Liquefaction-induced damage evaluation of earth embankment and corresponding countermeasure

Abstract

Liquefaction of sandy soils is a big threat to the stability and the safety of an earth embankment laid on saturated soils. A large number of liquefaction-induced damages on embankment due to different types of earthquakes have been reported worldwide. In this research, the dynamic behaviors of earth embankment and the reinforcement effects of grouting as remediation method, subjected to moderate earthquake EQ₁ and strong earthquake EQ₂, were numerically investigated. The seismic behaviors of ground composed of cohesionless sandy soil and cohesive clayey soil were uniformly described by the cyclic mobility (CM) model, which is capable of describing accurately the mechanical property of the soil due to monotonic and cyclic loadings by accounting for stress-induced anisotropy, over-consolidation, and soil structure. It is known from the numerical investigation that the embankment would experience destructive deformation, and that the collapse mode was closely related to the properties of input seismic motion because high intensities and long durations of an earthquake motion could lead to significant plastic deformation and prolonged soil liquefaction. Under the strong seismic loading of EQ₂, a circular collapse surface, combined with huge settlement and lateral spread, occurred inside the liquefication zone and extended towards the embankment crest. In contrast, in moderate earthquake EQ₁, upheaval was observed at each toe of the embankment, and instability occurred only in the liquefied ground. An anti-liquefaction remediation via grouting was determined to significantly reduce liquefaction-induced deformation (settlement, lateral spreading, and local uplift) and restrain the deep-seated circular sliding failure, even though the top sandy soil liquefied in both earthquakes. When the structure was subjected to EQ₂ motion, local failure occurred on the embankment slope reinforced with grouting, and thus, an additional appropriate countermeasure should be implemented to further strengthen the slope. For both input motions, the surface deformation of the considered embankment decreased gradually as the thickness of reinforcement was increased, although the reinforcement effect was no longer significant once the thickness exceeded 6 m.

Graphical abstract

Keywords

dynamic response / earth embankment / damage pattern / liquefaction / ground improvement

1 Introduction

Failures in earth embankment could adversely affect thousands of lives and threaten property security, and thus the safety evaluation of these structures should be conducted seriously and carefully. Earthquake is one of the major causes of the embankments destruction (examples of earthquakes include the 1995 Hyogoken-Nambu Earthquake and the 2011 Tohoku Earthquake off the Pacific Coast). A large number of earthen infrastructures, such as coastal dikes, levees, earth dams, and embankments, can be seriously devastated by earthquakes, mainly through Earthquake-induced liquefaction [1,2]. In most cases, large deformation and destructive failure occur when the underlying saturated loose sand liquefies, causing excessive subsidence, cracking, lateral spreading, slumping, and slope instability [3,4]. During liquefaction, significant excess pore water pressure (EPWP) and associated shearing strain occur in the liquefiable ground and extend to the geo-structures. The shearing region, combined with large settlement and lateral movement, could cause local instability failure or complete collapse [5]. Moreover, the characteristics of earthquakes (duration, frequency, peak ground acceleration (PGA)) have been proven to be critically influential to the seismic behaviors of structures [6].

Thus far, several evaluation studies on the seismic behaviors of embankments have been conducted through laboratory tests and numerical analyses. Experimental investigations have involved shaking table tests [7,8] and dynamic centrifuge tests [9–11]. Case studies and numerical investigations have been conducted to analyze the dynamic responses of embankment founded on liquefiable ground [12–16]. However, in most of the aforementioned research endeavors, dynamic constitutive models tended to be limited to describing granular sand, whereas the basic parameters involved were complex, numerous, and difficult to determine. Moreover, these studies focused mainly on embankments located on homogenous sandy deposits and did not sufficiently account for embankments on liquefiable ground consisting of cohesionless sand with low-permeability, cohesive, clayey layers. In addition, there exist only small amounts of data on the failure paths and collapse mechanisms of embankments subjected to a variety of moderate to strong input motions.

In most cases, earth embankment might be constructed on liquefiable soil layer, necessitating the implementation of applicable reinforcement measures [17]. These reinforcement measures include drainage, densification, solidification, and inclusion, which are applied to alleviate the destruction of earthquake-induced liquefaction in geo-structures. Ground-improvement measures have been analyzed via centrifuge tests or numerical analyses [18–20], whereas remediation measures via grouting have been confirmed as a valid and economical solution for resisting liquefaction-induced damage.

