SOILS AND FOUNDATIONS

Japanese Geotechnical Society

Vol. 51, No. 2, 239251, Apr. 2011

STRENGTH DEVELOPMENT IN CEMENT ADMIXED BANGKOK CLAY:

LABORATORY AND FIELD INVESTIGATIONS
SUKSUN HORPIBULSKi), RUNGLAWAN RACHANii), APICHAT SUDDEEPONGiii) and AVIRUT CHINKULKIJNIWATiv)
ABSTRACT
The in-situ deep mixing technique has been established as an eective means to eect columnar inclusions into soft
Bangkok clay to enhance bearing capacity and reduce settlement. In this paper, an attempt is made to identify the critical factors governing the strength development in cement admixed Bangkok clay in both the laboratory and the eld. It
is found that clay-water/cement ratio, wc/C is the prime parameter controlling the laboratory strength development
when the liquidity index varies between 1 and 2. Based on this parameter and Abrams' law, the strength prediction
equation for various curing times and combinations of clay water content and cement content is proposed and veried.
This will help minimize the number of trials necessary to arrive at the quantity of cement to be admixed. Besides the
wc/C, the strength of deep mixing column is controlled by the execution and curing conditions. For low strength improvement (laboratory 28-day strength less than 1,500 kPa), the eld strength of the deep mixing columns, quf, made
up from both dry and wet mixing methods is higher than 0.6 times the laboratory strength, qul. The quf/qul ratios for the
wet mixing columns are generally higher than those for the dry mixing columns. This higher strength ratio is due to the
dissipation of the excess water in the column (consolidation) caused by the eld stress. The water to cement ratio,
W/C, of 1.0 is recommended for the wet mixing method of the soft Bangkok clay. A fast installation rate was shown
to provide high quality for low strength columns. Suggestions are made for improving the deep mixing of soft Bangkok clay, which are very useful both from economic and engineering viewpoints.
Key words: Bangkok clay, clay-water/cement ratio, curing stress, deep mixing technique, low-swelling clayey soil, unconned compression test (IGC: D6/H1/K6)
the role of montmorillonite. As such, it is possible that
some clayey soils can be classied as non-swelling or low
swelling types, even if the primary clay mineral in clay
fraction (2 mm) is montmorillonite (Horpibulsuk et al.,
2007).
One of the eective ground improvement techniques
for the soft Bangkok clay is in-situ deep mixing. Cement
is commonly used as a cementing agent since it is readily
available at a reasonable cost in Thailand. The resistance
to compression and consequent strength development of
the cement admixed clay increases with curing time. The
columnar inclusions in the soft ground transform all such
soft ground to composite ground.
The fundamental mechanical properties of cement admixed clays have been extensively investigated by Terashi
et al. (1979, 1980); Kawasaki et al. (1981), Kamon and
Bergado, (1992) and Horpibulsuk et al. (2004a, b), etc.
Even though many previous investigations (Nagaraj and
Miura, 1996; Uddin, 1994; Yin and Lai, 1998; and
others) have been focused on the eect of water content
and cement content on the strength development in ce-

INTRODUCTION
Bangkok clay is well-known as a soft clay with a high
water content close to its liquid limit. It has large potential for settlement with low inherent shear strength. This
clay is classied as non- to low swelling (Horpibulsuk et
al., 2007) as per free swelling test (Prakash and Sridharan, 2004). Its swelling potential increases with depth. Besides Bangkok clay, non- to low swelling soils are generally found in many lowlands, such as Ariake bay in Japan
(El-Shafei, 2001; Modmoltin, 2002). Data provided by
Tanaka et al. (2001) indicate that some marine clays
(Pusan, Singapore, Drammen, Louiseville clays) are classied as inactive and normal clays, and are thus non- to
low swelling. Even though the clay mineralogy of many
clayey soils is primarily montmorillonite, the montmorillonite in those soils might not be the dominant
parameter controlling the soil expansivity. Other
parameters, such as other clay minerals (kaolinite, and illite, etc), the non-clay fraction (0.002 mm) and pore
medium chemistry, can also play a great role, masking
i)
ii)
iii)
iv)

Professor, School of Civil Engineering, Suranaree University of Technology, Thailand (suksung.sut.ac.th).

Assistant Professor, Department of Civil Engineering, Mahanakorn University of Technology, Thailand.
Post Graduate Researcher, School of Civil Engineering, Suranaree University of Technology, Thailand.
Assistant Professor, Construction Technology Research Unit, School of Civil Engineering, Suranaree University of Technology, Thailand.
The manuscript for this paper was received for review on September 29, 2010; approved on December 7, 2010.
Written discussions on this paper should be submitted before November 1, 2011 to the Japanese Geotechnical Society, 4-38-2, Sengoku,
Bunkyo-ku, Tokyo 112-0011, Japan. Upon request the closing date may be extended one month.
239

240

HORPIBULSUK ET AL.

ment admixed clays, the combination eect of both water

content and cement content has never been well integrated. Moreover, no dosage methodologies based on rational criteria have been suggested like those which are used
in concrete technology, where the water/cement ratio
plays a fundamental role in the assessment of target
strength. In concrete technology, Abrams' law (Abrams,
1918) is broadly applied: for a given concrete with certain
content, age and curing conditions, the strength of the
hardened concrete is determined exclusively by the ratio
of free water content to the cement content in the mix.
Strength is independent of the absolute contents of free
water and cement content in the mix.
As an analogy, Horpibulsuk and Miura (2001), Horpibulsuk et al. (2005) and Miura et al. (2001) have identied the clay-water/cement ratio, wc/C as a parameter for
analyzing and assessing laboratory strength development
in cement admixed clays. It is dened as the ratio of clay
water content to cement content (both reckoned in percentage). While the clay water content reects the
microfabric of soft clay, the cement content inuences
the level of bonding of that fabric. Based on this
parameter and Abrams' law (Abrams, 1918), Horpibulsuk et al. (2003) have revealed that for a given set of cement admixed clays, the strength development depends
only on the clay-water/cement ratio, wc/C. They have
also introduced a generalized equation for predicting
laboratory strength development in cement admixed clays
at various water contents, cement contents, and curing
times.
In addition to laboratory studies, studies on the performance of composite ground are vital. Experimental investigations have been done by Miyake et al. (1991),
Hashizume et al. (1998) and Kitazume et al. (1999). Field
observations on the deep mixing column are also necessary to investigate the eld strength development in order to
provide an understanding of the dierence between eld
and laboratory strengths and consequently, estimates of
the optimal input of cement to achieve the target eld
strength. Horpibulsuk et al. (2004c) and Nishida et al.
(1996) have concluded that the strength dierence between eld and laboratory improvement is mainly caused
by the non-uniformity in mixing in-situ clay with cement.
Horpibulsuk et al. (2004c) have classied the deep mixing
column into low and high strength (cement content)
columns based on the diculty in eld mixing. They have
shown that for the low strength column (qul(28days)1500
kPa, where qul(28days) is the laboratory 28-day strength),
both fast and slow installation rates yield practically the
same quality. In the case of the high strength column,
however, a low installation rate of less than 0.7 m/min is
required. Its eld to laboratory strength ratio is about
0.33 to 0.67, which is much lower than that of the low
strength column.
In practice, many laboratory trial mixes are needed to
arrive at a proper strength before the execution of the
deep mixing column. This laboratory strength must be
high enough to compensate for execution and curing conditions. At the service time, the eld strength must meet

