Key Factors for Deep Cement Mixing Construction
for Undredged Offshore Land Reclamation
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
K. S. Yin 1; L. M. Zhang, F.ASCE 2; H. F. Zou 3; H. Y. Luo 4; and W. J. Lu 5
Abstract: Deep cement mixing (DCM) is an environmentally friendly technique for offshore ground improvement without dredging or
much disturbance to the marine ecological system. Several field construction factors can influence the unconfined compressive strength
(UCS) of cement-stabilized soil. In this study, key construction factors are evaluated referring to site investigation records, construction
records, and quality test results of a large offshore DCM construction project. The key factors were attributed to geological conditions,
construction procedures, and curing conditions. Specific field construction factors include fluctuations of tidal level, original soil type, volume fraction of injected water, volume fraction of injected cement slurry, injection rate of cement slurry, penetrating and mixing time per
meter, curing age, and moisture content. The importance of these construction factors on the DCM strength has been quantified using a
statistical method based on construction records. The injected water volume and original soil type are noted to be the two most dominant
factors on the UCS of the treated soils. Longer mixing time improved the strength of the treated soils. DOI: 10.1061/(ASCE)GT.19435606.0002848. © 2022 American Society of Civil Engineers.
Author keywords: Deep cement mixing (DCM); Ground improvement; Land reclamation; Offshore construction; Unconfined compressive
strength (UCS); Marine deposit.
Introduction
Land reclamation is one way to ease the land shortage problem in
coastal areas. Deep mixing is one of the most preferred techniques
for undredged land reclamation due to its capability to improve the
strength and stiffness of soft marine clays and avoid up-heave problems. This technique also minimizes disturbance to the natural
marine ecological system from dredging activities. The deep mixing technique was originated in Japan in the late 1960s and adopted
worldwide starting in the 1970s. The injected binders, either in the
form of dry powder or wet slurry containing cement or lime, are
mixed with in-situ soils during construction, forming columns or
clusters of cemented binder–soil mixtures. This study concerns
1
Postdoctoral Research Associate, Dept. of Civil and Environmental
Engineering, Hong Kong Univ. of Science and Technology (HKUST),
Kowloon, Hong Kong 999077, China. ORCID: https://orcid.org/0000
-0003-1350-4497
2
Chair Professor and Head, Dept. of Civil and Environmental Engineering, Hong Kong Univ. of Science and Technology, Kowloon, Hong Kong
999077, China; Professor, HKUST Shenzhen-Hong Kong Collaborative
Innovation Research Institute, Shenzhen 518000, China (corresponding
author). ORCID: https://orcid.org/0000-0001-7208-5515. Email: cezhang@
ust.hk
3
Assistant Engineer, AECOM Asia Limited Company, 138 Shatin Rural
Committee Rd., Hong Kong 999077, China.
4
Postdoctoral Research Associate, Dept. of Civil and Environmental
Engineering, Hong Kong Univ. of Science and Technology, Kowloon,
Hong Kong 999077, China. ORCID: https://orcid.org/0000-0002-2986
-9418
5
Postdoctoral Research Associate, Dept. of Civil and Environmental
Engineering, Hong Kong Univ. of Science and Technology, Kowloon,
Hong Kong 999077, China. ORCID: https://orcid.org/0000-0002-0526
-7471
Note. This manuscript was submitted on August 22, 2021; approved on
April 25, 2022; published online on June 7, 2022. Discussion period open
until November 7, 2022; separate discussions must be submitted for individual papers. This paper is part of the Journal of Geotechnical and
Geoenvironmental Engineering, © ASCE, ISSN 1090-0241.
© ASCE
deep cement mixing (DCM) widely used in offshore undredged
land reclamations. DCM improves the consistency, shear strength,
deformation characteristics, and permeability of natural marine
soils for practical engineering purposes, including bearing capacity
improvement, settlement reduction, seepage control, and excavation support. Famous land reclamation projects involving largescale DCM include the second-phase artificial island of Kansai
International Airport (Furudoi 2005) and the D-runway project
of Haneda International Airport (Watabe and Noguchi 2011).
Laboratory tests and numerical analyses have been performed to
evaluate the mechanical characteristics of DCM-improved soil considering numerous factors, including water content (Horpibulsuk
et al. 2012), stress concentration ratio (Jiang et al. 2013), column
diameter, length and distribution (Liu et al. 2012), column stiffness
(Yapage et al. 2014), column material (Abusharar et al. 2009), column permeability (Yin and Fang 2006), surrounding soil properties
(Rogers and Glendinning 1997), and penetration ratio (Yang et al.
2014). Other studies investigated key contributory factors to DCM
mixture strength, including soil type, water content, binder amount
and type, curing period, mixing time, and humidity (Kitazume
et al. 2015; Mohammadinia et al. 2019; Disfani et al. 2021;
Subramaniam and Banerjee 2020; Bellato et al. 2020). The unconfined compressive strength (UCS) of cement–soil mixtures exhibited significant spatial variation (Lee et al. 2006; Chen et al. 2011;
Wijerathna and Liyanapathirana 2018, 2019), with measured autocorrelation distances ranging from far less than a cluster’s diameter
to around 50 m (Honjo 1982; Larsson et al. 2005a, b; Navin 2005),
indicating the heterogeneous nature of DCM strength (Chen et al.
2016). Some empirical relations were proposed to predict the UCS
of cement–soil mixtures (Horpibulsuk et al. 2003; Lorenzo and
Bergado 2004; Lee et al. 2005).