This research aims to numerically study the dynamic behavior of an embankment and the effect of ground improvement with grouting method for soft soils, especially the solidification for preventing excessive ground deformation. First, a general background regarding dynamic response is described in detail, focusing on liquefaction-induced damage and failure for two different types of seismic excitations with different intensities, based on the constitutive model named as cyclic mobility (CM) model [21]. Afterward, a typical liquefaction vulnerability analysis for the earth embankment with/without ground improvement is performed, including the detailed discussions on stress path, EPWP, acceleration, deformation, failure path, and collapse mechanism, when the embankment is subjected to two different seismic motions (moderate EQ₁ and severe EQ₂). Finally, comparisons of the embankment performance with and without ground improvement are presented, which can provide a detailed design base for seismic enhancement of earth embankments.

2 General formulation for soils

2.1 Constitutive model

In this research, the finite element method (FEM) code DBLEAVES [22] is employed for a 2D fully soil-water coupled seismic analysis of the liquefaction-induced deformations of the embankment subjected to different seismic motions. The Green–Naghdi stress rate tensor is first applied to this numerical analysis to describe finite deformation that occur in a series of moderate to huge earthquakes. A simple constitutive model, which involves only a few parameters, is insufficient for the liquefaction analysis, and therefore, this study utilizes a dynamic constitutive model, specifically the kinematic hardening elastoplastic model developed by Zhang et al. [21], which describes the monotonic or cyclic behaviors of liquefied sand and non-liquefied clay for drained or undrained conditions uniformly. To determine the seismic performance of cohesionless sand and cohesive clay, the stress-induced anisotropy, over-consolidation, and soil structure are taken into consideration. There exist enormous differences in the evolutions of over-consolidation and soil structure for sand and clay [23]. For clayey soil, the collapse rate of the soil structure is slower than the loss rate of over-consolidation. Conversely, for sandy soil, the collapse rate of the soil structure is much faster compared with loss rate of over-consolidation.

Based on finite element (FE) spatial discretization, the formulation becomes

(1)f=\ln \frac{p^{\prime }}{{\tilde{p}}_0^{\mathrm{\prime }}}+\ln \frac{M^2-{\zeta }^2+{\eta }^{\ast 2}}{M^2-{\zeta }^2}+\ln R^{\ast }-\ln R-\frac{{\epsilon }_{\text{v}}^{\text{p}}}{C_{\text{p}}}=0,

(2)\text{d}{\beta }_{ij}=\frac{M}{C_{\text{p}}}br\left(b_{\text{l}}M-\zeta \right)\text{d}{\epsilon }_{\text{d}}^{\text{p}}\frac{{\hat{\eta }}_{ij}}{{\hat{\eta }}_{ij}},

(3){\eta }^{\ast }=\sqrt{\frac{3\hat{\eta }\times \hat{\eta }}{2}},\hat{\eta }={\eta }_{ij}-{\beta }_{ij},{\eta }_{ij}=\frac{S_{ij}}{p^{\prime }},p^{\prime }=\frac{{\sigma }_{ii}^{\mathrm{\prime }}}{3},

(4)\eta =\sqrt{\frac{3}{2}{\eta }_{ij}{\eta }_{ij}},\zeta =\sqrt{\frac{3}{2}{\beta }_{ij}{\beta }_{ij}},Cp=\frac{\lambda -k}{1+\text{e}},

(5)\text{d}R=U\text{d}{\epsilon }_{ij}^{\text{p}}+R\frac{\eta }{M}\frac{\mathrm{\partial }f}{\mathrm{\partial }{\beta }_{ij}}\text{d}{\beta }_{ij},U=-\frac{mM}{C_{\text{p}}}\left(\frac{{\left(p_{\text{m}}/p_{\text{m0}}\right)}^2}{{\left(p_{\text{m}}/p_{\text{m0}}\right)}^2+1}\right)\ln R,

(6)\text{d}{R}^{\ast }=U^{\ast }\text{d}{\epsilon }_{\text{d}}^{\text{p}},U^{\ast }=\frac{aM}{C_{\text{p}}}{R}^{\ast }\left(1-R^{\ast }\right),

(7)R^{\ast }=\frac{{\tilde{p}}^{\mathrm{\prime }}}{{\overline{p}}^{\mathrm{\prime }}}=\frac{\tilde{q}}{\overline{q}},R=\frac{p^{\prime }}{{\overline{p}}^{\mathrm{\prime }}}=\frac{q}{\overline{q}},

where M is the stress ratio at the critical state, η is the shearing stress ratio, ζ is the magnitude of the stress-induced anisotropy, R (0 < R ≤ 1, R = 1/OCR) is the over-consolidation ratio, R^{\ast } (0 < R^{\ast } ≤ 1) refers to the structural ratios, λ is the compression index, and k is the expansion index. All these parameters can be determined via laboratory tests. A detailed description has been provided by Zhang et al. [21].

