific coast of Tohoku earthquake. Tech. Rep. Nat. Res. Inst. Agric. Eng. 213, 175–199 (in Japanese). Jeong, K., Shibuya, S., Kataoka, S., Baek, J., Kawabata, T., Sawada, Y., 2016a. Shaking table test of a small sized earth-fill dam model with bentonite sheet. In: 51th Japan National Conference on Geotechnical Engineering. The Japanese Geotechnical Society, Okayama, Japan, pp. 1007–1008 (in Japanese). Jeong, K., Shibuya, S., Kataoka, S., Baek, J., Kawabata, T., Sawada, Y., Nakazawa, H., 2016b. Seismic Behavior and Numerical Simulation of a Full-Scale Earth-Fill Dam in Shaking Table Test. 60th Geotechnical Symposium. The Japanese Geotechnical Society, pp. 249–254 (in Japanese). Koyama, A., Suzuki, M., Kochi, Y., Urabe, T., 2014. Deformation behavior of irrigation pond that employs geosynthetic clay liner by shaking table test. Gr. Eng. 32 (1), pp137–141 (in Japanese). Ministry of Agriculture Forestry and Fisheries, 2009. Planning and Design Criteria of Land Improvement Project “Pipeline. ” Designcode (in Japanese). Ministry of Agriculture Forestry and Fisheries in Japan, 2015. Guidelines for Land Improvement Projects, Maintenance of Earth Fill Dam. (in Japanese). Ministry of Agriculture Forestry and Fisheries, 2017. Small earth dam (in Japanese). http://www.maff.go.jp/j/nousin/bousai/bousai_saigai/b_tameike/. Mohri, Y., Masukawa, S., Hori, T., Ariyoshi, M., 2014. Damage to agricultural facilities. Soils Found. 54 (4), 588–607. Nakazawa, H., Sawada, Y., Oda, T., Kobayashi, T., Kobayashi, S., Kawabata, T., Shibuya, S., Kataoka, S., Yamashita, T., 2017. Investigation of residual deformation of small earth dams in full-scale shaking table tests. J. Jpn Soc. Civil Eng., Ser. A1 (Struct. Eng. Earthq. Eng. (SE/EE)) 73 (4), I_815–826 (in Japanese). Oda, T., Moriwaki, K., Tanigaki, K., Nomura, Y., Sawada, Y., Kawabata, T., 2015. Seismic diagnosis and maintenance works of irrigation ponds in Hyogo prefecture. J. Jpn. Soc. Irrig., Drain-Age Reclamat. Eng. 83 (12), 7–10 (in Japanese). Oda, T., Sawada, Y., Nakazawa, H., Kobayashi, S., Shibuya, S., Kawabata, T., 2016. Influence of geosynthetic clay liner laid in a staircase shape in embankment on seismic behavior of small earth dam. Geosynth. Eng. J. 31, 175–182 Japan. Plaxis Manual https://www.plaxis.com/support/verifications/free-vibration-analysis-ofa-dam/. Sasaki, T., Kawaguchi, T., Kawajiri, S., Shibuya, S., 2015. Experimental study on friction resistance between soil and bentonite geosynthetic clay liner. Geosynth. Eng. J. 30, 133–140 Japan. Sawada, Y., Nakazawa, H., Andy Take, W., Kawabata, T., 2019b. Effect of installation geometry on dynamic stability of small earth dams retrofitted with a geosynthetic clay liner. Soils Found. https://doi.org/10.1016/j.sandf.2019.08.007. Published online. Sawada, Y., Nakazawa, H., Kataoka, S., Kobayashi, S., Oda, T., Kobayashi, T., Shibuya, S., Yamashita, T., Tani, K., Kajiwara, K., Kawabata, T., 2016. Full-scale shaking table tests for small earth dams with sloping core zone and geosynthetic clay liner. Geosynth. Eng. J. 31, 167–174 Japan. Sawada, Y., Nakazawa, H., Oda, T., Kobayashi, S., Shibuya, S., Kawabata, T., 2018. Seismic performance of small earth dams with sloping core zones and geosynthetic clay liners using full-scale shaking table tests. Soils Found. 58 (3), 519–533. Sawada, Y., Nakazawa, H., Andy Take, W., Kawabata, T., 2019a. Full scale investigation of GCL damage mechanisms in small earth dam retrofit applications under earthquake loading. Geotext. Geomembranes 47 (4), 502–513. Suzuki, H., 2015. Study on damage factors to earth dams for irrigation due to earthquake motions-In the Case of the 2011 Offshore Pacific Coast Tohoku Earthquake. Bull. Nat. Inst. Rural Eng. 54, 73–155 (in Japanese). Tanaka, T., Tatsuoka, F., Mohri, Y., 2012. Earthquake-induced failure of Fujinuma dam, on dams for a changing world, 24th congress ICOLD. Proc. Intern. Symp., Kyoto 2012 47-6.52. Tani, S., 1996. Damage to earth dams. Soils Found. 36, 263–272 Special Issue. Tani, S., 2000. Earthquake damage to fill dams in Japan. Jpn. Agric. Res. Q. 34 (1), 39–48. Tatsuoka, F., Koseki, J., Takahashi, A., 2017. Earthquake-induced damage to earth structures and proposal for revision of their design policy-Based on a case history of the 2011 off the Pacific Coast of Tohoku Earthquake. J. JSCE 5, 101–112.
