Ground Improvement Proceedings of the Institution of Civil Engineers

Volume 166 Issue GI1 Ground Improvement 166 February 2013 Issue GI1
Pages 44–58 http://dx.doi.org/10.1680/grim.11.00028
Consolidation settlement of floating- Paper 1100028
column-improved soft clayey deposit Received 14/09/2011 Accepted 26/06/2012
Published online 08/11/2012
Pongsivasathit, Chai and Ding Keywords: columns/embankments/geotechnical engineering

ICE Publishing: All rights reserved

Consolidation settlement of
floating-column-improved soft
clayey deposit
j
1 Supasit Pongsivasathit PhD j
3 Wenqi Ding PhD
Department of Civil Engineering, Faculty of Engineering, Professor, Department of Geotechnical Engineering, Tongji
Rajamangala University of Technology, Thunyaburi, Thailand University, Shanghai, China
j
2 Jinchun Chai Dr Eng
Professor, Department of Civil Engineering and Architecture, Saga
University, Japan

j
1 j
2 j
3

The interaction behaviour between a floating column and the surrounding soil, with or without a surface cement
stabilised slab, has been investigated by laboratory model tests as well as finite-element analysis (FEA) using an
axisymmetric unit cell model. Based on the test and the FEA results, a method for calculating the consolidation
settlement–time curve of floating-column-improved soft clayey subsoil has been developed. This method has been
modified from the earlier method of Chai and Pongsivasathit by treating a part of the column improved layer with a
thickness of Hc as an unimproved layer in settlement calculation to consider the effect of possible penetration of the
column into the soft soil layer below the column. An explicit equation for calculating the value of Hc has been
proposed as a function of area improvement ratio (Æ), depth improvement ratio (), load intensity (p) and the
undrained shear strength (su ) of the soft clayey soil. The validity of the method has been checked using the
laboratory model test results as well as four field case histories in Japan.

Notation e0 i initial void ratio in the corresponding subsoil layer

Ac area of the column H thickness of soft clayey layer
Ae area of the unit cell which represents a column and HL length of the column
its improvement area H1 thickness of the upper layer for the s(t ) calculation
c9 effective cohesion of soil H1c thickness of the upper layer for the U(t ) calculation
cv coefficient of consolidation H1 i thickness of subsoil layers in H1
cv1 coefficient of consolidation for layer H1c H2 thickness of the lower layer for the s(t ) calculation
cv2 coefficient of consolidation for layer H2c H2 i thickness of subsoil layers in H2
Dc constrained moduli of the column H2c thickness of the lower layer for the U(t ) calculation
Dc i constrained moduli of the column of the sub-layer H20 initial thickness of lower layer
H1 i kc hydraulic conductivity of the column
Ds constrained moduli of the surrounding soil kh hydraulic conductivity in the horizontal direction
Ds i constrained moduli of the surrounding soil of the ks hydraulic conductivity of the smear zone
sub-layer H1 i kv hydraulic conductivity in the vertical direction
dc diameter of column kv1 coefficient of permeability for layer H1c
de diameter of unit cell which represents a column kv2 coefficient of permeability for layer H2c
and its improvement area Ls length from the end of the column to a point at
ds diameter of smear zone which the settlement ratio (SR) satisfies a
E Young’s modulus pre-specified criterion
Ei Young’s modulus in the corresponding subsoil LR length ratio
layer LR9 length ratio including the thickness of a slab
e0 initial void ratio M slope of the critical state line in (q, p9) plot

44
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

mv1 constrained compressibility for layer H1c 1. Introduction

OCR overconsolidation ratio Deep cement mixing, normally forming stronger and stiffer
p loading intensity or consolidation pressure columns in the ground, is a widely used soft clayey ground
pa atmospheric pressure improvement method (e.g. Broms and Boman, 1979; Bergado et
pc consolidation yield stress al., 1994). Recently, to reduce the construction cost and minimise
p0 pre-consolidation pressure the impact on the ground environment, a method of improving
p90 effective mean stress on isotropic consolidation soft clayey deposits by floating soil-cement columns, with or
line corresponding to the current stress state without a cement stabilised slab on the ground surface, is
p9 mean effective stress increasingly being used (Chai et al., 2009; Shen et al., 2001).
q deviatoric stress One of the important aspects in the design of such a system is the
qu unconfined compression strength of the calculation of the consolidation settlement of the improved soft
soil-cement column deposit.
RE relative error
SR settlement ratio Chai et al. (2009) studied the interaction behaviour between the
sf final consolidation settlement of the system floating column–slab system and the surrounding soft soil by
(sf )cal calculated final settlement finite-element analysis (FEA) using an axisymmetric unit cell
(sf )mea measured final settlement model. The results of FEA revealed that the relative penetration
s(t ) total consolidation settlement at any time of the column into the surrounding soft clayey soil is influenced
s1 (t ) consolidation settlement at any time for layer H1 by (a) area improvement ratio (Æ ¼ Ac /Ae ; Ac being the cross-
s2 (t ) consolidation settlement at any time for layer H2 sectional area of the column and Ae the cross-sectional area of
su undrained shear strength of the soft soil the unit cell which represents a column and its improvement
U(t ) degree of consolidation at any time area); (b) depth improvement ratio ( ¼ HL /H; HL being the
U1 i (t ) average degree of consolidation of the subsoil length of the column and H the thickness of soft clayey layer);
layers in H1 at time t (c) load intensity ( p), and (d ) the strength and stiffness of the
U2 i (t ) average degree of consolidation of the subsoil soft deposit. A method was then proposed to calculate the final
layers in H2 at time t consolidation settlement of the floating column–slab system con-
wL liquid limit sidering the effect of Æ and  only. Based on the the Chai et al.
wP plastic limit (2009) method, Chai and Pongsivasathit (2010) developed a
Æ area improvement ratio method for calculating the consolidation settlement–time curves
 depth improvement ratio of the floating column–slab improved deposit.
9 depth improvement ratio including the thickness of
a slab There are some shortcomings in the method proposed by Chai et
ªt total unit weight al. (2009). First, it does not include the effect of the load
˜p1 i total vertical stress increments in sub-layer H1 i intensity and the undrained shear strength (su ) of the soft soil into
˜p2 i total vertical stress increments in sub-layer H2 i the equations for calculating Hc which is the thickness of the part
c settlement of the column at a point considered of the column improved layer to be treated as an unimproved
s settlement of soil at periphery of unit cell at same layer in settlement calculations. As a result, it will overpredict the
elevation selected for measuring c settlement for a stiffer deposit under a lower surcharge load (for
 stress ratio example 50 kPa), and underpredict the settlement for a very soft
k slope of unloading–reloading line in e ln p9 plot deposit under a higher surcharge load (p . 150 kPa) (Chai et al.,
ki slope of unloading–reloading line in e ln p9 plot in 2009). Second, in engineering practice there are cases where the
sub-layer H2 i soft deposit is only improved by column inclusions and without a
º slope of virgin compression line in e ln p9 plot cement-stabilised surface slab, but the method cannot be directly
ºi slope of virgin compression line in e ln p9 plot in applied to this kind of situation. Third, in this method, for
the corresponding subsoil layer simplicity the effect of Æ is represented by a linear function, but
 Poisson’s ratio the numerical results indicate that the relation between Æ and the
i Poisson’s ratio in the corresponding subsoil relative penetration of a column into the surrounding soft soil is
layer non-linear. As a result, the method under-predicts the settlement
9av i average effective vertical stress in the for Æ , 15%.
corresponding subsoil layer
9v i initial vertical effective stress in the sub-layer In this paper, laboratory model tests as well as further FEA were
H2 i conducted to investigate the interaction behaviour between the
0 , 1 , 2 embankment loads calculated by the Osterberg floating column and the surrounding soft clayey soil. Based on
(1957) method the results, a modified method for predicting the settlement–time
9 effective internal friction angle curve of floating-column-improved soft clayey soil with or with-