2.2 Soil-water coupling approach

For saturated soil, the soil-water coupling scheme should be considered in the liquefaction analysis, which is on account of the effective stress method. The two-phase mixture theory introduced by Akai and Tamura [24] is employed using the u−p (deformation of solid grain-pore water pressure) equation. The relative acceleration of fluid compared to the soil skeleton is much smaller and is therefore considered negligible [25]. This equation is effective for low-frequency phenomenon, such as earthquakes, which decrease the number of freedom degrees. The control formulation for the two-phase mixture theory, which involves the dynamic equilibrium of soil, and the fluid continuity formula are as follows:

(8)\mathrm{E}\mathrm{q}\mathrm{u}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{b}\mathrm{r}\mathrm{i}\mathrm{u}\mathrm{m}\mathrm{e}\mathrm{q}\mathrm{u}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}:\rho {\ddot{u}}_i^{\text{s}}=\frac{\mathrm{\partial }{\sigma }_{ij}}{\mathrm{\partial }{x}_j}+\rho b_i,

where ρ and u are the soil density and deformation, respectively, whereas b_i is the body force.

(9)\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{u}\mathrm{i}\mathrm{t}\mathrm{y}\mathrm{e}\mathrm{q}\mathrm{u}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}:{\rho }^{\text{f}}{\ddot{\epsilon }}_{ii}^{\text{s}}-\frac{{\mathrm{\partial }}^2{\rho }_{\text{d}}}{\mathrm{\partial }{x}_i^2}+\frac{{\gamma }_{\text{w}}}{k}\left[\frac{{\ddot{\epsilon }}_{ii}^{\text{s}}}{n}-\frac{1}{K^{\text{f}}}{\dot{\rho }}_{\text{d}}\right]=0,

where ρ^f is the density of liquid phase, ρ_d is the EPWP induced by dynamic motion, γ_w is the gravity of the liquid phase, k is the permeability coefficient, n is the soil porosity, and K^f is the volumetric compression modulus in the liquid phase.

3 Finite element model

3.1 Background

In this paper, special emphasis is given to the dynamic response of an earth embankment in Aichi Prefecture, Japan. The geometric profile of the earth embankment is presented in Fig.1. The underground water table is on the ground surface. This embankment is 6 m in height, with a slope inclination equal to 1:1.5 (vertical: horizontal) and a crest of 5 m. The embankment body is dry plain backfill with an average diameter (D₅₀) of 0.23 mm, and it consists of gravel, sand, silt and clay. The embankment is located on soft deposits, consisting of a sandy layer with the thickness of 10 m on top and followed by a clayey layer with thickness of 20 m. The standard penetration value (N value) of the sand is approximately 12, and its relative density is D_r ≈ 50%. D₅₀ of the soft clay deposit is around 0.0035 mm with Ip = 15, below which a rigid bedrock is formed as a supporting layer in the simulation.

When subjected to earthquake motions, the earth embankment will be heavily damaged with large deformations. The most devasting failure observed for this specific pattern of soil profile was when such a structure due to the 2011 Tohoku Earthquake, where the saturated liquefiable zone was bounded by the impermeable clayey layer [4]. Remediation of earthquake-induced liquefaction on a constructed site could be implemented by the decrease of the liquefaction potential via foundation reinforcement techniques [26]. In this study, a liquefaction countermeasure via grouting is adopted to eliminate or reduce liquefaction-induced damages in the embankment due to two specific types of ground motion.

3.2 Boundary conditions

The FE mesh is composed of 4-node quadrangle elements (4859 nodes, 4710 elements), as shown in Fig.2. The two specific earthquakes are input to the bottom of the FE model, and a plane-strain condition is applied. The horizontal deformation and vertical deformation are both fixed at the bedrock. Equivalent boundary conditions have been applied to the two side nodes of the lateral boundary (i.e., the deformations of nodes at the same depth on the two lateral sides are identical), and thus the embankment will vibrate horizontally, just as in shaking table tests, when input motion is imposed on the base. Although the horizontal length of the free ground on each side of the embankment (30 m) was less than twice the width of the embankment bottom (23 m), the boundary effect has insignificant impact on the seismic property of the embankment and that it satisfies the requirement of a free-field condition in the lateral boundary [27]. The bottom and lateral boundaries on both sides are impermeable, whereas the ground surface is assumed to be permeable.

