Sunteți pe pagina 1din 236

Technical Committee 211

Ground Improvement
Comit technique 211

Amlioration des sols

General Report of TC 211


Ground Improvement
Rapport gnral du TC 211
Amlioration des sols
Huybrechts N.

Belgian Building Research Institute, BBRI & KU Leuven, Belgium

Denies N.

Belgian Building Research Institute, BBRI, Belgium

ABSTRACT: The present General Report highlights the significant contributions of the papers of the Session of the XVIII ICSMGE
dedicated to Ground Improvement. All papers that have been reviewed are referred (in bold) in the General Report in order to provide
a balanced overview of the entire Technical Session.
This General Report discusses the latest developments and current researches in the field of Ground Improvement (GI) works. The
various GI techniques are classified considering the recent classification proposed by Chu et al. (2009). The papers are then tackled
according to the described GI technique and with regard to the topics that are assessed: execution process, mechanical characterization
of the treated material (in laboratory or in situ), case history, Quality Assurance/Quality Control (QA/QC) activities and design
aspects. Conceptual works and numerical modeling are supported by laboratory and field investigations - with in situ monitoring and
large scale tests. Finally, other references on the topics discussed are also given in the report.
RESUME : Le prsent rapport gnral met en vidence les contributions significatives des articles de la session amlioration des
sols de la 18me CIMSG. Tous les articles revus ont t rfrencs (en gras) dans le rapport gnral de manire fournir une vue
densemble quilibre du contenu de cette session.
Ce rapport discute des derniers dveloppements et des recherches actuelles dans le domaine des travaux damlioration des sols. Les
diffrentes techniques sont classes selon la rcente classification propose par Chu et al. (2009). Les articles sont ensuite abords en
tenant compte de la technique dexcution dcrite et du sujet choisi par les auteurs : procd dexcution, caractrisation mcanique
du matriau trait (en laboratoire ou in situ), cas pratique, activits de contrle et dassurance du point de vue de la qualit et aspects
lis au dimensionnement. Les approches de conception et la modlisation numrique sont supportes par des recherches en laboratoire
et par lexprience de chantier apporte par le monitoring in situ et par les essais en grandeur relle. Finalement, dautres rfrences
concernant le domaine de lamlioration des sols sont aussi indiques.
KEYWORDS: ground improvement/reinforcement, deep mixing, drainage, geosynthetics, grouting, inclusions, vacuum consolidation
1

INTRODUCTION

Ground improvement (GI) is one of the major topics in


geotechnical engineering. With regard to the world population
growth and in response to the expansion needs of our society, it
has become a fast growing discipline in civil engineering as an
alternative allowing construction on soft/weak/compressible
soils. Various specialized ground improvement conferences
have been frequently held in the past and recent years such as
the International Symposium on Ground Improvement
organized by the Technical Committee 211 of the ISSMGE and
recently held in Brussels (Denies and Huybrechts, 2012)
especially with more than 140 papers and 7 General Reports
focusing on GI works. A number of books covering various
topics on ground improvement have been also published in the
past. Most of them are referred in Chu et al. (2009). During the
last decades the importance of the ground improvement market
has enormously increased. New methods, tools and procedures
have been developed and applied in practice. In order to support
this evolution in a scientific way, research programs have been
and are being carried out worldwide, leading to more and better
insights and delivering the basis for the establishment of design
methods, quality control procedures and standards. As a result,
many technical papers on GI works were published in journals
and conference proceedings. It is not possible to mention all.
Separate lists are given on the TC211 website
(www.bbri.be/go/tc211). Major GI techniques have been

documented by the Working Groups of TC211 and are currently


available on this website.
TC211 adopts a classification system as shown in Table 1 in
Chu et al. (2009) with the following categories (and methods):
- A. GI without admixtures in non-cohesive soils or fill
materials (dynamic compaction, vibrocompaction,)
- B. GI without admixtures in cohesive soils (Replacement,
preloading, vertical drains, vacuum consolidation,)
- C. GI with admixtures or inclusions (Vibro replacement,
stone columns, sand compaction piles, rigid inclusions,)
- D. GI with grouting type admixtures (Particulate and
chemical grouting, Deep mixing, jet grouting,)
- E. Earth reinforcement (geosynthetics or MSE, ground
anchors, soil nails,)
This classification is based on the broad trend of behaviors of
the ground to be improved and whether admixture is used or
not. In the following sections, the papers of the Session of the
XVIII ICSMGE dedicated to GI works will be reviewed
according to this classification and with regard to the topics that
are assessed: execution process, mechanical characterization of
the treated material, case history, QA/QC activities and design
aspects. It can already be noted that there is no paper
considering GI without admixtures in non-cohesive soils
(category A) in the present Technical Session.

2417

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

GI WITHOUT ADMIXTURES IN COHESIVE SOILS

In the present Technical Session, six papers can be put in the


category B: GI without admixtures in cohesive soils. They are
mainly related to the subject of consolidation acceleration by
vertical drains combined with surcharge or Vacuum. The
interest seems to be oriented to the approach of smear. ParsaPajouh et al. (2013) address this delicate topic so difficult to
model due to the lack of field parameters. According to the
authors, the smear zone varies between 1.6 and 7 times the drain
radius or 1 to 6 times the mandrel equivalent diameter.
Numerical models are used within the framework of case
studies. Parameters studies confirm their validity. As a result of
their researches, it is recommended to assess the smear zone on
the basis of trial construction with the help of back calculation
process.
Chai and Carter (2013) present a theoretical approach of
Prefabricated Vertical Drains (PVD) and consolidation
combining vacuum pressure and surcharge loading. Using
Hansbos (1981) solution, consolidation parameters of the
smear zone and the undisturbed zone were derived using a
simple equation. Adopting an average well resistance and with
some approximation, the dimensionless parameter quantifying
the effects of PVD spacing, smear zone and well resistance can
be expressed. The study was performed in uniaxial
consolidation condition, which is not in agreement with the real
isotropic character of deformation under Vacuum. Moreover,
the classical assumption of uniform smear zone cannot be
measured. However the pore pressure measurements of the
tested samples are in extreme close concordance with the
prediction confirming the validity of the approach and the
selected parameters.
Indraratna et al. (2013) treat similar subject in conjunction
with a real construction site in the Port of Brisbane where the
consolidation of thick Holocene clays was performed with
PVDs under surcharge and/or Vacuum loading. Variable drain
spacing was selected and analytical solutions were proposed.
For the excess pore pressure dissipation, the same equation as in
Chai and Carter (2013) was adopted. The results demonstrate
that Vacuum combined with preloading would speed up
consolidation compared to preloading alone. Moreover,
Vacuum results in isotropic consolidation increasing the
stability of the surcharge fill (decreasing lateral displacements).
In a similar way, Lee et al. (2013) have also studied the
effect of the smear zone for a consolidation case history in
Busan (South Korea). Modification of Hansbo's analysis is
proposed to study the degree of consolidation considering the
properties of the soil within the smear zone.
As another case history, Islam and Yasin (2013) present an
application of PVDs coupled with preloading used for the
construction of a large container yard in Bangladesh. The soil
profile consists of 4 to 6 m thick silty clay, 8 to 10 m of sand
and silt and 16 m of clayey silt. On the basis of design
requirements, GI of the upper soft clay layer was considered
essential. Five alternatives were assessed and compared. A
solution combining PVD and preloading was adopted for this
site. The settlement under preloading was monitored during the
consolidation phase. Pre and post consolidation SPT tests are
presented to illustrate the efficiency of the technique. It is
believed that dynamic compaction although economical would
not have been technically feasible due to the clayey nature of
the upper fill. However, dynamic replacement in the upper 4 m
with densification of the lower silty sand might have been
technically and financially optimal.
For their part, Jebali et al. (2013) have assessed the theory
of Carillo using three different oedometer tests carried on Tunis
soft soil. Oedometer tests were conducted, conventionally (NF
P94-90-1) for the first test, with a vertical drain allowing only
radial drainage for the second one and finally with a drain

2418

allowing vertical and radial drainage for the last one. Defining
Cr and Cv as the radial and vertical coefficients of consolidation
and Kr and Kv as the coefficients of radial and vertical
permeability, they observed that the often-made assumption of
the equality between the ratios Cr/Cv and Kr/Kv is only valid at
high levels of stress conditions. Moreover, on the basis of
experimental results, the authors demonstrated that the global
degree of consolidation computed with respect of the Carillos
theory can lead to underestimated consolidation times.
The paper of Weihrauch et al. (2013) describes a
combination of GI methods for the improvement of roads in the
HafenCity area in Hamburg. Indeed, in the Hamburg Harbour
area, many roads are lifted with almost 3 m to ensure safety in
case of flooding. Special measures are necessary when the
subsoil contains compressible layers. At the Hongkongstrasse,
three different construction methods have been applied, namely:
- installation of PVD and preloading with sand (settlements of
more than 30 cm have been measured);
- filling with lightweight aggregate: expanded clay (almost no
settlement was observed);
- pile supported embankment including geogrid-reinforced
sand layer (measurements are discussed in another paper).
The different aspects of each method are described. The
conclusion is that when comparing different methods, not only
the absolute costs must be ascertained, but also the project
specific reconstruction, protection and follow-on measures, as
well as the time and flexibility for individual measures, and
their technical feasibility under local conditions.
3
3.1

GI WITH ADMIXTURES OR INCLUSIONS


Rigid inclusions

Moving towards category C, GI with admixtures or inclusions,


the paper presented by Kirstein and Wittorf (2013) is an
interesting transition between categories B and C. Indeed, the
authors describe the improvement of soft fat clay using rigid
inclusions combined with vertical drains, preloading and the use
of geotextile. The aim of the project was the construction of a
bridge for a new road in Germany including 1.5 to 7 m high
embankments. Vertical drains were first used to accelerate the
consolidation under the embankments (preloading condition).
Even using 600 kN/m woven geotextiles, vertical settlement of
around 1.5 m and horizontal displacement up to 27 cm were
measured throughout one year of monitoring. Because the
bridge could not tolerate residual settlements, Controlled
Modulus Columns (CMC) were designed and executed. The
design of the transition interface between the bridge and the
embankment, referred as the Load Transfer Platform (LTP), was
confirmed by the monitoring.
Cirin et al. (2013) set the constructive procedures and
bases of design of rigid inclusions including the LTP. The
ASIRI guidelines (IREX, 2012) were not yet published at the
time of preparation of this paper. The paper highlights the
difference with pile foundation. In rigid inclusion solutions,
there is no mechanical link between the pile and the structure. A
LTP is usually placed between the inclusions and the structure.
This distribution layer spreads the acting loads from the
structure towards the underlying soil-inclusions setup. As
indicated by the authors, isolated or continuous footings can
possibly be used to directly transmit the loads to the soilinclusions setup. This GI technique can also be applied for
embankments and landfills.
The following paper constitutes a good transition with the
next topic concerning stone columns. According to Carvajal et
al. (2013), dealing with the design of Column Supported
Embankments (CSE), a clear distinction has to be made
between rigid inclusions (e.g. concrete type columns)
characterized by a brittle behavior in its Ultimate Limit State

Technical Committee 211 / Comit technique 211

(ULS) and stone columns (made of gravel and sand) which


demonstrate a ductile behavior in its Serviceability Limit State
(SLS) due to its compressibility and drainage characteristics
(influence of the consolidation process on the design). Due to
the brittle behavior of concrete type columns, larger safety
factors have to be introduced, certainly for very slender
elements. The General Reporters fully agree that similar
approaches cannot be applied for very slender concrete type
columns and for stone columns. However it has to be remarked
that it is not common to consider stone columns as drainage
elements.
3.2

Stone columns

In the present Technical Session, Vlavianos et al. (2013)


propose technical solutions for the design of a road project in
the Region of Western Greece. The geology of the site consists
of soft silty clays and silty sands with high liquefaction
susceptibility. The high ground water table and the seismicity of
the area result in a design solution including GI. The installation
of stone columns followed by preloading was selected. For the
design of the bridge embankments and the pile foundations for
bridge piers, a comparative parametric study was performed
with or without stone columns. As discussed by the authors, the
main aim of the preloading was the increase of the undrained
shear strength of the superficial fine-grained soil layer. With the
installation of the stone columns, the following requirements
were met:
- increase of the general stability of the embankments;
- increase of the bearing capacity;
- reduction of the internal forces in the classical pile
foundations;
- acceleration of the consolidation process;
- mitigation of the liquefaction susceptibility.
Although stone columns is nowadays a well-known GI
method, installation effects arising during the execution still
remains poorly understood. In order to investigate this question,
Klimis and Sarigiannis (2013) describe the numerical analysis
of the installation of stone columns with a diameter of 0.8 m
and a depth of 23 m by means of the FLAC 3D Finite
Difference code. The excavation stage has been modeled in one
unique step and the realization of the stone column as follows:
- a) vibration and compaction, modeled by the application of
an equivalent radial pressure against the internal wall of the
cylindrical excavation;
- b) filling with a linear elastic geomaterial.
This numerical sequence was necessary to correctly determine
the area in the surrounding soil influenced by the installation of
the stone column and hence to assess with more accuracy the
effective diameter of this latter.
Poon and Chan (2013) present another methodology to
design stone columns. In this analysis, stone columns are
replaced by equivalent strips, as illustrated in Fig. 1. The
equivalent friction angle of the strips is dependent of the stress
concentration ratio which is defined as the ratio of the average
applied vertical stress within stone column to the average
vertical stress of the surrounding soil at the same level. A
method is proposed to compute this ratio by means of an
axisymmetric Finite Element Model (FEM) containing one
column and the surrounding soil. Numerical results obtained
with this methodology (2D FEM with strips) have been
compared with the results of a 3D FEM and with the results of a
conventional 2D FEM analysis in which the entire soil is
represented by a single block with equivalent properties. The
authors conclude that the strip model is preferable to the block
model for the assessment of the horizontal displacements.
Further research is still necessary to investigate the question of
the equivalent strength of the interface in the 2D strip method.

Figure 1. 2D stone column strips, from Poon and Chan (2013)

3.3

Geotextile confined columns

Rigid inclusions are a common GI technique for foundations of


embankments in soft soils. Nevertheless, when the soft soil does
not provide enough lateral support, the columns can be encased
with a geotextile. The following papers mainly focus on the
geotextile confined columns, also defined as geoencased
granular columns (GECs).
Castro et al. (2013) describe and compare analytical and
numerical analyses considering the behavior and the
performances of geotextile confined columns (GECs).
Parametric studies of the settlement reduction and stress
concentration show the efficiency of GECs for GI purposes.
This efficiency is mainly related to the contrast of stiffness
between the encasement and the soil. As another conclusion, it
is found that the settlement reduction is nearly the same for
different replacement ratios but decreases with the applied load.
Finally, columns with smaller diameter are better confined.
If GECs are often used to reduce settlements induced by the
construction of large embankments on soft soils, up to now no
rational displacement based design approach has been
introduced. For the purpose of investigating this question, Galli
and di Prisco (2013) first review the most common design
standards and then focus on the interaction between the
embankment and the geoencased columns. The main
contribution of the paper resides in the consideration of the
deformable base of the embankment. Indeed, real embankments
are characterized by a deformable base, as illustrated in Fig. 2.
As a consequence, different values of settlement are expected
for the top of the column (uc) and for the soil (us) at the base of
the embankment. As explained by the authors, vertical stresses
are redistributed at the base of the embankment between the
internal zone of the cell (above the column characterized by an
average stress i) and the external one (a circular crown above
the soil, characterized by an average stress e) due to the arch
effect. Shear stresses are then activated at the GEC-soil
interface, and differential settlements are expected even at the
top of the embankment.

a)

b)

Figure 2. Mechanical response of the system in case of (a) rigid and (b)
deformable embankment, from Galli and di Prisco (2013)

Hataf and Nabipour (2013) have designed a reduced-scale


model in such a way to identify the parameters governing the
behavior of the GECs installed in clayey soils. As a result, they
propose to encapsulate only the upper half of the column.

2419

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Geosynthetic reinforced column or pile supported


embankment the use of geogrids

Another way to use geosynthetic material for GI application is


the design of geogrids for the support of embankment, land
levees, yards and structure foundations (slabs and superficial
isolated or continuous footings).
Investigating the use of geosynthetics for reinforcement
under ground mass collapse, Ponomaryov and Zolotozubov
(2013) compare the method outlined in British Standard BS
8006 and several design approaches with numerical
calculations. On the basis of experimental elongation results,
they introduce the ratio of actual tensile force to deformation.
Computational assumptions are proposed for the description of
the mechanisms of stress-strain development in the reinforced
ground mass. The authors finally present a comparison between
experimental measurements and the results of seven different
methods used for the calculation of the tensile force in the
geosynthetic, its deflection and the surface settlement.
Mihova and Kolev (2013) analyze the benefit of a
geosynthetic reinforced pad of crushed stone used for the
foundation of a hall in Sofia over soft saturated soil. Field tests
were performed to estimate the E-moduli before and after
improvement. The authors also conducted Finite Element
analysis to model the consolidation process and to confirm the
design stability under static and seismic conditions.
Dimitrievski et al. (2013) present a history case of soil
reinforcement with geosynthetics for the construction of a sixstorey structure in Ohrid (Republic of Macedonia). Multi layers
geogrids were designed and the effects of the geostatic,
hydrostatic and dynamic loading conditions were studied with
the help of FEM calculations. The validity of the analysis was
demonstrated with the help of in situ measurements obtained for
a close similar structure.
3.5

Sand compaction piles (SCPs)

In the sand compaction pile (SCP) method, sand is fed into the
ground through a casing pipe and is compacted by vibration,
dynamic or static compaction to form columns. In practice,
SCPs are mainly used to prevent liquefaction and reduce
settlement with similar success in sandy and clayey soils. With
the help of laboratory and field tests, Burlacu et al. (2013)
investigate the potential of columns made of loess-sandbentonite mixture for the reinforcement of collapsible loess
deposits in Romania. Indeed, as explained in the paper of
Alupoae et al. (2013), these collapsible soils require GI works.
They are characterized by high water sensitivity: when its water
content increases, important deformations in the soil can be
observed. In such a way to illustrate this phenomenon, the
authors present a case study of differential settlement of
buildings founded on loess sensitive to wetting. In spite of the
good realization and control of the foundation, important
differential settlements were measured thereafter as a result of
the defective rainwater recovery system.
3.6

Microbial methods

The use of microbially induced carbonate precipitation (MICP)


to cement cohesionless soils has recently received substantial
attention from geotechnical researchers. The most common
MICP mechanism is hydrolysis of urea. MICP via ureolytic
hydrolysis relies on microbes to generate urease enzyme, which
then serves as a catalyst for the calcium carbonate (CaCO3)
precipitation reaction. If it is to date well known that the
mechanical properties of the treated soils are directly correlated
to the amount of (CaCO3) precipitation, a gray area still remains
concerning the influence of the original nature of the granular
material on the resulting properties of the treated soil. Within
the framework of a laboratory campaign, Tsukamoto et al.

2420

(2013) investigate the influence of the relative density of sand


samples on the MICP. As a result of their study, the MICP tends
to increase as the relative density of the soil decreases.
Nevertheless, considering the results of triaxial tests, maximum
principal stress differences were obtained for the samples with
the highest relative density. In light of these results, this
technique seems to be very promising for the future but due to
the bioplugging (permeability reduction) of the granular
material and to the generation of toxic product (ammonium
salt); soil stabilization using ureolytic MICP remains currently
unusual. According to Hamdan et al. (2013), the use of plant
derived urease to induce the carbonate cementation could be the
solution to avoid these drawbacks.
4
4.1

GI WITH GROUTING TYPE ADMIXTURES


Deep Mixing Method (DMM) and soil stabilization

The deep mixing method (DMM) is nowadays a worldwide


accepted GI technology. In this method, the ground is in situ
mechanically (and possibly hydraulically or pneumatically)
mixed while a binder, based on cement or lime, is injected with
the help of a specially made machine. Numerous reviews and
recent progresses of the DMM are referred in Denies and
Van Lysebetten (2012). In the recent years, the DMM is
undergoing rapid development, particularly with regard to its
range of applicability, cost effectiveness and environmental
advantages, as illustrated by the papers of this paragraph.
In the deep mixing projects, the design can be based on
laboratory mixing tests. Soil-cement samples are then prepared
and tested to study the mechanical properties of the stabilized
soil. But, up to now, many laboratories prepared these samples
without standardized procedure. Actually, molding techniques
have a great influence on the mechanical characteristics of the
stabilized material. According to Grisolia et al. (2013), this
influence is strictly correlated to the workability of the soilcement mixture and this latter can be quantified with the
measurement of the torque required to turn an impeller in the
mixture. Five molding techniques have been studied and the
authors propose the abacus illustrated in Fig. 3 to define the
range of applicability of these techniques in function of this
torque.
Applicable

Marginally Applicable

Not Applicable

3 6

No Compaction

Molding technique

3.4

65

75

Tapping

Rodding
Static
Compaction
25kPa

10 15

Static
Compaction
50kPa
Dynamic
Compaction
0

10

20

30

40

30

40

50

60

70

80

120
90

...
100

mixture's workability, Torque Mt (Nm)


High
workability,
liquid

Low
workability,
consistent

Figure 3. Ranges of applicability of the different molding techniques,


from Grisolia et al. (2013)

The applicability of each molding technique was evaluated by


an Applicability index, related to densest specimens with the
highest strength and results repetitiveness.
Since several decades, DMM has been used for GI works.
But in recent years, this technique has been increasingly used

Technical Committee 211 / Comit technique 211

for structural applications. Standardized guidelines for the


design of this kind of applications are not currently available. If
the previous work allows the construction of standardized and
international test procedures for laboratory mix samples, the
Quality Control (QC) of the execution process is generally
based on the results of Unconfined Compressive Strength
(UCS) tests performed on cored material. As part of the semiprobabilistic design approach presented in Eurocode 7, it is thus
essential to define the UCS characteristic value that can be
taken into account in the design. Denies et al. (2013) discuss
the definition of this value. In the first category of approaches,
the characteristic strength is defined as an X% lower limit value
computed either on the basis of a statistical distribution function
or based on the cumulative frequency curve of the original
experimental dataset of UCS values obtained from tests on
cored samples. A second approach to determine the UCS
characteristic value is the use of the average value of the dataset
in combination with a safety factor. For the first category of
approaches, a value for the X% has to be defined. Actually, one
major issue is the representativeness of the core samples with
regard to the in situ executed material. For the purpose of
investigating this question, the authors present the results of a
study on the influence of soil inclusions and then they discuss
the topic of the scale effect with regard to large scale UCS tests.
The following papers concern the investigation of the
mechanical properties of the soil mix material under the field of
laboratory or in situ experiments and with the help of numerical
modeling.
In such a way to investigate the properties of the soil mix
material, Szymkiewicz et al. (2013) have carried out a
parametric study on lab soil-cement mixtures. The influences of
the particle size, the clay content and the water content on the
strength of the material were considered. They propose an
abacus relating the UCS of the specimens to the cement content.
Six zones are identified in the abacus depending on the nature
of the soil. In addition, the authors also propose a formula valid
for granular soils for the estimation of the UCS at 28 curing
days. This formula takes into account the water, the cement and
the fine contents.
In a similar way, Correia et al. (2013) have performed
laboratory tests to study the improvement of soft clayey silt
with high organic content by mixing it with a binder made up of
75% Portland cement (PC) and 25% blast furnace slag. They
first give a formula for the assessment of the UCS at 28 days in
function of the binder content and the liquidity index (LI) of the
soil. A normalized UCS is then introduced as follows:
UCSLI = UCS x LI. In a second step, the applicability of the
normalized UCS approach is analyzed for seven other cementstabilized soft soils with successful result.
If the water/cement (w/c) ratio is often used in attempt to
understand soil-mix properties, it can be found limited since in
practice execution is mostly performed in soils in the presence
of water (unsaturated or saturated conditions). A well-adapted
governing parameter could be then the porosity/cement index
defined as the ratio of porosity to the volumetric cement content
(n/Civ). Rios et al. (2013) highlight the influence of this index
on the mechanical properties of cemented Porto silty sand.
Unique trend was obtained between the UCS and an adjusted
porosity/cement ratio (n/Civ0.21), proposed by the authors.
Similar observation was also made with indirect tensile
strength. Triaxial tests resulted in two peak strength envelopes
for each predetermined (n/Civ)0.21 and finally, oedometer tests
establish this ratio as the governing parameter of the behavior of
the soil-cement specimen in one-dimensional compression in
lieu of the cement content or the initial void ratio.
A major advance in DMM could be found in the contribution
of Yi et al. (2013) with the investigation of the carbonation of
reactive magnesia (MgO) for soil stabilization. Nowadays,
Portland cement (PC) is the most common binder used in the

deep mixing applications. However, there are significant


environmental impacts associated with its production in terms
of high energy consumption and CO2 emissions. In their
laboratory study, reactive MgO was used as a binder and the
MgO-soil samples were carbonated by CO2 to improve the
mechanical properties of the soil and reduce the CO2 emission.
As an evident result, the UCS values of the uncarbonated MgOstabilised soils were much lower than those of the PC-stabilised
soils; both mixes took ~28 days to finish most of their strength
development.
Nevertheless,
the
carbonation
process
significantly increased the UCS of MgO-stabilised soils in a
very short time, this latter fast reaching the UCS value of the
28-day PC-stabilised soils, indicating that it could be used to
support a structure just after the completion of the carbonation
procedure.
Another type of binder largely used for soil stabilization is
lime. Mesri and Moridzadeh (2013) discuss the results of a
laboratory study focusing on the improvement of the Brenna
clay (high plastic lacustrine clay of North Dakota) by adding
lime. Lime contents varying between 3 and 10 % of the dry
weight of the clay have been considered. The authors observed
a decrease of the measured pH with time and an increase of the
Liquid Limit and the Plasticity Index with time when 5 % of
lime was added. Adding 3 to 8 % of lime, the residual friction
angle (in drained conditions) increases between 3 to 6 %.
Unfortunately the laboratory test results were not compared
with full scale test results.
Extensive laboratory tests have been performed by Szendefy
(2013) for the purpose of determining the effect of lime
stabilization on 21 Hungarian clayey soils. In addition, some in
situ stabilized soils have also been analyzed. According to his
study, the improvement of the clayey soil with the lime is
mainly related to the coagulation of the clay particles related to
the cation exchange. Indeed, during the stabilization with lime,
Ca2+ ions attach to the surface of clay particles. As a result of
this high charging, the clay particles coagulate resulting in a
material characterized by an increased internal friction angle.
The pozzolanic reaction would play then a secondary role in the
stabilization.
Soil stabilization can also be performed with fiber
reinforcement, such as discussed in Madhusudhan and Baudet
(2013). In their study, laboratory tests have been performed to
determine the influence of adding polypropylene fibers on the
shear strength characteristics of completely decomposed granite
(CDG). In Hong Kong, CDG is regularly used for landscaping
and as green cover of existing shotcrete slopes. The test results
clearly indicate an important increase of the UCS when adding
0.5% of fibers and compacting the CDG at the water content
close to the optimum Proctor value. In triaxial drained tests, the
addition of fibers seems to increase the shear strength of the
CDG and its stiffness. Dilation is also reduced.
In Singapore, laboratory tests have been performed by Xiao
et al. (2013) in order to determine the characteristics of the
Singapore upper marine clay when mixed with 20 to 50%
Portland cement (PC) and up to 0.32% fibers of different types.
As a result of their study, strength and ductility of cementtreated clay were improved by fiber reinforcement. There is an
optimum fiber content with regard to performance and
workability of the material. Polyvinyl alcohol (PVA) fibers are
generally more efficient than polypropylene (PP) fibers except
for low cement and water contents. The length of the fibers has
a significant effect on the ductility of the cement-treated clay for
both fiber types. Concerning the strength, the influence of the
fiber length is more significant for PVA reinforcement than for
PP reinforcement.
Cuira et al. (2013) present the results of numerical models
simulating an axial Static Load Test (SLT) on a soil-cement
column. Numerical and experimental results are compared with
the help of three Finite Element models and one simplified

2421

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

semi-analytical model. Numerical results are in agreement with


the experimental observations all along the SLT but especially
regarding to the fracture pattern: structural failure localized in
the upper part of the column. This numerical study highlights
the nonlinear behavior of the soil-mix material. In comparison
with classical rigid piles, the contrast of strength (and
stiffness) between the column and the soil is lower and has a
huge influence on the global behavior.
Originally, DMM was developed for GI applications in soft
clays and organic soils. But more recently, it was also dedicated
to various structural and environmental applications such as
illustrated by the following case histories.
Recently, the DMM has been chosen for several Hungarian
railway projects involving soft soils, such as the restoration of
the Srrt railway line crossing an area where the subsoil
consists of soft chalky silt. For the foundation of a 4m high
embankment, two DMM were taken into account: the mass
stabilization and the soil-cement columns. Koch and
Szepeshsi (2013) firstly describe results of laboratory tests on
chalky silt samples mixed with cement for different w/c
contents. Both DMM are then assessed using 3D-FEM
considering the site requirements in term of stability and
settlement.
In a similar way, DMM have been widely used in Japan for
the improvement of soft clays and organic soils.
Matsui et al. (2013) introduce the concepts of an hybrid
application of soil-cement columns combined with soil mix
walls (SMW) designed for the foundation of an embankment.
The concept is illustrated in Fig. 4. The authors propose a
conceptual method allowing the control of ground deformation
and ensuring an optimization of the volume of treated soil. The
method is supported by 2D-FEM and in situ monitoring is
performed for the validation of the concept.

39.2m

Ac2- 2

Inside piles

Ac2- 3
Dvc
Dvs

Walls

Figure 6. Time-dependent load sharing between raft and piles, from


Yamashita et al. (2013)
36.2m

1.9m

21.2m

As2

10.0m

Ac1- 2

6.7m 4.5m

1:1.8

9.8m

37.2m

Figure 5. Measured vertical ground displacements below raft, from


Yamashita et al. (2013)

Section view

5.2m

7.0m

12.0m

presented, namely: settlement reduction, improvement of slope


stability, reduction of active pressure on retaining walls and
decrease of liquefaction susceptibility. The two last topics are
then illustrated with case histories.
Other case history tackles the topic of liquefaction
susceptibility restrained with the DMM. Yamashita et al.
(2013) deal with the measurements performed underneath a
piled raft completed with SMW to reduce the risks of
liquefaction. It concerns a 12-storey office building. The load
distribution between piles, SMW and the surrounding soil has
been monitored during a period of three years. After the end of
the construction, settlements of 20 mm have been recorded, as
illustrated in Fig. 5. As another result, 70 % of the load was
taken by the piles, 14 % by the SMW and 15% by the soil, as
shown in Fig. 6. The measurements also learned that the
Tohoku earthquake of March 2011 had almost no influence on
the settlements and on the load distribution.

Outside piles

Figure 4. GI with soil-cement columns and SMW, from Matsui et al.


(2013)

In Lund (southern Sweden) a new generation synchrotron


radiation facility, called MAX IV, is under construction.
According to Lindh and Rydn (2013), it should be 100 times
more efficient than any existing comparable synchrotron
radiation facility in the world. For this kind of facility, the
vibration requirements are very stringent. Various alternatives
were discussed and simulated during the conception. The
optimum solution was achieved with a four meter thick layer of
stabilized soil below the concrete foundation. A combination of
quicklime and ground granulated blast furnace slag (GGBFS)
was found to be in agreement with both design and construction
requirements.
Jeanty et al. (2013) describe the use of the CSM and the
Trenchmix methods for the realization of SMW. Both
techniques are explained in details and different applications are

If the foundation of embankments and buildings are become


both common applications of the DMM, underpinning with soil
mix material constitutes an interesting emerging technique, such
as illustrated in the following paper.
Traditional DMM are commonly restricted for underpinning,
limitations being mainly related to the capacity of the machine
to pass existing foundation structures as reinforced slabs or
footings, the reduced working spaces and the possible low
headroom conditions. Melentijevic et al. (2013) present a case
history of underpinning of an existing floor slab in an industrial
building using DMM. The soil-cement columns were installed
with the new Springsol tool. After the realization of a contact
grouting between the slab and the soil, the slab and the contact
grouting layer are cored. The spreadable Springsol tool is then
introduced into the gap. Finally, its blades are opened and the
soil-cement column is executed until the predetermined depth.
The conception is supported by numerical modeling and
QA/QC aspects of the project are related to the testing of core
and wet grab samples.
4.2

Use of stabilized dredged material for construction

As previously discussed in Chu et al. (2009), dredging and land


reclamation have increasingly become important parts of
construction activities that involve heavily geotechnical
knowledge. If dredging provides low cost construction material,

2422

Technical Committee 211 / Comit technique 211

it is sometimes necessary to resort to additional GI methods in


order to obtain a product meeting the design requirements. The
following paper illustrates how GI and dredging are
complementary construction processes.
Loh et al. (2013) present considerations in the design and
construction of a containment bund made of modified geotextile
tubes (M-GT) filled with cement-mixed soil. In the port of
Singapore, dredged soil mixed with cement was used as in-fill
material in the M-GTs and as the core of a large geotextile
containment bund, as illustrated in Fig. 7. Field instrumentation
and monitoring were carried out with the help of strain
measurements, hydrographic survey, inclinometers and
extensometers during and after the construction to verify the
design and the performance of the system.

Figure 7. Geotextile containment bund, from Loh et al. (2013)

4.3

Recent advances in the jet grouting applications

If special devices have been developed in the past to measure


the diameter of the jet grout columns executed in situ,
considerable effort should be made in the understanding of the
physical processes governing this parameter. Bzwka et al.
(2013) analyze excavated jet grout columns. The experimental
results are used to model the bearing capacity of the columns by
means of the Z-soil software. Indeed, although the jet grout
columns have been realized in compacted medium sand
underlain by stiff clay, almost all columns had an irregular
shape influencing its bearing capacity.
In the grouting applications, the bleed capacity is another
indicator of grout effectiveness, since it is representative of the
volume of voids filled by cement. The grouts water-to-cement
ratio (W/C) and the maximum cement grain size (dmax) are two
important parameters controlling the cement grout bleed
capacity. Pantazopoulos et al. (2013) provide some insights on
the effect of grout bleed capacity on the mechanical properties
of ordinary and microfine cement grouted sands, in conjunction
with the effect of the W/C ratio. They demonstrate that the
distinction between stable and unstable grouts (see EN 12715)
may not be an indicator of grout effectiveness since similar
effects may be produced by both stable and unstable grouts: e.g.
same coefficients of permeability were obtained for a bleed
capacity ranging from 5 (stable) to 30 % (unstable suspension).
Bleed capacity correlates very well with some grouted sand
properties (i.e. unconfined compression strength and cohesion)
and not at all with other properties (i.e. internal friction angle
and damping ratio).
5
5.1

EARTH REINFORCEMENT

In a similar way, Tabarsa and Hajiesmaeilian (2013) have


studied the influence of sand encapsulated non-woven
geotextile (sandwich technique) on the stability of clay
embankment. Using FLAC 2D Finite Difference model, the
authors highlight the efficiency of the method with regard to the
geotextile-reinforced and the unreinforced embankments.
5.2

A significant element in the reclamation of landfills is the


reinforcement and biological stabilization of the slopes which
can be very sensitive to surface erosion. According to Koda
and Osinski (2013), landfill stability improvement activities
can be divided in two phases: the first one consists in the
technical reclamation of the landfill and the second one is the
biological restoration of the vegetation cover. For both phases,
the authors argue it is possible to use recyclable materials such
as fly ash or sewerage sludge. They discuss the improvement of
slope stability of a solid waste disposal with the help of this
approach. On the one side, fly ashes can be considered as
impermeable and present good compaction properties. Mixed
with cohesive soil, it could be therefore used for the capping of
the waste disposal. On the other side, the sewage sludge protects
the seeds from erosion and excessive drying. Moreover the
sewerage sludge presents a high nutrition content supporting the
development of the vegetation cover. Unfortunately, no
information is given in the paper concerning the installation
procedures of the fly ashes and sewerage sludge and how the
influence of vegetation can be introduced in the stability
calculations.
6

Centrifuge tests have been performed by Bo et al. (2013) in


order to study the reinforcement of low plastic brown weathered
shale with polypropylene fibers for the construction of an
embankment. Vertical and horizontal displacements deduced
from the centrifuge tests have been compared with those
obtained from FEM analyses. Both approaches demonstrate the
contribution of the fibers on the stability.