the designed strength. To facilitate the determination of

the proper quantity of cement to be admixed, which compensates for strength reduction in the eld, a geotechnical
engineer needs a simple and rotational method to assess
laboratory and eld strengths with time for various combinations of water content and cement content by minimum laboratory trials. In this paper, we analyze the
strength development in the cement admixed Bangkok
clay in both laboratory and eld improvements. The
framework of the analysis of the laboratory strength development is the clay-water/cement ratio hypothesis
(Horpibulsuk and Miura, 2001; Miura et al., 2001) and
the Abrams' law (Abrams, 1918). The eect of the execution and curing conditions on the eld strength development of the deep mixing columns is examined. Finally,
we make some suggestions to improve deep mixing for
soft Bangkok clay based on the laboratory and eld investigations. This method could also be applied to other
non- to low swelling clay deposits.
EXPERIMENTAL INVESTIGATION
This investigation consists of laboratory and eld studies to formulate a simple and rational method of assessing laboratory and eld strength development in the cement admixed Bangkok clay.

Laboratory Investigation
Since Bangkok clay is non- to low swelling, and the
swelling potential increases with depth (Horpibulsuk et
al., 2007), kaolin and Bangkok clays, with free swell ratios of 0.6 and 1.3, were used to represent non- and low
swelling clays, respectively. The free swell ratio, FSR, is
dened as the ratio of equilibrium sediment volume of
10-g oven-dried soil passing a 425 mm sieve in distilled
water (Vd) to that in carbon tetra chloride or kerosene
(Vk). Soil is classied as non-swelling and low swelling
when the free swell ratio is (less than or equal to) 1.0 and
between 1.0 and 1.5, respectively (Prakash and Sridharan, 2004). Kaolin clay was obtained from a commercial
company. Its consistency limits were liquid limit, LL
43z and plastic limit, PL of 34z. Its specic gravity was
2.78. Bangkok clay was collected from Ladkrabang district, Bangkok, Thailand at a depth of 3 to 4 meters. Its
natural water content was 85z, the liquid and plastic
limits were in the order of 89 and 30z, and the specic
gravity was 2.71. The groundwater level was at about 1.0
m from the surface. The overconsolidation ratio was 1.2
and the eective strength parameters in triaxial compression were c?0 and q?229
. The chemical properties of
the two clays and the Type I Portland cement are shown
in Table 1. The grain size distribution curves for the two
clays and the cement are shown in Fig. 1. The specic
gravity of the cement was 3.15.
The main aims of the laboratory test are to analyze the
strength development of the cement admixed Bangkok
clay using the wc/C as a prime parameter and to develop
its generalized strength prediction equation. For the rst
aim, both Bangkok and kaolin clays were passed through

DEEP MIXING IN BANGKOK CLAY

241

Table 1. Chemical composition of the cement, kaolin clay, and Bangkok clay
Chemical
compounds

Portland cement
Type I

Kaolin clay

Soft Bangkok
clay

zSiO2
zAl2O3
zFe2O3
zMgO
zCaO
zNa2O
z K 2O
zSO3

20.90
4.76
3.41
1.25
65.41
0.24
0.35
2.71

59.79
31.84
1.59

3.05
0.05

63.83
21.34
8.41
1.54
0.94
0.28
2.45
1.22

Fig. 1. Grain size distribution of Bangkok and kaolin clays and Portland cement

a 2-mm sieve to remove any shell pieces and other larger

particles. The water content was adjusted to a range of
liquidity indices (LI), i.e., 1.0, 1.5, and 2.0. The liquidity
index has been used in this investigation as an indicator to
refer the initial water content of the clays in relation to
their plasticity characteristics before the cement is admixed. This intentional increase in water content is to simulate water content increase taking place in the wet method
of dispensing cement admixture in deep mixing. Both
clays with their water content corresponding to the above
levels of LI were thoroughly mixed with the cement at
dierent cement contents to attain the target wc/C. The
target wc/C values were 3.5 and 6.5 for kaolin clay and 4,
7, 12, and 14 for Bangkok clay. The mixing time was arbitrarily xed at 10 min as recommended by Miura et al.
(2001). This uniform paste was transferred to cylindrical
containers 50 mm in diameter and 100 mm in height as
well as to oedometer rings of 50 mm in diameter and 20
mm in height, taking care to prevent any air entrapment.
After 24 hours, the cylindrical samples were dismantled.
All the cylindrical samples and the oedometer samples
with the rings were wrapped in vinyl bags and they were
stored in a humidity room of constant temperature (20
29C). After 7 and 28 days of curing, the cement admixed
kaolin and Bangkok clays were taken for unconned
compression (UC) test. The rate of vertical displacement
in UC tests was 1 mm/min. Consolidation tests were conducted after 28 days of curing.

Fig. 2.

Soil prole of the test sites

To achieve our second aim, the two clays were used to

develop a simple and rational method of predicting
laboratory strength development. The clays at LI1 to 2
were admixed with 10 to 30z cement. After 7, 14, 28, 60,
90, and 120 days, the UC tests were conducted on the admixed samples. Based on the analysis of the test results,
the strength prediction equation for the cement admixed
Bangkok clay was developed. The trial mix results for the
eld investigation were used to verify the developed
strength prediction equation.

