What Is Static Liquefaction Failure of Loose Fill Slopes?
What Is Static Liquefaction Failure of Loose Fill Slopes?
What Is Static Liquefaction Failure of Loose Fill Slopes?
Charles W. W. Ng
The Hong Kong University of Science and Technology, Hong Kong SAR
ABSTRACT: Static liquefaction failure of soil slopes has often been reported in literature. It appears that
some researchers and engineers use different criteria to define and describe static liquefaction and they refer to
different failure mechanisms. What is static liquefaction? How is it triggered? How can we identify and define
static liquefaction failures? Does a strain-softening material necessarily mean static liquefaction? These are
not all easy questions to answer and some of them may be even controversial. Based on some centrifuge model and triaxial element tests, suggested answers to some of these questions are explored, discussed and verified in this paper.
1 INTRODUCTION
Slope failures occur in many parts of the world. A
slope will become unstable when its shear resistance
is smaller than any external driving shear stress,
which may be induced by mechanical and hydraulic
means such as rainfall, earthquake, vibration and
seepage. Alternatively, a slope will also become unstable if its shear resistance is deteriorated and reduced due to weathering and any other mechanisms
such as static liquefaction. Very often the terminology static liquefaction is used to describe soil slope
failures and reported in literature. However, it is evident that different researchers and engineers may refer to different failure mechanisms. Some use debris
mobility (travel angle or run out distance) to judge
whether a slope failure is caused by liquefaction or
not. Clearly there is no direct relationship between
liquefaction and mobility. For instance, a level
ground can liquefy (at zero/small effective stress under seismic loading) with zero run out distance. On
the contrary, a steel ball can run down a bare slope to
reach a very long travel distance and this is nothing
to do with liquefaction or not (Ng 2007).
What is static liquefaction? How is it triggered?
What is the effective stress at failure, if the slope is
fully saturated initially such as undersea slopes?
How can we identify and define static liquefaction
failures? Does a strain-softening material necessarily
mean static liquefaction? Is there any difference between slide failure and flow failure? What is the role
of hydrofracture? How the angle of a slope affects
the so-called static liquefaction? Is there any difference between fluidization and liquefaction? Will
Deviator stress
(a)
B Strain hardening
Dilation
Strain softening
C
Strain hardening
Limited
liquefaction
A
Axial strain
Deviator stress
haviour observed in the laboratory is rather confusing and, strictly speaking, incorrect. Would it be
clearer and more precise to describe the material behaviour of the loose specimen, A, and a dense
VSHFLPHQ % DV VWUDLQ-VRIWHQLQJ DQG VWUDLQKDUGHQLQJUHVSHctively, in the deviator stress-axial
strain space (see Fig. 1a)? In the mean effective
stress-deviator stress space (see Fig. 1b), would it be
more precise to use the terms XQGUDLQHG VWUHQJWK
UHGXFWLRQ RU VR-called collapse (Sladen et al.
DQGXQGUDLQHGVWUHQJWKLQFUHDVHWRGescribe
the strength changes of specimen A and specimen B,
respectively? Of course, it is well-recognised that a
reduction and an increase in undrained shear strength
are caused by the respective tendency of sample contraction and dilation, leading to a respective increase
and a reduction in pore water pressure ('u) for
specimens A and B during undrained shearing (see
relationship between 'u and axial strain in Figure
1c). It must be pointed out that these are just material
element behaviour that does not necessarily capture
and represent the global behaviour of an entire fill
slope or an earth structure.
B Dilation
(b)
Undrained
strength increase
Phase transC
due to dilative
formation
tendency
point
Limited
liquefaction
contractive tendency
A
Liquefaction Undrained strength reduction
due to contractive tendency
Mean effective stress
Contractive tendency Liquefaction
(c)
A
Axial strain
Limited li- C
quefaction
contractive tendency
B
60
40
20
0
0.001
LVDT
0.01
0.1
Particle size (mm)
10
Drainage
board
Temporary
reservoir
(a)
700
Model
container
Inlet hole
PPT7
Reflector
y
PPT5
q (kPa)
PPT6
PPT4
PPT2
PPT1
600
500
e0=0.973
Model scale
305
80
29.4
100
Outlet hole
PPT3
Sand
1130,7
1130.7
400
e0=0.970
300
Quasi-steady state
Quasi-steady state
200
e0=0.983
100
e0=0.992
0
0
10
20
30
40
Hq (%)
700
(b)
500
300
100
Figure 3. Contractive behaviour of loose LB sand under consolidated undrained tests (a) in the Hq - q and (b) in pc - q planes
(modified from Zhang 2006, data from Cai 2001).
collapsed rapidly (undrained) which was then followed by the collapse of Z3 (undrained) without inducing obvious deformation in the lower part. The
collapses of Z2 and Z3 were due to the strainsoftening associated with the significant strength reduction (i.e. high liquefaction potential) of the loose
LB sand as illustrated in Figure 3. The rapid
undrained collapses of Z2 and Z3 were evident from
the measured large excess positive pore pressures at
PPT7 (see Fig. 6).