CONCLUSIONS

In the present General Report, 47 papers of the Technical


Session on GI of the XVIII ICSMGE are reviewed. It can be
noted that 40% of these papers deal with Deep Mixing and soil
stabilization, proving the huge interest in these techniques.
Similar percentage was already observed in the Proceedings of
the TC211 IS-GI 2012 (Denies and Huybrechts, 2012) but this
is not surprising, as these methods constitute outstanding and
cost-effective sustainable construction processes.
Finally, beyond the choice of the GI solution, the necessity
of monitoring was also highlighted by several authors of this
Technical Session. For example, van der Stoel et al. (2013)
discuss a well-documented case history concerning the
realization of two deep excavations in the courtyards of a
historical building in Amsterdam. Based on 2D FEM
calculations, an extensive monitoring program has been
proposed and performed (including levelling point
measurements, inclinometers and the use of a permanent
webcam). Thanks to this monitoring process the consequences
of two important accidents during the execution of the
excavations could be limited as much as possible. Most
important was that the time delay remained very small. The
authors conclude that the costs of the meticulous and proactive
monitoring were minor in comparison with the potential costs of
a delayed opening of the hotel.
If ground improvement is really become an efficient and
controllable cost-effective alternative to classical foundation
technique, measure still remains treasure.
7

Geosynthetics

Vegetation methods

ACKNOWLEDGEMENTS

The authors wish to thank the chairmen of the TC211 Jan


Maertens and Serge Varaksin for their contribution to the
review of the papers of the Technical Session on GI works.

2423

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

REFERENCES

Alupoae, D. Auencei, V. and Rileanu, P. 2013. Time - dependent


behaviour of foundations lying on an improved ground.
Bo, L., Linli, J., Ningyu, Z. and Sinong, L. 2013. Centrifugal and
numerical analysis of geosynthetic-reinforced soil embankments.
Burlacu, C., Olinic, E., Manea, S. and U, P. 2013. Compacted soil
columns for foundations on collapsible soils. Laboratory and in-situ
experimental study.
Bzwka, J., Juzwa, A. and Wanik, L. 2013. Selected problems
connected with the use of the jet grouting technique.
Carvajal, E., Vukoti, G., Sagaseta, C. and Wehr, W. 2013. Column
Supported Embankments for Transportation Infrastructures:
Influence of Column Stiffness, Consolidation Effects and Cyclic
Loading.
Castro, J., Sagaseta, C., Caizal, J., Da Costa, A. and Miranda, M. 2013.
Foundations of embankments with encased stone columns.
Chai, J.-C. and Carter, J. P. 2013. Consolidation theory for combined
vacuum pressure and surcharge loading.
Chu, J., Varaskin, S., Klotz, U. and Meng, P. 2009. Construction
Processes, Proceedings of the 17th International Conference on Soil
Mechanics and Geotechnical Engineering, 5-9 October 2009,
Alexandria, Egypt, M. Hamza et al. (Eds.), IOS Press, Amsterdam,
Vol. 4, pp. 3006-3135.
Cirin, A., Pauln, J., Racinais, J. and Glandy, M. 2013. Displacement
rigid inclusions. Proceedings of the 18th ICSMGE, Paris, France.
Correia, A.A.S., Venda Oliveira, P.J. and Lemos, L.J.L. 2013.
Prediction of the unconfined compressive strength in soft soil
chemically stabilized.
Cuira, F., Costa dAguiar, S., Grzyb, A., Pellet, F., Mosser, J.F.,
Guimond-Barrett, A. and Le Kouby A. 2013. Numerical modeling
of a soil-mixing column behavior and comparison with a full-size
load test.
Denies, N. and Huybrechts, N., editors. 2012. Proceedings of the
International Symposium of ISSMGE - TC211. Recent research,
advances & execution aspects of ground improvement works. 31
May-1 June 2012, Brussels, Belgium (www.bbri.be/go/tc211).
Denies, N. and Van Lysebetten, G. 2012. General Report. Session 4
Soil Mixing 2 Deep Mixing. Proceedings of the International
Symposium of ISSMGE - TC211. Recent research, advances &
execution aspects of ground improvement works. 31 May-1 June
2012, Brussels, Belgium (www.bbri.be/go/tc211).
Denies, N., Van Lysebetten, G., Huybrechts, N., De Cock, F., Lameire,
B., Maertens, J. and Vervoort, A. 2013. Design of Deep Soil Mix
Structures: considerations on the UCS characteristic value.
Dimitrievski, Lj., Ilievski, D., Dimitrievski, D., Bogoevski, B. and
Strasheski, A. 2013. Method of improvement of the subsoil under
Adora facility Ohrid, Republic Of Macedonia.
Galli, A. and di Prisco, C. 2013. Geoencased columns: toward a
displacement based design.
Grisolia, M. , Leder, E. and Marzano, I.P. 2013. Standardization of the
molding procedures for stabilized soil specimens as used for
QC/QA in Deep Mixing application.
Hamdan, N., Kavazanjian, E and ODonnell, S. 2013. Carbonate
Cementation via Plant Derived Urease.
Hataf, N. and Nabipour, N. 2013. Experimental investigation on bearing
capacity of geosynthetic encapsulated stone Columns.
Indraratna, B., Rujikiatkamjorn, C. and Geng, X. 2013. Performance
and Prediction of Vacuum Consolidation Behavior at Port of
Brisbane.
IREX. 2012. Projet national ASIRI. Recommandations pour la
conception, le dimensionnement, l'excution et le contrle de
l'amlioration des sols de fondation par inclusions rigides. Presses
des Ponts. France.
Islam, M. S. and Yasin, S. J. M. 2013. Improvement of a Clay Deposit
using Prefabricated Vertical Drains and Pre-loading - A Case
Study.
Jeanty, J.M., Mathieu, F. and Benhamou, L. 2013. Importance and
practical examples of inertial soil improvement.
Jebali, H., Frikha, W. and Bouassida, M. 2013. Assessment of Carillos
theory for improved Tunis soft Soil by geodrains.
Kirstein, J. F. and Wittorf, N. 2013. Improvement of soft fat clay using
rigid inclusions and vertical drains.
Klimis, N.S. and Sarigiannis, D.D. 2013. Interaction of stone column
and surrounding soil during its construction: 3D numerical analysis.

Koch, E. and Szepeshzi, R. 2013. Laboratory tests and numerical


modeling for embankment foundation on soft chalky silt using
deep-mixing.
Koda, E. and Osinski, P. 2013. Assessment of bio-mechanical
reinforcement materials influencing slope stability, based on
numerical analyses.
Lee, C., Choi, Y., Lee, W. and Hong, S. J. 2013. Evaluation of Vertical
Drain-enhanced Radial Consolidation with Modified Analytical
Solution.
Lindh, P. and Rydn, N. 2013. Adjusting the soil stiffness with
stabilisation to minimize vibration at Maxlab IV a synchrotron
radiation facility in Sweden.
Loh, C.K., Chew, S.H., Tan, C.Y., Lim, S.K. and Lam, J.P.W. 2013.
Construction and Performance of Containment Bund Using
Geotextile Tubes Filled With Cement Mixed Soil in Singapore.
Madhusudhan, B.N. and Baudet, B.A. 2013. Reinforcement of
completely decomposed granite with discrete fibres.
Matsui, H., Ishii, H. and Horikoshi, K. 2013. Hybrid Application of
Deep Mixing Columns Combined with Walls as a Soft Ground
Improvement Method Under Embankments.
Melentijevic, S., Arcos, J.L. and Oteo, C. 2013. Application of cement
deep mixing method for underpinning.
Mesri, G. and Moridzadeh, M. 2013. Lime Remediation of Reactivated
Landslides.
Mihova, L. and Kolev, Ch. 2013. Improvement of the Soil under the
Concrete Pavement of a Plants Hall.
Pantazopoulos, I.A., Atmatzidis, D.K., Basas, V.G. and
Papageorgopoulou, S.K. 2013. Effect of Grout Bleed Capacity on
the Engineering Properties of Cement Grouted Sands.
Parsa-Pajouh, A., Fatahi, B. and Khabbaz, H. 2013. Numerical Analysis
to Quantify the Influence of Smear Zone Characteristics on
Preloading Design in Soft Clay.
Ponomaryov, A. and Zolotozubov, D. 2013. Technique of reinforced
soil base calculation under fall initiation in ground mass.
Poon, B. and Chan, K. 2013. Stress Concentration Ratio and Design
Method for Stone Columns using 2D FEA with Equivalent Strips.
Rios, S. and Viana da Fonseca, A. 2013. Porosity/cement index to
evaluate geomechanical properties of an artificial cemented soil.
Szendefy, J. 2013. Impact of the soil-stabilization with lime.
Szymkiewicz, F., Le Kouby, A., Reiffsteck, P., Mosadegh, A. and
Tacita, J-L. 2013. Laboratory parametric study of the Deep Mixing
material.
Tabarsa, A.R. and Hajiesmaeilian, S. 2013. Investigation of failure
analysis of clay reinforced with sand encapsulated.
Tsukamoto, M., Inagaki, Y., Sasaki, T. and Oda, K. 2013. Influence of
relative density on microbial carbonate precipitation and
mechanical properties of sand.
Van der Stoel, A.E.C., Vink, D. and Bouma, J. 2013.
Conservatoriumhotel Amsterdam, geotechnical design and
monitoring.
Vlavianos, G.J., Marinelli, A.K., Andrianopoulos, K. and Foti, S. 2013.
Bridge foundation on very soft alluvia with stone column ground
improvement.
Weihrauch, S., Oehrlein, S. and Vollmert, L. 2013. Subgrade
improvement measures for the main rescue roads in the urban
redevelopment area HafenCity in Hamburg.
Xiao, H.W., Lee, F.H., Zhang, M.H., Yeoh, S.Y. 2013. Fiber Reinforced
Cement Treated Clay.
Yamashita, K., Wakai, S. and Hamada, J. 2013. Large-scale Piled Raft
with Grid-Form Deep Mixing Walls on Soft Ground.
Yi, Y.L., Liska, M, Unluer, C. and Al-Tabbaa, 2013. A. Initial
investigation into the carbonation of MgO for soil stabilisation.

2424

Time-dependent behaviour of foundations lying on an improved ground


Temps-comportement dpendant de fondations reposant sur un sol amlior
Alupoae D., Auencei V., Rileanu P.

"Gheorghe Asachi Technical University of Iasi, Faculty of Civil Engineering and Building Services, Department of Roads
and Foundations,43 Dimitrie Mangeron Bd, 700050, Iasi, Romania

ABSTRACT: The paper presents some aspects concerning time-dependent behaviour of the improved foundation soils. The
foundation soils can develop favourable or inappropriate resistance properties under the loads submitted by constructions. Engineers
and investors encounter more and more difficult foundation soils, in their desire to efficiently use the construction sites. In this case,
physical and mechanical properties of the soil have to be improved, in order to sustain the infrastructure and structure of a building.
The paper reviews some improvement methods, after presenting difficult foundation soils from Iai area. The paper presents a case
study regarding problems caused by difficult foundation soils that are present in the region, during the operating period of structures.
The presence of water in the foundation soil created a negative impact in its behaviour, which led to differential settlements and,
consequently, the buildings were switching from their initial vertical position. The study also analyzes time-dependent settlements of
a construction. Finally the paper presents some conclusions resulting from studies both bibliographic and practical.
RSUM : Le document prsente quelques aspects concernant le comportement en temps des sols amliors pour les fondations. Le
terrain de fondation peut avoir un comportement favorable ou par contre dfavorable sous laction des charges donner par les
constructions. Pour utiliser efficacement les terrains des constructions, les ingnieurs et les investisseurs rencontrent souvent des sols
de fondation de plus en plus difficile. Dans ce cas, les proprits physiques et mcaniques du sol doivent tre amliores, afin
dassurer des bonnes conditions dappuis pour l'infrastructure et la structure d'un btiment. Le document passe en revue les sols de
fondation difficiles de la zone de Iai et des mthodes de les amliores. Il est aussi prsent une tude de cas concernant les
problmes qui peuvent apparaitre au cours de la priode d'exploitation de structures, a cause des ces sols de fondation difficiles. La
prsence de l'eau dans le terrain de fondation a eu un impact ngatif sur son comportement, ce qui a produit des tassements
diffrentiels, ca veut dire que les btiments furent commuts de leur position initiale, verticale. Pour conclure, le document prsente
des conclusions issues de ltude bibliographique et pratique la fois.
KEYWORDS: leaning structure, expansive clay, loess, differential settlement.
1

INTRODUCTION

As a result of the analysis performed over time on a large


variety of soils and taking into account soil behaviour in the
presence of external factors, the foundation soils can be divided
in two categories, considering their capacity to support loads
from constructions: good and difficult foundation soils.
The entire existence of the building system depends on the
stability and strength of the foundation soil and this is the main
reason why a special interest is given to the second category of
soils and therefore to the specific issues that must be considered
in the design, execution and operating period of a construction.

DIFFICULT FOUNDATION SOILS

The sites that have a construction soil with good geotechnical


characteristics are rapidly decreasing. Large urban areas are a
particular problem because, due to the lack of space, it is
necessary to reconsider the possibility of placing a building on a
soil that was unsuitable for constructions until now.
2.1

Difficult foundation soils - classification

These soils are classified as follows:


macroporous soils (present large cavities in their
structure and have the ability to suffer large settlements
when are subjected to a wetting process);
collapsible soils (are characterised by the fact that when
in high humidity develop large deformations);

2.2

liquefiable soils (especially non-cohesive soils


consisting of saturated fine sand which under the action
of a dynamic load suddenly loose their shear strength);
expansive soils (cohesive soils such as clays, which
change their volume when water content varies);
soils that during the freezing and thawing phenomena
change their structure and properties;
peaty soils (organic matter is present in its structure,
have a high and very high compressibility and a low
shear strength);
eluvium (formed as a result of decomposition and
alteration of existing rocks);
saline soils (are characterized by the settlement
phenomena that occurs during a long lasting wetting);
fillings (occur as a result of unconsolidated alluvial
deposits) (Iliesi 2012).
Methods of soil improvement

Given the frequency of soils that present unfavourable


characteristics for constructions over time were developed
methods to improve their mechanical properties, such as:
soil compaction which can be made on surface (rolling,
dynamic, cushions) or in depth (columns, pre-wetting,
dynamic);
chemical soil stabilization (cementation, silicatization,
jet grouting, bentonite etc.);
electrochemical
methods
(electrophoresis
and
electroosmosis);
thermal treatment of soils.

2425

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

The case study refers to the problems that can occur with an
improved foundation soil. The method used for improvement is
soil replacing (soil cushion).
3

CASE STUDY

Within the areas with difficult foundation soils, Iai City is the
largest urban settlement located in the eastern part of Romania.
The city lies at the contact between Jijia Meadow and the
Moldavian Plateau. The landscape is varied, forming a region
consisting of eroded hills crossed by Bahlui plateau.
3.1

Soil types in the studied area

Studies performed over the last decades into the existing


terraces of the region show that almost 70% from the current
area of the city have medium and low suitability for
construction purpose, this reason being more or less a natural
barrier for city expansion. Theoretical and practical solutions
offered for solving the issues caused by these types of soils
present a special interest in the current context (Vieru 2010).
Among these types of soils there are two specific categories:
loess and expansive soils.
The different types of soils existing in the studied area have
either a normal behaviour under loads, or an atypical one.
Therefore, upper and medium terraces consist of a
succession of coarser sediments at the bottom followed by a
loess soil sensitive to wetting. Loess layer is yellow-brown with
variable thickness from 8.00 m to 15.00 m lying in the highest
areas of the city. Loess deposits usually consist of silty clay and
clayey silt.
The loess of Iai region has medium plasticity with the liquid
limit LL = 3050%. Grain-size distribution is: 2529% clay,
4347% silt and 2432% sand. Regarding the uniformity
coefficient, the loess of Iai City is considered to have a good
uniformity (Ciornei & Rileanu 2000).

of the particles making up the clay fraction and the nature of the
absorbed ions;
- hydro-geological conditions groundwater is present both
through deep under pressure aquifers and also through free flow
ones. Deep layers have a high mineralization, being intercepted
only by drilling. They have an ascending nature, sometimes an
artesian one. Shallow drillings revealed the presence of captive
water with low mineralization, which can be used locally;
- layer thickness the thicker the layer is, the bigger its
swelling;
- moisturized area if the wet surface under an existing
building is insignificant, the deformations increase and the
probability of deterioration grows (Alupoae et al. 2011).

Figure 2. Contraction breaks and cracks

3.2

On site situation

The case study follows a residential area placed on one of the


hills in Iai City, Romania. The increase of water content inside
the foundation soil determined a differential settlement and the
buildings placed on site were switched from their initial vertical
position.

Bahlui Meadow is characterized as a mixture of sand and


gravel layers at the base of the stratification, followed by a layer
of fat swelling and shrinking clay. The sand layer has a
thickness of almost 4.00 m and the clay is between 5.00 m and
6.00 m. This clay is actually the foundation soil from the area,
requiring good knowledge of soil characteristics.
As far as the soil properties are concerned, Bahlui clay falls
within the category of high swelling and shrinking soils.
Climatic conditions of the area, with temperatures decreasing in
the summer with 10C...20C from day to night and heavy
rainfall, lead to changes in soil volume. To avoid foundation
deterioration the minimum foundation depth has been set at 2.00 m, as deep as the effects of seasonal variations in moisture
content and temperature may not be felt (NE 001-96 1996).

3.2.1 Data regarding the constructions from the studied area


The constructions were built during two different time frames:
Stage 1 between 1994 and 1998, consists of a two
section building 22.0 x 12.0 meters (Section I and
Section II), has a total ground surface of 530 m2, a
structure made of reinforced concrete frames placed on
network of foundation beams. The foundation rests on a
soil cushion, 1.0 meter thick. In 1998 the foundation
system was checked and the results showed that the soil
cushion placed under the foundation had a degree of
compaction of 95.15%. Thus it can be stated that the
operations of soil improvement using mechanical means
were correctly carried out. On site, a layer of loess,
sensitive to wetting, was intercepted in drillings up to
9.0 meters from the ground surface. Under the soil
cushion the thickness of the loess layer is about 5.0
6.0 meters.
Stage 2 the construction of Section III started in 2001,
with a built area of about 850 m2 and a structure and
height similar to the initial sections. This section is not
entirely finished and the main problem is the fact that no
systematization works are carried out. Also, the systems
of rainwater collection and disposal are not finished.
Because of this, in 2010 a movement was observed.

Other factors influencing the volume variation are:


- soil activity volume variation is influenced by molecular
and electro-molecular phenomena reflected by adhesive and
capillary water, their size depending on the mineralogical nature

After the initial observations, measures have been taken to


analyze the technical condition of the building and to establish
the necessary actions to ensure a proper exploitation for the
constructions.

Figure 1. Structural rearrangement for collapsible soils

2426

Technical Committee 211 / Comit technique 211

27

Humidity [%]

25
23
Section I
21

Section II
Section III

19
17
15
1998

Table 1. Humidity evolution on the site filling layer


Section

[%]

Humidity
2010

1998

[%]
I
II

20.70
33.42

22.99
32.85

26.90

+19.80

III

---

---

18.31
33.04

24.40

---

Table 2. Humidity evolution on the site soil cushion


Humidity
2010
[%]

Average
humidity
2010
[%]

Increase
of
humidity
[%]

21.15
24.15

21.29

16.37
23.81

20.53

-3.70

II

17.11
23.69

20.63

18.40
24.45

21.77

+5.50

III

---

---

17.61
20.64

19.03

---

2002

2006

2010

Figure 4. Humidity variation on the site, in the earth pillow, for the three
sections of the building

20.70
33.42

22.45

[%]

18

+4.20

17.18
24.15

Average
humidity
1998
[%]

Section III
19

25.83

II

Humidity
1998

Section II

19.57
30.07

24.78

Section

Section I

20

Increase
of
humidity
[%]

16.98
29.16

21

Average
humidity
2010
[%]

[%]

2010

22

Humidity [%]

3.2.2 Causes that led to differentiated settlements


The main cause that led to settlements on the studied case was
determined by the increased humidity in the foundation soil.
This happened as a result of a deficient vertical systematization:
no sidewalks, there were no gradients on site to discharge the
water and also there were not built ditches and surface drainage
systems. The lack of systematization works led to rainwater
infiltrations in the filling layer above the soil cushion used as an
improving method for the loss soil on the site. Water bags were
formed in the filling layer which supplied the permanent
moisture on the layer above the cushion. The humidity of the
cushion became 3.14% higher then the optimum compaction
humidity (19.40%). Also the filling layer recorded higher values
for humidity: 25.07% 27.52%.
Average
humidity
1998
[%]

2006

Figure 4. Humidity variation on the site, in the filling layer, for the three
sections of the building

Figure 3. Photos showing on site displacements

Humidity
1998

2002

Table 3. Humidity evolution on the site surrounding area


Section

Humidity
1998

Average
humidity
1998
[%]

Humidity
2010

Average
humidity
2010
[%]

Increase
of
humidity
[%]

26.85

18.90
29.36

22.30

-20.0

26.85

18.90
29.36

22.30

-20.0

[%]

Topographic measurements were made, on the site, for


verifying settlements that appeared due to moistening of the
foundation soil. By analyzing the results obtained after four
cycles of measurements, the following conclusions can be
drawn:
for section I the measured values of settlements are
insignificant. This happed because the values fall within
the margin of error of the measurements and also
because the variations determined at the markers
considered stationary must be taken into consideration;
for section II were found higher values of the
settlements at the joint between section II and III. This
occurs where the surface water penetrated the ground
and produced a pronounced moistening of the
foundation soil;
for section III were also found higher values of the
settlements at the joint between section II and III. This
occurs where the surface water penetrated the ground
and produced a pronounced moistening of the
foundation soil.
Established settlements have small values and pose no
danger to the behaviour of the building in time. Relative
settlements have also small values, 3.65105 millimetres, much
lower than the admissible relative settlement, which is,
according to Romanian Standards, 0.001 millimetres.

2427

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

The settlement speed decreases from 0.213 mm/day, after 9


days, to 0.061 mm/day after 22 days and further to 0.006
mm/day after 83 days. This led to the conclusion that
settlements are slowing down.
3.2.3 Proposed solutions
Continuous monitoring of building settlements and conducting
topographic readings at least every three months until the
constructions are stabilized.
Efficient vertical and horizontal systematization can be done
by making sidewalks, gradients for water discharge, ditches and
surface drains.
For stopping water infiltration in the foundation soil is
mandatory to check utility networks and repair them where is
necessary.
4

CONCLUSIONS

Difficult foundation soils are frequently found on sites located


in large urban areas.
Over time, a series of methods and techniques for improving
the difficult foundation soils were developed. The
implementation of this methods and techniques must take into
consideration the soil characteristics intercepted on the site.
In the case of loess soils that are improved using soil
cushions a good vertical and horizontal systematization is
required to drain the rainwater or the water from other surface
sources and to avoid the appearance of settlements.

2428

ACKNOWLEDGEMENTS

This paper was realised with the support of POSDRU


CUANTUMDOC DOCTORAL STUDIES FOR EUROPEAN
PERFORMANCES IN RESEARCH AND INNOVATION
ID79407 project funded by the European Social Fund and
Romanian Government.
6

REFERENCES

Alupoae D. Baron A. Rotaru A. and Rileanu P. 2011.


Geomorphological characteristics of the Bahlui riverbed soils in
the metropolitan area of Iasi city, Romania, 11th International
Scientific Conference VSU, VI 141-146.
Ciornei A. and Rileanu P. 2000. How to dominate the macroporous
soil sensitive to wetting. Junimea, Iai.
Iliesi A.T. 2012. Geotechnical risk when building on collapsible soils.
PhD. Thesis. Iai.
Vieriu F. 2010. The study of Sarmatian clay and covering formations
from Iai City, seen as foundation soils. PhD Thesis. Iai.
NE 001-96. 1996. Design and building execution on high swelling and
shrinking soils.

Centrifugal and numerical analysis of geosynthetic-reinforced soil embankments


Etude par centrifugeuse et analyse numrique des remblais renforcs par gotextile
Bo L., Linli J. Ningyu Z., Sinong L.

School of Civil Engineering and Architecture, Chongqing Jiaotong University, Chongqing, China

ABSTRACT: Centrifuge models and numerical analysis of geosynthetic-reinforced and unreinforced soil embankments are presented.
The results obtained from the centrifuge tests were compared with those from the numerical analysis. It is found that the filamentous
fiber (polypropylene) is effective in constraining lateral displacement and reducing vertical settlement for the case of geosyntheticreinforced soil embankments. Also, the distribution of stress in the geosynthetic-reinforced soil embankment is significantly
ameliorated compared with the unreinforced. The presence of geosynthetic filamentous fibers in reticular structure provides the
reinforced soil embankments strength to resist crack.
RSUM : Dans cet article, les rsultats de modles de centrifugeuse et les analyses numrique des remblais renforcs par gotextile
et non-renforcs sont prsents. Les rsultats obtenus laide de la centrifugeuse sont compars avec ceux des analyses numriques.
Les fibres filamenteux (polypropylne) sont efficaces pour restreindre les dplacements latrales et rduire les tassements verticaux
dans le cas du remblais renforc. De plus, la rpartition des contraintes dans le remblai renforc est amliore de faon significative
compar avec celle du remblai non-renforc. La prsence des fibres dans une structure rticulaire dans le remblai renforc donne une
rsistance contre la fissuration.
KEYWORDS: Embankment ; Geosynthetic-reinforcement ; Centrifuge test ; Numerical analysis
1

INTRODUCTION

The concept and design theory of reinforced soil were proposed


by the French engineer Henri Vidal from model tests in the
1960s. The reinforcement materials include metal strips,
concrete slabs, bamboo ribs and geosynthetic materials, etc.
Now-a-days, geosynthetics was commonly used in reinforcing
soil owing to its easy-controlled properties of structure type and
size, strength, impermeability, acid dissolution and durability.
Cohesion of filamentous fiber reinforced soil comes from
friction between soil and fibers, as well as the constraint force
of the fiber network. The magnitudes of CBR and unconfined
compressive strength(UCS) increase with augment of
filamentous fibers linearly(Xiong Youyan 1989). Soil
reinforced with continuous filamentous fibers is obviously
effective in reducing the vertical deformation of sand under the
vertical pressure; it is superior in reducing horizontal tension
than geogrids(A.F.L.Hyde and M.Ismail 1988). In recent years,
this technique has applied successfully by reinforcing the
embankment using filamentous fibers in embankment projects,
and datum are available from researches (Bao Chenggang and
Ding Jinhua 2012). However, the interaction micro-mechanism
of interface between soil and filamentous fibers is still unclear
(Tang Chaosheng, Shi Bin and Gu Kai 2011, Jie Yuxin and Li
Guangxin 1999).
In this paper, the behavior of geosynthetic-reinforced
embankments has been explored using centrifugal and finite
element modeling. The objectives of this paper include: (1) to
probe the mechanism of filamentous fibers in improving the
stability of the embankment, and (2) to examine the
effectiveness of filamentous fiber reinforcement.
2

CENTRIFUGE TESTS

Centrifuge model testing, because of its ability to reproduce


same stress levels, same deformation and same failure
mechanism in an 1/ n scale model as in a full-scale prototype, is
widely used in studying geotechnical problems. Jie Yuxin and
Guang-Xin Li studied the stability of cohesive soil slope and
fiber-reinforced soil slope with different densities through
centrifugal model tests; Yang Xiwu and Ouyang Zhongchun

obtained the deformation behavior of embankments which


reinforced with various fiber styles. It should be pointed out that
idealized conditions may be created in centrifuge models
carefully to avoid problems caused by stress errors, boundary
effects, particle scale effects and geometrical scale effects.
2.1

Centrifuge testsEquipment and procedure

2.1.1
Equipment
In the present study, centrifuge model tests were performed
using the TLJ60 centrifuge in Chongqing Jiaotong
University. The main parameters of the centrifuge are indicated
in Table 1.
Table 1. The main parameters of the centrifuge
Characteristic
Maximum volume weight
Maximum load

Value
60gt
600kg(100g)
300kg(200g)

Effective radius

2.0m

Maximum acceleration

200g

Acceleration control accuracy


Model box size

0.5%FS
600mm350mm
500mm

2.1.2 Model scale


Due to the inherent symmetry of the embankment about its
centerline, only one half of it was modeled. In order to simulate
the actual project accurately and satisfy the boundary effects,
1:90 scale centrifuge model was constructed. Fig.1 shows the
details of test model and its full-scale prototype.

2429

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

(a)

(a) 1:90 scale centrifuge model (cm)

(b)
Figure 2. The marked model

2.2

(b) Full-scale prototype (m)


Figure 1. Arrangement of model for centrifuge test and its prototype

2.1.3
Parameters of soil and fiber
1The physical parameters of the soil
Table 2 gives the parameters of the brown weathered shale that
obtained from compaction test and liquid and plastic limit
combined test.
Table 2. Brown weathered shale material properties
Liquid
Plastic
Optimm
Maximm
limit (%)
water
dry density limit (%)
3
content (%)
(g/cm )
8.9

2.15

26.928

20.193

Plasticity
index
6.735

Before the centrifuge test, soil sample was experienced


airing and grinding, then sieved by 6mm sieve to remove
impurities.
2The parameters of the fiber
Polypropylene fiber with 19mm length was proposed to
construct the fiber reinforced soil embankment model. Table 3
gives the triaxial test strength of the embankment soil with the
fiber ratio of 1 , 2 and 3 respectively.

Centrifuge testssummary of results

2.2.1 Comparison analysis of deformation and displacement


In this section, the results obtained from unreinforced
embankment test are compared with the results obtained from
reinforced embankment test. The deformation of unreinforced
embankment was slightly larger than the deformation of
reinforced embankment. The settlements under the shoulder of
the unreinforced embankment and the slope gradient were
considerably greater than those of reinforced embankment. Two
cracks on the top of the unreinforced embankment and (heave)
beyond the toe of the unreinforced embankment were observed
at the end of the centrifuge tests. Fig.3 and Fig.4 show the
displacement vectorgraph of unreinforced and reinforced
embankment respectively. From the close comparison between
unreinfroced and reinforced embankments, it is evident that
fiber reinforcement reduced the displacement of embankment,
and enhanced the embankment obviously.

Table 3. Embankment soil material parameters


Embankment soil

Cohesion(kPa)

Friction
angle(degrees)

Unreinforced soil

49.167

34.077

19mm-0.1%Polypropylene-reinforced soil

94.005

35.717

19mm-0.2%Polypropylene-reinforced soil

138.294

36.362

19mm-0.3%Polypropylene-reinforced soil

228.356

35.951

2.1.4 Deformation measuring


An array of pins was installed on the front face of the
embankment model as deformation marker. This was used for
measuring the model vertical and horizontal displacement from
coordinate difference between beginning of test and the end of
test through the front perspex window. Fig.2 shows the details
of the marked model.

2430

Figure 3. Deformation of the unreinforced model

Figure 4. Deformation of geosynthetic-reinforced model

Technical Committee 211 / Comit technique 211

2.2.2 Comparison analysis of settlement


The maximum lateral displacement of unreinforced model was
72cm, which located in the distance of 8m from the toe of the
embankment. The maximum settlement was 48.6cm, which
located in the distance of 10.8m from the centerline; For
reinforced case, the maximum lateral displacement emerged in
the distance of 8.2m from the toe of the embankment with
35cm, and the maximum settlement was 41.2cm (located in the
distance of 10.4m from the centerline). It is safely to conclude
that the maximum displacement of both unreinforced and
reinforced embankment approximately close to the same point,
whereas the maximum lateral displacement of reinforced
embankment is approximately equal to 48.6% of the maximum
lateral displacement of unreinforced embankment and the
maximum settlement of reinforced embankment is
approximately equal to 84.8% of the maximum settlement of
unreinforced embankment.
The comparison between computation analysis and centrifuge
tests of the embankment discloses that fibers help to resist the
lateral thrust and lateral deformation of the embankment
effectively. This is due to the fact that fibers unified the overall
redistribution of stress and reduced asymmetric settlement of
embankment.
3

Fig.6 present computed displacement and stress contours of the


unreinforced and reinforced models respectively.

(a) computed displacement of unreinforced model

FINITE ELEMENT MODELLING

(b) computed displacement of reinforced model


Figure 5. Computed displacement of unreinforced and reinforced model

3.1

Assumptions of computing

In the analysis presented in this paper, the unreinforced and


reinforced embankments are modeled using the Drucker-Prager
constitutive model (D-P model).
Two-dimensional plane strain models were constructed with
boundary conditions similar to those of centrifuge models. The
modeling based on follow assumptions: (1) taking geotextile
reinforced soil as homogeneously isotropic material, the
parameters obtained from triaxial tests; (2) without considering
the influence of temperature to embankment; (3) consolidation
was completed under its gravity, and without considering the
impact of pore pressure.
3.2

Parameters

Table 4. Material parameters specified for the finite element analysis


PolypropyleneUnreinforced reinforced
Characteristic
Foundation
embankment
embankment
(19mm-0.1%)
Density
2150
2180
2150
(kg/m3)
Cohesion
49.167
94.005
49.167
(kPa)
Friction
34.077
35.717
34.077
angle(degrees)
Poissons
0.27
0.23
0.27
ratio
Depth
of
18
18
36
embankment(m)

3.3

(a) stress contours of unreinforced model

(b) stress contours of reinforced model


Figure 6. Stress contours of unreinforced and reinforced model

Displacement comparison

The computed results indicated that the values of deformation


and stress as well as its fluctuation range were marginally less
for reinforced embankment than for unreinforced embankment.
The maximum lateral displacement of unreinforced model was
79.442cm (located in the distance of 11.4m from the toe of the
embankment), and the maximum settlement was 51.498cm
(located in the centerline); The maximum lateral displacement
of reinforced model was 38.246cm (located in the distance of
11.4m from the toe of the embankment), and the maximum
settlement was 48.318cm (located in the centerline). Fig.5 and

RESULTS AND COMPARISIONS

Fig.7 shows the variation of lateral displacement and vertical


displacement of unreinforced embankment from centrifuge
tests. Superimposed on the measured variation are the variations
computed by numerical modeling analysis. It can be seen from
Fig.7 that there is a close agreement between the observed and
computed displacements for centrifuge test and numerical
analysis.

2431

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

effectively, and the distribution of stress and deformation was


harmonious comparing with the unreinforced.
When using geosynthetic fibers to reinforce embankment, it
also shows two advantages: (1) reinforced embankment can
resist cracks due to the network of intertwined fibers, and(2) the
fiber reinforced soil is closer to a homogeneous, isotropic
material than unreinforced soil.
REFERENCES
(a)

(b)
Figure 7. measured and computed displacement of unreinforced model

Vidal, M.H. 1978. The development and future of reinforced earth.


Proceedings of a SymposiumReinforcement at the ASCE Annual
Convention. Pittsburgh, Pennsylvania, 1-61.
Xiong Youyan.1989. Geosynthetic-reinforced soils. Chongqing
Highway Science Research Institute, Chongqing, China.
Tang Chaosheng, Shi Bin and Gu Kai. 2011. Microstructural study on
interfacial interactions between fiber reinforcedment. Journal of
Engineering Geology19(4), 610-614.
Jie Yuxin and Li Guangxin. 1999. A study on colculation method of
texsol. China Civil Engineering Journal2(5), 51-55.
Bao Chenggang and Ding Jinhua. 2012. Researches and applications of
fiber reinforced soils. Soil Engineering and Foundation26(1), 80-83.
Jie Yuxin, Li Guangxin and Chen Lun. 1998. Study of centrifugal
model tests on texsol and cohesive soil slopes. Chinese Journal of
Geotechnical Engineering20(4), 12-15.
Yang Xiwu and Ouyang Zhongchun. 2000. Experimental study on the
strengthened sreep slopes. China Civil Engineering Journal33(5),
88-91.

The comparison between the computed and observed


displacement both in the horizontal direction and in the vertical
direction for the reinforced embankment are shown in Fig.8.
The computed displacement is quite close to the observed
values for both lateral displacement and vertical displacement.