Although the relationships between DCM strength and strengthimprovement factors have been investigated, most were based on
laboratory tests under controlled conditions, which could significantly deviate from conditions in the field. It is necessary to investigate and quantify the impacts of actual field conditions,
facilities, specifications, and techniques on DCM-treated offshore
04022063-1
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
(a)
DCM Construction
500
BH-5
Test DCM
Borehole
BH-4
Longitude distance (m)
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
soils. The significance level of impacts from the actual construction process may be evaluated based on the recorded construction data.
This study aimed to identify the mechanisms and key construction factors that influence DCM quality indicated by the homogeneity and magnitude of UCS. The construction process of a
practical offshore DCM project was investigated in close association with cement–soil chemical and physical reactions. The main
objective was to assess the impact of key construction factors on the
DCM strengths in dominant types of natural marine deposits referring to site investigation records, construction monitoring records
and quality assessment test results.
400
Site Condition
BH-3
300
BH-2
200
BH-6
BH-7
BH-1
100
BH-8
0
0
200
400
600
800
1000
1200
Latitude distance (m)
(b)
Fig. 1. Study site information: (a) soil stratification records; and
(b) plan view of test DCM clusters and boreholes.
Table 1. Properties of the marine deposit at study site
Properties
Range
Plastic limit (%)
Liquid limit (%)
Plasticity index (%)
Clay content
Silt content
Sand content
Gravel content
Void ratio at p00 ¼ 1 kPa
Compression index
Natural moisture content (%)
Saturated density (kN=m3 )
21–59
44–91
22–59
0.39–0.55
0.40–0.42
0.12–0.19
0.01–0.09
1.54–1.93
0.42–0.53
43–124
12.5–18.4
The study site was at the Pearl River estuary in the south China
Sea, where the stratigraphy of seabed soil can be divided into five
categories along depth: marine clay, alluvial crust, lower alluvium deposit, completely decomposed granite, and bedrock.
Subsoil profiles from eight boreholes at the site prior to DCM
construction are shown in Fig. 1(a). Soft silty clay and sandy–
silty clay dominated above −25 mPD, beneath which lay granular
deposits, including sand, silt, and gravel. To lower disturbances to
the marine ecological system, the seabed mud was not dredged
but improved with more than 4,000 DCM columns involving
630,000 m3 of marine soil. The properties of the marine deposits
prior to DCM construction are summarized in Table 1. The deposits
were mainly composed of clayey and silty soils with natural moisture content occasionally over 100%. The average depth of DCM
clusters was −30mPD, with the majority located within the clay
layer. Selected DCM clusters were cored and tested after construction for quality check. The layout of the test columns is shown in
Fig. 1(b).
DCM Construction Procedure
The offshore DCM was conducted using special barges equipped
with mixing units, leaders, pumping units, slurry agitators, binder
silos, power plants, and operation rooms, shown in Fig. 2(a). Wet
cement slurry instead of dry powder was applied as the DCM
cementing material in this project to achieve higher homogeneity
of the stabilized soil, more uniform strength along depth, and
smoother penetration in competent strata. The primary components
Fig. 2. Sketches of (a) DCM barge; (b) DCM mixing unit; and (c) arrangement of mixing shafts for a DCM cluster.
© ASCE
04022063-2
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
Table 2. Numbers of coring samples for quality testing
Curing age group
Dominant soil type
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
Clayey soils
Silty soils
Sandy soils
Gravelly soils
Total
Fig. 3. Pattern of installed DCM panels.
of the mixing unit are shown in Fig. 2(b). A single mixing cluster
involved four shafts with an overlapped area between two adjacent
clusters, as illustrated in Fig. 2(c).
The DCM clusters were designed as separate panels to enhance
the stability of the seawall in the tangent direction (Fig. 3). The plan
view of the DCM panels and their designed embedded depths, levels of installation and competent stratum, required unconfined compressive strengths (RUCSs), and cement slurry injection rates are
shown in Fig. 4. The design installation depth followed the tentative
level of competent strata. The cement was ordinary portland cement.
In the mix design, the water–cement ratio (w=c) of the cement slurry
was set at 0.8, which corresponded to a wet density of 1,641 kg=m3 .
The injected water during the penetration stage and natural soil water
were not included in the design w=c ratio. The cement dosage for
mixing cement slurry was approximately 260 kg=m3. The volume
fraction of injected cement slurry measures the cement content per
DCM cluster in practice.
The construction started by installing a geotextile layer and sand
blanket. Subsequently, the DCM barge was anchored at a designated location, followed by the penetration of the mixing units into
the marine soils with the mixing shafts rotating. During the penetration stage, the blades attached to the mixing shaft end broke
and disturbed the soil, reducing its strength, allowing the mixing
shafts to be driven by their self-weight. The average volume fraction of injected water during the penetration stage reached 12% for
28–40 days
40–60 days
60–80 days
1,071
48
57
0
1,176
623
28
39
32
722
446
38
17
16
517
a single cluster in this study. The penetration was terminated when
the mixing shafts reached the designed embedded depth within the
competent stratum. The bottom treatment was then conducted to
assure a sufficient shaft resistance of the DCM cluster in the competent soil layer. Finally, in the withdrawal stage, the mixing unit
mixed the cement slurry and the surrounding soil horizontally at a
fixed rotating speed, providing a cross-section of the DCM clusters,
as shown in Fig. 2(c). The cement slurry injection rate for bottom
treatment and withdrawal was set as equal, from 0.48 to 0.57 m3
per minute. The penetrating and withdrawal mixing rates were estimated at 0.45 and 0.28 m per minute.
DCM Quality Tests
After the completion of DCM construction, the stabilized cement–
soil mixture was sampled and tested for uniformity, continuity,
dimensions, and permeability. The verification of quality was conducted with full-depth coring and unconfined compression tests at
various curing ages, from 28 to 80 days. The detailed sample sizes
are summarized in Table 2. The measured results and frequency
statistics are shown in Fig. 5. The Kolmogorov–Smirnov test was
performed for normality checking, suggesting that the measurements do not strictly follow the normal distribution at confidence
level of 0.05. From the measured UCS values of cored DCM specimens, the design strength requirements were exceeded substantially.
