See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303921276

Installation of underground columns for mitigation of seismic liquefaction

Conference Paper · November 2012

CITATIONS READS

0 22

4 authors, including:

Naoki Takahashi Rouzbeh Rasouli

Sumitomo Mitsui Construction Co., Ltd., Japan Amec Foster Wheeler, Canada
28 PUBLICATIONS   54 CITATIONS    32 PUBLICATIONS   45 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Rouzbeh Rasouli on 13 June 2016.

The user has requested enhancement of the downloaded file.

 Installation of underground columns for mitigation of seismic
liquefaction

Ikuo Towhataa, Naoki Takahashib, Yoshiyuki Morikawac, Rouzbeh Rasoulia

a
The University of Tokyo, Japan
b
Sumitomo Mitsui Construction Company, Japan
c
Port and Airport Research Institute, Japan
towhata@geot.t.u-tokyo.ac.jp

Abstract: Traditionally, countermeasures against seismic liquefaction aimed to prevent

significant development of excess pore water pressure. Although this aim was achieved by a
variety of measures, the limitation has been understood as well in the recent times. The
limitation is typically found in lifelines and transportation lines (road embankment) together
with river levees for which financial requirement is more strict and also residual deformation
is allowed to occur to a certain extent. In this regard, the present study addresses installation
of stable cement-mixed soil columns in liqeufaction-prone subsoil so that ground deformation
may be constrained during earthquakes and lateral flow of liquefied sand may be reduced. For
its validation, shaking-table model tests have been conducted in both 1-G and centrifugal
environments.

Keywords: Liquefaction, Deformation, Mitigation, Model test

1. Introduction
Deep soil mixing is often employed today for mitigation and prevention of earthquake
damage induced by subsoil liquefaction. In general, the configuration of improvement is of
either block (uniform) or square-grid shapes. The grid type is less costly and, together with
the surface dry soil, confines cyclic shear deforamtion of liquefaction-prone sand so that
exdcess pore pressure therein would be prevented from significant development1).
The present study attempts to further reduce the construction cost by reducing the
amount of soil improvement by installilng grouted soil columns in place of grid walls.
Although columnar improvement has such advantages as reduced raw materials, easier
construction and allowanve for existing underground constructions, it has been thought to be
of limited mitigation of liquefaction damage as compared with those of block or grid shapes2).
This situation is changing because of recent studies3)4) that demonstrated reduced cyclic
shear strain during shaking and hence the residual deformation was made smaller, leading to
mitigation of liquefaction, when columnar improvement was installed. One of the previous
studies by the authors5) compared grid and columnar improvements to show that columnar
improvement is able to reduce the development of excess pore water pressure and to mitigate
liquefaction when the improvement ratio (cross section of columns / entire area to be
improved) exceeds 3.5%, although the grid improvement can achieve similar mitigation under
lower improvement ratio. Further studies by the authors6) demonstrated that a new
construction machine with four soil mixing rods combined into one can construct a bigger
columns and that the increased bending stiffness of the column is able to achieve more
efficiency.
Moreover, the latest studies changed the configuration of columns from the conventional
square or triangular ones to an irregular one in which the geometrical effect reduces the lateral
displacement of subsoil. Shaking table tests on liquefaction-prone backfill of sheet-pile quay
walls indicated that the irregular configuration was able to reduce the lateral displacement of
backfill soils as well as quay walls7)8)9)10). This innovation was applied in the present paper to
lateral flow of a sloaping ground in which lateral flow of liquefied subsoil was expected to be
reduced by irregular configuration of columns. To demonstrate this effect, shaking table tests
were conducted in both gravitational and centrifugal environments.

2. Irregular configuration of underground columns

Figure 1 illustrates three kinds of column configurations that are addressed in this paper. The
irregular configuration in Fig. 1(a) shows that a 2*2 square grid with a spacing of "d" is
shifted either by 2d or d/2 distance in X and Y directions, respectively. This configuration
does not allow free passing of liquefied subsoil through column spaces as compared with the
square and triangular configurations (Figs.1(b) and (c)) where a straight flow passage is
available; see red arrows in these figures. This lack of free and open space in the irregular
configuration is further expected to restrain cyclic shear straining during shaking and reduce
the probability of liquefaction.