(a)

(b)
Figure 8. Measured and computed displacement of reinforced model

The behavior of reinforced embankment and unreinforced


embankment was successfully investigated using centrifuge
modeling and finite element analysis. The comparisons between
the centrifuge tests and computed results indicated the utility of
fibers can enhance overall stability of embankment. For the case
of reinforced embankments with fibers, it was found that the
deformation, the magnitude of stress, and their variation range
was considerably less than those for unreinforced case. Also,
the fiber reinforcement constrained the lateral displacement

2432

Compacted soil columns for foundations on collapsible soils. Laboratory and in-situ
experimental study
Colonnes de sols compacts utilises pour des fondations sur sols effondrables.
tude exprimentale mene en laboratoire et in-situ
Burlacu C., Olinic E., Manea S.

Technical University of Civil Engineering Bucharest, Department of Soil Mechanics and Foundations

U P.

SC Geosond SA Bucharest

ABSTRACT: Moisture-sensitive or collapsible soils are materials with high porosity that under the loads transmitted by the
foundations present additional settlements once the soil is saturated. This category includes loess deposits and other high silt content
soils with uneven porosity. A method often used for foundation on these soils is the realization of local loessoid material compacted
columns. According to the Romanian legislation, it is forbidden to use granular material in loessoid soils. A compromise may be
reached by using a mixture of granular material and local loessoid soil in columns. This paper presents the experimental laboratory
program aiming to achieve an optimal mixture of local material (loess) and monogranular sand in order to improve the values of the
mechanical soil parameters while keeping the permeability coefficient values as low as possible. This objective can be achieved by
adding bentonite. On the experimental polygons, 1:5 scale compacted soil columns were made using a dynamic penetrometer. The
aim of the dynamic penetration tests performed in the center and between the columns was to obtain results concerning the
improvement of the mechanical characteristics of the columns and the foundation soil.
RSUM : Les sols effondrables sont des matriaux avec une porosit leve, qui, suite la saturation, prsentent des tassements
supplmentaires sous leffet des charges transmises par les fondations. Cette catgorie inclue les dpts de loess et dautres sols ayant
un contenu lev de silt avec une porosit irrgulire. Une des mthodes de fondation souvent utilise sur ce type de sols est la
ralisation de colonnes de matriel loessique compact. Bien que la lgislation roumaine interdise lutilisation de matriaux sableux
dans des sols loessiques, ceux-ci peuvent toutefois tre utiliss pour la ralisation de colonnes dans un mlange avec du sol loessique.
Cet article prsente un programme exprimental de laboratoire qui vise raliser un mlange optimal de matriaux lssiques avec du
sable pour amliorer les valeurs des paramtres mcaniques du sol, en maintenant toutefois les valeurs du coefficient de permabilit
le plus bas possible. On peut atteindre cet objectif par laddition de bentonite. Dans le cadre dun programme exprimental, on a
realis des essais l'aide d'un pntromtre dynamique sur des colonnes de sol compact une chelle de 1:5. Le but des essais de
pntration dynamique a t dobtenir des rsultats concernant lamlioration des caractristiques mcaniques des colonnes et du
terrain de fondation.
KEYWORDS: collapsible soils, compacted soil columns, dynamic penetration test, soil mixtures.
1

INTRODUCTION

Moisture-sensitive or collapsible soils are unsaturated


macroporous cohesive soils that, upon saturation with water,
undergo sudden and irreversible changes of the internal
structure, reflected by additional settlements with collapsing
character and decreases in the values of geotechnical parameters
of mechanical behaviour (NP 125: 2010).

In Romania, moisture-sensitive soils cover about 19% of the


countrys territory (approx. 40.000 km2) and it is common
particularly in the eastern part of the country (Figure 1).
In order to characterize a soil as moisture sensitive, it must
meet at least one criterion regarding the physical characteristics
and one criterion regarding the mechanical behaviour, the main
criteria being the following:
A. Criteria regarding physical characteristics:
- ratio of silt fraction: 50 80%
- degree of saturation: Sr < 0,8
- porosity in natural state: n > 45%
B. Criteria regarding mechanical behaviour:
- the index of the additional settlement caused by saturation
under a loading of 300 kPa (in oedometric test): im300 2%.
2 IMPROVEMENT METHODS FOR COLLAPSIBLE
SOILS

Figure 1. Collapsible soil spreading in Romania (Bally,Antonescu 1971)

Difficult foundation soil improvement methods are


continuously progressing, not only quantitatively, but also
qualitatively, as a result of both the development of new
technologies and the recognition of economic and
environmental protection benefits of modern methods.
A significant number of techniques aimed at improving the
mechanical characteristics of difficult foundation soils have

2433

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

been developed. Methods are divided into two wide categories


(Schlosser 1997):
- physical methods soil improvement technologies, by which
soil structure is mainly improved in what concerns contacts
between particles by additives or by reducing porosity in
order to increase the tamping state - physical characteristics
improvement methods;
- mechanical methods soil reinforcing technologies, by
which structural elements are introduced in the ground in
order to increase the mechanical strength mechanical
characteristics improvement methods.
Classification of results sought by difficult foundation soil
improvement (Kirsch, Sondermann 2003):
- increasing density and shearing strength
- reducing compressibility
- influencing permeability in order to reduce infiltrated water
flow or to speed up consolidation process
- improving homogeneity.
3

percentage of sand in the mixture (from 20% to 40%), the


maximum density in dry condition increases. At the same time,
the optimal compaction moisture of the mixtures decreases
The synthesis of the oedometre compressibility tests
depending on the oedometric moduli values indicated that the
same values Eoed 200-300 could be obtained for the mixture
containing an addition of sand of 20%, at smaller humidity
values and at a better tamping state than in case of the natural
loess samples. This trend disappeared once the percentage of
sand in the mixture was increased (40%). In what concerns
samples with bentonite, similar values of oedometric moduli
were obtained at a better tamping state that in case of medium
loess samples, but at a reduced tamping state than in case of
samples with sand, which was also confirmed by the values
obtained following Proctor tests.

LABORATORY TESTS

In the experimental programme, various mixtures of loessoid


material with different natural mineral materials have been
proposed, in view of eliminating moisture sensitiveness,
improving geotechnical parameters of mechanical behaviour
and limiting permeability (Burlacu 2012).
To this purpose, a series of mixtures have been proposed:
loess with sand 1-2 mm (Cu = 1.5) and loess with sand and
bentonite powder addition in two variants of mixture. The
obtained mixtures are presented below:
Mixture 1: 80% loess + 20% sand (1-2 mm);
Mixture 2: 60% loess + 40% sand (1-2 mm);
Mixture 3: 50% loess + 40% sand (1-2 mm) + 10%
bentonite;
Mixture 4: 50% loess + mixture from (40% sand (1-2 mm) +
10% bentonite);
The difference between the last two mixtures consisted in the
way they were mixed. In the first case, all the three materials
were simultaneously mixed and then water was added to reach
different degrees of humidity in order to perform the normal
Proctor test. In case of the last mixture, the sand was first mixed
with the bentonite and with water and then, after this mixture
had dried, it was also mixed with the loess (Olinic 2012).
As a first step, the optimal compaction characteristics of the
proposed mixtures were determined and then, based on the
compacted samples, the compressibility and shearing
mechanical characteristics and the possible moisture-sensitivity
of the compacted material were determined. The samples used
for carrying out the mechanical tests were the ones surrounding
the optimum compacted sample. In order to reach uniform
results, the variation of the density in dry condition depending
on the height of the compacted sample was analyzed and
confirmed (Figure 2). This is why a certain sampling order was
followed.

Figure 3. the results of the Proctor trial for all the mixtures obtained.

If, in case of mixture 3, the Proctor diagram has a maximum


point (dmax, wopt), in case of mixture 4, the same tamping state
was obtained for humidity values between 11% and 15%.
The Proctor diagram resulted for mixture 4 indicated that
sample 3 could have represented a maximum point. Therefore,
in order to validate the results, tests on this sample were carried
out again and similar values were obtained (Figure 4). Given
that, humidity plays a key role in the real scale compaction
process, the last indication regarding mixture 4 is important
because it allows compaction at humidity values belonging to
higher humidity domains.

Figure 4. Results of Proctor test for mixtures 3 and 4.

Figure 2. Dry density teoretical variation depending on the height of the


compacted sample.

As a result of the Proctor test outcome analysis (Figure 3), it


has been observed that along with adding up and increasing the

As to the values of the permeability coefficient, these have


been of the order of 10-5 cm/s for the average loess sample
rising up to values of 10-4 cm/s in case of the mixture containing
40% sand, while in case of the mixtures containing an addition
of bentonite, the measured values were below 10-9 cm/s.

2434

Technical Committee 211 / Comit technique 211

IN SITU TESTS

In-situ tests first aimed at identifying the effect of the cone


shape on: the rate at which the cone advances in the ground, the
tamping of the surrounding ground and the compaction degree
of the material in the column body. Three types of cones with a
diameter of 7 cm were made. (Figure 5).

Then, for determining the cone with a wider influence


radius, DPM tests were carried out at a distance of 2Dc=14 cm
towards the columns. These tests indicated [as expected given
its shape (the smallest angle at the top)], that cone no. 1 (300)
had the greatest influence on the tamping state of the soil
around the column.
The tests indicated that cone 1 shape (30) was optimal for
soil columns execution.
4.3

Compacted loess columns

Therefore, 2 m long columns were executed, arranged as an


equilateral triangle network (Figure 7) with a distance of 3Dc 21 cm between the columns.
After finishing the group of columns, average dynamic
penetration tests were conducted both between the columns and
at different distances towards them.
Figure 5. Cones made: a) C1 30; b) C2 60; c) C3 hemisphere.

4.1

Column execution technology

Collapsible soils improvement by soil columns is regulated by


normative C29 - 85. The experimental polygon met the column
execution methodology described in the normative but adapting
it to the equipment that has been newly proposed for their
execution (LMSR-Hk dynamic penetrometer).
Column execution steps are: column hole execution, filling
by fill material portions and fill material compaction until
rejection.
The fill material portion was set for a column with a
diameter of 7 cm and for a height of the compacted material of
21 cm (3 diameters), resulting 1,5 kg of material having optimal
compaction humidity.
Successive tests regarding the obtained compaction degree
indicated that rejection (compaction stopping) was reached after
an advance of maximum 7mm/blow.
4.2

Figure 7. Columns and DPM tests disposal.

Optimal cone shape

On the experimental polygon, columns were executed by using


the three types of cones. Figure 6a presents the blow number
variation per an advance of 10 cm in DPM tests performed in
the centre of the columns and Figure 6b presents the same tests
carried out at a distance of 2 diameters towards the column.

Figure 6. Results of DPM tests carried out in the centre of the columns
and at a distance of 2Dc (14cm) towards the columns.

Figure 8. DPM tests results: a) between the columns; b) @ 1Dc;


c) @ 2Dc; d) @ 3Dc.

In case of DPM tests carried out in the centre of the columns,


the results obtained were similar for all the columns. Therefore,
an optimal shape of the cone that leads to a better compaction of
the column body could not be found.

It may be observed that, at a distance of 3Dc near the


columns, the improvement effect has no longer been perceived
(Figure 8d). When the DPM test was carried out in the centre of
the column group (Figure 8a), the improvement effect recorded
an obvious increase.

2435

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

4.4

Loess and sand mixture compacted columns

Finally, on the experimental polygon columns of 60% local


material (loess) and 40% sand were executed. For the execution
of these columns, cone 1 (30) was also used as rammer.

Bally R.J. and Antonescu I. 1971. Loessoid soils in constructions.


Tehnica Publishing House, Bucharest. (in Romanian)
Burlacu C. 2012. Contributions to improvement solutions for weak
foundation soils. PhD Thesis. Technical University of Civil
Engineering Bucharest, Romania.
Kirsch K. and Sondermann W. 2003. Geotechnical engineering
handbook, Volume 2: Procedures, Chapter 2.1. Ground
improvement, 1 - 50. Ernst & Sohn, Berlin, Germany.
Olinic E. 2012. Personal comunication.
Schlosser. 1997. Expos sur la them: Amliorqtion et renforce,ent des
sols. Proceedings of the Fourteenth International Confrence on Soil
Mechanics and Fountation Engineering, Volume 4, 2445 2466,
Hamburg, Germany.
NP 125:2010. Normative for foundation of buildings on moisture s
sensitive, collapsible soils. (in Romanian)

Figure 9. DPM tests results for the column executed from a mixture of
loess and sand (N1): a) DPM in the centre of C1 and N1 columns; b)
DPM at 2Dc towards C1 and N1 columns

Figure 9a indicates that the results of the DPM tests carried


out in the centre of the column executed from a mixture of loess
and sand are better that those of the column made entirely of
loess. Moreover, it may be noticed that, unlike the compacted
loess column, the one made of mixture led to the improvement
of the material under the column's body.
Nevertheless, the results of the medium dynamic penetration
test carried out near the columns at a distance of 2Dc (~14 cm)
indicate that the tamping effect is higher that in case of the loess
column.
5

REFERENCES

CONCLUSIONS

Laboratory tests aimed at identifying a mixture of loess and


natural mineral materials, with better mechanical characteristics
and with reduced permeability compared to the one the loess
has in its natural state. From all the solutions proposed
(compacted loess, mixture of loess and sand and mixture of
loess, sand and bentonite) the last one (mixture 4 - sand and
bentonite, mixed with loess after drying) seems to be the
optimal one due to the wide domain in which optimal
compaction parameters are reached.
Concerning mechanical characteristics, no significant
differences seem to exist between the analysed mixtures, but
one can notice that water sensitivity is significantly reduced and
that, compared to the flooded loess, the values obtained are
significantly better.
In-situ tests, performed with a penetrometer, simulated the
execution of loess columns and of loess with compacted sand
columns, at a scale of 1:5. Both the quality of the material in the
column body and the effect on the surrounding ground were
verified by typical tests. The sand improves the mechanical
behaviour of the material in the column body, without
significantly exceeding the mechanical behaviour of natural
loess that has not been flooded.
By executing columns of compacted local material with
natural mineral materials, the mechanical behaviour of the
columns - loess complex that has not been flooded does not
improve, but this technique leads to some nuclei capable of
reducing the negative effect of the accidental flooding of loess.

2436

Selected problems connected with the use of the jet grouting technique
Certains problmes lis lapplication de la technologie dinjection de jet
Bzwka J., Juzwa A., Wanik L.

The Silesian University of Technology, Gliwice, Poland

ABSTRACT: The paper presents selected problems connected with the use of the jet grouting technique. It is one of the most popular
methods for subsoil strengthening, enhancement for existing foundation, vertical and horizontal waterproof cut-off walls. Columns
made using this method feature a high bearing capacity (very high friction on the shaft). The newest achievements of the numerical
explanation of the interaction between jet grouting columns and subsoil are presented in the paper. The created models will be used to
verify engineering methods of jet grouting columns dimensioning. Computational analyses are conducted using software based on the
finite element method (Z_Soil).
The computational model describes the interaction between a group of jet grouting columns and soil. The main element of this
analysis consists of selection and calibration of computational model of the "group of jet grouting columns subsoil" interaction.
The model space is divided into three zones: columns, soil and the contact layer formed between the columns and the soil massif.
The computational model allows for a plastic character of deformation under load and especially for a non-linearity of contact zone.
The description of shape of a shaft surface of jet grouting columns is very difficult, so the fractal theory is used to describe this shape.
Fractal and box dimensions are used to estimate the irregular surface. This model allows a precise selection of formation parameters,
like the injection rod pull out velocity and number of rotations, injection pressure and the water/cement ratio, which define the
geometry of jet grouting columns.
RSUM : Cet article prsente quelques problmes lis lutilisation de la technique damlioration des sols : le jet grouting. Cette
technique est une de mthodes les plus utilises pour renforcer le sous-sol, les fondations dj existantes et on sen sert aussi comme
les parois verticales et horisontales tanches (impermables leau). Les colonnes ralises par cette mthode se caractrisent par la
grande capacit portante (coefficient de frottement trs lev sur la surface latrale). Dans cet article, on prsente les dcouvertes les
plus rcentes lies aux modlisations numriques de linteraction entre les colonnes de jet et du sol. Les modles dvelopps seront
utiliss pour vrifier les mthodes dingnierie et pour dimensionner les colonnes de jet grouting. Les analyses numriques sont
effectues par les programmes bass sur la mthodes des lments finis (Z_Soil).
Le modle de calcul dcrit linteraction entre un groupe de colonnes de jet grouting et le sol. Lapport le plus important de cette
analyse rside dans le choix et le calage du modle de calcul pour linteraction groupe de colonnes de jet grouting sol". Lespace
du modle est divis en trois zones: colonnes, sol et couche de contact forme entre les colonnes et le massif du sol. Le modle de
calcul permet davoir des dformations plastiques et en particulier de dformation non-linaire de la zone de contact.
La description de la forme des surfaces latrales de ces colonnes est extrmement difficile, alors on a introduit la thorie de fractales
pour la dcrire. Les dimensions de type fractal et de bote, sont utilises pour estimer la surface latrale irrgulire des colonnes. Ce
modle permet de choisir dune manire prcise des paramtres de formation des colonnes tels que: vitesses - de rotation et
davancement en descente de la tige de forage, le nombre de rotation, la pression de coulis inject, rapport eau/ciment, qui dfinissent
la gomtrie des colonnes de jet grouting.
KEYWORDS: jet grouting technique, interaction between columns and subsoil, shape and dimensions of jet grouting column.
1

JET GROUTIMG COLUMNS INTERACTION


WITH SUBSOIL

The jet grouting method is frequently used in the engineering


practice. It may be used for nearly all types of soils, both natural
and man-made. It does not work only for a subsoil built of
organic soils. The method consists in a high-pressure injection
into the subsoil of an injectant stream (most often being a
cement grout), which cuts and disintegrates the soil body,
forming after binding with soil fractions a petrified soilcement composite of any geometrical form, e.g. close to a
column cylinder shape. This solution because of the speed of
performance and very good parameters of subsoil strengthening
is frequently used to strengthen a weak subsoil under high
transport embankments or bridge abutments (Bzwka 2009;
Juzwa 2012b; Modoni and Bzwka 2012).
To explain the interaction between the jet grouting columns
and the strengthened subsoil it is suggested to apply numerical
methods and to build models reflecting the operation of a single
column and the interaction of jet grouting columns group in

transferring the load to deeper soil layers. A solution is sought,


which would allow optimising design solutions of jet grouting
columns, would ensure safety of a structure designed this way
and at the same time would contribute to the works costs
cutting. The authors emphasise especially as precise as possible
reflection of real conditions, existing on a site.
A single column and a group of columns are the subject of
numerical and in situ analysis. A single column is an idealised
form, seldom existing in practice. However, the analysis of its
behaviour is a starting point to make models more realistic and
built of a group of columns. For the needs of analysis of
interactions occurring between jet grouting columns
strengthening a weak subsoil and the soil body numerical
models were constructed, considering the environment division
into three material zones: the soil-cement material of jet
grouting columns the contact layer the subsoil (Bzwka
2009, 2010).
Because of a physical inhomogeneity and of a complicated
geometrical arrangement the finite element method was used to

2437

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

build models and the Z_Soil software was used for


computations. An elastic ideally plastic model of Coulomb
Mohr boundary surface with non-associated law of flow were
adopted to describe the mechanical behaviour of the soil
environment and the jet grouting columns material.
To perform computer simulations it is necessary to give the
following parameters: angle of internal friction , angle of
dilatancy , cohesion c, modulus of elasticity E and Poissons
ratio . Values of parameters for soils building the model
subsoil were taken based on in situ tests on a test site. The
following values were taken, for sand: E = 55.5 MPa, = 0.3,
= 31.8, c = 1 kPa, for a cohesive soil interbedding: E = 33.8
MPa, = 0.3, = 18.0, c = 30 kPa. The value of angle of
dilatancy was introduced from the range of values =
(0.350.40). Determination of material parameters for a
cementsoil material depends on the subsoil ground
characteristics, cement type in the grout, the method of columns
performance. To determine them it is necessary to take core
samples from the column performed (Fig. 1). These samples are
then tested for uniaxial and triaxial compression. For the needs
of this study 10 samples were tested for each case, obtaining
results of significant scatter (Bzwka 2009). A statistical
analysis of result values was carried out and after approximation
with the first type regression function the following parameters
were taken for calculations: E = 9888 MPa, = 0.186, =
59.3, c = 1772 kPa. Values of soil parameters (E, , , c) were
taken for the contact zone based on CPT sounding performed in
this area. Their values equal to soil parameters reduced by 1/3.

GEOMETRY OF JET GROUTING COLUMNS

The shape of columns made by the jet grouting technique, due


to specific nature of this technology, is very diversified and
difficult to predict. It depends inter alia on the type and
condition of soils making the subsoil, the injection system used
(single, double or triple) and on technological parameters
(injection pressure, size and shape of injection nozzles, speed of
injection rod pulling out and rotations and others) (Wanik and
Bzwka 2012).
To determine precisely the geometry of jet grouting columns
they are excavated, making their measurement and macroscopic
visual inspection possible. The shaft may have various shapes
(Fig. 4) depending on the aforementioned factors.
a)

b)

Figure 2. Model deformations [m] under influence of the load of


embankment: a) h=2.0 m; b) h=4.0 m high (Z_Soil) (Bzwka et al.
2012; Juzwa 2012a).
Figure 1. Core samples for strength tests (Bzwka, 2009).

A 2D model was built cutting from the space around


columns an area large enough, allowing idealisation of
boundary conditions. Boundary conditions were taken in the
form of: full fixing of the base of the halfspace cut and partial
fixing, allowing a vertical shift, on side surfaces of the halfspace
In the model of a flat system a group of 3 columns was
taken, each of them 4.0 m long and 0.8 m in diameter, arranged
at a distance of 2.5 m, while the subsoil is stratified. Division
into quadrilateral isoparametric elements was assumed. The grid
was concentrated in the area of contact zone. An incremental
load (uniform for all columns) was applied to such system,
reflecting a real transport embankment 4.0 m high, laid at fixed
intervals in layers 0.5 m thick.
The image of system deformations caused by columns
loading is presented for two stages in Fig. 2. Corresponding
stresses are shown in Fig. 3. The stress maps perfectly show the
range of transition zone, which parameters affect the
distribution of internal forces values in the system (Bzwka et
al. 2012; Juzwa 2012a).

Fractal theories may be used to describe an irregular surface


of jet grouting columns. Using a fractal and a box dimension it
is possible to describe better an irregular shaft surface of a jet
grouting column, its shape and roughness. A more precise
description of roughness and geometrical parameters of soil
particles allows a more detailed determination of such
properties as: porosity, density and shear strength (Bzwka and
Skrzypczyk 2011).
The paper presents an example of fractal dimension and box
dimension calculation for an excavated jet grouting column
made in a single system (see Fig. 68). Results of studies
presented in papers (Kawa and Wieczorek 2005; Wanik 2012a,
2012b; Wanik and Bzwka 2012) have been used.
The described jet grouting column was made in average
compacted medium sand, under which a stiff silty clay was
situated. After column excavating and cleaning, an irregular
shaft surface was disclosed and also a clear change of column
diameter on the boundary of two layers forming the subsoil (see
Fig. 5).

a)

2438

Technical Committee 211 / Comit technique 211

a)

a)

b)

b)

c)

d)

Figure 3. Map of vertical stresses of the model under influence of the


load of embankment: a) h=2.0 m; b) h=4.0 m high (Z_Soil) (Bzwka et
al. 2012; Juzwa 2012a).
Figure 4. Different shapes of excavated jet grouting columns
(photos: J. Bzwka, and K. Wanik).

2439

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Figure 5. Structure
(photo: J. Bzwka).

of

excavated

jet

grouting

column

SUMMARY

Issues presented in the paper show the scale of problems related


to the representation of actual interaction of jet grouting
columns with the surrounding subsoil. Theoretical models
require repetitions and calibration, making the obtained results
realistic. It is especially important to determine the thickness
and parameters of the contact zone formed at the contact of
column material and the subsoil.
The shape and dimensions of formed jet grouting columns
depend on the type and condition of soils building the subsoil
and on technological parameters of columns forming, such as:
the injection pressure, the injection rod pulling out and rotation
speed, the density of injected cement grout as well as the
number and size of injection nozzles.
A large number of factors affecting geometry and hence
related columns bearing capacity and the soilcement material
strength causes problems in designing. To verify geometry of
columns made it is necessary to perform excavations and to
measure the diameter, circumference shape and to assess the
shaft structure. Mathematical issues from the field of fractal and
box dimension allow creating a clear description of
a complicated shape of jet grouting columns shaft.
4

ACKNOWLEDGEMENT

The coAuthors: Anna Juzwa and Lidia Wanik received a grant


of the DoktoRIS project a scholarship program for innovative
Silesia region cofinanced by the European Union of the
European Social Fund.
5
Figure 6. Method for determining fractal dimension for column
(Kawa and Wieczorek 2005; Wanik 2012a).

Figure 7. Method for determining box dimension for column


(Kawa and Wieczorek 2005; Wanik 2012a).

Figure 8. Fractal dimension and box dimension for jet grouting column.

2440

REFERENCES

Bzwka J. 2009. Interaction between jet grouting columns and subsoil.


Monograph published by the Silesian University of Technology,
Gliwice (in Polish).
Bzwka J. 2010. FEM analysis of interaction of jet grouting column
with subsoil. Scientific Conference on Natural and Technical
Problems of Environmental Engineering Soil parameters from in
situ and laboratory tests, Pozna 27-29 September 2010, 445455.
Bzwka J. and Juzwa A. and Wanik L. 2012. Selected problems of jet
grouting application. Inynieria Morska i Geotechnika, No. 4,
514519 (in Polish).
Bzwka J. and Skrzypczyk J. 2011. Fractal dimensions in geotechnics.
Proc. of the 9th International Conference on New Trends in Statics
and Dynamics of Buildings, 20-21 October 2011, Bratislava,
Slovakia, 2124 (in Polish).
Juzwa A. 2012a. Computational description of interaction between
group of jet grouting columns and subsoil. Monograph:
Experimental and theoretical tests in Civil Engineering published
by the Silesian University of Technology, Gliwice, 6774 (in
Polish).
Juzwa A. 2012b. Subsoil strengthening by using jet grouting
technology. 9th fib International PhD Symposium in Civil
Engineering, Karlsruhe Institute of Technology, Germany, 22-25
July 2012.
Kawa K. and Wieczorek W. 2005. Fractals application in geotechnics.
Master thesis, The Silesian University of Technology, Faculty of
Civil Engineering, Gliwice (in Polish).
Modoni G. and Bzwka J. 2012. Analysis of foundations reinforced
with jet grouting. ASCEJournal of Geotechnical and Geoenvironmental Engineering.
Wanik L. 2012a. Application of fractals to describe shape of jet
grouting columns. Monograph: Experimental and theoretical tests
in Civil Engineering published by the Silesian University of
Technology, Gliwice, 133141 (in Polish).
Wanik L. 2012b. Fractal and box dimensions in description of jet
grouting columns geometry. Inynieria Morska i Geotechnika, No.
4, 432434 (in Polish).
Wanik L. and Bzwka J. 2012. Influence of various factors on geometry
of jet grouting columns. Zeszyty Naukowe Politechniki
Rzeszowskiej, Budownictwo i Inynieria rodowiska, z.59
(3/12/IV), No. 283, t. 4, 117124 (in Polish).

Column Supported Embankments for Transportation Infrastructures: Influence of


Column Stiffness, Consolidation Effects and Cyclic Loading
Remblais sur sols renforcs avec de colonnes ballastes pour les infrastructures de transport:
Influence de la rigidit des colonnes, des effets de consolidation et du chargement cyclique
Carvajal E., Vukoti G.

Kellerterra S.L., Madrid, Spain

Sagaseta C.

University of Cantabria, Santander, Spain

Wehr W.

Keller Holding GmbH, Offenbach, Germany


ABSTRACT: Ground improvement methods based on column-type elements are analyzed regarding the influence of the column
properties on serviceability and safety of the Column Supported Embankments (CSE). Particularly, treatments made by rigid
inclusions are analyzed and compared with stone columns. Stiffness of column-type elements determines the design and risks
involved. Rigid inclusions are analyzed according to the recent French national project ASIRI. In the case of these elements, a
considerable mobilization of negative skin friction and punching effects governs their behavior in the Ultimate Limit State, which
represents a non-ductile mechanism of failure. Whereas stone columns present a ductile behavior determined in the domain of
Serviceability Limit State (SLS). It is pointed out, that possible damages on CSE systems may extend settlement stabilization due to
the consolidation process, if no drainage elements are adopted. It is also noted that risks related to rigid columns in the SLS under
cyclic loading, may be decisive in the design of CSE composed by low-heights embankments. Briefly, it could be stated that rigid
inclusions present higher risks, increasingly when their diameters are smaller than 30 cm.
RSUM : On analyse les mthodes d'amlioration des sols avec des colonnes pour la fondation des remblais sur sols mous. En
particulier, on analyse les inclusions rigides selon les recommandations du rcent projet national franais ASIRI, et on prsente la
comparaison avec des colonnes ballastes. La rigidit de la colonne dtermine la conception et les risques associs. Dans le cas des
inclusions rigides, une mobilisation considrable du frottement ngatif et la portance rsultante gouvernent leur comportement dans
l'tat limite ultime, ce qui reprsente un mcanisme non-ductile de rupture. Au contraire, les colonnes ballastes prsentent un
comportement ductile dtermine dans le domaine de l'tat limite de service. Il a t observ que les risques de colonnes rigides dans
les ELS peut tre retards moins que on installe quelques lments de drainage. On a remarqu aussi que les risques associs aux
inclusions rigides soumises aux chargements cycliques peuvent tre dcisives pour remblais de faible hauteur. Ainsi, les inclusions
rigides prsentent des risques plus levs, de plus en plus lorsque leur diamtre est plus petit que 30 cm.
KEYWORDS: Load Transfer Platform, geosynthetic, embankment, rigid inclusion, stone columns, risk, stiffness, arching effect

INTRODUCTION

Column Supported Embankments (CSE) represent an


innovative solution for transport infrastructure over soft soils, in
order to reduce execution time and general earthworks. Hence,
the use of low-height embankments based on column-type
elements tends to be preferred, whenever possible, instead of
direct soil replacement or preloading with or without vertical
drains. Recently, the use of CSE is increasing, and consequently
growing interest in developing reliable and unified criteria for
their design and construction is observed.
However, due to the possibility of application of a wide
range of ground improvement techniques, further risk
assessment has to be done. Risks and reliability related to CSE
could be largely analyzed considering the influence of column
stiffness in Ultimate and Serviceability Limit States.
Furthermore, column stiffness also affects consolidation process
and the system behavior against cyclic or dynamic loading, very
often decisive for safety and serviceability.
2
2.1

order to optimize the solution, ground improvement methods


have been increasingly used in the last years.
Ground improvement methods should intent not to take the
entire action by the supporting elements, but only the difference
between the required and existing bearing capacity without
improvement (Wehr et al. 2012). This is applicable to stone and
sand columns, which take important part of the foundation load,
and make the most of soil confinement to ensure its own
capacity. These two types of columns accelerate the
consolidation process and do not need any embedment to
transfer the loads to stiffer soil layers; thereby they can be
considered as authentic ground improvements.
On the other side, the columns made by the addition of
bonding agents, mortar or concrete into the ground, do not
accelerate consolidation. The improvement introduced by such
columns mainly consists of the load transfer to the stiffer layers
in the same way as piles, thus, to ensure their correct application
the largest embedment is frequently desired.
Load Transfer
Platform (LTP)

COLUMNS SUPPORTED EMBANKMENT SYSTEMS

HC

Embankment
Plane with S = 0

Geosynthetics

Pile cap

End Bearing
Columns

Type of columns

Soft Soil

Typical elements of CSE systems are shown in Figure 1.


Initially, reinforced piles with concrete cap were applied, in
order to absorb the largest load of embankment as possible. In

2441

Soil with
intermediate
Stiffness

Floating Columns

Firm Soil

Figure 1. Elements of Column Supported Embankment Systems

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

(b) (b)

(a)

(c)

(d)

Figure 2. Mechanism of load transfer in the CSE: (a) approaches of arching-effect shape; (b) resultsof laboratory test performed by Chen et al.
(2008); (c) load transfer mechanism proposed by Combarieu (1974, 1988); (d) influence of confined modulus on improvement factor (Kirsch 2004).

These kinds of columns, with predominantly round crosssections of 25 cm to 80 cm diameter, are denominated Rigid
Inclusion according to the French national research project
ASIRI (Amliorations de Sols par Inclusions Rigides). Rigid
inclusions may be arranged in a regular grid, although, due to
horizontal stresses sometimes have to be distributed in wall or
panel form in order to overcome slope and internal instability.
2.2

Load Transfer Platform

The design and operation of CSE is largely influenced by the


load transmission mechanism toward the columns, through a
Load Transfer Platform (LTP) laid out at the base of
embankment. LTPs are generally composed by a layer of
compacted granular material that in many cases has to be
reinforced by geosynthetics, or composed by layers treated with
hydraulic binder.
LTP behavior is essentially determined by two parameters.
The efficacy or efficiency E, defined as the ratio between load
on the column head QP and the total load on the surrounding
soil within a unit cell (W + Q), where W is the weight of
embankment and Q is the force due to surcharge on the surface;
and the critical Height HC, which indicates the height of
embankment where differential settlements in between column
head and middle of the grid are negligible. As stated by several
authors, E and HC depend on many factors such as column
rigidity, shear strength of LTP layers, spacing between columns,
and soft soil stiffness (Zaeske and Kempfert 2001, Okay 2010).
Most theoretical methods focus on the requirements of the
geosynthetic within LTPs for piled embankments, considering a
void between rigid elements. The geosynthetic takes the load
that remains in the middle of columns and delivers it to the
column heads by means of membrane effect. Consequently
almost all load is acting on the columns heads. According to
these methods only a minor part or even any soil reaction is
considered. Several guidelines or recommendations documents
deal with these methods (BS8006 2010, EBGEO 2010, Nordic
Handbook 2005). Such approaches could be classified
according to the shear stress form-distribution that governs the
mechanism of arch load-transfer and differential settlements
within the LTP (Han and Colling 2005), see Figure 2a.
According mentioned approaches HC varies from 0.7 to 1.6
times the clear distances between columns (s - a).

Figure 3. Influence of column modulus on the


differential settlements within Load Transfer Platform
(Gangakhedar 2004).

