Experimental case study of seismically induced loess liquefaction and landslide

Accepted Manuscript Experimental case study of seismically induced loess liquefaction and landslide

Xiangjun Pei, Xiaochao Zhang, Bin Guo, Gonghui Wang, Fanyu Zhang PII: DOI: Reference:

S0013-7952(16)30555-5 doi: 10.1016/j.enggeo.2017.03.016 ENGEO 4529

To appear in:

Engineering Geology

Received date: Revised date: Accepted date:

29 October 2016 1 February 2017 23 March 2017

Please cite this article as: Xiangjun Pei, Xiaochao Zhang, Bin Guo, Gonghui Wang, Fanyu Zhang , Experimental case study of seismically induced loess liquefaction and landslide. The address for the corresponding author was captured as affiliation for all authors. Please check if appropriate. Engeo(2017), doi: 10.1016/j.enggeo.2017.03.016

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Experimental case study of seismically induced loess liquefaction and landslide

a

CR

IP

T

Xiangjun Peia, Xiaochao Zhanga,*, Bin Guoa, Gonghui Wangb, Fanyu Zhangc

State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu

US

University of Technology, Chengdu, 610059, China b

AN

Research Center on Landslides, Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto, 611-0011, Japan c Key Laboratory of Mechanics on Disaster and Environment in Western China, The Ministry of Education of China; School of Civil Engineering and Mechanics, Lanzhou University, Lanzhou,

ED

M

Gansu 730000, China

AC

CE

PT

*Corresponding: [email protected] (Xiaochao Zhang)

ACCEPTED MANUSCRIPT Highlights

A large loess landslide triggered by the 1920 Haiyuan earthquake was studied



Liquefaction characteristics of saturated loess were analyzed



Steady state strength and deformation characteristics of saturated loess were discussed



Low steady state strength of liquefied loess had play a key role on loess landslide

US

CR

IP

T



mobility

AN

Abstract

M

The 1920 Haiyuan Ms 8.5 earthquake induced a large number of fluidized loess landslides in

ED

China, characterized by low slope angles, long run-out distances, and fluidized movement. The mechanism of these landslides has aroused considerable interest, although additional research is

PT

needed to understand more fully the behavior of the loess and the failure mechanism. Field investigation was conducted on the Shibeiyuan landslide, and loess samples collected for later

CE

laboratory analysis by conventional geotechnical tests, triaxial compression tests, and ring shear

AC

tests. The field survey revealed that the Shibeiyuan landslide occurred on a concave slope (<5°) with a long run-out distance, indicating a small apparent friction angle. It was also found that standing water was present in the landslide area and that the loess had high porosity. Experimental results showed that application of cyclic shear stress to a loose and saturated loess specimen causes excess pore pressure to develop gradually, resulting in a decrease of effective stress until liquefaction. The steady state strength of the loess showed no correlation with stress path, but it was closely related to the degree of saturation, loading rate, and void ratio. This

ACCEPTED MANUSCRIPT indicates that excess pore pressure can accumulate under seismic loading and that plastic deformation can develop rapidly within a shearing zone, resulting in loess liquefaction and a reduction of shear strength. In the Shibeiyuan landslide, the steady state strength was near zero with the large deformation related to the low angle, long distance, and fluidized movement.

IP

T

Keywords: fluidized landslide; loess liquefaction; steady state strength; stress path; apparent

CR

friction angle

US

1. Introduction

Malan loess has geological characteristics of loose structure, high porosity, high water sensitivity,

AN

high compressibility, and collapsibility. The Loess Plateau in China is an area of high seismic

M

intensity, where earthquakes have triggered large numbers of loess landslides, collapses, ground

ED

fissures, and other types of geological hazard. In 1920, a large (Ms 8.5) earthquake occurred in Haiyuan in China, triggering many fluidized landslides. The majority of these landslides

PT

occurred on slopes with low gradients, but they showed high mobility with long sliding distances.

CE

Conventionally, the liquefaction potential of a soil can be related to its grain-size distribution

AC

curve (Yoshimi and Ariizumi, 1991). Soils composed of mainly fine grains are considered to have low liquefaction potential during an earthquake or under cyclic loading; thus, loess is classified with low liquefaction potential. However, field evidence has indicated that loess liquefaction can occur under certain conditions. Based largely on field investigations, Bai and Zhang (1990), Wang and Zhang (1999), and Wang et al. (2001) conducted research on the failure process and mechanism of the Shibeiyuan landslides in China. They postulated that loess liquefaction triggered by strong earthquakes was the main reason for the low angles and long

ACCEPTED MANUSCRIPT sliding distances of the landslides. This work was followed by a series of dynamic liquefaction tests studying the liquefaction characteristics of loess (Wang et al., 2000; Liu and Wang, 2002; Yuan et al., 2004; Deng et al., 2012). Based on the findings of these tests, the criteria governing