X d 2d
Y
d
2d

(a) Irregular geometry (b) Regular geometry (c) Regular geometry

Fig. 1 Geometry of column installation

3. Methodology of shaking model tests

As stated before, the present study carried out both 1-G and centrifugal shaking model tests in
which a sloping liquefiable sandy ground with different column configurations was subjected
to horizontal shaking in the direction of slope. In 1-G tests the vertical columns were placed
in sand and their bottoms were fixed without any translation and rotation allowed. This
installatoin is equivalent to a very stiff bearing layer at the bottom and directly demonstrated
the mitigative effects of configuration on liquefaction. The top of columns was of limited
constraint as will be described later and rotation was possible to a certain extent. In reality, it
is likely that the base soil is not so sitff and minor translation and/or tilting of columns are
possible in the sloping direction. This deformation of column configuration may affect the
mitigative effects and was studied by centrifugal tests. Herein, the bottom tip of columns was
simply embedded in unliquefiable layer where limited translation and rotation were possible,
while the top was either embedded in the surface crust, similar to what was mentioned above,
or totally free without such a crust layer.

4. Preparation of 1-G model tests

The 1-G model tests were conducted at the University of Tokyo by using a soil container that
measured 2,650 mm in length, 390 mm in width, and 600 mm in depth. The scale of
modelling was supposed to be 1/20. As Fig. 2 indicates, the model ground consisted of a base
 Fig. 2 Schematic view of 1-G model ground

Fig. 3 Side view of model ground

unliquefiable layer of 150 mm in thickness, and a sloping and liquefiable sandy layer at the
top. The surface gradinet was set equal to 10%. The entire ground was made of Toyoura sand
(Gs=2.684, emin=0.605, emax=0.974, D50=0.21mm) and was submerged in water.
The base layer was prepared by air pluviation of dry Toyoura sand, followed by
compaction to the relative density of 75 %. This dry layer was then saturated by one-hour
slow percolation of water. The upper liquefiable layer was the prepared by water pluviation to
attain 40% relative density. The height of fall was maintained constant, irrespective of the
ongoing height of sand surface, while the water depth was also controllled to be 20 cm that
was expected to remove pore air from the falling sand and help achieve high degree of
saturation. During shaking, the water surface was set at the same elevation as the top of the
slope (Fig. 2).
The embedded columns were modelled by acryl pipes that measured 26 mm in the outer
diameter and 20 mm in the inner diameter, respectively, implying the equivalent diameter of
520 mm in the prototype. The bottom of the pipes were fixed, as stated above, by screwing
into a PVC (polyvinyl chloride) plate of 20 mm thickness. On the other hand, the top of the
pipes were connected with another PVC plate of 5 mm thickness by two O-rings. Thus,
rotation was easy to occur at the top.
During shaking, acceleration, excess pore water pressure, and lateral displacement of
liqeufied soil were recorded (Fig. 2). The lateral deformation was recorded by photographs
and motion pictures of colored sand in the cross section (Fig. 3) and on the surface. As Fig. 2
illustrates, the central part of the slope model had vertical columns and the time history of
lateral soil displacement was recorded at both upstream and downstream sides ofthe columns
by using embedded inclinometers. Moreover, some of the columns (acryl pipes) were
equipped with strain gages to record bending strain therein.
Horizontal shaking took place in the longitudinal direction of the slope with 10Hz and
 300

200

Acce le r a tio n ( G a l)
10 0

- 10 0

-200

-300
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Time ( se c)
Fig. 4 Base input motion

Table 1 Details of 1G model tests

Test case CASE1 CASE2 CASE3 CASE4 CASE5
Configuration of columns - Irregular Regularly triangular
Improvement ratio (%) 0 25 35 25 35
Maximum acceleration (Gal) 200

Table 2 Similitude law for dyanmic tests

Scaling factors
Unit
(λg)
Length m 1/λ
3
Density t/m 1
Force N 1/λ2
Stress kPa 1
-6
Strain ×10 1
Displacement M 1/λ
Bending rigidity MN・m2 1/λ4
Frequency Hz λ
Acceleration Gal λ
Time（dynamic） sec 1/λ
Time（consolidation） sec 1/λ

200 Gal at the maximum, while the duration time was 6 seconds(Fig.4). More details of tests
are summarized in Table 1 where configuration of columns and improvement ratio are varied
from tests to tests.