3.3 Input earthquake loadings

The properties of the input seismic motions are demonstrated to be crucial for the dynamic performance of the earth embankment. The input motions for the nonlinear liquefaction analysis and foundation reinforcement effect are defined based on two distinct earthquakes with different amplitudes: moderate (EQ₁, a_max = 0.11g, duration t = 116.0 s) and strong signals (EQ₂, a_max = 0.76g, duration t = 168.0 s). The acceleration-time history curve and response spectra of these two earthquakes are depicted in Fig.3 and Fig.4. The strong EQ₂ motion is characterized by a larger maximum acceleration with longer duration. These two earthquakes were based on records chosen from the Geo-disaster database of the Chubu Branch of the Japan Geotechnical Society (JGS). A liquefaction-induced damage analysis of the embankment model under two different types of earthquake loadings is performed, and the effect of the properties of the earthquakes on the collapse mechanism is investigated.

3.4 Soil behaviors and numerical parameters

The embankment is composed of a dry-compacted fill soil, whereas the foundation consists of liquefiable sand and thick cohesive clay. The property parameters, outlined in Tab.1, were determined via laboratory triaxial tests. To demonstrate the capacity of the CM model, the experimental results and numerical predictions for foundation soils subjected to undrained cyclic triaxial loadings were compared. The obtained graphs for cyclic stress ratio (CSR = σ_d-cyc/2p_0^{\prime }, where σ_d-cyc is the cyclic vertical stress applied in the undrained cyclic triaxial test) in the isotropic consolidation as a function of the cyclic loading number, to generate liquefaction at p_0^{\prime } = 100 kPa, are shown in Fig.5. The dynamic responses obtained by the CM model matched quite well with the test data. The dynamic strength curves show that the property parameters for sandy soil were representative of liquefiable sand, and that the cyclic shearing strength of soft clayey soil was larger than that of sand.

To explore the seismic behavior of the embankment subjected to the seismic motions, two elements (E₁ and E₂) in the liquefiable sandy layer, illustrated in Fig.1, are selected for a demonstration of their effective stress paths. Fig.6 illustrates the mean effective stress paths of these two representative elements under EQ₁ and EQ₂ loading. The mean effective stress and deviator stress gradually declined because of the earthquake motions, whereas the stress path moved toward the critical state line (C.S.L). The mean effective stress at E₁ (5 m below the surface in the free field) was observed to gradually decrease to zero with the occurrence of liquefaction for both ground motions, where the phenomenon of CM was more prominent under EQ₂ loading. For soil E₂ (4.3 m right below the embankment body), its initial mean effective stress and dilative stress were higher because of the gravity of the overlying embankment, whereas E₂ maintained a large confining pressure after either of the two seismic motions. Despite its similar composition to that of the sandy layer at E₁, the soil at E₂ (beneath the embankment) would not liquefy when subjected to either of the two seismic loadings.

4 Liquefaction analysis for embankment

The objectives of this research are to study the seismic behaviors of the embankment laid on liquefiable soil subjected to moderate earthquake (EQ₁) and strong earthquake (EQ₂) numerically, particular attention has been paid to assessing the accumulation of EPWP and the deformation evolution during the liquefaction-induced failure.

4.1 Excess pore water pressure of natural soils

Fig.7 shows the EPWP distribution after the completion of each seismic motion (moderate EQ₁ and strong EQ₂). The EPWP was observed to build up in the foundation soil situated in depth during either of the two seismic motions and was not entirely dissipated after either of the earthquakes because of the relatively low-permeability clayey layer in the lower part of the liquefiable sandy layer. Furthermore, high-amplitude seismic motion (EQ₂) resulted in a larger EPWP in the foundation soils than that which resulted from the moderate seismic motion. Because of the difference in permeability between the sandy and clayey layers, the water flow generated by liquefaction was obstructed by the clayey layer, resulting in accumulation of high EPWP, leading to the generation of a thin water-rich region. The EPWP gathered mainly in the liquefied sandy layer at the interface and gradually extended to the clayey substratum as the earthquake intensity increased. The maximum value of EPWP was approximately 81 kPa for EQ₁, and approximately 100 kPa for EQ₂.