T

initial loess liquefaction were discussed but the deformation pattern after initial liquefaction was

IP

not considered. It was proposed that large deformation might not necessarily occur after initial

CR

liquefaction. This was because, in addition to the conditions of the stress–strain relationship, the

US

development of excess pore pressure after the application of the dynamic load is vital to the occurrence of large deformation after liquefaction. Undrained triaxial compression tests have

AN

been undertaken to obtain the steady state strength of undisturbed Malan loess from Lanzhou

M

(China), and the criteria controlling loess liquefaction failure and flowslide were analyzed (Yang

ED

et al., 2004, 2005). Furthermore, triaxial shear tests and large ring shear tests have been conducted by many researchers at Kyoto University to study the apparent friction angle. The

PT

results showed that the water-saturated loess soil was highly susceptible to flow liquefaction

CE

failure (Wang and Sassa, 2002, Zhang and Wang, 2007; Wang et al., 2014a, 2014b). In the

AC

current study, in order to analyze the seismically induced loess liquefaction and failure mechanism associated with the low slope angle and long sliding distance of the Shibeiyuan landslide, a detailed field exploration, geological survey, and number of laboratory tests were undertaken. 2. Shibeiyuan flowslide The Shibeiyuan landslide is a lateral spread landslide located near the city of Guyuan in Ningxia (Fig. 1a), China. Lateral spread landslides occur rapidly and they have long runout distances and

ACCEPTED MANUSCRIPT low angles. The study site is located in the southeast of the Haiyuan meizoseismal area (90 km north of the Haiyuan seismic fault zone), which is an area classified as Level X (i.e., Extreme) on the Modified Mercalli Intensity Scale. An overview of the landslide area is shown in Fig. 1(b). The wavy surface visible in the accumulation zone reflects characteristics commonly caused by

T

liquefaction (Fig. 1c). As shown in Fig. 2, the landslide was 1.2 km long and 2.2 km wide with a

IP

principal sliding direction of 270°. It affected an area of 2.64 km2 with a volume of over 37 × 106

CR

m3 of material deposited with an average thickness of 10–35 m. A longitudinal section (A–B) is shown in Fig. 3. The elevation in the study area ranges from 1640 to 1710 m with a slope

US

gradient of 3–5° (max. <10°). The main recharge of groundwater in the area is via precipitation,

AN

and the groundwater depth ranges from 8 to 20 m. The loess stratum comprises Late Pleistocene

M

Malan loess and Middle Pleistocene Lishi loess in sequence with laminated paleosol layers.

ED

3. Experimental study

To analyze the failure mechanism of the landslide, a series of laboratory tests was conducted in

PT

addition to the fieldwork. The dynamic properties of the saturated loess and its steady state

CE

strength were analyzed, from which the apparent friction angle was obtained, and the mechanism related to the low slope angle and long sliding distance of the landslide was investigated.

AC

The interplay between strength and deformation after initial liquefaction under a constant shearing force has direct impact regarding long run-out flow slides (Wang and Sassa, 2010; Xu et al., 2012a, 2012b; Wang, B. et al., 2013). An analysis of loess liquefaction should not only focus on the criteria controlling initial liquefaction but also on the dissipation of excess pore pressure, stress–strain behavior, and steady state strength. The correlations of steady state deformation and mechanical behavior after the cessation of ground movement were analyzed using dynamic triaxial compression tests. To study the mechanism governing the high mobility

ACCEPTED MANUSCRIPT and long run-out of loess landslides, ring shear tests were conducted on undrained loess samples with shear displacements of up to several meters. 3.1 Samples and apparatus Undisturbed loess samples were used in the conventional geotechnical tests to determine the

IP

T

physics and mechanical parameters, as well as to analyze the microstructural characteristics of

CR

the loess. However, reconstituted samples were used in the triaxial and ring shear tests because of the difficulties involved in sampling, transporting, and preparing undisturbed samples. Thus,

US

minor differences between the steady state undisturbed and reconstituted soil samples were neglected in this study. The undisturbed samples obtained from the source area of the landslide

AN

consisted mainly of silt-sized grains with specific gravity of 2.70 (Fig. 4). The soil layers at

M

different elevations had different void ratios ranging from 0.56 to 1.02. The void ratios of the

ED

loess samples from the back scarp were about 0.83–1.02, i.e., larger than the slip zone samples that were about 0.56–0.69 (Table 1).

PT

The microstructural characteristics of low density, high porosity, and low plasticity of the

CE

undisturbed samples from the back scarp were captured using an S-3000N scanning electron microscope (Fig. 5). MTS810 triaxial dynamic test equipment (Zhang X. et al., 2014) was used

AC

to examine the possible occurrence of liquefaction, and the DPRI-5 large ring shear test apparatus of Kyoto University was used to analysis the undrained shear strength at the ―real steady state‖ (Poulos, 1981). The size of the triaxial test samples was 50 × 100 mm (D × L), and the ring shear samples had inner and outer diameters of 120 and 180 mm, respectively, height of 38.3 mm, and cross-sectional area of 141.37 mm2 (Wang and Sassa, 2002).