The actual average UCS of 3.12 MPa was more than twice the design
value [Fig. 4(d)]. The interval of the natural soil moisture content
(Table 1) was also reduced from 0.43–1.24 to 0.036–0.956, demonstrating the effectiveness of DCM.
Fig. 4. Construction layout of DCM clusters in longitudinal direction: (a) design installation level; (b) design embedded depth; (c) tentative level of
competent stratum; (d) required unconfined compressive strength (RUCS); and (e) cement slurry injection rate.
© ASCE
04022063-3
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
0.16
Mean: 3.12 MPa
COV: 0.45
p-value (KS test): 1.31E-13
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0
(a)
2
4
6
8
UCS (MPa)
0.16
Mean: 45.95 %
COV: 0.27
p-value (KS test): 5.7E-6
Relative Frequency
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0
20
40
60
80
Moisture content (%)
(b)
0.16
0.14
Relative Frequency
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
Relative Frequency
0.14
0.12
Mean: 1639.43 kg/m3
COV: 0.046
p-value (KS test): 0.047
0.10
0.08
0.06
0.04
0.02
0.00
1400 1500 1600 1700 1800 1900 2000 2100
Density (kg/m3)
(c)
Fig. 5. Quality assurance test results from sampled DCM clusters: (a) unconfined compressive stress (UCS); (b) moisture content; and (c) density.
Identification of Construction Impact Factors
Micromechanism of Cement–Soil Strength Formation
Micromechanisms of strength formation of cement-stabilized soil
can refer to the chemical reactions (i.e. cement hydration, cation
exchange, and pozzolanic reactions) and physical contacts between
soil particles and cement slurry shown in Fig. 6. Cement hydration
is a collection of chemical reactions operating in series, primarily
the reaction between water and the cement calcium silicate with
tricalcium aluminate, which produce an alkaline environment with
calcium hydroxide, hydrated calcium silicate, and calcium aluminate gels that harden with time (Bullard et al. 2011). The amounts
of basic reactants, such as cement dosage, cement slurry quality,
and water volume, can influence the hydration reactions.
Negatively charged silicate particles from natural marine soils
are attached with some cations, including Naþ and Kþ . Equivalent
exchange can occur immediately between these positive ions and
calcium ions released from the calcium hydroxide, reducing water
adsorption of the clay particles. The increased electric charge decreases the repulsion between the negatively charged clay particles
and the thickness of the adsorbed water layer, gradually forming
© ASCE
flocculation. Factors that impact the cation exchange with flocculation during deep mixing include the natural soil type and water
content.
With the calcium ions continuously accumulating, the active calcium ions can react with the silica and alumina from the soil in an
alkaline environment, generating hydrated calcium silicate (CSH)
and hydrated calcium aluminate (CAH) cementitious matrices with
a fibrous crystal structure (Sherwood 1993). The exposed positive
ions continuously adsorb hydroxide ions, forming solidification
wrapping around the soil particle. This process is the so-called pozzolanic reaction, which is sensitive to soil type with different concentrations of silica, alumina, and water as reactants. Furthermore,
the long-period reaction in an alkali environment can be impacted
by disturbance to the alkalinity during DCM construction.
Among the previous chemical reactions, physical contact
[Fig. 6(b)] plays an important role. Various contact degrees of cement slurry with soil particles can occur during DCM construction.
When the soil particle surface was isolated from the adhesion of
other substances, the voids between them were filled with cement
slurry, leading to an ideal reinforcement effect. The soil grains and
pore water can obstruct the pore channel, resulting in partial filling
with less contact. Some soil aggregates may not be fully crushed
04022063-4
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
Fig. 6. Mechanism of strength gain of cement stabilized soil referring to: (a) chemical reactions; and (b) physical contact between cement slurry and
soil aggregates.
during the penetration and mixing process, and the cement slurry
only wraps on their interfaces without filling the voids.
Construction Impact Factors
The UCS of DCM mixtures can be influenced by binder characteristics, soil conditions, mixing conditions, and curing conditions
(Terashi 1997) based on laboratory studies. However, many of the
conditions cannot be well controlled in field conditions, and how
the strength of cement-stabilized soil behavior is influenced by construction techniques and site conditions has not been fully explored
in the previously reported studies in the literature. In this study,
given the illustrated DCM construction workflow and the mechanisms of DCM strength formation shown in Fig. 7, several key construction factors that may influence the DCM final strength are
identified. These factors are associated with construction site conditions, construction procedures, and curing conditions.
Impact factors related to construction site conditions include
fluctuation of tidal level (m/hour) and soil type. Large fluctuations
of tidal level can disturb the mixing shafts and blades during the
penetration, lifting, and mixing stages of DCM construction, leading to partial contact between cement slurry and soil aggregates.
Different soil types provide varied cation exchange capacities
because the cations can be held by the negatively charged particles
of fine-grained soils with thin and plate structures, and sand cannot
exchange cations without electrical charge. Hence, soils with a
© ASCE
higher clay fraction adsorb metal ions better than sandy soils, leading to a higher concentration of adsorbed calcium ions during cation exchange (e.g., Manrique et al. 1991). Furthermore, the varying
silica and aluminate and pH levels in different soils can affect the
pozzolanic reaction. The soil pore structure provides distinct permeability properties, which may also affect the heterogeneity of the
strength of DCM treated soil. Macropores (>0.075 mm) are more
abundant in coarse-grained soils (sandy soils), accounting for about
35%–50% of the overall pore volume, whereas fine-grained soils
(silty and clayey soils) are dominated by micropores (<0.03 mm)
and have smaller particle spacing (SSGTC and SSSA 2008).