In this study, the effective stress decreasing ratio, expressed as ESDR = 1 − {\sigma }_{\mathrm{m}}^{\prime }/{\sigma }_{\mathrm{m}0}^{\prime } [28], where {\sigma }_{\mathrm{m}}^{\prime } is the current mean effective stress, and {\sigma }_{\mathrm{m}0}^{\prime } is the initial mean effective stress, is employed to define the occurrence of liquefaction. When liquefaction occurs, {\sigma }_{\mathrm{m}}^{\prime } becomes zero, and ESDR = 1.0. Fig.8 illustrates the ESDR distributions for the two seismic loadings. The ESDR of sandy deposits in both sides of the embankment were calculated to be quite high (i.e., ESDR ≈ 1.0), revealing that the soils in those regions had completely liquefied. The phenomenon of liquefaction extended from each toe of the embankment to the free field. The values of ESDR were at the maximum in the free field and at the minimum below the embankment during seismic loading. Although high EPWP were concentrated in the sandy soil below the embankment center, the ESDR was quite low because of the high initial overburden vertical pressure for both seismic motions. In the case of huge motion (EQ₂), the ESDR underneath the cohesive clayey layer was higher, indicating that the clayey soil lost more strength compared to the loss due to moderate motion (EQ₁).

4.2 Responding acceleration of natural ground

In Fig.9, the horizontal acceleration−time profiles for the two seismic motions at the three representative positions (N₁–N₃) illustrated in Fig.2 are depicted. It is shown that the PGA experienced by the embankment had a significant influence on the amount of acceleration. For EQ₂, the larger the PGA, the larger the acceleration response. Nevertheless, the acceleration responses at these three locations were similar for both moderate EQ₁ and strong EQ₂. The high-frequency component and peak value of the seismic motions were damped out by the liquefied soils during earthquakes. The acceleration response at location N₃ was slightly lower than that at N₂, indicating that liquefaction would suppress the propagation of earthquake waves and attenuate the acceleration response to some extent. However, the maximum acceleration at the embankment crest (N₁) increased significantly compared to those of both input waves, indicating that the acceleration response was obviously amplified when the ground motion propagated through the embankment body.

4.3 Deformation of natural ground

Fig.10 presents the deformed shape of the embankment model for each seismic loading. Because the embankment is bilaterally symmetric in structure, only the deformation in the left part was studied. Based on the liquefaction-vulnerability study for the two input earthquakes, the seismic behavior of the earth embankment was determined to be closely related to the characteristics of the earthquakes. Each seismic excitation resulted in a distinct failure path and destruction level. Furthermore, the high-amplitude motion (EQ₂) would lead to much larger deformation at the crest and on both sides of the embankment than those by the moderate motion (EQ₁). Meanwhile, slightly disparate damage patterns were observed for the strong motion, implying the eventual occurrence of circular slipping surfaces in both sides of the embankment, which could lead to the collapse failure of the system. This type of failure path was also presented by Ishikawa et al. [29] and Sasaki and Tamura [30] and identified as lateral spreading combined with settlement at the earth embankment. Meanwhile, in the case of moderate motion EQ₁, the structure underwent instability only inside of the liquefiable sandy layer, whereas the upheavals produced at each toe of the embankment and in the soil beneath the embankment moved toward the free field. Lateral movement and heaving of the ground surface near each toe of the embankment suggested squeezing-out of the underlying liquefiable sandy soils. Rapid liquefaction on the two sides of the embankment decreased the constraint of the soils underneath the embankment and permitted the lateral extension of those soils toward the free field. No further post-liquefaction effects were observed, which were consistent with the findings of Coelho et al. [31], in which the most significant deformation developed simultaneously with the shaking.

5 Strengthening effect

The liquefaction susceptibility of the earth embankment could be effectively improved via a decrease in the liquefaction potential through various ground-reinforcement methods. Current techniques for seismic mitigation include: (1) excavation and substitution of liquefied soil, (2) in situ densification, (3) in-situ strengthening, and (4) improvement of drainage system, among others. Marcuson et al. [17] provided a summary of those reinforcement techniques. Infiltration grouting is a strengthening technique by injecting cement into a granular sandy layer with relatively low grouting pressure (0.2–1.5 MPa), on condition that the sandy layer is permeable enough to allow a non-fracture injection of cement with the ground. Because of a relative-uniformed distribution of cement injection, the increase in the stiffness and the strength of the reinforced region due to the infiltration grouting could significantly improve the seismic capacity, especially reducing huge horizontal shearing strain and vertical settlement when EPWP accumulated during earthquake. In the numerical analysis performed in this research, the reinforced sandy soil was assessed to be a linear elastic material with a Young’s modulus of E = 100 MPa, unit weight of γ = 19.9 kN/m³, and Poisson’s ratio of ν = 0.20. The permeability of the reinforced soil was assessed to be very low and was 100 times less than that of silty sand. The parameters for reinforced sand were similar to those applied in the study by Gu et al. [32] on improving the supporting sandy layer of the piled-raft-soft ground subjected to cyclic train loading. The two purple rectangular areas beneath the embankment depicted in Fig.1 represent the ground-improved zones. The dimensions of the treatment areas, i.e., width of 8 m, and height of 10 m, were determined according to the handbooks of the Level I earthquake established in the Seismic Design Code of Japan.

