Professional Documents
Culture Documents
2D Deep Mixing Column
2D Deep Mixing Column
MANAGEMENT,
SECOND CYCLE, 30 CREDITS
STOCKHOLM, SWEDEN 2017
BIEKE PEETERS
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Disposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Literature review 4
2.1 Deep mixing method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 Current state of practice of the deep mixing method . . . . . . . . . 4
2.1.1.1 Installation and production of deep mixing columns . . . . 5
2.1.2 Properties of mixing materials . . . . . . . . . . . . . . . . . . . . . 8
2.1.3 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Failure of deep mixing stabilized ground . . . . . . . . . . . . . . . . . . . 10
2.2.1 Current Swedish design method for stabilized ground . . . . . . . . 10
2.2.2 All failure mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Research on embankment supported by deep mixing columns . . . . . . . . 14
2.3.1 Numerical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.1.1 Continuous shear panels . . . . . . . . . . . . . . . . . . . 18
2.3.2 Laboratory tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Research on excavation supported with rows of dry deep mixing columns . 24
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3 General methodology 27
3.1 Geometrical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 Material properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4 Analytical calculations 32
4.1 Extrusion of soft ground between shear walls . . . . . . . . . . . . . . . . . 32
4.2 Internal stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2.1 No reduction of strength and stiffness at the overlap . . . . . . . . . 38
4.2.1.1 Undrained shear strength . . . . . . . . . . . . . . . . . . 38
4.2.1.2 Drained shear strength . . . . . . . . . . . . . . . . . . . . 39
4.2.2 Reduction of strength and stiffness at the overlap . . . . . . . . . . 41
i
4.2.2.1 Undrained shear strength . . . . . . . . . . . . . . . . . . 41
4.2.2.2 Drained shear strength . . . . . . . . . . . . . . . . . . . . 43
4.2.3 Calculation safety factor . . . . . . . . . . . . . . . . . . . . . . . . 44
5 Numerical calculations 47
5.1 Geometrical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2 Material model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2.1 Mohr-Coulomb model . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2.2 Material properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.2.2.1 Properties of the overlap of the DM columns . . . . . . . . 52
5.2.2.2 Properties of the column and overlap zones in the 2D model 53
5.2.3 Definition of failure . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2.4 Safety analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3.1 Analysis of the failure mode . . . . . . . . . . . . . . . . . . . . . . 56
5.3.1.1 Quality of overlap: 50% . . . . . . . . . . . . . . . . . . . 57
5.3.1.2 Quality of overlap: 25% . . . . . . . . . . . . . . . . . . . 60
5.3.1.3 Quality of overlap: 100% . . . . . . . . . . . . . . . . . . . 60
5.3.2 Safety factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6 General discussion 66
6.1 Risk for extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6.2 Influence of overlap between adjacent columns . . . . . . . . . . . . . . . . 66
6.3 How representative is LEM compared to numerical analysis? . . . . . . . . 67
7 Conclusions 69
ii
Summary
In geotechnical engineering, a technique called ”deep-mixing method” is often used to im-
prove soil. The deep-mixing method, a deep in-situ stabilization technique using cement
or lime as a binder has been applied to improve soft soils [1]. The internal failure of the
improved ground will be investigated in this study. The internal failure of the deep mixing
columns is the failure taking place in the columns due to their low strength [2]. Shear walls
constructed by overlapping individual deep mixing columns contain vertical joints caused
by the reduced width of the wall at the overlap between adjacent columns. These vertical
joints can be weaker by misalignment during construction, which reduces the efficiency
of the overlap [3]. To analyse the shear failure of the improved soil, the improved soil is
assumed to behave as a composite soil volume with an average strength and stiffness. The
strength and the stiffness of the composite soil are calculated using the weighted average
of the stiffness and strength of the lime-cement columns and the surrounding soil. The
internal failure can be investigated using the limit equilibrium method or two-dimensional
numerical analysis. When the shear walls are widely spaced, a failure mechanism in which
the slip surface will develop between the shear panels, can occur. Extrusion of the soft
ground between the shear walls is checked in this study.
The aim of this study consist of two main subjects: 1) The acceptable distances between
the shear panels against extrusion are investigated to check when the composite soil model
against internal failure is valid. 2) The potential influence from the overlap zones between
adjacent columns is investigated. This will be done by performing calculations with both
the limit equilibrium method and two-dimensional numerical analysis with varying strength
of the overlap zones.
The main conclusions of this study are:
• The factor-of-safety-values are smaller for the numerical analyses compared to the
limit equilibrium method. The overestimation of the safety factor is higher for lower
embankment loads. It is recommended that this should be further studied in future
work.
• There is a possibility that the composite soil volume will not fail in shear so that
an analysis using the limit equilibrium method will not be enough. A shear failure
mode develops when the quality of overlap is weak, an external sliding failure occurs
when the quality of overlap is strong.
• It can be concluded that it is better to include the reduced quality of overlap in the
analytical calculations, following the method proposed in this study, in order to get
a safe design because the safety factor decreases when the quality of overlap reduces.
Keywords: Lime, cement, deep mixing columns, limit equilibrium method, finite element
analysis.
iii
Sammanfattning
ȧ Inom geotekniken anvnds vanligen kalk-cement pelare för att förbttra Jorden. Metoden
använder cement eller kalk som bindemedel för att förstärka jorden och utgörs av pelare
som placeras i olika mönster beroende pȧ typ av förstärkning. I samband med dessa typer
av förstärkningar kan den förstärkta jorden gȧ till olika typer av brott, där en ett inre
skjuvbrott i form av en glidyta har studerats i detta arbete. Denna typ av brott utvecklats
i den förstärka jorden när dess skjuvhȧllfasthet är lȧg. När kalk-cement pelarna placeras i
rader uppstȧr en överlappning mellan pelarna, vars hȧllfasthet kan vara lägre pȧ grund av
avvikelser i sidled vilket reducerar hȧllfastheten för detta omrȧde.
Säkerheten mot brott beräknas vanligtvis med en förenklad jämviktanalys där glidytan
betraktas som en stel kropp, men kan även beräknas med numeriska analyser för att
erhȧlla en mer korrekt bild av händelseförloppet. I bȧda dessa tvȧ-dimensionella analyser
betraktas raderna av pelarna med en kompositmodell bestȧende av jord med tillhörande
kalk-cementpelare. Ett annat brott som kan inträffa är utglidning mellan raderna av pelare
om de placeras för brett isär.
Syftet med detta arbete har varit att: (1) beräkna acceptabla avȧstnd mellan kalk-cementpelarna
för det undersöta fallet för att förhindra utglidning mellan pelarna, (2) undersöka hur vari-
ation i hȧllfastheten i överlappzonen pȧverkar säkerheten mot glidning och (3) undersöka
skillnaden i beräknad säkerhet jämfört med jämviktanalys och numerisk tvȧ-dimensionell
beräkning.
Resultaten frȧn det utförda arbetet visar att:
• Säkerhetsfaktorn är mindre för de numeriska analyserna jämfört med analytisk jämvik-
tanalys. Skillnaden är väsentligt högre vid lȧga belastningar. Det rekommenderas
att detta studeras vidare i framtida arbete.
• De numeriska analyserna visar att det är mjligt att den förstärkta jorden inte gȧr
till brott genom en utvecklad glidyta sȧsom antas i jämviktanalysen. Ett glidbrott
förefaller att utvecklas om hȧllfastheten i den överlappande zonen är lȧg, medan
detta inte sker dȧ hȧllfastheten i den överlappande zonen inte reduceras.