The key factors related to the construction procedure include the
volume fraction of injected water, volume fraction of injected
cement slurry, cement injection rate (m3 =min), and penetrating and
mixing time per meter (min=m). The volume fraction of injected
water is the ratio of the injected water volume during penetration
to the bulk volume of a single DCM cluster, representing the water
quantity for each cluster. Water is a reactant involved in hydration,
cation exchange, and pozzolanic reactions. Nevertheless, excessive
water can reduce the concentration of other reactants and the environmental alkalinity in chemical reactions and increase water pockets not filled by cement slurry, leading to strength reduction. The
volume fraction of injected cement slurry indicates the cement
quantity as a basic reactant for hydration. The produced calcium
hydroxide also participates in subsequent cation exchange and
pozzolanic reactions, contributing to the strength gain of the
04022063-5
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
Fig. 7. Workflow of DCM mixing unit and associated key factors affecting strength formation.
cement–soil mixture. A higher cement injection rate can provide
greater initial kinetic energy of cement slurry from nozzles, facilitating the cement slurry to permeate into the voids between soil
particles more thoroughly at the bottom treatment, lifting, and mixing stages. At the penetrating and mixing stages, the mixing units
crushed the natural soil mass into small aggregates through cutting
blades, allowing sufficient contact between the soil particles and the
injected cement slurry. Increasing penetration and mixing time can
increase the contact surface area and accelerate the second phase of
the cement hydration and pozzolanic reaction.
The influence of the curing condition is observed through test
samples’ age and moisture content. The pozzolanic reaction can
last months to years for long-term strength formation, leading to
the variation of sample strength at different ages. The moisture
contents of cored samples indicate wetness at the curing stage.
Excessive water can reduce the alkalinity and concentration of
silica and alumina for pozzolanic reactions during the curing stage.
Considering water injection during construction, the multiple
stages of pozzolanic reaction, and the hydrate crystal products,
the curing moisture content at time of coring is used as a direct
field indicator to indicate the water content in DCM samples rather
than the moisture content in natural soil.
Interpretation of Construction Effects Based on
Field Records
In order to quantify the influence of each key in-situ construction
factor, project field records were analyzed along with the construction conditions and measured DCM strengths.
Processing of Field Records
According to the site borehole records and geotechnical site investigations, the average UCS of cored DCM-stabilized soil mixtures
was calculated for four principal soil types: clayey, silty, sandy, and
gravelly soils. This study grouped curing ages into four sets from
© ASCE
28 to over 80 days in 20-day intervals. The increase of cementstabilized soil strength tended to level off beyond 60 days when
the initial water content of the original soil was high. The whole
DCM construction was performed in the marine area and under the
seabed level, and water was injected while penetrating the soil for
protecting the mixing units and smoothing the penetration process.
The injected water and cement slurry volumes were recorded in the
penetrating and mixing stages. These volumes were transformed
into volume fractions by dividing the individual bulk volume of
the tested DCM column to normalize the influence of the DCM
cluster size.
The scatterplots of the average UCS versus seven influence factors for marine clay are presented in Fig. 8, reflecting the quantity
and fluctuation of UCS with these factors. The coefficients of variation (COV, ratio of standard deviation and mean) of UCS associated
with these factors are shown in Fig. 9. The COV reflects the homogeneity level of UCS in the construction site. A greater value of
COV indicates decreased homogeneity of UCS and greater uncertainty from the influence factors. The average UCS and the COV
for dominant natural soil types are summarized in Fig. 10.
Importance of Influence Factors
Sensitivity analysis was performed to identify the key construction
factors that impact the strength and uniformity of DCM constructions, based on the out-of-bag (OOB) importance estimation. The
OOB estimation measures how influential the predictor variables
(i.e., DCM construction factors) are in predicting the response
(i.e. observed UCS) by permutation using a random forest model
(e.g., Altmann et al. 2010). The influence of a particular construction factor can be manifested by the variation of model error by
random permutations. If a construction factor has a greater influence on the UCS, then permutating its value should affect the
model error significantly, whereas if a construction factor is not
influential, there will be little variation of the model error after permutating its value. The error difference assuming a random forest
model of T trees and P predictors is shown as follows:
04022063-6
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
dtj ¼ εtj − εt ¼
Pn
i¼1
ðyi − ŷbag;tj Þ2
−
n
Pn
i¼1
ðyi − ŷtj Þ2
n
observations for each predictor variable j ¼ 1; : : : ; P, the generated UCS ŷbag and model error εtj can be estimated. The OOB predictor importance can be quantified by the ratio between the
difference of errors averaged over all trees and the standard
deviation dj =σj . The larger the ratio dj =σj , the more influence
the key factor has on the DCM. The results of the predictor
ð1Þ
Average UCS (MPa)
8
8
8
8
7
7
7
7
6
6
6
6
5
5
5
5
4
4
4
4
3
3
3
3
2
2
2
2
1
1
0
0
0.2
0.3
0.4
0.48
0.50
(a)
Average UCS (MPa)
0.52
0.54
0.56
0.58
1
1
0
0
Injection rate (m3 /min)
Tidal fluctuation (m/h)
7
7
7
6
6
6
5
5
5
4
4
4
3
3
3
2
2
2
1
1
1
2.5
0.15
3.0
3.5
4.0
0
3.0
3.3
Penetration unit time (min/m)
3.6
0
10
3.9
Mixing unit time (min/m)
20
0.25
0.30
0.27
0.30
30
40
50
0.33
0.36
0.39
0.42
Injected slurry volume fraction
(d)
60
70
80
Moisture content (%)
(f)
(e)
0.20
(c)
8
2.0
0.10
(b)
8
1.5
0.05
Injected water volume fraction
8
0
1.0
(g)
COV of UCS
Fig. 8. Relationships between the average UCS of marine clay and key DCM construction factors: (a) tidal fluctuation (m=h); (b) injection rate
(m3 =min); (c) injected water volume fraction; (d) injected slurry volume fraction; (e) penetration unit time (min/m); (f) mixing unit time (min/m); and
(g) moisture content (%).