Fig. 15. Amplification ratio of horizontal response acceleration; (a) results of shaking table test (modified from Sawada et al., 2016) (b) results of numerical simulation.
failure was observed in either case. Relatively large longitudinal cracks developed in the earth-fill dam with a geosynthetic clay liner. After these cracks developed, the amplification of the response acceleration increased further. In the case of the dam with a geosynthetic clay liner, there is a phase difference in acceleration on the horizontal plane between the upstream side and the downstream side. However, the tension cracks and the phase difference of acceleration on the horizontal could not be replicated by numerical analysis.
Acknowledgments This work was part of a collaborative research project between Hyogo prefecture and the National Research Institute for Earth Science and Disaster Resilience in Japan, and a cooperative research project between Hyogo prefecture and Kobe University. Financial supports from National Research Institute for Earth Science and Disaster Resilience and Hyogo Prefecture are greatly appreciated.
8
Contents lists available at ScienceDirect
Geotextiles and Geomembranes journal homepage: www.elsevier.com/locate/geotexmem
Seismic performance and numerical simulation of earth-fill dam with geosynthetic clay liner in shaking table test Kyungbeom Jeonga,∗, Satoru Shibuyaa, Toshinori Kawabatab, Yutaka Sawadab, Hiroshi Nakazawac a
Graduate School of Engineering, Kobe University, 1-1, Rokkodai, Nada, Kobe-shi, 657-8501, Japan Graduate School of Agricultural Science, Kobe University, 1-1, Rokkodai, Nada, Kobe-shi, 657-8501, Japan c National Research Institute for Earth Science and Disaster Resilience (NIED), 3-1,Tennodai, Tsukuba-shi, 305-0006, Japan b
A R T I C LE I N FO
A B S T R A C T
Keywords: Earth-fill dam Clay liner Shaking table test Numerical analysis Aseismicity
In this paper, shaking table tests were carried out on both a small-scale and a full-scale earth-fill dams with geosynthetic clay liners to examine their seismic performance. The behavior of these fully instrumented earth-fill dams when subjected to seismic loading was also simulated by numerical analysis. Firstly, in the small-scale shaking table test, no failure was observed along the geosynthetic clay liner when the earth-fill dam was subjected to seismic motion. Numerical analysis confirmed that the behavior of the model earth-fill dam was unaffected by the geosynthetic clay liner. Secondly, a comparative shaking table test was carried out on full-scale earth-fill dams, one with a sloping core zone and another with a geosynthetic clay liner. Both model dams showed similar acceleration response and deformation behavior. It should be mentioned that the acceleration response increased gradually toward the top of the dam, and the deformation, after shaking, was relatively large near the foot of the slope. These observations were successfully simulated by the numerical analysis.
1. Introduction There are approximately 200,000 small earth-fill dams (i.e., reservoirs) in Japan. However, about 70% of these dams were constructed over 150 years ago, before modern design standards and compaction techniques were established (Ministry of Agriculture, Forestry and Fisheries, 2017). Furthermore, there are many small earth-fill dams where the fill material and construction method are unknown. Since the seismic performance of these dams cannot be assumed, there are concerns that they have been or will be damaged by seismic activity. Accordingly, it is considered necessary to prevent water leakage and improve seismic performance for these old earth-fill dams which have insufficient earthquake resistance. A large number of small earth-fill dams have been severely damaged during previous earthquakes in Japan. Almost all those damaged were constructed before the implementation of modern seismic design standards (Ministry of Agriculture, Forestry and Fisheries, 2009). Tani (1996, 2000) investigated the features of dams damaged in past earthquakes, including the 1995 Hyogo-ken Nanbu Earthquake, and classified the types of damage and the damage rate. The survey results indicated that large earth-fill dams with heights above 15 m constructed using current dam designs exhibit high earthquake resistance ∗
and are sufficiently safe (Tani, 2000). In 2011, the Tohoku Earthquake occurred. Though its epicenter was in the Pacific Ocean, the earthquake severely damaged numerous small earth-fill dams in Japan (Fukushima Prefecture, 2011). In Fukushima prefecture, a total of 745 out of 3730 earth-fill dams were damaged by sliding failure of upstream or downstream slopes, crest settlement, and cracks to the longitudinal section at the crown (Hori et al., 2012). Among them, three dams collapsed completely. In particular, the 18.5 m high Fujinuma earth-fill dam, which was completed in 1949 to supply irrigation water, collapsed, resulting in casualties (see in Fig. 1). Several researchers have investigated the condition and features of each damaged dam (Hori et al., 2012; Tanaka et al., 2012; Mohri et al., 2014). None were well-compacted or waterproof. It was reported that the Fujinuma dam had a low degree of compaction, which brought about a deterioration of undrained shear strength under high water levels by applying stronger cyclic loadings (Tanaka et al., 2012). One of the lessons of the 2011 earthquake was the importance of fill material compaction in the design of earth structures (Tatsuoka et al., 2017). Japan expects a strong earthquake to occur along the Nankai Trough within the next 30 years. Improving the earthquake resistance of earth-fill dams is paramount. To this end, the Ministry of Agriculture and the Waters Ministry undertake aseismicity inspections of earth-fill
Corresponding author. E-mail address: [email protected] (K. Jeong).