• Det kan konstateras att den potentiellt reducerade hȧllfastheten i den överlappande
zonen bör tas med i beräkningarna dȧ den leder till en nȧgot lägre säkerhetsfaktor.
iv
List of symbols
v
Es . . . . . . . . . . . . . Young’s modulus of the soil between columns
su,o2D . . . . . . . . . . . composite shear strength of the overlap zone
Eo2D . . . . . . . . . . . . composite Young’s modulus of the overlap zone
σt,c2D . . . . . . . . . . . composite tensile strength of the column zone
σt,s . . . . . . . . . . . . tensile strength of the soil between columns
σt,o2D . . . . . . . . . . . . composite tensile strength of the overlap zone
a . . . . . . . . . . . . . . the ratio between the area of the column to the area of the
soil
Acolumn,1 . . . . . . . . . . the area of the column including overlap
Asegment . . . . . . . . . . half of the area of overlap between adjacent columns
Aslice . . . . . . . . . . . ratio of the column area to the centre-to-centre distance of
the columns
c1 . . . . . . . . . . . . . centre-to-centre distance of the columns
c2 . . . . . . . . . . . . . distance between two deep mixing rows
τclay . . . . . . . . . . . . drained shear strength of the clay
τcol . . . . . . . . . . . . . drained shear strength of the deep mixing columns
uw . . . . . . . . . . . . . pore water pressure
σv0 . . . . . . . . . . . . . effective vertical stress
σv . . . . . . . . . . . . . total vertical stress
τaverage . . . . . . . . . . average drained shear strength
ΣM omentsresisting . . . . the sum of the resisting moments for extrusion of soft
ground between shear walls
ΣM omentsactivating . . . . the sum of the activating moments for extrusion of soft
ground between shear walls
A . . . . . . . . . . . . . the area of the sliding mass of clay
r . . . . . . . . . . . . . . the radius of the critical slip circle
x . . . . . . . . . . . . . . weighting factor, you weight the area of the column without
overlap to the total column area
y . . . . . . . . . . . . . . weighting factor, you weight the area of the two overlaps to
the total column area
su,overlap . . . . . . . . . . undrained shear strength of overlap
E . . . . . . . . . . . . . Young’s modulus
ν . . . . . . . . . . . . . . Poisson’s ratio
ψ . . . . . . . . . . . . . angle of dilatancy
e . . . . . . . . . . . . . elastic part of the strain rate
p . . . . . . . . . . . . . plastic part of the strain rate
vi
vii
Chapter 1
Introduction
1.1 Background
To build a highway or railway embankment, a soil that can carry the applied loads without
collapsing is needed. Since in Sweden, the surface consists mainly of soft soil, soil improve-
ment is often necessary to be sure that the applied loads can be carried without failure
of the soil. In geotechnical engineering, a technique called ”deep-mixing method” is often
used to improve soil. The deep-mixing method, a deep in-situ stabilization technique using
cement or lime as a binder has been applied to improve soft soils [1]. Binders like lime or
cement are injected and mixed with the soil using special equipment. After the mixture
process, the binder hydrates and reacts with the soil to form hard soil columns which are
much stronger and stiffer than the ambient soft soil. It has been used extensively since
the 1970s to reduce settlement and to improve stability of for example railway embank-
ments [4]. Lime columns were mainly used till the 1990s. Lime-cement colums replaced
the lime columns since the 1990s and they are used a lot nowadays [5]. The properties of
the improved soil depend on the amount and characteristics of the cement-lime binder, the
characteristics of the soil and the mixing and curing conditions [6].
Despite the succesful use of these lime-cement columns in the design of highay and railway
embankments, column failures or unexpected deformations occured in a few cases, even
when the design was made with sufficient high safety factors [8]. Analytical design methods
used nowadays to calculate the stability of the lime-cement columns can overestimate
the stability. In the majority of the cases where limit equilibrium methods are used to
analyse the stability, the limit equilibrium method will overestimate the factor of safety
of the structure compared to numerical analyses [7]. This happens because in addition to
shear failure, numerical analysis can capture failure mechanisms not included in the limit
equilibrium method.
The current Swedish design method for stabilized ground, described in TK Geo 13 [9]
and [10] assumes that the improved soil volume behaves as a composite soil volume. The
strength and the stiffness of the composite soil are calculated using the weighted average
1
Objectives 2
of the stiffness and strength of the lime-cement columns and the surrounding soil. Failure
should be checked in internal stability and external stability. The external stability is
the global failure of the improved ground area without any failure in the columns. The
internal failure of the deep mixing columns is the failure taking place in the columns due
to their low strength [2]. In internal stability, the columns will fail by shearing all at
the same time. The stability of the embankment is analysed using a slip circle analysis
with an average strength and stiffness used for the composite soil volume. Shear walls
constructed by overlapping individual deep mixing columns contain vertical joints caused
by the reduced width of the wall at the overlap between adjacent columns. These vertical
joints can be weaker by misalignment during construction, which reduces the efficiency of
the overlap [3]. The Swedish design method for stabilized ground does not take the effect
of weaker overlap zones into account. Extrusion of the soft ground between the columns
should also be checked. Extrusion of the soft clay ground could occur if the shear walls are
widely spaced.
1.2 Objectives
The objectives of the study are summarized as follows:
1. Investigate the acceptable distances between the shear panels against extrusion to
check when the composite soil model against internal failure is valid.
(a) An analytical method to calculate the safety factors for extrusion of soft ground
is made.
2. Investigate the potential influence from the overlap zones between adjacent columns.
This will be done by performing calculations with both the limit equilibrium method
and two-dimensional numerical analysis with varying strength of the overlap zones.
(a) Compare the safety factors for internal stability calculated according to the
Swedish design code for stabilized ground using the limit equilibrium method
with the safety factors calculated using a 2D finite element model.
(b) A method to calculate the safety factors for internal stability taking weaker
overlaps into account is made. Compare the safety factors for internal stability
calculated using the limit equilibrium method considering the strength of the
overlap zones with the safety factors calculated using a 2D finite element model.
(c) Investigate the influence of the reduction of strength and stiffness in the overlap
zones on the 2D numerical model.
1.3 Disposition
This thesis contains a review of studies and research performed concerning the failure
mechanisms of deep mixing columns. Further, it contains a number of chapters that
Limitations 3
describe the methodology used to calculate safety factors for internal failure.
1.4 Limitations
This study is not based on any specific site condition. The typical ground condition
was chosen from literature with the care of choosing suitable parameters. Therefore, the
results obtained cannot be compared to any measurements of a real life situation. Another
limitation of this study is that the results of the 2D numerical model cannot be compared
to the results of a 3D numerical model. Therefore, it is insecure in which cases the 2D
models are good or bad representations of the 3D real situations. One failure mode will
not be checked by using the numerical 2D model, namely the extrusion of clay between
the shear wall panels.
Chapter 2
Literature review
4
Current state of practice of the deep mixing method 5
There are some main reasons why this ground improvement method has expanded a great
deal around the world. First of all, ground improvement by deep mixing allows, simul-
taneously, an increase in stability and a reduction of settlements. Further, deep mixing
has a minor environmental impact and a great speed of execution. At last, because of
the development of machinery, it is possible to execute the deep mixing method for larger
depths (greater than 40m), in non-homogeneous soil formations, including soft soils, sand,
gravel, overconsolidated clays and even weathered soft rocks [15].
Dry deep mixing method A large number of dry deep mixing methods are developed
all over the world, especially in Japan and the Nordic countries. Examples of the main
tools used nowadays will be given in this section. The standard dry mixing tool in Japan
is shown in Figure 2.1.1. The blades of the tool rotate into the soil and form a cavity. The
binder is spread over columns cross section in this cavity. When the rotation speed of the
tool increases, the volume of the cavity increases and a vacuum is created, which facilitates
the process when manufacturing large diameter columns [17].
Figure 2.1.1: Standard mixing tool by the Japanese dry mixing method [17]
Current state of practice of the deep mixing method 6
In Scandinavia, the tool shown in Figure 2.1.2 is typically used. The mixing tool is rotated
down to a particular depth (dependent on the design length of the column), and once the
desired depth is reached the rotation of the mixing tool is reversed. The tool is withdrawn
at a constant speed. During withdrawn, the binder is transported into the soil using
compressed air pushing through a hole located just above the mixing tool [13], as shown
in Figure 2.1.2. The binder quantity injected into the soil is controlled by changing the
rotation speed of the feeding wheel [14]. The improved soil mixture is compacted by the
inclined blades of the mixing tool. Using the dry mixing method, there is almost no waste
of material since the dry mixing method brings almost no spoils to the ground surface [13].
Figure 2.1.2: Tools used in Scandinavia to produce deep mixing columns [14]
Current state of practice of the deep mixing method 7
Wet deep mixing method The wet deep mixing method has been extensively used
in Japan, particularly in marine projects [17]. Nowadays, it also has subsequently been
used for land applications. There are different techniques to produce columns by injecting
water-based binders into the soil. The most common approach is to inject part of the
binder as the mixing tool is penetrating the soil. While the mixing tool penetrates into the
soil, it disaggregates the soil and at the same time lifts the soil slightly in order to facilitate
incorporation. The remainder of the binder is injected as the mixing tool is withdrawn from
the soil [17]. The intensity by which the slurry is injected depends on the soil condition.
