Вы находитесь на странице: 1из 10

WATER FILM IN LIQUEFIED SAND AND ITS EFFECT

ON LATERAL SPREAD
By Takeji Kokusho1

ABSTRACT: A 1D saturated sand layer of 2 m in thickness, in which a silt seam is sandwiched, is liquefied
by an instant shock. It is found that a water film is easily formed beneath the silt seam with a thickness as thin
as a few millimeters just after liquefaction in loose sand and that the film lasts longer than the post-liquefaction
settlement. The effect of the water film on pore-pressure distribution and sand settlement is intensively studied.
1g shake table tests are then carried out for 2D models with or without seams of silt within a saturated sand
layer. In the former case, water films formed beneath silt seams just after liquefaction enable the soil mass above
them to glide due to an unbalanced force along the water films, not only during but also after shaking. In the
latter case, the soil deforms continuously, mostly during shaking, and stops afterward. Thus, a significant effect
of water films formed beneath thin, low-permeability sublayers in a liquefied loose sand, on the failure mode
and timing in lateral spread, is clearly demonstrated by these simple model tests.

INTRODUCTION portant role in a slope failure or lateral spread failure, only a


small number of research projects have focused on water films
Structural damages due to lateral spread of a liquefied sand as the mechanism of the lateral spread failure. Seed (1987)
layer have increased in number during recent earthquakes in indicated a possibility of emergence of a water interlayer be-
Japan, such as those in Niigata in 1964, Nihonkai-Chubu in neath an impervious thin layer when he discussed residual
1983, and Hyogoken-Nanbu in 1995. Seed (1987) collected a strength in slope failure. Because of this possibility, the same
worldwide case histories of slope failures for embankment author pointed out the difficulty in evaluating a reliable resid-
dams, natural slopes, and so on, caused by soil liquefaction ual shear strength in liquefied sand by a laboratory soil ele-
during recent earthquakes, in order to establish an empirical ment test. Fiegel and Kutter (1994) performed a centrifuge
evaluation method for determining residual strength of lique- shake table test to demonstrate formation of water films in
fied sand layers. In some of these events, the lateral spread or layered sand. Dobry, Tabaoada, and Liu (1995) tried to explain
slope failure was witnessed not only during but also after the cause of the lateral spread at the Pence Ranch site during
earthquake shaking. For example, during the Niigata earth- the 1983 Borah Peak earthquake by the effect of a water in-
quake (Kawakami and Asada 1966; Hamada 1992) in Hakusan terlayer just beneath an impervious surface layer. Except for
District in the Niigata city, a wide area of 250 by 150 m started these examples, a majority of research on the lateral spread
to move toward the Shinano River during the shaking and mechanism has concentrated on a homogeneous sand layer
continued to move after the shaking ended. In the earthquake rather than a nonuniform layer consisting of horizontal sub-
at the Lower San Fernando Dam near Los Angeles, slope fail- layers. Shake table tests under 1g or in a centrifuge apparatus
ure took place 1 min after shaking stopped (Seed 1987). More in such research all indicate that the soil movement can occur
recently, in an earthquake in New Zealand, lateral spread of only during shaking and stops completely after the end of
the foundation ground put a bridge out of service about one shaking.
hour after the earthquake shaking ceased (Berrill et al. 1997). In the first part of this research, a 1D model of saturated
These facts imply that a lateral spread or slope failure in a loose sand with a seam of silt sandwiched in it was liquefied
liquefied ground may not necessarily be caused by the inertia to investigate in detail the mechanism of water film generation
force of an earthquake, but by the gravity force sustained af- and related postliquefaction behavior. Based on a high possi-
ter it. bility of emergence of water films in liquefied loose sand, 1g
A close look at a natural sand deposit readily reveals that it shake table tests were then carried out to understand the mech-
normally consists of a set of sublayers with different soil par- anism of lateral spread in a heterogeneous sand layer where
ticles, including both silty and coarse sands. For example, bor- water films tend to be formed beneath less permeable sublay-