https://doi.org/10.1016/j.geotexmem.2019.11.006
0266-1144/ © 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: Kyungbeom Jeong, et al., Geotextiles and Geomembranes, https://doi.org/10.1016/j.geotexmem.2019.11.006
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
leakage was observed in either dams. Therefore, earth-fill dams with geosynthetic clay liners are not considered to display inferior seismic performance compared to dams with sloping core zones (Sawada et al., 2018). Another full-scale shaking table test was carried out on earth-fill dams with geosynthetic clay liners, comparing the installation shapes: one was installed straight, the other was installed in a staircase shape with overlapping joints (Sawada et al., 2019a,b). Based on the previous research, this paper first examined the sliding-type failure and seismic behavior of a small earth-fill dam with a geosynthetic clay liner in a small-scale shaking table test, as well as by numerical analysis. Secondly, a comparative shaking table test were carried out on a full-scale earth-fill dam with a sloping core zone and another with a geosynthetic clay liner in order to examine their seismic performance. In addition, the behavior of these fully instrumented earth-fill dams was simulated by numerical analysis.
Fig. 1. Collapsed Fujinuma earth fill dam after earthquake (Hori et al., 2012).
dams nationwide (Suzuki, 2015). Boring surveys, sounding tests, and laboratory tests on fill materials are performed to evaluate the sliding failure stability, residual displacement, and liquefaction risk. If deemed necessary, remedial works are carried out to improve waterproofing and earthquake resistance. A widely-used method for improving small earth-fill dams is to build a sloping core zone with low-permeability clay on the upstream side of the dam (Ministry of Agriculture, Forestry, and Fisheries in Japan, 2015). Recently, because of a shortage of low-permeability clay and logistics making bringing a large amount of soil to a site impractical (Oda et al., 2015), geosynthetic clay liners, which boast outstanding waterproofing abilities, have been used to prevent water leakage and improve dams’ earthquake resistance (see in Fig. 2). However, geotechnical engineers are concerned about the potential for a sliding-type failure at the interface between the soft clay liner and the fill material. It is therefore necessary to research the seismic performance of small earth-fill dams with geosynthetic clay liners. Sasaki et al. (2015) carried out direct shear tests under low confining pressures to find out the frictional interaction between a geosynthetic clay liner and the typical soil in an embankment. The shear strength of the geosynthetic clay liner and the interface between the woven or non-woven geotextile of the geosynthetic clay liner and soil were evaluated, respectively. Other researchers have undertaken shaking table tests (Koyama et al., 2014; Jeong et al., 2016a). Their research did not corroborate the concerns about a sliding-type failure at the interface between the liner and the fill material. However, because the shaking table tests were qualitative studies on small-scale models, the results were not conclusive. Thus, full-scale shaking table tests were carried out on 3 m high earth-fill dams with either a sloping core zone or a geosynthetic clay liner to examine the seismic behavior of both (Sawada et al., 2016; Oda et al., 2016) with prior numerical analysis (Jeong et al., 2016b). Large longitudinal cracks had developed at the crest of the dam with the geosynthetic clay liner. Nonetheless, after subjection to seismic, the amount of residual settlement in each dam was similar and no water
2. Small-scale shaking table test and numerical simulation 2.1. Shaking table test The sliding-type failure of a small-sized earth-fill dam model with a geosynthetic clay liner was examined in a small-scale shaking table test. In this test, a Perspex container (length = 1600 mm, height = 1000 mm, width = 800 mm) with a steel frame was used to observe the shape and location of the failure surface. Well-graded soil from an earth-fill dam with a mean diameter D50 = 0.35 mm, mixed with poorly graded clean sand with D50 = 0.3 mm was employed in the test. The D50 of the mixed soil was 1.7 mm. The model fill of 400 mm in height was prepared by compacting the soil at every layer of 100 mm in thickness to the degree of compaction, Dc, ranging from 80% to 85%. The moisture content of the fill material was approximately 4%. Fig. 3 shows the model tests