An example is shown in Figure 2.1.3.
Figure 2.1.3: Tools used to produce columns by wet mixing method [14]
Properties of mixing materials 8
• Mixing conditions
• Curing conditions
The Federal Highway Administration Design Manual [18] gives values for the unconfined
compressive strength of laboratory-mixed materials in the range of 0,01MPa to 2,8MPa
when the dry mixing method is used and 0,1 to 28MPa when the wet mixing method is
used. The strength depends on the characteristics of the soil, the binder type and the
amount of mixing binder. Compared to field-mixed material the strength of the laboratory
mixed material can be greater at the same mixture proportions. The strength of field-mixed
material can range from 20% to 100% of the laboratory mixed material. The strength of
the mixed material increases with an increased content of binder [2]. In Figure 2.1.4 the
results of different studies concerning the unconfined compressive strength of soil cement
mixing [19] are showed. The tensile and bending strength of the improved material is quite
small compared to the shear strength [20], [2] (see Figure 2.1.5).
Figure 2.1.5: Tensile and bending strength of stabilized soils [20], [2]
2.1.3 Applications
The deep mixing method is currently used for both on-land and marine constructions for
different purposes. On land, the deep mixing columns are used for embankment stability,
reduction of settlement of embankments, bridge abutment, cut slope stability, excavation
support, nearby structure impact reduction as well as liquefaction mitigation. Applications
of deep mixing columns for marine constructions are revetment, seawalls, foundations,
breakwater slope stability as well as tunnels and road construction [13].
In Sweden, the most frequent use of deep mixing columns for soil stabilization is for infras-
tructure, for example railway and roadway embankments, and for excavations [21]. The
purpose of these deep mixing columns is to increase the stability and reduce the settlement.
For example, in Sweden 83% of deep mixing columns produced are for supporting highway
or roadway embankments [21].
Deep mixing columns can also be used for the approach embankments to bridges. In some
projects the foundation of the bridge itself is supported by deep mixing columns. With
deep mixing columns, supporting both the bridge and the approach embankments, it is
possible to reduce differential settlements. An additional advantage of using deep mixing
columns to support the foundation of bridges is that the deep mixing stabilization can also
be used to improve the stability of the excavation for the bridge foundation [21].
Deep mixing columns are also used to improve properties of soft soil in such a way that
vibrations are reduced. For example, deep mixing columns installed in panels along railway
tracks can reduce the vibrations.
Examples of deep mixing columns used for marine applications can be found in construc-
tions of a waterfront as well as seepage cut-off. Furthermore, the deep mixing method
was also used as a countermeasure against liquefaction [2], [22]. Recently, the deep mixing
method is also applicable to environmental issues such as dredging soil and landfill struc-
tures [2].
Failure of deep mixing stabilized ground 10
Different improvement patterns of deep mixing columns can be used according to their
purpose. Different improvement patterns are shown in Figure 2.1.6. Typically, a group
of isolated columns is used when the horizontal loads are small. Wall, grid and block
type patterns, made by overlapping deep mixing columns, are mostly applied for a greater
external load as well as high stability requirements [2].
Figure 2.2.1: The peak shear strength of the columns is mobilized at the same time as the
the peak shear strength of the unstabilized soil between the lime-cement columns[5], [29]
The strength and the stiffness of the composite soil are calculated using the weighted
average of the stiffness and strength of the lime-cement columns and the surrounding soil.
The undrained shear strength can be calculated using equation 2.2.1.
Where
• su,col is the undrained shear strength of the lime-cement columns
• a is a weighting factor
The drained shear strength of the composite soil volume is calculated according to equation
2.2.2
Where
• τu,col is the drained shear strength of the lime-cement columns
• a is a weighting factor
Two failure patterns are assumed: sliding failure in external stability and rupture breaking
failure in internal stability [1]. The external stability is the global failure of the improved
ground area without any failure in the columns. The internal failure of the deep mixing
columns is the failure taking place in the columns due to their low strength [2]. In internal
stability, the columns will fail by shearing all at the same time. The stability of the
Current Swedish design method for stabilized ground 12
embankment is analysed using a slip circle analysis with an average strength and stiffness
used for the composite soil volume. To calculate the safety factors for shear failure, a
slip surface analysis can be performed with limit equilibrium method using the program
SLOPE. In external stability, the columns and clay will fail by sliding. The composite
soil moves horizontally as a block on a stiff layer. Extrusion of the soft clay ground could
occur if the shear walls are widely spaced. Therefore, extrusion of the soft ground between
shear walls should also be checked. Rupture breaking failure in internal stability is shown
in Figure 2.2.2 and sliding failure in external stability is shown in 2.2.3.
The failure pattern is dependent on the ground condition, the characteristics of the column
and the loading condition. The Swedish design method for stabilized ground assumes that
the shear failure pattern will occur (failure mode h).
The main failure patterns in external stability are a sliding pattern and a tilting pattern.
Mostly, a tilting failure pattern (Figure 2.2.3) will take place as discussed in paragraph
2.3. In the tilting failure mode, also called collapse failure mode, the columns and the clay
are assumed to deform as simple shear due to the imbalance between active and passive
earthe pressures acting on the side boundaries of the improved area [23]. When the shear
strength of the clay ground decreases with depth, a sliding failure pattern can take place.
Research on embankment supported by deep mixing columns 14
A sliding failure pattern can also occur using floating type improved ground. This cases are
more seldom. Although, the Swedish design method for stabilized ground assumes that a
sliding failure pattern will occur. The current Swedish design method for stabilized ground
provides reasonable failure mechanisms as long as the column strength is a relatively low
value where the improved ground is expected to fail by a shear pattern [23], as discussed
in paragraph 2.3. However, when the column strength is a relatively high value, where the
improved ground is expected to fail by tilting pattern, the current Swedish design method
for stabilized ground will overestimate the stability of the embankment [23].
Ground deformation at ground failure for unimproved ground as well as for improved
ground are shown in Figure 2.3.2 and 2.3.3. In the unimproved ground, a large deforma-
tion can be noticed at the shallow layer while there are no deformations at deeper layers.
A slip circle ground deformation takes place. Contrary to the unimproved ground, in the
improved ground it can be seen that all the columns tilt at the toe with negligible defor-
mation at the bottom.
Numerical analysis 15
In order to create a sliding failure pattern instead of a collapse failure pattern, soil prop-
erties were adapted. When the shear strength of the clay ground decreases with depth
instead of being constant throughout the depth of the clay layer, a sliding failure pattern
can be observed (Figure 2.3.4). A ground condition in which the shear strength decreases
with depth is seldom found in practice [1].
A sliding failure pattern can also be obtained by using floating type improved ground. The
deep mixing columns do not reach the stiff layer but penetrate partially in the clay ground
[1]. Under this condition, a sliding failure pattern can take place instead of a collapse
failure pattern when the column height is relatively small (Figure 2.3.5).
In a parametric study, performed by Kitazume M. [23], was found that the failure mode
of the improved ground is dependent on the column strength. A sliding failure mode is
Numerical analysis 16
Figure 2.3.4: Sliding failure pattern-shear strength decreases with depth [1]
Figure 2.3.5: Sliding failure pattern when a floating type improved ground is used [1]
developed when the column strength is relatively low and a collapse failure mode is devel-
oped when the column strength is relatively high [23].
External stability For the isolated columns, without the shallow layer reinforcement,
sliding is not observed as the main failure mechanism but a tilting failure is. When a
shallow layer is used to reinforce the isolated high-strength columns, a sliding pattern in
external failure is observed instead of tilting failure pattern. Reinforcement of the columns
by using a shallow layer changes the failure pattern from a tilting pattern to a sliding
pattern.
The failure pattern of isolated columns supporting an embankment with shallow layer
(with same compressive strength as the columns) are studied afterwards. In this case, the
columns experience a bending failure with failure points at the connection between the
shallow layer and the columns. Except for this failure point at the connection there were
no more failure points observed along the column depth. A tilting deformation took place
[2]. This is shown in Figure 2.3.8.
Figure 2.3.6: Model setup using a shallow layer to reinforce a group of isolated columns
[2]
Figure 2.3.7: Horizontal displacement varying with column strength of the first column
(a) and the last column (b) [2]
Figure 2.3.8: Deformation of low-strength columns (a) and high-strength columns (b)
with shallow layer [2]
Adams investigated a number of cases studies, in which shear walls were used to support
levees, through numerical analyses. Adams showed that the factor of safety for a slip fail-
ure pattern is higher when shear panels with overlapping columns are used compared to
the case in which isolated columns at the same area replacement ratio are used [28], [3],
[30], [31].