Field Investigation
The eld study investigates the strength reduction due
to several eld factors such as non-uniformity in mixing
in-situ clay with cement, and the dierence in curing conditions between laboratory and eld improvement. Deep
mixing was performed in the Sukhaphiban 3, and Ladkrabang districts in Bangkok, and the Bangpee district in
Sumutphakarn province, Thailand. The soil proles for
the three test sites are presented in Fig. 2. It was found
that the uppermost soil of about 2 meters in thickness,
was weathered zone. The soft Bangkok clay was more

242

HORPIBULSUK ET AL.

than 13 meters thick for the three sites. The maximum

thickness was found at the Bangpee district since it is
close to the Gulf of Thailand. Underlying this soft clay,
medium to sti clay was found. The medium to sti clay
was deposited above the rst sand layer (1st sand) overlying hard clay (Horpibulsuk et al., 2007). The natural
water content of the soft Bangkok clay is close to liquid
limit (slightly higher than liquid limit) near the ground
surface, and tends to decrease with depth. There was no
signicant dierence in the specic gravity of the soft
Bangkok clay with depth for the three sites: it was about
2.670.02. The yield stress increased with depth, and the
OCR was close to 1.0.
The tested deep mixing columns were made up from
both the dry and wet methods. The input of cement was
obtained from a trial mix to attain a 28-day strength of
about 1000 and 900 kPa for Sukhaphiban 3 and Ladkrabang districts, respectively. These tested deep mixing
columns were thus classied as low strength as per Horpibulsuk et al. (2004c). Tables 2 to 4 show the laboratory
strength test results from the trial mix at both sites for
dierent depths together with the application of the
proposed strength prediction equation. The cement contents are 150 and 185 kg/m3 for the dry and the wet mixing methods, respectively at Sukhaphiban 3 district. It is
175 kg/m3 at Ladkrabang district for the wet mixing
method. In the prediction, wc is the total water content in
the clay while being mixed with cement. For dry mixing,
it is equal to the natural water content, wn but for wet
mixing, it is the sum of the natural water content and the
additional water content from the cement slurry. The cement content, C, is dened as the ratio of weight of cement to dry weight of soil. Figure 3 shows the mixing
machines for both the dry and the wet mixing methods.
The dry mixing columns were made up with the penetration rate (PR) of 1.0 m/min, withdrawal rate (WR) of 4.0
m/min, and a rate of rotation of 140 rpm. They were executed only at the Sukhaphiban 3 district. In the wet mixing method, the wing rotation was 40 rpm for all three
test sites. For the Sukhaphiban 3 and Ladkrabang districts, both the PR and WR were 1.0 m/min and the
water to cement ratio (W/C) was 1.0. The eect of W/C,
the installation rate, and the cement content on the eld
strength development of the wet mixing columns were investigated at the Bangpee district. The test results were
compared with those previously reported by Horpibulsuk
et al. (2004c) for Ariake clay in order to provide a better
understanding of eld strength development. All the tested dry and wet mixing columns of 0.6 m diameter and 12
m length were installed with 1.5 m spacing at the three
sites. After 3, 14, and 28 days of curing, the cored samples were taken by a coring machine. These samples were
trimmed to a diameter of 50 mm and a height of 100 mm,
which are the same as those prepared in the laboratory.
Since the samples are hard and carefully cored and
trimmed, the eect of sample disturbance on the strength
can be neglected. The eld curing water content was also
measured to compare with the laboratory one.

Fig. 3.

mixing machines; a) dry method and b) wet method

LABORATORY TEST RESULTS

Figures 4 and 5 show the role of wc/C on the strength
characteristics of cement admixed kaolin and Bangkok
clays. They show the typical stressstrain relationships
in unconned compression tests of samples with dierent
initial water contents and dierent levels of cementing
agent but the same wc/C values. It is noted that the lower
the wc/C, the higher the cementation bond strength,
which leads to higher strength. Similar stressstrain behavior of all the admixed samples, which had the same
wc/C, was found.
Figure 6 presents the compression behavior of the cement admixed Bangkok clay samples, all with a wc/C
value of 12 but with dierent combinations of water content and cement content after 28 days of curing. The yield
stress was obtained as the intersection point of the two
straight lines which extend from the linear parts on either
end of the compression curve plotted as log (1e) against
log s?v (Buttereld, 1979; Sridharan et al., 1991). The
clay-cement mixtures were made up from three clay water
contents; namely, 89z, 119z, and 148z. The (ev, log
s?v) relationship is plotted so as to take care of the eect
of the dierence in the void ratio for the eective vertical
stresses less than the yield stress. In this range of the eective vertical stress, the cementation component is the

DEEP MIXING IN BANGKOK CLAY

243

Fig. 4. Unconned compression test results of kaolin samples, having

the same wc/C

Fig. 6. Compression behavior of cement admixed Bangkok clay samples for wc/C12 after 28 days of curing

Fig. 5. Unconned compression test results of Bangkok clay samples,

having the same wc/C

dominant factor which resists compression. It is found

that the yield stress and the deformation behavior at preyield stress of all samples with identical wc/C are practically the same. The samples with higher clay water content, however, are stable at higher void ratios and provide
higher compression index beyond yield stress. This is due
to the break-up of the cementation bond (Miura et al.,
2001, Horpibulsuk et al., 2010), which is similar to the
behavior of naturally cemented clay. This implies that the
cement admixed sample with higher clay water content
undergoes higher settlement at the post-yield state. The
role of cement admixture is to increase the yield stress,
resulting in an increase in the yield surface and the failure
envelope.
From the strength and the compression test results, it
was found that when the clay water contents are in the
range of LI1 to 2, the yield stress and the strength are
practically the same as long as the wc/C is identical. As
such, it is logical to relate the yield stress and the unconned compressive strength, as successfully done by Horpibulsuk et al. (2004a). The eect of the clay water content plays a dominant role on compressibility at the postyield state in which the cementation bond is broken