performed in this study. Case 1 refers to the test without the geosynthetic clay liner. Case 2 shows the fill with the geosynthetic clay liner installed. Fig. 4 shows the structure of the geosynthetic clay liner used in this study. A wave frequency of 5 Hz was employed to mimic the input motion of the full-scale shaking table. A sine wave horizontal acceleration was applied in one direction for a period of 8 s. The maximum acceleration was gradually increased to 6 m/s2, 10 m/s2, and 12 m/s2 until the slope failed. As seen in Fig. 5, no failure was observed in either case when the maximum acceleration of 6 m/s2 was applied. When the maximum acceleration of 10 m/s2 was applied, a shallow circular failure developed over the central portion of the fill slope in Case 1 (the case without the geosynthetic clay liner). A localized failure was observed at the foot of the slope in Case 2 (the case with the geosynthetic clay liner). When the maximum acceleration was increased to 12 m/s2, the failure at the foot of the slope in Case 2 gradually expanded and a circular failure eventually occurred above the geosynthetic layer. However, no slip was observed at the interface between the geosynthetic clay liner and the fill underneath.
Fig. 2. Construction of geosynthetic clay liner.
Fig. 3. A small-scale shaking table test with geosynthetic clay liner. 2
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Table 1 Material properties.
Model γt (kN/m3) E (kN/m2) ν c (kN/m2) φ (°)
Fig. 4. Structure of geosynthetic clay liner employed (refer to Sasaki et al., 2015).
earth-fill dam
GCL
Mohr-Coulomb 18.5 5000 0.34 0.5 38
Mohr-Coulomb 16.0 15,000 0.35 15 28
2.2. Numerical analysis In order to examine the seismic behavior and the shape of the failure surface of the earth-fill dams with and without the geosynthetic clay liner, 2D-FEM eigenvalue analysis, dynamic analysis and pseudo-static limit equilibrium analysis were carried out on the two cases (see Fig. 3). The material properties used in this analysis are shown in Table 1. First, in the eigenvalue analysis for finding the natural frequency of the earth-fill dams with and without the geosynthetic clay liner, a static analysis was performed during the first phase with a static force acting laterally on top of the crest (Plaxis manual, 2017). Then, free vibration analysis was carried out over a 2 s period after reducing the static force by assuming the property of the test fill being the linear elastic. As a result, the time history of horizontal displacement at the top of the slope was obtained. Fig. 6 shows a time history of horizontal displacement during the free vibration analysis, suggesting that the natural frequency of both cases was approximately 20 Hz. It was manifested that no difference was observed in terms of the natural frequency for these two cases, with and without the geosynthetic clay liner. Second, the time history seismic response analysis was performed in order to examine the response characteristics of the test fills using PLAXIS2D. In this analysis, a sine wave (frequency, f = 5 Hz) horizontal acceleration was input at the base boundary over a period of 8 s, which was similar to the shaking table test. The maximum response acceleration, together with the amplification ratio (maximum response acceleration/input acceleration) of horizontal acceleration at the top as well as the foot of slope is summarized in Table 2. This numerical simulation verified that the geosynthetic clay liner does not affect the seismic response characteristics. The effects of amplification were insignificant in both cases. Lastly, a pseudo-static limit equilibrium analysis was performed along the circular slip by following the modified Fellenius's slice method to compare failure shapes and the factor of safety by using the COSTANA program (Jeong et al., 2016a). The geosynthetic clay liner was modeled as a solid element. In this paper, the interface between the soil and the geosynthetic clay liner was not considered. The parameters of the Mohr-Coulomb model for the geosynthetic clay liner were obtained from the direct shear test and used for the pseudo-static and FEM analysis (see Sasaki et al., 2015). A parametric study in which the seismic coefficient (kh) varied 0.15, 0.3, 0.6, and 1.0 was performed. The values of 0.6 and 1.0 correspond to the maximum horizontal acceleration of 6 m/s2 and 10 m/s2 of input motion in the test, respectively. Table 3 shows the results, suggesting no differences between the
Fig. 6. Time history of horizontal displacement at the top of slope.