5. Preparation of centrifuge model tests

The centrifugal tests were conducted by using the device at the Port and Airport Research
Institute11). A soil container, which measured 708 mm in length, 200 mm in width, and 512
mm in depth, was placed in the device for shaking tests. Table 2 summarizes the law of
similitude concerning the presen study. The size was reduced to 1/50 in the present tests (λ =
50). To achieve the consistent similarity in dynamic and consolidation time scales, pore water
was made 50 times more viscous by using water solution of cellulose ether.
 AC2 AC1

100
45 15

100
AC2

75 Top plate
250
t=5mm

172
50
10% slope AC11
AC16-AC17 AC15
AC5 50 PW9
PW6

400
PW3 AC14

300
AC4 AC12 AC7 AC10
229.2

PW8
PW2 PW5 AC13 Un-liquefiable
AC3 Liquefiable layer AC6 layer AC9
PW1 PW4 PW7
Base plate t5mm

40
40

AC1
708
(mm)
Fig. 5 Schematic view of centrifuge model ground

Fig.6 Side view of model ground for centrifugal test

Figure 5 indicates the tested model. The model was made of Sohma silica sand No. 5 (Gs
= 2.65, emin = 0.71, emax = 1.115). The slope was composed of the unliquefiable base layer of
40 mm in thickness and 90 % of relative density and the upper liquefiable slope was of 10 %
slope and 40 % relative density. The entire model was prepared by air pluviation of sand.
After sand deposition, the viscous liquid was percoalted from the bottom under 30-G
centrifugal environment. The final liquid level was placed at the top of the slope.
The vertical columns of 2-m prototype diameter was reporduced in the model. The
model columns were made of acryl pipes that measured 40 mm in the outer diameter and 28
mm in the inner diameter. The weight of the pipes was adjusted to satisfy the similitude law
by installing steel rods in the inner space and filling the remaining space by silicon. Note that
similitude of the bending stiffness in both the elastic stage and the post-yielding stage wasout
of scope in order to avoid unnecessarily complicated data interpretation. Thus, the present
series of model tests concerns elastic behavior of the columns and the effects of column
configuration was investigated.
The pipes were placed in holes (44mm in diameter) in a thin (5mm thick) acryl templates.
The remaining 2-mm space between the plate and pipes allowed free rotation as well as minor
 Table 3 Specification of shaking
Unit Prototype Model
Centrifugal acceleration g 1 50
Model scale 1/1 1/50
Horizontal excitation Sinusoidal wave, 50 cycles
Duration of shaking sec 25 0.5
Frequency Hz 2 100
Maximum acceleration m/sec2 2（200 Gal） 100

Table 4 Details of tested cases in centrifuge model tests

Test case CASE6 CASE7 CASE8 CASE9
Improvement pattern - Irregular Regular
Improvement ratio (%) 0 35 35 35
Top boundary conditions of columns - Template Free Template
200

200
(mm) (mm)
250 250

(a) Regular saur configuration (b) Irregular configuration

Fig. 7 Configuration of columns in centrifuge model tests

lateral translation of pipes. Note that this space was filled with silicon grease in order to
prevent sand from entering this space. In contrast at the top, another acryl template (270 mm
*198 mm in size) was installed at the elevation of 50 mm lower than the pipe top for
installation of the pipes. Each hole in the plate had three point connections with the pipe to
allow rotation.
Figure 5 indicates the location of employed sensors. Acceleration and excess pore water
pressure were recorded in the model. The piles with hatchied cross section in thie illustration
had strain gauges. Colored Sohma sand made it possible to observe the development of lateral
displacement in both vertical and horizontal planes (Fig. 6). Shaking took place in a
sinusoildal manner of 2 H; see Table 3.
Table 4 summarized the details of tested cases. The improvement ratio was set equal to
35%. Note that configuration of columns, square or irregular, as well as boundary condition at
their tops (entirely free or loose conncetion to an acrylic template) were varied. Fig. 7
illustrates two kinds of column configuration that are namely irregular and regular square.
Note that those columns in the irregular configuration that had to be placed out of the
specified improvement area (indicated by dashed lines) were not installed to avoid
complication. Accordingly, the same number of columns (12) were installed in both kinds of
configurations.
 Ex ce ss p o r e wa te r p r e ss u r e r a tio