ACCEPTED MANUSCRIPT 3.2 Test procedures To ensure homogeneity of the samples, a series of consolidated–undrained triaxial compression tests was conducted on the reconstituted samples. A loess sample with designated water content was deposited into a compaction container in several layers, where each layer was tamped to

IP

T

achieve the desired density. The size of the reconstituted specimens was 50 × 100 mm (D × L). The samples were saturated with de-aired water assisted by carbon dioxide. The degree of

CR

saturation was reflected by BD, which is the ratio between the increment of excess pore pressure

US

(Δu) and the increment of normal stress (Δσ) under undrained conditions (BD = Δu/Δσ) (Skempton, 1954; Sassa, 1985; Yoshimi and Ariizumi, 1991). The samples were considered fully

AN

saturated when BD ≥ 0.95.

M

Firstly, the saturated specimens were consolidated with a confining pressure of 100 kPa. After

ED

consolidation, 1 Hz sinusoidal cyclic loading was imposed with a dynamic shear stress ratio of 0.2. When the dynamic strain reached 1%, 2%, and 3%, the cyclic load was changed to

PT

undrained monotonic shear at the velocity of 0.5 mm/min. The test was terminated if the axial

CE

strain reached 25% or if the sample reached steady state. Stress, strain, and pore pressure were monitored throughout the tests.

AC

In addition, conventional undrained triaxial compression tests were performed on other samples following the same preparation method above at a velocity of 0.5 mm/min, to obtain the steady state strength under static loading for consolidation stresses of 100, 150, 200, and 300 kPa. Compared with the ring shear tests, the strain level in a triaxial test is limited and thus, the ―real steady state,‖ as defined by Poulos (1981), might not be reached by the end of the triaxial compression test. Therefore, the ring shear apparatus was used for low porosity samples at large

ACCEPTED MANUSCRIPT shear displacements of up to several meters under undrained conditions. Ring shear apparatus is used to test the residual shear strength of remolded annular soil samples and to simulate the deformation mechanism of landslides and other large strains. The main advantage of this method over the direct shear test is that the area of continuous shear is kept constant throughout the

T

experiment. Using this method, one can accurately reproduce the residual shear stress field in the

IP

laboratory. The samples were consolidated and tested at different confined pressures, e.g., 100,

CR

200, and 300 kPa, following which the average void ratio was 0.56 and the BD values were 0.56, 0.83, and 0.95, respectively. Shear tests were performed at a shearing velocity of 0.1 mm/s until

US

reaching a steady state (i.e., with displacement of 100 mm about). Stress, strain, and pore

AN

pressure were monitored throughout the tests. The effects of effective normal stress, saturation,

M

and shearing velocity on the steady state strength were considered.

ED

3.3 Test results

Figures 6–13 present the results of the undrained triaxial shear tests on saturated loess samples

PT

under both dynamic and static loads. Figures 14–17 present the results of the ring shear tests on

CE

loess samples at a normal pressure of 200 kPa, with variations in the shear stress, pore water pressure, axial strain, and effective stress path. It can be seen that excess pore pressure was

AC

generated, which resulted in significant reduction in the effective stress during shearing. Irrespective of the type of loading or test, many differences are apparent in the results, which can be attributed to other reasons such as degree of saturation and the void ratio.

ACCEPTED MANUSCRIPT 4. Liquefaction and steady state strength of loess 4.1 Liquefaction under dynamic loading Figure 6 shows the stress–strain hysteresis loop of the saturated loess, which indicates that during dynamic liquefaction, the loess developed viscoelasticity at the onset of the seismic wave.

IP

T

The sample had low pore pressure, slow strain increase, and a narrow stress–strain hysteresis

CR

loop. During the process of dynamic loading, the sample appeared to demonstrate plasticity. Excess pore pressure was generated instantaneously at the beginning of the collapse phenomenon,

US

when the sample showed characteristics of plasticity that were more evident. Strain increased continually and the hysteresis loop widened gradually with the seismic cycle. The pore pressure

AN

increased to up to 70% of the initial cell pressure when the strain was near 3% (Fig. 7), which

M

can be considered as one of the critical points for initial liquefaction (Wang et al., 2000). Loess

ED

liquefaction has been reported in earlier works and liquefaction-induced loess flowslides have received considerable attention (Wang et al., 2000, 2014a, 2014b; Zhang F. et al., 2014; Zhang X.

PT

et al., 2014). However, fundamental understanding of the factors governing the changes of stress,

CE

strain, and pore pressure after dynamic loading is lacking; thus, further analyses based on special undrained shear tests were undertaken in this study.