ing logs obtained in liquefied ground during the Niigata earth- ers. A 2D saturated loose sand layer with a sloping surface or
quake indicated the soil consists of various sublayers of silty under an embankment was liquefied to study the effect of wa-
to gravelly sands [e.g., Kawakami and Asada (1966)]. In some ter films formed beneath seams of silt introduced in sand by
cases, a sand deposit may sandwich silt or clay layers of a comparing the results with those for homogeneous sand.
thickness too thin to be identified as independent soil layers
by a normal geotechnical survey. If such a sand layer liquefies, MODEL TEST FOR WATER FILM DEVELOPMENT
an upward flow of excess pore water squeezed from the re- AND DECAY
consolidated sand rushes toward the ground surface. On its
way, it is highly probable that the water flow is trapped by A saturated loose sand layer was made by pouring dry Toy-
relatively impermeable sublayers and forms a water film be- oura sand into water in a lucite tube of 13 cm in inner diameter
neath it. and 211.5 cm in height, as illustrated in Fig. 1. The sand layer
Despite a possible involvement of water films and their im- of 200 cm in depth sandwiches a seam of nonplastic silt in
the middle (at X = 96 cm, where X is the vertical distance
1
Prof., Civ. Engrg. Dept., Facu. of Sci. and Engrg., Chuo Univ., 1-13- from the bottom) to separate the upper and lower sand layers.
27 Kasuga, Bunkyo-ku, Tokyo, 112 Japan. In most of the test series, only 50 g of silt were used to make
Note. Discussion open until March 1, 2000. To extend the closing date a silt seam about 4 mm thick. The permeability constant for
one month, a written request must be filed with the ASCE Manager of this silt seam was estimated as 1 ⫻ 10⫺4 cm/s from the test
Journals. The manuscript for this paper was submitted for review and
possible publication on May 5, 1998. This paper is part of the Journal
measurement, whereas that for the sand was measured as on
of Geotechnical and Geoenvironmental Engineering, Vol. 125, No. 10, the order of 10⫺2. This 1D sand layer was instantaneously
October, 1999. 䉷ASCE, ISSN 1090-0241/99/0010-0817–0826/$8.00 ⫹ liquefied by a shock given by a steel hammer powered by a
$.50 per page. Paper No. 18279. controlled spring force. The typical acceleration time history
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999 / 817
recorded by an oscilloscope is shown in Fig. 2, in which the 1996) or in the Port Island during the 1995 Hyogoken Nanbu
maximum acceleration is about 270g. This acceleration value earthquake.
may seem extraordinarily large, but the shock only causes a Five electric pore-pressure gages were installed at five levels
maximum liquefaction settlement of 3% of the total soil thick- of the tube, as shown in Fig. 1, to measure excess pore-pres-
ness, which is actually about the same settlement observed in sure during and after liquefaction. The settlement of the sand
the Niigata city during the 1964 Niigata earthquake (Ishihara and thickness of the water film were measured by two digital
video cameras to obtain time-dependent records. The relative
density of the sand was Dr = 39 and 14% in the lower and
upper sand layer, respectively. To compare the results, a sand
layer of exactly the same size with almost the same density
without the silt seam was also tested. Grain size curves of the
sand and silt employed in the test are shown in Fig. 3. The
maximum and minimum density of the sand were evaluated
by the standard test method by the Japanese Geotechnical So-
ciety.
In Fig. 4(a), the settlements at the top of the upper and lower
layers (at X = 200 cm and X = 96 cm, respectively) sand-
wiching the silt are plotted against the elapsed time, starting
at the moment of shock. The same figure also plots the vari-
ation in thickness of the water film beneath the silt seam. It is
readily understood that the water film starts to build up just
after the onset of liquefaction. This indicates that the undrained
condition does not hold, but a migration of pore-water starts
to occur locally in a very short time period. As shown in the
photograph in Fig. 5, which was taken 35 s after the shock,

FIG. 1. 1D Saturated Loose Sand Model Liquefied by Hammer

FIG. 2. Typical Acceleration Time History of Shock Given by


Hammer

FIG. 4. Time-Dependent Variation in Water Film Thickness


and Soil Settlement in Sand Layer (a) with Seam of Silt; (b) with-
FIG. 3. Grain Size Curves for Soil Materials Used in Research out Seam of Silt