Exce s s p o r e wa te r p r e ssu r e r a tio

Exce ss p o r e wa te r p re ssu r e ra tio
1.6 1.6 1.6
1.4 PW7 1.4 PW7 1.4 PW7
1.2 1.2 1.2
1.0 1.0 1.0
0 .8 0 .8 0 .8
0 .6 0 .6 0 .6
0 .4 0 .4 0 .4
0 .2 0 .2 0 .2
0 .0 0 .0 0 .0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Time ( se c) Time ( se c) Time ( sec)
Ex ce ss p o r e wa te r p r e ss u r e r a tio

Exce ss p o r e wa te r p r e ssu r e r a tio

Exce ss p o r e wa te r p re ssu r e ra tio
1.6 1.6 1.6
1.4 PW6 1.4 PW6 1.4 PW6
1.2 1.2 1.2
1.0 1.0 1.0
0 .8 0 .8 0 .8
0 .6 0 .6 0 .6
0 .4 0 .4 0 .4
0 .2 0 .2 0 .2
0 .0 0 .0 0 .0
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Time ( se c) Time ( se c) Time ( sec)

(a) No improvement (Case1) (b) Irregular configuration (Case2) (c) Regular configuration (Case4)
Fig.8 Time history of excess pore water pressure ratio (P6 among columns and P7 on the
upstream side of the columns) in 1G tests

6. Results of 1-G shaking tests

The recorded time history of excess pore water pressure was treated by removing cyclic
components and only the average trend of its develoment is shown in Fig. 8. The concerned
configuration of columns are no column (Case 1), irregular configuration (Case 2), and
regularly triangular (Case 4). The location of the records are P6 (among columns) and P7
(upstream side of the columns). Note that the excess pore water pressure achieved 100%
development, whether or not columns were installed. The pore pressure ration greater than
unity in P7 records was induced by subsidence of the transducer after onset of liquefaction.
Deofmration of testes models is illlsutrated in Figs. 9-11. While deformation of vertical
columns of colored sand in contact with the side window is shown by black lines, the
inclinometer data in the central part of the mode is indicated by black dots. First, the
deformation of ground without columnar improvement (Case 1) shows that the most part of
deformation occurred in the upper liquefied layer, while the deformation in the unliquefiable
base layer is insignificant. The higher upstream side subsided and the lower downstream side
uplifted, conseauently reducing the slope gradient. Second, the Case 2 test wth the irregular
column configuration developed uplift in the upsream proximity of the improved area (+200
to +500 mm) probably becasue the columns redeced and dammed the lateral flow of liquefied
sand. This finding is a good contrast with the deformation of Case 4 (triangular configuration)
where the lateral flow of liquefied sand was easier. Similar difference was observed in the
cases of 35% improvement ratio as well.
Figures 12 and 13 compare the lateral displacement at the top (at the surface) of colored
sand as indicated by black dots in Figs. 9 to 11. It is first found that cases with any kind of
column configuration (Case 2 to 5) was of less extent of displaccement than in Case 1 without
columns. Thus, columns mitigate the lateral displacement of liquefied subsoil. Further, the
cases with the irregular configuration (Cases 2 and 3) showed the displacement even smaller
than in the cases of triangular configuration (Cases 4 and 5). Thus the mitigative effect of the
irregular configuration of columns is superior to that of the regular triangular configuration.
 Vertical position (mm)
600
CASE1 : No improvement 200Gal Subsidence
500 ■： Inclinometer Initial surface
400
Uplift
300
200
100
0
-1200 -1000 -800 -600 -400 -200 0 200 400 600 800 1000 1200
Holizontal position (mm)

Fig.9 Deformation of model ground of Case1

Vertical position (mm)

600
CASE2 : Irregular 25% 200Gal Improved area Initial surface Subsidence
500 ■： Inclinometer
400
Uplift
300
200
100
0
-1200 -1000 -800 -600 -400 -200 0 200 400 600 800 1000 1200
Holizontal position (mm)

Fig.10 Deformation of model ground of Case2

Vertical position (mm)

600
CASE4 : Regular 25% 200Gal Improved area Subsidence
500 ■： Inclinometer Initial surface
400
300
200
100
0
-1200 -1000 -800 -600 -400 -200 0 200 400 600 800 1000 1200
Holizontal position (mm)