45
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

out a surface cement-stabilised slab has been proposed. The
proposed method has been applied to four field cases in Japan,
and its usefulness is demonstrated.
2. Laboratory model tests
2.1 Set-up and test procedure
Laboratory model tests were conducted to investigate the behav-
iour of a model floating soil-cement column improved ground.
The cylinder mould is 0.45 m in diameter and 0.9 m in height
and made of PVC. The test set-up is shown in Figure 1. The two
types of soil sample used were reconstituted Ariake clay, and
Ariake clay–sand mixture (mixed soil). The properties of the soils
are listed in Table 1. The undrained shear strength of the soils
was measured by both laboratory vane shear test (vane: 20 mm in
Drainage
Air pressure Shaft
O-ring
0·12 m
diameter and 40 mm in height) and unconfined compression test
using reconstituted soil samples under a consolidation pressure of
about 38 kPa (a pressure used to make model ground for the
model test). The su values from both tests are very close and the
values shown in Table 1 are average ones.
The cement used for making the model column was US10, a
typical cement used for ground improvement in Japan. US10 is a
type of cement produced by Ube Material Industries, Japan, using
waste concrete as part of the raw material. Its compression
strength is almost the same as commonly used Portland cement.
As for the properties of the model column, based on the results
of unconfined compression tests, the Young’s modulus of the
model column was estimated as 6 3 104 kPa in the case of the
Ariake clay soil and 1 3 105 kPa for the mixed soil, and the
Poisson’s ratio () was assumed as 0.2. Non-woven geotextile
with a thickness of about 3 mm (under zero confining pressure)
was used as drainage material at the bottom and the top of the
model ground. The geotextile was made of polypropylene and
weighed about 130 g/m2 : To accelerate the process of consolida-
tion during the preloading (making model ground) process, a
mini-prefabricated vertical drain (mini-PVD) was installed in the

middle of the model ground. The mini-PVD was made by folding

the geotextile in three layers with a cross-section of 30 mm by
9 mm.

Geotextile The detailed procedures of testing can be found in Chai and

Pongsivasathit (2010) and the main steps are briefly explained as
Column

follows.

Soil
sample
2.1.1 Test set-up and pre-consolidation
0·78 m

First, three layers of the geotextile were put at the bottom of the
model as drainage layers. A thin layer of silicon grease was
0·55 m

Piezometer painted on the inside wall of the mould to reduce friction. Then,
the remoulded soil sample with a water content higher than its
0·40 m

liquid limit (about 120% and 60% for the Ariake clay and the
mixed soil respectively) was put in the mould layer by layer.
0·20 m

Geotextile
Piezometers were installed at locations of 0.2 m, 0.4 m and
0.55 m height from the bottom. When the thickness of the soil
sample reached 0.78 m, the mini-PVD was installed (pushed in
by a stainless steel rod) in the centre of the model, and three
0·45 m
layers of the geotextile were placed on the top of the model. Then
Figure 1. Set-up of the laboratory model test the loading system as shown in Figure 1 was set up, and air
pressure of 40 kPa was applied. Considering the effect of the

Soil type wL : % wP : % ªt : kN/m3  k º e0 su : kPa kh : 3104 m/day kv : 3104 m/day

Ariake clay 110 60 15.5 0.3 0.033 0.331 3.20 10.0 3.651 3.651
Mixed soil 58 36 15.9 0.3 0.015 0.149 1.52 13.1 2.740 2.740

Note: wL is liquid limit; wP is plastic limit; ªt is total unit weight; º is the slope of virgin compression line in eln p9 plot (p9 being the mean
effective stress); k is the slope of unloading–reloading line in eln p9 plot; e0 is initial void ratio; kh and kv are hydraulic conductivity in the
horizontal and the vertical directions respectively.