B
20
Water flow
Sand movement
Gully head
A
Water
Z1
Z2
15
PPT7
10
Z3
PPT5
PPT4
PPT6
PPT2
PPT1
Slope profile
before failure
Z4
PPT3
Z5
A
0
0
10
15
20
25
30
35
40
45
50
55
60
65
(m)
0.6
PPT7
0.4
PPT4
PPT5
'
0.8
0.2
0.0
-0.2
PPT2
PPT1
PPT6
-0.4
-0.6
PPT3
-0.8
-1.0
37.8 38.2 38.6 39.0 39.4 39.8 40.2 40.6 41.0 41.4 41.8 42.2 42.6 43.0
Duration (min)
CU050
CU100
CU200
CU300
CU400
(a)
250
Mins=1.12 (I'ins=28.2o)
M=1.54 (I'=37.8o)
200
150
Instability line
100
e=0.78
e=0.82
e=0.85
50
e=0.94
e=1.05
0
0
50
100
150
200
250
300
350
Effective mean normal stress, p' (kPa)
400
450
300
CU050
CU100
CU200
CU300
CU400
(b)
250
sion tests, the soil specimens were consolidated isotropically to different initial mean effective stresses
before shearing. Figure 8a shows the effective stress
paths of five isotropically consolidated undrained
compression tests with the initial pc ranging from 50
kPa to 400 kPa (corresponding to void ratios varying
from 1.05 to 0.78). The effective stress path of each
loosely compacted specimen is characterized by its
initial increasing q with decreasing pc, due to an increase in pore water pressure during undrained
shearing resulting from the contractive tendency of
the soil. After a peak is reached, q decreases with a
further reduction in pc until the critical state
(M=1.54, I o) is reached, illustrating the unstable nature of the specimen. By joining the stress origin and the peak of each stress path, an instability
line (Lade 1992) can be identified with its slope
equal to 1.12, corresponding to Iins=28.2o. Strainsoftening behaviour with very small liquefaction potential but without any phase transformation phenomenon was noted in these tests (see Fig. 8b).
During the cyclic tests (Ng et al. 2004b), a cyclic
deviator stress of equal magnitude in compression
and extension was applied to the specimens. Figure
8c shows a typical result of CDG (e=0.821) from a
cyclic triaxial test with a cyclic stress ratio (CSR) of
0.1, where CSR is defined as the single amplitude
cyclic shear stress (d) divided by twice the initial effective confining pressure (
3), i.e. CSR=d
3).
In the test, pc decreased monotonically but the rate of
the pore water pressure build-up decreased as the
number of cycles increased, due to the relatively low
CSR. Eventually, the pore water pressure ceased to
develop further as the contractive and dilative tendency of the soil specimen balanced out. The total
deviator strain developed was less than 0.2% at the
end of the test. On the other hand, for a cyclic test on
CDG with CSR=0.15 (e=0.821) as shown in Figure
8d, the pore water pressure accumulated continuously and resulted in a continuous decrease in pc, illustrating a typical cyclic mobility phenomenon
(Castro 1969).
e=0.78, Vc=400kPa
200
e=0.82, Vc=300kPa
150
e=0.85, Vc=200kPa
100
e=0.94, Vc=100kPa
50
e=1.05, Vc=50kPa
0
0
10
15
20
Axial strain, Ha (%)
25
30
(c)
(d)
Figure 10. The observed changes of soil structure of the crest region due to rainfall infiltration (Take et al. 2004).
18.900
Prototype Scale
e
0.600
Unit in metre
Scale
6.240
24.240
LVDT-v3
LVDT-v2
LVDT-v1
LVDT-h1
LS-h1
660
30
ACC-T-X,Y,Z
150
X
PPT4
ACC4-Y
ACC4-X
ACC3-Y
PPT3 ACC5-Y ACC3-X ACC2-Y
PPT2
ACC2-X
ACC5-X
ACC1-X
ACC1-Y
PPT1
712
Figure 12. Configuration of the model slope and instrumentation (Ng et al. 2004b).