Filz et al. [7] compared the safety factor obtained by limit equilibrium analysis and nu-
merical stress-strain analysis. The limit equilibrium analysis only permits a shear failure
while the numerical stress-strain analysis also allows bending failure of the columns. When
isolated columns are used, Filz et al [7] showed that the factor of safety value is higher from
limit equilibrium analyses than from numerical analysis [19]. This is shown in Figure 2.3.9.
The factor of safety values were calculated again when continuous panels of overlapping
columns instead of isolated columns are used. The area replacement ratio when using
continuous panels is the same as in the case in which isolated columns are used. The
continuous panels of overlapping columns consist of weaker joints located at the overlap
positions of the columns. The joints were assumed to have half the strength of the intact
portions of the wall [19]. Limit equilibrium analysis gives the same safety factor as in the
case in which isolated columns are used. This results from the fact that limit equilibrium
analysis only considers composite shearing and the area replacement factor in both cases
Numerical analysis 19
is the same so the limit equilibrium method does not distinguish between the geometries
of boths cases. [19]. Numerical analysis takes the difference between both cases in consid-
eration since it also allows bending failure. The factor of safety calculated by numerical
analysis is higher when continuous panels are used [19]. This is shown in Figure 2.3.10.
It is verified that shear walls are more effective than isolated columns at the same area
replacement ratio.
Filz et al. [19] performed a study on the use of deep mixing shear panels to support levees
and floodwalls. Based on a number of case studies, Filz et al. presents simplified methods
to analyse and design for external and internal stability of deep mixed shear walls to sta-
bilize levees and floodwalls [19].
Figure 2.3.9: Safety factor calculated with limit equilibrium method and numerical
analysis. Isolated deep mixing columns are used [19]
Laboratory tests 20
Figure 2.3.10: Safety factor calculated with limit equilibrium method and numerical
analysis. Continuous deep mixing panels are used [19]
The shear resistance and failure mechanisms of lime-cement columns in soft clay were in-
vestigated by Larsson et al. [32], [33]. by performing laboratory shear box model tests.
The shear box test is shown schematically in Figure 2.3.11. The main failure mechanisms
of the lime-cement columns when subjected to lateral loading is bending failure. The shear
box model tests verified that using a shear wall pattern with overlapping columns is more
effective than using isolated columns in the clay [32],[28]. The shear wall pattern with
overlapping columns increased the shear resistance [33].
A centrifuge model test consists of making a scale model and load it onto the end of a
centrifuge. The centrifuge will produce equal confining stresses in model and prototype.
Therefore, the reasonable assumption that strains and deformations are also equal in model
and prototype can be made.
Kitazume et al. [1] performed a series of centrifuge model tests, slip circle analyses and
FEM analyses to establish the failure envelope of each failure pattern [1]. The deformation
of the improved ground, as well as the the deep mixing columns were investigated through
a series of centrifuge model tests and numerical calculations. The conclusions of the tests
performed by Kitazume [1] will be summarized here.
A series of centrifuge model tests were performed. An example of the model ground setup
Laboratory tests 21
is shown in Figure 2.3.12. More detailed description about the model setup can be found
in the study of Kitazume et al. [1].
External stability In the unimproved ground model, a slip circle deformation can be
seen (Figure 2.3.13). Contrary to the unimproved ground, the improved ground model
does not show a slip circle failure (Figure 2.3.14).
In the improved ground model it can be seen that the columns tilt like dominos. The
improved area deforms uniformly as a shear failure since the inclination angle is the same
throughout the columns. The use of deep mixing columns changes the ground failure
pattern from a slip circle failure to a collapse failure [1].
The results of the FEM analysis (described in paragraph 2.3.1) correspond well to the
results observed in the centrifuge model tests.
Internal stability Similar centrifuge model tests were performed by Kitazume et al.
[34] as described in Figure 2.3.12. More detailed description about the model setup can
be found in the study of Kitazume et al. [34]. The development of the bending moment
distribution in the deep mixing columns were measured in detail to investigate the failure
mechanisms.
It can be concluded from the performed tests that the deep mixing columns fail one by one
by bending failure mode. The deep mixing columns do not fail simultaneously as assumed
by the Swedish design method for stabilized ground. Bending points can be noticed in
Figure 2.3.15. When the improvement width is small, the forefront column will fail first.
Afterwards, the other columns will fail in sequence from the forefront column towards the
rearmost column. When the improvement width becomes larger, the forefront column will
fail in bending first. Afterwards, the second and third columns will fail. Subsequently, the
Laboratory tests 23
rearmost column fails due to large ground settlement. It is of interest to note that the lo-
cation of the bending failure is shallower in the low strength column than the high strength
column, and shallower in the rear side columns (furthest away from the embankment toe)
than the front side columns (closest to the embankment toe) [34].
Nguyen Binh T. T. [2] performed a centrifuge model tests to investigate the effect of a
shallow layer on the failure mechanisms of the deep mixing columns. A detailed description
of the model set up can be found in the study of Nguyen Binh T.T. [2].
External stability An agreement with the study of Kitazume et al. was confirmed,
namely that the collapse failure mechanism occurs when isolated columns are used to
reinforce an embankment (without shallow layer). When a shallow layer is applied, a
sliding failure pattern was found instead of a tilting failure pattern [2].
Yang et al. [35] investigated by numerical analyses the behaviour of an embedded im-
proved soil raft. An embedded improved soil raft (a raft of overlapping short columns) is
constructed by using soil cement columns that form a continuously improved composite
ground [28]. During excavation, the columns are loaded laterally by the inward-moving re-
taining wall. Thus,the mobilised mass properties in lateral direction are important for the
design [35]. Yang et al. examined the mechanisms of how mass properties of the improved
are mobilized and their relation to the elemental properties, the variation of properties
within a column and the geometrical arrangement of the columns [35]. The analysis and
the simulation of a case study showed that soil-cement columns used for supporting exca-
vations should be constructed with overlap rather than just in contact with each other [28].
Figure 2.4.1: 3D-model of an excavation supported with rows of dry deep mixing columns
[4]
2.5 Summary
This chapter provided an overview of the deep mixing method, specifically about the cur-
rent state of practice of the deep mixing method and its applications. The main application
of the deep mixing method in Sweden is to support railway or roadway embankments us-
ing isolated deep mixing columns or continuous shear panels. Recently, the deep mixing
columns are also used in rows to support excavations. In the Scandinavian countries, the
dry deep mixing method is mostly used to install the columns. The wet deep mixing
method has been extensively used in Japan, particularly in marine projects [17]. Nowa-
days, it also has subsequently been used for land applications.
The current Swedish design method for stabilized ground assumes that the improved soil
volume behaves as a composite soil volume.Two failure patterns are assumed: sliding
failure in external stability and rupture breaking failure in internal stability [1]. In internal
stability, the current Swedish design method for stabilized ground assumes that the columns
will fail by shearing all at the same time. In external stability, the columns and clay
will fail by sliding according to the Swedish design method for stabilized ground. The
composite soil moves horizontally as a block on a stiff layer. A number of studies showed
that sliding failure as internal failure mechanism and rupture breaking failure as external
failure mechanism are not the only failure mechanisms that can occur. The majority of
the studies about deep mixing columns supporting an embankment are numerical tests or
small scale model laboratory tests. The results of the most important studies performed
about this topic were summarised in this chapter. In most studies, a tilting failure pattern
could be observed as the main external failure mechanism and a bending failure as the
main internal failure mechanism of the improved ground. A sliding failure pattern can
occur when the deep mixing columns have low strength properties or when floating type
improved ground is used. Reinforcement of the columns by using a shallow layer changes
Summary 26
the failure pattern from a tilting pattern to a sliding pattern. There was showed that
continuous shear panels oriented perpendicular to the centreline of the embankment are
more efficient for the stability than isolated columns.
The analytical method according to the Swedish design method for stabilized soil represents
the overlapping columns and clay in between the deep mixing rows as a composite soil
volume with an average strength and stiffness. The Swedish design method does not
take into account any weaker vertical joints caused by the reduced width of the wall
at the overlap between adjacent columns. These vertical joints can be made weaker by
misalignment during construction, which reduces the efficiency of the overlap [3]. Ignat et
al. [4] verified the importance of the quality of overlap in a model on the failure mechanism
that will develop. Therefore, a comparison will be made in this study between an analytical
model that does not take weaker vertical joints into account and an analytical model in
which the overlap is represented by a decreased strength and stiffness. A comparison will
also be made with a numerical model that consists of a composite soil volume containing
vertical weaker joints.