AC

4.2 Deformation after dynamic loading The accumulated strain reflects the level of dynamic damage for the same soil type. The shearing curves of the same loess soil under three different initial dynamic strains are shown in Fig. 8(a– c). It is clear that the pore pressure ratio increased rapidly under seismic loading and reached 0.7 and >0.8 at 1% and 3% strain, respectively. Pore pressure did not dissipate immediately but

ACCEPTED MANUSCRIPT continued to increase until it achieved the same steady state for all three curves. Conversely, saturated loess strength decreased gradually until it achieved steady state. In order to understand the influence of stress path on strength, the relation curves of strain, stress, and pore pressure under different loading conditions were analyzed (Figs. 9 and 10). Figure 9

IP

T

reveals that the loess structure was damaged by dynamic loading. The strength decreased continuously with the growth of strain, although the rate of decrease varied for the different

CR

samples. After dynamic loading, strain continued to grow with monotonic shearing until all

US

samples achieved steady state at the same value of strength. This implies that dynamic load has no effect on the steady state strength of loess under similar conditions (e.g., saturation, effective

AN

normal stress, and void ratio) except for accelerating the process. Pore pressure follows the same

M

relation with stress path, showing insignificant effects. Figure 10 shows the samples had various pore pressures after dynamic loading, but with the growth of strain, ultimately, they all reached

ED

the same stable value. This means that dynamic load has no effect on pore pressure at steady

PT

state but that it significantly accelerates the growth velocity. Therefore, it can be inferred that a new structure and steady state were formed during the process of large deformation. Thus, static

AC

method.

CE

load could be used to obtain the steady state strength regardless of the stress history or loading

4.3 Steady state strength and deformation under static loading The steady state of deformation for any mass of particles is that state in which the mass is continuously deforming at constant volume, constant normal effective stress, constant shear stress, and constant velocity. Steady state strength plays a very important role in both the evaluation of stability and the analysis of deformation of liquefiable soils affected by earthquakes. Low steady state strength after liquefaction was deemed the key reason for the

ACCEPTED MANUSCRIPT occurrence of large deformation and flowslides on loess slopes (Wang and Sassa, 2002; Zhang and Wang, 2007; Wang et al., 2014a, 2014b). Laboratory tests can be conducted to determine the apparent friction angle from the steady state strength (Zhang and Wang, 2007). The stress–strain, pore pressure, and stress path curves are shown in Figs. 11–13. The tests

IP

T

revealed that under the same void ratio of 1.02, although the confining pressures were different (<200 kPa), the effective stress decreased with the accumulation of excess pore pressure and the

CR

steady state strength decreased continuously to zero. Flowslides can be triggered easily by a

US

small shearing force under such conditions. It was found that the initial normal stress has no effect on steady state strength, although significant differences exist in peak strength. Therefore,

AN

greater initial consolidation stress means that a lower apparent friction angle is required for

M

liquefaction-induced failure.

ED

Large ring shear tests were conducted using samples with a low void ratio, initial consolidation normal stress of 200 kPa, and a void ratio after consolidation of 0.56. To study the effects of

PT

saturation on steady state strength, values of BD of 0.56, 0.83, and 0.95 were selected. The

CE

shearing curves and paths are shown in Figs. 14 and 15. Based on the monitoring of the pore pressure during shearing, the pore pressure ratio curves are shown in Fig. 16.

AC

4.3.1 Effect of saturation on steady state strength The effect of saturation on steady state strength was analyzed under constant consolidation stress. The results revealed that the peak value of pore pressure and steady state strength have greater trends of decrease with increasing saturation. Under fully saturated conditions (BD = 1), the pore pressure ratio increased rapidly at the onset of shearing to reach 0.6 after 50 s. The rate of change then reduced as the pore pressure ratio reached 0.7 at 200 s before becoming stable soon after.

ACCEPTED MANUSCRIPT Under the condition of BD = 0.56, the pore pressure ratio increased slowly to reach 0.25 after 1000 s. The rate of change and the peak value of pore pressure decreased with decreasing saturation. This indicates that high initial saturation might result in a quicker response, generating higher pore pressure that in turn causes liquefaction in loess.

IP

T

4.3.2 Effect of shearing velocity on steady state strength

CR

Several shearing tests under constant consolidation stress were conducted with different shearing velocities to study the effects of shearing velocity on the steady state strength. Figure 17 shows

US

the influence of shearing on steady state strength. While its influence is insubstantial when the velocity is <0.1 mm/s, the shearing strength decreases with an increase of shearing velocity in

AN

the range 0.1–1.0 mm/s. However, the shearing strength exhibits minimal changes when the

M

velocity is >10 mm/s. In general, shearing velocity has greater influence on shearing strength at

ED

higher degrees of saturation and little influence under dry conditions.