1.0
1.0
PPT1
PPT2 (Z=100mm)
PPT2
0.5
'u / Vcv
PPT1 (Z=145mm)
0.0
0.5
PPT4
-0.5
0.0
0.2
0.4
0.6
Time (s)
0.8
1.0
0.0
PPT4 (Z=10mm)
Figure 16. General view of the slope (from Tang & Lee 2003).
-0.5
0
10
15
Time (s)
Figure 14. Measured excess pore-water pressure ratios in biaxial shaking test M2D-0.3 (Ng et al. 2004b).
Laser sensor
Laser sensor
Laser sensor
Laser sensor
Figure 17. General view of slope after failure (from Tang &
Lee 2003)
Coarse soil
PPT1 PPT2
PPT3
Loose CDG (WTS)
Downstream drainage
board
LS3
PPT4
Downstream temporary
reservior
reservoir
LS2
Upstream temporary
reservior
reservoir
PPT5
LS1
Wood block
Coarse soil
PPT6
Inlet hole
PPT7
PPT B
Outlet hole
PPT8
PPT9 PPT C
Figure 18. Comparisons of measured soil displacements without (CG45) and with soil nails (CGN45) in two centrifuge tests
using CDG loose fill at 60 g (dimensions in metres at prototype
scale) (Ng et al. 2002b).
Figures 20 and 21 show the occurrence of a nonliquefied slide and the measured excessive pore water pressure during two failures, respectively. The
slide was initiated near the crest. Based on the observed failure mechanisms and the small excessive
pore water pressures measured, it was concluded that
non-liquefied slide of loose shallow CDG fills slopes
could occur but static liquefaction was very unlikely
to happen in the slopes.
6.2 Destabilisation of loose shallow CDG fill slope
at the toe in centrifuge (Take et al. 2004)
Take et al. (2004) also carried out centrifuge model
tests to investigate the possible slide-flow failure
mechanism of a loose thin CDG fill layer. The CDG
used was taken from Beacon Hill. Figure 22 shows
the geometry adopted. The slope angle was 33o. At
30 g, the model corresponded to a fill slope of 9 m in
height, with a vertical depth of fill of 3 m. The chosen soil profile for the model fill also represents an
idealized case of layering in which the CDG fill material has been sieved and separated into its coarse
of the fill slope. As intended, the rate of water transfer into the toe region exceeded the seepage velocity
through the model fill, causing a transient increase in
the pore water pressure at the toe. The local pore water pressure was observed to increase at a nearly constant rate reaching a maximum value of 16 kPa at
point B in Figure 23a. As this seepage front progressed towards the toe, the slope was slowly creeping (Fig. 23b).
After time B, the slope mass is observed to accelerate (points B-C on Figure 23b). By analysing images captured by PIV (White et al 2003) at the onset
of more rapid failure, it is found that the toe accelerated horizontally with an average velocity of approximately 6 mm/s (Fig. 24). The observed displacement field over this time interval indicates that
the surface of the model fill moved down-slope at a
slower velocity. When the fill material finally came
to rest, it formed a low-angle run-out. This failure
mechanism differs from that of the slope destabilised
by downward seepage in the test for the Housing
Department in which the slope was not blinded hydraulically at the toe (see Fig. 19). The initiation of
the non-liquefied slides differed in these two slopes.
(a)
(b)
(c)
(d)
7 CONCLUSIONS
Both static and dynamic model tests on LB and CDG
were carried out. In-flight rainfall infiltration, rising
ground water and dynamic loadings were simulated.
Based on the tests, it can be concluded that static
liquefaction/fluidization of the loose LB sand fill
slope due to a rising ground water table was successfully created in the centrifuge. The occurrence of
liquefaction in sand was observed by in-flight video
cameras and verified by the significant and sudden
build-up of excessive positive pore water pressures
measured at various locations in the slope. It is
found that strain softening of the material is a necessary but not a sufficient condition to cause flow liquefaction. A trigger such as seepage force or additional loading is needed.
No liquefied flow and slide was observed in thick
loose CDG fill slopes when they were subjected to
rising ground water tables, heavy rainfall infiltration
and very strong bi-axial shaking. Only excessive soil
settlements were induced. Consistency was found
between centrifuge model tests and full-scale field
trial of a loose CDG fill slope. The significant difference between the observed physical test results on
the LB sand and CDG models may be attributed to
the difference in the fine contents, gradation and liquefaction potential of the two materials.