Fig.11 Deformation of model ground of Case4

Lateral displacement (mm)

△ :CASE1(No improvement)
● :CASE2(Irregular 25%)
160
□ :CASE4(Regular 25%)
140
120 Improved area
100
80
60
40
20
0
-1200 -1000 -800 -600 -400 -200 0 200 400 600 800 1000 1200
Holizontal position (mm)

Fig.12 Lateral displacement at the surface at the end of shaking (improvement ratio 25%)
Lateral displacement (mm)

△ :CASE1(No improvement)
● :CASE3(Irregular 35%)
160
□ :CASE5(Regular 35%)
140
120 Improved area
100
80
60
40
20
0
-1200 -1000 -800 -600 -400 -200 0 200 400 600 800 1000 1200
Holizontal position (mm)

Fig.13 Lateral displacement at the surface at the end of shaking (improvement ratio 35%)
7. Results of centrifugal shaking tests
Time histories of excess pore water pressure are indicated in Fig. 14 for Case 6 without
improvement and Fig. 15 for Case 7 with the irregular configuration of columns. Similar to
the previous figure, the average trend of poressure development is illustrated here as well.
Note further that possible effect of sinking or floating of piezometers during liquefaction is
not corrected for. Hence, any pore pressure ratio greater than unity or any residual pore
pressure at a long time after the shaking were the consequence of these kinds of problems.

1.6 1.6 1.6

1.4 PW3 1.4 PW6 1.4 PW9
Excess pore water

Excess pore water

Excess pore water

pressure ratio

pressure ratio

pressure ratio
1.2 1.2 1.2
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0
-50
PW3 0 50 100 150 200 250 300 -50
PW6 0 50 100 150 200 250 300 -50
PW9 0 50 100 150 200 250 300
Time (sec) Time (sec) Time (sec)
1.6 1.6 1.6
1.4 PW2 1.4 PW5 1.4 PW8
Excess pore water

Excess pore water

Excess pore water

pressure ratio

pressure ratio

pressure ratio
1.2 1.2 1.2
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0
-50
PW2 0 50 100 150 200 250 300 -50
PW5 0 50 100 150 200 250 300 -50
PW8 0 50 100 150 200 250 300
Time (sec) Time (sec) Time (sec)
1.6 1.6 1.6
1.4 PW1 1.4 PW4 1.4 PW7
Excess pore water

Excess pore water

Excess pore water

pressure ratio

pressure ratio

pressure ratio
1.2 1.2 1.2
1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0
-50
PW1 0 50 100 150 200 250 300 -50
PW4 0 50 100 150 200 250 300 -50
PW7 0 50 100 150 200 250 300
Time (sec) Time (sec) Time (sec)

Fig.14 Time history of excess pore water pressure ratio of Case 6 without improvement
1.6 1.6 1.6
1.4 PW3 1.4 PW6 1.4 PW9
Excess pore water

Excess pore water

Excess pore water

pressure ratio

pressure ratio

pressure ratio

1.2 1.2 1.2

1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0
-50
PW3 0 50 100 150 200 250 300 -50
PW6 0 50 100 150 200 250 300 -50
PW9 0 50 100 150 200 250 300
Time (sec) Time (sec) Time (sec)
1.6 1.6 1.6
1.4 PW2 1.4 PW5 1.4 PW8
Excess pore water

Excess pore water

Excess pore water

pressure ratio

pressure ratio

pressure ratio

1.2 1.2 1.2

1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0
-50
PW2 0 50 100 150 200 250 300 -50
PW5 0 50 100 150 200 250 300 -50
PW8 0 50 100 150 200 250 300
Time (sec) Time (sec) Time (sec)
1.6 1.6 1.6
1.4 PW1 1.4 PW4 1.4 PW7
Excess pore water

Excess pore water

Excess pore water

pressure ratio

pressure ratio

pressure ratio

1.2 1.2 1.2

1.0 1.0 1.0
0.8 0.8 0.8
0.6 0.6 0.6
0.4 0.4 0.4
0.2 0.2 0.2
0.0 0.0 0.0
-50
PW1 0 50 100 150 200 250 300 -50
PW4 0 50 100 150 200 250 300 -50
PW7 0 50 100 150 200 250 300
Time (sec) Time (sec) Time (sec)