PT

4.3.3 Effect of void ratio on steady state strength The steady state strength of loess has a close relationship with the void ratio (Fig. 18). A larger

CE

void ratio and smaller density can lead to lower steady state strength. For a known stable state

AC

curve, a flowslide could occur when the driving stress is above the curve, whereas it would not occur when the driving stress is below the curve. According to the steady state strength and the initial stress state, the apparent friction angle can be estimated. The void ratio of the loess at the slip surface of the Shibeiyuan slope is 0.91–1.05 (Fig. 19) with low steady state strength near zero; therefore, a flowslide could be triggered easily by a driving shearing force located above the curve.

ACCEPTED MANUSCRIPT 5. Liquefaction and flowslide of saturated loess under seismic loading Some previous research has explained the deformation process after liquefaction under seismic loading (Zhang and Sassa, 1996; Zhang and Wang, 2007; Zhang et al., 2009). The stress path curve of a seismically induced landslide is shown in Fig. 19 (initial pore pressure is neglected),

IP

T

where I is the initial shear stress of the slip surface and θ is the slope angle. Deformation of a loess layer after an earthquake can be either large (S) or small (R) depending on the excess pore

CR

pressure due to the seismic loading. In Fig. 19, PFL is the collapse curve for the peak value of

US

effective stress space and RFL is the steady state curve for the stable value of effective stress space.

AN

In the analysis of landside mobility, the concept of travel angle (Áa) has been used widely

M

(Scheidegger, 1973; Zhang and Wang, 2007), where Áa is defined as tanÁa = H/L (H is the

ED

height of the landslide, i.e., the difference in elevation between its crown and tip, and L is its horizontal distance). Additionally, Áa is known as the apparent friction angle and a low apparent

PT

friction angle means high mobility. Sassa (1985) emphasized that the apparent friction angle

CE

during landslide motion is chiefly a combination of the real internal friction angle (ψm) and the pore pressure (u), which suggests that the approximate apparent friction angle (ψa) can be

AC

expressed as in Eq. (1):

tanψa ≈ (σ0 – u)/σ tanψm

ψa—apparent friction angle; ψm—internal friction angle at steady state; σ0—initial static normal stress;

(1)

ACCEPTED MANUSCRIPT μ—excess pore pressure at steady state. In the nineteen eighties (Sassa,1985,1988), the value was estimated based on direct shear and drained ring shear tests. After the undrained ring shear apparatus was developed (Sassa, 1996,1998), ψa could be obtained from Eq. (2), where τss is the shear strength at steady state and

IP

T

σ0 is the initial consolidation normal stress:

(2)

CR

tanψa= τss/σ0.

When loess is fully saturated, excess pore pressure accumulates and the effective stress decreases

US

under a dynamic load. Failure occurs when the soil stress state reaches the peak strength F (Fig.

AN

19). As the excess pore pressure continues to increase, and stress path develops along the steady state curve RFL until reaching point S. Meanwhile, deformation continues if the slope gradient is

M

greater than the apparent friction angle ψa. With low saturation of loess, excess pore pressure

ED

might be not generated under dynamic loading and therefore, failure occurs along F'-R, where the deformation is terminated at point R. No long run-out flowslide occurs if the slope gradient is

PT

less than the apparent friction angle.

CE

The experimental results showed that continuous deformation after an earthquake is affected by

AC

the steady state shear strength, which depends on the development of pore pressure. Although the slope gradient at Shibeiyuan was very low before the earthquake, the potential sliding surface had a large void ratio with a shallow groundwater level. As the results showed, the excess pore pressure of the saturated loess layer, triggered by the earthquake, failed to dissipate immediately after the seismic event. This resulted in a continuous decrease in effective stress, and the shear resistance of the saturated loess layer near the sliding surface dropped to a very small value because of liquefaction. Therefore, as the steady state strength was near zero, even a small

ACCEPTED MANUSCRIPT driving force could trigger a flowslide because of the low apparent friction angle. It is estimated that the shear stress acting on the potential sliding surface (τ = Hγd sinθcosθ) (Wang et al., 2014b) before the slide was about 19.5–32.5 kPa (a possible mean thickness (H) was about 25 m, the weight (γd) was about 15.0KN/m3, and the slope angle (θ) was about 3–5°). Thus, the

T

considerable difference between the driving shear stress and the lowered shear resistance resulted

IP

in a low apparent friction angle, long run-out distance, and high mobility of the loess landslide.

CR

6. Conclusions

US

Based on field investigations and the laboratory experiments performed at the State Key Laboratory of Geohazard Prevention and Geoenvironment Protection and Kyoto University, the

AN

conclusions based on this research can be summarized as follows:

M

(1) Excess pore pressure of the Shibeiyuan loess was generated and continued to increase in the

ED

monotonic shearing test after dynamic loading with no observable dissipation. The shearing strength exhibited a tendency of reduction until a steady state was achieved. The steady state

PT

strength of saturated loess was not affected by the historical stress path at the same void ratio and

CE

effective mean normal stress.