Otherwise, the method proposed by Combarieu (1974,


1988), and adopted in the ASIRI Recommendations, deals not
only with the load transfer into LTP but also along the entire
length of rigid columns. Furthermore, ASIRI project's
recommendations are based on various physical and numerical
modelling (Jenck 2005, Chevalier et al. 2008). 1
Figure 2c shows the mechanism of load transfer proposed in
the ASIRI, where differential settlements between soil and
columns produce negative skin friction in the upper part of the
column; at certain depth where settlements are the same in soil
and columns, the skin friction is equal to zero, and below this
neutral plane the load in the columns is transferred through
positive skin friction and tip resistance. It can be noted that such
mechanism is quite similar to those exhibited by the combined
pile-raft foundations (CPRF).
3
3.1

INFLUENCE OF THE COLUMN CHARACTERISTICS


Columns stiffness

Unfortunately, so far there is not any analytical method


(commonly used) that takes into account the variation of
column stiffness, and accordingly numerical modelling usually
have to be performed to analyze the influence of column
stiffness. However, even the most relevant numerical modelling
that can be found in the literature has no focus on the risks and
suitability aspects related to the column stiffness.
Kirsch (2004) analyzed the influence of the ratio between
confined modulus of columns and soil on the improvement
factor ratio of settlements with and without improvement).
Results indicate that confined modulus ratios beyond 40 to 50
do no suppose considerable increments on improvement factor
, (Figure 2d). Similarly, Gangakhedar (2004) performed a
numerical analysis of the influence of Youngs modulus of the
columns, on the differential settlements at the base of
geosynthetic reinforced embankment. Figure 3 shows that
differential settlements increase with increasing column
modulus. Although it can be noted that there exists a greater
increase of differential settlements when modulis are higher
than those usually obtained for stone columns, of about 80 to
120 MPa, and that differential settlements tends to be much
higher with the increase of column modulis if no geosynthetic
reinforcement is considered.
Therefore, the cost-operating inefficiency of columns may
be stated when column modulus are higher than 120 MPa, or
modulus ratio are larger than 40 to 50, approximately. If
columns rigidity exceeds this limits, CSE system requires an
increase on the capacity of geosynthetic-reinforcement and the
additional improvement is negligible.
It is well known that stone columns have a load-carrying
mechanism by lateral bulging, whereas rigid inclusions transmit
the load by skin friction and punching effect on their tip and
head. In the latter case, the usual amount of differential
settlement obtained in the column head implies a behavior
controlled by its ultimate limit state (ULS), and governed by
mobilization of negative skin friction. Figure 3 depicts that such

2442

Technical Committee 211 / Comit technique 211

Figure 4. (a) Estimation of evolution of negative skin friction with degree of consolidation (Alonso et al. 1984); (b) chart for geosynthetic design
based of allowable differential settlement (Lawson 2000); (c) influence of height and friction angle of embankment on Efficacy factor (Jenck 2005).

deformation in the head of rigid inclusions may suppose the


failure state, as settlements may reach levels corresponding
tolarge percent of column diameter. Furthermore, the punching
failure in the head and toe of columns occurs immediately after
the application of embankment load, and associated risk
increases with smaller diameters of rigid inclusion, being quite
sensitive to the variation of the soil parameters also.
On the other side, flexible elements like stone columns tend
to reduce the punching effects at the base of embankment. In
this case the system gives a ductile behavior, whereas, due to
column compressibility and its drainage characteristics, the
ultimate limit state is reached after large deformation and at the
end of consolidation. Therefore, the behavior of such system
takes place in the domain of serviceability limit state (SLS).
Wehr et al. (2012) proposed three categories of increasing
risks, in order to assess the reliability of ground improvement
methods according to their ductility and sensitivity to the
variation of soil and materials parameters, taking as a reference
the standards DIN 1054 and Eurocode 7. Thus, regarding to
columns-type elements, flexible columns with small risks (stone
columns, vibro compaction, sand columns) are in category A;
rigid columns with diameter larger than 30 cm, which presents
an average risk, are in category B; and rigid inclusions with
diameters less than 30 cm and non-ductile behavior, which
represent a high risk, are in category C.
3.2

Consolidation process

The addition of cement agents disables the drainage capacity of


rigid columns, whereby settlements stabilization is obtained
only due to a high load concentration on the columns. However,
during the consolidation of pore pressures produced by the
remaining part of embankment load that act on the soil, an
important negative skin friction is generated in the part of
columns above the neutral plane, very similar to piles, but
without any capacity and structural connections. Consequently,
the risk should be assessed due to possible reduction or loss of
the load concentration on columns (or efficiency factor) along
the lifetime of the CSEs. This situation could occur if certain
loss of arching effect happens, as a consequence of possible
LTPs deteriorations, e.g. due to internal failure of geosyntheticreinforcement. In this case, the consolidation would occur in the
long term, according to the permeability of the natural soil.
Moreover, it would involve the evolution of neutral plane
over the time, dominated by the increase of negative friction.
Figure 4a shows an example of this complex mechanism
reported by Alonso et al. (1984).
In the case of stone columns, the rapid settlements
stabilization is expected due to their drainage capability. Castro
and Sagaseta (2009) analyzed the evolution of stress
concentration on the stone columns, showing that in the very
beginning entire load is carried by the soil, and the final load
concentration on the columns is obtained after consolidation
(Figure 5). However, after short period of consolidation,

2443

effective stress of soil tends to increase, and additionally


provides greater confinement to the columns. Such results
suppose an improvement of the whole column-soil system.

Figure 5. Time development of soil and column stresses, (Castro and


Sagaseta 2009)

GENERAL ASPECTS OF SAFETY VERIFICATION

There is a range of recommendations that attempt to unify


design of LTPs composed by geosynthetic-reinforcement layers,
basically used in piled embankments (BS8006 2010, EBGEO
2010, Nordic Handbook 2005). However, the design of columntype elements is redirected to typical pile standards. As it was
mentioned in section 2.2, it has to be emphasized, that these
recommendations deal with systems where almost entire load is
transferred to bearing elements heads, hence negative skin
friction is practically negligible. According to what has been
stated here about the higher level of risk exhibited by the rigid
inclusions with small diameter, the most important safety
aspects of such elements will be commented.
4.1

Large-height embankment

The ASIRI recommendations define two different situations:


Domain 1: if the ULSs are not guaranteed without
improvement, rigid inclusions are used to ensure the global
stability, and bearing capacity of rigid inclusions for both ULSs
and SLSs have to be checked, similarly to the French Eurocode
7 application for piles.
Domain 2: if the ULSs are analyzed for the situation without
improvement, then rigid inclusions are used as settlement
reducers, and only SLSs have to be proceeded.
Taking into account the ASIRI recommendations, it could be
distinguished that when the CSE system comprises
embankments with more than 3 to 5 m height, the design is
usually focused to guarantee the ULSs. Regarding to the
external bearing capacity (GEO) for rigid inclusions, the most
important checks against the permanent loads will be punching
at their heads and tips, as well as the horizontal stresses,
bending moments and shear stresses due to slope
failures.Buckling effects have to be checked when soft soil has
pressuremeter modulus smaller than 3 MPa.

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

(a)

(b)

(c)

Figure 6. (a) Factor of soil arching reduction (Heitz et al. 2008); (b) stress conditions in the subgrade due to moving load on the pavement surface;
(c) pavement deformation due to hard-point effects associated with the presence of rigid inclusions.

Regarding to the structural bearing capacity (STR), a


minimum compressive strength of 7 MPa has to be adopted, and
no shear stress is allowed for unreinforced columns smaller than
30 cm. Besides, if tension can develop, for Domain 1 the rigid
inclusions have to be reinforced, whereas for Domain 2 only an
adequate tensile strength of concrete could be adopted.
On the other hand, Katzenbach et al. (2012) have compared
the safety checks outlined in the ASIRI recommendations with
other guidelines for similar foundation systems usually used in
Germany (CSV, CRPF), according to the partial safety factor
approach. They reported that ASIRI has lower values of safety
factors than those the compared guidelines indicate.
4.2

Low-height embankment

In the case of embankments with heights less than 3 meters,


the design is usually aimed to guarantee the SLSs, according to
the Domain 2. Basically, the geometry of the CSE systems has
to be set to avoid excessive deformation in the surface of the
embankments, in order to allow an adequate traffic operation.
For this objective Lawson (2000) proposed the chart depicted in
Figure 4b, for the design of the height and geosyntheticreinforcement of LTP layers considering the columns as hard
points, and according to typical thresholds adopted in transport
projects related to differential settlements.
The differential settlements also depend on the LTP strength.
Figure 4c shows the analysis of Jenck (2005) related to the
influence of the height of the embankment and the strength of
unreinforced LTPs in terms of friction angle. Results indicate
that efficiency factor E increase with height of embankments
until a maximum value similar to the critical height HC. Also, it
can be seen that when LTP is composed by materials with
friction angle less than 20 degree the efficiency factor is
drastically reduced, and practically negligible when = 0.
So far it is not fully analyzed the behavior of CSE against the
cyclic loading of traffic. Heitz et al. (2008) have demonstrated
that the arching mechanism to transfer load of LTP can only be
formed in a very limited extent if geosynthetic reinforcement is
not placed. Based on laboratory model tests under cyclic
loading, they proposed a soil arching reduction factor, k.
Figure 6a shows this factor depending on the ratio of fill
height and column spacing h/s, the frequency f and amplitude of
the cyclic load c.For rigid inclusion application negative
influence of the traffic loading has to be considered during
construction and operation stages. Figure 6b illustrates that
cyclic loading of traffic can generate the rotation of principal
stresses in the subgrade layers, which could cause severe
damages to the rigid inclusions and pavement serviceability in
the long term, especially for low-height embankments.
Finally, Figure 6c shows an example of pavement
deformation due to a combination of the effects mentioned.
5

CONCLUSIONS

The influence of columns stiffness commonly used on the


Column Supported Embankment (CSE) systems has to be
rigorously investigated in order to establish the implications on
the safety and serviceability issues. The facts that indicate the
higher risks of rigid inclusions compared with flexible ground

2444

improvement methods like stone columns are exposed,


especially when diameters of rigid inclusions are smaller than
30 cm. Moreover, the requirements of LTPs in terms of strength
and thickness, has to be more strict for rigid inclusion
comparing with stone columns, in order to ensure the arching
load transfer in the long term behavior of the CSEs, for both
static and cyclic loading.
6

REFERENCES

Alonso E. Josa A. and Ledesma A.1984. Negative skin friction on piles:


a simplified analysis and prediction procedure. Geotechnique 34.
No. 3. pp 341-357.
ASIRI National Project. 2012. Recommendations for the design,
construction and control of rigid inclusion ground improvements.
British Standard 8006. 2010. Code of practice for strengthened/
reinforced soils and other fills. British Standard Institution. London.
Castro J. and Sagaseta C. 2009. Consolidation around stone columns.
Influence of column deformation. Int. J. Num. Anal. Meth.
Geomech. 33(7): 851-877. doi:10.1002/nag.745.
Chen Y. M. Cao W. P. and Chen R. P. 2008. An experimental
investigation of soil arching within basal reinforced and
unreinforced piled embankments. Geotex. and Geom. 26. 164-174.
Chevalier B. Combe G. and Villard P. 2008. Modlisation discrte:
tude du report de charge. Rapport 3-08-4-01.
Combarieu O. 1988. Amlioration des sols par inclusions rigides
verticals. Application ldification des remblais sur sols
mdiocres. Revue franaise de gotechnique No. 44. pp 57-79.
EBGEO. 2011. Recommendation for design and analysis of earth
structures using geosynthetic reinforcement. Ernst & Sohn. Berlin.
Gangakhedar R. 2004. Geosynthetic reinforced piled-supported
embankments. Master thesis. University of Florida.
Han J. and Collin J.G. 2005. Geosynthetic Supported System over Pile
Foundations. ASCE. G.S.P. 130-142. pp. 3949-3953
Heitz C. Lking J. and Kempfert H.G. 2008. Geosynthetic reinforced
and pile supported embankments under static and cyclic loading.
Proceedings EuroGeo 4. Edinburg. United Kindong.
Jenck O. 2005. Le renforcement des sols compressibles par inclusions
rigides verticales. Modlisation physique et numrique. Thse de
Doctorat. INSA Lyon.
Kirsch F. 2004. Experimentelle un numerische Untersuchungen zum
Tragverhalten von Rttelspopfsulen, Dissertation am Institut fur
Grundbau un Bodenmechanik. Heft 75. Braunschweig.
Katzenbach R. Bohn C. Wehr J. 2012. Comparison of safety concepts
for soil reinforcement methods using concrete columns. Technische
Universitt Darmstadt. Institut un Versuchsanstalt fr Geotechnik.
Lawson C. R. 2000. Serviceability limits for low-height reinforced piled
embankment. Proceedings GeoEng 2000. Melbourne. Australia.
NGG. 2005. Nordic Handbook Reinforcedment soil and fills, Nordic
Geotechnical Society. Stockholm
Okay U.S. 2010. Etude exprimentale el numrique des transferts de
charge dans un massif renforc par inclusions rigides. Application
des cas de chargements statiques et dynamiques. PhD in the scope
of ASIRI. INSA Lyon and Universit Claude Bernard.
Wehr W. Topolnicki M. And Sonderman W. 2012. Design Risks of
ground improvement methods including rigid inclusions.
International Symposium Ground improvement. Brussels.
Zaesk D. and Kempfert H.G. 2001. Wirkungsweise von unbewehrten
und unbewehrten mineralischen Tragschichten ber pfahlartigen
Grndungselementen. Universitt Gh Kassel. Heft 10.

Foundations of embankments using encased stone columns


Fondations de remblais avec des colonnes ballastes entoures de gotextile
Castro J., Sagaseta C., Caizal J., Da Costa A., Miranda M.
University of Cantabria, Santander, Spain

ABSTRACT: Stone columns are a common improvement technique for foundations of embankments in soft soils. When the soft soil
does not provide enough lateral support, the columns are encased with a geosynthetic. This paper presents a closed-form solution to
study soft soil improvement, both reduction of settlement and consolidation time, by means of encased stone columns. An end-bearing
column and its surrounding soil, is modelled in axial symmetry under a rigid and constant load. Soil is assumed as elastic but plastic
strains are considered in the column. An elasto-plastic behaviour is also considered for the encasement by means of a limit tensile
strength. Parametric studies of the settlement reduction and stress concentration show the efficiency of encasing the columns, which is
mainly ruled by the encasement stiffness compared to that of the soil. The analytical results are in good agreement with numerical
analyses. Finally, the encasement length is analysed using the closed-form solution.
RSUM: Les colonnes ballastes sont une technique d'amlioration de sol pour les remblais en sols mous. Lorsque le sol mou ne
fournit pas assez de soutien latral, les colonnes sont entoures avec un gosynthtique. Cet article prsente une solution analytique
pour tudier l'amlioration des sols mous, la rduction des tassements ainsi que le temps de consolidation, au moyen des colonnes
entoures en gotextile. Une colonne ne reprenant les efforts que par la pointe et le sol environnant sont modliss en axisymtrie
sous une charge constant. Le comportement du sol est suppos lastique mais les dformations plastiques sont considres dans la
colonne. Un comportement lasto-plastique est galement pris pour le gosynthtique au moyen d'une rsistance la traction limite.
Des tudes paramtriques de la rduction du tassement et de concentration de contraintes montrent l'efficacit de l'enveloppe
gosynthtique des colonnes, ce qui est principalement rgie par la rigidit de lenveloppe gosynthtique par rapport celle du sol.
Les rsultats analytiques prsentent une bonne concordance avec les analyses numriques. Finalement, la longueur de lenveloppe
gotextile est analyse en utilisant la solution base sur une cellule lmentaire constitue dune colonne et dun volume lmentaire
de sol.
KEYWORDS: soft soils, ground improvement, encased stone columns, analytical solution, numerical analyses.
1

INTRODUCTION

Stone columns, either by the vibro-replacement or vibrodisplacement methods, are one of the most common
improvement techniques for foundation of embankments or
structures on soft soils. The inclusion of gravel, which has a
higher strength, stiffness and permeability than the natural soft
soil, improves the bearing capacity and the stability of
embankments and natural slopes, reduces total and differential
settlements, accelerates soil consolidation and reduces the
liquefaction potential. Alteration of the natural soft soil caused
by stone column installation (Guetif et al. 2008, Castro and
Karstunen 2010) is not usually considered in their design.
Stone columns may not be appropriate in very soft soils that
do not provide enough lateral confinement to the columns. It is
generally accepted that those are soils with undrained shear
strengths below 5-15 kPa (Wehr 2006). To increase the lateral
confinement of the columns, and consequently their vertical
capacity, encasing the columns with geotextiles has proved to
be a successful solution in recent years.
A high tensile stiffness of the encasement is recommended
as it will be shown in this paper; and therefore, other
geosynthetics, such as geogrids, are also used to encase the
column (Sharma et al. 2004, Gniel and Bouazza 2009).
However, geogrids do not act as a filter and do not avoid
contamination of the column with fines.
The development of encased stone columns as a ground
improvement technique has come with an increasing number of
studies in the last decade. However, most of the research is done
using numerical methods (e.g. Murugesan and Rajagopal 2006,
Malarvizhi and Ilamparuthi 2007, Smith and Filz 2007, Yoo
2010, Lo et al. 2010) and there are very few analytical solutions
available in the literature (Raithel and Kempfert 2000, Pulko et
al. 2011). That recently motivated the authors to develop a new
closed-form solution to study the deformation and consolidation

around encased stone columns (Castro and Sagaseta 2011). That


solution is an extension of another previous analytical solution
developed for non-encased stone columns (Castro and Sagaseta
2009).
This paper analyses the main features of that closed-form
solution, showing its limitations and range of applicability, the
influence of the key parameters for routine design and a
comparison with numerical analyses.
2
2.1

CLOSED-FORM SOLUTION
Model

The vertical capacity of the columns is a fundamental issue


when the applied load is concentrated on the columns.
Therefore, column encasement is very useful in those cases
(Murugesan and Rajagopal 2010, Khabbazian et al. 2010); but
also under distributed loads, such as tanks or embankments,
because the increase of lateral confinement reduces the
settlement.
The authors' closed-form solution (Castro and Sagaseta
2011) is limited to distributed uniform loads because it is based
on a unit cell model, i.e. only one column and its surrounding
soil are studied in axial symmetry. Furthermore, the column is
assumed to be fully penetrating in the soft soil and the applied
load is considered as rigid, i.e. uniform settlement. The area of
soft soil, Al, that is improved by each column, Ac, is generally
expressed by the area replacement ratio, ar=Ac/Al, but
sometimes is also defined in terms of the relation between
diameters or radii, N=rl/rc=1/ar.
The solution is developed for a horizontal slice at a depth z
of the unit cell, and consequently, shear stresses between slices
at different depths are not considered (Figure 1). The overall

2445

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

behaviour of the whole unit cell is obtained by means of


integration of the solution at the different depths.

Tg J g

pa

Axis

at any depth, z

zs

rs

rc

Column

rc

Encasement

rl

2.4
Figure 1. Analytical model.

2.2

Consolidation

The analysis of consolidation around encased stone columns as


a fully coupled problem is difficult to deal with. As a
simplifying assumption, the solution uses the average value of
the excess pore pressure along the radius, u , which is a simple
way of getting a reasonably accurate solution. The details of this
kind of approach can be found in Castro and Sagaseta (2009).
Multiple instantaneous load steps may be considered. The
column (drain) is considered to be fully permeable, which is
doubtful for conventional stone columns but is reasonable if the
columns are coated with a geotextile. In this way, consolidation
around encased stone columns is studied using any conventional
solution for radial consolidation (e.g. Barron 1948) and a
modified coefficient of consolidation that accounts for the
influence of column and encasement.
2.3

Encasement

The encasement is modelled as a cylindrical shell of negligible


thickness around the column. Therefore, it is valid for different
types of coating, such as geotextiles, geogrids... Encasement
behaviour is supposed to be linear elastic-perfectly plastic and
characterized by a tensile stiffness, Jg, and a maximum tensile
strength, Tg,max. During column installation, the encasement is
pre-stressed to an initial tensile stress, Tg,i. The encasement
tensile stress obtained with the analytical solution is the
increment from that value, Tg.

sr

Tg

rs
rc
rc

J g sr
rc2

rs

(3)

Those simple equations (Eq. 2 and 3) show how the


encasement influence depends on its stiffness and radius.

Soil

rc

(2)

where sr is the radial displacement of the interface.


Combining these two equations, the radial equilibrium
between soil and column at their interface depends on the
encasement properties (stiffness and radius) and its radial
expansion.

Horizontal slice
zc

sr
rc

Tg

Formulation

The detailed formulation of the solution can be found in Castro


and Sagaseta (2009, 2011). Three different possible phases are
identified: (a) soil, column and encasement in the elastic range,
(b) column yielding and (c) encasement yielding, which will
occur after column yielding in a real situation.
A sensible design should cause yielding of the column but
not of the encasement. Therefore, the last phase of the solution
may not be considered and it is just necessary to check that the
tensile stress of the encasement does not exceed its strength.
The solution considers just one instantaneous load step, but it
is quite straightforward to generalize it for multiple loading
steps (Castro and Sagaseta 2008), taking the initial stresses as
the final ones of the previous load step. However, modelling the
real loading steps is only necessary to study the consolidation
process but not for the final values as it gives the same results.
2.5

Drained solution

The studied closed-form solution models the consolidation


process. However, consolidation around stone columns,
especially if the columns are coated with a geotextile, may be
nearly as fast as the loading pace, which means that for these
cases drained condition is a more reasonable assumption.
In any case, depending on the soil permeability and the
loading pace, the real behaviour is between drained condition
and an undrained loading followed by consolidation.
Fortunately, both cases yield very similar final values as can be
shown numerically.
Nonetheless, analytical solutions use
simplifying
assumptions that have different consequences in each situation.
The most evident example is disregarding the elastic strains in
the column once it has reached its active state. This assumption
gives acceptable results for non-encased columns or when the
consolidation process is modelled but not if drained conditions
are considered for encased columns (Castro and Sagaseta 2011).
Hence, in that last case it is necessary to account for those
elastic strains in the column (Pulko et al. 2011).
3

PARAMETRIC STUDY AND NUMERICAL ANALYSES

Figure 2. Equilibrium and compatibility conditions of the encasement.

3.1

The encasement is compressed in vertical direction, and as it


can only take tension, it does not have any influence in vertical
direction. Its equilibrium and compatibility conditions (Figure
2) are those of a thin tube under internal, rc, and external
pressure, rs.

Numerical simulations are included in the parametric study to


evaluate the accuracy of the closed-form solution and the
influence of its simplifying assumptions, such as neglecting the
shear stresses and using an average pore water pressure along
the radius. Coupled numerical analyses of the unit cell were
performed using the finite element code Plaxis v8.6 (Brinkgreve
2007). For comparison purposes, the same boundary conditions
and material properties of the analytical solution were chosen
for the numerical models. Therefore, a rigid plate was set on top

rc

Tg
rc

rs

(1)

2446

Numerical model

Technical Committee 211 / Comit technique 211

of the unit cell, the soil was modelled as elastic and the
encasement and the column as elastic-perfectly plastic.
Stress concentration

The ratio between the vertical stress on the column and on the
soil is usually called the stress concentration factor
(SCF=zc/zc) and gives an idea of the part of the applied load
that the soil transfers to the column. Figure 3 shows its variation
with time. The vertical stresses on the soil and on the column
may vary with the radius, and therefore, their averaged values
are used to calculate the SCF.
A higher encasement stiffness provides a better lateral
confinement to the column, and hence, the column supports a
higher load. A good agreement is found between the analytical
and the numerical results. However, as it happens for the stone
column solution (Castro and Sagaseta 2009), the agreement for
low degrees of consolidation (<30%) is not very good due to
inherent assumptions of Barrons solution.

SCF, zc/zs

c=40

Column
yielding

K0s=0.6

pa/(z')=10

FE

0
0.0001

c=40

0.6

=0.53

0.5
0.4

FE

Closed-form solution
Jg/(rcEs)=0
Jg/(rcEs)=0.75
Jg/(rcEs)=2

0.3
=0.233-0.235

0.5

1.0

1.5

Normalised applied load, pa/(L's)

Figure 4. Settlement reduction. Influence of the applied load.

Closed-form solution

Closed-form solution
Jg/(rcEs)=0
Jg/(rcEs)=2

0.001

0.01

0.1

0.6
Ec/Es=40

c=s=0.3

0.4

c=40

's='c
pa/(L's)=1 K0s=0.6
c=10

0.2

Jg/(rcEs)=5
Jg/(rcEs)=2
Jg/(rcEs)=0.75
Jg/(rcEs)=0

0.8

0.05

Time factor, Tr

0.10

0.15

0.20

0.25

0.30

Area replacement ratio, ar

Figure 3. Stress concentration on the column with time.

Figure 5. Settlement reduction. Influence of the encasement stiffness for


different area replacement ratios.

Settlement reduction

The settlement reduction decreases with the applied load, pa,


from an elastic value, e, and approaches a plastic one, p, at
the same rate as plastic strains develop in the column (Figure 4).
The applied load is normalized by the initial vertical stress
because column yielding depends on that factor, pa/(L's).
On the other hand, the settlement reduction introduced by the
encasement is nearly the same for different area replacement
ratios (Figure 5), which means that column encasement is
equally useful for different area replacement ratios, yet columns
of smaller diameters are better confined. In Figures 4, 5 and 6,
the numerical results validate the accuracy of the analytical
solution, but the agreement gets slightly worse as the tensile
stiffness of the encasement increases. Hence, the only
assumption that has a slightly noticeable effect in the results is
neglecting the elastic strains in the column during its plastic
deformation. A future improvement of the analytical solution
including those elastic strains is currently being developed.

'c='s=10 kN/m

Ec/Es=40
Es=1 MPa

pa=100 kPa = =0.3


c
s
ar=0.11

100
Settlement, sz [mm]

3.3

=0.62

c=10

FE

c=10

=0.70

1.0

c=s=0.3

rc=0.5 m

's='c
c=s=0.3 K =0.6
0s
ar=0.11

Ec/Es=40
Es=1 MPa
ar=0.25

Ec/Es=40

0.7

0.2

Settlement reduction factor,

10

Settlement reduction factor,

3.2

0.8

200

c=40

300

L=10 m
K0s=0.6

c=10

FE
400

500
0.0001

Closed-form solution
Jg/(rcEs)=0
Jg/(rcEs)=0.75
Jg/(rcEs)=2
0.001

0.01

0.1

Time factor, Tr

Figure 6. Time-settlement curve.

3.4

Encasement length

The effectiveness of encasing the columns in reducing the


settlement is directly related to the tensile stress of the
encasement, which provide lateral support to the column. Some
authors (e.g. Khabbazian et al. 2010, Gniel and Bouazza 2009,
Murugesan and Rajagopal 2006) have proposed a partial
encasement of the columns, limiting it to the upper part where

2447

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

the initial lateral stresses are lower. Then, the analysis focuses
on the length of the column that should be encased. Here, a
preliminary study of the encasement length is presented using
the authors' closed-form solution.
The closed-form solution provides the vertical strain of the
column at different depths. Figure 7 shows that those strains are
higher at shallow depths and linearly decrease with depth, as
initial horizontal stresses increase. If the column is encased,
those strains are lower but follow a similar pattern. Therefore,
encasing the columns is more effective in their upper part but
that varies linearly with depth and there is not a critical length
of the encasement that should specifically be used.
0

Ec/Es=40

's='c
c=s=0.3 K =0.6
0s
ar=0.11

Depth factor, z 's / pa

c=40

c=10

Jg/(rcEs)=0
Jg/(rcEs)=2

Elastic column
(high depths)
0

Vertical strain, z (%)

Figure 7. Vertical strain at different depths.

CONCLUSIONS

The main features of a closed-form solution, recently developed


by the authors (Castro and Sagaseta 2011), to study soft soil
improvement, both reduction of settlement and consolidation
time, by means of encased stone columns are presented. The
analytical solution pretends to be a simple and useful tool for
design. Therefore, only a unit cell, i.e. an end-bearing column
and its surrounding soil, is modelled in axial symmetry under a
rigid and constant load.
Parametric studies of the settlement reduction and stress
concentration show the efficiency of encasing the columns,
which is mainly ruled by the encasement stiffness compared to
that of the soil. Therefore, encasing stone columns is
recommended in very soft soils and the encasement should be
stiff enough. Besides, the settlement reduction decreases with
the applied load. Column encasement is equally useful for
common area replacement ratios but columns of smaller
diameters are better confined.
The results of the closed-form solution agree well with
numerical analyses. The only assumption of the solution that
has a slightly noticeable effect in the results is neglecting the
elastic strains in the column during its plastic deformation.
Therefore, including those elastic strains is an improvement of
the presented solution under development.
Finally, a preliminary analysis of the encasement length
shows that is more efficient to encase the columns in the upper
part, as expected, but there is not a critical length of the
encasement that should specifically be used.

ACKNOWLEDGEMENTS

The work presented is part of a research project on "An


integrated calculation procedure for stone columns, considering
the influence of the method of installation", for the Spanish
Ministry of Science and Innovation (Ref.: BIA2009-13602).
6

REFERENCES

Balaam N.P. and Booker J.R. 1981. Analysis of Rigid Rafts supported
by Granular Piles. International Journal for Numerical and
Analytical Methods in Geomechanics 5: 379-403.
Barron R.A. 1948. Consolidation of fine-grained soils by drain wells.
Transactions ASCE 113: 718742.
Brinkgreve R.B.J. 2007. Plaxis finite element code for soil and rock
analysis, 2D, version 8. Rotterdam: Balkema.
Castro J. and Karstunen M. 2010. Numerical simulations of stone
column installation. Canadian Geotechnical Journal 47(10): 11271138.
Castro J. and Sagaseta C. 2008. Influence of stone column deformation
on surrounding soil consolidation. In M. Karstunen and M. Leoni
(ed.), Proc. of the 2nd International Workshop on Geotechnics of
Soft Soils, Glasgow, pp. 333-338. Leiden: Balkema.
Castro J. and Sagaseta C. 2009. Consolidation around stone columns.
Influence of column deformation. International Journal for
Numerical and Analytical Methods in Geomechanics 33: 851-877.
Castro J. and Sagaseta C. 2011. Deformation and consolidation around
encased stone columns. Geotextiles and Geomembranes 29, 268276.
Gniel J. and Bouazza A. 2009. Improvement of soft soils using geogrid
encased stone columns. Geotextiles and Geomembranes 27: 167
175.
Guetif, Z., Bouassida, M. and Debats, J.M. 2007. Improved soft clay
characteristics due to stone column installation. Computers and
Geotechnics 34(2): 104-111.
Kempfert H.-G. 2003. Ground improvement methods with special
emphasis on column-type techniques. In: Proceedings of the
International Workshop on Geotechnics of Soft Soils-Theory and
Practice, SCMEP, Noordwijkerhout, Netherlands, pp. 101-112.
Khabbazian M., Kaliakin V.N. and Meehan C.L. 2010. Numerical study
of the effect of geosynthetic encasement on the behaviour of
granular columns. Geosynthetics International 17: 132143.
Lo S.R., Zhang R. and Mak J. 2010. Geosynthetic-encased stone
columns in soft clay: A numerical study. Geotextiles and
Geomembranes 28: 292302.
Malarvizhi S.N. and Ilamparuthi K. 2007. Comparative study on the
behaviour of encased stone column and conventional stone column.
Soils and Foundations 47: 873885.
Murugesan S. and Rajagopal K. 2006. Geosynthetic-encased stone
columns: Numerical evaluation. Geotextiles and Geomembranes
24: 349358.
Murugesan S. and Rajagopal K. 2010. Studies on the Behavior of Single
and Group of Geosynthetic Encased Stone Columns. Journal of
Geotechnical and Geoenvironmental Engineering 136: 129-139.
Pulko B. and Majes B. 2005. Simple and accurate prediction of
settlements of stone column reinforced soil. In: Proceedings of the
16th International Conference on Soil Mechanics and Geotechnical
Engineering, Osaka, Japan, vol. 3, pp. 1401-1404.
Raithel M. and Kempfert H.G. 2000. Calculation models for dam
foundations with geotextile coated sand columns. In: Proceedings
of the International Conference on Geotechnical & Geological
Engineering, GeoEngg2000, Melbourne.
Sharma S.R., Phanikumar B.R. and Nagendra, G. 2004. Compressive
load response of granular piles reinforced with geogrids. Canadian
Geotechnical Journal 41: 187192.
Smith M. and Filz G. 2007. Axisymmetric numerical modeling of a unit
cell in geosynthetic-reinforced, column-supported embankments.
Geosynthetics International 14: 1322.
Wehr J. 2006. The undrained cohesion of the soil as criterion for the
column installation with a depth vibrator. In: Proceedings of the
International Symposium on vibratory pile driving and deep soil
vibratory compaction, TRANSVIB 2006, Paris.

2448

Consolidation theory for combined vacuum pressure and surcharge loading


Thorie de la consolidation sous laction combine du vide et dun pr-chargement
Chai J.-C.

Saga University, Japan

Carter J. P.

The University of Newcastle, Australia

ABSTRACT: Atheory describing the consolidation of a uniform clayey deposit with and without prefabricated vertical drain (PVD)
improvement under the combination of a vacuum pressure and a surcharge load has been developed and expressed as closed-form
equations. For the case of a soil layer without PVD improvement, both one-way and two-way drainage boundary conditions are
considered. Laboratory consolidation tests using combinations of vacuum pressure and surcharge load were conductedunder
oedometer conditions with vertical or radial drainage. The measured excess pore water pressures are compared with values predicted
by the theory presented in the paper. It has been demonstrated that the theory is valid and can be used for designing preloading
projects that involve the combination of a vacuum pressure and a surcharge load.
RSUM : Une thorie dcrivant la consolidation dun dpt argileux uniforme avec et sans amlioration par drains verticaux
prfabriqus (DVP) sous laction combine du vide et dun pr-chargement a t dveloppe avec un systme ferm dquations. Pour
le cas dune couche de sol sans amlioration par DVP, des conditions aux limites drainantes par un ct et par deux cts sont
considres. Des essais de consolidation au laboratoire sous des conditions oedomtriques ont t raliss sous vide et pr-chargement
avec des drains verticaux ou radiaux. La surpression interstitielle mesure est compare avec les valeurs prvues par la thorie
prsente dans le prsent article. Il a t dmontr que la thorie est valable et peut tre utilise pour dfinir des projets de prchargement qui impliquent lutilisation combine du vide et dun pr-chargement.
KEYWORDS:consolidation, vacuum pressure, embankment, laboratory test, soft clay
1

INTRODUCTION

Preloading a soft clayey deposit with the combination of a


vacuum pressure and a surcharge load (embankment fill) has
several advantages, such as increasing the preloading pressure
and reducing lateral displacements of the deposit, etc. (e.g.,
Chai et al. 2006). Its use in engineering applications has
increased in recent years (e.g., Kelly and Wong 2009; Hirata et
al. 2010; Indraratna et al. 2011).
Vacuum consolidation has different characteristics compared
with consolidation induced by direct application of a surcharge
load (Chai et al. 2009). For a soil deposit without any
improvement in consolidation performance that might result
from the installation of prefabricated vertical drains (PVDs),
when a vacuum pressure is applied water is drained out of the
soil layer only at the boundary where that vacuum pressure is
applied. This applies for both cases of one-way and two-way
drainage conditions. However, for a deposit with one-way
drainage constrained to deform under one-dimensional (1D)
conditions, the final state is a uniform vacuum pressure
distribution throughout the deposit and consequently zero flow
rate. But for a deposit with two-way drainage, at the bottom
boundary the excess pore water pressure is fixed at zero and
effectively no vacuum pressure can be applied at this location,
and so the final state involves the steady flow of pore water
toward the boundary at which the vacuum pressure is applied.
Considering these complicating factors, Chai and Carter (2011)
recently derived a consolidation theory for soils subjected to
vacuum pressure. However, their theory cannot be applied
directly for cases that involve a combination of vacuum
pressure and surcharge loading, and therefore there is a need to
develop a reliable theory for such cases.
This paper presents a newly developed consolidation theory
applicable to soils subjected to a combination of vacuum
pressure and surcharge loading. This theory is applicable to the

2449

case of a uniform soil deposit with or without PVD


improvement. Predictions obtained using this theory are
compared with the results of laboratory tests conducted under
oedometer conditions, for cases that involve both vertical and
radial drainage conditions, with the latter designed to simulate
the consolidation of a deposit improved by PVDs. It has been
shown that the theory is valid and can be used for
designingpreloading projects that involve a combination of
vacuum pressure and surcharge loading.
2

CONSOLIDATION THEORY

2.1Uniform layer without PVDs


Under the same assumptions as those made in Terzaghis 1D
consolidation theory (Terzaghi 1943), the governing equation
and the boundary conditions for the generation and dissipation
of excess pore water pressure in a saturated soil layer under a
combination of vacuum pressure and surcharge load are as
follows:

2u u

z 2 t

(1)

u (0, t ) pvac

(2)

cv

u H , t
0 for t 0 (one-way drainage)
z

u H , t 0 for t 0 (two-way drainage)

(3)
(3a)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

wherez = the spatial coordinate; t = time; u = the excess pore


water pressure; cv = the coefficient of consolidation of the soil;
pvac = the magnitude of the applied vacuum pressure at z = 0;
and H is the thickness of the deposit.
With the presence of a vacuum pressure, the final state is not a
condition with zero excess pore pressure in the deposit.
Therefore, the solution to the governing equation must consist
of two parts, namely the steady state solution (Y(z)) and the
transient solution (v(z,t)) (Chai and Carter 2011). With the
boundary condition defined by Eq. (2), u(z, t) can be expressed
in the following form:

u ( z, t ) pvacY ( z ) pvac v1 ( z, t ) ps v2 ( z, t )

(4)

whereps = the magnitude of the applied surcharge load. The


term -pvacY(z) is the final steady state excess pore water pressure
distribution and (pvac v1(z, t)+ psv2(z, t)) is the time-dependent
component of the excess pore water pressure.
2.1.1One-way drainage
For this case the excess pore water pressure distribution is given
by:

u pvac ( pvac ps )

n 1

sinan z e a c t (5)
2n 1

2
n v

wherean = (2n-1)/(2H). In this case, Y(z) = 1, and the v1(z, t) =


v2(z, t) and its expression is given in the last set of parentheses
of Eq. (5). The average degree of consolidation is given by:

U 1

n 1

c t

2 n 1
1
4H
e
2
2n 1
v

(6)

2.1.2Two-way drainage
In this case the excess pore water pressure distribution in the
soil is given by:

pvac ps

sin n z
z 2

u ( z , t ) pvac 1

e c t
p
H

n 1

s
n sin n ( H z )

2
n v

(7)
where n n H . The average degree of consolidation is
given by:

(t ) 1
U

n 1

c t

2 n 1
1
H
e
2
2n 1
vs
2

(8)

2.2 Uniform layer with PVD improvement


The theory for a PVD-improved soil deposit is derived here for
the case of one-way drainage conditions using a unit cell model,
as shown in Fig. 1. The governing equation for consolidation is
as follows:

u 1 u
u
ch 2
t
r r r
2

(9)

wherer = the radial distance and ch = the coefficient of


consolidation in the horizontal direction. The boundary
conditions are:

2450

Figure 1. Unit cell model and boundary conditions

u re , z , t
0
r

(10)

u r ,0, t
u ' r ,0, t
0,
0
z
z

(11)

u r , l , t
u ' r , l , t
0,
0
z
z

(12)

u ' rw ,0, t pvac

(13)

whereu and u' = the excess pore water pressures in the


undisturbed zone and the smear zone, respectively (Fig. 1), z =
depth from the ground surface, rw = equivalent radius of a PVD,
and re = radius of the unit cell. The solutions for u and u' can be
expressed as:

u (r , z, t ) pvac ( pvac ps )v(r , z, t ) for (rs<rre) (14)


u ' (r, z, t ) pvac ( pvac ps )v' (r , z, t ) for (rw<rrs) (15)
wherers = radius of the smear zone. The additional conditions
for getting explicit expressions for v and v' are the following
water flow continuity conditions.
(1) The total inflow of pore water through the boundary of a
cylinder with a radius of r has to be equal to the change in
volume of the hollow cylinder with outer radius of re and
inner radius of r.
(2) The pore water flow into the PVD from a horizontally cut
soil slice is equal to the change of vertical flow rate in the
PVD.
At the interface between the smear zone and the undisturbed
zone, the radial flow rate from the undisturbed zone is equal to
the flow rate into the smear zone.
With these conditions and using the same assumptions as
those adopted in obtaining Hansbos (1981) solution, it can be
shown that the expressions for v(r, z, t) and v(r, z, t) are as
follows:

2 r r 2 rw2

re ln r 2

kh
w

exp 8Th
v' r , z , t
2

k s re k s 2
2

n 12lz z
k
w

fo

r ( rw r rs )(16)

Technical Committee 211 / Comit technique 211

re2 ln s
2
2
r 2 ln r r rs kh 2 2
e
r r
2
rs
ks s w 8Th
1
for
vr, z, t 2
2 exp

re

kh n2 12lz z 2

kw

( rs r re )(17)
where: n = re/rw, s = rs/rw,kh and ks = the hydraulic
conductivities in the horizontal direction of the undisturbed
zone and the smear zone respectively, kw = the hydraulic
conductivity of the drain (PVD), l = the drainage length of a
PVD, and Th = cvt/(4re2). Parameter represents the effects of
PVD spacing, smear zone and well resistance. Adopting an
average well resistance and with some approximation, the
expression for is as follows (Hansbo 1981):

ln(n / s ) (k h / k s ) ln(s )

3 2l 2 k h

4 3rw2 k w

(a) Vertical drainage test

(18)

The average degree of consolidation (Uh) of the unit cell is


(Hansbo 1981):

U h 1 exp 8Th /

(19)

3 COMPARISON OF TEST RESULTS AND PREDICTIONS


Laboratory consolidation tests involving the combination of a
vacuum pressure and a surcharge load have been conducted
under oedometer conditions with both verticaland radial
drainage (the latter to simulate the effects of PVD drainage),
and the measured excess pore pressures have been compared
with the predicted values.