Fig.15 Time history of excess pore water pressure ratio of Case7 with irregular improvement
 In Case 6 without improvement (Fig. 14), the excess pore water pressure developed
100 % except the slightly less value in PW7. Even in Case 7 with columns (Fig. 15), the pore
pressure developed significantly as well.
Figures 16 to 19 illustrates the ground deformation as depicted by colored sand. In Case
6 without improvement (Fig. 16), the lateral displacement was maximal at the surface while
being null in the compacted base layer. Being similar to the aforementioned 1-G tests but
more evident, the subsidence on the upstream side and uplift in the lower part of the slope
were observed, which is deformation towards a level shape to achieve the minimal potential
energy.
When columns were installed with the irregular configuration (Fig. 17 for Case 7 and Fig.
18 for Case 8), the surface of unimproved parts became nearly level after shaking. In contrast,
the surface around the columns remained inclined similar to the pre-shaking gradient. In more
details, the lateral displacement in this area has a linear variation in the vertical direction.
Deformation in the deeper elevation is more significant here than in the unimproved parts.
Figure 19 indicates the deformation of Case 9 test in which columns were installed in a
triangular configuration. Similar to other tests, the model ground became level on the
upstream and downstream sides of the model, while the improved part maintained the original
surface gradient. The lateral displacement in the deeper part of the improved area exhibits
fluctuation.

500 500
Vertical position (mm)
Vertical position (mm)

CASE6 No improvement CASE7 Irregular

Subsidence
400 400 Improved area
Subsidence
Uplift Uplift
300 300

200 200

100 100

0 0
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Holizontal position (mm) Holizontal position (mm)
Fig.16 Deformation of model ground of Case 6 Fig.17 Deformation of model ground of
Case 7 with template at the top of
irregular columns

500 500
Vertical position (mm)
Vertical position (mm)

CASE8 Irregular CASE9 Regular

Subsidence Subsidence
400 400 Improved area
Improved area
Uplift Uplift
300 300

200 200

100 100

0 0
-300 -200 -100 0 100 200 300 -300 -200 -100 0 100 200 300
Holizontal position (mm) Holizontal position (mm)
Fig.18 Deformation of model ground of Case 8 Fig.19 Deformation of model ground of
with free top of irregular columns Case 9 with template at the top of
regular triangular column configuration
 　 :Case6(No improvement)
2.0
　 :Case7(Irregular)
　 :Case8(Irregular)
　 :Case9(Regular)

Lateral displacement (m)

1.5

1.0

0.5
Improved area

0.0
-20 -15 -10 -5 0 5 10 15 20
Holizontal position (m)

Fig.20 Lateral displacement at the surface

Figure 20 demonstrates the lateral displacement at the surface in the prototype scale.
Case 6 without columns developed the maximum displaceement in the middle part of the
container, while showing very small displacement at two ends of the slope. Similar
distribution of displacement was observed in 1-G tests as well (Figs. 12 and 13 in this paper
together with the 6-metermodel tests by Sasaki et al.12)). It is important to find the reduced
extent of displacement in Case 7 and Case 8 where the irregular ocnfiguration of columns was
employed. The regular triangular configuration (Case 9) exhibits reduced displacement as
well.
More details are discussed in this paragraph. First, Case 7 (irregular) and Case 9
(triangular) had the column tops connected to the same template to allow a certain extent of
rotation but reduced lateral translation of columns. Near the upstream edge of the improved
area (0 to 10 m), Case 7 achieved less horizontal displacement than Case 9. In contrast, the
greater displacement of Case 7 near the downstream edge (-5 to -2 m) was probably induced
by the lack of columns near the side wall and easy flow of liquefied sand therein (Fig. 7b),
and does not have an important implication.
Second, the effect of different conditions at the top of columns are discussed; Case 7
with template and Case 8 with free end. Basically there is not significant difference in the
magnitude of lateral displacement. In both tests, all the columns maintained their relative
positions and tilted to some extent towards the downslope direction. Although the significance
of this tilting is not well understood, the maintained irregular configuration in Case 8 worked
well to mitigate the lateral displacement.