cases. In concurrence with the results from the shaking table test, no sliding was observed at the boundary of the ground and the geosynthetic clay liner; the circular sliding surface formed well above the geosynthetic clay liner (Fig. 7). 2.3. Discussion In summary, no failure was observed along the geosynthetic clay liner, implying that it did not act as a weak layer in the shaking table test. This was confirmed by the numerical analysis. Furthermore, the geosynthetic clay liner did not affect the seismic behavior of the earthfill dam. However, since these results were obtained from small-scale models without ground water conditions, it is necessary to carry out a full-scale shaking table test and numerical simulation. Also, since compacting the soil on the soft geosynthetic clay liner was difficult, it may be surmised that the soil adjacent to the geosynthetic liner was not well-compacted, which may cause the progressive type of failure seen in Case 2 during the shaking table test. 3. Full-scale shaking table test and numerical simulation 3.1. Outline Shaking table testing using full-scale earth-fill dams, one with a sloping core zone and one with a geosynthetic clay liner was carried out using a huge three-dimensional shaking table at the Hyogo Earthquake Engineering Research Center. Numerical simulation was also performed in order to examine the seismicity of the two earth-fill dams. In this study, two steel containers were placed on the shaking table to test both model fills simultaneously.
Fig. 5. Results of shaking table test; (a) Case1 without the GCL (b) Case2 with the GCL. 3
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Table 2 Maximum response acceleration and amplification ratio. Input acceleration
Maximum response Acceleration (m/s2)
Case1 Case2
Amplification ratio
Case1 Case2
Lateral seismic coefficient
Case1 Case2
kh = 0.3
kh = 0.6
kh = 1.0
1.54 1.54
1.19 1.19
0.76 0.77
0.46 0.46
10 m/s2
top of slope toe of slope top of slope toe of slope
6.12 5.81 6.13 5.81
9.91 9.68 10.17 9.68
top of slope toe of slope top of slope toe of slope
1.02 0.97 1.02 0.97
0.99 0.97 1.02 0.97
time history analysis was performed. In this analysis, as seen in Fig. 10, two types of sine wave (frequency, f = 5 Hz) of maximum acceleration of 1.77 m/s2 and 4.71 m/s2 (hereafter referred to as level 1 earthquake motion and level 2 earthquake motion, respectively), which were used in the tests, were applied at the base boundary for 12 s by simulating the dynamic motion applied in the shaking table test. The sloping core zone and the random soil were both modeled as a solid element by means of a hardening soil model with small-strain stiffness (Table 4). The geosynthetic clay liner was modeled using a Mohr-Coulomb model (Table 1). In this paper, the interface properties between the geosynthetic clay liner and the fill material are not properly considered. The strength parameters for both the random soil and the sloping core zone were obtained by a triaxial compression test (see Sawada et al., 2016).
Table 3 Factor of safety.
kh = 0.15
6 m/s2
3.1.1. Test model Fig. 8 shows the full-scale 3 m high model fills. The water level upstream was maintained at 0.5 m below the fill top. The model fills were compacted to over 95%. Cohesive soil was used for the sloping core zone. The random soil mixed clean sand with the cohesive soil was employed in the shaking table test. Fig. 9 shows the grain size distribution curves of the materials used. In these tests, two types of sine waves (frequency, f = 5 Hz) of the maximum acceleration of 1.77 m/s2 and 4.71 m/s2 were applied, which correspond to level 1 and level 2 earthquake motions, respectively (see Fig. 10). A wave frequency of 5 Hz was chosen in order to avoid the sloshing frequency (1.25 Hz) of water in the reservoir and also to avoid the natural frequency (3 Hz) of the model dams in the full-scale shaking table test (see Sawada et al., 2016). The full-scale shaking table test has been described in detail by Sawada et al. (2016).