5.1 Excess pore water pressure after ground improvement

Fig.11 illustrates the time histories of ESDR in typical positions (E₁, E₂) in the top sandy layer, for natural and improved ground and for two input motions. For the natural ground, EPWP accumulated rapidly at E₁ (sandy foundation in the free field) within a few minutes, whereas the highest values of ESDR were approximately 1.0, revealing that the topsoil in this zone was totally liquefied under either of these two earthquake loadings. Its value remained high until EPWP started dissipating in the post-consolidation stage. By comparison, after reinforcement via grouting, the soil at E₁ also turned into its liquefied phase, whereas EPWP dissipated only during the subsequent post-consolidation stage. Meanwhile, the time history of ESDR at E₂ was fairly different from that at E₁. During the EQ₁ shaking, ESDR gradually built up at E₂, and then started to decrease because of shearing dilation. During the strong shaking of EQ₂, oscillation occurred. Generally, the variation trends of ESDR for the reinforced ground were similar to those for the natural ground, whereas the required dissipation time for EPWP became much longer because of the emergence of two impervious reinforced zones.

Fig.12 illustrates the distribution of EPWP for the reinforced ground at the end of the two seismic motions. High EPWP occurred in two distinct areas, i.e., in the interface between the sandy and clayey layers in the free field, and in the lower part (clayey soil) of the two reinforced zones, as the clayey and reinforced sandy layers acted as obstructions for vertical EPWP dissipation. With larger amplitude and PGA, higher pore water accumulation occurred during the seismic motion of EQ₂. A negative EPWP was observed in the upper part of the reinforced zone below the embankment, as volumetric dilatation occurred when the soil underwent shearing deformation to a certain degree.

Fig.13 presents ESDR distribution for the reinforced ground after the completion of each seismic motion. The ESDR of the sandy layer in the free field clearly remained quite high, indicating that the sand in the two sides of the embankment were liquefied even with grouting. Therefore, the challenge with designing the grouting reinforcement for the safe seismic performance of the embankment located on liquefied ground is to securely restrict deformation to an acceptable level, even when the foundation soil is liquefied. Engineering practice has certified that liquefaction countermeasures based on deformation criteria are more cost-effective for earthquake-resistant designs [33].

5.2 Deformation after ground improvement

Fig.14 presents the distributions of the displacement vector of the reinforced embankment for the two input ground motions. For EQ₁ motion, heaving at each toe of the embankment was greatly inhibited, and a stable state was achieved using grouting reinforcement. By comparison, under the huge seismic loading of EQ₂, the circular slip surface disappeared, and deep-seated failure was effectively prevented, as the two grouting-reinforced regions restricted the lateral spreading of the embankment to the free field. It is concluded that grouting had a prominent reinforcement effect against heaving, settlement, and lateral spreading. Nonetheless, instability due to the earthquake and large displacements may occur on the reinforced embankment slope under the seismic loading of EQ₂, and thus an additional countermeasure should be implemented to reinforce the embankment slope.

Different reinforcement thicknesses have distinct effects in alleviating earthquake-induced surface deformation. Fig.15 shows the deformations on the surface of the embankment for various reinforcement thicknesses (H = 0, 3, 6, 10 m). The high-amplitude motion EQ₂ led to more noticeable values for settlement and horizontal deformation (lateral spreading to the free field) at the crest and on both sides the embankment. For natural soil, the largest values for the settlement, at 1.78 m, and lateral movement, at 2.60 m, at the crest were observed after EQ₂ loading; by contrast, the corresponding values were only 0.70 m and 1.20 m under EQ₁ loading. For both seismic motions, the largest upward displacement (approximately 0.50 m) occurred at a position approximately 18 m away from the centerline. Generally, the foundation moved horizontally toward the free field, whereas natural foundation soil at a far distance (approximately 16.5 m away from the toe part) moved toward the embankment because of compatible deformation under the earthquake loadings. After ground reinforcement was applied, the settlement, lateral movement, and uplift were greatly reduced. However, according to the deformation curves for a variety of moderate to strong input earthquakes, the strengthening effect was no longer significant once the reinforcement thickness exceeded 6 m.