Chapter 3
General methodology
• Analytical model: Internal stability- Reduction of strength and stiffness of 75% and
50% in the overlap zones.
An analytical method to obtain safety factors for shear failure is made, taking into
account the reduced strength and stiffness at the overlap between adjacent columns.
Safety factors for shear failure are calculated with limit equilibrium method using
program SLOPE.
• Numerical model.
A numerical 2D plane strain model is made to calculate the safety factors for internal
stability and to see which failure mechanism will develop.
• Sensitivity analysis.
The influence of the strength and stiffness in the overlap zones on the failure mech-
anisms obtained are discussed.
This study is not based on any specific site condition, so a typical ground condition is chosen
from literature with the care of choosing suitable parameters for the size and dimension of
the model ground, deep mixing columns, embankment etc. as well as the soil properties.
27
Geometrical model 28
The properties used for both the analytical model and the numerical model are represented
in this chapter.
The soil consists of a normally consolidated very soft clay layer of 10 m overlaying a stiff
frictional soil at the bottom. The stiff frictional soil is modelled with a thickness of 4m.
The ground water table is set 1m below the ground surface. In order to avoid boundary
effects, the length of the model is chosen to be 60 m. The geometry of the embankment
and underlaying soil layers is shown in Figure 3.1.1.
The length of the deep mixing columns is 9,5m. During the installation of the columns,
it is very unlikely that the installation tool will be able to penetrate the moraine layer
deeply. Since the binder is transported into the soil through a hole located above the
mixing tool (as explained in paragraph 2.1.1.1 and shown in Figure 2.1.2), the deep mixing
columns created by this tool will probably not penetrate the moraine layer but will end
approximately 30-50 cm above the moraine layer. Therefore are the columns shown in
Figure 3.1.1 not reaching the moraine layer but end 0,5m above the moraine layer. The
columns are installed in rows with a distance of 3m (srow ) between the different rows. The
diameter of the deep mixing columns is equal to 0,6m. The centre-to-centre distance of
the columns is equal to 0,5m (scol ) so there is an overlap of 0,1m. The arrangment of the
deep mixing columns in rows is shown in Figure 3.1.2
Material properties 29
Figure 3.1.1: Geometry of the modelled embankment and underlaying soil layers.
3.2.2. In real soils, the stiffness depends significantly on the stress level, which means that
the stiffness generally increases with depth. The start of the increase in this case is at
ground surface, where z equals 0m.
Table 3.2.1: Material properties used for the analytical model and the numerical model.
For the deep mixing columns, the Young’s modulus and undrained shear strength can be
deduced from the compressive strength following equation 3.2.1 and 3.2.2.
E = 100σc (3.2.1)
σc
su = (3.2.2)
2
The tensile strength can be found by using equation 3.2.3.
σt = 0, 10 · su = 0, 05 · σc (3.2.3)
The earth pressure at rest can be calculated following equation 3.2.4.
0
K0 = sin(φ ) (3.2.4)
For the soft clay, a value of 0,6 is used for the earth pressure at rest since the clay is a very
soft, normally consolidated clay.
Poisson’s ration can be calculated following equation 3.2.5
K0
ν= (3.2.5)
1 + K0
The properties of the embankment material are summarised in Table 3.2.2. The shear
strength of the embankment is set to a minor value of 1kPa to avoid local tensile failure
at the embankment slope during the Finite Element analyses [2].
Material properties 31
The above explained condition is used as the standard condition for the analytical model,
as well as the numerical model.
Chapter 4
Analytical calculations
In this chapter, first the safety factors for extrusion of soft ground between shear panels are
calculated in sequencing construction stages. The safety factors are calculated for different
distances between the rows of deep mixing columns. In this way, you can determine the
minimum distance srow that has to be present. The method to obtain this analytical
solution will be described here, as well as the obtained results.
32
Extrusion of soft ground between shear walls 33
When extrusion of soft ground between shear walls happens, the clay between the rows
of deep mixing columns will fail in shear. A cross section is made according to Figure
4.1.1 (green line) and the most critical slip surface is determined with the limit equilibrium
method, using the program SLOPE.
Figure 4.1.1: The shear failure of the cross section at the green line is determined using
limit equilibrium method [4].
To calculate the factor of safety with the limit equilibrium method, using the program
SLOPE, a combined analysis was performed. In the combined analysis the lowest strength
of either the undrained or the drained case is used. The parameters of the clay can be
found in Table 3.2.1 in paragraph 3.2. The drained parameters of the clay are given in
equation 4.1.1.
The factor of safety for extrusion can be calculated using equation 4.1.2.
srow · ΣM omentsresisting + 2 · A · su,clay · c
Fp = (4.1.2)
srow · ΣM omentsactivating
Where
• A is the area of the sliding mass of clay, shown in Figure 4.1.2b in light blue colour
• c is the distance from the centre point of the slip circle down to the centre of gravity
of the area, shown in Figure 4.1.2b
Extrusion of soft ground between shear walls 34
To calculate the area of the sliding mass of clay and the distance c, equation 4.1.3-equation
4.1.5 and figure 4.1.2 are used.
p
α = 2 · arccos(1 − ) (4.1.3)
r
Where p is the distance from the ground surface to the bottom of the slip circle, shown in
figure 4.1.2 and r is the radius of the slip circle, obtained from SLOPE.
A = 0, 5 · r2 · (α − sin(α)) (4.1.4)
4 · r · sin3 (α/2)
c= (4.1.5)
3 · (α − sin(α))
Extrusion of soft ground between shear walls 35
(b) Critical slip surface for the shear failure of clay ground for an embankment height of 7m.
Figure 4.1.2: Parameters used to calculate the factor of safety for extrusion of clay.
Extrusion of soft ground between shear walls 36
The factor of safety is reduced to account for 3D effects according to equation 4.1.6, [11].
Fp
F3D = F2D + 0, 75 · ( − 1) (4.1.6)
F2D
Where
For the undrained shear strength is the minimum value of 10kPa used in the calculation
of the safety factor in equation 4.1.2. Therefore, the safety factor will be slightly under-
estimated since the undrained shear strength of the clay normally increases with depth.
The shear strengths on the vertical sides of the slip surface in the embankment are not
considered. This is a conservative assumption.
The parameters used to calculated F3D when the embankment has a height of 3m are
summarised in table 4.1.1.
Parameters values
r [m] 13,4
p [m] 4,94
α [radians] 1,775
A[m2 ] 71,45
ΣM omentsresisting [kNm] 6492,3
ΣM omentsactivating [kNm] 9377
c [m] 10,47
srow [m] 3
Fp 1,224
F2D 0,692
F3D 1,27
The safety factors for extrusion for sequencing construction stages are calculated for a
distance between the shear panels of 1,5m; 3m and 4,5m, shown in Table 4.1.2.
Internal stability 37
Table 4.1.2: Safety factors for extrusion of clay between shear walls for different distances
between shear panels.
The factor of safety for a distance of 3m and 4,5m between the shear panels increases when
the embankment height decreases. The factor of safety for a distance of 1,5m between the
shear panels fluctuates around the same values. This happens because srow is small so the
area of the sliding mass of the clay, the radius and the distance p have a great influence
on the safety factor. The critical slip circle for each construction stage is different and
determined by SLOPE. The factor of safety for extrusion has to be more than 1,3 to have
a safe design according to the Swedish design code [9] and Filz et al. [19]. Since the factor
of safety is slightly underestimated, a safety factor of 1,27 for a distance of 3m between
shear panels can be enough. Therefore, a distance between the shear panels of 3m can be
chosen. The following calculations in this study are done for srow equal to 3m.
volume due to a reduction of strength and stiffness in the overlap between adjacent columns.
The method used by the Swedish design code is described in paragraph 4.2.1. The method
used to include reduction of strength and stiffness of the overlap between adjacent columns
in the undrained and drained shear strength of the composite soil volume is described in
paragraph 4.2.2.
2
Dcolumn,1
Asegment = · (α − sin(α)) (4.2.3)
8
2
π · Dcolumn
Acolumn,1 = − 2 · Asegment (4.2.4)
4
Acolumn,1 is the area of the column including overlap (green area in Figure 4.2.1).