1.0
1.0
1.0
1.0
0.8
0.8
0.8
0.8
0.6
0.6
0.6
0.6
0.4
0.4
0.4
0.4
0.2
0.2
0.2
0.2
0.0
0.2
0.3
0.0
0.4
Tidal fluctuation (m/h)
0.48
0.50
0.52
0.54
Injection rate
0.56
0.58
0.0
(m3 /min)
1.0
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
2.0
2.5
3.0
3.5
Penetration unit time (min/m)
(e)
0.15
4.0
0.0
3.0
3.3
0.20
0.25
0.30
(c)
1.0
1.5
0.10
Injected water volume fraction
1.0
0.0
1.0
0.05
(b)
(a)
COV of UCS
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
where for each tree t ¼ 1; : : : ; T; n = number of samples; εt = OOB
error estimated by the mean squared error; yi = observed UCS; and
ŷ = predicted UCS from the model. After randomly permutating the
3.6
3.9
0.0
10
Mixing unit time (min/m)
(f)
20
30
40
50
0.0
0.27
0.30
0.33
0.36
0.39
0.42
Injected slurry volume fraction
(d)
60
70
80
Moisture content (%)
(g)
Fig. 9. Relationships between the COV of UCS of marine clay and key DCM constructions factors: (a) tidal fluctuation (m=h); (b) injection rate
(m3 =min); (c) injected water volume fraction; (d) injected slurry volume fraction; (e) penetration unit time (min/m); (f) mixing unit time (min/m); and
(g) moisture content (%).
© ASCE
04022063-7
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
28-40days
40-60days
60-80days
0.7
3
2
28-40days
40-60days
60-80days
0.6
COV of UCS
Average UCS (MPa)
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
4
0.5
0.4
0.3
0.2
1
0.1
0
Clayey
CLAY
soils
Silty
SILT
soils
Sandy
SAND
soils
0.0
Gravelly
GRAVEL
soils
Dominant soil type
Clayey
CLAY
soils
Silty
SILT
soils
Sandy
SAND
soils
Gravelly
GRAVEL
soils
Dominant soil type
Fig. 10. Average and COV of UCS from DCM samples for varied natural dominant soil types.
Fig. 11. Out-of-bag permuted predictor importance estimates for key factors from offshore DCM construction. A larger value suggests a greater
influence.
importance estimates for all key factors are summarized in Fig. 11.
As can be seen, for the DCM offshore construction project, the
amount of water injected during drilling and type of native soil
are the most important factors, whereas the difference in grouting
rate in the range of 0.48–0.58 m3 =min does not affect the strength
substantially.
Influence of Key Factors
The influence of nine key factors on the UCS was demonstrated
referring to the strengths of cored samples and strength gain mechanisms of cement–soil mixtures.
Original dominant soil type. Based on borehole records, four
dominant soil types were involved: clayey, silty, sandy, and gravelly soils. The average UCS of the DCM samples was approximately 2.5 MPa. The UCS of the samples originating in marine
clay was the highest, leading by almost 0.5 MPa for all ages compared to DCM mixtures in other soil types [Fig. 10(a)]. The inherent difference in dominant soils’ cation exchange capacity (CEC)
could be considered a contributing factor for strength gains in different types of soil. The marine clays had the highest clay fraction,
composed of negatively charged clay particles with greater CEC
than other dominant soil types. Therefore, a higher concentration
of calcium ions adsorbed to clay particles can be achieved from
© ASCE
cation exchange, which further influences the pozzolanic reactions
that form long-term strength. There was an ascending trend in the
COV from marine clay to gravel, indicating that the homogeneity
level of UCS becomes lower with increasing particle size or coarse
content. It is reasonable because large voids are more likely to form
when the soil aggregates are large, which easily introduce varied
cement slurry saturation degrees when the same construction technique is applied to all soil types during DCM construction. An exception was the COV for silt soil, which was larger than the
COV for sandy soil. One explanation is that silt soil has a dualporosity structure compared to clays and sand, and its mechanical
behavior is controlled by soil aggregates and coarse particles
together (e.g., Zhao et al. 2013). This structure with large interaggregate pores leads to varying degrees of permeation of cement
slurry into the voids during construction [Fig. 6(b)], possibly leading to heterogeneity in the treated soil.
Fluctuation of tidal level. Another potential disturbance of offshore DCM construction is the tidal impact on the mixing units on
the DCM barge. Reduction of the average UCS and large fluctuations in the COV could be observed after tidal fluctuation reached
0.38 m per hour [Fig. 8(a)]. Large tidal fluctuations may introduce
vibrations of the mixing shafts and blades that operate in the lifting
and rotating stages. Such vibrations may disturb the mixing unit
04022063-8
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
to varying extents and produce uneven contact conditions between
the soil particles and cement slurry, leading to uncertainty in the
homogeneity of cement-stabilized soil.
Volume fraction of injected water. The volume fraction of injected water, primarily in the range of 0.05–0.30, represents the
volume of injected water for protecting the mixing unit and
smoothening the penetration activity. It does not include the water
in the natural soils and cement slurry. The average UCS decreases
with increasing injected water volume, from an initial 3.5 MPa to
around 2.5 MPa [Fig. 8(c)]. Although the injected water constitutes
the primary reactant for the hydration and pozzolanic reactions in
the cement soil mixtures, excessive water injection can reduce the
alkalinity of the environment and concentration of other reactants
involved in the cation exchange, hydration, and pozzolanic reactions. The redundant water contributes to the separation of soil
aggregates by water pockets, leading to reduced UCS.