3.2. Results and discussions 3.2.1. Damage and residual deformation In the shaking table test, crest settlements were measured at nine points. Fig. 12 shows time histories of vertical deformation at the crest of the model fill obtained from the results of the test and the numerical simulation (+means settlement). When the level 1 earthquake motion was applied, a residual deformation of less than 1 mm developed at the crest and on the slope face in both cases. No water leakage or cracking was observed. In concurrence with the shaking table test results, a crest settlement of approximately 2.6 mm occurred in the fill with a sloping core zone. Another of approximately 4.2 mm occurred in the geosynthetic clay liner case. Based on these results, it can be inferred that there was no harmful deformation in either of the cases as a result of the level 1 earthquake motion. However, when the level 2 earthquake motion was applied, a couple of cracks developed at the crest and at the slope face in both cases. As seen in Fig. 10(a), in the earth-fill dam with a sloping core zone, small cracks –about 1 mm in width and 100 mm in depth–developed on both the upstream and downstream slope faces. As seen in Fig. 10(b), in the earth-fill dam with the geosynthetic clay liner, large longitudinal cracks –about 10 mm in width –developed at the crest. When the cross-section of the embankment was inspected after the shaking event, the cracks were found to have developed because of the geosynthetic clay liner
3.1.2. Numerical simulation 2D-FEM dynamic analysis using the program PLAXIS2D was carried out for the two cases, as shown in Fig. 6. In the analysis, the test models and the shaking table were both modeled as a solid element. The base of the shaking table was fixed horizontally and vertically, and the viscous boundaries were used to absorb the reflected wave of dynamic loading in the base as well as the side boundary of the model. As seen in Fig. 11, the ground water line was set with reference to the phreatic lines of both the model fills, which were estimated by measuring pore water pressure in the tests (Sawada et al., 2016). After setting the boundary conditions together with the ground water conditions, a plastic analysis was performed during the first phase to generate the initial stress conditions. In the next phase, the
Fig. 7. Circular sliding surface in the case of kh = 0.6; (a) Case1 without the GCL (b) Case2 with the GCL. 4
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Fig. 8. Cases of full-scale shaking table test (Sawada et al., 2016); (a) Case 1 with the SCZ (b) Case 2 with the GCL.
Table 4 Material properties.
Model Unsaturated unit weight, γt (kN/m3) Saturated unit weight, γsat (kN/m3) Secant stiffness in standard drained triaxial test, E50 (kN/m2) Unloading/reloading stiffness from drained triaxial test, Eur (kN/m2) Poisson's ratio for unloading-reloading, νur Reference shear modulus at very small strains (ϵ < 10–6), G0 (kN/m2) Threshold shear strain at which Gs = 0.722G0, γ0.7 Cohesion, c’ (kN/m2) Friction angle, φ’ (°)
random soil
core zone
HS-small 20.04 21.34 11,000
HS-small 20.86 21.24 11,000
33,000
33,000
0.2 40,000
0.2 40,000
0.2E-3 6.1 35.5
0.2E-3 38.4 33.2
Fig. 9. Grain size distribution curve (Sawada et al., 2016).
numerical analysis, the average residual crest settlement was 10.0 mm and 10.5 mm, respectively. Although there is some difference in the net displacement between these two cases, no appreciable difference in the deformed shape was observed (Fig. 14). In the numerical simulation, although the observed residual deformation was less than in the shaking table test, the crest settlement and the deformations near the foot of the slope on the upstream side were successfully simulated (Figs. 12(b), Fig. 14(b)). It is necessary to consider more details of strength and stiffness deterioration in saturated as well as unsaturated conditions by applying stronger cyclic loadings in order to get more accurate simulation.
3.2.2. Acceleration response Fig. 15 shows the amplification ratio of the horizontal acceleration response along the vertical, which were obtained from the results of the tests and the numerical simulations, respectively, noting that the height is dimensionless. As shown in Fig. 15, when the level 1 earthquake motion was applied, the response acceleration increased gradually toward the crest. The amplification of the horizontal response acceleration was approximately 1.3 at the crest in both the model fills, implying that the response was successfully simulated by numerical analysis (Fig. 15(b), Table 5). When the level 2 earthquake motion was applied, a large amplification was observed at a lower part of the model fills than in the shaking table test (Fig. 13). The amplification ratio at the crest was approximately 1.4 in the fill with a sloping core zone. In the fill with a geosynthetic clay liner, the ratio was approximately 2.3 before the development of cracks and 3.3 after. The response characteristics between the upstream side and the downstream side separated by the geosynthetic clay liner were noticeably different from each other (the phase difference of acceleration on the horizontal plane). Regarding
Fig. 10. Input sine waves acceleration data (Sawada et al., 2016).
(see Oda et al., 2016). The crack at the crest developed at the interface of the geosynthetic clay liner and the fill materials. Nevertheless, it should be noted that no circular failure developed in either case or no water leakage was observed. After the level 2 earthquake motion was applied in the shaking table test, crest settlement and lateral deformation at the foot of the slope were observed (Figs. 10(a), Fig. 14(a)). These residual deformations were larger on the upstream side than the downstream side. In the shaking table test, an average crest settlement of 21.4 mm was observed in the fill with a sloping core zone. An average crest settlement of 16.7 mm was observed in the fill with the geosynthetic clay liner. In the
Fig. 11. Ground water line in the numerical analysis (Jeong et al., 2016b); (a) with the SCZ (b) with the GCL. 5
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Fig. 12. Time histories of crest settlements.; (a) results of shaking table test (refer to Sawada et al., 2018) (b) results of numerical simulation.