(b) Radial drainage test


Figure 2. Sketch of the set-up of the tests

3.1Test details
Figures 2(a) and (b) show the set-up of the tests, with vertical
(V-test) and radial (R-test) drainage conditions,respectively.
During testing, the settlement, the excess pore water pressure at
the bottom of the sample (V-test) or the middle height of the
consolidation ring (R-test), and the horizontal earth pressure at
the middle height of the consolidation ring can be measured.
For the R-test, the centre drainage porous stone tube has an
outer diameter of 8 mm, which is inserted into a predrilled hole
at the center of a sample with a filter paper placed between the
soil sample and the tube. The soil samples were re-consolidated
from Ariakeclay slurries under a surcharge pressure of 20 kPa.
Two series of tests, V-tests and R-tests, were conducted.
Here only one test from each serieshas been chosen to compare
with the values predicted by the theory presented above. In the
case of the V-test, the test with one-way drainage conditions has
been selected, because for two-way drainage conditions no pore
water pressures were measured with the device used. The two
series of tests were conducted at different times and different
soil samples were used. Some of available soil properties as
well as the test conditions are listed in Table 1. In this table, the
vertical effective stress, 'v0, indicates that the soil sample was
first consolidated under 'v0 (simulating the initial effective
stress of the soil sample at a specified depth in the deposit) and
then the consolidation test was conducted by applying
additional incremental consolidation pressures (vacuum
pressure and surcharge load).

2451

Figure 3. Comparison of predicted and measured ub values

3.2Comparison of measured and predicted pore pressures


3.2.1V-test
After initial consolidation under 'v0 = 40 kPa, the thickness of
the sample was 18.7 mm (or compression of about 1.2 mm).
Further, under 80 kPa vacuum pressure and 80 kPa surcharge
load, the additional compression was about 3.2 mm. Since the
thickness of the sample is also the vertical drainage path length,
in the predictions an average sample thickness of 17.2 mm was
adopted. Comparison of the measured and the predicted excess
pore water pressures at the bottom of the sample (ub) is shown

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Table 1. Some soil properties and test conditions


Test

Soil

Plasticity
limit,
Wp
(%)

Liquid
limit,
WL(%)

Coefficient
of
consolidation
cv or ch (m2/min)

Cc

e 0

'v0
(kPa)

pvac
(kPa)

ps
(kPa)

V-test
Ariake clay-1
60.3
120.5
2.310-5#
0.75
2.5
40*
80
80
R-test
Ariake clay-2
56.8
120.3
5.010-6
0
80
80
*: Initial vertical effective stress in the sample; #: The value was obtained by fitting the measured consolidation rate; : After pre-consolidation under
20 kPa pressure.

in Fig. 3. Except for the fact that the measured initial value of
ubof about 72 kPa is slightly lower than the 80 kPa applied
surcharge load, the prediction almost matches the measured data.
The slightly lower initial ub value may indicate that the
specimen was not 100% saturated.
Although two-way drainage test was not conducted, using
the same soil parameters as for one-way drainage test, and
assuming the thickness of the soil sample is 20 mm, the
predicted excess pore water pressure (u) distribution within the
sample at different elapsed times are given in Fig. 4 to
demonstrate the capacity of the proposed theory.

Depth, D (mm)

0
pvac = 80 kPa
ps = 80 kPa

water pressure at the periphery of the sample starts to reduce.


Comparison of the excess pore water pressures at the periphery
of the sample (ure) is given in Fig. 5. For this case, during the
consolidation period the measured excess pore water pressure
initially decreased but then increased for a brief period before
finally exhibiting further dissipation.
Furthermore, the
measured final excess pore water pressure did not reach the
applied vacuum pressure of 80 kPa. Nevertheless, the trends of
both the measured and the predicted dissipation curves are
similar.
From the above comparisons, it can be seen that the theory
provides reasonable predictions of the measured soil behaviour
and so it should be able to be used reliably for designing
preloading projects that adopt a combination of vacuum
pressure and surcharge load to consolidate the soil deposit.
4 CONCLUSIONS

10
15

20
-80

1 min

-40
0
40
Excess pore pressure, u (kPa)

80

Figure 4. Predicted u variation in soil sample under two-way drainage


boundary condition

A consolidation theory, expressed in closed-form equations, for


soil consolidation under the combination of a vacuum pressure
and a surcharge load has been developed for a uniform clayey
deposit with and without prefabricated vertical drain (PVD)
improvement. For cases without PVD improvement, both oneway and two-way drainage boundary conditions have been
considered.
Laboratory consolidation tests were conducted, adopting a
combination of vacuum pressure and surcharge loading under
oedometer conditions with both vertical and radial drainage.
The excess pore water pressures measured in these test were
compared with values predicted by the suggested theory. It has
been demonstrated that the theory is valid and can be used for
designing preloading projects that adopt a combination of
vacuum pressure and surcharge load to pre-consolidate soft soil
deposits.
REFERENCES

Figure5.Comparison of predicted and measured ure

3.2.2 R-test
The geometric parameters required to calculate the predictions
for this case are: re = 30 mm; rw = 4 mm; and l = 20 mm. The
assumed radius of the smear zone, rs = 7 mm; the hydraulic
conductivity ratio, kh/ks = 5; kh = 10-9 m/s; and kw = 10-4 m/s. In
the case of radial drainage, with Eqs. (16) and (17) the initial
condition of a uniform excess pore water pressure (u0)
distribution in a unit cell is not satisfied (which is a particular
limitation of this theory). These equations only ensure that the
average initial value of u0 is the same as the applied value. The
predicted initial value at the periphery of the sample (unit cell)
is higher than the applied value. The predicted values are
compared with the measured data from the time at which the
predicted value at the periphery was equal to the applied initial
value. In the physical test at the corresponding time, the pore

2452

Chai, J. C., Carter, J. P. and Hayashi, S. 2006. Vacuum consolidation


and its combination with embankment loading.Canadian
Geotechnical Journal 43(10), 985-996.
Chai, J.-C., Matsunaga, K., Sakai, A. and Hayashi, S. 2009. Comparison
of vacuum consolidation with surcharge load induced consolidation
of a two-layer system. Gotechnique 59(7), 637-642.
Chai, J.-C. and Carter, J. P. 2011. Deformation analysis in soft ground
improvement. Springer, p. 247.
Hansbo S. 1981. Consolidation of fine-grained soils by prefabricated
drains. Proceedings of 10th International Conference on Soil
Mechanics and Foundation Engineering, Stockholm 3, 677-682.
Hirata, M., Kitoh, M., Yamada, K., Iizuka, A. and Arai, K.
2010.Deformation behavior and counter measures of expressway
embankment on super-soft ground.Journal of Japan Society of Civil
Engineers 66(2), 356-369 (in Japanese).
Indraratna, B., Rujikiatkamjorn, C., Ameratunga, J. and Boyle, P. 2011.
Performance and Prediction of Vacuum Combined Surcharge
Consolidation at Port of Brisbane.Journal of Geotechnical and
Geoenvironmental Engineering ASCE137(5), 550-554.
Kelly, R. B., Wong, P. K. 2009. An embankment constructed using
vacuum consolidation. Australian Geomechanics 44(2), 55-64.
Terzaghi K. 1943. Theoretical soil mechanics. New York, John Wiley
and Sons.

Displacement rigid inclusions


Inclusions rigides refoules
Cirin A., Pauln J.

Soletanche-Bachy-CIMESA, Mexico

Racinais J.

Menard, France

Glandy M.

Soletanche-Bachy-Pieux, France

ABSTRACT: In soils with poor mechanical properties and in areas where the generation of excavation debris is an issue, given the
restrictions regarding its disposal, the solutions of massive soil improvement with displacement rigid inclusions solve both needs. In
this paper we describe the basis of the constructive procedure of displacement rigid inclusions. We explain the concept of
improvement with this kind of inclusions; we itemize the bases of their design, and describe their construction sequence, highlighting
the controls during the execution to ensure quality.
RSUM: Dans les sols ayant des proprits mcaniques faibles comme dans les zones o llimination des matriaux produits des
travaux reprsente un problme, les Inclusions Rigides avec refoulement de sol donnent des solutions ces deux situations. Larticle
explique le concept des solutions damlioration des sols en utilisant la technique des Inclusions Rigides, donne les bases du
dimensionnement, et dcrit la squence de construction des inclusions Rigides en insistant sur les contrles utiliss pour assurer la
qualit finale.

KEYWORDS: soft soil, rigid inclusion, displacement of soil, excavation debris.

INTRODUCTION

When studying what type of foundation is best suited to


withstand the shock that a new building (structure) will impose
on the soil, it is necessary to check not only the limit conditions
for failure, but also the limit conditions of service, including
total and differential settlements.
Being successful in the choice and design of the type of
foundation to be built largely depends on the control of two
variables: load and settlement. Nevertheless, there are additional
parameters that also play an important role in the decision
process, such as the cost of the foundation with respect to the
total cost of the project, construction time and increasingly
the impact on the environment.
The foundations based on rigid inclusions (system structure
massive soils improvement) have experienced a boom in recent
years, especially in works on large areas subjected to uniform
vertical loads. While this is not a new concept (wooden
inclusions were used since prehispanic times in Mexico see
Auvinet, G., 2006), there is now specialized equipment
capable of building concrete rigid inclusions following special
procedures that not only achieve higher production results, but
also greater depths and better loadbearing capacities. They also
respect strict quality controls. This gives us the possibility to
propose foundations based on the installation of grids of rigid
inclusions made of poor concrete that meet specific technical
requirements regarding load bearing capacities and the
reduction of settlements. They are also attractive: economically
and for their constructive feasibility, as well as for their reduced
construction times and the quality of their execution.
Displacement inclusions in particular have the great
advantage of not generating construction debris, which benefits
the environment and reduces or eliminates the cost of its
removal. In soils with a large frictional component, the ratio of
voids surrounding the inclusion is reduced by the incorporation
of the concrete so that the relative compactness of the material

increases, as well as the perimeter friction of the inclusionground. The construction process of the displacement rigid
inclusion guarantees quality control in the execution, so the
concrete is placed continually and safe from contamination.
2

BASES OF DESIGN

The goal is to install a set of inclusions in soils with low bearing


capacity and/or highly compressibility to create a layer of
compound soil-inclusions material that has better mechanical
properties.
The improvement or reinforcement of soils with rigid
inclusions is commonly used to ensure bearing capacity and/or
reduce settlements in the following types of work:

Slabs,
Superficial footings (isolated or continuous),
Embankments, landfills,
Work or storage yards.

The solution is characterized by the fact that the traditional


mechanical link between the pile and the structure in deep or
mixed foundations does not exist. A distribution layer, also
called a Load Transfer Platform (LTP), is usually placed
between the inclusions and the structure to be supported, and
this is what marks the difference between piles and inclusions.
The distribution layer spreads the acting loads on the slab or
other covering surface towards the underlying soil-inclusions
setup. The system described is configured as shown in Figure 1.
If there are concentrated vertical loads from one column,
isolated or continuous footings can be used to directly transmit
the loads to the soil-inclusions setup.

2453

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

3
Transfer layer
Compressible soils
Rigid inclusions

Hard layer

Rigid inclusions
d

Trasfer layer

Figure 1. Inclusion under a load uniformly distributed on the surface.


Side and top views.

CONSTRUCTIVE SEQUENCE

The equipment used for the construction of displacement rigid


inclusions kind must circulate over a flat working platform,
drained and stable, generally constructed of granular material.
The inclusions are built from this platform.
The drilling equipment consists of a crane supported on
caterpillars with a cab for the operator and a mast that supports
a cylindrical auger of a defined length. The auger is hollow and
has a special geometry see Figures 4a, 4b, capable of
displacing soil laterally when drilling. This is the most
important feature of displacement rigid inclusions because the
surrounding soil becomes laterally compressed. Lateral friction
increases in the case of mainly granular soils or soils with a
large content of sand.
At the bottom part of the tip there is a hinged lid that remains
closed during the drilling phase to prevent the entry of material
into the inner tube and which opens to allow the exit of the
concrete to form the inclusions.
Besides the necessary drilling equipment there has to be a
concrete pump which feeds the upper side of the drilling tool
through flexible hoses.

In this case the LTP may not be required and a significant


portion of the load from the superstructure will be supported by
the grid of inclusions and the remainder will be supported by
the soil surrounding the inclusions see Figure 2.
q

Compressible soils
Rigid inclusions

Hard layer

Figure 2. Model setup of inclusions under an acting strut load on a


footing. Side view.

Figure 4a. Diagram of the typical point of the hollow auger for
displacement rigid inclusions.

In the same way that the inclusion-soil system supports


vertical loads uniformly distributed or concentrated from
buildings, this application can be extended to the case of
embankments and landfills in which the system will receive the
weight of the material that forms the embankments or landfills.
A particular case occurs when the embankment or landfill is
significantly high and the soil reinforced with inclusions
participates in its stabilization see Figure 3.
Embankment

Failure surface

Figure 4b. Point of the hollow auger for displacement rigid inclusions
developed for Soletanche-Bachy, RefSol system.

Compressible
soils
Hard layer
Rigid inclusions

Figure 3. Inclusions that help stabilize an embankment or landfill


constructed on the surface.

The inclusions will generally be subjected to the action of


vertical forces caused by discharges from the building or due to
the weight of the embankment or landfill. However, in cases
where the inclusions participate in the stabilization of
embankments or landfills, or when they are subjected to the
action of seismic forces, the generation of lateral forces will
also have to be taken into account in the design.
Several approaches and ways of analyzing and designing
inclusions have been developed. Some of them have been
recently brought together in the ASIRI (Amlioration des Sols
par Inclusions Rigidessee ASIRI National Project, 2012).

With the topographic location of the inclusion to be built, the


process begins by placing the mast of the crane upright and
lowering the auger into the ground. A rotor torque and a
descending vertical force are applied to the auger to cut,
penetrate and displace the soil laterally. This action is
performed continuously until the drill reaches the specified
depth see Figure 5A.
At this point, the concrete is pumped from the tank of the
pump through a flexible hose to the upper part of the hollow
auger to fill it completely and to generate sufficient pressure on
the concrete.

2454

Technical Committee 211 / Comit technique 211

(A)

(B)

(C)

Figure 5. Execution sequence of a displacement rigid inclusion.

Then the auger is lifted a few centimeters from the soil at the
bottom of the perforation, which causes the lid at the lower end
of the auger to open. The concrete, subject to pressure, pours
into the bottom of the hole, filling it. While still pouring
concrete and controlling the pressure, at this point the operator
lifts the auger continuously by means of a rotor torque and a
vertical pulling force see Figure 5B. This process continues
until the auger is fully above ground see Figure 5C. The
concrete is poured continuously from the bottom of the
perforation until it reaches the level defined as the head of
inclusion, which can be between the working platform level and
a few dozen centimeters below it.
Throughout the process of building an inclusion (Figures 5A,
5B and 5C) real time and continuous monitoring of the
parameters that intervene in its execution are done with
electronic devices located in the cab of the crane. They detect
the signals sent by various sensors installed at strategic points of
the construction equipment. Through this monitoring, the
operator has control of the different construction parameters and
can ensure the quality of the construction of the inclusion at all
times and along its entire height. Among the parameters
controlled are: the drilling depth, the pressure and the volume of
the concrete, the upward and downward speeds, rotation and the
auger's torque.
The equipment is also able to store the record of the controls
for each inclusion, to be processed later on a personal computer.
Continuous records are obtained along the depth(see Figure 6).

The procedure described is a clean process that leaves


practically no perforation debris on the work platform. There
are also no vibrations or damage to the surface layers, which
makes working in areas adjacent to sensitive structures possible.
Additionally, the method is capable of achieving high industrial
production compared with traditional methods of pile
construction.
For quality control, it is also necessary to carry out strength
tests on samples of the concrete used. There will be as many
tests as are needed or as required by local regulations. The
common values of resistance to compressive strength of the
concrete used for the construction of displacement rigid
inclusion range from 10 to 15 MPa at 28 days, with modules of
elasticity usually set between 5,000 and 10,000 MPa, although
higher resistance and rigidity levels can be used according to the
needs of each project.
The commercial diameters of displacement rigid inclusion
construction range between 250 and 500 mm and can reach
depths of up to 30 m.
To guarantee the quality of the implementation and the
design criteria, this construction procedure has been certified by
the international bureau of control and certification Bureau
Veritas.
4

CONCLUSIONS

Soil improvement and reinforcement with displacement rigid


inclusions kind solves a great number of foundations in which
not only increasing bearing capacity, reducing settlements or
ensuring slope stability play an important role, but where also
cost and execution times are factors to be considered.
Given the type of auger used in the construction these
inclusions are defined as displacement inclusions where the
surrounding soil is displaced and laterally compressed at the
moment of drilling, which increases the compactness of soils
whose frictional component is significant.
During construction of displacement rigid inclusion there is
real-time monitoring of parameters such as drilling depth,
pressure and volume of the poured concrete, advancement speed
and auger rotation, downward force of the rotor torque of the
auger, which ensures a high quality control of the construction.
Due to the advantages provided by the design of soil
improvement systems with rigid inclusions, plus the
geotechnical and environmental benefits of displacement rigid
inclusions, numerous projects worldwide are being approached
with this technique.
5

REFERENCES

Auvinet, G. (2006). Rigid inclusions in Mexico City soft soils: history


and perspectives, International Symposium Rigid inclusions in
difficult soft soil conditions, Instituto de Ingeniera, UNAM, Cd.
de Mxico.
Combarieu, O. (1988). Amlioration des sols par inclusions rigides
verticales Application ldification des remblais sur sols
mdiocres. Revue Franaise de Gotechnique N 44, 5779.
ASIRI National Project (2012). Recommendations for the design,
construction and control of rigid inclusion ground improvements.
Bureau Veritas. Cahiers des charges CMC.
Bureau Veritas. Cahiers des charges Refsol.
Figure 6. Record of monitoring in continuous real time: (A) Inclusion
profile (mm), (B) Perforation energy (bar), (C) Perforation speed (m/h),
(D) Rotation torque (t.m), (E) Rotation speed / Bearing force.

The start and stop of the concrete pump is wirelessly


controlled by the crane operator from the cab. The speed at
which the auger advances, the rotor torque, the rotation speed
and down force or extracting force of the auger is controlled
manually through the hydraulic system of the crane.

2455

th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Prediction of the unconfined compressive strength in soft soil chemically


stabilized
Prediction of the unconfined compressive strength in soft soil chemically stabilized
Prvision
Prvision de
de la
la rsistance
rsistance
la
la compression
compression non
non confine
confine dans
dans sols
sols mous
mous chimiquement
chimiquement stabilises
stabiliss
A.A.S. Correia; P.J. Venda Oliveira & L.J.L. Lemos
Correia
A.A.S.;
Venda
Oliveira
P.J., Lemos
L.J.L. Portugal
Department
of Civil
Engineering
University
of Coimbra,
Department
of Civil Engineering
University
of Coimbra, Portugal
aalberto@dec.uc.pt;
pjvo@dec.uc.pt
& llemos@dec.uc.pt

ABSTRACT: The chemical stabilization of soils is a ground improvement technique consisting on the mechanical mixing of the in
situ natural soil with binders. The chemical stabilization of soils can be applied with either slurries (wet method) or powder (dry
method) binders. When the stabilizing binders are mixed with the soil, physico-chemical interactions take place and are responsible
for the stabilization effect, which has a major influence on the mechanical behaviour of the improved material. This stabilizing effect
is dependent on a range of parameters which should be analysed through a long and extensive laboratory and field trial test program,
as stated in the european standard (EN 14679:2005). In order to minimize the number of tests during the optimization process, this
paper presents a simple method to predict the unconfined compressive strength, which is independent of the binder content and state
(powder or slurry). The method is successfully applied to a wide range of soils, showing its versatility (Correia, 2011). Applying the
generalised relationship of the method, it is possible to predict the unconfined compressive strength for any binder content and state
from one single unconfined compression test.
RSUM : La stabilisation chimique de sols est une technique de l'amlioration des sols qui consiste en le mlange mcanique dans
situ du sol naturel avec liants. La stabilisation chimique de sols peut tre applique avec coulis (mthode mouille) ou poudre
(mthode sec) liants. Quand les liants stabilisateurs sont mlangs avec le sol, ils produisent interactions physique-chimique lesquels
sont responsables pour l'effet de la stabilisation, qui a une influence majeure sur le comportement mcanique du matriel amliore.
Cet effet stabilisateur est dpendant dune gamme de paramtres qui devraient tre analyss travers dun long et tendu programme
dessais en laboratoire et sur terrain, comme nonc dans la norme europenne (EN 14679:2005). Pour minimiser le nombre dessais
pendant le processus de l'optimisation, cet article prsent une mthode simple de prdire la rsistance la compression simple, qui est
indpendant du contenu de liant et tat (poudre ou coulis). Le mthode est applique avec succs une grande gamme de sols,
montrant sa versatilit (Correia, 2011). Appliquant la version gnralise du mthode, c'est possible prdire la rsistance la
compression simple pour tout contenu de liant et tat bas d'une seule essais la compression simple.
KEYWORDS: chemical stabilization, unconfined compression test, soft soils, strength prediction.
1

INTRODUCTION.

Over the last few decades, infrastructure requirements and land


occupation policies have demanded construction on soils with
poor geotechnical properties (in particular, soft soils). These
soils are usually characterized by low strength and high
compressibility, demanding from geotechnical engineers new
and challenging solutions to overcome these undesirable
engineering characteristics. One of the ground improvement
techniques that have been used with success in practice is the
chemical stabilization, where the natural soil is mechanically
mixed in situ with binders (usually called Mass Stabilization, or,
Deep Mixing when applied in depth). This technique has given
good results when applied to soft soils, becoming a prominent
subject nowadays, rapidly growing and wide spreading around
the world due to its technical and economical benefits when
compared with other ground improvement techniques.
At first the chemical stabilization of soils used the quicklime
as the hardening agent. Later on, the use of Portland cement has
permanently been outpacing the use of quicklime, not only
because Portland cement is readily available at reasonable cost
but also because cement is more effective than quicklime
(Horpibulsuk et al 2011, hnberg 2006, Lorenzo and Bergado
2004, Kitazume and Terashi 2002). However, additives such as
granulated blast furnace slag, fly ash, gypsum and silica dust,
among others, may be used specially for the improvement of
soft soils with high water content or organic soils (Kitazume
and Terashi 2002, Edil and Staab 2005).

The chemical stabilization of soils can be applied with either


slurries (wet method) or powder (dry method) binders. When
the stabilizing binders are mixed with the soil, physico-chemical
interactions take place and are responsible for the stabilization
effect, which has a major influence on the mechanical behaviour
of the improved material. This stabilizing effect is dependent on
a range of parameters which should be analysed through a
laboratory and field trial test program, as stated in the european
standard (EN 14679:2005).
The fundamental mechanical properties of cement based
admixed soft soils have been experimentally investigated by
many researchers (Correia 2011, hnberg 2006, HernandezMartinez 2006, Lorenzo and Bergado 2006 and 2004,
Horpibulsuk et al 2004, Kamruzzaman 2002, Horpibulsuk 2001,
Miura et al 2001, Uddin et al 1997, Locat et al 1996). Most of
these previous investigations mainly focus on the influence of
the water content and binder content, as well as on the ratio
between them. Based on some of these parameters, Horpibulsuk
et al (2003 and 2011) and Lorenzo and Bergado (2006) have
introduced phenomenological models for predicting laboratory
strength development in cement based stabilized soft soils. This
paper presents a new simple model which aims to predict the
laboratory strength (expressed by the unconfined compression
test) for various combinations of water content and cement
content. This model intents to minimize the number of
laboratory tests needed to specify the quantity of cement and
water to be admixed with the soft soil. Although the model is

2457

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

developed for a particular soft soil, its versatility is


demonstrated for a wide range of soils. A generalized strength
equation is presented, which allows the strength prediction
based, at lower limit, on a single unconfined compression test.
2
2.1

EXPERIMENTAL TESTS
Materials

Table 1 presents the geotechnical and chemical properties of the


soft soil deposit of Baixo Mondego (located near Coimbra
city, Portugal), used in the study. In general, the soil is
predominantly clayey-silt with a high organic matter content,
which has a strong influence on some characteristics of the soil,
namely, low unit weight, high plasticity, high natural water
content, high void ratio, low undrained shear strength and high
compressibility although this fact is not consistent with the
grain size distribution, particularly due to the low clay content,
(Coelho 2000, Venda Oliveira et al. 2010).
Table 1. Principal properties of the soft soil of Baixo Mondego.
80
Natural water content, wnat (%)
Unit weight, sat (kN/m3)
14.6
Natural void ratio, enat (-)
2.1
Clay fraction (%)
8-12
Silt fraction (%)
71
Sand fraction (%)
17-21
Density, G (-)
2.55
Organic matter content, OM (%)
9.3
Liquid limit, wL (%)
71
Plastic limit, wP (%)
43
Undrained shear strength, cu (kPa)
< 25
CaO (%)
0.74
SiO2 (%)
62
Al2O3 (%)
16
Fe2O3 (%)
4.8
MgO (%)
1.1
pH (-)
3.5

the laboratory procedure presented in EuroSoilStab (2001) with


the modifications proposed by Correia (2011). During the
curing time, fixed as 28 days, all samples were subjected to a
vertical pressure of 24 kPa and stayed submerged in a water
tank at a controlled temperature (202C). After this period, the
samples were submitted to the unconfined compression test in
order to evaluate its strength (qu max).
Tabel 3 and Figure 1 summarizes the main results of the
chemical stabilization of the soft soil of Baixo Mondego. The
results show that, as expected, the unconfined compressive
strength increases with the binder content and with the
decreasing of the water content (or liquidity index). As the
binder content increases, more binder is admixed with the soil
allowing the construction of a stronger skeleton matrix. As the
water content increases the void ratio also increases, promoting
the particles spacing with obvious reflects on the fabric of the
stabilized soil and on its strength.
Table 3. Unconfined compressive strength results of the chemical
stabilization of the soft soil of Baixo Mondego.
IL
aw
qu max
(-)
(%)
(kPa)
1.35
9
209
12
644
15
1143
18
1618
21
1831
24
1936
27
1995
1.96
9
118
15
694
21
1266
27
1383
2.49
9
90
15
552
21
965
27
1032
2100

Table 2. Composition and specific surface of the binders.


CEM I 42.5R
CaO (%)
63.02
SiO2 (%)
19.70
Al2O3 (%)
5.23
Fe2O3 (%)
2.99
MgO (%)
2.38
Specific surface, S (m2/kg)
321.5

2.2

SLAG
37.02
38.74
11.59
0.85
6.75
363.0

qu max (kPa)

The binders used in the present study to produce stabilized


Baixo Mondego soft soil samples were a Type I Portland
cement, designated CEM I 42.5 R (EN 197-1 2000), and a blast
furnace granulated slag, here simply designated as SLAG.
These two binders, on a dry weight proportion of 75/25 as
proposed by Correia (2011), were thoroughly mixed to obtain a
uniform binder. The binder added to the soil was defined by the
parameter binder content, aw (ratio of the dry weight of binder
used in the mixture to the dry weight of the soil). The
composition and the specific surface of the binders are
presented in Table 2.

1800

IL = 1.35

1500

IL = 1.96

1200
IL = 2.49

900
600
300
0

10

15

aw (%)

20

25

30

Figure 1. Unconfined compressive strength results of the chemical


stabilization of the soft soil of Baixo Mondego.

Chemical stabilization of the soft soil of Baixo


Mondego

In order to evaluate the influence of the water content and


binder content on the chemical stabilization of the soft soil of
Baixo Mondego, several samples were prepare for various
water contents (equivalent to a liquidity index IL of 1.35, 1.96
and 2.49) and binder contents (from 9 to 27, step 3). The
laboratorial procedure to produce stabilized samples followed

From Figure 1 it can be seen that the curves for different


liquidity index exhibit a similar shape (are homothetic). Thus
the unconfined compressive strength (qu max) can be normalised
by the liquidity index (IL) multiplying both parameters (qIL =
qu max IL). Figure 2 presents these results which are well fitted
by a linear logarithmical regression. This is a simple way to
predict the unconfined compressive strength at 28 days of
curing time for the cement based stabilized softy soil in study.
As it is a linear regression it only requires two test data made
for different binder contents.

2458

Technical Committee 211 / Comit technique 211


th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
8000

Ariake clay
7000 Tokyo clay
Chiba clay

3000

6000

2500

5000

2000

qIL (kPa)

qIL = qu max x IL (kPa)

In order to validate this simple method, it will be applied to


other soft soils as presented in the next section.

qIL = 2337.63.ln(aw) - 4854.76


2

1500

(R = 0.97)

4000
3000
2000
1000

1000

500
0

qIL = 3855.12.ln(aw) - 83 27.90 ( R = 0.91)


2
qIL = 4649.59.ln(aw) - 9450.40 (R = 0.95)
2
qIL = 34 41.90.ln(aw) - 5932.30 (R = 0.98)

10

15

aw (%)

20

25

30

8000

Kangawa clay
7000 Hiroshima clay
Aichi clay
Osaka clay
6000

3 DATA FROM OTHER CEMENT BASED STABILIZED


SOFT SOILS
Table 4 presents the main results of 7 other cement based
stabilized soft soils whose geotechnical properties are described
in Horpibulsuk (2001) and Kawasaki et al (1981). Figure 3
presents the results of the unconfined compressive strength
normalized by the liquidity index, from which it can be
concluded that each cement based stabilized soft soil has its
normalization (fitting curve). Thus the method here proposed is
versatile as it is valid for other soft soils.
Table 4. Unconfined compressive strength results of cement based
chemical stabilization of other 7 soft soils (Horpibulsuk 2001, Kawasaki
et al 1981).
aw
qu max
Soft soil
IL
(-)
(%)
(kPa)
Ariake clay
1.0
10
833
15
1798
1.5
10
434
15
1286
20
2343
2.0
15
839
20
1736
Tokyo clay
1.0
10
1085
20
4941
30
6072
Chiba clay
1.0
10
2063
20
4189
30
5894
Kangawa clay
1.0
10
1068
20
3120
30
5047
Aichi clay
1.0
10
887
20
1889
30
2159
Osaka clay
1.0
10
595
20
1707
30
1976
Hiroshima clay
1.0
10
748
20
2436
30
3952

2459

qIL (kPa)

Figure 2. Normalized unconfined compressive strength results of the


chemical stabilization of the soft soil of Baixo Mondego.

10

15

aw (%)

20

25

30 35

qIL = 3550.61.ln(aw) -7217.88 (R = 0.96)


2
q IL = 2864.59.ln(aw) -5928.20 (R = 0.97)
2
qIL = 1188.81.ln(aw) -1802.35 (R = 0.94)
2
qIL = 1294.43.ln(aw) -2327.64 (R = 0.93)

5000
4000
3000
2000
1000
0

10

15

aw (%)

20

25

30 35

Figure 3. Normalized unconfined compressive strength for other cement


based stabilized soft soils.

GENERALIZING THE PROPOSED METHOD

As it was observed in Figures 2 and 3, the method proposed can


be applied satisfactory to a wide range of soft soils. However,
each cement based stabilized soft soil has its own fitting
parameters (see the equations for qIL in Figures 2 and 3),
different for each soil.
In order to find a generalized strength equation, independent
of the soft soil, the qIL data of a particular soft soil was
normalized by the unconfined compressive strength defined for
a liquidity index of 1.0 and for a constant binder content (it was
considered the value 18% for all soft soils), qIL=1 (aw=18%). For
each soft soil, this last value was evaluated from the fitting
curves presented in Figures 2 and 3. All data are presented in
Figure 4, where it can be seen that the values are in a narrow
linear band, fitted relatively well by a linear logarithmical
regression (R2 = 0.94). Thus, the method proposed in this paper
seems to be independent of the soft soil type, being valid for the
prediction of the unconfined compressive strength at 28 days of
curing time of cement based stabilized soft soils, which is
helpful for the laboratory optimization process of the chemical
stabilization. The number of unconfined compression tests
required can be reduced to one if it is applied the generalized
equation and the binder content chosen is 18%.

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
3,0

qIL / [qIL=1 (aW = 18%)]

2,5

Baixo Mondego clayey-silt


Ariake clay
Tokyo clay
Chiba clay

Kangawa clay
Aichi clay
Osaka clay
Hiroshima clay

2,0
1,5

qIL/[qIL=1(aw=18%)] = -2.26 + 1.12.ln(aw)


2

(R = 0.94)
1,0
0,5
0,0

10

aw (%)

20

30

40

Figure 4. Generalized strength equation for the unconfined compressive


strength of cement based stabilized soft soils.