6 Conclusions

In this research, reinforcement effects of cement grouting in liquefiable ground, on which an earth embankment is laid, as a liquefaction mitigation method, was numerically investigated via soil−water coupled FE analysis based on the sophisticated CM model. The seismic performance of the embankment was evaluated in terms of its dynamic behaviors, namely, generation of EPWP, effective stress path, responding acceleration, deformation and failure mode, when the embankment was subjected to two distractive seismic motions. The main conclusions obtained are as follows.

1) The liquefaction vulnerability of the earth embankment laid on liquefied ground is largely dependent on the seismic intensity and characteristics of input earthquake motion. Two different seismic motions were considered, in which different failure paths and destruction levels were clearly identified. In the case of the strong motion of E₂, a sliding surface occurred inside the liquefiable area and extended towards the crest in both sides of the earth embankment, together with settlement and lateral spreading. For the moderate motion E₁, however, a local heaving failure at each toe of the embankment was identified, totally different failure mode.

2) In the reinforced ground by cement-grouting, though the soils at both sides of the embankment after earthquake motions still kept a high EPWP, the liquefaction-induced settlement and lateral movement were greatly reduced, because of the high stiffness of the reinforced zones that resisted the further deformation towards free field. Therefore, the contribution of the reinforcement is keeping the deformation of the embankment to an acceptable value but not to prevent soil liquefaction, which is much more difficult to be achieved. In the viewpoint of seismic design, the main purpose of preventing the complete failure of an embankment is achieved if the deformation of the embankment is kept to an acceptable value.

3) Local instability may occur on the embankment slope with grouting remediation when it is subjected to the strong seismic motion EQ₂. Therefore, further measures are necessary to control the sliding failure of the embankment. Meanwhile, the surface deformations like settlement, lateral spreading and uplift were significantly reduced as the thickness of the reinforcement region increased. Yet, in both seismic motions, the reinforcement effect was no longer significant once the thickness exceeded 6 m.

References

[1] Adalier K, Sharp M K. Embankment dam on liquefiable foundation-dynamic behaviour and densification remediation. Journal of Geotechnical and Geoenvironmental Engineering, 2004, 130(11): 1214–1224

[2] Unjoh S, Kaneko M, Kataoka S, Nagaya K, Matsuoka K. Effect of earthquake ground motions on soil liquefaction. Soil and Foundation, 2012, 52(5): 830–841

[3] Singh R, Roy D, Jain S K. Analysis of earth dams affected by the 2001 Bhuj Earthquake. Engineering Geology, 2005, 80(3−4): 282–291

[4] Oka F, Tsai P, Kimoto S, Kato R. Damage patterns of river embankments due to the 2011 off the Pacific Coast of Tohoku Earthquake and a numerical modeling of the deformation of river embankments with a clayey subsoil layer. Soil and Foundation, 2012, 52(5): 890–909

[5] Okamura M, Tamamura S, Yamamoto R. Seismic stability of embankments subjected to pre-deformation due to foundation consolidation. Soil and Foundation, 2013, 53(11): 11–22

[6] Rapti L, Lopez-Caballero F, Modaressi-Farahmand-Razavi A, Foucault A, Voldoire F. Liquefaction analysis and damage evaluation of embankment-type structures. Acta Geotechnica, 2018, 13(5): 1041–1059

[7] ParkYKimSKimSKimM. Liquefaction of embankments on sandy soils and the optimum countermeasure against the liquefaction. In: Proceedings of 12th World Conference on Earthquake Engineering. Auckland: New Zealand, 2000

[8] Koga Y. Applicability of the dynamic centrifuge model test method in developing countermeasures against soil liquefaction. In: Proceedings of International Conference of Centrifuge 91. Colorado: Boulder, 1991, 431–438

[9] Koga Y, Matsuo O. Shaking table tests of embankments resting on liquefiable sandy ground. Soil and Foundation, 1990, 30(4): 162–174

[10] Fern E J, de Lange D A, Zwanenburg C, Teunissen J A M, Rohe A, Soga K. Experimental and numerical investigations of dyke failure involving soft materials. Engineering Geology, 2017, 219: 130–139

[11] Dashti S, Bray J D, Pestana J M, Riemer M, Wilson D. Mechanisms of seismically induced settlement of buildings with shallow foundations on liquefiable soil. Journal of Geotechnical and Geoenvironmental Engineering, 2010, 136(1): 151–164

[12] Matsuo O. Damage to river dikes. Soils and Foundations, 1996, 36(1): 235–240

[13] Elgamal A, Parra E, Yang Z, Adalier K. Numerical analysis of embankment foundation liquefaction countermeasures. Journal of Earthquake Engineering, 2002, 6(4): 447–471