8. Conclusions
The mitigation of lateral displacement of liquefied ground was studied by running 1-G and
centrifugal model tests with a special interest in the irregular configuration of vertical
embedded columns. The following conclusions were drawn from the present study.
(1) 1-G tests showed heaving on the upstream side of the improved area. This deformation
was induced by the damming effect of the irregula configuration of columns.
(2) The triangular configuration did not exhibit such a damming-up effect because its
mitigative effect is less significant and passing-through of liquefied sand was easier.
(3) Consequently, the lateral displacement was reduced more efficiently to some extent by the
irregular configuration than by the regular triangular configuration of columns.
(4) The centrifugal testsindicated that the surface of the 10% slope became level after shaking,
 while the surface within the column installation did not change much.
(5) In case there is a free space among columns where flow is easy, the lateral displaceemtn
increased and occurred even at the deeper elevation.
(6) The irregular configuration mitigated the lateral displacement more efficiently than the
triangular configuration.

References
[1] Suzuki, Y., Tokito, K., Suzuki, Y. and Babasaki, R. : Introduction examples of soil
improvement method by stabilization, The Foundation Engineering and equipment, Vol.17,
No.9, 87-95, 1989. (In Japanese)
[2] Koga, Y., Taniguchi, E., Nakasumi, I. and Awanami, K. : Shaking table tests on deep
mixing method against liquefaction of sand deposit, Proc., 41th Annual Conf. of the Japan
Society of Civil Engineering, 3, 201-202, 1986. (In Japanese)
[3] Yasuda, S., Murasawa, Y., Konishi, T., Tanaka, T. and Uchiyama, J. : Effect of soil-cement
column with a triangular grid pattern as a countermeasure against liquefaction, Proc., of the
38th Japan National Conference on Geotechnical Engineering, 1881-1882, 2003. (In
Japanese)
[4] Tanaka, T., Yasuda, S., Murasawa, Y., Konishi, T and Uchiyama, J. : Effects of Column
type deep mixing method on liquefaction prevention, Proc., of the JSCE Earthquake
Engineering Symposium, Vol.27, No.210, 2003. (In Japanese)
[5] Yamamoto, Y., Takahashi, N. and Kurokawa, Y. : Analytical Study on the Liquefaction
Countermeasure Effect of the Pile Type Deep Mixing Method, Reports of technical research
institute of Sumitomo Mitsui Construction Co., Ltd., No.4, 45-53, 2006. (In Japanese)
[6] Takahashi, N., Bahmanpour, A., Towhata, I., Yamada, S. and Yamamoto, Y. : Shaking
model tests on column type deep mixing method on liquefaction prevention, Proc., of the 45th
Japan National Conference on Geotechnical Engineering, 1439-1440, 2010. (In Japanese)
[7] Takahashi, N., Bahmanpour, A., Towhata, I., Yamada, S and Yamamoto, Y. : Effective
configuration of columns for deep mixing method against lateral flow by liquefaction, Proc.,
65th Annual Conf. of the Japan Society of Civil Engineering, 1, 107-108, 2010. (In Japanese)
[8] Derakhshani, A., Takahashi, N. Baumanpour, A. and Towhata, I. : Influence of Deep
Mixing Soil Stabilization Pattern on Liquefaction Mitigation, Proc., of the 4th Japan-Taiwan
Joint Worlshop on Geotechnical Hazards from Large Earthquakes and Heavy Rainfalls, 2010.
[9] Towhata, I., Takahashi, N., Bahmanpour, A., Yamada, S. and Liu, B.A.: Shaking model
tests on mitigation of liquefaction-induced ground flow by underground columns,
International Symposium on Recent and Future Technologies in Coastal Development,
Yokohama, 2010.
[10] Derakhshani, A., Takahashi, N., Bahmanpour, A., Yamada, S. and Towhata, I. :
Experimental study on effects of underground columnar improvement on seismic behaviour
of quay wall subjected to liquefaction, Proc., of the 2011PanAm CGS Conference, 2011.
[11] Kitazume, M. : Development of PHRI Mark II Geotechnical Centrifuge, Technical Note
of the Port and Harbour Research Institute Ministry of Transport, Japan, No.812, 1-35,
1995.(In Japanese)
[12] Sasaki, Y., Towhata, I., Tokida, K., Yamada, K. Matsumoto, H., Tamari, Y. and Saya, S. :
Mechanism of Permanent Displacement of Ground Caused by Seismic Liquefaction, Soils
and Foundations, Vo.32, No.3, 79-96, 1992.

View publication stats