Table 1. Properties of soils used in laboratory model tests

46
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

shaft fixed on the piston, the pre-consolidation pressure (p0 ) was settlement at the top of the model ground and excess pore water
about 38 kPa. The consolidation was under a two-way drainage pressures were monitored.
condition. During the test, the settlement at the top of the model
ground and the excess pore pressure at three different locations In total five cases were tested with conditions as listed in Table 2.
were monitored.
2.2 Comparison of measured and calculated values
Chai et al. (2009) proposed a method to calculate the final
2.1.2 Soil-cement column installation
consolidation settlement of floating column–slab system improved
After the degree of pre-consolidation reached about 90%, judged
clayey ground, which treats a part of the soil-cement column
on the measured shape of the settlement–time curves and excess
improved layer, with thickness Hc (see Figure 2), as an unimproved
pore pressures, the test was stopped, the loading system was
layer and to calculate its compression using the properties of the
dismounted and the mini-PVD was withdrawn. Then at the centre
soft soil only. The core of the method is the proposed equation for
of the model ground, a hole with a diameter of 150–246 mm was
calculating the value of Hc , which is a function of Æ and .
made by an augur to a pre-designed depth. A frame was used to
fix the position of the augur to make sure the hole could be made
p
in the correct location. At the bottom of the hole a small hole
was left after withdrawal of the mini-PVD, which was carefully
filled with the clay. The soil removed by the augur from the Slab
model ground was mixed with the cement (16.6% by dry weight)
and put back into the hole to form the model soil-cement column.
For ease of mixing with the cement, the water content of the
excavated soil was adjusted to about 120% and 60% for the
Ariake clay and the mixed soil respectively. Then the model was
left for 2 weeks to cure the column before the further consolida- H1
tion test. According to laboratory unconfined compressive test HL
results of cement-stabilised Ariake clay samples at different
curing times, the strength with 2 weeks curing time is more than
H
90% of the strength with 28 days curing time (Taguchi et al.,
2007). To shorten the time period for one test, 2 weeks curing
time was adopted before the further consolidation test. HC

2.1.3 Consolidation test H2

After 2 weeks curing, three layers of geotextile were installed on
the top of the model ground, and the loading system was set up
again. Then a pressure of 38 kPa was applied first for about 1 day
to ensure firm contact between the piston and the top of the
model ground and to bring the model ground to a normally
consolidate state, and then the pressure was increased to 95 or Figure 2. Floating-column-improved soft clayey subsoil
190 kPa and the consolidation test was started. During the test,

Case Soil type Æ: % : % p0 : kPa P: kPa Settlement: mm RE: %

(sf )mea (sf )cal

L1 Ariake clay 11 70 38 95 35.26 31.21 11.5

L2 Ariake clay 30 50 38 95 37.86 37.16 1.8
L3 Ariake clay 11 70 38 190 68.38 59.95 12.3
L4 Mixed soil 30 50 38 95 27.32 27.93 2.2
L5 Mixed soil 30 70 38 95 18.98 19.22 1.3

Note: p is consolidation pressure; (sf )mea and (sf )cal are the calculated and measured final settlement respectively.

Table 2. Cases studied and comparison of measurements with

calculated values from Chai et al. (2009)

47
Ground Improvement
Volume 166 Issue GI1
The measured final settlements are compared with the calculated
values by the Chai et al. (2009) method. Although the applied
surcharge pressure was 95 kPa, it was increased to that level from
a pressure of about 38 kPa. Therefore, the increment was only
57 kPa. Then considering the degree of consolidation (of about
90%) under 38 kPa, an incremental consolidation pressure of
about 61 kPa was evaluated for calculating the final settlement. In
the case of a 190 kPa surcharge load, the incremental consolida-
tion pressure was about 156 kPa. The results are listed in Table 2.
In the table, the relative error (RE) is defined as the percentage of
the difference between the calculated and the measured settle-
ment divided by the measured settlement (Chai et al., 2009).
Based on the results in Table 2, the following comments can be
made.
(a) For the model tests using Ariake clay (L1, L2 and L3), the
Chai et al. (2009) method resulted in a good prediction of the
final settlement for L2 (Æ ¼ 30%) but under-evaluated the
settlement for L1 and L3, which had a small Æ value of 11%.
Further, the under-evaluation increased with the increase in
load intensity (comparing L1 and L3).
(b) Although relatively small, the calculation method over-
evaluated the final settlement for L4 and L5 that used the
mixed soil, which is stronger and stiffer than the reconstituted
Ariake clay.
All these results confirmed the tendency reported by Chai et al.
(2009). Further, the Chai et al. (2009) method is only applicable
for a column slab system, but in the field there are cases where
the soft deposit was only improved by column inclusions (e.g.
Igaya et al., 2010). Therefore, the interaction behaviour between
the soil-cement column with or without a surface cement
stabilised slab and surrounding soft soil was further investigated
by FEA focusing on the effect of the slab, load intensity, and su
value of the soil.
3. Numerical investigation of some
influencing factors
3.1 Assumed conditions for numerical investigation
The axisymmetric unit cell model used for numerical investiga-
tion is illustrated in Figure 3 together with the boundary
conditions adopted. In FEA, 15-node triangular elements were
used to repress the soil, the column and the slab. A fine mesh
was adopted which can simulate a shear band of about 20 mm in
thickness. Figure 4 shows a typical mesh for numerical investiga-
tion. The soft clayey soil was represented by the Soft-Soil model
(Brinkgreve, 2002). The column and slab were modelled as a
linear elastic material. The assumed thickness of the soft layer is
16 m. In the case where a slab was used, the thickness of the slab
is 1.0 m. The diameter of the column is 1.0 m. The other
parameters were assumed by referring to the site investigation
results of soft Ariake clay in the Saga Plain, Japan (Hino et al.,
2008) and the parameter values adopted by Chai et al. (2009); the
values are given in Table 3. The unconfined compression strength
48
Consolidation settlement of floating-
column-improved soft clayey deposit
Pongsivasathit, Chai and Ding
(qu ) of the soil-cement column and the slab was assumed as
550 kPa, the Young’s modulus (E) was estimated as 100qu and
the Poisson’s ratio at 0.2. It was assumed that the groundwater
level was 1.0 m below the ground surface. The program used for
the FEA is Plaxis 2D V.8.
The purpose of the analysis was to identify the interaction
mechanism between the column and the surrounding soft soil as
well as to quantify it. Thus, the results focus on the relative
settlement between the column and the surrounding soil. For ease
of presentation, a parameter called length ratio (LR) (Chai et al.,
2009) is used with definition as follows
1: LR ¼ Ls = H L
where Ls is the length from the end of the column to a point on
the column at that elevation. The settlement of the soft soil (s )
at the middle between two adjacent columns equals a pre-defined
portion (for example 0.95) of the settlement of the column (c ).
In other words, LR is a measure of the percentage length of the
column which has considerable relative movement (pre-defined
criterion) with the surrounding soft soil. A larger value of LR
means a longer portion of the column has considerable relative
movement with the surrounding soil. s /c is designated as the
settlement ratio (SR) (Chai et al., 2009).
3.1.1 Numerical results
3.1.1.1 EFFECT OF THE CEMENT-STABILISED SURFACE SLAB
With the same improvement depth, the length of the column for
the with-slab case will be smaller than for the without-slab case.
Then for a given improvement depth, for both cases, the LR value
will be different. To make a direct comparison of with- and
without-slab cases under the same improvement depth, the
following parameters are introduced for the with-slab case
HL þ Hs
9 ¼
2: H þ Hs
Ls
LR9 ¼
3: HL þ Hs
where Hs is the thickness of the slab and H is the thickness of the
soft layer below the slab.
The relationships between  (or 9) and LR (or LR9) are
compared in Figure 5. It can be seen that the without-slab case
has a smaller LR value, which means a smaller relative penetra-
tion of the column into the surrounding soil. The difference is
reduced with increase of  (or 9) value. The reasons for less
relative penetration of the column are: (a) less stress concentra-
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