818 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999


the water film consists solely of clear water, about 8 mm thick settlement in the upper layer continues at a slower rate up to
at the maximum in this case. In Fig. 4(b), the settlements at point P5. The water film beneath the silt seam grows to its
the same level of the sand layer without a silt seam are plotted maximum thickness at P2 with almost the same speed as the
for comparison. In these figures, starts, ends, and turning settlement of the lower layer. This indicates that almost all of
points of the curves are benchmarked as P1 through P5. It is the excess pore water coming out of the saturated sand beneath
evident that P2 and P3 correspond to the completion of the the silt seam forms the water film. After that, the water film
post-liquefaction settlement in the lower and upper layers, re- decreases its thickness due to the seepage flow through the silt
spectively. The trend is not so different up to P3 in both cases. and the upper sand layer until it vanishes at point P4.
The clear difference is that, in the case with the silt seam, the Figs. 6(a) and (b) illustrate time-dependent changes in ex-
cess pore pressure for the cases with and without the silt seam,
respectively. The dots of P1 through P5 in the figures indicate
the same benchmarks as those in Fig. 4. In the case without
the silt seam, the pore pressure starts to decrease from the
bottom of the layer and completely dissipates by point P3. In
contrast, if there is a silt seam, the pore-pressure dissipation
occurs concurrently, both in the upper and lower layers, as
indicated by the pore-pressure data (obtained from piezometers
c and e in Fig. 6(a). The pressure continues to decrease until
the end of settlement of liquefied sand in the upper layer, P3,
and after that maintains an almost constant value up to the
point P4. During that period, between P3 and P4, note that
piezometers d and e, which are located below the silt seam,
measure almost the same value, while upper piezometers a, b,
and c indicate values almost proportional to their depth from
below the surface. This implies that the stationary seepage
flow continues in this period, from the water film to the soil
surface through the silt seam and the upper sand layer. After
P4, the remainder in excess pressure completely dissipates
from the lower layer and finally returns to the initial effective
FIG. 5. Photograph of Water Film Consisting of Clear Water stress condition near P5.
Formed beneath Silt Seam Based on the time-dependent pressure change, pressure dis-

FIG. 6. Time-Dependent Variation in Excess Pore-Pressure at Five Levels of Sand Layer (a) with Seam of Silt; (b) without Seam of Silt

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999 / 819


tributions along the depth are plotted in Figs. 7(a) and (b). In from the test results as indicated below if there were no fric-
the case without a seam of silt, the hydraulic gradient may be tion, as in the field condition.
approximated as zero in a section where liquefaction has al- In Fig. 8(a), the pressure distribution along the depth for
ready settled. However, if there is a silt seam, the pressure in the model with the skin friction is shown schematically, while
the upper layer are evidently varied with depth, indicating in Fig. 8(b), a similar chart is drawn for an ideal field condition
clear influence of upward flow from the water film, as men- without skin friction. These figures use the same benchmarks
tioned before. Note also that, by extrapolating the pressure as do the previous figures. The difference in the field is that,
gradient from the upper and lower layers, a distinctive gap in once the post-liquefaction settlement is completed in the lower
pressure exists at the silt seam, introducing a high hydraulic layer, the excess pressure there is expected to stay at the value
gradient in it. of P2. In the meantime, the excess pressure in the upper layer
It can readily be understood that, if a water film forms in a decreases with postliquefaction pressure dissipation, resulting
soil layer continuously in a horizontal direction, the excess in an excessive hydraulic gradient applied to the seam of silt.
pore pressure in the soil layer below should be equal to the This may lead to a breakout of the silt seam at P2⬘, and at
initial effective stress applied by the soil above the water film. that instant, a part of the sand layer is again exposed to a
However, the measurements in piezometers d and e in the time hydraulic gradient higher than the critical gradient, which may
interval between P3 and P4 are actually much lower than this again cause local liquefaction. The timing of the breakout will
value, as indicated in Fig. 7(a). This can be attributable to the depend not only on the thickness but also the plasticity of the
skin friction between the lucite tube and the sand. It may not silt. For a nonplastic silt, the breakout will take place at a
be so easy to conduct this type of model test, in which the lower hydraulic gradient, while it may not occur for a high-
effect of the friction of a soil container can be effectively can- plasticity clayey silt. The breakout event will decrease the
celed. However, it may be possible to estimate the difference pressure in the lower layer down to the level of P2⬙ and make
the decay time of the water film shorter than in this model
test. Despite such quantitative differences due to the existence
of skin friction mentioned here, the fundamental behavior of
the sand layer observed in the model test will also be appli-
cable to the field condition.
Fundamental mechanisms for a water film to emerge and
decay have thus been clarified. It is of interest to investigate
the effect of basic parameters on the dimensions and time du-
ration of the water film. For this purpose, a series of tests have
been carried out under the same conditions as shown in Fig.
1, but with different density in the lower sand layer. The total