CONCLUSION

The paper presents a new simple method to predict the


unconfined compressive strength at 28 days of curing time of
cement based stabilized soft soils, whatever be the water content
and binder content. The method was initially developed for the
soft soil of Baixo Mondego chemically stabilized, and then
was sucessfully applied to a wide range of cement based
stabilized soft soils. Thus, this new method seems to be
independent of the soft soil, which allows the definition of a
generalize relationship (presented in Figure 4). At limit, the
number of unconfined compression tests required can be
reduced to one if it is applied the generalized equation and the
binder content chosen is 18%. The method proposed in this
paper is helpful for the laboratory optimization process of the
chemical stabilization at the pre-design stage.

ACKNOWLEDGEMENTS

The authors would like to express their thanks to CIMPOR for


supplying the binders used in the work and to the institutions
that supported the research financially: University of Coimbra,
CIEC and FCT (PTDC/ECM/101875/2008).
7

REFERENCES

hnberg H. 2006. Strength of stabilised soils a laboratory study on


clays and organic soils stabilised with different types of binder.
PhD Thesis, University of Lund, Sweden.
Coelho P.A.L.F. 2000. Geotechnical characterization of soft soils. Study
of the experimental site of Quinta do Foja, MSc Dissertation,
University of Coimbra (in portuguese).
Correia A.A.S. 2011. Applicability of deep mixing technique to the soft
soils of Baixo Mondego, Ph.D. Dissertation, University of Coimbra,
Portugal (in Portuguese).
Edil T.B. and Staab D.A. 2005. Practitioners guide for deep-mixed
stabilization of organic soils and peat. Final Report. The National
Deep Mixing Research Program, Project Number NDM302, p. 60.
EN 197-1 2000. Cement - Part 1: Composition, specifications and
conformity criteria for common cements. IPQ, Portuguese edition
from April of 2001, 35 p.
EN 14679 2005. Execution of special geotechnical works deep mixing.
CEN, English version, April of 2005, p. 52.
Eurosoilstab 2001. Development of design and construction methods to
stabilise soft organic soils. Design guide soft soil stabilization.
CT97-0351, EC Project No. BE 96-3177, Industrial & Materials

2460

Technologies
Programme
(BriteEuRam
III),
European
Commission, p. 94.
Hernandez-Martinez F.G. 2006. Ground improvement of organic soils
using wet deep soil mixing. PhD Thesis, University of Cambridge,
United Kingdom.
Horpibulsuk S. 2001. Analysis and assessment of engineering behavior
of cement stabilized clays. PhD Dissertation, Saga University, Saga,
Japan.
Horpibulsuk S., Runglawan R. and Suddeepong A. 2011. Assessment of
strength development in blended cement admixed Bangkok clay.
Construction and Building Materials, Vol. 25, No. 4, p. 1521-1531.
Horpibulsuk S., Miura N. and Bergado D.T. 2004. Undrained shear
behavior of cement admixed clay at high water content. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 130,
No. 10, p. 10961105.
Horpibulsuk S., Miura N. and Nagaraj T.S. 2003. Assessment of
strength development in cement-admixed high water content clays
with Abrams' law as a basis. Gotechnique, Vol. 53, No. 4, p. 439
444.
Kamruzzaman A.H.M. 2002. Physico-chemical and engineering
behavior of cement treated Singapore marine clay. PhD Thesis,
Department of Civil Engineering, National University of Singapore.
Kawasaki T., Niina A., Saitoh S. and Honjyo Y. 1981. Deep mixing
method using cement hardening agent. Proceedings of the 10th
International Conference on Soil Mechanics and Foundation
Engineering, Vol 3, p. 721-724.
Kitazume M. and Terashi M. 2002. The deep mixing method principle,
design and construction. Edited by Coastal Development Institute
of Technology, Japan. Balkema.
Locat J., Trembley H. and Leroueil S. 1996. Mechanical and hydraulic
behaviour of a soft inorganic clay treated with lime. Canadian
Geotechnical Journal, Vol. 33, p. 654 669.
Lorenzo G.A., and Bergado D.T. 2004. Fundamental parameters of
cement-admixed clay- New approach. Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, Vol. 130, No. 10, p. 10421050.
Lorenzo G.A., and Bergado D.T. 2006. Fundamental characteristics of
cement-admixed clay in deep mixing. Journal of Materials in Civil
Engineering, ASCE, Vol. 18, No. 2, p. 161-174.
Miura N., Horpibulsuk S. and Nagaraj T.S. 2001. Engineering behavior
of Cement stabilized clays at high water content. Soils and
Foundations, Vol. 41, No. 5, p. 33-45.
Uddin K., Balasubramaniam A.S. and Bergado D.T. 1997. Engineering
behavior of cement-treated Bangkok soft clay. Geotechnical
Engineering Journal, Southeast Asian Geotechnical Society, Vol.
28, No. 1, p. 89-119.
Venda Oliveira P.J., Lemos L.J.L., and Coelho P.A.L.P. 2010. Behavior
of an atypical embankment on soft soil: field observations and
numerical
simulation.
Journal
of
Geotechnical
and
Geoenvironmental Engineering, ASCE, Vol. 136, No. 1, p. 35-47.

Modlisation numrique du comportement dune colonne de soil-mixing


et confrontation un essai de chargement en vraie grandeur
Numerical modeling of a soil-mixing column behavior and comparison with a full-size load test
Cuira F.

TERRASOL, Paris, France

Costa dAguiar S.

SNCF I&R, Paris, France

Grzyb A., Pellet F.


INSA, Lyon, France

Mosser J.-F.

SOLETANCHE BACHY, Rueil-Malmaison, France

Guimond-Barrett A., Le Kouby A.


IFSTTAR, Paris, France

RSUM : Cet article prsente les rsultats dun travail de modlisation numrique visant simuler un essai de chargement axial sur
une colonne de sol-ciment, ralise par soil-mixing en voie humide. Quatre modles ont t btis dans le cadre de ce travail : trois
modles en lments finis, et un modle semi-analytique simplifi. Les rsultats des quatre modles sont confronts ceux obtenus
par un essai de chargement monotone en vraie grandeur, ralis sur une colonne de soil mixing de 400 mm de diamtre et 5 m de
hauteur, mise en uvre dans un sol limoneux sablo-graveleux. Ces rsultats permettent notamment de reproduire le mode de rupture
observ lors de lexcavation de la colonne, et mettent en vidence la ncessit de modliser correctement le comportement non
linaire du matriau sol-ciment qui influe significativement sur le comportement global, la diffrence des pieux rigides .
ABSTRACT: This article shows the results of a numerical modelling study aiming at simulating an axial load test on a soil-cement
column, carried out using the wet soil-mixing method. Four models were built as part of this study: three finite element models and
one simplified semi-analytical model. The results from the four models were compared with those from a full-size monotonic load
test, performed on a 400 mm diameter and 5m high soil-mixing column, installed in silty to sandy-gravelly soils. These results
allow to reproduce the failure mode observed during the column excavation, by emphasizing the need of an accurate modelling of the
non-linear soil-cement material which has a significant influence on the general behaviour, unlike with rigid piles.
KEYWORDS: Numerical modelling, soil reinforcement, Soil-mixing, load test.
1

INTRODUCTION

La technique du soil mixing permet damliorer les


caractristiques dun sol meuble par mlange mcanique in situ
avec un liant hydraulique. Le sol initialement prsent sur le
chantier est alors valoris comme un matriau de construction,
avec le double intrt de diminuer les dchets (sols excavs) et
de rduire la consommation de matriaux et dnergie.
Le procd Springsol utilise un outil ouvrant, pour raliser des
colonnes de sol-ciment de diamtre variable, ce qui limite au
maximum limpact des travaux sur les existants. Ce procd
offre des perspectives aussi bien dans le domaine de la
maintenance des plateformes ferroviaires (possibilit de
renforcer les structures ferroviaires sans avoir dposer les
voies), que dans celui du renforcement des fondations existantes
(empreinte des forages sur les structures limite au diamtre de
loutil ferm).
Cette souplesse et le caractre conome en dchets et en
matriaux permettent au procd Springsol de rpondre aux
nouvelles exigences environnementales et conomiques des
projets. Cest donc dans le but de dvelopper ces solutions que
Soletanche Bachy, la SNCF et Terrasol collaborent au sein du
projet de recherche RUFEX (Renforcement et rUtilisation des
plateformes et Fondation Existantes) avec lIFSTTAR, lINSA
de Lyon et lEcole des Ponts ParisTech.
Cet article prsente les rsultats dun travail de modlisation
numrique dun essai de chargement monotone conduit jusqu
la rupture sur une colonne de soil mixing ralise sur un
chantier de validation du projet RUFEX.

ESSAI DE CHARGEMENT EN VRAI GRANDEUR

Un essai de chargement sur une colonne de soil-mixing de 400


mm de diamtre et 5 m de hauteur a t ralis sur le site
exprimental du projet Rufex situ sur la commune de
Vernouillet (78). Le contexte gotechnique est caractris par
des remblais en surface suivis dune couche de limon sableux
reposant sur un sable graveleux. Aucune nappe phratique na
t rencontre lors des reconnaissances. Le Tableau 1 rcapitule
les caractristiques go-mcaniques issues des essais raliss.
Tableau 1. Caractristiques gotechniques des sols du site.
Limon
sableux

Sable
graveleux

0,5

3,5

Pression limite nette pl* (MPa)

2,5

Module pressiomtrique Em (MPa)

10

20

Cohsion c (kPa)

Angle de frottement ()

27

37

Nature
Profondeur du toit (m)

La colonne teste a t fore avec loutil Springsol. Le


malaxage du sol avec le liant hydraulique (ciment de type CEM
III) a t effectu in situ par voie humide. La quantit de ciment
injecte dans la colonne sous forme de coulis tait denviron
230 kg / m3.

2461

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Lessai de chargement a t ralis 90 jours aprs


linstallation de la colonne, avec des paliers de chargement de
50 kN maintenus pendant 30 min. La courbe de chargement
obtenue est prsente sur la Figure 1. Le chargement a t arrt
pour une charge maximale de 400 kN, lorsque le tassement de la
tte de la colonne a dpass 40 mm (1/10ime du diamtre).
Charge en tte Q (KN)
0

100

200

300

3 MODLISATION DE LESSAI DE CHARGEMENT :


MODLES LMENTS FINIS
Le rsultat de lessai de chargement est utilis comme rfrence
pour lvaluation de quatre modles numriques dont trois bass
sur un traitement complet en lments finis. Ces modles sont
construits selon la coupe schmatique suivante (Figure 3) et sur
la base des paramtres donns dans les Tableaux 1 et 2.
Q

400

Tassement s (mm)

0
10

0,5 m

Remblai

20
30

2,0 m

1,0 m

1,5 m

Limons
sableux

40
50

Transition

Figure 1. Courbe de chargement sur colonne de soil-mixing.

Aprs 180 jours, lexcavation de la colonne teste a permis


de distinguer une partie suprieure (de 0 2,5 m) constitue de
limon trait et une partie infrieure compose de sable trait.
Une zone de transition faite dun mlange de limon et de sable
trait a t observe entre 2,5 et 3,5 m. Des fragments de la
colonne ont t prlevs, carotts en laboratoire, et soumis des
essais mcaniques pour dterminer la rsistance et le module de
dformation du matriau constitutif de la colonne (Tableau 2).
Tableau 2. Rsultats des essais en laboratoire sur les prouvettes
provenant de la colonne excave.
Limon
trait (1)

Transition
(2)

Sable
trait (3)

0,5 - 2,5

2,5 3,5

3,5 - 5,0

3,7

7,6

11,9

Module local E50 (MPa)

1280 Rc

1280 Rc

1280 Rc

Angle de frottement ()

42

42

42

Cohsion (kPa)

700

1700

2800

Nature
Profondeur (m)
Rsistance Rc (MPa)

Vers 1 m de profondeur, la colonne de limon trait apparat


particulirement fissure et fracture (Figure 2), ce qui fait
suggrer que la rupture sest produite au sein du matriau
constitutif de la colonne. Ce constat est confort par le fait que
la portance de la colonne (estime partir des rsultats des
essais pressiomtriques) tait a priori suprieure sa rsistance
interne (estime partir des rsultats des essais sur les
prouvettes carottes).

Sable
graveleux

= 400 mm

Figure 3. Coupe de calcul retenue

3.1

Modle 1 (logiciel GEFDyn)

Ce modle est bti laide du logiciel lments finis GEFDyn.


Il sagit dun modle tridimensionnel o la colonne est
modlise par des lments volumiques 8 nuds formant un
quart de cylindre de 400 mm de diamtre et 5 m de long, noy
dans un milieu continu de 20 m de profondeur. Des lments
dinterface ont t introduits entre la structure de la colonne et
les lments de sol.
Le modle de comportement choisi est un modle linaire
lastique parfaitement plastique avec un critre de rupture de
Mohr Coulomb (not MC par la suite) pour les diffrents
matriaux (sol et colonne). La loi de linterface est celle
dAguiar et al (2011) dont la formulation est base sur les
mmes hypothses de comportement que le modle dit de
Hujeux (1985) qui se rvle trs adapt au comportement non
linaire des sols. La richesse de ce type de modle rside dans la
possibilit, en fonction des paramtres de la surface de charge et
du type dcrouissage, de modliser des comportements qui
peuvent aller du simple lastique parfaitement plastique un
comportement lasto-plastique crouissage dviatorique et
volumique. Pour le prsent calcul, les paramtres de linterface
sont choisis de manire reprsenter une surface de charge de
Mohr Coulomb avec un comportement lastique parfaitement
plastique. Langle de frottement linterface a t choisi gal
celui du sol environnant. Cela est justifi par le mode de
ralisation de la colonne (malaxage local) qui produit une
interface rugueuse mobilisant ainsi un mcanisme de rupture
mettant en jeu la rsistance intrinsque du sol (Figure 4).

Zone de limon
trait fissure et
fracture vers 1 m
de profondeur

Figure 2. Excavation de la colonne aprs lessai de chargement.


Figure 4. Etat de la surface dun tronon de colonne excave

2462

Technical Committee 211 / Comit technique 211

Modle 2 (logiciel ABAQUS)

3.2

Le modle 2 est bti sous le logiciel ABAQUS. Il sagit dun


modle axisymtrique qui intgre une loi de comportement
avance de type Drucker-Prager modifie avec cap (DPC).
Ce modle permet de prendre en compte leffet dcrouissage
du sol, lhistorique de contraintes, ainsi que leffet de
compaction en pointe sous la colonne. La surface de charge est
compose de trois parties : une limite de rupture en cisaillement
de type Drucker-Prager, un cap elliptique, et une zone de
transition (Figure 5).

Ce modle permet par ailleurs de tenir compte de la variation


du module de dformation scant E50 ( 50% de la contrainte de
rupture) avec ltat de contraintes. Cette variabilit est contrle
par un paramtre puissance not m, quon prend usuellement
gal m = 0,5 pour le sol en place (le module scant est
proportionnel la racine de la contrainte applique). Pour la
colonne, ce paramtre est pris gal m = 0 (pas de variation du
module scant avec ltat de contraintes). Enfin, cette loi est
combine avec un critre de rupture de type Mohr Coulomb.
Des lments dinterface ont t par ailleurs introduits sur
toute la frontire de la colonne avec une loi de contraintesdformations et un critre de rupture identiques ceux des sols
environnants.
Mise en uvre et rsultats

3.4

Figure 5. Loi de Drucker-Prager modifie avec cap (DPC)

Les paramtres et d sexpriment en fonction de langle de


frottement interne et la cohsion c laide de la relation (1) :

6 sin
3 sin

et d

18c cos
3 sin

Les autres paramtres sont choisis soit par calage, soit dune
manire forfaitaire sur la base dlments bibliographiques. En
particulier, le paramtre pb qui dlimite la surface dcrouissage,
doit en toute rigueur tre cal sur le rsultat dun essai de
consolidation isotrope. Dans le prsent exercice, ce modle
(DPC) a t considr pour caractriser le comportement des
sables graveleux. Pour les autres matriaux (colonne et limons
sableux), il a t retenu une loi linaire lastique parfaitement
plastique avec critre de rupture de Mohr Coulomb (MC). Pour
la dfinition de la loi DPC dans les sables graveleux, les
paramtres suivants ont t considrs : R = 0.10, = 0,01 et pb
fonction de la dformation volumique plastique selon la loi
dcrouissage des sables dOttawa (Helwany 2000). Enfin, des
lments dinterface ont t introduits avec une loi MC.
Modle 3 (logiciel PLAXIS)

3.3

Charge en tte (kN)

(1)

Le 3 modle est un modle axisymtrique bti sous le logiciel


PLAXIS. Les lments utiliss sont des lments triangulaires
15 nuds et 30 degrs de libert. Pour le sol et la colonne, on
choisit une loi de contraintes-dformations de type HSM
(Hardening Soil Model - Figure 6) qui est une loi hyperbolique
tenant compte de lcrouissage en cisaillement et en
compression.

50

100

150

200

250

300

350

400

5
10

Tassement en tte (mm)

tan

Pour les trois modlisations ralises, les donnes


gotechniques ont t compltes par les valeurs du module de
dformation scant E50h/2 mi-paisseur dans chaque couche :
E50h/2 = 25 MPa pour les limons sableux et E50h/2 = 100 MPa
pour les sables graveleux. La Figure 7 prsente la courbe de
chargement simule laide des trois modles prsents cidessus. Une trs bonne concordance est observe entre la
modlisation et les mesures jusqu 300 kN (75% de la charge
de rupture mesure). Les modles 2 et 3 mettent en vidence un
palier de rupture net situ entre 350 et 400 kN, ce qui
correspond, 10% prs, au palier obtenu par lessai de
chargement sur site.

15
20

Modle 1 (GEFDyn - MC)


Modle 2 (ABAQUS - DPC / MC)
Modle 3 (PLAXIS - HSM)

25

Essai

30

Figure 7. Simulation de la courbe de chargement par modlisation


numrique en lments finis

Dans les modles 2 et 3, le palier de rupture obtenu


correspond au dveloppement dune zone de plastification
conique dans la partie suprieure de la colonne vers 1 m de
profondeur (Figure 8). Ce constat est corrobor par les
observations faites sur site lors de lexcavation de la colonne.

0,5m

1,0m

1,5m

Figure 8. Dveloppement dun mcanisme de rupture localis dans la


colonne (modle 3)

Figure 6. Principe de la loi HSM sous PLAXIS

2463

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

4
4.1

Principe du modle

On utilise prsent un modle semi-analytique simplifi bas


sur la mthode dite t-z : la colonne est assimile une
poutre verticale travaillant en compression axiale, tandis que le
frottement latral et la contrainte en pointe q suivent une loi de
mobilisation de Frank et Zhao (1982), et sont donc fonctions du
dplacement vertical de la colonne w. Chaque loi est
caractrise laide de deux paramtres : un paramtre de pente
(Kt ou Kp) et une contrainte unitaire limite (qs ou qp). Ces lois
sont couramment utilises en France pour estimer le tassement
dun lment de fondation profonde, et se rvlent trs efficaces
dans les exercices de calage par rapport un essai de
chargement en vraie grandeur.
Frottement latral

Rsultats

4.3

MODLE SEMI-ANALYTIQUE SIMPLIFIE

Contrainte en pointe

La Figure 11 prsente les rsultats obtenus (courbe de


chargement). Deux cas ont t tudis : cas dun comportement
linaire lastique de la colonne (E = E50), et cas dun
comportement non linaire (E = f()) selon la loi dcrite dans la
Figure 10. Pour chaque cas, deux situations sont examines :
frottement de type sol/sol (qs = f(v)) et frottement de type
bton/sol (qs = f(pl*)). Les rsultats obtenus confirment la
pertinence dune loi de comportement non linaire pour le
matriau constitutif de la colonne, et montrent que le choix dun
frottement de type sol/sol est plus reprsentatif du
comportement rel observ. Avec ces hypothses, le rsultat du
modle semi-analytique se rvle trs concordant avec celui de
lessai de chargement jusquau palier de rupture. Celui-ci est
obtenu par plastification en tte de la colonne (contrainte
applique proche de la rsistance la compression simple Rc).
Charge en tte (kN)

50

100

150

200

250

300

350

400

450

500

5
10
15

Figure 9. Lois de mobilisation de type Frank et Zhao

Asymptote = Rc

Rc

0,5 Rc
E50

2 E50 Rc

4.2

Mise en uvre

La loi de contrainte-dformation de la Figure 10 est construite


partir des paramtres (E50, Rc) qui Figurent dans le Tableau 2.
Les paramtres de pente des lois de Frank et Zhao (Figure
9) sobtiennent par corrlation avec le module
pressiomtrique EM : Kt = 0,8 EM / B et Kp = 4,8 EM / B, o B
dsigne le diamtre de la colonne. Ensuite, pour le choix du
frottement latral limite qs, deux hypothses enveloppes sont
examines : la premire est celle dun contact bton/sol pour
laquelle la valeur de qs sobtient par corrlation avec la pression
limite pl* ; la deuxime hypothse est celle dun contact sol/sol
pour laquelle la valeur de qs est celle du cisaillement limite de
Mohr Coulomb : qs = tan().K0.v, avec K0 = 0,5 et v la
contrainte verticale effective initiale linterface de la colonne.
Enfin, la contrainte limite en pointe qp est prise gale 4 MPa.

2464

Essai

25
30

E = f() et qs = f(v)

35

E = f() et qs = f(pl*)

40

E = E50 et qs = f(v)

45

E = E50 et qs = f(pl*)

50

Figure 11. Courbe de chargement Modlisation analytique simplifie

CONCLUSION

Les enseignements tirs de lessai de chargement en vraie


grandeur ont permis dorienter le choix des paramtres dentre
des diffrents outils numriques dvelopps dans le cadre du
projet RUFEX. Les rsultats obtenus, tant par les modles
numriques que par le modle semi-analytique, permettent de
bien reproduire le comportement observ lors de lessai, aussi
bien sur le comportement avant rupture que sur le mode de
rupture. Ils mettent en vidence la ncessit de modliser
correctement le comportement non linaire du matriau solciment qui influe significativement sur le comportement global,
la diffrence des pieux rigides .
6

Figure 10. Loi de contrainte-dformation retenue pour la colonne

20

Tassement (mm)

Les courbes de mobilisation ci-dessus sont combines avec


la loi de comportement de la colonne, qui relie la contrainte
applique au taux de dformation axiale de la colonne . A la
diffrence des pieux classiques pour lesquels le contraste de
rigidit pieu/sol est tel que lessentiel des tassements est obtenu
en pointe, la particularit dune colonne de soil-mixing rside
dans un contraste de rigidit colonne/sol plus faible et une
sensibilit notable de la raideur globale en tte vis--vis du
comportement local. Ces lments ont justifi le recours une
loi de contrainte-dformation non linaire pour la colonne : il a
t retenu une loi de forme hyperbolique (Figure 10) construite
laide de deux paramtres : le module scant E50 et la
rsistance la compression simple Rc. Lors du prsent exercice,
cette loi sest rvle apte retranscrire le comportement
observ, la diffrence dune loi linaire lastique.

REMERCIEMENTS

Les auteurs tiennent remercier la DGCIS (Direction Gnrale


de la Comptitivit et des Services) et les Conseils Gnraux de
la Rgion Ile de France et 93 qui cofinancent cette recherche.
7

REFERENCES BIBLIOGRAPHIQUES

DAguiar, S. C., A. Modaressi-Farahmand-Razavi, J. A. dos Santos, and


F. Lopez-Caballero (2011). Elastoplastic constitutive modeling of
soil-structure interfaces under monotonic and cyclic loading.
Computers and Geotechnics 38(-), 430447.
Frank, R. & Zhao, S. R. (1982), Estimation par les paramtres
pressiomtriques de lenfoncement sous charge axiale des pieux
fors dans les sols fins, Bull. Liaison Labo. P. et Ch. 119 :17-24.
Helwany S. 2000 Applied Soil Mechanics with ABAQUS Applications,
J. WILEY & SONS, INC pp. 61-67
Hujeux, J. C. (1985). Une loi de comportement pour le chargement
cyclique des sols en gnie parasismique, pp. 278302. V.
Davidovici, Presses ENPC.

Design of Deep Soil Mix Structures: considerations on the UCS characteristic value
Dimensionnement des structures en soil mix : considrations sur la valeur caractristique UCS
Denies N., Van Lysebetten G., Huybrechts N.

Belgian Building Research Institute, BBRI, Belgium

De Cock F.

Geotechnical Expert Office Geo.be, Belgium

B.
Lameire
Lameire
B.

Belgian Association of Foundation Contractors ABEF, Belgium

Maertens J.

Jan Maertens bvba & KU Leuven, Belgium

Vervoort A.

KU Leuven, Belgium
ABSTRACT: Since several decades, the deep soil mix (DSM) technique has been used for ground improvement works. But in recent
years, this technique has been increasingly used for structural applications. Standardized guidelines for the execution and the design
of this kind of applications are not currently available. For the purpose of developing such guidelines, mechanical characteristics of
DSM material were investigated. Within the framework of a Flemish regional research program (IWT 080736), DSM material from
38 Belgian construction sites, with various soil conditions and for different execution processes, has been tested. Internationally
QA/QC activities are commonly related to tests on core samples for the determination of the Unconfined Compressive Strength
(UCS) and the modulus of elasticity (E) of the material. Both values allow an approach of the design which takes into account the
bending characteristics (EI), the deformation (E), the arching effect (UCS) and the structural resistance (UCS) of the element. For the
semi-probabilistic design approach presented in Eurocode 7, a characteristic value of the UCS has to be defined as part of the
design of DSM structures. The present paper discusses the definition of this value.
RSUM : Depuis plusieurs dcennies, la technique du soil mix est utilise comme procd damlioration du sol. Mais ces dernires
annes, elle est de plus en plus utilise pour des applications structurelles. Aucune directive nest actuellement disponible pour
lexcution et le dimensionnement de telles applications. De manire dvelopper de telles directives, les caractristiques mcaniques
du matriau soil mix ont t investigues. Dans le cadre dun programme de recherche financ par lIWT, lagence gouvernementale
flamande pour linnovation, des chantillons de soil mix de 38 sites de construction ont t tests pour diffrents types de sol et
diffrents systmes. La qualit du matriau soil mix est gnralement contrle laide dessais, raliss sur des chantillons carotts
in situ, par lesquels sont dtermins la rsistance la compression simple (UCS) et le module dlasticit (E) du matriau. Ces deux
grandeurs permettent une approche du dimensionnement tenant compte de la rigidit flexionnelle (EI), des dformations (E), de leffet
de vote (UCS) et de la rsistance structurelle (UCS) de llment. Au vue de lapproche semi-probabiliste de lEurocode 7, il est
important de dfinir la valeur caractristique de la rsistance du soil mix (UCS) prendre en compte dans le dimensionnement. Le
prsent article discute de la dfinition de cette valeur caractristique.
KEYWORDS: Deep soil mix wall, structural design, ucs characteristic value
1

INTRODUCTION

The Deep Soil Mix (DSM) process was introduced in the 70s
in Japan and in the Scandinavian countries. Since several
decennia, DSM has been known as a ground improvement (GI)
technique. According to the classification of GI methods
adopted by the ISSMGE TC 211, DSM can be classified as
ground improvement with grouting type admixtures. Numerous
reviews and recent progresses of the DSM technique are
referred in Denies and Van Lysebetten (2012). The results of
national and European research programs have also been
published in multiple interesting reports (such as Eurosoilstab
2002), while the European standard for the execution of deep
mixing Execution of special geotechnical works Deep
Mixing (EN 14679) was published in 2005. Most of these
research projects focused on the global stabilization of soft
cohesive soils such as clay, silt, peat and gyttja (result of the
digestion of the peat by bacteria). More recently, DSM is
increasingly being used for structural applications such as soil
mix walls (SMW) for the retaining of soil and water in the case
of excavations.
In the DSM process, the ground is mechanically mixed in
place, while a binder, based on cement, is injected. For SMW
applications, the DSM cylindrical columns or the rectangular

panels are placed next to each other, in a secant way. By


overlapping the different soil mix elements, a continuous SMW
is realized. Steel profiles are inserted into the DSM fresh
material to resist the shear forces and bending moments. The
main structural difference between SMW and the more
traditional secant pile walls is the constitutive DSM material
which consists of a soil cement mixture instead of concrete.
Elements such as piles or diaphragm walls only comprise
standardized components and their characteristic strength can be
defined by the strength class of concrete. The design approach
for the DSM material is very different since the existing soil is
used as an essential component of the final product. Moreover,
the DSM strength depends not only on the soil type, but also on
the DSM technique, the amount and the type of binder, etc.
Within the framework of the BBRI Soil Mix project
initiated in 2009 in collaboration with the KU Leuven and the
Belgian Association of Foundation Contractors (ABEF),
numerous tests on in situ DSM material have been performed. A
good insight has been acquired with regard to mechanical
characteristics that can be obtained with the CVR C-mix, the
TSM and the CSM systems in several Belgian soils as reported
in Denies et al. (2012). BBRI information sheets (BBRI, 2012a
and b) have been published for the purpose of helping
contractors to improve the quality control (QC) of their finished

2465

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

product, but guidance rules for the design of SMW are still
lacking in particular for the determination of a characteristic
value representative of the strength of the soil mix material.
Neither in the Eurocode 7 nor in the European standards for
grouting (EN 12715), jet-grouting (EN 12716) or deep-mixing
(EN 14679), specifications are given for the internal strength of
the material.
In practice, Quality Assurance (QA) and Quality Control
(QC) activities are commonly related to tests on core samples
for the determination of the Unconfined Compressive Strength
(UCS) and the modulus of elasticity (E) of the material. Both
values allowing an approach of the design taking into account
the bending characteristics (EI), the deformation (E), the
arching effect (UCS) and the structural resistance (UCS) of the
element. For engineering purposes and as part of the semiprobabilistic design approach presented in Eurocode 7, it is thus
essential to define the UCS characteristic value that can be
taken into account in the design of DSM structures. The
following paragraphs discuss the definition of this value.
2 DETERMINATION OF THE UCS CHARACTERISTIC
VALUE OF DSM MATERIAL
On the basis of an X% lower limit value

2.2

Figure 1. a) Distribution of the UCS values of 41 cores of DSM material


from a site in Gent (Belgium) and the corresponding theoretical
Gaussian curve. b) Distribution of the logarithm of the UCS values
increased with = 0.6 from the same site and the corresponding
Gaussian curve. The vertical line indicates the 5% lower limit value,
after Denies et al. (2012).
35
30
25

UCS (MPa)

The first methodology consists in the calculation of the


characteristic strength as the X% lower limit on the basis of a
statistical distribution function. Nevertheless, in practice, the
wrong assumption is often made that the datasets of UCS values
of soil mix material are normally distributed (see Fig. 1a). The
characteristic UCS value is then erroneously calculated as the
X% lower quantile of the normal distribution with parameters
corresponding to the dataset. Moreover, this often results into
negative and thus useless characteristic UCS values. The
mathematically correct solution would be to apply the best
fitting standard distribution function, for example a lognormal
distribution in case the distribution is skewed and/or does not
contain subpopulations. The X% lower limit can then be
calculated on the basis of this theoretical distribution function,
as illustrated in Denies et al. (2012) for a lognormal distribution
(see Fig. 1b). Possibly, a factor has to be added to the values
to obtain an optimal fit with a normal distribution after
transformation. However, this way of working is probably too
complex to apply in practical situations.
The second methodology to determine the X% lower limit is
based on the cumulative frequency curve of the original
experimental dataset and thus independent of any theoretical
distribution function. Note that to apply this method, enough
data points have to be available (for an accurate determination
of the 5% lower limit without extrapolation, at least 20 samples
are necessary). This approach seems rather simple but any other
method probably results in a large uncertainty. Figure 2 presents
the cumulative frequency curve for the UCS values of the
dataset illustrated in Fig. 1.

f c ,k quf

5
0

a.

5 10

50
90
Cumulative percentage

99

5
4
3
2
1
0

b.

5
10
Cumulative percentage

50

Figure 2. Cumulative frequency curve of all UCS values of the dataset


from the site in Gent: a) Full curve. b) Zoom on the part below 50%:
presentation of the construction for the evaluation of the 5% lower limit
value.

(1)

where q uf is the mean UCS value and a factor representing a


certain confidence and safety level ( < 1).
In the formalized design approach (DIN 4093, August 2012)
used in Germany, the UCS characteristic value is defined as the
minimum value of three parameters:

15
10

On the basis of an average value with safety factor

A second approach to determine the UCS characteristic value is


the use of the average value of the dataset in combination with a
safety factor:

20

UCS (MPa)

2.1

f c,k min f m,min ; f m,mittel ; 12 MPa

(2)

where fm,min is the minimum UCS value and fm,mittel the


arithmetic mean UCS value from a series of at least 4 samples.
is determined in function of fc,k: equals 0.6 for fc,k 4 MPa
and 0.75 for fc,k = 12 MPa (linear interpolation is required for
intermediate values). This method is described in more detail by
Topolnicki and Pandrea (2012).

2466

Technical Committee 211 / Comit technique 211

If the characteristic value fc,k is smaller than 4 MPa,


additional creep tests have to be conducted with a load of fc,k/2
as described in the annex B of the DIN 4093.
The design strength for calculations with the concept of
partial safety factors is then computed as follows:
f c ,d 0.85

f c ,k

(3)

where 0.85 is a factor to consider permanent situations and m is


the material safety factor as defined in Eurocode 7 (1.5 for
permanent and temporary load cases and 1.3 for accidents). For
temporary situations, the design strength is computed without
the 0.85 coefficient.
As reported in Topolnicki and Pandrea (2012), if
independent and separate design calculations are performed for
compressive and shear stresses (i.e. no 3D stress analysis), the
maximum allowed compressive stress is 0.7 x fc,d and the
maximum allowed shear stress is 0.2 x fc,d.
For comparison with the previous version of the DIN 4093
(published in September 1987), Table 1 presents cumulated
safety factors on material strength (fm,mittel) and equivalent
global safety factors (m x G,Q)/( x 0.85 x (0.7 or 1)) computed
with the new DIN 4093 for permanent design situations. An
increase in the number of test samples has no effect on the
safety factors.

etc.) should depend on the type of the distribution of the dataset.


Second, problems may arise with limited number of samples,
skewed populations and in the presence of subpopulations.
Figure 3 compares the UCS characteristic value computed
with the help of the cumulative frequency curve (CC method) or
with respect to the DIN approach. The ratio of the two
characteristic values is presented as a function of the number of
tested samples for each considered dataset. Minimum 20
samples are necessary in order to conduct the statistical analysis
on the cumulative frequency curve. As observed in Fig. 3, the
UCS characteristic value is always greater when computed with
the help of the cumulative frequency curve (all the values are
larger than 1). In Fig. 3, results are given for two different X%
lower quantiles: X = 5% and 10%. Indeed, for the first category
of approaches (based on the lower limit value), a value for the
X% has to be defined. A more detailed analysis is necessary to
determine if a 5% lower limit, as often stated in Eurocode 7, is a
representative characteristic value for the strength of the soil
mix material. Actually, one major issue is the representativeness
of the core samples with regard to the in situ executed DSM
material.

Table 1. Cumulated safety factors on material strength (fm,mittel) and


equivalent global safety factors in permanent design situation according
to DIN 4093 August 2012 (m = 1.5).
For

With 3D analysis
Cumulated safety factor
Permanent actions (G=1.35)
Equivalent global safety factor
Variable actions (Q=1.50)
Equivalent global safety factor
Without 3D analysis
Cumulated safety factor
Permanent actions (G=1.35)
Equivalent global safety factor
Variable actions (Q=1.50)
Equivalent global safety factor

For

=0.6

=0.75

2.94

2.35

3.97

3.18

4.41

3.53

4.20

3.36

5.67

4.54

6.30

5.04

Figure 3. Ratio of the characteristic values (fc,k (CC) and fc,k (DIN4093))
as a function of the number of tested samples.