Volume fraction of injected cement slurry. The average UCS
increased with the injected volume fraction of cement slurry in the
range of 0.27–0.42 [Fig. 8(d)]. Cement slurry is the main reactant
for generating hydrated calcium-based gels and forming cementitious matrices with a fibrous crystal structure, which dominates the
strength formation of cemented soils. The increased volume fraction of cement slurry directly improves the principal components
contributing to the strength. The COV reached a minimum value at
around 0.2 when the injected slurry increased to 0.41 [Fig. 9(d)].
Injection of sufficient cement slurry ensures the cement slurry fills
the off-contact gaps between soil aggregates, contributing to higher
physical contact degree and homogeneity of strength. The COV
was high initially. When the amount of cement slurry injected is
low, only part of the pores in the soil are filled, thus producing
a nonuniform distribution of hydration products. When the injected
volume rises, more pores are filled and the reactants are in good
contact, leading to a reduced COV.
Cement slurry injection rate. A higher cement slurry injection
rate provides larger kinetic energy of cement slurry from the nozzles of the mixing units, improving the contact degree of cement
slurry and soil aggregates. The injection rate ranges between 0.48
and 0.57 m3 =min, leading to approximately 0.35 m3 =m of the
grout volume fraction under identical mixing time per meter. The
average UCS and COV [Figs. 8(b) and 9(b)] varied little within this
range of injection rate, as did the grout penetration in soil aggregates. The effect of the change in grout volume per unit time on the
UCS can be obscured by the variation of the volume fraction of
cement slurry. Hence this factor is the least sensitive construction
factor in terms of the estimated OOB importance (Fig. 11).
Penetrating time per meter. When mixing units are drilled
through a stiff soil layer or a hard obstruction during the penetration
stage, additional time is required, and a larger amount of water will
be injected to protect the mixing units and smooth the penetration
process. Therefore, the average UCS and COV variations with penetration time per meter [Figs. 8(e) and 9(e)] are similar to those of
the injected water volume fraction. As mentioned, the drilling process broke up large soil clods into smaller pieces and improved the
contact conditions of the cementation reactants. However, the drilling process was accompanied by water injection, so its effect on
strength was overshadowed by the dominant factor, water volume,
as illustrated in Fig. 11.
Mixing time per meter. With the increased mixing time from
3.1 to 3.9 min per meter, the average UCS improved from an initial
3 to 3.5 MPa [Fig. 8(f)], and the homogeneity level of UCS also
improved, indicated by a decrease in the COV [Fig. 9(f)]. The
increased mixing time per meter ensures the mixing blades thoroughly cut the original soil aggregates into small pieces and
increases the contact area of the reactants for chemical reactions
© ASCE
associated with the strength gain of cement–soil mixtures. The increased mixing time also reduces the incomplete filling of the voids
between soil aggregates by the cement slurry, leading to improved
homogeneity.
Age. Cement-stabilized soil strength grows slowly over 60 days
when the initial water content of the natural soils is high (Kawasaki
et al. 1981). Indeed, in this project, the magnitude and homogeneity
level of strength increased substantially during the curing ages of
28–60 days for marine clayey, silty, and sandy soils, as shown in
Fig. 10. With the increasing curing age, the pozzolanic reactions
tend to complete, with more cemented calcium silicate and calcium
aluminate hydrate accumulating and surrounding the soil particles,
improving the cementation strength. After 60 days, the UCS for
the silty and gravelly marine soils reached a stabilized level, but
those for the clayey and sandy soils continued to grow. This can be
explained by the difference in the pore structure. Both the dualporosity structure of the silty soil with coarse particles and the
macropores of gravelly soils introduce high permeability characteristics, allowing ease of fluid flow. With water infiltration, the environmental alkalinity decreases, and the concentration of activated
calcium ions is diluted. Therefore, the reduced concentration of
reactants in the pozzolanic reaction leads to a mitigated reaction
degree and stabilized strength.
Moisture content. The moisture content of the cement–soil
mixture was measured during the UCS test [Figs. 8(g) and 9(g)].
The average UCS gradually decreased to around 2 MPa at a
moisture content of 80% in the cement–soil mixtures. The initial
water in the cement–soil mixture serves as a reactant for pozzolanic
reactions. However, the excessive amount of water can decrease the
concentration of reactants and the alkaline chemical reaction environment, which then delays the long-term strength formation process of cement-stabilized soils. Large fluctuations in COV are
observed after the moisture content is greater than 60% for different
curing ages. High moisture content can cause uneven distribution
of excessive water, which requires additional time to react with the
silica and alumina from the soil and calcium hydroxide during
pozzolanic reactions, leading to the varying COV at different curing
ages.
Summary and Conclusions
Compared to traditional dredged land reclamation methods, offshore DCM construction is more environmentally friendly without
dredging or disturbance to the marine ecological system. An offshore DCM project for undredged land reclamation was evaluated
to identify key construction factors that affect the homogeneity and
magnitude of UCS. These factors are related to geological conditions, construction procedures, curing conditions, and chemical and
physical mechanisms of cement-stabilized soils. Several conclusions can be drawn:
1. The impact on the UCS from the volume fraction of injected
water and the natural moisture content is most significant among
all the evaluated factors from offshore DCM construction. The
volume of injected water must be carefully calibrated, especially
when the marine soils to be treated have high natural water contents. Additional injected water at the penetration stage may exceed the hydration and pozzolanic requirements and decrease
the concentration of other reactants and alkalinity of the environment, affecting the strength formation of cement-stabilized
soils.
