You are on page 1of 6

Physical Modelling in Geotechnics – 6th ICPMG ’06 – Ng, Zhang & Wang (eds)

© 2006 Taylor & Francis Group, London, ISBN 0-415-41586-1

Seismic responses of column and grid-type improved grounds

A. Ishikawa & Y. Asaka


Shimizu Corporation, Tokyo, Japan

ABSTRACT: The seismic responses of column and grid-type improved grounds were investigated experimen-
tally and by numerical calculations using an electromagnetic shaking table and the homogenization method
respectively. The first-order resonant frequencies depended on the shape of the improvements. Experimental
resonant frequency values for column-type improvement, grid-type improvement and non-improved soil were
7.2 Hz, 7.8 Hz and 4.0 Hz, respectively. The equivalent stiffness calculated using the homogenization method
was dependent on the shape, stiffness and replacement ratio of the improvements. Static eigenvalue analysis
using equivalent stiffness can reasonably express the first-order resonant frequencies of improved soils.

1 INTRODUCTION improvement, as illustrated in Figure 1(b). The grid


divided the shear box into 120 mm × 168 mm cells
The partial soil-cement mixing method for improv- with grid thicknesses of 20.0 mm and 21.5 mm in the
ing seismic response of soft clay is often used in two directions. The volumes of the improvements were
Japan. Although the effect against seismic response is 20% of the internal volume of the shear box in both
sometimes confirmed through FEM calculations, sim- cases. The improvements were formed by a mixture
ple evaluation methods ref lecting shape, stiffness and of cement, Toyoura sand and kaolinite clay, and the
replacement ratio of the improvements have not yet resulting stiffness (Vs ) and ρd values were 650 m/s
been proposed. Since seismic responses are reflected and 1.90 g/cm3 , respectively. The soil layer was pre-
in the overall stiffness of the improved soils, a rea- pared in the shear box by placing the improvements,
sonable evaluation of equivalent stiffness has prime and then filling withToyoura sand using the dry air plu-
importance. viation method. Non-improved soils were also tested
In this study, seismic response tests of column and as shown in Figure 1(c) and compared with the results
grid type improved soils were carried out using a of improved ones.
centrifuge with the aim of developing a simple seis-
mic evaluation method. The equivalent stiffness and
the corresponding first resonant frequencies were also 2.2 Instrumentation
calculated with an existing numerical method.
Accelerometers ACC1-4 were placed at the center
point of the shear box at 0.5, 10, 20, 30 cm (i.e. 0.15,
3.0, 6.0 and 9.0 m in prototype scale) below the surface
2 CENTRIFUGE MODELS of soil layer.ACC5 and 6 were placed on the surfaces of
the soil and improvement respectively, to monitor spa-
2.1 Shear box, soils, and improvement tial variance of accelerations, while ACC7 and 8 were
A 30 cm high soil layer of Toyoura sand and improve- fixed at the sides of shear box to measure vertical and
ments was prepared in a shear box, as shown in horizontal accelerations.
Figure 1. The internal dimensions of the box were: Bender elements BE1 and BE2 were placed at 10
745 mm length, 445 mm width and 360 mm height. and 20 cm (i.e. 3.0 and 6.0 m inprototype scale) below
The bulk density (ρd ) and relative density (Dr ) of the the surface to measure the S-wave velocities of the soil
soil were 1.53 g/cm3 and 65%, respectively. layer.
Two types of improvements were used in the tests.
The first was column-type, as shown in Figure 1(a).
2.3 Base input waves
Twelve columns were placed at regular intervals in the
shear box while the dimensions of one column were The radius of centrifuge from its central axis to the
74 mm × 74 mm × 300 mm. The second was grid-type shaking table was 311cm and the applied centrifugal

521
Table 1. Test cases.

Case no. Improvements Wave type α(g)

1 Column-type Sinusoidal wave 5


2 Grid-type Sinusoidal wave 3
3 No improvement Sinusoidal wave 5

10

αres/αinput
Column-type
5

0
10

αres/αinput
Grid-type
5

0
10

αres/αinput
No improvement
5

0
0 2 4 6 8 10
Frequency (Hz)

Figure 2. Transfer functions of acceleration.

acceleration was 30 g at the surface of the soil layer.


The shaking table was vibrated horizontally in the lat-
eral direction of the shear box. Sinusoidal waves were
used as an input motion to investigate the frequency
response characteristics of the improved soil layer. The
wave frequencies were swept up first from 30 Hz to
300 Hz under base input acceleration (α) of 3 to 5 g as
shown in Table 1.