down. The lower the wc/C, the greater the strength and
the yield stress. This nding is the same as that reported
by Horpibulsuk et al. (2005) and Miura et al. (2001) for
the cement admixed Ariake clay. As such, it is possible to
adopt the wc/C and Abrams' law to analyze the strength
development in the cement admixed Bangkok clay.
Figures 7 and 8 show an analysis of laboratory strength
development in the cement admixed kaolin and Bangkok
clays for dierent curing times. It is found that the wc/C
can be applied to analyze the strength development of cement admixed clay with dierent cement content and
water content at a specic curing time. The test data are
represented well by a power function in the form:

qu

A
(wc/C )B

(1)

where qu is the unconned compressive strength, A and B

are empirical constants. The test data can also be
represented by the exponential function as done by Horpibulsuk (2001) and Horpibulsuk et al. (2003) for various
clays. However, it is found from this study that for cement admixed Bangkok clay, the power function is the
most appropriate since it has the highest degree of correlation. This power function was successfully used to assess the strength development of compacted cement stabilized coarse-grained soils by Horpibulsuk et al. (2006). In
all cases, the parameter A varies widely depending upon
soil type and curing time. However, the parameter B only

HORPIBULSUK ET AL.

244

Fig. 7. Analysis of strength development in cement admixed kaolin

clay using wc/C

Fig. 9. Strength development with time for dierent low-swelling

clayey soils and their generalization

with a high degree of correlation (0.908):

Fig. 8. Analysis of strength development in cement admixed Bangkok
clay using wc/C

varies in a narrow band between 1.25 and 1.31, irrespective of the swelling potential and curing time considered.
The parameter B can thus be taken as a constant for both
cement admixed clays (non- to low-swelling clays) for 7 to
120 days of curing. When the parameter B is 1.27, the
result is the following relation:

q(w /C) A/(wc/C )B1

(wc/C )2

q(w /C) A/(wc/C )B2

(wc/C )1
c

1.27

(2)

where q(w /C) is the strength estimated at a clay-water/cement ratio of (wc/C )1, and q(w /C) is the strength value at
clay-water/cement ratio of (wc/C )2.
At a particular wc/C, the strength development with
time is controlled only by the value of A since B is regarded as constant. Even though the parameter A depends on clay type, the rate of strength development with
time is identical for various admixed clays since it is
predominantly inuenced by the hydration process (Horpibulsuk et al., 2003). As such, it is possible to generalize
the strength development using the 28-day strength, q28 of
cement admixed clays as a reference value (vide Fig. 9). A
linear regression analysis gives the following relationship
c

qD
0.0390.283 ln D
q28

(3)

where D is the curing time (days), and qD is the strength at

D days of curing. This normalization accounts for the
eects of dierence in clay type, water content, and cement content. It is valid for the range of curing time between 3 and 180 days in which the usual service time of
deep mixing column is within 30 days. This relationship is
close to that proposed for cement admixed Ariake clay
(Horpibulsuk et al., 2003) and for cement stabilized silty
clay (Horpibulsuk et al., 2009).
The generalized interrelationship among strength, curing time, and wc/C for predicting the laboratory strength
development in the cement admixed Bangkok clay at LI
1 to 2 for wc/C ranging from 2.5 to 15 is obtained by
combining Eqs. (2) and (3).

{ }
q(w /C)
q(w /C)
c

28

(wc/C )28
(wc/C )D

1.27

(0.0390.283 ln D)

(4)

where q(w /C) is the strength of the cement admixed Bangkok clay to be estimated at clay-water/cement ratio of (wc
/C ) after D days of curing, and q(w /C) is the strength of
the cement admixed Bangkok clay at clay-water/cement
ratio of (wc/C ) after 28 days of curing.
Tables 2 to 4 show that the predicted strengths are in
c

28

DEEP MIXING IN BANGKOK CLAY

Table 2.
Depth
(m.)
3

12

245

Trial mix for determination of input of cement for Sukraphiban 3 district for dry mixing method
Claywater/cement
ratio,
wc/C

Laboratory
strength,
qul (kPa)

Predicted
strength,
qup (kPa)

106.7
106.7
106.7

5.09
5.09
5.09

873
1167
1453

857
1142
Reference

1.83
2.17
0

24.44
24.44
24.44

106.7
106.7
106.7

4.37
4.37
4.37

1008
1378
1745

1042
1389
1735

3.40
0.78
0.54

7
14
28

27.93
27.93
27.93

106.7
106.7
106.7

3.82
3.82
3.82

1262
1628
1969

1235
1645
2056

2.13
1.10
4.42

250

7
14
28

34.92
34.92
34.92

106.7
106.7
106.7

3.06
3.06
3.06

1596
1996
2250

1639
2185
2730

2.70
9.42
21.34

150

7
14
28

21.45
21.45
21.45

114.5
114.5
114.5

5.34
5.34
5.34

803
985
1246

807
1076
1344

0.53
9.17
7.92

175

7
14
28

25.03
25.03
25.03

114.5
114.5
114.5

4.58
4.58
4.58

992
1181
1463

982
1308
1635

1.05
10.83
11.74

200

7
14
28

28.6
28.6
28.6

114.5
114.5
114.5

4.00
4.00
4.00

1224
1446
1694

1163
1550
1937

4.98
7.19
14.33

250

7
14
28

35.75
35.75
35.75

114.5
114.5
114.5

3.20
3.20
3.20

1676
1932
2168

1544
2058
2572

7.86
6.52
18.62

150

7
14
28

16.12
16.12
16.12

66.6
66.6
66.6

4.13
4.13
4.13

1022
1246
1549

1118
1490
1862

9.36
19.60
20.15

175

7
14

18.81
18.81

66.6
66.6

3.54
3.54

1209
1429

1360
1812

12.50
26.76

Input of
cement, C?
(kg/m3)

Curing
time,
(days)

Cement
content,
C (z)

150

7
14
28

20.95
20.95
20.95

175

7
14
28

200

Clay water
content,
w c (z )

Mean Absolute Percent Error, MAPE (MAPE

1
n

i 1

|qupqul|
100)
qul

|qupqul|
100
qul
z

8.24z

Remark: for dry mixing, the natural water content, wn and the clay water content, wc are the same.