CL CL
R R

p p

Permeable Permeable

Slab HS
r

Column
HL
HL
Impermeable

Impermeable

Impermeable

Impermeable
Column

H
H

Hc Hc

Sand Sand

Permeable Permeable

(a) With slab (c) Without slab

Figure 3. Unit cell model for floating-column-improved soft soil

tion at the top of the column; and (b) more surface settlement of on the relative penetration of the column. Since the su value of a
the surrounding soil. The different deformation modes for the deposit varies with depth, we propose to use the value at the
with- and without-slab cases are illustrated in Figure 6. The elevation of the end of the column, which will have the most
illustration in Figure 6b ignores stiffness of embankment or any significant influence on the penetration of a column into the
other upper-earth structure. surrounding soft soil. For the Soft-Soil model (Brinkgreve, 2002)
used, the su value can be calculated by the following equations
The effect of Æ is shown in Figure 7. When Æ is larger than 20%,
the LR of the without-slab system differs little from LR9 of the 1
su ¼ Mp90 (OCR)¸
with-slab system. However, when Æ is lower than 20%, the 4: 2¸þ1
difference between the two systems increases with reducing Æ
value.
 ¸
3.1.1.2 EFFECT OF p AND su VALUES OF THE SURROUNDING SOIL M 2 þ 2
p90 ¼ ( p9 þ c9 cot 9)
For a given soil condition, the relative penetration of the column 5: M2
into the underneath soft soil will increase with the increase of the
magnitude of the load at the end of the column, while for a given
load, the weaker the surrounding soil the larger the relative q
¼
penetration. Therefore, it is considered that the ratio of p and su 6: p9 þ c9 cot 9
(p/su ) can be a parameter to represent the influence of p and su

49
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

C.L. β or β⬘: %
40 50 60 70 80 90
p 0
Permeable
1m

0·2

LR or LR⬘: %
0·4

0·6 For the case of α ⫽ 20%

With slab: Without slab:
Impermeable

Impermeable

SR ⫽ 1·00 SR ⫽ 1·00
0·8
SR ⫽ 0·95 SR ⫽ 0·95
SR ⫽ 0·90 SR ⫽ 0·90
1·0

Figure 5. Comparison between with and without-slab cases on

the effect of 
15 m

Impermeable

Impermeable

Column
Column

(a) With a slab (b) Without a slab

0·913–1·118 m
Figure 6. Deformation modes of column–slab and floating-column
Permeable systems

Figure 4. A typical FEM mesh

where p90 is the effective mean stress on isotropic consolidation
line corresponding to the current yielding stress state, p9 is mean
effective stress, q is deviatoric stress, M is the slope of the critical
state line in (q, p9) plot,  is stress ratio, and ¸ equals (1  k/º).
Figure 8 shows the effect of p/su on LR, an index for the relative
penetration of the column. In the case of varying p, the su at the
end of the column was fixed. In the case of varying su , the
constant p value of 100 kPa was applied. The change of su was
achieved by varying cohesion, c9, in Equation (5). For SR ¼ 1.0,
LR increases with increase of p/su : However, when p/su . 8, the

Depth: m Soil layer  E: kPa k º c9: kPa 9: deg OCR ªt : e0 kh : 3108 m/s kv : 3108 m/s
kN/m3

0–4.0 Clay-1 0.3 — 0.065 0.65 5–20 30 4–1.1 13.6 3.3 3.0 2.0
4.0–16.0 Clay-2 0.3 — 0.065 0.65 5–20 30 1 .
13 6 .
33 .
30 2.0
Column 0.2 55 000 The same as the corresponding soil layer
Slab 0.2 55 000

Note: c9 is the effective cohesion of soil; 9 is the effective internal friction angle.