AC

(2) Analysis of the steady state strength and deformation characteristics of saturated loess revealed that steady state strength decreased with increases of both the degree of saturation and the void ratio. The void ratio had a unique relationship with stress state under similar saturated conditions. (3) Shearing velocity influenced the steady state strength. This influence was minimal when velocity was <0.1 mm/s, but shearing strength decreased as the velocity increased between 0.1 and 1.0 mm/s. However, shearing strength tended to become stable with little change when the

ACCEPTED MANUSCRIPT shearing velocity was >10 mm/s. Samples with different degrees of saturation showed different shearing behaviors. In general, velocity had greater influence on shearing strength at higher degrees of saturation. (4) Excess pore pressure of the Shibeiyuan loess accumulated under seismic load. Plastic

IP

T

deformation developed rapidly in the shearing zone, which led to loess liquefaction, and the shearing strength decreased continually toward a steady state with large deformation. The

US

out distance, and fluidized behavior of the landslide.

CR

apparent friction angle was very low (i.e., near zero), which resulted in the low angle, long run-

AN

Acknowledgements

This study was partially supported by the Natural Key Basic Research Development Plan of

M

China (No. 2014CB744703) and the National Natural Science Foundation of China (No.

ED

41402240).

PT

References

Bai, M., Zhang, S. 1990. Landslide induced by liquefaction of loess during earthquake of high

CE

intensity. Geotechnical Investigation and Surveying 6: 1-5 (in Chinese).

AC

Deng, L., Fan, W., HE, L., 2012, Liquefaction property of loess under stochastic seismic load. Chinese Journal of Rock Mechanics and Engineering 31(6): 1274-1280 (in Chinese). Liu, H., Wang, L., 2002. Study on the liquefaction mechanism and pore microstructure of saturated loess. Northwestern Seismological Journal24 (2): 135-139 (in Chinese). Poulos, S. J., 1981. The Steady State of Deformation. Geotechnical Engineering Division, ASCE, 107(5): 553-561.

ACCEPTED MANUSCRIPT Sassa, K., 1985. The mechanism of debris flows. In Proceedings of the 11th International Conference on Soil Mechanics and Foundation Engineering, San Francisco, Calif. Vol. 3, pp. 1173-1176. Sassa, K., 1988. Geotechnical model for the motion of landslides. Special Lecture of 5th

IP

T

International Symposium on Landslides. Landslides, vol. 1. Balkema, Rotterdam, pp. 37-55.

CR

Sassa, K., 1996. Prediction of earthquake induced landslides, Special Lecture of 7th International Symposium on Landslides. Landslides, Balkema, Rotterdam, 1:115-132.

US

Sassa, K., 1998. Mechanisms of landslide triggered debris flow. In: Sassa, K. (Ed.), Proc. of

AN

IUFRO Div. 8 Conference, Environmental Forest Science, Kluwer Academic Publisher, Kyoto, pp. 471-490.

ED

Rock Mechanics,5(4),231-236.

M

Scheidegger, A.E., 1973. On the Prediction of the Reach and Velocity of Catastrophic Landslides.

PT

Skempton, A.W., 1954. The pore-pressure coefficients A and B. Géotechnique 4, 1751-7656.

CE

Wang, B., Zen, K., Chen, G. Q., et al., 2013. Excess pore pressure dissipation and solidification after liquefaction of saturated sand deposits. Soil dynamics and earthquake engineering 51,47-

AC

57.

Wang, F., Sassa, K., 2010. Landslide simulation by a geotechnical model combined with a model for apparent friction change. Physics and Chemistry of the Earth 35,149-161. Wang, J., Zhang, Z., 1999.A study of the mechanism of high speed loess landslide induced by earthquake. Chinese Journal of Geotechnical Engineering 21(6): 670-674(in Chinese).

ACCEPTED MANUSCRIPT Wang, J., Bai, M., Xiao, S. 2001. A study of compound mechanism of earthquake-related sliding displacements on gently inclined loess slope. Chinese Journal of Geotechnical Engineering 23(4): 445-449 (in Chinese). Wang, G., Li, T., Xing, X., Zou, Y., 2014a. Research on loess flowslides induced by rainfall in

IP

T

July 2013 in Yan’an, NW China. Environmental Earth Sciences, 1-12.

CR

Wang, G., Sassa, K., 2002. Post-failure mobility of saturated sands in undrained load-controlled ring shear tests. Canadian Geotechnical Journal 39, 821-837.

US

Wang, G., Zhang, D., Furuya, G., Yang, J., 2014b. Pore-pressure generation and fluidization in a

AN

loess landslide triggered by the 1920 Haiyuan earthquake, China: A case study. Engineering Geology 174, 36-45.