very good agreement with the laboratory ones. In this

prediction, even though index properties of the soft clay
vary with depth, it is assumed that this variation insignicantly aects the strength prediction. As such, a
strength value for a particular wc/C at a particular depth
can be used as a reference. It is found that the error from
the prediction is acceptable for engineering practice with
the mean absolute percent error less than 9.6z for both
sites. This reinforces the application of the strength
prediction equation (Eq. (4)) for the cement admixed
Bangkok clay at various curing times and combinations
of water content and cement content.
FIELD TEST RESULTS
Figure 10 shows the eld 3-day curing water content
for the dry mixing, compared with the natural water content, wn, and the laboratory 3-day curing water content,
w(lab)3days. Figure 11 shows the eld 28-day curing water
content for the wet mixing, compared with the natural

water content, wn, the clay water content (after mixing

clay with water to attain W/C of 1.0), wc, and the laboratory 28-day curing water content, w(lab)28days. It is found
that for the whole depth, the w(lab)3days for the dry mixing is
lower than the wn and the w(lab)28days for the wet mixing is
lower than the wc. This reduction in water content is due
to cement hydration. It is of interest to mention that the
eld curing water contents for the dry mixing are close to
the w(lab)3days while the eld curing water contents of the
wet mixing are very much lower than the w(lab)28days. The
dierence between the eld and the laboratory curing
water contents for the wet mixing is due to the eld curing
stress. This shows that the inuence of the eld curing
stress is more for the wet mixing. This is a fact that immediately after dispensing cement slurry, the natural water
content suddenly increases. Such excess water must be
dissipated to reach the natural equilibrium state under the
existing overburden stress. Due to the dispensing pressure
and the wing rotation, cracks on the surrounding clay develop. These cracks accelerate the consolidation of the

HORPIBULSUK ET AL.

246

Table 3.
Depth
(m.)
3

Trial mix for determination of input of cement for Sukraphiban 3 district for wet mixing method

Input of
cement, C?
(kg/m3)

Curing
time,
(days)

Cement
content,
C (z)

Water
content,
w c (z )

Claywater/cement
ratio,
wc/C

Laboratory
strength,
qul (kPa)

Predicted
strength,
qup (kPa)

175

7
14
28

27.05
27.05
27.05

122.62
122.62
122.62

4.53
4.53
4.53

631
758
1053

596
794
992

5.56
4.77
5.75

185

7
14
28

28.78
28.78
28.78

124.00
124.00
124.00

4.31
4.31
4.31

703
874
1121

635
847
1058

9.61
3.11
5.60

200

7
14
28

31.40
31.40
31.40

126.07
126.07
126.07

4.02
4.02
4.02

793
920
1175

695
926
1158

12.35
0.69
1.49

175

7
14
28

24.63
24.63
24.63

108.33
108.33
108.33

4.40
4.40
4.40

565
722
999

619
825
1031

9.60
14.30
3.22

185

7
14
28

26.20
26.20
26.20

109.59
109.59
109.59

4.18
4.18
4.18

634
773
1119

660
879
Reference

4.08
13.76
0

200

7
14
28

28.58
28.58
28.58

111.48
111.48
111.48

3.90
3.90
3.90

668
918
1236

721
961
1201

7.95
4.68
2.85

175

7
14
28

20.87
20.87
20.87

89.60
89.60
89.60

4.29
4.29
4.29

552
658
807

638
851
1063

15.64
29.28
31.73

185

7
14
28

22.19
22.19
22.19

90.67
90.67
90.67

4.09
4.09
4.09

638
775
940

680
906
1132

6.52
16.86
20.40

200

7
14
28

24.19
24.19
24.19

92.28
92.28
92.28

3.82
3.82
3.82

705
900
1090

742
988
1235

5.19
9.81
13.30

Mean Absolute Percent Error, MAPE (MAPE

1
n

|qupqul|
100)
qul
i 1

|qupqul|
100
qul
z

9.56z

Remark: W/C1.0

column before the onset of the hydration (prior to cementation bonds formation) (Miura et al., 1998; Shen and
Miura, 1999; Shen et al., 2003; Shen et al., 2008). This
consolidation results in a signicant reduction in the eld
water content, as experimentally observed by Consoli et
al. (2006) and Rotta et al. (2003).
Figures 12 and 13 show the strength development in
the dry and the wet mixing columns at Sukhaphiban 3
district. Figure 14 shows the strength development in the
wet mixing columns at Ladkrabang district. For the same
input of content in kg per cubic meter of clay for the
whole depth, the weight of cement per dry weight of clay,
C, decreases with depth since water content decreases
(unit weight increases) with depth. The decrease in both
wc and C yield almost the same wc/C value, and therefore
practically the same laboratory strength, qul, for the
whole improved depth. The laboratory strengths shown
in Figs. 12 to 14 are thus the average values from dierent
depths. It is found that the eld strength is between 0.6
and 1.5 times the laboratory strength for the wet mixing
and between 0.6 and 1.0 times the laboratory strength for
the dry mixing. It must be kept in mind that the quf/qul

ratios vary wildly, depending on clay types, binder types,

executing machines, etc. Consequently, the quf/qul ratios
presented in this paper are valid only for the soft Bangkok clay and the employed machines. For the wet mixing,
the average eld strength is close to the laboratory
strength, while the average eld strength is close to the 0.7
times the laboratory strength for the dry mixing. For the
dry mixing, most of the eld strengths are lower than the
laboratory strengths for all curing times, possibly due to
non-uniformity of mixing in-situ clay with cement (Horpibulsuk et al., 2004c; Nishida et al., 1996). For the wet
mixing, even with the non-uniformity of the mixing, most
of the eld strengths are higher than the laboratory
strengths. This high eld strength (qufqul) is due to the
eld curing stress, resulting in the consolidation before
the onset of the hydration. This consolidation causes the
reduction in the eld void ratio (water content), which increases the strength. The eect of the reduction in the
void ratio due to the curing stress (consolidation) on the
strength development is experimentally depicted by Consoli et al. (2006) and Rotta, et al. (2003).
Generally, for a clay subjected to the same shear stress,

DEEP MIXING IN BANGKOK CLAY

Table 4.
Depth
(m.)
3

12

247

Trial mix for determination of input of cement for Ladkrabang district for wet mixing method

Input of
cement, C?
(kg/m3)

Curing
time,
(days)