Table 3. Assumed model parameters

50
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

α: % 0·6
0 10 20 30 40 50
0 Modified f (α)
0·5 Bilinear function
0·2 (Chai et al., 2009)
0·4
LR or LR⬘: %

0·4

f (α)
0·3

0·6 For the case of β ⫽ 81·25%

With slab: Without slab: 0·2
SR ⫽ 1·00 SR ⫽ 1·00
0·8
SR ⫽ 0·95 SR ⫽ 0·95
0·1
SR ⫽ 0·90 SR ⫽ 0·90
1·0
0
Figure 7. Comparison between with-slab and without-slab on the 0 10 20 30 40 50 60
α: %
effect of Æ
Figure 9. Modified function of Æ

p /su
0 2 4 6 8 10 12
0

without-slab case, the effect of Æ can be expressed by a function

0·2 f1 (Æ) as follows
SR ⫽ 0·90

0·4
8: f 1 (Æ) ¼ f (Æ)(0:775 þ 0:5Æ)(0:10 < Æ < 0:45)
LR: %

0·6
SR ⫽ 1·00
Most empirical constants in Equations 7 and 8 were back-fitted
α ⫽ 20%, β ⫽ 80%
0·8 from the numerical results.
Varying p (su ⫽ 19·1 kPa)
Varying su ( p ⫽ 100 kPa) 4.2 Considering the effect of p and su
1·0
To include p and su in the equation for evaluating the Hc value,
Figure 8. Effect of p/su on LR the following approach was taken.

(a) Find an Hc value (by trial and error) which can result in the
rate of increase is reduced. For SR ¼ 0.90, LR increased almost same settlement as the measured value from the model test or
linearly with increase of p/su value. from FEA for column–slab cases, designated as Hc(mea) and
Hc(FEA) respectively.
(b) Determine H9c as follows
4. Modification of the Chai et al. (2009)
method
4.1 Modification of the effect of Æ 9: H9c ¼ H L f (Æ) g()
Based on the results reported by Chai et al. (2009) and the
numerical and laboratory model test results from this study, the
function f (Æ) has been modified as follows where g() is a function expressing the effect of  and can be
expressed as follows (Chai et al., 2009)
8
< 0:75  2:5(Æ) (Æ > 0:20)
f (Æ) ¼ 0:4  1:0(Æ) (0:20 , Æ < 0:45) 
: 1:62  1:6 (0:2 <  < 0:7)
7: 0 (Æ . 0:45) g() ¼
10: 0: 5 (0:7 <  < 0:9)

The comparison of the modified function with the original one

for a column–slab system is given in Figure 9. Referring to the (c) Then a ratio of Hc(mea) /H9c or Hc(FEA) /H9c can be obtained.
results in Figures 5 and 7, it is further proposed that for the Define a dimensionless parameter, ª, as

51
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

: Permeable
ppa 1 5 Slab
ª¼
11: s2u:5

Column

Column

Column

Column
kv1, mv1, cv1 H1c
HL
where pa is the atmospheric pressure. The relationship
between the ratio of Hc(mea) /H9c or Hc(FEA) /H9c and ª is shown H
Hc /2
in Figure 10. The FEA values are scattered. A possible reason
is that the proposed functions of f (Æ) and g() are kv2, mv2, cv2 H2c
approximations of the real situation. It has been confirmed
that for fixed values of Æ and , the Hc(FEA) /H9c , ª
relationship is much less scattered. Permeable or impermeable
(d ) Finally, an empirical equation has been proposed to consider
the effect of ª on the calculated Hc value as follows Figure 11. Two-layer model for calculating the degree of
consolidation
H c(FEA)
hðªÞ ¼ ¼ 0:27ln (ª)  0:41
12: H9c
The methods for determining kv1 and mv1 for the upper layer
(Figure 11), and the thicknesses (H1c and H2c ) of both layers are
Then it is proposed that as follows

13: H c ¼ H L f (Æ)g()h(ª) for with-slab case 1

mv1 ¼
15: ÆDc þ (1  Æ)Ds

14: H c ¼ H L f 1 (Æ)g()h(ª) for without-slab case !

2:5 H 21c k h
k v1 ¼ 1þ
16: d e 2 k v
5. Method for calculating the settement–
time curve
where Dc and Ds are the constrained moduli of the column and
5.1 Degree of consolidation the surrounding soil, H1c is the thickness of the upper layer, and
Chai and Pongsivasathit (2010) presented a method for determin-  can be expressed as follows (Hansbo, 1981)
ing the average degree of consolidation (U(t )) of floating-
column-improved soft soil deposit using a two-soil layer con- n kh 3 8 H 21c k h
solidation theory (Zhu and Yin, 1999) as illustrated in Figure 11.  ¼ ln þ ln ð sÞ  þ
17: s ks 4 3d 2c k c
1·4

1·2 where n ¼ de /dc , s ¼ ds /dc (dc , ds and de are diameter of column,

smear zone and unit cell which represents a column and its
Hc(FEA)
⫽ 0·271 ln(γ) ⫺ 0·41 improvement area, respectively), kc and ks are the hydraulic
Hc(FEA) /H ⬘c or H c(mea) /H ⬘c

1·0 H ⬘c
conductivities of the column and the smear zone, respectively. It
pp 1a· 5
γ⫽ should be noted that even for kc ¼ kv , kv1 is larger than kv
0·8 s 2u· 5 because a flow mode toward the column is assumed in calculating
the value of kv1 . For a soil–cement column formed by an in situ
deep mixing method, the authors suggest not considering the
0·6
effect of the smear zone when applying Equation 17 to a soil-
Measurement cement column.
0·4 In range of
α ⫽ 20–30% FEA
β ⫽ 50–70% Considering the higher stiffness, and therefore the coefficient of
0·2 consolidation of a stabilised slab, it is proposed to exclude the
0 100 200 300 400 500 600 700 slab from Hc , which implies that the bottom of the slab is
γ
permeable. So, for the thickness of the upper layer and the lower
Figure 10. Variation of Hc(FEA) /H9c and Hc(mea) /H9c layer, Chai and Pongsivasathit (2010) proposed (by trial and error)
that H1c ¼ HL  Hc /2. As for H2c , to consider the relatively large