Factor a can be calculated as the ratio between the column area and the area of the soil,
using equation 4.2.5.
Aslice
a= (4.2.5)
c1 · c2
Where:
No reduction of strength and stiffness at the overlap 39
Figure 4.2.1: Angle α, area of a circle segment in purple, area of the column, Acolumn,1 in
green [10].
Parameters values
Dcolumn,1 [m] 0,6
c1 [m] 0,5
c2 [m] 3
α [radians] 1,171
Asegment [m2 ] 0,0113
Acolumn,1 [m2 ] 0,260
Aslice [m2 ] 0,520
a 0,173
The cohesionintercept in the deep mixing columns is dependent on the place of the columns:
whether they are used in the active zone, direct shear zone or passive zone. The factor
β will differ dependent on where the composite soil volume is situated. This is shown in
Figure 4.2.2. The factor β is equal to 0,4 in the active zone, 0,1 in the direct shear zone
and 0 in the passive zone. To find those different zones in the calculation with SLOPE,
several iterations were performed.
Figure 4.2.2: Active zone, direct shear zone and passive zone for a soil volume developing
a shear failure [10].
The drained shear strength of the clay and the deep mixing columns can be calculated
according to equation 4.2.7 and 4.2.8.
σv0 = σv − uw (4.2.9)
z is equal to zero at the ground surface level. The water level is lying 1 meter below the
ground surface.
The total vertical stress is calculated using equation 4.2.11
For an embankment with a height of 7m, the undrained shear strength of the clay, the
deep mixing columns and the average shear strength is summarized in Table 4.2.3.
Where:
• su,column is the undrained shear strength of the deep mixing columns, equal to 100kPa
• su,overlap is the undrained shear strength of the overlap, assumed to be equal to 50%
of su,column = 50kP a or 25% of su,column = 25kP a
• su,clay is the undrained shear strength of the clay ground, equal to 10+1,5z [kPa]
asc · (Ac − 2 · Ao )
x= (4.2.14)
Ac
Reduction of strength and stiffness at the overlap 42
The weighting factor y is calculated using equation 4.2.15. In this formulas you weight the
area of the column without overlap and the area of the two overlaps to the total column
area.
aso · 2 · Ao
y= (4.2.15)
Ac
Where:
• asc is the area replacement ratio of the columns
e = d − scol (4.2.16)
c = d · sin(α) (4.2.18)
The area of the overlap zone between the columns, A0 , is calculated following equation
4.2.19.
d (d − e) · c
Ao = 2 · (α · ( )2 − ) (4.2.19)
2 4
The area of the columns without overlap, Ac , is calculated following equation 4.2.20.
d
Ac = π · ( )2 − 2Ao (4.2.20)
2
The effective width of overlap zone, bo2D , is calculated following equation 4.2.21.
Ao
bo2D = (4.2.21)
c
The effective width of columns without overlap, bc2D , is calculated following equation
4.2.22.
Ac
bc2D = (4.2.22)
d
Reduction of strength and stiffness at the overlap 43
The area replacement ratio of the overlap zone, aso , is calculated following equation 4.2.23.
c
aso = (4.2.23)
srow
The area replacement ratio of the columns, asc , is calculated following equation 4.2.24.
Ac
asc = (4.2.24)
srow · (d − e − bo2D )
Figure 4.2.3: Representation of the different factors needed to calculate the weighting
factors x and y [4].
The undrained shear strength of the composite soil volume when a reduction of strength
and stiffness of 50% in the overlap zones is assumed is calculated using equation 4.2.25.
su = x · su,column + 0, 5y · su,column + (1 − x − y) · su,clay = 24, 22 + 1, 24z[kP a] (4.2.25)
The undrained shear strength of the composite soil volume when a reduction of strength
and stiffness of 75% in the overlap zones is assumed is calculated using equation 4.2.26.
su = x · su,column + 0, 25y · su,column + (1 − x − y) · su,clay = 23, 69 + 1, 24z[kP a] (4.2.26)
The sum of the weighting factors x and y is equal to 0,173, the weighting factor a calculated
in equation 4.2.5. This means that when there is no reduction of strength and stiffness in
the overlap between adjacent columns that the undrained shear strength calculated using
this method is equal to the undrained shear strength calculated with the method according
to the Swedish design code for stabilized ground described in paragraph 4.2.1.
Parameters Values
d [m] 0,6
srow [m] 3
scol [m] 0,5
e [m] 0,1
α [◦ ] 33,557
c [m] 0,332
Ao [m] 0,023
Ac [m] 0,238
bo2D [m] 0,068
bc2D [m] 0,396
aso 0,111
asc 0,183
x 0,15
y 0,021
Table 4.2.5: Safety factors calculated with the limit equilibrium method, combined
analysis.
Height of the embankment (m) Safety factor Safety factor Safety factor
no reduction 50% reduction 75% reduction
Phase 15 7 1,10 1,07 1,05
Phase 14 6,5 1,16 1,12 1,11
Phase 13 6 1,23 1,19 1,17
Phase 12 5,5 1,31 1,27 1,25
Phase 11 5 1,41 1,36 1,34
Phase 10 4,5 1,52 1,47 1,45
Phase 9 4 1,77 1,71 1,68
Phase 8 3,5 1,84 1,78 1,75
Phase 7 3 2,08 1,99 1,96
Phase 6 2,5 2,40 2,31 2,28
Phase 5 2 2,88 2,54 2,52
Phase 4 1,5 3,06 2,96 2,98
Phase 3 1 4,45 4,37 4,36
Phase 2 0,5 7,38 7,23 7,17
Figure 4.2.4: Example of shear failure in both the stabilized soil and the clay. The
embankment has a height of 7m and no reduction of strength and stiffness of the overlap
is included in the undrained shear strength of the composite soil in this case.
Calculation safety factor 46
Table 4.2.6: Safety factors calculated with the limit equilibrium method, undrained
analysis.
Height of the embankment (m) Safety factor Safety factor Safety factor
no reduction 50% reduction 75% reduction
Phase 15 7 1,10 1,07 1,05
Phase 14 6,5 1,16 1,12 1,11
Phase 13 6 1,23 1,19 1,17
Phase 12 5,5 1,31 1,27 1,25
Phase 11 5 1,41 1,36 1,34
Phase 10 4,5 1,52 1,47 1,45
Phase 9 4 1,77 1,71 1,68
Phase 8 3,5 1,84 1,78 1,75
Phase 7 3 2,08 1,99 1,96
Phase 6 2,5 2,40 2,31 2,28
Phase 5 2 2,88 2,77 2,73
Phase 4 1,5 3,62 3,47 3,42
Phase 3 1 5,33 5,11 5,02
Phase 2 0,5 9,63 9,26 9,12
It can be seen that the safety factor is lower when a reduction of strength and stiffness is
present in the overlap zones due to misalignment during construction. The safety factors
decrease with an increasing reduction in strength and stiffness in the overlap zones. The
safety factors of the undrained and combined analysis are the same up to an embankment
height of 2m. This make sense because the undrained shear strength is always lower
then the drained shear strength so the undrained shear strength is used in the combined
analysis to calculate the safety factor. When the embankment has a height lower than
2m, the drained shear strength is lower than the undrained shear strength. Therefore, the
safety factor calculated in the undrained analysis is higher than the safety factor calculated
in the combined analysis.
It can be concluded that it is better to include the reduced quality of overlap in the
analytical calculations, following the method proposed in this study, in order to get a safe
design because the safety factor decreases when the quality of overlap reduces.
Chapter 5
Numerical calculations
Numerical analysis using a 2D plane strain model was performed to investigate the failure
behaviour of deep mixing columns supporting a highway or railway embankment. The
three-dimensional columns are represented by a composite soil volume consisting of vertical
joints representing the overlap between adjacent columns. A comparison is made between
the analytical solution and the results from the numerical model. Afterwards, a sensitivity
analysis of the numerical model is performed in which the influence of the quality of the
overlap of adjacent columns is studied. The condition explained in paragraph 3.1 is used
as the standard.
The finite element program PLAXIS 2D 2016 was used in this study to model the embank-
ment supported by deep mixing columns.
In order to calculate the dimensions of the 2D model, namely the width of the columns
excluding the overlap zone, bc2D , and the width of the vertical joints representing the over-
lap zone, bo2D , the procedure described by Adams [3] and Ignat et al. [4] was used.