Cement
content,
C (z)

Water
content,
w c (z )

Claywater/cement
ratio,
wc/C

Laboratory
strength,
qul (kPa)

Predicted
strength,
qup (kPa)

150

7
14
28

21.78
21.78
21.78

110.54
110.54
110.54

5.08
5.08
5.08

359
500
669

395
526
Reference

10.01
5.19
0

175

7
14
28

25.81
25.81
25.81

113.83
113.83
113.83

4.41
4.41
4.41

453
614
865

471
628
785

4.03
2.26
9.19

200

7
14
28

29.95
29.95
29.95

117.12
117.12
117.12

3.91
3.91
3.91

593
770
1010

549
732
915

7.31
4.96
9.47

250

7
14
28

38.57
38.57
38.57

123.70
123.70
123.70

3.21
3.21
3.21

762
940
1197

707
942
1177

7.24
0.14
1.66

150

7
14
28

21.16
21.16
21.16

100.34
100.34
100.34

4.74
4.74
4.74

439
556
758

430
573
716

2.07
3.09
5.53

175

7
14
28

25.08
25.08
25.08

103.53
103.53
103.53

4.13
4.13
4.13

517
673
928

513
684
854

0.75
1.51
7.94

200

7
14
28

29.12
29.12
29.12

106.72
106.72
106.72

3.67
3.67
3.67

596
808
1061

596
795
993

0.04
1.68
6.35

250

7
14
28

37.52
37.52
37.52

113.10
113.10
113.10

3.01
3.01
3.01

743
959
1261

764
1019
1273

2.92
6.25
0.96

150

7
14
28

18.69
18.69
18.69

95.66
95.66
95.66

5.12
5.12
5.12

463
601
792

390
520
650

15.72
13.41
17.89

175

7
14
28

22.13
22.13
22.13

98.51
98.51
98.51

4.45
4.45
4.45

544
712
946

466
621
776

14.32
12.74
18.00

200

7
14
28

25.65
25.65
25.65

101.35
101.35
101.35

3.95
3.95
3.95

646
836
1101

542
722
903

16.10
13.58
18.00

250

7
14
28

32.97
32.97
32.97

107.04
107.04
107.04

3.25
3.25
3.25

788
1051
1348

696
927
1159

11.73
11.78
14.03

Mean Absolute Percent Error, MAPE (MAPE

1
n

i 1

|qupqul|
100)
qul

|qupqul|
100
qul
z

7.72z

Remark: W/C1.0

its excess pore water pressure development decreases with

the increase in eective stress (shear strength). For the
soft Bangkok clay deposit where natural water content
decreases and undrained shear strength increases with
depth, the excess pore water pressure development due to
wet mixing in the soft Bangkok clay decreases with depth.
Hence, the eld strength development due to the consolidation (dissipation of excess pore pressure) decreases
with depth. In other words, the quf/qul value for the wet
mixing tends to decrease with depth as shown in Figs. 13
and 14.
The eect of water to cement ratio (W/C) on the eld
strength development of the wet mixing columns is shown

in Fig. 15. The strength development at three levels of

water to cement ratio (W/C) is shown. The W/C of 1.0,
which is commonly used for soft Bangkok clay, gives the
highest strength. Hence, a W/C of 1.0 is recommended
for the wet mixing method based on the range of W/C
tested.
The eect of the installation rate on the strength development in the low strength columns is shown in Fig.
16. The strength proles were almost the same for dierent installation rates. This result is in agreement with that
reported by Horpibulsuk et al. (2004c), and is typical for
low strength columns. This means that both slow and fast
installation rates result in practically the same quality

248

HORPIBULSUK ET AL.

Fig. 13. Strength prole for wet mixing columns at Sukhaphiban 3

district
Fig. 10. Field curing water content of dry mixing columns at Sukhaphiban 3 district

Fig. 14.

Strength prole of wet mixing columns at Ladkrabang district

Fig. 11. Field curing water content of wet mixing columns at Sukhaphiban 3 district

Fig. 15. Eect of water to cement ratio, W/C, on the strength development of wet mixing columns
Fig. 12. Strength prole of dry mixing columns at Sukhaphiban 3 district

columns (quf/qul0.6 to 1.5). The same is not for the high

strength columns where the reduction in water content
due to hydration has a remarkable eect on the eectiveness of the mixing: a very low installation rate is required

in this case.
Figure 17 shows the strength development of the high
strength columns (C220 kg/m3) compared with that of
the low strength column (C220 kg/m3) for the same
W/C and installation rate (PRWR1.0 m/min). It is
clearly noted that the strength of all the wet mixing

DEEP MIXING IN BANGKOK CLAY

Fig. 16. Eect of installation rate on the strength development of wet

mixing columns

249

Fig. 18. Suggested procedure of wet mixing method for soft Bangkok
clay

columns than a few high strength columns.

SUGGESTED METHOD FOR DEEP MIXING
IMPROVEMENT

Fig. 17. Eect of cement content on the strength development of wet

mixing columns

columns made up from dierent cement contents is practically the same. In other words, the strength is irrespective of the cement content for such a high installation
rate. This result is the same as that reported by Horpibulsuk et al. (2004c) for improvement with high cement contents with installation rates close to and higher than 1.0
m/min. As such, it is not advantageous or economical to
opt for the high cement content columns particularly at
high installation rate. Horpibulsuk et al. (2004c) concluded that an enhancement of strength does occur when the
installation rate is less than 0.7 m/min. The higher the cement content, the greater the reduction in water due to
larger hydration, making mixing dicult.
From this study and the work of Horpibulsuk et al.
(2004c), wet mixing is suitable for executing low strength
columns since the fast installation rate can be adopted
and the quf/qul is high. For high strength columns, high
input of cement is required with the very low installation
rate. The quf/qul value is about 0.330.67, which is much
lower than that of low strength columns (Horpibulsuk et
al., 2004c). As such, for a particular design (dead and
live) load on the soft Bangkok clay, it might be more economical to improve soft ground with many low strength