52
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

strain in the unimproved layer, it has been proposed to use the
average thickness (before and after consolidation) of the lower
layer – that is, H2c ¼ H20  sf /2. H20 is the initial thickness of the
lower layer and sf is the final consolidation settlement of the
system (assuming that most compression is from the lower layer).
Note that the thicknesses of the layers for the degree of consolida-
tion are different from those for the settlement calculation.
5.2 Settlement–time curve
Generally, the settlement consists of two parts: the compression
of the column improved layer (s1 ) with a thickness of
H1 ¼ HL Hc and the compression of the unimproved layer plus
Hc layer (s2 ) with a total thickness of H2 : The equations for
calculating s1 and s2 values are as follows
where Ei is Young’s modulus,  i is Poisson’s ratio and 9av i is the
average effective vertical stress in the corresponding subsoil layer.
In Equations 19 and 21, ki , the slope of the unloading-reloading
line in the eln p9 plot, is used instead of º i , in case the
subsoil is in an overconsolidated state – that is, [9v i þ ˜p1i U1i (t )]
or [9v i + ˜p2 i U2 i (t )] is less than pc (the consolidation yield stress).
Finally, the settlement (total compression), s(t ), can be expressed
as follows
22: s(t) ¼ s1 (t) þ s2 (t)

Xn
˜ p1i H 1i U1i (t) Considering the complicated expression for calculating the excess
s1 (t) ¼
18: i¼1
Dci Æ þ (1  Æ)Dsi pore pressure distribution in a two-layer system, it was found that
using the average degree of consolidation to all sub-layers, an
acceptable settlement–time curve can be predicted, and it is
suggested this be used for most practical purposes.
X
n  
ºi ˜ p2i
s2 (t) ¼ H 2i ln 1 þ U 2i (t)
19: i¼1
1 þ e0i  v9 i 6. Application of the proposed method to
laboratory results
The parameters used for calculation are listed in Tables 2 and 4.
The calculated results are compared with the measured ones in
where H1 i and H2 i are the thickness of subsoil layers in H1 and Figure 12.
H2 respectively; U1i (t ) and U2i (t ) are the average degree of
consolidation of the subsoil layers in H1 and H2 at time t The proposed method yields a good prediction of the test data for
respectively; 9v i is the initial vertical effective stress in the sub- the cases using both the Ariake clay and the mixed soil. The
layer H2 i ; e0i is the initial void ratio; º i is the slope of virgin calculated final consolidation settlement of case L1 is less than
compression line in eln p9 plot in the corresponding subsoil the measured settlement due to the poor quality of the column
layer; ˜p1i and ˜p2 i are the total vertical stress increments in around its bottom end, which was verified by the post-test
sub-layer H1 i and H2i , respectively; Dci and Ds i are the con- investigation. For case L3 (higher loading case), the calculated
strained moduli of the column and the surrounding soil of sub- result by the proposed method is much better than that by the
layer H1i respectively and they can be calculated as follows method of Chai and Pongsivasathit (2010). For this case, the
value of h(ª) is about 1.27 and if the effect of p and su is not
E i (1  ı i ) considered, the Hc value will be underestimated.
Dci ¼
20: (1 þ ı i )(1  2ı i )
7. Application of the proposed method to
case histories
(1 þ e0i ) a9vi The modified method has been applied to four case histories –
Dsi ¼ three from Fukuoka, Japan (Fukuoka cases), and one from Saga,
21: ºi
Japan (Saga case).

Case º e0 cv1 : 3102 m2 /day cv2 : 3102 m2 /day kv1 : 3104 m/day kv2 : 3104 m/day H1c : m H2c : m

L1 0.331 2.383 10.71 0.34 1.38 1.02 0.39 0.24

L2 28.17 0.35 1.37 1.02 0.31 0.34
L3 10.72 0.34 1.38 1.02 0.41 0.25
L4 0.149 1.240 56.08 0.32 1.59 0.65 0.32 0.35
L5 43.25 0.30 1.38 0.65 0.45 0.24

Table 4. The adopted parameters for calculating degree of

consolidation

53
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

Elapsed time: day

proposed method only considers the compression of the soft
0 10 20 30 40 50 60
0 layer, the settlement differences (˜s) between measurement
By proposed method points S-1 and S-2 in Figures 13–15 are used to compare with
10
By Chai and Pongsivasathit (2010) the calculated values. For calculating the degree of consolidation,
20 Measurement the Young’s moduli of the columns and the slabs were assumed
Settlement: mm

Case L1: α ⫽ 11%, β ⫽ 70% (p ⫽ 100 kPa) as 100 times the corresponding design value of unconfined
30 compression strength (except for case-F3), and a Poisson’s ratio
of 0.2. For case-F3, the amount of the cement used was between
40
the values of case-F1 and case-F2, but the qu value is obviously
50 lower. Since there are no field measured data available, and
Case L3: α ⫽ 11%, β ⫽ 70% (p ⫽ 200 kPa)
based on local experience, E ¼ 70 000 kPa was assumed and
60
used in calculations. The Zhu and Yin (1999) solution considers
70 linear variation of the total stress increment in a two-layer
(a) Ariake clay system as shown in Figure 16. Under embankment load, 0 , 1
and 2 can be approximately calculated by the Osterberg (1957)
Elapsed time: day method. Based on the test results using the undisturbed samples
0 10 20 30 40 50
0 from the sites (FNHO, 2003), the other parameters for calculat-
ing the degree of consolidation (U(t )) are listed in Table 6. cv
5 and kv are the values corresponding to the average consolidation
By proposed method pressure.
Settlement: mm