The parameters that have to be chosen are the diameter of the columns, d, the distance
between the column panels, srow , and the distance between the adjacent columns, scol . The
47
Geometrical model 48
other parameters needed to calculate the width of the columns in the 2D model as well
as the width of the vertical joints, can be deduced from this three main parameters. The
calculation method is explained in 4.2.2.1. Equation 4.2.16-4.2.24 should be used. The
different parameters calculated are summarized in Table 5.1.1.
The three dimensional situation, as well as plan view are shown in Figure 5.1.1. The two
dimensional model is shown in 5.1.1c.
Geometrical model 49
Figure 5.1.1: Plan view, three dimensional situation and two dimensional model [3], [4]
Material model 50
The linear elastic perfect plastic Mohr-Coulomb model involves five input parameters:
• Angle of dilatancy, ψ
For undrained materials,the Mohr-Coulomb model may be used with the friction angle, φ
set to zero and the cohesion set to the undrained shear strength, su .
The basic idea of an elastic perfectly plastic model is that the strain rates are decomposed
into an elastic and plastic part, [36], following equation 5.2.1. This is shown in Figure
5.2.2.
= e + p (5.2.1)
Figure 5.2.2: Strain rates are decomposed in an elastic and plastic part [36]
Plasticity is associated with the development of irreversible strains. Plastic yielding will
occur when the yield function, which is a function of stress and strain, equals zero. The
condition that the yield function equals zero can be presented by a surface in principal
stress space. A perfectly plastic model means that the model has a fixed yield surface,
which means that the yield surface is fully defined by model parameters and not affected
by straining. For stress states represented by points within the yield surface, the behaviour
is elastic so all the strains are reversible, and obeys Hooke’s law for isotropic linear elasticity
[36], [37]. The Mohr-Coulomb yield condition is an extension of Coulomb’s friction law
to general states of stress. In fact, this condition ensures that Coulomb’s friction law is
obeyed in any plane within a material element [36]. The Mohr-Coulomb failure criterion
expressed in terms of effective stress follows equation 5.2.2.
0 0
τf = c0 + σ · tan(φ ) (5.2.2)
Material properties 52
Where
• c’ is the cohesion
0
• φ is friction angle, based on effective stress
0
• σ normal effective stress on the failure plane
• τf is the shear strength
Failure from shear will occur when the shear stresses on a plane reaches the value given
by equation 5.2.2. When formulated in terms of principal stresses, the full Mohr-Coulomb
yield condition consists of six yield functions. They will not be discussed in detail here,
but can be found in literature [36], [37].
Because soil can sustain none or only very small tensile stresses, a tension cut-off is used
in PLAXIS. The tension cut-off introduces three more yield functions. Therefore, it is
important to give the tensile strength of the material as an extra parameter in PLAXIS.
You have to keep in mind that there are some problems with the Mohr-Coulomb model.
The Mohr-Coulomb model cannot generate pore pressure and stresses correctly. Further-
more, the columns are in fact heavily consolidated compared to the normally consolidated
soft clay. Therefore, there should be a dilatant behaviour (you should get negative pore
pressures) but the Mohr-Coulomb model can not considerate this.
Where
The composite Young’s modulus of the columns can be calculated following equation 5.2.4.
Where
The composite shear strength of the overlap zone can be calculated following equation
5.2.5.
Where
The composite Young’s modulus of the overlap zone can be calculated following equation
5.2.6.
Eo2D = Eo · aso + Es · (1 − aso ) (5.2.6)
Where
• Eo2D is the composite Young’s modulus of the overlap zone
• Eo is the Young’s modulus of the overlap
• aso is the area replacement ratio of the overlap zone
• Es is Young’s modulus of the soil between the columns
The composite tensile strength of the columns can be calculated following equation 5.2.7.
σt,c2D = σt,c · asc + σt,s · (1 − asc ) (5.2.7)
Where
• σt,c2D is the composite tensile strength of the column zone
• σt,c is the tensile strength of the columns
• asc is the area replacement ratio of columns
• σt,s is the tensile strength of the soil between the columns, equal to zero
The composite tensile strength of the overlap zone can be calculated following equation
5.2.8.
σt,o2D = σt,o · aso (5.2.8)
Where
• σt,o2D is the composite tensile strength of the overlap zone
• σt,o is the tensile strength of the overlap
• aso is the area replacement ratio of the overlap zone
The properties of the composite soil volume are summarised in Table 5.2.2 when using deep
mixing columns with a compressive strength of 200kPa, a Young’s modulus of 20 000kPa,
an undrained shear strength of 100kPa and a tensile strength of 10kPa.
Reduction su,c2D [kPa] Ec2D [kPa] su,o2D [kPa] Eo2D [kPa] σt,c [kPa] σt,o [kPa]
0% 26,5+1,2z 5709,0+306,2z 19,9+1,3z 4434,7+333,5z 1,8 1,1
50% 26,5+1,2z 5709,0+306,2z 14,4+1,3z 3329,2+333,5z 1,8 0,6
75% 26,5+1,2z 5709,0+306,2z 11,7+1,3z 2776,4+333,5z 1,8 0,3
Definition of failure 55
The global safety factor is defined as the ratio of the collapse load to the working load, as
seen in Equation 5.2.9 [38].
Smaximumavailable
SF = (5.2.9)
Sneededf orequilibrium
c − σn tan(φ)
SF = (5.2.10)
cr − σn tan(φr )
Where c and φ are the input strength parameters and σn is the actual stress component.
The parameters cr and φr are reduced strength parameters that are just large enough to
maintain equilibrium. The principle described above is the basis for a safety analysis in
PLAXIS to calculate the global safety factor [38]. In this approach, the cohesion and the
tangent of the friction angle are reduced in the same proportion, as seen in Equation 5.2.11
[38].
c tan(φ) X
= = Msf (5.2.11)
cr tan(φr )
P
The reduction of the strength parameters is controlled by the total multiplier Msf . This
parameter is increased in a step-by-step procedure
P in PLAXIS until failure occurs. The
safety factor is then defined as the value of Msf at failure [38].
5.3 Results
The results obtained by the numerical model are given in this chapter.
The deviatoric strain contourplot is shown in Figure 5.3.2. You can see that a shear failure
zone develops. A slip circle can be noticed. The largest strains are situated beneath the
composite soil volume. When you zoom in on the composite soil volume, you can see that
a large amount of the deviatoric strain can be noticed in the overlap zones close to the
embankment toe.
Analysis of the failure mode 58
Figure 5.3.2: Deviatoric strain at failure when there is a reduction of strength and
stiffness in the overlap zones of 50% .
The Mohr-Coulomb plastic points at failure are shown in Figure 5.3.3. The plastic points
will develop first in the overlap zones, when increasing load, plastic points will develop in
the column zones of the composite soil volume as well.
Analysis of the failure mode 59
The relative shear stress, τrel , is the ratio between the mobilized shear stress, τmob , and the
maximum shear stress, τmax . When the relative shear stress reaches the value 1, the full
shear strength of the material is mobilized and the material behaviour changes from linear
elastic to perfect plastic [4]. The relative shear stress in the composite soil volume in the
phase in which the load is equal to 25% (phase 4), 50% (phase 7) and 75% (phase 10) of
the ultimate load are shown in Figure 5.3.4a, Figure 5.3.4b and Figure 5.3.4c respectively.
The legend is the same for all figures and is shown in Figure 5.3.4a. The area in the black
rectangle in Figure 5.3.3 is shown in detail in these figures. The ultimate load, qult is
equal to approximately 132kPa. The relative shear stress is always higher in the overlap
zones compared to the column zones. With an increasing load, yielding will be reached in
more and more overlap zones. Yielding is reached first in the overlap zones close to the
embankment toe, and it propagates more and more to the middle of the embankment when
the load increases. When yielding is reached in the overlap zones, further load increase
will result in plastic deformations in the column zones of the composite soil volume.
Analysis of the failure mode 60
(a) Relative shear stress when the applied load is equal to 25% of the ultimate load.
(b) Relative shear stress when the applied load is equal to 50% of the ultimate load.
(c) Relative shear stress when the applied load is equal to 75% of the ultimate load.
Figure 5.3.4: Relative shear stress of the composite soil volume in different phases,
quality of overlap: 50%. The area of the composite soil volume going form x=10m until
x=40m is shown.
fail in bending. In this case, one plastic hinge in each column zone exists. The change of
inclination is marked on figure 5.3.5 using a little cross. The plastic hinges start to develop
closest to the embankment toe and extend until failure occurs in the entire structure. The
horizontal deformation at the point of the embankment toe is equal to 0,10m at failure.