2. The varying soil types at the construction site significantly impacted the UCS of offshore DCM in terms of both homogeneity
and magnitude. Due to the inherent variation of the cation
04022063-9
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
exchange capability, fine-grained marine soils with high clay
fractions generate greater cement–soil mixture strength compared to coarse-grained sandy soils. Hence specific injection
and mixing methodologies should be recommended for different
soils, including using extra mixing time and cement slurry injection to ensure adequate cutting effects of mixing blades and
increase the contact degree between the cement slurry and soil
aggregates.
3. Longer mixing time during construction did not reduce the
UCS but promoted it, whereas large tidal-level fluctuations
may lead to vibrations of the mixing units and produce strength
variations.
4. The strength of DCM increases with curing age within a curing
age up to 80 days for the dominant clayey and sandy marine
soils. For silty and gravelly soils, the UCS appeared to reach
a stabilized level 60 days after the DCM treatment.
Data Availability Statement
Some or all data, models, or code generated or used during the
study are available from the corresponding author by request.
Acknowledgments
The work presented in this paper was substantially supported by
Eunsung O&C Offshore Marine and Construction (No. EUNSUNG19EG01).
References
Abusharar, S. W., J. J. Zheng, and B. G. Chen. 2009. “Finite element modelling of the consolidation behavior of multi-column supported road
embankment.” Comput. Geotech. 36 (4): 676–685. https://doi.org/10
.1016/j.compgeo.2008.09.006.
Altmann, A., L. Toloşi, O. Sander, and T. Lengauer. 2010. “Permutation
importance: A corrected feature importance measure.” Bioinformatics
26 (10): 1340–1347. https://doi.org/10.1093/bioinformatics/btq134.
Bellato, D., I. P. Marzano, and P. Simonini. 2020. “Microstructural analyses
of a stabilized sand by a deep-mixing method.” J. Geotech. Geoenviron.
Eng. 146 (6): 04020032. https://doi.org/10.1061/(ASCE)GT.1943-5606
.0002254.
Bullard, J. W., H. M. Jennings, R. A. Livingston, A. Nonat, G. W. Scherer,
J. S. Schweitzer, and J. J. Thomas. 2011. “Mechanisms of cement
hydration.” Cem. Concr. Res. 41 (12): 1208–1223. https://doi.org/10
.1016/j.cemconres.2010.09.011.
Chen, E. J., Y. Liu, and F. H. Lee. 2016. “A statistical model for the
unconfined compressive strength of deep-mixed columns.” Géotechnique
66 (5): 351–365. https://doi.org/10.1680/jgeot.14.P.162.
Chen, J., F. H. Lee, and C. C. Ng. 2011. “Statistical analysis for strength
variation of deep mixing columns in Singapore.” In Geo-frontiers:
Advances in geotechnical engineering, GSP No. 211, edited by J. Han
and D. E. Alzamora, 576–584. Reston, VA: ASCE.
Disfani, M. M., A. Mohammadinia, A. Arulrajah, S. Horpibulsuk, and
M. Leong. 2021. “Lightly stabilized loose sands with alkali-activated
fly ash in deep mixing applications.” Int. J. Geomech. 21 (3): 04021011.
https://doi.org/10.1061/(ASCE)GM.1943-5622.0001958.
Furudoi, T. 2005. “Second phase construction project of Kansai International
Airport—Large-scale reclamation works on soft deposits.” In Vol. 16 of
Proc. of the Int. Conf. on Soil Mechanics and Geotechnical Engineering,
313. Amsterdam, Netherlands: IOS Press BV.
Honjo, Y. 1982. “A probabilistic approach to evaluate shear strength of
heterogeneous stabilized ground by deep mixing method.” Soils Found.
22 (1): 23–38. https://doi.org/10.3208/sandf1972.22.23.
Horpibulsuk, S., A. Chinkulkijniwat, A. Cholphatsorn, J. Suebsuk, and
M. D. Liu. 2012. “Consolidation behavior of soil-cement column
© ASCE
improved ground.” Comput. Geotech. 43 (Jun): 37–50. https://doi
.org/10.1016/j.compgeo.2012.02.003.
Horpibulsuk, S., N. Miura, and T. S. Nagaraj. 2003. “Assessment of
strength development in cement-admixed high-water content clays with
Abrams’ law as a basis.” Géotechnique 53 (4): 439–444. https://doi.org
/10.1680/geot.2003.53.4.439.
Jiang, Y., J. Han, and G. Zheng. 2013. “Numerical analysis of consolidation
of soft soils fully penetrated by deep-mixed columns.” KSCE J. Civ.
Eng. 17 (1): 96–105. https://doi.org/10.1007/s12205-013-1641-x.
Kawasaki, T., A. Niina, S. Saitoh, Y. Suzuki, and Y. Honjo. 1981. “Deep
mixing method using cement hardening agent.” In Vol. 3 of Proc., 10th
Int. Conf. on Soil Mechanics and Found. Eng., 721–724. London:
ISSMGE.
Kitazume, M., M. Grisolia, E. Leder, I. P. Marzano, A. A. S. Correia, P. J. V.
Oliveira, and M. Andersson. 2015. “Applicability of molding procedures in laboratory mix tests for quality control and assurance of the
deep mixing method.” Soils Found. 55 (4): 761–777. https://doi.org/10
.1016/j.sandf.2015.06.009.
Larsson, S., M. Dahlström, and B. Nilsson. 2005a. “A complementary field
study on the uniformity of lime-cement columns for deep mixing.”
Proc. Inst. Civ. Eng. Ground Improv. 9 (2): 67–77. https://doi.org/10
.1680/grim.2005.9.2.67.