10 By Chai and Pongsivasathit (2010)

Measurement 7.2 Saga case histories
15 In 2007, in Saga, Japan, a full-scale test embankment was
Case L5: α ⫽ 30%, β ⫽ 70%
constructed on soft clay (Ariake clay) deposit improved by
20
floating columns. At the test site, the thickness of soft Ariake soil
is about 11 m. The embankment had a fill thickness of 6.5 m,
25
Case L4: α ⫽ 30%, β ⫽ 50% with a base dimension of 45.1 m 3 23.7 m, and top dimension of
30 33.4 m 3 12.0 m (Igaya et al., 2010). The cross-section of the
(b) Mixed soil embankment is shown in Figure 17.

Figure 12. Comparison of results on the settlement–time curves
7.1 Fukuoka case histories
The Fukuoka cases were described by Chai et al. (2009). The
cross-sections of the three cases are shown in Figures 13–15 and
the construction time as well as some of design parameters are
given in Table 5. The field measurements indicate that there were
settlements below the soft layers for these three cases. Since the
The columns were constructed with a diameter of about 1.2 m
and arranged in a square pattern. The spacing was about 1.9 m
which results in an Æ value of about 31%. The length of the
column was about 8.5 m ( ¼ 76%). The backfill used was
decomposed granite soil and the compacted backfill had a total
unit weight of approximately 18.2 kN/m3 : The construction time
was about 109 days. The parameters for calculating the degree of
consolidation (U(t )) are included in Table 6. For the value of
Young’s moduli of the column, E ¼ 100 000 kPa was adopted

15 m

·8
Sand mat 1:1 8m
0·5 m
S-1
Slab 0·5 m

HL ⫽ 6·5 m
H ⫽ 8·5 m

α ⫽ 21·7% S-2 Soft clayey silt

Stiff gravelly silt

Figure 13. Cross-section of case-F1

54
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

22 m

·8
1:1
Sand mat 8·3 m
0·5 m
S-1
Slab 2·5 m

HL ⫽ 5·5 m HL ⫽ 5·0 m
H ⫽ 6·5 m

α ⫽ 9·0% Soft clayey silt

S-2 Tuff clay and fine sand
(⫺9·7 m)

Figure 14. Cross-section of case-F2

22 m
·5
1: 0

Reinforced
10 m
Sand mat
0·5 m
S-1

HL ⫽ 5 m

H ⫽ 10·6 m
α ⫽ 30·0%

Soft clayey silt

S-2 Stiff gravelly clay
(⫺11·2 m)

Figure 15. Cross-section of case-F3

Case F1 F2 F3 F4

Unit weight of backfill, ªt : kN/m3 19 19 19 18.2

The design compressive strength of column, qu : kPa 700 700 400 600
The amount of the cement mixed with soil for column: kg/m3 140 100 130 150
Column length: m 6.5 5.5 5 8.5
Æ: % 21.7 9 30 31
: % 76 85 47 76
Thickness of slab: m 0.5 2.5 — —
The design compressive strength of slab, qu : kPa 300 300 — —
The amount of the cement mixed with soil for slab: kg/m3 80 80 — —
Time of construction: day 96 90 124 109
Final embankment thickness: m 8 8.3 10 6.5

Table 5. Some design and geometry parameters for case histories

55
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

Total stress increment

referring to the test results using the samples retrieved from the
σ0 field.

H1 Layer 1 7.3 Comparing the results

The comparisons of the settlement curves are given in Figure 18.
The calculated results using the Chai and Pongsivasathit (2010)
σ1 methods are also included in the figures for comparison. It can be
seen that the modified method yielded a better fit to the measured
H2 Layer 2 results. The lower predicted settlement for case-F1 by the
modified method is due to the consideration of the effect of p and
σ2 su : The h(ª) value is about 0.71. For case-F2, the value of h(ª) is
less than unity but the modified f (Æ) function results in a larger
Depth
value, and the overall effect on the Hc value is about the same as
Figure 16. Linear variation of the total stress increment the previous method. For case-F3, the h(ª) value is about 1.03.
As for case-F4, both the effects of h(ª) and the without-slab

Case Fukuoka cases Saga case F4

F1 F2 F3

H: m 8.5 6.5 10.6 11.2

H1c : m 6.23 4.92 4.68 8.30
H2c : m 2.12 1.44 5.63 2.74
0 : kN/m2 152.00 157.70 190.00 118.3
1 : kN/m2 146.96 153.23 186.95 105.10
2 : kN/m2 140.24 150.97 168.87 98.40
Æ: % 21.7 9.0 30.0 31.0
: % 76.5 84.6 47.2 76.0
su at the end of column: kN/m2 22.50 20.00 16.25 23.00
Calculated final settlement: m 0.289 0.279 0.575 0.310
Layer H1 cv1 : m2 /day 0.596 0.181 1.258 2.064
kv1 : 3104 m/day 3.387 2.347 5.210 6.050
Layer H2 cv2 :m2 /day 0.035 0.031 0.061 0.040
kv2 : 3104 m/day 2.668 2.022 1.844 2.320

Table 6. Parameters for calculating the degree of consolidation

12 m

·8
1:1 6·5 m
S1

HL ⫽ 8·5 m
H ⫽ 11·18 m

α ⫽ 30% Soft clay

S2
Sand

Figure 17. Cross-section of the test embankment in Saga, Japan

56
Ground Improvement Consolidation settlement of floating-
Volume 166 Issue GI1 column-improved soft clayey deposit
Pongsivasathit, Chai and Ding

Elapsed time: day Elapsed time: day

0 50 100 150 200 250 300 0 50 100 150 200 250 300
0 0

0·05 Fukuoka case (case-F1) 0·05 Fukuoka case (case-F2)