geosynthetic clay liner and the fill (Oda et al., 2016). In other words, the phase difference in the horizontal acceleration triggered cracks to develop at the interface between the soil and the geosynthetic clay liner. The upstream slope and the downstream slope, when separated
this observation in particular, the following two points could be considered; the difference in dynamic characteristic between the saturated soil on the upstream side and the unsaturated soil on the downstream, and the strength characteristic at the interface between the
Table 5 Amplification ratio of horizontal response acceleration at the crest. Water proof method
Level 1 earthquake motion
Level 2 earthquake motion
Shaking table test
sloping core zone geosynthetic clay liner
1.3 1.3
1.4 2.4 (before development of the crack) 3.3 (after development of the crack)
Numerical analysis
sloping core zone geosynthetic clay liner
1.3 1.4
1.2 1.3
6
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
Fig. 13. Cracks observed on the surface of earth-fill dam (refer to Nakazawa et al., 2017; Oda et al., 2016); (a) with sloping core zone (b) with geosynthetic clay liner.
earth-fill dams with the geosynthetic clay liner. Despite concerns that the geosynthetic clay liner may act as a weak layer, no failure was observed along the geosynthetic clay liner when the small-scale earth-fill model was subjected to seismic motion. This finding was confirmed by the pseudo-static limit equilibrium analysis. Moreover, time history FEM analysis demonstrated that the geosynthetic clay liner did not affect seismic behaviors, namely, the natural frequency and response characteristics. In the full-scale shaking table test, when the maximum acceleration of 1.77 m/s2 (level 1 earthquake motion) was applied, no harmful deformation was observed in either the fills with a sloping core zone or a geosynthetic clay liner. There was no difference in seismic behaviors in either, such as the response characteristics or residual deformation. Numerical analyses verified these findings. When the maximum acceleration of 4.71 m/s2 (level 2 earthquake motion) was applied, some cracks developed in the fills; however, no
by the geosynthetic clay liner, moved opposite sides, which may generate tensile stress at the interface (Sawada et al., 2018). The acceleration response increased when the high-level earthquake motion was applied. It should be mentioned that the phase difference of acceleration on the horizontal plane, together with the effects of tension cracks on the response characteristic were not properly simulated by numerical analysis. It is necessary to consider more details of strength and stiffness deterioration in saturated as well as unsaturated conditions by applying greater amplitude of cyclic loading. 4. Conclusion Shaking table tests were carried out on small-scale and full-scale earth-fill dams with geosynthetic clay liners in order to evaluate their seismic performance. In addition, numerical analyses were carried out in order to examine the seismic behavior and the shape of failure of
Fig. 14. Residual deformation after shaking; (a) results of shaking table test (modified from Sawada et al., 2016) (b) results of numerical simulation. 7
Geotextiles and Geomembranes xxx (xxxx) xxx–xxx
K. Jeong, et al.
References Fukushima Prefecture, 2011. Damage to the Public Facilities of the Ministry of Agriculture by the 2011 off the Pacific Coast of Tohoku Earthquake. (in Japanese). http://www.pref.fukushima.lg.jp/download/1/230427_aff_higai-02.pdf. Hori, T., Ueno, K., Matsushima, K., 2012. Damages of small earth dams induced by the 2011 off the pacific coast of Tohoku earthquake. Tech. Rep. Nat. Res. Inst. Agric. Eng. 213, 175–199 (in Japanese). Jeong, K., Shibuya, S., Kataoka, S., Baek, J., Kawabata, T., Sawada, Y., 2016a. Shaking table test of a small sized earth-fill dam model with bentonite sheet. In: 51th Japan National Conference on Geotechnical Engineering. The Japanese Geotechnical Society, Okayama, Japan, pp. 1007–1008 (in Japanese). Jeong, K., Shibuya, S., Kataoka, S., Baek, J., Kawabata, T., Sawada, Y., Nakazawa, H., 2016b. Seismic Behavior and Numerical Simulation of a Full-Scale Earth-Fill Dam in Shaking Table Test. 60th Geotechnical Symposium. The Japanese Geotechnical Society, pp. 249–254 (in Japanese). Koyama, A., Suzuki, M., Kochi, Y., Urabe, T., 2014. Deformation behavior of irrigation pond that employs geosynthetic clay liner by shaking table test. Gr. Eng. 32 (1), pp137–141 (in Japanese). Ministry of Agriculture Forestry and Fisheries, 2009. Planning and Design Criteria of Land Improvement Project “Pipeline. ” Designcode (in Japanese). Ministry of Agriculture Forestry and Fisheries in Japan, 2015. Guidelines for Land Improvement Projects, Maintenance of Earth Fill Dam. (in Japanese). Ministry of Agriculture Forestry and Fisheries, 2017. Small earth dam (in Japanese). http://www.maff.go.jp/j/nousin/bousai/bousai_saigai/b_tameike/. Mohri, Y., Masukawa, S., Hori, T., Ariyoshi, M., 2014. Damage to agricultural facilities. Soils Found. 