For comparison, in the previous version of the DIN 4093


(September 1987), the design value was computed as follows:
f c ,d

f m ,mittel
5

(4)

for samples with UCS values expected larger than 5 MPa and
tested according to the DIN 1048 standard for concrete material,
or with the help of:
f c ,d

q u
3

(5)

for samples with UCS values expected smaller than 5 MPa and
tested according to the DIN 18 136 for soil material. q'u is the
UCS value computed according to the DIN 18136.
Considering the safety factor of 5 and the reduction factor of
0.7 related to the 3D character of the loading, the previous
version of the DIN 4093 resulted in a global safety factor of
7.14.
For this second approach based on an average value with
safety factor, Denies et al. (2012) have remarked that first, the
definition of the most suitable mean (arithmetic mean, median,

INFLUENCE OF THE UNMIXED SOIL INCLUSIONS

There is mainly the question of the influence of unmixed soft


soil inclusions on the mechanical behaviour of the DSM
material. Indeed, as a natural material (i.e. soil) is being mixed,
it is to be expected that the entire wall is not perfectly mixed
and homogeneous: inclusions of unmixed soft soil are present.
As a result, Ganne et al. (2010) have proposed to reject all test
samples with soil inclusions > 1/6 of the sample diameter, on
condition that no more than 15% of the test samples from one
particular site would be rejected. This possibility to reject test
samples results from the reflexion that a soil inclusion of 20 mm
or less does not influence the behaviour of a soil mix structure.
On the other hand, a soil inclusion of 20 mm in a test sample of
100 mm diameter significantly influences the test result. Of
course, this condition is only suitable if one assumes that there
is no soil inclusion larger than 1/6 of the width of the in situ
DSM structure. For the purpose of studying this question, 2D
numerical simulations were performed at KU Leuven with the
aim to quantify the effect of soil inclusions on the DSM strength
and stiffness. The following parameters are being considered:
size, number, relative position and percentage of soil inclusions.
The results of this study are presented in Vervoort et al. (2012)
and Van Lysebetten et al. (2013). As illustrated in Fig. 4, they
confirm that DSM samples with soft soil inclusions larger than
1/6 have a considerable influence on the deduction of the
engineering values. Based on this numerical analysis, the rule
of 1/6 as proposed by Ganne et al. (2010) seems to be justified.

2467

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

the determination of the X% lower quantile for DSM


material (in case of statistical calculation),
- the presence of the unmixed soft soil inclusions potentially
considering the rule of 1/6 (Ganne et al. 2010),
- the scale effect (with regard to the full-scale factor of 0.7),
- the possibility of 3D analysis,
- and the time effects (with the help of creep test or based on
experience with similar technique and soil conditions).
The curing and creep phenomena are currently investigated
within the framework of the BBRI Soil Mix project. Indeed,
while SMWs were previously used only for temporary
excavation support, permanent retaining and bearing
applications with soil mix are increasingly applied in Belgium.
For the evolution of the UCS value with time, it is suggested to
consider the value of the UCS at 28 days as the value of
reference for the strength of the DSM material.
-

Figure 4. Influence of the dimensions of the soil inclusions on the UCS


of soil mix material. Results of 2D numerical simulations performed
with the help of the Universal Distinct Element Code UDEC of Itasca.
Details of the model are available in Van Lysebetten et al. (2013). H is
the ratio between the height of the soil inclusion and the sample
diameter.

INFLUENCE OF THE SCALE EFFECT

Apart from traditional core samples (with a diameter around 10


cm), large scale UCS tests were conducted on rectangular
blocks with approximately a square section, with a width
corresponding to the width of the in situ SMW (about half a
meter) and with a height approximately twice the width
(Vervoort et al. 2012). The results of all the tests performed in
KU Leuven are presented in Fig. 5 for various soil conditions
and different execution systems: the CSM and the TSM.
As observed in Fig. 5, a linear relationship is observed
between the test results obtained from the typical core samples
and the large rectangular blocks. Although there is a scatter in
the test results, the UCS of the full-scale blocks is about 70% of
the average UCS of the typical core samples. It is to note that
similar conclusion was observed for DSM columns in Japan
(CDIT 2002).

Figure 5. Scale effect: relationship between the results of UCS tests on


typical cylindrical core samples (10 cm diameter) and on large
rectangular blocks tested in KU Leuven (after Vervoort et al. 2012).

CONCLUSIONS

Based on the results of the BBRI Soil Mix project, a Belgian


design methodology for the DSM structures is currently
developed. On the one hand to determine the UCS characteristic
value of the DSM material and on the other hand to design the
SMW as a retaining wall according to the requirements of the
Eurocode 7. According to the results presented in this paper, the
calculation of the UCS characteristic value should consider:
- the number of tested core samples,
- the possibility to use a statistical approach (based on the
cumulative curve) or an approach such as in the DIN,

2468

ACKNOWLEDGEMENTS

This research is financially supported by the Agency for


Innovation by Science and Technology of the Flemish Region
IWT (BBRI Soil Mix project, 2009-2013).
7

REFERENCES

BBRI. 2012a. Infofiche. Parois de type Soil mix de type 1 : parois


faites de colonnes. BBRI information sheet 56.5, www.bbri.be, July
2012 (in Dutch and French).
BBRI. 2012b. Infofiche. Parois de type Soil mix de type 2 : parois
faites de panneaux. BBRI information sheet 56.6, www.bbri.be, July
2012 (in Dutch and French).
CDIT. Coastal Development Institute of Technology. 2002. The Deep
Mixing Method Principle, Design and Construction. Edited by
CDIT, Japan. A. A. Balkema Publishers/Lisse/Abingdon/Exton
(PA)/Tokyo.
Denies, N. and Van Lysebetten, G. 2012. General Report Session 4
SOIL MIXING 2 DEEP MIXING. International Symposium of
ISSMGE - TC211. Recent research, advances & execution aspects
of ground improvement works. N. Denies and N. Huybrechts (eds.).
31 May-1 June 2012, Brussels, Belgium, Vol. I, pp. 87-124.
Denies, N., Huybrechts, N., De Cock, F., Lameire, B., Vervoort, A.,
Van Lysebetten, G. and Maertens, J. 2012. Soil Mix walls as
retaining structures mechanical characterization. International
Symposium of ISSMGE - TC211. Recent research, advances &
execution aspects of ground improvement works. 31 May-1 June
2012, Brussels, Belgium, Vol. III, pp. 99-115.
DIN 4093:2012-08. Design of ground improvement Jet grouting, deep
mixing or grouting. August 2012 (in German).
Eurosoilstab. 2002. Development of design and construction methods to
stabilise soft organic soils. Design Guide Soft Soil Stabilisation. EC
project BE 96-3177.
Ganne, P., Huybrechts, N., De Cock, F., Lameire, B. and Maertens, J.
2010. Soil mix walls as retaining structures critical analysis of the
material design parameters, International conference on
geotechnical challenges in megacities, June 07-10, 2010, Moscow,
Russia, pp. 991-998.
Topolnicki, M. and Pandrea, P. 2012. Design of in-situ soil mixing.
International Symposium of ISSMGE - TC211. Recent research,
advances & execution aspects of ground improvement works. 31
May-1 June 2012, Brussels, Belgium, Vol. III, pp. 309-316.
Van Lysebetten G., Vervoort A., Denies, N., Huybrechts, N., Maertens,
J., De Cock, F. and Lameire B. Numerical modeling of fracturing in
soil mix material. International Conference on Installation Effects
in Geotechnical Engineering. March 24 27, 2013. Rotterdam. The
Netherlands.
Vervoort, A., Tavallali, A., Van Lysebetten, G., Maertens, J., Denies,
N., Huybrechts, N., De Cock, F. and Lameire, B. 2012. Mechanical
characterization of large scale soil mix samples and the analysis of
the influence of soil inclusions. International Symposium of
ISSMGE - TC211. Recent research, advances & execution aspects
of ground improvement works. 31 May-1 June 2012, Brussels,
Belgium, Vol. III, pp. 127-135.

Method of improvement of the subsoil under Adora facility Ohrid, Republic Of


Macedonia
Mthode damlioration du sous-sol sous le btiment Adora Ohrid, Rpublique de Macdoine
Dimitrievski L.

Faculty of Civil Engineering, Skopje, Republic of Macedonia

Ilievski D., Dimitrievski D., Bogoevski B., Strasheski A.

GEING Krebs und Kiefer International and others ltd. Skopje, Republic of Macedonia

ABSTRACT: Adora residential building is a 6-storey structure, built nearby Ohrid Lake (Ohrid, Republic of Macedonia). The
foundation depth of the building is approximately 1,5 m (foundation construction foundation slab). The foundation soil consists of
soil materials which have a poor strength properties and low bearing capacity. The ground water table (GWT) on the location is on 1,0
m bellow the ground surface. On such geotechnical conditions a big settlements are expected. Therefore, a project on soil
improvement was prepared. Several preliminary solutions were considered, but most appropriated was the one which involves
geosyntetic reinforcement as subsoil improvement measure. In order to evaluate the settlements, performance of the building, axial
forces developed in the geogrids and stress-strain condition in the subsoil during static and dynamic loads, detailed analyses were
conducted. The software models developed in Plaxis 2D clearly showed the effectiveness on the applied measures for soil
improvement.
RSUM : le btiment de rsidence Adora est-une construction de 6 tages, bti cot du lac dOhrid (Ohrid, Rpublique de
Macdoine). La profondeur des fondations du btiment est denviron 1,5 m (construction de fondation dalle de fondation). Le sol de
fondation est compos de sols de mauvaise qualit et faible capacit portante. La nappe phratique (NP) du site est situe 1,0 m en
dessous de la surface de sol. Avec ces conditions gotechniques des tassements du sol sont attendus. Un projet damlioration des sols
a donc t prpar. Plusieurs solutions prliminaires ont t considres, mais la plus approprie est celle qui implique le
renforcement par gosynthtiques, comme mesure damlioration du sous-sol. Des analyses dtailles ont t menes afin dvaluer
les tassements du sol, la construction du btiment, les forces axiales dveloppes dans le gogrille et la relation contraintedformation dans le sous-sol sous chargements statique et dynamique. Les modles logiciels dvelopps dans Plaxis 2D montrent
clairement lefficacit des mesures appliques pour lamlioration des sols.
KEYWORDS:soil imporovement, geogrid, geotextile
1

INTRODUCTION

Adora residential building is foreseen to be built on a site which


hasvery poor geomechanical properties or in other words the
geotechnical conditions on the site are very unfavorable. In such
casesalways major problems are low bearing capacity of the
subsoil and large differential settlements. The city of Ohrid is
located in active seismic area which is classified in the 9th
seismic zone according to MCS. Moreover, on the site there are
layers of loose uniform sand. Having in mind these two facts a
liquefaction becomes also a serious danger for the structure. In
order to adopt a solution for soil improvement and to check the
liquefaction potential of the soil,comprehensive analyses were
conducted.

According to the geological formations on the site, there are


present sediments from the Quaternary period (Pleistocene
epoch), i.e. lake and swamp sediments, represented with gravel,
sand, sandy clays, silt, different types of clay and clayey-silty
sediments, as well as occasional presence of peat. The
sediments are well sorted, so that they have heterogenic particle
size distribution and heterogenic mineralogy composition, i.e.
they are fine to medium gravels and sands; fine silty sands,
sandy clays and soft lake/swamp clays with low to medium
plasticity. The thickness of these sediments varies between 50.0
80.0 m. According to the hydrogeological properties, they
belong to the group of low permeability sediments with
interparticle porosity. A closed type of springs with free level is
present in the sediments, at depth to 20.0 m with GWT = 5.0
7.0 m, and with permeability k = nx10-4 nx10-5 m/s.
Also there is a closed type of springs under pressure (artesian
springs) at depth from 20.0 60.0 m with discharge of Q = 1.0
5.0 l/sec.
From engineering geological point of view, these sediments
belong to the group of weathered rocks, well placed and sorted
with heterogenic particle size distribution, low compacted, with
smooth surfaces, fully saturated with water. In other words they
present materials with poor physical and mechanical properties
that have different strength and deformability parameters.
3

Figure 1.Excavation pit on a site

GEOLOGICAL SITE PROPERTIES

GEOMECHANICAL SITE PROPERTIES

The site for construction of the new building is located


approximately 200 m from the shore of Ohrid Lake, so, as it
was mentioned before, the soils found on the site are with

2469

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

sediment nature. In order to get precise geotechnical profile of


the ground, extensive geotechnical field investigations were
conducted. On the other hand, soil samples were taken for
laboratory testing. With such extensive scope of field
investigations, clear view of the ground profile was obtained.
Based on the performed field investigation works, it was
concluded that the site is composed of different layers of
sedimentary soils. With foreseen depth of the boreholes, no
bedrock was detected. On this site there is vast variety of soil
materials, from gravels and sands to silts and clays. Because of
the high heterogeneity of the ground profile, layers are grouped
into two extinguishing layers. The surface layers are low
plasticity clays and clayey sands, which are highly compressive
and they are present up to approximately 2.0 m from the
surface. While the next deeper layer is clayey sand with higher
compaction than the previous layer, and they are present up to
10.0 m from the surface. It is supposed that this second layer
continues up to depth for which loading stresses have impact on
the settlements.
According to the conducted field and laboratory tests,
present soil materials are loose and they have very poor strength
properties. In addition, the level of ground water table is very
high, approximately 1.0 m from the ground surface, because of
the nearby Ohrid Lake.
Table 1. Geomechanical properties
Fill Material

CL/SFc

SFc

g (kN/m3)

22.0

18.4

18.5

c(kN/m2)

0.0

15.0

10.0

35.0

10.0

20.0

80000.0

5000.0

10000.0

13

Soil

f (0)
Mv (kPa)
SPT

substructure) is 69.0x41.5 m. The foundation slab under the


superstructure is 0.9 m thick and on the extension parts it is 0.5
m thick.The contact pressure transferred on the subsoil varies in
range between 100 kPa (on cross-sectionsin the middle of the
building) and 20 kPa (on cross-sections in the extensions).

Figure 3. Cross section of the foundation structure

THE SOLUTION

As it was mentioned before, the improvement of the subsoil was


done by means of soil replacement and usage of geosynthetic
materials.
Because of the foundation level of -1.5 m from the ground
surface and the depth of soil replacement of 2.0 m, total of 3.5
m deep foundation pit was excavated. The excavation pit was
done by constructing2:1 slopes. In addition, dry conditions for
execution of the construction works in the excavation pit were
ensured by dewatering the excavation by extraction wells. On
the other hand the excavation pit had greater dimensions in plan
view, 2.0 m greater than the contours of the foundation slab.
Hence the loading stresses can be spread in the fill material by
angle of max 45.
Table 2. Properties of the geotextile
Raw material

PP multicolored/PET

Method of production

Mechanically bonded
300 gr/m2

Weight

Considering all these facts, it is obvious that ground


improvement is necessary under the foundation of the new
construction. Moreover, the problem with the settlements is
inevitable, so the serviceability of the construction is also an
important issue.

Thickness under 2 kPa load


Ultimate tensile strength

Longitudinal 4.0 kN/m


Transversal 7.5 kN/m

Strain
strength

at

ultimate

tensile

1300 N (-300 N)

Opening size O90


Water permeability
normal to the plane

5.1
Fondation slab 0.9 m

Figure 2.Adora building cross section

CONSTRUCTION DETAILS

The Adorabuildingis built on the sitewhich is very close to the


Ohrid Lake. The superstructure of the building has 5 floors and
the substructure has 1 floor (see Figure 2). The substructure is
extended out of the superstructure and it is actually a parking
lot. The size of the superstructure in plan view is 52.0x22.0 m,
and the total size of the building (including the extensions of the

Longitudinal 120% (40%)


Transversal 80% (40%)

CBR puncture resistance

Fondation slab 0.5 m

3 mm

0.10 mm (0.02 mm)


index

85 x 10-3 m/s
(-15 x 10-3 m/s)

Geotextile

At the bottom of the excavation pit, geotextile was used to


ensure separation of the fill material from the subsoil. By the
separation, it is meant that the geotextile will prevent mixing of
the different soils, but it will enable complete water
permeability, so with this, complete preservation of the
properties of the later placed fill materials will be ensured. Used
geotextile has the physical and mechanical properties, listed in
the Table 2.
The geotextile was placed all over the bottom of the
excavation pit as well as on the excavation slopes. The overlap
oftwo adjacent panels is 60 cm, and it completely wraps the fill
material up. Used geotextile with the properties given in the
Table 2 has the ability to withstand burst and puncture, and has
enough tensile strength to serve a separation function, without
being destroyed.

2470

Technical Committee 211 / Comit technique 211

5.2

Geogrids and fill material

For increasing the bearing capacity of the subsoil under the


foundation slab, geogrids are inserted as reinforcement. For
reinforcement three layers of geogrids with different properties
are placed in the fill material at different elevations.
The first layer of the geogrids was installed under the
middle section of the building where the contact pressure has
maximummean value up to P=100 kPa.The geogrids installed
on this position give the geotextile anadditional tensile strength,
as well as bursting and puncturing resistance.

bottom of the excavation pit up to the foundation slab.


Furthermore, this geogrid will ensure reaching of the requested
modulus of compressibility of the upperlayers of fill material.
The properties of the second layer of geogrid are shown in the
Table 4.
Table 4. Properties of the geogrid 80/30 (second layer of geogrids)
Raw material

Geogrid 20/20

PET

Coating

Polymer

Weight

~ 350 gr/m2

Ultimate tensile strength

Longitudinal 80 kN/m
Transversal 30 kN/m

Geogrid 80/30

Ultimate tensile strength at


3% strain

Longitudinal 22 kN/m

Ultimate tensile strength at


5% strain

Longitudinal 40 kN/m

Strain
strength

at

nominal

tensile

Transversal < 8.5%

Mesh size
Geotextile

Geogrid 40/40

Figure 4. Cross section of the soil improvement measures

On other hand this first layer provides initial stiffness of the


low lifts of the fill material. The overlap oftwo adjacent panels
is 50 cm and anchoring length of 3.0 m. The properties of the
first layer of geogrid are shown in the Table 3.
Table 3. Properties of the geogrid 40/40 (first layer of geogrids)
Raw material

PP

Coating

Polymer

Weight

~ 330 gr/m2

Ultimate tensile strength

Longitudinal 40 kN/m
Transversal 40 kN/m

Tensile strength at 2% strain

Longitudinal 16 kN/m
Transversal 16 kN/m

Tensile strength at 5% strain


Strain
strength

at

Mesh size

nominal

tensile

Longitudinal < 8.5%


20 x 20 mm

On top of the second geogrid two lifts of fill material are


done with thickness of 30 cm, total of 60 cm. The required
modulus of compressibility on top of these two layers of fill
material should be at least 100 MPa and minimum 98%
compaction after Proctor. The fill material is crushed stone base
aggregate.
On the extended parts of the building where the contact
pressure has maximummean value up to P=20 kPa, also a
geogrids is installed. This geogrid has an ultimate tensile
strength of 20 kN/m in both directions. The overlap of two
adjacent panels is 60 cm and the anchoring length is 1.0 m.
After installation of this geogrid the fill material is placed and
compacted in 30 cm thick lifts. The final layer of the fill
material at the extended parts of the building should reach at
least 80 kPa and minimum 98% compaction after Proctor.
After completion of the earth works ground improvement
measures were completely finished. So the works for hydro
insulation and constructing the structure commenced.
Table 5. Properties of the geogrid 40/40 (first layer of geogrids)
Raw material

Longitudinal 32 kN/m

PP

Transversal 32 kN/m

Coating

Polymer

Longitudinal < 7%

Weight

~ 190 gr/m2

Transversal < 7%

Ultimate tensile strength

40 x 40 mm

Over this geogrid a layer of drainage fill material was


placed with thickness of 50 cm. This material has particlessize
from 16 to 32 mm.
This layer of drainage fill material is foreseen to reduce the
possibility of liquefaction. So in case of earthquake, the
building up of the pore water pressure will be reduced by
draining the water from the subsoil layers into the drainage
layer placed with the subsoil replacement works.
Over this drainage fill material, another geogrid was placed
but this time with higher strength properties. The use of this
geogrid is to reinforce the fill material as well as to ensure
additional stiffness, which is gradually increasing from the

Longitudinal 20 kN/m
Transversal 20 kN/m

Tensile strength at 2% strain

Longitudinal 8 kN/m
Transversal 8 kN/m

Tensile strength at 5% strain

Longitudinal 18 kN/m
Transversal 18 kN/m

Strain
strength

2471

at

Mesh size

nominal

tensile

Longitudinal < 7%
Transversal < 7%
40 x 40 mm

Proceedings of the 18th International Conference on Soil Mechanics


and
Geotechnical
Paris 2013
Figure 6. Plaxis
model
of the AdoraEngineering,
building

Figure 5. Axis forces in geogrid 80/30 (first layer of geogrids)


Figure 5. Axis forces in geogrid 80/30 (first layer of geogrids)

The Plaxis model was subjected to several load cases which


involves geostatic, hydrostatic and dynamic load. The analyses
are conducted with and without applied geosyntetics under the
7 CONCLUSION
construction. Comparing the results of analyses of both models
CONCLUSION
it7 was
obvious that improvement of the subsoil using
geosyntetics is fully justified.
The maximum total settlements of the subsoil after the
construction of the hotel are estimated at 30 cm. The estimated
differential settlements during the seismic analyses were 0.1 cm.
The maximum axial forces developed in the geogrids for
7 geostatics
CONCLUSION
load case are 13,86 kN/m in the middle geogrid and
10,14 kN/m in the lowest geogrids. When the model is
subjected to dynamic loads (seismic activity) the axial force in
the middle geogrid 35,17 kN/m and in the lowest geogrid the
axial force is 19,12 kN/m.
Additional analyses were carried out in order to estimate a
liquefaction potential of the subsoil. These analyses were
necessary due to the presence of saturated, uniform sand in the
subsoil which has a relative density in the range of Dr=15-40%.
The results showed that the subsoil has a liquefaction potential.

In the last 2,5 years this is the second bigger project of soil
In 7.
theusing
last stresses
2,5
years
this in
is the second
biggerThe
project
improvement
geosyntetics
Ohrid area.
first,of soil
Figure
Shear
distribution
6 GEOTECHNICAL ANALYSES
improvement
using
the Ohrid
area.Hotel
The first,
very similar,
case was
soilgeosyntetics
improvementin under
the new
6 GEOTECHNICAL ANALYSES
very
similar,
case
was
soil
improvement
under
the
new
Park
2,5
years
ago.
The
two
buildings
are
approximately
2.0
km Hotel
In order to adopt the measures for improvement of the subsoil,
yearsof
ago.a The
two datafrom
buildings are
approximately
On 2,5
a basis
obtain
monitoring
of the2.0 km
In analyses
order to were
adopt conducted.
the measures
for improvement
of the
subsoil,apart. Park
detailed
These
analyses involve
flow
On a basis
obtain
datafromthat
monitoring
settlements
it of
cana be
concluded
the totalof the
7 apart.
CONCLUSION
detailed analyses
conducted. These
analysesdue
involve
net analyses,
analyses were
of soil-structure
interaction
to flowbuildings
Figure
Axis dynamic
forcesanalyses
in geogrid
80/30
(first layer
of geogrids)
buildings
settlements are
it can
be theconcluded
that the total
of bothbuildings
below
initially estimated
net 5.and
analyses,
interaction
geostatic
load of
as soil-structure
well
as analyses
of thedue tosettlements
settlements
of
bothbuildings
are
below
the
initially
estimated
settlements.
In
the
last
2,5
years
this
is
the
second
bigger
project
of soil
geostatic
and dynamic
load as
as model
analyses
liquefaction
potential
of the subsoil.
Thewell
ground
wasof the
settlements.
For
example, for
monitoring
of theinsettlements
the The
Hotelfirst,
improvement
using
geosyntetics
the Ohrid of
area.
liquefaction
potential
of
the
subsoil.
The
ground
model
was
developed
according
to
the
adopted
soil
properties
shown
in
the
6 GEOTECHNICAL ANALYSES
For example,
for monitoring
of Those
the settlements
of the
Hotel
8 survey
points
were
positioned.
points
similar,
case was
soil
improvement
under
the initially
new
Hotel
developed
according
to the adopted
properties
shown in theParkvery
Table 1,and
also the
exact geometry
of the soil
structure
was applied
Park
survey
were
positioned.
Those
points2.0initially
measured
after
placing
them
(during
the
fundaments
Park
2,5 8years
ago.points
The
two
buildings
are approximately
km
Table to
1,and
also
themeasures
exact geometry
of theout
structure
to the
Most
ofthe
the
analyses
were
carried
inofPlaxis
2Dappliedwere
In model.
order
adopt
for improvement
thewas
subsoil,
wereOnmeasured
after
them (during
the fundaments
later
in the
of constructing
the
structure
and
apart.
a basis
of time
a placing
obtain
datafrom
monitoring
of the
to the
model. Most
ofconducted.
the analyses
wereanalyses
carried out
in Plaxis
software
suitusingfinite
element
method.
detailed
analyses
were
These
involve
flow2Dconstructing),
constructing),
laterwhole
initthecan
time
constructing
the structure
finishing
of the
construction.
The that
first
initialtotaland
settlements
beof concluded
the
suitusingfinite
method. interaction due to afterbuildings
netsoftware
analyses,
analyses ofelement
soil-structure
after finishing
the whole
construction.
The estimated
first
of
the ofelevation
the fixed
points
was initial
settlements
of bothbuildings
areof below
the initially
geostatic and dynamic load as well as analyses of the measurement
measurement
of the
fixed points was
on 11.07.2011,
and elevation
the last oneofon the
23.03.2012.
settlements.
liquefaction potential of the subsoil. The ground model was undertaken
11.07.2011,
andofthe
one on 23.03.2012.
In undertaken
both
cases on
cost-benefit
analyses
conducted
during
For example,
for
monitoring
thelast
settlements
of thethe
Hotel
developed according to the adopted soil properties shown in the
both cases
cost-benefit
analyses
conducted
during
process
showed
soil
improvement
Park 8In survey
points
werethat
positioned.
Those
points using
initiallythe
Table 1,and also the exact geometry of the structure was applied designing
designing
process
showedmethod
that (during
using
is most
economical
insoil
suchimprovement
geotechnical
were
measured
after
placing
them
the
fundaments
to the model. Most of the analyses were carried out in Plaxis 2D geosyntetics
geosyntetics
is most
economical
method
such
geotechnical
conditions.
Regardless
aspect
thisinmethod
has aand
constructing),
later
inthetheeconomic
time of constructing
the
structure
software suitusingfinite element method.
conditions.
Regardless
theshort
economic
this first
method
has a
major
advantage
which
period aspect
of installation.
after
finishing
of is
thea very
whole
construction.
The
initial
major advantage
which
is a very of
shortthe
period
of installation.
measurement
of the
elevation
fixed
points was
8 undertaken
REFFERENCES
on 11.07.2011, and the last one on 23.03.2012.
Figure 5. Axis forces in geogrid 80/30 (first layer of geogrids)
8In both
REFFERENCES
cases cost-benefit analyses conducted during the
Brinkgreve.
R. last
&process
Vermeer
P.
1998.
Plaxis
Code
for Soil
In the
2,5 years
this is
the
second
bigger
project
ofusing
soil
designing
showed
thatFinite
soilElement
improvement
andBrinkgreve.
Rock Analyses.
R.
&
Vermeer
P.
1998.
Plaxis
Finite
Element
Codefirst,
for Soil
improvement
using
geosyntetics
in
the
Ohrid
area.
The
geosyntetics is most economical method in such geotechnical
6 GEOTECHNICAL ANALYSES
Geing-KuK.
2011.
Geomechanical
report
for construction
of residential
and
Rock
Analyses.
very similar,
case
was the
soileconomic
improvement
under
new has
Hotela
conditions.
Regardless
aspect
this the
method
building
inyears
Ohrid.
SkopjeGeing
Kuk. report
Geing-KuK.
2011.
Geomechanical
forapproximately
construction of 2.0
residential
Park
2,5
ago.
The
two
buildings
are
km
major advantage
which
is
a
very
short
period
of
installation.
In order to adopt the measures for improvement of the subsoil, Geing-KuK.
2011.Project
on
improvement
of
the
foundation
soil
under
building
in Ohrid.
SkopjeGeing
Kuk.
apart.
On
a
basis
of
a
obtain
datafrom
monitoring
of
the
detailed analyses were conducted. These analyses involve flow
theGeing-KuK.
residential building
in Ohrid.
SkopjeGeing of
Kuk.
2011.Project
on improvement
the foundation soil under
settlements
it canOhrid.
be
concluded Kuk.
that edition.
the total
8buildings
REFFERENCES
net analyses, analyses of soil-structure interaction due to Robert
M.Koerner.1997.
Geosynthetics.
the residentialDesigning
building inwith
SkopjeGeingFourth
settlements
of Inc.
bothbuildings
are below
the initiallyFourth
estimated
Prentence-Hall,
geostatic
load
as well as analyses of the
Robert M.Koerner.1997.
Designing
with Geosynthetics.
edition.
Figure
6. Plaxis and
modeldynamic
of the Adora
building
Brinkgreve.
R. & Vermeer
settlements.
Prentence-Hall,
Inc.P. 1998. Plaxis Finite Element Code for Soil
liquefaction
potential
theAdora
subsoil.
The ground model was
Figure 6. Plaxis
model of
of the
building
and example,
Rock Analyses.
The
Plaxis according
model wastosubjected
to several
load cases
which
For
for monitoring of the settlements of the Hotel
developed
the adopted
soil properties
shown
in the
Geing-KuK.
2011.points
Geomechanical
report for construction
of residential
The
Plaxis
model
was
subjected
to
several
load
cases
which
involves
geostatic,
hydrostatic
and
dynamic
load.
The
analyses
Park
8
survey
were positioned.
Those points
initially
Table 1,and also the exact geometry of the structure was applied
building
in
Ohrid.
SkopjeGeing
Kuk.
involves
geostatic,
hydrostatic
and
dynamic
load.
The
analyses
are to
conducted
with
and
without
applied
geosyntetics
under
the
were measured after placing them (during the fundaments
the model. Most of the analyses were carried out in Plaxis 2D
2011.Project on improvement of the foundation soil under
construction.
Comparing
results
of applied
analysesgeosyntetics
of both models
are conducted
withthe
and
without
under the Geing-KuK.
constructing),
later in the time of constructing the structure and
software
suitusingfinite
element
method.
the residential building in Ohrid. SkopjeGeing Kuk.
Comparing
the results
analyses
of both
it wasconstruction.
obvious that
improvement
of ofthe
subsoil
usingmodels Robert
after finishing
of theDesigning
whole with
construction.
TheFourth
first edition.
initial
M.Koerner.1997.
Geosynthetics.
it wasis fully
obvious
that improvement of the subsoil using measurement
geosyntetics
justified.
of Inc.
the elevation of the fixed points was
Prentence-Hall,
Figure 6. Plaxis model of the Adora building
Thegeosyntetics
maximum istotal
fullysettlements
justified. of the subsoil after the
undertaken on 11.07.2011, and the last one on 23.03.2012.
total
settlements
of
subsoil
construction
ofmaximum
themodel
hotel was
are
estimated
at
cm.the
The
estimated
TheThe
Plaxis
subjected
to 30
several
load
cases after
whichthe
In both cases cost-benefit analyses conducted during the
construction
of the
hotelthe
areseismic
estimated
at 30
cm.The
The
estimated designing process showed that soil improvement using
differential
settlements
during
analyses
were
0.1analyses
cm.
involves
geostatic,
hydrostatic
and
dynamic
load.
Thearemaximum
developed
ingeosyntetics
theanalyses
geogrids
for0.1
differential
settlements
during
the seismic
were
conductedaxial
with forces
and without
applied
under
thecm. geosyntetics is most economical method in such geotechnical
The load
maximum
forces
developed
in geogrid
theboth
geogrids
geostatics
case
areaxial
13,86
kN/m
in of
theanalyses
middle
and for conditions. Regardless the economic aspect this method has a
construction.
Comparing
the
results
of
models
are
13,86
kN/mWhen
in the
middle
geogrid
10,14
kN/mobvious
in load
the case
lowest
geogrids.
model
is and major advantage which is a very short period of installation.
it geostatics
was
that
improvement
of
thethe
subsoil
using
subjected
to dynamic
axial the
forcemodel
in
10,14
kN/m
inloads
the (seismic
lowest activity)
geogrids.theWhen
is
geosyntetics
is fully
justified.
subjected
to dynamic
loads
(seismic
the axial
force
the middle
35,17
and
in the
lowest
geogrid
the
The geogrid
maximum
totalkN/m
settlements
ofactivity)
the subsoil
after
the in 8 REFFERENCES
the middle
35,17
kN/m andatin30the
geogrid the
axial
force
is 19,12
kN/m.
construction
ofgeogrid
the
hotel
are estimated
cm.lowest
The estimated
Brinkgreve. R. & Vermeer P. 1998. Plaxis Finite Element Code for Soil
analyses
Additional
were
carried
out in order
to estimate
axial force
is 19,12
kN/m.
differential
settlements
during
the seismic
analyses
were 0.1acm.
and Rock Analyses.
analyses
Additional
carried
outininanalyses
order
to were
estimate
liquefaction
potential
of forces
thewere
subsoil.
These
The maximum
axial
developed
the
geogrids
for a
liquefaction
potential
of
subsoil.
Thesesand
analyses
were Geing-KuK. 2011. Geomechanical report for construction of residential
necessary
due load
to thecase
presence
of the
saturated,
uniform
in theand
geostatics
are 13,86
kN/m
in the
middle
geogrid
building in Ohrid. SkopjeGeing Kuk.
subsoil
which
hasdue
a relative
densitygeogrids.
in
range
ofuniform
Dr=15-40%.
necessary
to
presence
ofthe
saturated,
sand inisthe Geing-KuK. 2011.Project on improvement of the foundation soil under
10,14
kN/m
in
thethelowest
When
the model
subsoil
which
hasthe
aloads
relative
density
in the range
of Dr=15-40%.
Thesubjected
results
showed
that
subsoil
has a activity)
liquefaction
to
dynamic
(seismic
the potential.
axial
force in
the residential building in Ohrid. SkopjeGeing Kuk.
results
showed
that kN/m
the subsoil
hasthe
a liquefaction
potential.
theThe
middle
geogrid
35,17
and in
lowest geogrid
the
Robert M.Koerner.1997. Designing with Geosynthetics. Fourth edition.
axial
is 19,12
Prentence-Hall, Inc.
Figureforce
6. Plaxis
modelkN/m.
of the Adora building
Figure 7.
Shear stresses
distribution
analyses
Additional
were carried out in order to estimate a
Figure
7. Shear
stresses
distribution
The Plaxis
model
was
subjected
to several
cases which
liquefaction
potential
of
the subsoil.
Theseload
analyses
were
involves geostatic,
dynamicuniform
load. The
analyses
necessary
due to thehydrostatic
presence ofandsaturated,
sand
in the
are conducted
witha relative
and without
applied
under the 2472
subsoil
which has
density
in thegeosyntetics
range of Dr=15-40%.
construction.
Comparing
thesubsoil
resultshas
of aanalyses
of both
models
The
results showed
that the
liquefaction
potential.
it was obvious that improvement of the subsoil using

buildin
Geing-KuK
the resi
Robert M.K
Prenten

Geoencased columns: toward a displacement based design


Colonnes renforce par gotextiles: vers une conception base sur le dplacement
Galli A., Prisco di C.
Politecnico di Milano

ABSTRACT: As is largely testified by the scientific literature, in the last decade geoencased columns have become a quite common
alternative solution to standard stone columns. This is essentially due to the possibility of employing reinforcements to better the
mechanical response of the inclusions without reducing their drainage efficiency. Although GEC are often used to reduce settlements
induced by the construction of large embankments on soft soils, up to now a rational displacement based design approach has not yet
been introduced. This is thus the final objective of this paper, that, by starting from a critical review of the standards presently
available, will illustrate the results of a series of finite difference numerical analyses. The unit cell of an ideal reinforced soil
embankment placed on a soft soil stratum will be accounted for and the effect of the main geometrical/mechanical parameters, as well
as the response of the system during the construction stages, is discussed.
RSUM : Comme il est largement connu dans la littrature de ces vingt dernires annes, les colonnes en matriaux granulaires
renforce par gotextiles (GEC) sont devenues une solution trs utilise par rapport aux colonnes ballastes standard. Cela est
essentiellement d la possibilit d'employer des renforts pour amliorer la rponse mcanique des inclusions sans rduire leur
efficacit de drainage. Bien que les GEC soient souvent utilises pour rduire les tassements induits par la construction de remblais
importants sur sols mous, une approche rationnelle de conception base sur le dplacement n'a, jusqu' prsent, pas encore t mise en
place. Cela est donc l'objectif final de cette tude, qui, en partant d'une analyse critique des normes actuellement disponibles, illustrera
les rsultats danalyses numriques aux diffrences finies. Une cellule lmentaire d'un remblai idal de sol renforc plac sur un sol
mou sera prise en compte et l'effet des principaux paramtres gomtriques et mcaniques et la rponse du systme au cours des
diffrentes tapes de la construction seront discuts.
KEYWORDS: geoencased granular columns, geotextiles, numerical analyses, displacement based design, earth reinforced structures.
1

INTRODUCTION.