M

Wang, L., Liu, H., Li, L., et al., 2000. Laboratory study on the mechanism and behaviors of

ED

saturated loess liquefaction. Chinese Journal of Geotechnical Engineering 22(1), 89-94 (in

PT

Chinese).

Xu, L., Dai, F.C., Gong, Q.M., Tham, L.G., Min, H., 2012a. Irrigation-induced loess flow failure

CE

in Heifangtai Platform, North-West China. Environmental Earth Sciences 66, 1707-1713.

AC

Xu, L., Dai, F.C., Tham, L.G., Zhou, Y.F., Wu, C.X., 2012b. Investigating landslide-related cracks along the edge of two loess platforms in northwest China. Earth Surface Processes and Landforms 37, 1023-1033. Yoshimi, Y; Ariizumi, K.1991. Evaluation of liquefaction resistance of sand improved by deep vibratory compaction. International Journal of Rock Mechanics and Mining Sciences and Geomechanics Abstracts 28 (3):153-158.

ACCEPTED MANUSCRIPT Yang, Z., Zhao, C., Wang, L., et al., 2004.Liquefaction behaviors and steady state strength of saturated loess. Chinese Journal of Rock Mechanics and Engineering 23 (22): 3853-3860 (in Chinese).

IP

of Rock Mechanics and Engineering 24(5): 864-871 (in Chinese).

T

Yang, Z., Zhao, C., Wang, L. 2005.Testing study of saturated loess liquefaction. Chinese Journal

CR

Yuan, Z., Wang, L., Yasuda, S., et al. 2004.Further study on mechanism and discrimination criterion of loess liquefaction. Earthquake Engineering and Engineering Vibration 24 (4):

US

164-169.

AN

Zhang, D., Sassa, K., 1996. Study on the Mechanism of Loess Landslides Induced by Earthquakes. Japan Society of Erosion Control Engineering 49, 4-13 (in Japanese).

M

Zhang, D., Wang, G., 2007. Study of the 1920 Haiyuan earthquake-induced landslides in loess

ED

(China). Engineering Geology 94, 76-88.

PT

Zhang, D., Wang, G., Luo, C., Chen, J., Zhou, Y., 2009. A rapid loess flowslide triggered by

CE

irrigation in China. Landslides 6, 55-60. Zhang, F., Wang, G., Kamai, T., Chen, W., 2014. Effect of pore water chemistry on undrained

AC

shear behaviour of saturated loess. Quarterly Journal of Engineering Geology and Hydrogeology 47, 201-210. Zhang, X., Huang, R., Xu, M., et al., 2014. Loess liquefaction characteristics and its influencing factors of Shibeiyuan landslide. Rock and Soil Mechanics 35,801-810 (in Chinese). Table headings Table 1 Basic physical parameters of the tested loess

ACCEPTED MANUSCRIPT Figure captions

Fig. 1. The Shibeiyuan landslide: (a) location of study area, (b) overview of the Shibeiyuan landslide, and (c) deposition area

T

Fig. 2. Engineering geological map of the Shibeiyuan landslide

IP

Fig. 3. Engineering geological longitudinal section of the Shibeiyuan slope along line A–B in Fig.

CR

2

US

Fig. 4. Particle size distribution of tested sample

Fig. 5. Microstructural characteristics of loess sample under scanning electron microscope

AN

Fig. 6. Stress–strain hysteresis curve under 100 kPa confining pressure

M

Fig. 7. Strain–pore pressure ratio–cycle relation curve

ED

Fig. 8. Monotonic shearing curves with different initial dynamic strains: (a) 1%, (b) 2%, and (c) 3%

PT

Fig. 9. Stress path curves under different initial strains

CE

Fig. 10. Pore pressure and axial strain curves under different initial strains

AC

Fig. 11. Deviator stress and axial strain relation curves under consolidated–undrained shear test Fig. 12. Pore water pressure and axial strain relation curves under consolidated–undrained shear test

Fig. 13. Stress path curves under consolidated–undrained shear test Fig. 14. Time history curves of loess ring shear test under different saturations Fig. 15. Stress path curves of loess ring shear test under different saturations

ACCEPTED MANUSCRIPT Fig. 16. Pore pressure ratio curves of loess ring shear test under different saturations Fig. 17. Relationship between shear rate and shear resistance Fig. 18. Curve of void ratio and stable state strength relation

T

Fig. 19. Stress path of seismically triggered landslide (Zhang and Wang, 2007)

IP

Tables:

Initial moist bulk density (g/cm3)

Initial water content (%)

Void ratio

Specific gravity

1

1.36

1.86

1.02

2.7

2

1.48

10.02

1.01

2.71

3

1.45

2.8

0.91

4

1.55

5.3

0.83

5

1.79

11.14

0.69

6

2.09

21.11

0.56

Plastic limit (%)