54 (4), 588–607. Nakazawa, H., Sawada, Y., Oda, T., Kobayashi, T., Kobayashi, S., Kawabata, T., Shibuya, S., Kataoka, S., Yamashita, T., 2017. Investigation of residual deformation of small earth dams in full-scale shaking table tests. J. Jpn Soc. Civil Eng., Ser. A1 (Struct. Eng. Earthq. Eng. (SE/EE)) 73 (4), I_815–826 (in Japanese). Oda, T., Moriwaki, K., Tanigaki, K., Nomura, Y., Sawada, Y., Kawabata, T., 2015. Seismic diagnosis and maintenance works of irrigation ponds in Hyogo prefecture. J. Jpn. Soc. Irrig., Drain-Age Reclamat. Eng. 83 (12), 7–10 (in Japanese). Oda, T., Sawada, Y., Nakazawa, H., Kobayashi, S., Shibuya, S., Kawabata, T., 2016. Influence of geosynthetic clay liner laid in a staircase shape in embankment on seismic behavior of small earth dam. Geosynth. Eng. J. 31, 175–182 Japan. Plaxis Manual https://www.plaxis.com/support/verifications/free-vibration-analysis-ofa-dam/. Sasaki, T., Kawaguchi, T., Kawajiri, S., Shibuya, S., 2015. Experimental study on friction resistance between soil and bentonite geosynthetic clay liner. Geosynth. Eng. J. 30, 133–140 Japan. Sawada, Y., Nakazawa, H., Andy Take, W., Kawabata, T., 2019b. Effect of installation geometry on dynamic stability of small earth dams retrofitted with a geosynthetic clay liner. Soils Found. https://doi.org/10.1016/j.sandf.2019.08.007. Published online. Sawada, Y., Nakazawa, H., Kataoka, S., Kobayashi, S., Oda, T., Kobayashi, T., Shibuya, S., Yamashita, T., Tani, K., Kajiwara, K., Kawabata, T., 2016. Full-scale shaking table tests for small earth dams with sloping core zone and geosynthetic clay liner. Geosynth. Eng. J. 31, 167–174 Japan. Sawada, Y., Nakazawa, H., Oda, T., Kobayashi, S., Shibuya, S., Kawabata, T., 2018. Seismic performance of small earth dams with sloping core zones and geosynthetic clay liners using full-scale shaking table tests. Soils Found. 58 (3), 519–533. Sawada, Y., Nakazawa, H., Andy Take, W., Kawabata, T., 2019a. Full scale investigation of GCL damage mechanisms in small earth dam retrofit applications under earthquake loading. Geotext. Geomembranes 47 (4), 502–513. Suzuki, H., 2015. Study on damage factors to earth dams for irrigation due to earthquake motions-In the Case of the 2011 Offshore Pacific Coast Tohoku Earthquake. Bull. Nat. Inst. Rural Eng. 54, 73–155 (in Japanese). Tanaka, T., Tatsuoka, F., Mohri, Y., 2012. Earthquake-induced failure of Fujinuma dam, on dams for a changing world, 24th congress ICOLD. Proc. Intern. Symp., Kyoto 2012 47-6.52. Tani, S., 1996. Damage to earth dams. Soils Found. 36, 263–272 Special Issue. Tani, S., 2000. Earthquake damage to fill dams in Japan. Jpn. Agric. Res. Q. 34 (1), 39–48. Tatsuoka, F., Koseki, J., Takahashi, A., 2017. Earthquake-induced damage to earth structures and proposal for revision of their design policy-Based on a case history of the 2011 off the Pacific Coast of Tohoku Earthquake. J. JSCE 5, 101–112.
Fig. 15. Amplification ratio of horizontal response acceleration; (a) results of shaking table test (modified from Sawada et al., 2016) (b) results of numerical simulation.
failure was observed in either case. Relatively large longitudinal cracks developed in the earth-fill dam with a geosynthetic clay liner. After these cracks developed, the amplification of the response acceleration increased further. In the case of the dam with a geosynthetic clay liner, there is a phase difference in acceleration on the horizontal plane between the upstream side and the downstream side. However, the tension cracks and the phase difference of acceleration on the horizontal could not be replicated by numerical analysis.
Acknowledgments This work was part of a collaborative research project between Hyogo prefecture and the National Research Institute for Earth Science and Disaster Resilience in Japan, and a cooperative research project between Hyogo prefecture and Kobe University. Financial supports from National Research Institute for Earth Science and Disaster Resilience and Hyogo Prefecture are greatly appreciated.
8