3 TEST RESULTS

3.1 Seismic response of improved soils


Transfer functions of acceleration measured at 0.15,
3.0, 6.0 and 9.0 m below the surface along the central
axis of soil layers are shown in Figure 2. The first-
order resonant frequencies of non-improved soils were
4.0 Hz while the peak response factors varied from
1.0 to 4.3. In the case of column-type improvement,
the peak at 4.0 Hz, which was greatest in the non-
improved case, decreased and that at 7.2 Hz became
greatest instead. The peak response factors varied from
1.0 to 5.5 while the frequency peaks were steeper
than in the non-improved case. In case of grid-type
of improvement, the greatest response peak was seen
at 7.8 Hz and its shape was the steepest of the three
cases. These results indicated that the overall rigidi-
ties of the improved soils depended on the shape of
the improvements, and rigidities may be reflected in
Figure 1. Schematics of experimental setup.
resonant frequencies.

522
Table 2. Natural frequency of the models. 0

d f1 α U
Model (m) (Hz) (g) λ (mm) γ 2

4.5 × 10−4

Depth (m)
Column- 0.15 7.25 5.5 4.1
type 3 3.6 2.6 4
6 – –
9 1 0.7
6
Grid-type 0.15 7.8 3 9.0 3.4 3.7 × 10−4 Column-type
No 3 5.8 2.2 :Column-type
:Grid-type Grid-type
improvement 6 2.0 0.8 8 No improvement
9 1 0.4 :No improvement

0.15 5.4 5 4.3 5.5 1.1 × 10−3 0 200 400 0 200 400
3 4.2 5.5 Vs (m/s) Vs (m/s)
6 – – (a) Without stiffness reduction (b) With stiffness reduction
9 1 1.3
Figure 4. Vs distributions of soil layers.

Shear strains in all the cases were calculated using


Equation 2 and the results are illustrated in Table 2.
0

2 where, D: depth of soil layer (=9 m) and U : surface


resonant displacements (mm: prototype scale). Calcu-
lated shear strain values of the improved soils were
Depth (m)

4 4.5 × 10−4 and 3.7 × 10−4 , respectively, while the


value for non-improved soil was 1.1 × 10−3 . Since the
stiffness reduction due to the non-linear properties of
6 soils with these strain values cannot be neglected, the
overall stiffness of the improved soils was evaluated
Column-type
as discussed inthe following paragraphs.
8
Grid-type 3.2 S-wave velocity distributions of soil layer
No improvement
The S-wave velocity (Vs ) distributions in soils mea-
10 sured by bender elements with increasing centrifuge
0 2 4 6 8 10 12
Absolute displacement (mm) acceleration are shown in Figure 4(a). The distribu-
tions for column-type improvement and non-improved
Figure 3. Resonant displacement of soils. cases were almost identical, and they varied from 60 to
70 m/s, and 200 to 220 m/s, respectively, at the surface
and 6 m below the surface. Vs distributions in grid-type
The resonant absolute displacements were obtained improvement were obviously smaller than in the other
from Equation 1, two cases. This was considered to be the result of fric-
tion between sand and grid improvements. Thus, soils
in grid improvements had less rigidity than in the other
cases considered.
Kuribayashi et al. (1974) and Iwasaki et al. (1978)
where, f1 : resonant frequency, λ: acceleration response originally proposed a relationship between shear mod-
factor, and α: base input acceleration. Depth distribu- ulus (G) and shear strain (γ). A modified correlation
tions of U along the central axis of the soils are shown between these parameters for Toyoura sand following
in Table 2 and Figure 3. The displacement distributions Onoue et al. (1998) is illustrated in Figure 5, where
of column and grid-type improvements were almost at normalized shear modulus (Gmax ) is the G value cor-
the same level and fell along a straight line, as seen in responding to 1.0 × 10−6 shear strain. As shown in
Figure 3. The shear strains of soils in both cases were Figure 5, shear modulus is dependent not only on
thus considered to be uniform. The displacements of strain, but also on confining pressure (σc ). Stiffness
the original soils were obviously larger than those of reduction in each test was calculated from Figure 5.
improved soils. For example: the stiffness reduction in soil was set

523
1 As the compliance matrix of C H can be expressed
Column-type
as Equation 7, equivalent Vs and VsvH were calculated
0.8 0.58
σ'c =98kPac 0.46 using Equation 8.
σ'c =49kPac
0.34
0.6
G/Gmax

σ'c =24.5kPac No improvement


Grid-type
0.4 0.62 0.38
0.52 0.30
0.2 0.38 0.18

0 -6
10 10-5 10-4 10-3 10-2
Shear strain,γ

Figure 5. Relationship between normalized shear stiffness


and shear strain.