Based on the laboratory and eld study, our suggestions for improving deep mixing for the soft Bangkok
clay are summarized and presented (vide Fig. 18). The
method below is for low strength columns (qul(28days)
1500 kPa) by the wet mixing method.
Determination of input of cement compensating for eld
factors
1. From the soil prole and the design load on the soft
ground, determine the diameter, length, spacing
and eld strength at the service time of the deep
mixing column.
2. From the target eld strength at service time (open
square symbol), estimate the target eld strength at
7 days of curing (Target quf(7days)) (cross symbol),
which can be approximated using Eq. (3).
3. Determine the laboratory strength at 7 days of curing (qul(7days)), using the eld strength reduction of
1.7 (quf/qul0.6) (black circle symbol).
4. Determine the cement content to attain the laboratory strength at 7 days of curing (qul(7days)) and service time. This task can simply be done using Eq. (4).
Field execution and examination of eld strength
5. Execute the deep mixing column with a W/C of 1.0
and an installation rate (PR and WR) of about 1.0
m/min.
6. Core the soil-cement samples from a selected
column at dierent depths to determine the eld
strength at 7 days of curing for every section (every
3000 column).
7. If quf(7days)0.6 qul(7days), then the tested section meets
the requirement.
8. In the case that quf(7days)0.6 qul(7days),
8.1 If the actual quf(7days) is slightly lower than the
target quf(7days) (black square), the service time of
the section should be postponed to increase curing time.
8.2 If the actual quf(7days) is much lower than target

HORPIBULSUK ET AL.

250

quf(7days), conduct a pile load test.

8.2.1 If the measured load capacity is twice
higher than the design load capacity (factor of safety, FS2.0), the requirement
is met.
8.2.2 If FS2.0, the section must be
redesigned and reimproved to compensate for the strength loss.
CONCLUSIONS
This paper deals with the laboratory and eld strength
development in cement admixed Bangkok clay. An analysis of the laboratory strength development based on the
clay-water/cement ratio hypothesis and the Abrams' law
is carried out. The eld strength development in the deep
mixing column is studied. The following conclusions can
be drawn.
1. For the cement stabilization of soft Bangkok clay, in
which its water content varies in the range of liquidity
index between 1 and 2, the wc/C is the prime
parameter governing the strength and the compressibility at the pre-yield state. The cementation bond
strength increases as the clay-water/cement ratio,
wc/C, decreases.
2. Based on the clay-water/cement ratio and the
Abrams' law, we proposed a relationship between the
clay-water/cement ratio, the curing time and the
strength for the cement admixed Bangkok clay. This
proposed relationship is useful in estimating the
laboratory strength of the cement admixed Bangkok
clay wherein the water content and cement content
vary over a wide range by using the test results of a single trial. The formulation of the proposed relationship
is on sound principles. It is possibly applicable to
other non- to low-swelling clays, in which a constant B
of 1.27 is used. For other medium to high swelling
clays, this constant can be further rened with the
analysis of more data generated for this specic purpose.
3. The eld to laboratory strength ratio (quf/qul) is higher
than 0.6 for both the dry and wet mixing methods.
The average strength ratio is about 0.7 and 1.0 for the
dry and the wet mixing methods, respectively. The
consolidation of the wet mixing column before the onset of the hydration leads to an increase in the eld
strength.
4. From the eld test results, a water to cement ratio,
W/C, of 1.0 is recommended for the wet mixing improvement of soft Bangkok clay. For low strength
columns (qul1500 kPa), a high installation rate of
about 1.0 m/min should be adopted to execute the
quality column.
5. The suggested procedure for the wet mixing improvement for soft Bangkok clay is useful from both an engineering and economical viewpoint. The procedure
can save on sampling and laboratory testing and therefore cost. It can be applied to the other non- to low
swelling soft clay deposits.

ACKNOWLEDGEMENT
The authors would like to acknowledge the nancial
support provided by the Suranaree University of Technology. The eld test results of the soil-cement columns
obtained from the Bureau of Materials Analysis and Inspection, Department of Highways, Thailand are appreciated.
LIST OF NOTATIONS

Aempirical constant
Bempirical constant
Ccement content in terms of percentage
C?input of cement in terms of kg/m3
quunconned compressive strength
qufeld strength of the column
quf(7days)eld strength of the column at 7 days of curing
qullaboratory strength
qul(28days)laboratory 28-day strength
q(w /C) strength to be estimated at clay-water/cement
ratio of (wc/C )1
q(w /C) strength value at clay-water/cement ratio of
(wc/C )2
q(w /C) strength of the cement admixed Bangkok clay
to be estimated at clay-water/cement ratio after D days of curing
q(w /C) strength of the cement admixed Bangkok clay
clay-water/cement ratio after 28 days of curing
wnnatural water content
w(lab)3dayslaboratory 3-day curing water content
w(lab)28dayslaboratory 28-day curing water content
wc/Cclay-water/cement ratio
W/Cwater to cement ratio
FSfactor of safety
c

28

REFERENCES
1) Abrams, D. A. (1918): Design of concrete mixtures, Structural
Materials Research Material, Bulletin 1, Lewis Institute, Chicago.
2) Buttereld, R. (1979): A natural compression law for soils (an advance on e-log p?), Geotechnique, 29(4), 469480.
3) Consoli, N. C., Rotta, G. V. and Prjetto, P. D. M. (2006): Yielding-compressibility-strength relationship for an articially cemented soil cured under stress, Geotechnique, 56(1), 6972.
4) El-Shafei, A. (2001): Geo-environmental and geotechnical assessments of nitrate pollute soil, Ph.D. Dissertation, Saga University,
Saga, Japan.
5) Hashizume, H., Okochi, Y., Dong, J., Horii, N., Toyosawa, Y.
and Tamate, S. (1998): Study on the behavior of soft ground improved using Deep Mixing Method, Proceedings of International
Conference on Centrifuge, 98, 851856.
6) Horpibulsuk, S. (2001): Analysis and Assessment of Engineering
Behavior of Cement Stabilized Clays, Ph.D. Dissertation, Saga
University, Saga, Japan.
7) Horpibulsuk, S. and Miura, N. (2001): A new approach for
studying behavior of cement stabilized clays, Proc. 15th International Conference on Soil Mechanics and Geotechnical Engineering
(ISSMGE), Istanbul, Turkey, 3, 17591762.
8) Horpibulsuk, S., Miura, N. and Nagaraj, T. S. (2003): Assessment

DEEP MIXING IN BANGKOK CLAY

9)