[14] Aydingun O, Adalier K. Numerical analysis of seismically induced liquefaction in earth embankment foundations. Part I. Benchmark model. Canadian Geotechnical Journal, 2003, 40(4): 753–765

[15] Wang Z, Yang Y, Yu H, Muraleetharan K. Numerical simulation of earthquake-induced liquefactions considering the principal stress rotation. Soil Dynamics and Earthquake Engineering, 2016, 90: 432–441

[16] Ayoubi P, Pak A. Liquefaction-induced settlement of shallow foundations on two-layered subsoil strata. Soil Dynamics and Earthquake Engineering, 2017, 94: 35–46

[17] Marcuson W F III, Hadala P F, Ledbetter R H. Seismic rehabilitation of earth dams. Journal of Geotechnical Engineering, 1996, 122(1): 7–20

[18] LuJ CElgamalAYanL JLawK HConteJ P. Large scale numerical modeling in geotechnical earthquake engineering. International Journal of Geomechanics, 2011, 11(6): 490−503

[19] Bradley B A, Araki K, Ishii T, Saitoh K. Effect of lattice-shaped ground improvement geometry on seismic response of liquefiable soil deposits via 3-D seismic effective stress analysis. Soil Dynamics and Earthquake Engineering, 2013, 48: 35–47

[20] Xu L Y, Cai F, Wang G X, Ugai K, Wakai A, Yang Q Q, Onoue A. Numerical assessment of liquefaction mitigation effects on residential houses: Case histories of the 2007 Niigata Chuetsu-offshore earthquake. Soil Dynamics and Earthquake Engineering, 2013, 53: 196–209

[21] Zhang F, Ye B, Ye G L. Unified description of sand behavior. Frontiers of Architecture and Civil Engineering in China, 2011, 5(2): 121–150

[22] YeG L. DBLEAVES: User’s Manual, Version 1.6. 2011 (in Japanese and Chinese)

[23] Asaoka A, Nakano M, Noda T. Superloading yield surface concept for highly structured soil behavior. Soil and Foundation, 2000, 40(2): 99–110

[24] AkaiKTamuraT. Elastoplastic numerical analysis of soil-water coupled problem. Proceedings of JSCE, 1978, 269: 95−104 (in Japanese)

[25] Zienkiewicz O C, Chan A H C, Pastor M, Paul D K, Shiomi T. Static and dynamic behavior of soils: A rational approach to quantitative solutions. I. Fully saturated problem. Proceedings of the Royal Society of London Series A—Mathematical and Physical Sciences, 1990, 285–309

[26] Bradley B A, Araki K, Ishii T, Saitoh K. Effect of lattice-shaped ground improvement geometry on seismic response of liquefiable soil deposits via 3-D seismic effective stress analysis. Soil Dynamics and Earthquake Engineering, 2013, 48: 35–47

[27] Lopez-Caballero F, Modaressi-Farahmand-Razavi A, Stamatopoulos C A. Numerical evaluation of earthquake settlements of road embankments and mitigation by preloading. International Journal of Geomechanics, 2016, 16(5): C4015006

[28] Lu C W, Oka F, Zhang F. Analysis of soil-pile-structure interaction in a two-layer ground during earthquakes considering liquefaction. International Journal for Numerical and Analytical Methods in Geomechanics, 2008, 32(8): 863–895

[29] Ishikawa H, Saito K, Nakagawa K, Uzuoka R. Liquefaction analysis of a damaged river levee during the 2011 Tohoku earthquake. In: 14th International Conference of the International Association for Computer Methods and Advances in Geomechanics. Kyoto: IACMAG, 2015, 673–677

[30] Sasaki Y, Tamura K. Failure mode of embankment due to recent earthquakes in Japan. In: 4th International Conference on Earthquake Geotechnical Engineering. Thessaloniki: RLGEG, 2007, 1479

[31] Coelho P, Haigh S K, Madabhushi S P G. Centrifuge modelling of earthquake effects in uniform deposits of saturated sand. In: 5th International Conference on Case Histories in Geotechnical Engineering. New York: NY, 2004, 13–17

[32] Gu L L, Wang Z, Huang Q, Ye G L, Zhang F. Numerical investigation into ground treatment to mitigate the permanent train-induced deformation of pile-raft-soft soil system. Transportation Geotechnics, 2020, 24: 100368

[33] Finn W D L. State-of-the-art of geotechnical earthquake engineering practice. Soil Dynamics and Earthquake Engineering, 2000, 20(1−4): 1–15

RIGHTS & PERMISSIONS

Higher Education Press