FIG. 8. Schematic Chart on Time-Dependent Pore-Pressure


FIG. 7. Time-Dependent Variation in Pore-Pressure Distribu- Distribution in Sand Layer with Silt Seam (a) with Skin Friction
tion along Depth (a) with Seam of Silt; (b) without Seam of Silt in Laboratory Test; (b) without Skin Friction as in Field

820 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999


weight of the silt is kept constant at 50g, making the silt seam stances, multiple water films will show up after complete liq-
in these tests about 4 mm thick and the relative density of the uefaction of a loose sand deposit. Since these water films will
upper layer 13 to 15%. Fig. 9(a) indicates the effect of the survive for a while after liquefaction ceases, despite their
relative density of the sand layer below the silt seam on the breakage due to excessive hydraulic gradient, the excess pore
maximum thickness of the water film. Fig. 9(b) shows a sim- pressure, as shown in the figure, will remain. The pressure at
ilar relationship between the relative density and time duration each layer will be slightly inclined, reflecting the hydraulic
of the water film, which means from points P1 to P4 in Fig. gradient of the upward seepage flow from each water film.
4. Both charts indicate that the thickness and duration of the Such considerations reveal that a loose sand layer tends to
water film tend to decrease remarkably with the increase in remain unstable for some time, even after the initial settlement
relative density, implying that the influence of water films be- of suspended sand particles in each sublayer.
comes increasingly dominant for looser sand layers. Figs.
10(a) and (b) show the effect of the thickness of the silt seam
on the maximum thickness and duration of the water film,
respectively, under the similar test conditions shown in Fig. 1.
The weight of the silt is taken as 25, 50, and 100g, to make
about 2, 4, and 8 mm thick silt seams, respectively, and the
silt is sandwiched at either X = 96 or 140 cm. The relative
densities of the lower and upper sand layers are 39 to 43%
and 11 to 17%, respectively. Clearly, even with a thickness of
about 2 mm, the silt seam has a capacity to form a water film
for a short time, and the duration and maximum thickness of
the water film increase with the thickness of the silt. Although
these are very qualitative results under a simple test condition
inflicted with the skin friction, the dominant effects of the sand
density and a minimal thickness of the silt are clearly dem-
onstrated.
The model above considers only a single seam of silt. Mul-
tiple seams of silty sand may exist in actual field conditions,
as schematically illustrated in Fig. 11. Under such circum-

FIG. 10. Effect of Thickness of Silt Seam on (a) Maximum


Thickness and (b) Time Duration of Water Film

FIG. 9. Effect of Sand Density in Lower Layer on (a) Maximum FIG. 11. Schematic Chart on Postliquefaction Excess Pore-
Thickness; and (b) Time Duration of Water Film Pressure Distribution Considering Water Film Effect