Plasticity index (%)

31.1

19

12.1

32.8

19.49

2.69

26.89

2.7

ED PT CE AC

Cohesion(KPa)

friction angle(°)

13.31

/ 26.90

/ 25.15

20.13

6.76

/

/

31.32

21.15

10.17

2.72

33.85

20.48

13.37

37.71 21.52

26.97 25.2

2.7

29.4

21.1

8.3

16.70

17.56

AN

Liquid limit (%)

M

Sample number

US

CR

Table 1 Basic physical parameters of the tested loess

ACCEPTED MANUSCRIPT

270°

ED

M

AN

US

CR

IP

T

Figures:

AC

CE

PT

Fig. 1. The Shibeiyuan landslide: (a) location of study area, (b) overview of Shibeiyuan landslide, and (c) deposition area

ED

M

AN

US

CR

IP

T

ACCEPTED MANUSCRIPT

1-houses, 2-road, 3-ditch, 4-exploration section, 5-spring, 6-stratigraphic boundary,

PT

7-landslide boundary, 8-sampling locations, 9-shear outlet of landslide

AC

CE

Fig. 2. Engineering geological map of Shibeiyuan landslide

CR

IP

T

ACCEPTED MANUSCRIPT

1-Malan loess, 2-Lishi loess, 3-paleosol, 4-landslide deposits, 5-groundwater table,

US

6-original topography, 7-present topography, 8-slip surface, 9-sampling location

AC

CE

PT

ED

M

AN

Fig. 3. Engineering geological longitudinal section of the Shibeiyuan slope along Line A–B in Fig. 2

CR

IP

T

ACCEPTED MANUSCRIPT

(a)

AC

CE

PT

ED

M

AN

US

Fig. 4. Particle size distribution of tested sample

(b) Fig. 5. Microstructural characteristics of loess sample under scanning electron microscope (a) Original image magnified 200 times, (b) Binary image after processing

IP

T

ACCEPTED MANUSCRIPT

ED

M

AN

US

CR

Fig. 6. Stress–strain hysteresis curve under 100 kPa confining pressure

AC

CE

PT

Fig. 7.Axial strain–pore pressure ratio–vibration number relation curve

ACCEPTED MANUSCRIPT

60

0.6

0

0.4

Deviator stress Pore pressure ratio

0.2

-40 0

5

10 15 20 Axial strain(%)

25

30

0.0

CR

-60

T

-20

Pore pressure ratio

0.8

20

IP

Deviator stress(KPa)

1.0 40

60

AN

20 0 -20

ED

-60

0

5

10

15

0.8

0.4

Deviator stress (KPa) Pore pressure ratio

-40

1.0

0.6

M

Deviator stress (KPa)

40

20

Pore pressure ratio

US

(a) Initial dynamic strain of 1%

0.2 0.0 30

25

Axial strain(%)

60

1.0

40 20

0.8

0

0.6

-20

0.4

Deviator stress Pore pressure ratio

-40

0.2

Pore pressure ratio

Deviator stress(KPa)

AC

CE

PT

(b) Initial dynamic strain of 2%

-60 0

5

10

15

20

25

30

0.0

Axial strain(%)

(c) Initial dynamic strain of 3%

Fig. 8. Monotonic shearing curves with different initial dynamic strains: (a) 1%, (b), 2%, and (c) 3%

CR

IP

T

ACCEPTED MANUSCRIPT

Fig. 9. Stress path curves under different initial strains

US AN

0.8 0.6 0.4 0.2

ED

0.0 0

ε0=1% ε0=2% ε0=3%

M

Pore pressure ratio

1.0

5

10

15

20

25

30

Axial strain /%

AC

CE

PT

Fig. 10. Pore pressure and axial strain curves under different initial strains

Fig. 11. Deviator stress and axial strain relation curves under consolidated–undrained shear test

IP

T

ACCEPTED MANUSCRIPT

ED

M

AN

US

CR

Fig. 12. Pore water pressure and axial strain relation curves under consolidated–undrained shear test

AC

CE

PT

Fig. 13. Stress path curves under consolidated–undrained shear test

Fig. 14. Time history curves of loess ring shear test under different saturations

IP

T

ACCEPTED MANUSCRIPT

M

AN

US

CR

Fig. 15. Stress path curves of loess ring shear test under different saturations

AC

CE

PT

ED

Fig. 16. Pore pressure ratio curves of loess ring shear test under different saturations

Fig. 17. Relationship between shear rate and shear resistance

ED

M

AN

US

Fig. 18. Curve of void ratio and stable state strength relation.

CR

IP

T

ACCEPTED MANUSCRIPT

AC

CE

PT

Fig. 19. Stress path of seismically triggered landslide (Zhang, 2007)