Larsson, S., M. Dahlström, and B. Nilsson. 2005b. “Uniformity of limecement columns for deep mixing: A field study.” Proc. Inst. Civ. Eng.
Ground Improv. 9 (1): 1–15. https://doi.org/10.1680/grim.2005.9.1.1.
Lee, F. H., C. H. Lee, and G. R. Dasari. 2006. “Centrifuge modelling of wet
deep mixing processes in soft clays.” Géotechnique 56 (10): 677–691.
https://doi.org/10.1680/geot.2006.56.10.677.
Lee, F. H., Y. Lee, S. H. Chew, and K. Y. Yong. 2005. “Strength and modulus
of marine clay–cement mixes.” J. Geotech. Geoenviron. Eng. 131 (2):
178–186. https://doi.org/10.1061/(ASCE)1090-0241(2005)131:2(178).
Liu, S. Y., Y. J. Du, Y. L. Yi, and A. J. Puppala. 2012. “Field investigations
on performance of T-shaped deep mixed soil cement column–supported
embankments over soft ground.” J. Geotech. Geoenviron. Eng. 138 (6):
718–727. https://doi.org/10.1061/(ASCE)GT.1943-5606.0000625.
Lorenzo, G. A., and D. T. Bergado. 2004. “Fundamental parameters of
cement-admixed clay—New approach.” J. Geotech. Geoenviron. Eng.
130 (10): 1042–1050. https://doi.org/10.1061/(ASCE)1090-0241(2004)
130:10(1042).
Manrique, L. A., C. A. Jones, and P. T. Dyke. 1991. “Predicting cationexchange capacity from soil physical and chemical properties.” Soil
Sci. Soc. Am. J. 55 (3): 787–794. https://doi.org/10.2136/sssaj1991
.03615995005500030026x.
Mohammadinia, A., M. M. Disfani, D. Conomy, A. Arulrajah, S. Horpibulsuk,
and S. Darmawan. 2019. “Utilization of alkali-activated fly ash for
construction of deep mixed columns in loose sands.” J. Mater. Civ.
Eng. 31 (10): 04019233. https://doi.org/10.1061/(ASCE)MT.1943-5533
.0002878.
Navin, M. P. 2005. “Stability of embankments founded on soft soil
improved with deep-mixing-method columns.” Doctoral Dissertation,
Dept. of Civil Engineering and Environmental Engineering, Virginia
Tech.
Rogers, C. D. F., and S. Glendinning. 1997. “Improvement of clay soils in
situ using lime piles in the UK.” Eng. Geol. 47 (3): 243–257. https://doi
.org/10.1016/S0013-7952(97)00022-7.
Sherwood, P. T. 1993. Soil stabilization with cement and lime—State of the
art review. London: Transport Research Laboratory, Dept. of Transport,
HMSO publications.
Soil Science Glossary Terms Committee and Soil Science Society of
America. 2008. Glossary of soil science terms 2008. Madison, WI: Soil
Science Society of America.
Subramaniam, P., and S. Banerjee. 2020. “Dynamic properties of cementtreated marine clay.” Int. J. Geomech. 20 (6): 04020065. https://doi.org
/10.1061/(ASCE)GM.1943-5622.0001673.
Terashi, M. 1997. “Theme lecture: Deep mixing method—Brief state of the
art.” In Vol. 4 of Proc. 14th Int. Conf. on Soil Mech. and Found. Eng.,
2475–2478. Rotterdam, Netherlands: A.A. Balkema.
Watabe, Y., and T. Noguchi. 2011. “Site-investigation and geotechnical design of D-runway construction in Tokyo Haneda airport.” Soils Found.
51 (6): 1003–1018. https://doi.org/10.3208/sandf.51.1003.
04022063-10
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.
Wijerathna, M., and D. S. Liyanapathirana. 2018. “Reliability-based performance of embankments improved with deep mixing considering spatial variability of material properties.” ASCE-ASME J. Risk Uncertainty
Eng. Syst. Part A: Civ. Eng. 4 (4): 04018035. https://doi.org/10.1061
/AJRUA6.0000987.
Wijerathna, M., and D. S. Liyanapathirana. 2019. “Significance of spatial
variability of deep cement mixed columns on reliability of columnsupported embankments.” Int. J. Geomech. 19 (8): 04019087. https://doi
.org/10.1061/(ASCE)GM.1943-5622.0001473.
Yang, T., J. Z. Yang, and J. Ni. 2014. “Analytical solution for the consolidation of a composite ground reinforced by partially penetrated impervious columns.” Comput. Geotech. 57 (Apr): 30–36. https://doi.org/10
.1016/j.compgeo.2014.01.001.
© ASCE
Yapage, N. N. S., D. S. Liyanapathirana, R. B. Kelly, H. G. Poulos, and
C. J. Leo. 2014. “Numerical modeling of an embankment over soft
ground improved with deep cement mixed columns: Case history.”
J. Geotech. Geoenviron. Eng. 140 (11): 04014062. https://doi.org/10
.1061/(ASCE)GT.1943-5606.0001165.
Yin, J. H., and Z. Fang. 2006. “Physical modelling of consolidation behavior of a composite foundation consisting of cement-mixed soil column
and untreated soft marine clay.” Geotechnique 56 (1): 63–68. https://doi
.org/10.1680/geot.2006.56.1.63.
Zhao, H. F., L. M. Zhang, and D. S. Chang. 2013. “Behavior of coarse
widely graded soils under low confining pressures.” J. Geotech. Geoenviron. Eng. 139 (1): 35–48. https://doi.org/10.1061/(ASCE)GT.1943
-5606.0000755.
04022063-11
J. Geotech. Geoenviron. Eng., 2022, 148(8): 04022063
J. Geotech. Geoenviron. Eng.