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999 / 821


MODEL TEST ON LATERAL SPREAD FAILURE WITH
WATER FILM
Sloping Ground with Arc of Silt
The previous section shows that water films are easily
formed beneath relatively impermeable seams of silt as thin
as a few millimeters in relatively loose sand. The maximum
thickness of the water film is almost equal to the liquefaction
settlement in each liquefied sublayer beneath impervious silty
seams. More than a few of these relatively impervious sublay-
ers may not so infrequently be sandwiched in loose sand. Con-
sidering that some of the water films formed beneath these
sublayers may be continuous, they will surely have a great
influence on the failure mode, safety factor, and timing of fail-
ure if a sand layer has a potential for lateral spread or slope
failure. In order to study the role of water films in sliding
failure, 2D model tests of saturated loose sand have been car-
ried out.
A rectangular lucite soil box of 500 mm in height, 800 mm
in width, and 400 mm in thickness is filled with fine sand to
make a saturated slope containing an arc-shaped seam of silt,
as shown in Fig. 12. The grain size curves for the fine sand
and the silt are also available in Fig. 3. The model on a shake
table is liquefied by a kind of repetitive motion by a pneumatic
actuator. The direction of shaking is perpendicular to the slop-
ing direction, so that the effect of the inertia force can be
excluded from the lateral spread deformation. Two tests cases,
case 1 and case 2, are carried out, respectively, with and with-
out the arc of silt for a saturated sand slope of about 30%.
The relative densities for case 1 and case 2 are 21 and 19%,
respectively. The model is vibrated by the input motion of
three cycles in 3 Hz with an acceleration of 0.3g. The thick-
ness of the silt seam is about 6 mm on average, although it
varies from 4 to 9 mm from part to part. Failure modes in
these tests are visualized by movements of marker grids made
from Japanese noodles attached to the transparent wall of the
box.
In Figs. 13(a) and (b), the movements of the slope during
and after shaking are illustrated in two separate charts for case
1 and case 2, respectively. In case 2, where no silt arc exists,
the slope deforms continuously without any discontinuous slip
surface, and major movement stops as soon as the shaking
stops. In case 1, with the silt arc, a discontinuity in flow de-
formation at the arc becomes dominant, and the slope fails,
not only during but also after shaking. Figs. 14(a) and (b)
show relationships between the total deformation (the sum of
the length of deformation increment vector for each time in-
crement) at representative points of the slope shown in Fig.
14(c) and the elapsed time, in case 1 and case 2, respectively.
It is obvious from this figure that the slope with the silt arc

FIG. 13. Deformation of Sloping Sand during Shaking (Top)


FIG. 12. 2D Saturated Sand Slope with Silt Arc in Lucite Box and after Shaking (Bottom): (a) Case 1, with Silt Arc; (b) Case 2,
on Shake Table without Silt Arc
822 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999
continues to deform after the shaking end, whereas the slope
without it deforms mostly during shaking. As soon as the shak-
ing starts, a water film as thin as a hair appears in the lower
part of the silt seam, as shown in Fig. 15. The slide obviously
takes place along the water film, while it is sustained beneath
the arc.

Horizontal Ground with Horizontal Seams of Silt


Loaded by Embankment
Natural sandy ground is normally layered with relatively
impervious horizontal sublayers. Under these more realistic
circumstances, the water film’s role in sliding failure may not
be so simple as indicated in the test described above. Conse-
quently, a model test for horizontal ground with horizontal silt
seams is carried out for case 3 and case 4 in the same lucite
box used in the sloping ground model, and partially loaded
with an embankment made from coarse sand, as shown in Fig.
16. The sand and silt used here are exactly the same as in the
previous shake table test, and the mean grain size of the coarse
sand is 0.6 mm. Two seams of silt about 4 mm thick on av-
erage are introduced at different depths in a fine sand layer in
case 3, in contrast to no silt in case 4. The relative densities
for case 3 and case 4 are 14 and 19%, respectively. Four elec-
tric piezometers are installed at the locations shown in the
figure, as well as an accelerometer on the shake table. The
shake table is given the same input motion mentioned in the
previous test.
Figs. 17(a) and (b) illustrate the deformations in the two
cases during and after shaking for case 3 and case 4, respec-
tively. Note that only a continuous deformation takes place in
the sand layer without silt seams, while discontinuous defor-
mations occur after the shaking ends in the sand layer with
silt seams. This indicates that soil movement is still possible
even after shaking along the water films formed beneath the
silt seams. Figs. 18(a) and (b) show the changes in deforma-
tions at some representative points in the cross section shown
in Fig. 18(c) with elapsed time for the two cases with and
without seams, respectively. The postshaking deformation for
case 3 is obviously larger than a minimal deformation in case
4. Figs. 19(a) and (b) show the excess pore-pressure measure-
ments at the four locations in the sand layer for case 3 and
case 4, respectively. The dashed lines in the figure indicate the
theoretical value corresponding to a complete liquefaction
computed from the sand density. The pressure remains at its
FIG. 14. Soil Deformation versus Elapsed Time Relationship maximum in a longer period and after that rapidly decays to
for Representative Points in Sloping Sand: (a) Case 1, with Silt zero in case 4, whereas it takes longer for the pressure to
Arc; (b) Case 2, without Silt Arc; (c) Location of Representative dissipate in case 3. Note that the postshaking deformation
Points
shown in Fig. 18 takes place during the period of maximum