By proposed method By proposed method
0·10 By Chai and Pongsivasahit (2010) 0·10 By Chai and Pongsivasahit (2010)
Settlement: m

Settlement: m
Measurement Measurement
0·15 0·15

0·20 0·20

0·25 0·25

0·30 0·30

0·35 0·35
(a) (b)

Elapsed time: day Elapsed time: day

0 100 200 300 400 500 600 0 100 200 300 400 500 600
0 0

0·1 Fukuoka case (case-F3) 0·05 Saga case (case-F4)

By proposed method By proposed method
By Chai and Pongsivasahit (2010) 0·10 By Chai and Pongsivasahit (2010)
0·2
Settlement: m

Settlement: m

Measurement 0·15 Measurement

0·3
0·20
0·4
0·25
0·5
0·30
0·6 0·35
0·7 0·40
(c) (d)

Figure 18. Comparisons of settlement–time curve of case

histories

condition contributed to a smaller predicted settlement (Figure
18(d)) by the modified method.
Generally, the calculations resulted in a slower settlement rate at
the early stage for the Fukuoka cases. The possible reason is that
the cv values used correspond to the normally consolidated state,
but for the three cases initially the subsoil was in an over-
consolidated state.
8. Conclusions
Based on the laboratory model test and finite-element analysis
(FEA) results, a modified method for calculating the consolida-
tion settlement–time curve of floating soil-cement column im-
proved soft clayey soil deposit, with or without a surface cement-
stabilised slab system, has been proposed. The main modification
is in the method for determining the thickness (Hc ) of a part of
the column improved layer near the end of the column, which is
treated as an unimproved layer in settlement calculations. The
explicit equations for calculating the value of Hc have been
proposed as a function of area improvement ratio, depth improve-
ment ratio, load intensity and the undrained shear strength of the
soil.
The proposed method was applied to calculate the settlement–
time curves of the laboratory model tests and four field case
histories in Japan. Comparisons of the measured and calculated
results show that the proposed method yielded satisfactory
predictions. It is suggested that the proposed method can be used
to design the soft ground improvement using floating columns
with/without a cement-stabilised slab on the ground surface.
REFERENCES
Bergado DT, Chai JC, Alfaro MC and Balasubramaniam AS
(1994) Improvement Techniques of Soft Ground in Subsiding
and Lowland Environments. Balkema, Rotterdam, p. 222.
Brinkgreve RBJ (2002) Plaxis 2D. Scientific Manual V.8. Delft,
the Netherlands.
Broms BB and Boman PO (1979) Lime column: a new foundation
method. Journal of Geotechnical Engineering Division, ASCE
105(GT4): 539–556.
Chai JC and Pongsivasathit S (2010) A method for calculating
consolidation settlements of floating column improved clayey
subsoil. Frontiers of Architecture and Civil Engineering in
China 4(2): 241–251.
Chai JC, Miura N, Kirekawa T and Hino T (2009) Settlement

57
Ground Improvement
Volume 166 Issue GI1
prediction for soft ground improved by columns. Proceedings
of the Institution of Civil Engineers – Ground Improvement
163(2): 109–119.
FNHO (Fukuoka National Highway Office) (2003) Report on the
Field Monitoring of Test Embankments at Shaowa-Biraki,
Road No. 208, Fukuoka, Part 1. Kyushu Regional
Development Bureau, Ministry of Land, Infrastructure,
Transport and Tourism, Kyushu, Japan (in Japanese).
Hansbo S (1981) Consolidation of fine-grained soils by
prefabricated drains. Proceedings of the 10th International
Conference on Soil Mechanics and Foundation Engineering,
Stockholm, 3, 677–682.
Hino T, Taguchi T, Chai JC and Shimoyama S (2008) The Ariake Sea
coastal road project in the Saga lowlands: properties of soft
foundations and use of dredged clayey soil as an embankment
material. Proceedings of the International Symposium on
Lowland Technology 2008, Busan, Korea, pp. 467–472.
Igaya Y, Hino T and Chai JC (2010) Behavior of trial
embankments and the results of environmental monitoring for
the Ariake sea coastal road project. Proceedings of the 7th
58
Consolidation settlement of floating-
column-improved soft clayey deposit
Pongsivasathit, Chai and Ding
International Symposium on Lowland Technology, Saga,
Japan, 2010, 260–271.
Osterberg JO (1957) Influence values for vertical stresses in
semi-infinite mass due to embankment loading. Proceedings
of the 4th International Conference in Soil Mechanics and
Foundation Engineering, 1, 393.
Shen SL, Chai JC and Miura N (2001) Stress distribution in
composite ground of column–slab system under road
pavement. First Asian-Pacific Congress on Computational
Mechanics 2001. Elsevier Science, pp. 485–490.
Taguchi T, Hino T, Suzuki M and Yamamoto T (2007)
Consideration on applying of unconfined compressive
strength evaluation of soil-cement column cured under stress
to Saga lowland. Proceedings of the International Symposium
on Geotechnical Engineering, Ground Improvement and
Geosynthetics for Human Security and Environmental
Preservation, Bangkok, Thailand, pp. 659–667.
Zhu G and Yin JH (1999) Consolidation of double soil layers
under depth-dependent ramp load. Géotechnique 49(3): 415–
421.
WHAT DO YOU THINK?
To discuss this paper, please email up to 500 words to the
editor at journals@ice.org.uk. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as a
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and students.
Papers should be 2000–5000 words long (briefing papers
should be 1000–2000 words long), with adequate illustra-
tions and references. You can submit your paper online via
www.icevirtuallibrary.com/content/journals, where you
will also find detailed author guidelines.
Copyright of Ground Improvement: Proceedings of the Institution of Civil Engineers is the property of Thomas
Telford Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email articles for
individual use.