Ng, C.W.W., Van Laak, P., Tang, W.H., Li, X.S. & Zhang,
L.M. 2001. The Hong Kong Geotechnical Centrifuge. Proc.
3rd Int. Conf. Soft Soil Engineering, Hong Kong: 225-230.
Ng, C.W.W., Van Laak, P.A., Zhang, L.M., Tang, W.H., Zong,
G.H., Wang, Z.L., Xu, G.M. & Liu, S.H. 2002a. Development of a four-axis robotic manipulator for centrifuge modeling at HKUST. Proc. Int. Conf. on Physical Modelling in
Geotechnics, St. John's, Canada: 71-76.
Ng, C.W.W., Zhang, M. & Shi, X.G. 2002b. Keynote (In Chinese): An investigation into the use of soil nails in loose fill
slopes. Proc. of the 1st Chinese Symposium on Geoenvironment and Geosynthetics, Hangzhou, China: 61-80.
Ng, C.W.W., Kusakabe, O. & Leung, C.F. 2003. Theme lecture: Applications of centrifuge modelling technology in
geotechnical engineering practice. Proc. of 12th Asian Regional Conference on Soil Mechanics and Geotechnical
Engineering, August, Singapore, Vol. 2: 1277-1285.
Ng, C.W.W., Fung, W.T., Cheuk, C.Y. & Zhang, L.M. 2004a.
Influence of stress ratio and stress path on behaviour of
loose decomposed granite. J. Geotech. and Geoenviron.
Eng., ASCE 130(1): 36-44.
Ng, C.W.W., Li, X.S., Van Laak, P.A. & Hou, Y.J. 2004b.
Centrifuge modelling of loose fill embankment subjected to
uni-axial and bi-axial earthquake. Soil dynamics and earthquake Engineering 24(4): 305-318.
Ng, C.W.W, Zhang, L.M. & Wang, Y.H. 2006a. Proceedings
of 6th Int. Conf. on Physical Modelling in Geotechnics. Volumes 1 and 2. Publisher: Taylor & Francis. ISBN: 978-0415-41587-3 and 978-0-415-41588-0.
Ng, C.W.W., Zhang, E.M. & Zhou, R.Z.B. 2006b. Centrifuge
modelling of use of soil nails in loose and dense fill slopes.
GEO Report, CEDD of HKSAR.
Ng, C.W.W., Pun, W.K., Kwok, S.S.K., Cheuk, C.Y. & Lee,
D.M. 2007. Centrifuge modelling in engineering practice in
Hong Kong. 3URF RI WKH *HRWHFKQLFDO 'LYLVLRQV $QQXDO
Seminar, Hong Kong Institution of Engineers, 55-68.
Ng, C.W.W., Zhang, M., Pun, W.K., Shiu, Y.K. & Chang, G.
W.K. 2009. Investigation of static liquefaction mechanisms
in loose sand fill slopes. Submitted to CGJ.
6ODGHQ-$'+ROODQGHU 5' .UDKQ - The liquefaction of sands, a collapse surface approach. Canadian
Geotechnical Journal 22: 564-578.
Take, W.A., Bolton, M.D., Wong, P.C.P. & Yeung, F.J. 2004.
Evaluation of landslide triggering mechanisms in model fill
slopes. Landslides, Japan, Vol. 1: 173-184.
Tang, W.H. & Lee, C.F. 2003. Potential use of soil nails in
loose fill slope: an overview. Proceedings of the International Conference on Slope Engineering, Hong Kong, 974997.
White, D.J., Take, W.A. & Bolton, M. D. 2003. Soil deformation measurement using particle image velocimetry (PIV)
and photogrammetry. Gotechnique 53(7): 619-631
Zhang, M. 2006. Centrifuge modelling of potentially liquefiable loose fill slopes with and without soil nails. PhD Thesis, the Hong Kong University of Science and Technology.
Zhang, M. & Ng, C.W.W. 2003. Interim Factual Testing Report I SG30 & SR30. Hong Kong University of Science &
Technology.
Zhang, M., Ng, C.W.W., Take, W.A., Pun, W.K., Shiu, Y.K. &
Chang, G.W.K. 2006. The role and mechanism of soil nails
in liquefied loose sand fill slopes. Proc. of 6th Int. Conf.
Physical Modelling in Geotechnics, Hong Kong. Vol. 1:
391-396.