FIG. 15. Photograph of Hair-Like Water Film in Lower Part of FIG. 16. 2D Saturated Sand Layer with Horizontal Seams of
Arc-Shape Silt Seam Silt Loaded by Embankment in Lucite Box on Shake Table

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999 / 823


FIG. 18. Soil Deformation versus Elapsed Time Relationship
for Representative Points in Sand Layer: (a) Case 3, with Silt
Seams; (b) Case 4, without Silt Seams; (c) Location of Repre-
sentative Points

pore pressure. The emergence of water films beneath the silt


seams is also witnessed in this period, although they are some-
times ruptured at breakout points and rapidly fade away.
The test results thus shown are certainly of a qualitative
nature and cannot directly give quantitative answers to ques-
tions of the safety factor, or magnitude of lateral spread in the
field. Several undesirable factors actually influence the results,
such as the restraints of the model boundary on both sides of
the sand layer and the skin friction of the soil box. Despite
these limitations, the results give a valuable insight into how
the water film, once formed, will affect the way a sand layer
deforms due to liquefaction.

CONCLUSIONS
Two different types of model tests of a qualitative nature
FIG. 17. Deformation of Horizontal Sand Layer during Shaking
(Top) and after Shaking (Bottom): (a) Case 3, with Silt Seams; (b)
have been carried out in this research to investigate the effect
Case 4, without Silt Seams of water films formed beneath relatively impervious sublayers
824 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999
FIG. 19. Time-Histories of Shake-Table Acceleration and Pore Pressure at Four Points in Sand Layer: (a) Case 3, with Silt Seams; (b)
Case 4, without Silt Seams