to 0.38 at depths with σc less than 24.5 kPa, 0.52 for where:
σc = 24.5 ∼ 49 kPa and 0.62 for σc = 49 ∼ 98 kPa in ρeq : mean density of improved soils. The subscript “3”
grid-type improvements.The same stiffness reductions in the above equations denotes the vertical direction.
were applied in the other cases, and the resultant Vs dis- Relationships between calculated equivalent Vs of
tributions were translated using the Vs ∼ G relationship the improved soils after normalization with Vs of
(Eq. 3) as illustrated in Figure 4(b). original soils and replacement ratio are illustrated in
Figure 6. Vs of the improvements after normalization
with that of original soils are denoted by “a” in this
figure.
The calculated equivalent Vs values depended on the
shape of the improvements, and were different from the
4 CALCULATED RESULTS simply averaged values, as shown in Figure 6.
Equivalent Vs distributions of the soils were calcu-
4.1 Equivalent stiffness of improved soils by lated using modified Vs distributions shown in Fig-
homogenization method ure 4(b) and the Vs value of the improvements (i.e.
The overall stiffness of the improved soils can be eval- 650 m/s), and the results are illustrated in Figure 7.
uated by comparing experimental and calculated (from The results showed a significantly greater increase
equivalent stiffness) resonant frequencies. Equivalent in soil stiffness for grid-type improvement than for
shear stiffness values of the improved soils in this column-type improvement.
investigation were evaluated using the homogeniza-
tion method (Terada et al. 2003), where homogenized 4.2 Eigen value analysis of the models
stiffness matrix can be expressed in equation form as: Eigen value analysis of the models was carried out
using equivalent stiffness and the mean density of the
improved soils. The experiments in prototype scale
were modeled in the analysis (Figure 8) and the cal-
where, culated results are shown in Table 3. The calculated
C H : homogenized stiffness matrix and the first-order resonant frequencies were similar to the
superscript “H ” simply denotes corresponding experimental values although the dif-
“homogenized”, ference for column-type improvement was relatively
C : stiffness matrix of soils and improvements, large. The reason behind the big difference, which is
 : volume average in unit cell, under consideration, is unknown at the moment.
X : characteristic deformation matrix
expressed by 5 CONCLUSIONS
Equation 5 and calculated using Equation 6 as
shown below: The following conclusions have been drawn from this
study on the seismic response of improved soils:
1. The first-order resonant frequencies of improved
soils were dependent on the shape of improvements.
Resonant frequencies for column-type and grid-
where, type improvements and non-improved soils were
u: displacement of local unit cell. 7.2, 7.8 and 4.0 Hz respectively.

524
3 0
Equivalent Vs /Vs of the original soil

2
a=20 10
6
Column–type

Depth (m)
4 4
2
Column-type
6
Grid-type
2
No improvement
8

0 200 400
1
0 35 70 Vs (m/s)
Replacement ratio (%)
(a) column-type improvement Figure 7. Equivalent Vs distributions of improved soils.
20
Equivalent Vs /Vs of the original soil

a=20
18
16
Grid–type
14
12
10
10
8
6 Figure 8. Eigenvalue analysis of equivalent stiffness of
improved soils.
4
2
Table 3. Comparison of first-order resonant frequencies.
0
0 50 100 Experiment Eigen value analysis
Replacement ratio (%) Model (Hz) (Hz)
(b) grid-type improvement
Column-type 7.2 5.7
20 Grid-type 7.8 7.5
a=20 No improvement 4.0 3.8
Averaged Vs /Vs of the original soil

18
16
2. The equivalent stiffness of column and grid-type
14 improved soils was calculated by the homogeniza-
12 tion method. The equivalent stiffness values were
10 dependent on the shape, stiffness and replacement
10
ratio of the improvements.
8 3. Static eigenvalue analysis using equivalent stiff-
6 ness can reasonably express the first-order resonant
4
frequencies of improved soils.
2

0 ACKNOWLEDGEMENTS
0 50 100
Replacement ratio (%)
(c) averaged Vs by replacement ratio The authors would like to thank Dr. K. Terada and
Dr. T. Kyoya, associate professors of Tohoku Uni-
Figure 6. Correlations between equivalent Vs and
versity, for their suggestions and help on numerical
replacement ratio. calculations. Thanks are also due to Mr. Kusukame

525
and Mr. Yoshinari for his support in conducting Kuribayashi, E., Iwasaki, T., Tatsuoka, F. & Horiuchi, S. 1974:
experiments. Dynamic deformational properties of soils- Measurement
by resonant – column method. Tech. Memo., PWRI 912,
Tsukuba, Japan.
REFERENCES Iwasaki, T., Tatsuoka, F. & Takagi, Y. 1978: Shear moduli
of sands under cyclic torsion shear loading. Soils and
Onoue, A., Kazama, H. & Hotta, H. 1998. Seismic response Foundation 18(1): 39–56.
of a stacked-drift-type tunnel in dry sand. Proceedings of Terada, K. & Kikuchi, N. 2003. Kinshitsuka Nyumon (in
the international conference centrifuge 98, Tokyo 1998: Japanese): 57–66.
237–242. Rotterdam: Balkema.

526

You might also like