10)

11)

12)

13)

14)

15)

16)

17)

18)

19)

20)

21)

22)

of strength development in cement-admixed high water content

clays with Abrams' law as a basis, Geotechnique, 53(4), 439444.
Horpibulsuk, S., Bergado, D. T. and Lorenzo, G. A. (2004a):
Compressibility of cement admixed clays at high water content,
Geotechnique, 54(2), 151154
Horpibulsuk, S., Miura, N. and Bergado, D. T. (2004b): Undrained shear behavior of cement admixed clay at high water content, Journal of Geotechnical and Geoenvironmental Engineering,
ASCE, 130(10), 10961105.
Horpibulsuk, S., Miura, N., Koga, H. and Nagaraj, T. S. (2004c):
Analysis of strength development in deep mixingA eld study,
Ground Improvement Journal, 8(2), 5968.
Horpibulsuk, S., Miura, N. and Nagaraj, T. S. (2005): Claywater/cement ratio identity of cement admixed soft clay, Journal
of Geotechnical and Geoenvironmental Engineering, ASCE,
131(2), 187192.
Horpibulsuk, S., Katkan, W., Sirilerdwattana, W. and Rachan, R.
(2006): Strength development in cement stabilized low plasticity
and coarse grained soils: Laboratory and eld study, Soils and
Foundations, 46(3), 351366.
Horpibulsuk, S., Shibuya, S., Fuenkajorn, K. and Katkan, W.
(2007): Assessment of engineering properties of Bangkok clay,
Canadian Geotechnical Journal, 44(2), 173187.
Horpibulsuk, S., Rachan, R. and Raksachon, Y. (2009): Role of y
ash on strength and microstructure development in blended cement
stabilized silty clay, Soils and Foundations, 49(1), 8598.
Horpibulsuk, S., Liu, M. D., Liyanapathirana, D. S. and Suebsuk,
J. (2010): Behavior of cemented clay simulated via the theoretical
framework of the Structured Cam Clay model, Computers and
Geotechnics, 37, 19.
Kamon, M. and Bergado, D. T. (1992): Ground improvement techniques, Proc. 9th Asian Regional Conference on Soil Mechanics
and Foundation Engineering, 2, 526546.
Kawasaki, T., Niina, A., Saitoh, S., Suzuki, Y. and Honjo, Y.
(1981): Deep mixing method using cement hardening agent, Proc.
10th International Conference on Soil Mechanics and Foundation
Engineering, Stockholm, 721724.
Kitazume, M., Yamamoto, M. and Udaka, Y. (1999): Vertical
bearing capacity of column type DMM ground with low improvement ratio, Proc. International Conference on Dry Mix Methods
for Deep Soil Stabilization, 245250.
Miura, N., Shen S. L., Koga, K. and Nakamura, R. (1998):
Strength change of the clay in the vicinity of soil cement column,
Journal of Geotechnical Engineering, JSCE, No. 596/III43,
209221 (in Japanese with English abstract).
Miura, N., Horpibulsuk, S. and Nagaraj, T. S. (2001): Engineering
behavior of cement stabilized clay at high water content, Soils and
Foundations, 41(5), 3345.
Miyake, M., Akamura, H. and Wada, M. (1991): Deformation

23)

24)

25)

26)

27)

28)

29)

30)

31)

32)

33)

34)

35)

36)

251

characteristics of ground improved by a group of treated soil, Proc.

International Conference on Centrifuge, 295302.
Modmoltin, C. (2002): Mechanical properties and microstructure
of lime and cement stabilized Ariake clays, Ph.D. Dissertation,
Saga University, Japan.
Nagaraj, T. S. and Miura, N. (1996): Induced cementation of soft
groundA parametric assessment, International Symposium on
Lowland Technology, Institute of Lowland Technology, Saga University, Saga, Japan, 8597.
Nishida, K., Koga, Y. and Miura, N. (1996): Energy consideration
of the dry jet mixing method, Proc. 2nd International Conference
on Ground Improvement Geosystems, IS-Tokyo '96, 1, 643748.
Prakash, K. and Sridharan, A. (2004): Free swell ratio and clay
mineralogy of ne-grained soils, Geotechnical Testing Journal,
ASTM, 27(2), 220225.
Rotta, G. V., Consoli, N. C., Prietto, P. D. M., Coop, M. R. and
Graham, J. (2003): Isotropic yielding in an articially cemented soil
cured under stress, Geotechnique, 53(5), 493501.
Shen, S. L. and Miura, N. (1999): Soil fracturing of the surrounding clay during deep mixing column installation, Soils and Foundations, 39(5), 1322.
Shen, S. L., Miura, N. and Koga, H. (2003): Interaction mechanism between deep mixing column and surrounding clay during installation, Canadian Geotechnical Journal, 40, 293307.
Shen, S. L, Han, J. and Du, Y. J. (2008): Deep mixing induced
property changes in surrounding sensitivity marine clays, Journal
of Geotechnical and Geoenvironmental Engineering, ASCE,
134(6), 845854.
Sridharan, A., Abraham, B. M. and Jose, B. T. (1991): Improved
technique for estimation of preconsolidation pressure, Geotechnique, 41(2), 263268.
Tanaka, T, Mishima, O., Tanaka, M., Park, S. Z., Jeong, G. H.
and Locat, J. (2001): Characterization of Yangsan clay, Pusan,
Korea, Soils and Foundations, 41(2), 89104.
Terashi, M., Tanaka, H. and Okumura, T. (1979): Engineering
properties of lime treated marine soils and DMM, Proc. 6th Asian
Regional Conference on Soil Mechanics and Foundation Engineering, 1, 191194.
Terashi, M., Tanaka, H., Mitsumoto, T., Niidome, Y. and Honma, S. (1980): Fundamental of lime and cement treated soils,
Report of Port and Harbour Research Institute, 19(1), 3362 (in
Japanese).
Uddin, K. (1994): Strength and Deformation Behavior of Cement
Treated Bangkok Clay, Doctoral Dissertation, Asian Institute of
Technology, Bangkok Thailand.
Yin, J. H. and Lai, C. K. (1998): Strength and stiness of Hong
Kong marine deposit mixed with cement, Geotechnical Engineering
Journal, 29(1), 2944.