in a liquefied sand layer: a 1D saturated loose sand layer model 6. In a horizontally layered deposit, water films are formed
and a 2D model. In the former, the process of growth and horizontally and can be part of a potential slip surface
decay of the water film has been studied under simplified con- for flow failure. In this case, too, the deformation of sand
ditions. In the latter, the effect of a water film on lateral spread will take place even after the shaking ends, by exploiting
has been investigated by two soil models: a sloping ground some parts of water films.
and a level ground with a partial embankment. These model 7. Based on the above-mentioned experimental findings, it
tests have yielded the following major findings. is highly probable that the water file effect has a signif-
icant role in lateral spread or slope failure in a liquefied
1. A water film can be easily formed just after the complete loose sand layer in the field. Because water films are
liquefaction of a sand layer beneath a seam of silt as thin easily formed continuously beneath less permeable sub-
as a few millimeters. This indicates a high possibility that layers and no shear resistance is exerted along them, they
water films will actually be formed under low-permea- are surely chosen as part of a sliding surface for lateral
bility sublayers in the field due to complete liquefaction spreading failure. The failure may occur not only during
in loose sand. but also after earthquake shaking, whenever the safety
2. The maximum thickness and duration of a water film factor of the potential sliding surface exploiting water
may be approximated as inversely proportional to the films falls below unity.
sand density, indicating that the effect of a water film
will be more pronounced for looser soils.
3. A water film tends to outlast liquefaction settlement in Among differences between these qualitative model tests
each sublayer and thus may affect soil stability longer and a prototype, the significant difference in magnitude in the
than the apparent time period of liquefaction. effective overburden stress would be most controversial. In
4. If a water film is formed in a sand layer continuously general, however, the volume change characteristics in sheared
along a potential slip surface due to a low-permeability sand, which is the key mechanism controlling lateral spreading
sublayer, upper soil will deform discontinuously along and the generation of water films, tends to be more dilatant in
the water film. On the other hand, a uniform sand layer a scaled model than in a prototype because of the lower stress
without silty sublayers tends to deform quite continu- level in the model. Therefore, lateral spreading that exploits
ously without any distinct slip surface. water films, which has been actually observed in the small
5. If the water film is sustained, it is possible for a soil model under ultralow overburden stress, will be more pro-
mass to be driven only by the force of gravity, even after nounced in the field in the postliquefaction phase of loose sand
the shaking ends, as is sometimes observed in the field. deposits.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999 / 825
ACKNOWLEDGMENTS lifeline performance during past earthquakes. Vol. 1, Japanese Case
Studies, Nat. Ctr. for Earthquake Engrg. Res., State Univ. of New York,
The writer would like to show his sincere gratitude to K. Watanabe, Buffalo, N.Y., 3.1–3.123.
T., Sawano, T. Kojima, T. Nakano, and N. Nonaka, a research associate Ishihara, K. (1996). ‘‘Soil behaviour in earthquake geotechnics.’’ Settle-
and graduate students in the Civil Engineering Department at Chuo Uni- ment in sand deposits following liquefaction, Clarendon, Oxford, U.K.,
versity, for their hard work in implementing the laboratory tests. 308–315.
Kawakami, F., and Asada, A. (1966). ‘‘Damage to the ground and earth-
APPENDIX. REFERENCES structures by the Niigata earthquake of June 16, 1964.’’ Soils and
Foundations, Tokyo, VI(1), 14–30.
Berrill, J. B., Christensen, R. J., Keenan, R. J., Okada, W., and Pettinga, Kokusho, T., Kojima, T., Nakano, T., Nonaka, N., and Watanabe, K.
J. R. (1997). ‘‘Lateral spreading loads on a piled bridge foundation.’’ (1998). ‘‘Model tests on water film effect for lateral flow mechanism
Seismic Behavior of Ground and Geotech. Struct., Proc., Spec. Tech. in liquefied ground.’’ Special Symp. on Lateral Flow Failure in Liq-
Session on Earthquake Geotech. Engrg., Int. Conf. on Soil Mech. and uefied Ground, Japanese Geotechnical Society, Tokyo (in Japanese).
Geotech. Engrg., Balkema, Rotterdam, The Netherlands, 173–183. Kokusho, T., Sawano, T., Kawai, R., and Sugiyama, K. (1998). ‘‘Study
Dobry, R., Tabaoada, V., and Liu, L. (1995). ‘‘Modeling of liquefaction on formation of water film affecting lateral flow of liquefied sand
effects during earthquakes.’’ Proc., 1st Int. Conf. on Earthquake Geo- ground.’’ Spec. Symp. on Lateral Flow Failure in Liquefied Ground,
tech. Engrg., (Int. Symp., Japanese Geotech. Soc., Tokyo), Balkema, Japanese Geotechnical Society, Tokyo (in Japanese).
Rotterdam, The Netherlands, Vol. 2, 1291–1324. Kokusho, T., and Watanabe, K. (1997). ‘‘Water film effect on lateral flow
Fiegel, G. L., and Kutter, B. L. (1994). ‘‘Liquefaction mechanism for in liquefied ground.’’ Earthquake Engrg. Symp., Japan Society for Civil
layered soils.’’ J. Geotech. Engrg., ASCE, 120(4), 737–755. Engineers, Tokyo (in Japanese).
Hamada, M. (1992). ‘‘Large ground deformations and their effects on Seed, H. B. (1987). ‘‘Design problems in soil liquefaction.’’ J. Geotech.
lifelines: 1964 Niigata earthquake.’’ Case studies of liquefaction and Engrg., ASCE, 113(8), 827–845.

826 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / OCTOBER 1999

Вам также может понравиться