Figure 5.3.5: Deformed mesh for a quality of overlap of 100%, scaled 100 times.
You can see in Figure 5.3.6 that there is an indication of a shear failure zone but there is
no full development of the shear failure. The composite soil volume does not fail in shear,
but there is bearing failure. The load is transferred from the stabilized soil to the clay
ground below and on the right of the stabilized soil volume. When you zoom in on the
composite soil volume, you can see that the deviatoric strain in the overlap zones is not
much larger than the deviatoric strain in the column zones. There is an external failure,
the composite soil volume slides to the right.
Analysis of the failure mode 62
Figure 5.3.6: Deviatoric strain at failure when there is no reduction of strength and
stiffness in the overlap zones.
The Mohr-Coulomb plastic points at failure are shown in Figure 5.3.7 for a quality of
overlap of 100%. The plastic points are located mostly under the composite soil volume.
There is bearing failure because the composite soil volume slides to the right.
Analysis of the failure mode 63
The relative shear stress in the composite soil volume for different phases is shown in
Figure 5.3.8a, Figure 5.3.8b and Figure 5.3.8c. Yielding is reached first in the overlap
zones. Although, the difference between the relative shear stress in the overlap zones and
the column zones is quite small. The difference between the relative shear stress in the
overlap zones and the column zones is larger in the cases in which you have a quality of
overlap of 50% or 75%. Because of the strength and stiffness of the overlap zones, the load
will be transferred to the surrounding soil and there is a bearing failure.
Safety factors 64
(a) Relative shear stress when the applied load is equal to 25% of the ultimate load.
(b) Relative shear stress when the applied load is equal to 50% of the ultimate load.
(c) Relative shear stress when the applied load is equal to 75% of the ultimate load.
Figure 5.3.8: Relative shear stress of the composite soil volume in different phases,
quality of overlap 100%. The area of the composite soil volume going form x=10m until
x=40m is shown.
Table 5.3.1: Safety-factor-values calculated with limit equilibrium method and numerical
analysis.
General discussion
When the composite soil volume becomes stronger, you have to consider that there is a
possibility that the composite soil volume will not fail in shear. It can be seen from the
results of the numerical model that when columns with a shear strength of 100kPa and a
quality of overlap of 100% are used, the slip circle is not fully developed but bearing failure
occurs instead. The load is transferred from the stabilized soil to the clay ground below
and on the right of the stabilized soil volume. There is an external failure, the composite
soil volume slides to the right.
66
How representative is LEM compared to numerical analysis? 67
The difference between the deviatoric strain present in the column zones and overlap zones
is smaller than in the in which there is a quality of overlap of 50%. It can be seen from
the relative shear stress plot that yielding is reached first in the overlap zones. Although,
the difference between the relative shear stress in the overlap zones and the column zones
is smaller than the difference in the cases in which you have a quality of overlap of 50%.
Because of the great strength and stiffness of the overlap zones, the load will be transferred
to the surrounding soil and there is a bearing failure due to the sliding of the composite
soil volume.
Conclusions
• The factor-of-safety-values are smaller for the numerical analyses compared to the
limit equilibrium method. It is already proven by Kitazume et al. [1], [34], [23],
[20] that the limit equilibrium method slightly overestimates the safety factors. This
conclusion is strengthened here. The overestimation of the safety factor is higher for
lower embankment loads. It is recommended that this should be further studied in
future work.
• There is a possibility that the composite soil volume will not fail in shear so that an
analysis using the limit equilibrium method will not be enough. A shear failure mode
develops when the quality of overlap is weak (50% and 25%). An external sliding
failure occurs when the quality of overlap is strong.
• It can be concluded that it is better to include the reduced quality of overlap in the
analytical calculations, following the method proposed in this study, in order to get
a safe design because the safety factor decreases when the quality of overlap reduces.
The Swedish design code does not take any reduction of strength and stiffness in the
overlap zones between adjacent columns into account.
69
Bibliography
[1] Kitazume M. & Maruyama K. (2006) External stability of group column type deep
mixing improved ground under embankment loading. Soils and foundations Vol. 46,
No. 3, 323-340, June 2006.
[2] Binh Nguyen T. T. (2016) Failure of Deep Mixing Column Group reinforced by Shallow
Mixing Layer. Tokyo Institute of Technology, Japan, 2015
[3] Adams T.E. (2011) Stability of levees and floodwalls supported by deep mixed shear
walls: five case studies in the New Orleans area. PhD dissertation, Virginia Polytechnic
Institute and State University, Blacksburg VA.
[4] Ignat R. , Baker S., Larsson S. & Liedberg S. (2015) Two- and three-dimensional
analyses of excavation support with rows of dry deep mixing columns. Computers and
Geotechnics, Vol. 66, p16-30, January 2015.
[5] Charbit, B. (2009). Numerical Analysis of laterally loaded lime/cement columns. Royal
Institute of Technology, Stockholm, Sweden.
[7] Filz, G.M. & Navin, M.P. (2006) Stability of column-supported embankments. Virginia
Transportation Research Council, Charlottesville, Virginia, 64 p.
[8] Broms, B. B. (2004) Lime and lime/cement columns Chapter 8 in Ground Improvement
2nd edition, edited by Moseley M.P. & Kirsch K., London and New York, Spon Press,
2004.
[10] Rolf Larsson (2006) Rapport 17: Djupstabilisering med bindemedels- stabiliser-
ade pelare och masstabilisering. Swedish Deep Stabilization Research Centre, 2006,
Linkping.
70
BIBLIOGRAPHY 71
[25] Han J., Chen J., Hong Z., & Shen S. (2005) Mitigation of levee failures using deep
mixed columns and geosynthetics. Geom Geoeng, Vol. 1, p. 49-55.
[26] Han J., Oztoprak S., Parsons R.L. & Huang J. (2007) Numerical analysis of foundation
columns to support widening of embankment. Computers and Geotechnics, Vol. 34(6),
p.435-48.
[27] Huang J., Han J. & Porbaha A. (2006) Two and three-dimensional modeling of DM
columns under embankments. GeoCongress, Geotechnical engineering in the Informa-
tion technology age, ASCE, Atlanta.
[28] Ignat R. (2015) Field and laboratory tests of laterally loaded rows of lime-cement
columns. Royal Institute of Technology, Stockholm, Sweden.
[29] Kivelö M. & Broms, B.B. (1999) Mechanical behavior and shear resistance of
lime/cement columns. Proceedings of the International Conference on Dry Mix Methods
for Deep Soil Stabilization, Stockholm, pp. 115-121, October 1999.
[30] Adams T.E., Filz G.M. & Navin M. (2009) Stability of embankments and levees on
deep mixed foundations. Proceedings of the international symposium on deep mixing
and admixture stabilization, p.305-100, Kinawa.
[31] Adams T.E., Filz G.M., Cali PR., & Woodward ML. (2008) Stability Analyses of a
Levee on Deep Mixed Columns, Plaquemines Parish, Louisiana. ASCE Geotechnical
Special Publication No. 178; p. 708-15
[32] Larsson S. (1999) Shear box apparatus for modeling chemical stabilized soil introduc-
tory tests. Proceedings of the international conference on dry mix methods for deep soil
stabilization, p. 115-21, Stockholm
[33] Larsson S. & Broms B.B. (2000). Shear box model tests with lime/cement columns
some observations of failure mechanisms. Proceedings of geoeng, 6p Melbourne.
[34] Kitazume M. & Maruyama K. (2007) Internal stablity of group column type deep
mixing improved ground under embankment loading. Soils and foundations Vol. 47,
No. 3, 437-455, June 2007.
[35] Yang T., Tan T.S. & Leung C.F. (2011) Mass behaviour of embedded improved soil
raft in an excavation. Proceeding of he Institution of Civil EngineersGeotechnical En-
gineering, Vol. 164, p.11-25
[36] PLAXIS (2011) Plaxis: Materials Model Manual. Plaxis company, The Netherlands.
[37] Das Braja M. & Sobhan K. (2014) Principles of Geotechnical Engineering. Eight Edi-
tion. United States of America: Cengage Learning, 2014
[38] R. B. J. Brinkgreve, 2011 Plaxis Introductory: Student Pack and Tutorial Manual
2010, Publisher: Plaxis BV
TRITA 17:02
ISSN 1652-599X
www.kth.se