As it is well documented in the literature (see e.g. Raithel et


al. 2005), since mid-nineties the use of geoencased granular
columns (GEC) as foundations of earth structures on soft and
very soft soils has been progressively increased. GECs have
both mechanical and hydraulic functions: they work not only as
reinforcement inclusions, capable of preventing the global
collapse of the foundation and reducing differential settlements
within the structure, but they work additionally as vertical
drains, thus reducing the consolidation time of the soft soil.

Figure 1. sketch of an earth embankment on GECs.

The GEC foundation system is composed of an array of


granular columns of length L and diameter D, placed at a
regular spacing S below an embankment of height H (Figure 1).
The columns are encased by a geotextile with the double aim of
reinforcing the column and filtering to prevent the clogging of
the column itself. At the base of the embankment, to redistribute
vertical stresses, several layers of geotextile are also inserted
during the construction. The effectiveness of this foundation
system has been clearly proved both on real scale data (see e.g.

Kempfert 2003, where the response of the system is analyzed


by varying the spacing among columns and of the stiffness of
the encasing geotextile), and by means of numerical and
experimental researches (Murugesan and Rajagopal 2006, di
Prisco and Galli 2011). The fundamentals of the mechanical
behavior of the system is therefore quite well understood.
Nevertheless, common design standards are still based on too
simplified approaches, unable of capturing the actual
mechanical complexity of the system in particular, the
interaction between embankment and columns (this point will
be tackled in further details in the following section by critically
reviewing the most used design standards). Conversely to
traditional deep foundation systems (like reinforced concrete
piles or jet-grouted columns which can be considered axially
rigid with respect to the surrounding soil), GECs are axially
deformable inclusions, whose axial deformability is strictly
coupled with the stiffness of the surrounding soil. Moreover,
since this latter is very often characterized by a very low
permeability and high deformability, its mechanical response
should be modeled by properly taking into account the hydromechanical coupling (for the sake ofbrevity, this aspect will
however be disregarded in the following).
In the present paper the attention will be initially focused on
a critical review of the most common design standards. Then an
engineering displacement based approach will be briefly
introduced, and some numerical analyses, with the particular
aim of studying the distribution of differential settlement at the
top of the embankment, which are generally neglected by the
design approaches, will be presented.

2473

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

REVIEW OF DESIGN STANDARDS

The two most common design standards available for the design
of earth embankment on GECs are the British Standard BS
8006 (1995) and the German Standard EBGEO (Chapter 6.9;
2003). Both of them assume the column to be rigid and,
estimate the vertical stress distribution at the base of the
embankment to be independent of the mechanical interaction
with the foundation (i.e. the GEC and the soft soil). From an
engineering point of view, however, the vertical stress
redistribution at the base of the embankment is the main
parameter governing (i) the design of the reinforcement layers
(see Figure 1) and (ii) the evaluation of the differential
settlements.
According to BS8006, the vertical stress redistribution does
not depend on the mechanical properties of the embankment. In
particular, the average vertical stress acting at the columns top
(in the following this quantity will be called i) is determined
only as a function of the geometry (H, S, D), as was suggested
by the approach proposed by Martson (1913) for buried pipes
(see also Jones et al., 1990). The estimation of the average
vertical stress e acting on the soft soil is instead obtained by
means of empirical expressions, depending on the full or partial
formation of the arch effect, as a function of the ratio between
height H and difference S-D. According to EBGEO, on the
contrary, a rather complex analytical procedure, based on the
work proposed by Zaeske (2001), is employed to describe the
arch effect. This takes into account the geometry (H, S, D) and
the friction angle of the granular material constituting the
embankment, and imposes the equilibrium of one central slice
of a vault shell of the arch that it is supposed to develop within
the embankment. No estimation of the vertical stress i at the
top of the column is provided. For the sake of brevity, the
analytical expressions have not been reported here; for further
details, see BS8006 and EBGEO (Chapter 6.9).
As far as the evaluation of settlements is concerned, the
procedure prescribed by EBGEO follows the work proposed by
Ghionna and Jamiolkowski (1981) and consists in subdividing
the length L of the column in slices (each one of them is then
assimilated to an axisymmetric triaxial soil sample). The
following hypotheses are assumed: (i) the granular soil in the
column is at critical state (i.e. no changes in volume are possible
for the column), (ii) no relative settlement are considered
between the column and the soil. These two hypotheses
introduce very strong simplifications that can lead to unphysical
results. The second one, in particular, makes impossible the
superficial differential settlements to be estimated.
2.1

Parametrical analyses

In this section parametrical analyses on the values of e


obtained by employing BS8006, as well as some results
concerning the evaluation of the settlement and of the tensile
force in the encasing geo-membrane computed according to
EBGEO, are presented. In particular, the effect of the
embankment height H and of the material friction angle is
investigated for increasing values of the relative spacing S/D,
and by taking into account several diameters D of the column
(the authors are aware of the fact that some values of D and S/D
considered are unrealistic, nevertheless they have been chosen
in order to test even the asymptotic trend of the design
approaches).
2.1.1 Stress on the soft soil at the base of the embankment
Figure 2 shows the values of e computed according to BS8006,
and highlights that unphysical results of e<0 are obtained for
low values of the relative spacing S/D, independently of the
embankment height H. This result could in general lead to an
overestimation of the arch effect and thus to an unsafe design of
the georeinforcement layers at the base of the embankment.
The arch effect tends to vanish for increasing values of S/D,
and the value of e tends to the weight H of the embankment.

The corresponding values of i computed according to BS8006


(not reported here for the sake of brevity) are independent of
S/D, and only slightly dependent on H.

a)

b)

Figure 2. Evaluation of the stress e according to BS8006: (a) H=2.5m


and (b) H=10m.

It can be easily demonstrated that e and i (if computed


according to BS8006) do not even satisfy the total equilibrium
along the vertical direction with respect to the weight of the
embankment, and that the values of the tensile force in the
geosynthetic layers at the base of the embankment computed
according to BS8006 are not continuous with increasing H
(Moraci and Gioffr 2010).
2.1.2 Settlements and tensile force in the encasing membrane
With reference to the values of the mechanical parameters
listed in Table 1 (taken from an example of application
proposed by EBGEO), in this paragraph a parametrical analyses
on the values of the settlement is presented, for increasing
values of the embankment weight H and by taking into account
several values of stiffness J of the encasing geomembrane. The
values of the column length L and of the relative spacing S/D
are here considered to be constant and equal to 10 m and 2,
respectively (with D=80 cm and S=1.6m).
Table 1. Values of the mechanical properties of the materials considered
in the analyses.

Unit weight (kN/m3)


Friction Angle ()
Cohesion (kPa)
Young modulus (kPa), at a
reference pressure of 100kPa
Poisson coefficient (-)

Embankment

Column

20

19

Soft
soil
15

35
-

15
10

750

0.4

As it is evident from Figure 3a (where, for the sake of


generality, the value of s has been normalized with respect to
L), the presence of the encasing geomembrane induces a
stiffening effect of the foundation system, thus reducing the
expected value of the total settlement (which is considered,
according to the adopted hypotheses, to be uniform and
coincident with the settlement s at the top of the embankment).
The numerical procedure, however, for low values of H (i.e.
shallow or light embankments) leads to unrealistic results, for

2474

Technical Committee 211 / Comit technique 211

which negative settlements (i.e. uplift) are obtained. This


meaningless result is obtained even for a nil value of the
stiffness J of the encasing geomembrane. This essentially
derives from the assumption concerning the soil within the
column which is imposed to be at the critical state along the
entire column (similar results have been observed even for
other values of L and S/D, but they have not been reported here
for the sake of brevity). This is evident when the tensile force Tg
in the encasing geomembrane along depth z of the column is
considered (Figure 3b, where the case of a shallow embankment
is analyzed): at the base of the column the vertical stress is not
sufficient to induce an active state of stress, and the only
possibility for the column to satisfy the hypothesis of critical
state is to reduce its radius, thus inducing a compression (i.e.
Tg<0) in the encasing geomembrane.

activated at GEC-soil interface, and (iii) differential settlements


are expected even at the embankment top.

a)

b)

Figure 4. Mechanical response of the foundation system in case of (a)


rigid and (b) deformable embankment.

From a modeling point of view, by assuming an engineering


approach based on generalized variables, the mechanical
behavior of the embankment can be described by means of a
generalized constitutive relationship between the average
stresses at the base of the embankment and the differential
settlements (assumed to be uniform) between the column and
the soil:

i e f us uc ,

a)

(1)

where the values of i and e must satisfy the equilibrium


with respect to the weight of the embankment on the unit cell

i D2 e S2 D2 H S2 .
b)

Figure 3. EBGEO: evaluation of (a) settlements and (b) tensile force in


the encasing geomembrane.

A DISPLACEMENT BASED DESIGN APPROACH

In order to overcome the above cited limitations, a consistent


and physically based design would require a fully displacement
based approach. As was theoretically outlined by Galli and di
Prisco (2011), with reference to a single axisymmetric cell (i.e.
to a single column together with the surrounding soft soil), the
foundation system can be assumed to be composed by two
coupled springs, one representing the GEC and the second
representing the surrounding soft soil. The two springs work in
parallel if and only if the base of the embankment can be
considered to be rigid and no differential settlements to arise
(Figure 4a). Under this hypothesis, the vertical stress at the base
of the embankment is thus uniformly distributed (in Figure 4a
stands for the unit weight of the granular material constituting
the embankment), no differential settlement are observed at the
top of the embankment, and no shear stresses develop at GECsoil interface. The values of vertical stress both in the column
and in the soil then can be assumed to depend exclusively on the
axial stiffness of the column (KGEC) and on the vertical
compressibility of the soft soil (represented in Figure 4a by a
global stiffness KS).
Real embankments, however, are characterized by a
deformable base (Figure 4b), and different values of settlement
are expected for the top of the column (uc) and for the soil (us)
at the base of the embankment. Consequently: (i) vertical
stresses are redistributed at the base of the embankment
between the internal zone of the cell (above the column,
characterized by an average stress i) and the external one (a
circular crown above the soil, characterized by an average stress
e) due to the so called arch effect, (ii) shear stresses are

(2)

The constitutive relationship f can be in general assumed to


be described by means of a non-linear curve, whose average
stiffness depends (i) on the geometry of the system (S, D, H),
(ii) on the mechanical properties of the granular material
constituting the embankment and (iii) on the geo-reinforcements
at the base of the embankment. Its limit value corresponds
instead to the activation of a failure mechanism within the
embankment. Depending on the formation of the arch effect,
either a punching failure mechanism, or a dome failure
mechanism, with no (or very limited) superficial differential
settlements, might develop (Figure 5a-b).

a)

b)

Figure 5. Failure within the embankment: (a) punching mechanism and


(b) domed mechanism due to arch effect.

The pattern of superficial differential settlement s=s(r) could


then be formally described by a transfer function, ranging from
a discontinuous function (in case of punching), to a smooth
function (in case of formation of the arch effect).
4

NUMERICAL ANALYSES ON SETTLEMENT PROFILE

In order to investigate the settlement distribution s(r) at the top


of the embankment for increasing values of H, some
preliminary finite difference numerical analyses have been
performed by means of the commercial code FLAC. An
axisymmetric geometry has been chosen in order to model the
cell, and the simplifying hypothesis of rigid column has been

2475

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

assumed; the mechanical behavior of both the materials


constituting the embankment and the soft soil has been modeled
by assuming an elastic perfectly plastic relationship with a nonassociated Mohr-Coulomb failure condition. For the
embankment, two different types of material have been
considered: a loose sand (=20, with no dilatancy) and a
compacted sand (=40, with dilatancy =10). In both cases,
for the sake of simplicity the unit weight is assumed to be
equal to 20 kN/m3, the Young modulus equal to 3 MPa and the
Poisson coefficient equal to 0.25. No friction has been
considered at soil-column interface. For the soft soil, for the
sake of simplicity a dry condition has been assumed (i.e. no
hydro-mechanical coupling has been modeled). The values of
the mechanical parameters are listed in Table 1. Consistently
with the parametrical analyses previously discussed, the length
L of the column is 10 m and its diameter D is 80 cm. Two ratios
S/D have been considered, and the settlement distribution s(r)
has been normalized at each stage for the current value of H.

failure, whilst an increase in and tends to smoothen the


settlement profile.
5

CONCLUSIONS

The paper critically discussed some results obtained according


to the usual Design Standards, and proved that in some cases
these approaches lead to unrealistic results. The codes,
moreover, disregard the estimation of relative settlements at the
top of the embankment, which is actually one of the most
important parameters describing the efficiency of the
foundation. A consistent, displacement based conceptual
framework for describing the behavior of the system has been
formulated, and some preliminary numerical analyses have been
shown. These latter, in particular, showed on the contrary that
the top settlement profile is remarkably affected by both the
geometry and the mechanical properties of the embankment.
6

ACKNOWLEDGEMENTS

The Authors want to acknowledge TENCATE and ITASCA


Italy for financially supporting the research.
7
a)

b)

Figure 6. Normalized superficial settlements for S/D=2: (a) loose and


(b) dense material.

a)

b)

Figure 7. Normalized superficial settlements for S/D=4: (a) loose and


(b) dense material.

Figures 6 and 7 describe the evolution of s(r) at the top of the


embankment during the construction stages. It appears clearly
that the settlement profile ranges from a well localized punching
failure mechanism to a smooth distribution of settlements for
increasing H (witnessing the progressive mobilization of the
arch effect). The influence of relative spacing S/D and of the
mechanical properties of the embankment (in terms of both
and ) is opposite: an increase in S/D tends to localize the

2476

REFERENCES

British Standard BS 8006. 1995. Code of practice for


strengthened/reinforced soils and other fills. British Standards
Institution, London, UK pp.176.
di Prisco C. and Galli A. 2011. Mechanical behaviour of geo-encased
sand columns: small scale experimental tests and numerical
modeling. Geomechanics and Geoengineering: An International
Journal 6(4), 251263
EBGEO. 2003. Empfehlung 6.9 (2003). Bewehrte Erdkrper auf
punktoder linienfrmigen Traggliedern, Kapitel 6.9 fr die
Empfehlungen fr Bewehrungen aus Geokunststoffen, EBGEO,
DGGT (German Geotechnical Society).
Galli A. and di Prisco C. 2011. Un modello concettuale per la
progettazione di colonne granulari georinforzate a fondazione di
rilevati artificiali (in Italian). XXIV Convegno Nazionale di
Geotecnica, 231 246.
Ghionna V.N. and Jamiolkowski M. 1981. Colonne di ghiaia (in
Italian). X Ciclo di conferenze dedicate ai problemi di meccanica
dei terreni e ingegneria delle fondazioni: Metodi di miglioramento
dei terreni. Politecnico di Torino, Atti dellIstituto di Scienza delle
Costruzioni, n.507, pp.1-63.
Jones C.J.F.P., Lawson C.R., Ayres D.J. 1990. Geotextile reinforced
piled embankments. Geotextiles Geomembranes and Related
Products, Den Hoedt (ed.) 1990 Balkema, Rotterdam, ISBN 90
6191 119 2, pp 155-160.
Kempfert H.G. 2003. Ground improvement methods with special
emphasis on column-type techniques. Int. Workshop on
Geotechnics of Soft Soils-Theory and Practice- SCMEP.
Marston A. and Anderson A.O. 1913. The theory of load on pipes
ditches and tests of cement and clay drain tile and sewer pipes.
Bulletin 31. Iowa Engineering Experiment Station, Iowa State
College, Ames, Iowa.
Moraci N. and Gioffr D. 2010. La progettazione di rilevati su terreni
compressibili rinforzati con geosintetici (in Italian). Rivista Italiana
di Geotecnica, vol. 3/10, 67-100.
Murugesan, S. and Rajagopal, K. 2006. Geosynthetic encased stone
columns: Numerical Evaluation. Geotextiles and Geomembranes,
Vol. 24, 349-358.
Raithel M., Kirchner A., Schade C. and Leusink E. 2005. Foundation of
Constructions on Very Soft Soils with Geotextile Encased Columns
- State of the Art. Proceedings ASCE Geo-Frontiers 2005.
Zaeske D. 2001. Zur Wirkungsweise von unbewehrten und bewehrten
mineralischen
Tragschichten
ber
pfahlartigen
Grndungselementen, Schriftenreihe Geotechnik. Universitt
Kassel, Heft 10.

Design prediction of the strengthened foundation base deformation by field tests


data
La prvision de calcul des dformations de la base des fondements reports partir des
recherches prises en nature
Gotman A., Gotman N.

BashNIIstroy, Ufa, Russia

ABSTRACT: The paper presents the solution of the complicated practical geotechnical problem of the skeleton structure foundation
strengthening. The strengthening was done due to change of the spatial arrangement of a building and essential load increase. The
experience of a foundation strengthening with jet grouted piles is described based on results of the base deformations monitoring. The
main design principles of the foundations under strengthening are given. The results of the deformation design prediction based on jet
grouted piles test and the base deformation measuring are presented.
RSUM : Dans cet article on prsente la solution dun problme gotechnique pratique compliqu du renforcement des souches
dun btiment en carcasse construit sur les sols de fondation dangereux cause du karst, en raison du changement lors de la
construction de la conception de plan et de volume du btiment et de laccroissement signifiant de charge. Lexprience est dcrite du
renforcement des souches par des papillons dinjection de forage sur la base des rsultats du monitoring des dformations des sols de
fondation. Les rsultats sont prsents des pronostiques prvisionnelles des sols de fondation des souches renforces la base des
essais des papillons dinjection de forage et de la mesure des dformations du sol de fondation.

KEYWORDS: foundation strengthening, settlements, pile vertical load test


1

INTRODUCTION

Design prediction of the strengthened foundation base


deformation by field tests data was executed for the new
shopping centre located in Ufa (Russia). The 500x250m
shopping center was designed as a skeleton one-storey building
with column spacing 16x8 m. The building construction was
started in April, 2007, then restarted in June, 2009 and finished
in 2010. Since May, 2008 till August, 2009 the construction at
site has been not performed. In 20072008 the foundations and
the most part of the bearing structures have been constructed.
During construction time, the building part was changed (by
investors demand). The column spacing was increased (8x16m
to 16x16m or 12x16m) and foundation loads to the moment of
construction stoppage 1,21,5 times increased the design ones.
After construction restarting, other changes of building frame
design have taken place. At the significant area the number of
stories and floor loads have been increased. As a result, all these
changes provoked 3070% increase of the foundation loads
and the further foundation strengthening.
2

ENGINEERING-GEOLOGICAL CONDITIONS

Under the foundation base stiff clays, tough and soft loams
occur underlain with water saturated medium coarse sand and
gravel at the depth of 810 m. (table 3). Maximum predicted
ground water level is 2 m under the foundation base. The
building site applies to the third category of stability about karst
deformations and is divided into sections according to extent of
their risk in accordance with Russian Codes. At this site, areas
are located that are classified according to their karst risk as
potentially not dangerous and potentially dangerous (fig.1) with
the probable design diameter of karst hole 7 m.
Due to site severe engineering-geological conditions
characterized with lack of homogeneity and karst risk, the
following foundations were designed: post- and strip
foundations on the bed; piled foundations with the in-situ raft
(pile groups); strip foundations on the bed and piled foundations

with the strip in-situ raft reinforced considering a karst hole


formation.
Severe engineering-geological conditions of site demanded
foundation settlement observation and expert investigation of
construction.
3 TECHNICAL EVALUATION OF THE STRUCTURAL
CONCEPT ON FOUNDATION STRENGTHENING
When design working out and the way of the foundations
strengthening selecting, the following was taken into attention.
1. To the moment of the strengthening design development,
the building was 1,5 years. In axes 115 the bearing structures
were completely constructed and at the rest part of the area
foundations, columns and floors of the ground floor were built.
2. According to monitoring results (table 1), it was stated
that to the moment of strengthening design development
(August 2009), the settlements on the whole were stabilized.
The settlements of the column foundations with the loads
exceeding the design ones already in the process of construction
were 610cm; the settlements of the rest foundations were
34 cm. The settlements data were used to evaluate the
coefficients of subgrade reaction of strip- and post foundations
bases that demanded strengthening (fig.2, table 2).
Taking into account that the significant part of the structures
was constructed and more than half of the base loads have
already been transferred, when selecting the method of the
foundation strengthening the minimum digging out and
dismantling (drilling, cutting, etc.) of the existing foundation
should have been provided. Method of strengthening with jet
grouted piles was selected with loads transfer from the building
through the connection of the reinforced concrete column with
the in-situ raft rested on piles.

2477

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Mark number (fig.1)

Table 1
Dates of measurement (days) and settlements (mm) from the moment of last measuring
Settleme 24.
nt for the 11.
previous 2007
period,

20.

10.

07.

02.

15.

16.

20.

16.

16.

02.

04.

05.

09.

04.

11.

01.

05.

06.

2008

2008

2008

2008

2009

2009

2010

2010

2010

Settleme
nt for Strengthe Foundation
07.
958 days ning
type
2010
of
observati
23
on, mm
09.

mm

88

50

27

118

225

217

63

116

31

1,8

8,0

3,7

0,5

2,1

16,1

post

14

2,6

5,8

0,9

9,7

0,5

1,2

20,7

strip

26

1,9

2,5

2,1

0,7

0,9

9,1

strip

27

5,4

14,0

6,1

5,9

9,2

21,5

17,0

3,3

83,4

strip

29

6,2

11,7

5,2

2,3

10,2

13,8

2,1

1,2

52,7

strip

31

50

1,4

8,5

7,7

21,8

2,0

1,3

92,7

JGP

post

Fig.1. Combined plan of foundations, strengthening constructions and settlement marks

Fig. 2. Monitoring results of foundations settlements

2478

Technical Committee 211 / Comit technique 211

Table 2
Settlement Base pressure
Mark
before
before
Foundat
number
ion type construction construction
(fig.1)
restart, mm restart, kH/m2

Coefficient
of subgrade
reaction,
kH/m3

post

13,5

64

4700

14

strip

9,3

114

12200

26

strip

5,4

114

20000

27

strip

25,5

114

4400

29

strip

25,0

114

4500

31

post

60,0

144

2400

S=f(P)
0

100

200

300

400

0,001

500

600
N1

700

800

900

,k
N2

0,002
0,003
0,004
0,005 S1(3,2)
0,006
0,007 S1 (1)
0,008
0,009
0,010 S2(3,2)
0,011
0,012 S2 (1)
0,013
S,m
(test N.3)
(test N.2)
(test N.1)

Fig.3. Diagrams of field investigations


a base pressure-settlement of post- and strip foundations; b loadsettlement of test piles
Table 3

425 mm diameter and 1011 m length jet grouted piles were


deepened into gravel soil to the depth of 1 m and more. To
evaluate the pile design load, the pile vertical load test of trial
piles with the diameter 425 mm, 10,6 m length (pile No. 1),
10,88 m length (pile No. 2) and 11,5 length (pile No. 3) (table
3) have been carried out. The piling was realized with the unit
SBU 100 GA50.The engineering-geological characteristics of
soils are presented in table 3.
Pile vertical load tests have been carried out according to
standard method. The limit resistances while testing reached
980 kH. Figure 3 presents diagrams of pile tests.
Considering different structural concepts of the foundations and
the building, analysis have been carried out according to these
features and four types of the foundations strengthening have
been suggested (see figures 1 and 4).
I. Strengthening of post foundations of a building in axes 15
1. Practically all extra load is taken into account to be
transferred to jet grouted piles, i.e. the load is not transferred to
post foundation, as the construction of the reinforced concrete
raft strengthening is not absolutely stiff.
Thats why only insignificant part of the extra load is
transferred to the foundation base, so the foundation in
combination with the strengthening construction and piles
behaves as combined piled foundation.
II. Strengthening of strip foundations without piles by means of
geometrical dimensions increase with use of technology
HILTI.
III. Strengthening of pile group foundations with increased
loads was carried out by means of jet grouted piling around the
raft and including them into pile group behavior through the
reinforced concrete slab fixed with the column and the raft (with
the anchors HILTI). With such method of strengthening jet
grouted piles start to work in a pile group together with the
driven piles.
IV. Strengthening of foundations without piles in axes 1 29
was carried out by means of insignificant part of load transfer to
the foundation. Jet grouted piling use is based on insufficient
reinforcing with in situ reinforced concrete strip under the
columns, the load of which is more than twice increased
compared to design one. Such strengthening construction
partially loads the existing foundation including it into work.
The jet grouted piles together with the foundation accept the
ultimate design load. Pile strengthening is carried out along the
whole length of the strips, as otherwise the different stiffness of
the strip base will lead to its deterioration.
Irrespective of strengthening type, the main design requirement
is continuation of foundations loading only after completion of
all works on strengthening considering the terms of strength
increase of in-situ concrete of structures.
4

THE MAIN DESIGN PRINCIPLES

Analysis of foundation strengthening has been carried out


considering the deformability of the foundation base and jet
grouted piles. Due to special features of constructions of the
foundations under strengthening and different extent of works
completion on above foundations structures construction, the
following design assumptions were taken.
While the building post foundations strengthening in axes
115, analyses of loads transferred to the foundation after its
strengthening were carried out i.e. when construction restarting
considering the loads after the building starting (fig.5). Analysis
of the column joint and strengthening construction was done for
the total design load. Deformability indices of the foundation
base and jet grouted piles quantitatively evaluated with the
coefficients of subgrade reaction of the foundation base under
strengthening and pile stiffness respectively, were determined

2479

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Fig.4 Types of foundation strengthening with jet grouted piles (technical decision)

by results of pile test at site and monitoring of post foundations


settlement (tables 2, 4). While analysis of the number of jet
grouted piles, the total load was taken to be transferred to those
piles, so the coefficient of subgrade reaction of the foundation
base under strengthening was taken to be equal 0.
When piled foundations strengthening with jet grouted piles,
analyses have been carried out for the total design load. Pile
stiffness was determined by data of pile vertical load test of the
driven- and jet grouted piles.
The authors of the paper made predictions of the foundation
settlement in axes 1 15, P-S due to foundations load
increments, strengthened with the jet grouted piles after the
construction restart. The deformations are calculated with
regard for the different deformability of the post foundation
base and a pile. The coefficient of subgrade reaction in the post
foundation base (settlement marks M7 and M31 in fig.1) was
determined by results of settlements monitoring (table 2, fig.3a)
and stiffness coefficient of jet grouted piles by tests data (table
4, fig. 3b). The design scheme is presented in fig.5. By results
of analysis of the base deformation of the most loaded post
foundations strengthened with the jet grouted piles, the
settlements after the construction restart with regard for the total
load were 13,8 and 12 mm, respectively. At present, when the
construction of the shopping center is completed, but the
building is not put into operation, the measured base
deformations of these foundations are 10,5 and 5 mm,
respectively. Such conformity of predicted and measured
deformations confirms the efficiency of the base strengthening
and high accuracy of the analysis methods based on in-place
tests.
Table 4
Test
number
(table
3)

Pile
(JGP)
length, m

1
2
3

10,6
10,88
11,50

Pile (JGP)
length in
soil, m
10
10,58
11,10

Pile
settlement
according
to test, mm

Stiffness ratio
of pile (JGP)
base, kN/m

13
11
10,35

50000
60000
60000

N 2 N1
S 2 S1

2480

Fig.5. Design diagram of foundation strengthening


(strengthening types 1 and 4)

CONCLUSION

1. The complicated practical geotechnical problem of


strengthening of the skeleton building foundations under
construction, the necessity of which was provoked by change
of space-planning decision in the process of construction is
solved.
Within one 500x250 m building, four types of foundations have
been designed: post foundations on the bed without karst
protection; strip foundations on the bed designed for 7 m
diameter karst hole; piled foundations in kind of pile groups of
driven piles without karst protection and pile group foundations
combined with karst protected reinforced concrete strips on
piles.
2. Four types of foundations strengthening has been developed
with the use of jet grouted piles taking into attention loads
increase compared to design, acting (already imposed) loads
and foundations settlements at the moment of their
strengthening, structural concepts of foundations and the extent
of karst risk of the base.
3. Analysis of strengthening constructions and base
deformations was carried out according to data of jet grouted
piles vertical load test and settlements measurements of the
foundations under strengthening.
4. The results of the strengthened foundations settlements
measurement after the building implementation showed the
good precision with analysis data. This proves the correctness of
the taken structural and design schemes.

Standardization of the molding procedures for stabilized soil specimens as used for
QC/QA in Deep Mixing application
Normalisation des procdures pour la production dprouvettes de sols stabiliss utilises dans les
processus de QC/QA pour des applications de Deep Mixing
Grisolia M., Leder E., Marzano I.P.

Department of Civil and Environmental Engineering (DICEA), Sapienza University of Rome

ABSTRACT: An international collaborative research has been undertaken to establish common understanding of the key issues
involved in Quality Control/Quality Assurance (QC/QA) of Deep Mixing technique and propose international standards on design,
execution and execution control. The aim of the study is to investigate the influence of the laboratory procedures on the mechanical
properties of stabilised soil specimens and develop an innovative method to select the appropriate molding technique. A large
laboratory testing program was carried out on seven types of heterogeneous natural soils, as found in Rome, and on Kawasaki clay
stabilised with Portland cement. Thirty soil-binder mixtures with different workability were prepared using five different molding
techniques, varying initial water content of the soils, water to cement ratio and binder amount. Unconfined compression tests have
been carried out systematically on over 800 specimens. The applicability of different molding techniques in function of the
workability of the mixture has been investigated and from the results it was possible to define an applicability index and therefore
the range of applicability for each technique in function of the mixtures workability.
RSUM: Une tude internationale a t entreprise dans le but de dfinir des orientations communes pour les procdures QC/QA lis
aux travaux effectus par Deep Mixing et proposer des normes internationales relatives la conception, l'excution et le contrle
des oprations. Le but de cette tude est d'tudier l'effet des procdures de laboratoire pour la ralisation des prouvettes de sols
stabiliss et de dvelopper une mthode innovante pour slectionner chaque fois la technique de ralisation approprie. Un vaste
programme d'essais en laboratoire a t ralis en analysant plus de trente mlanges diffrents de ciments et sols partir de huit sols
naturels de Rome et Tokyo. Cinq techniques de ralisation ont t utilises pour la confection dprouvettes testes avec des essais de
compression simple.. L'applicabilit des diffrentes techniques de ralisation a t tudie selon l'usinabilit du mlange. A partir des
rsultats, il a t possible de dfinir un index dapplicabilit et donc un champ d'application de chaque technique en fonction de
lusinabilit du mlange..
KEYWORDS: Deep mixing, workability, laboratory procedures, operational abaci.
1

INTRODUCTION

The Deep Mixing Method is a widely spread in situ ground


improvement technique using different kind of binders to
enhance mechanical and physical properties of soils (Terashi
1997; CDIT 2002).
Laboratory mixing tests are essential to QC/QA processes
and performed to obtain the mechanical and physical properties
of stabilized soil samples. The laboratory test results provide
crucial information for the estimation of the mix design and insitu properties to utilize in the geotechnical design. (Bruce et al.
2000; Larsson 2005; Marzano et al. 2009; Terashi and Kitazume
2011; Filz et al. 2012). At the moment many laboratories
produce and test soil-binder specimens without a standard
procedure, therefore the results for the same soil-binder mixture
could be very different and not usefully compared. In fact
molding techniques have a great influence on the mechanical
and physical properties of the stabilized soil specimens (Grisolia
et al. 2012; Marzano et al. 2012). This influence is strictly
correlated to the workability of the soil-binder mixture defined
as the property of the mixture of being easily mixed in the bowl
and placed in the mold. High workability refers to liquid type
mixtures (easier to place and handle), while low workability to
sticky and stiff type ones.
Workability represents diverse characteristics of fresh
mixture that are difficult to measure quantitatively, because a
soil-binder mixture is a complex material with a wide range of
particle sizes and time-dependent properties. The definition of a
parameter representative of the mixtures workability and an
univocal method for the evaluation of the mixtures workability
are currently not available (Koehler and Fowler, 2003) even if

such parameter could be well related to consistency when


considering mixtures made up of cohesive soils.
A new method for the evaluation of the mixtures
workability was introduced and applied in the study. It is based
on the measure of the torque required to turn an impeller in soilbinder mixture through a commercial device which is applied
directly on the mixer. This method has the advantage to provide
the possibility of measuring the workability for each mixable
mixture, independently on the type of the materials used.
Furthermore, the study develops a procedure to select,
through an applicability index function of the initial mixture
workability, the molding technique that provides densest
specimens with highest strength and results repetitiveness in
order to obtain very useful reference values to set specification
limits to be achieved in field applications (ratio between
laboratory and field target strength is reported for instance by
JGS 0821-2000 and EuroSoilStab 2002).
2

MATERIALS AND METHODS

The experimental work consisted in a laboratory investigation


on the effect of different molding techniques on the unconfined
compressive strength, UCS (measured according to the JIS A
1216:2009) and wet density, (defined as the specimens
weight divided by the volume of the mold) of cement stabilised
soil specimens under various mixing conditions.
2.1 Materials
Eight types of natural soils stabilised with Portland cement
added in wet or dry form were used. The tests were performed
on: Kawasaki Clay (KC), manmade Silty Deposit (SD), Silty

2481

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

clayey Sand (SS), Sand and Gravel (SG), Pliocene Clay (PC),
Black Pozzolana (BP), Red Pozzolana (RP) and Argillified Tuff
(AT). For each soil, different mixtures were produced, varying
the initial water content and keeping constant the cement
content, ac (defined as the weight of the introduced dry cement
divided by the dry weight of the soil to be stabilized).
Specimens with 5 cm diameter and 10 cm height were
employed. Each soil was sieved through a 9.5mm sieve, so that
the maximum grain size of the soil sample would be less than
1/5th of the inner diameter of the mold. The properties of the
soil-binder mixtures analyzed are shown in Table 1.
Table 1. Soil properties and Testing conditions.
Soil
type

GravelSandSiltClay

KC

0-1442-44

SD

18-2434-24

SS

22-4020-18

SG

33-4014-13

PC

00-0064-36

BP

08-4938-05

RP

11-5824-07

AT

02-4739-12

Water
content,
wn

Cementitious
grout

(%)
72
66
60
60
60
54
54
49
49
20
30
40
35
40
45
6
8
10
50
60
70
25
30
35
20
26
32
44
48
53

w/c = 0
ac = 5%
ac = 20%
ac = 30%
ac = 20%
ac = 30%
ac = 20%
ac = 30%

ac = 10%
w/c = 1

ac = 10%
w/c = 0,5

Workability
parameter,
Torque, Mt
(Nm)
5,32
8,47
17,40
29,00
40,00
61,00
75,00
96,00
120,00
13,55
4,81
2,23
9,08
4,88
3,76
11,28
5,11
3,51
10,16
5,76
2,34
6,97
2,37
0,21
8,34
1,60
1,08
8,22
0,60
0,20

Afterwards the stabilized soil was placed into plastic molds


in three layers and compacted using several molding techniques
(Figure 1):
_ No Compaction (namely N.C.): It simply consisted in
placing the stabilized soil into the mold with a spoon.
_ Tapping, namely (TA.): For each layer, the mold was
tapped against the floor 50 times.
_ Rodding, (namely RO.): It consisted in tamping each layer
with a 8mm diameter steel rod for 30 times and eventually
pushing down the material attached to the rod.
_Static Compaction (namely S.C.25 and S.C.50): Each layer
was statically compressed for 10 seconds by using a heavy rod,
49 mm in diameter. 25 or 50 kPa pressure were applied.
_Dynamic Compaction (namely D.C.): Each layer was
compacted by a falling weight (1.5 kg) using a special
apparatus. Fall height was set at 10 cm, number of blows at 5.
These techniques are those currently used in most of the
laboratory all over the world (JGS 0821, 2000; EuroSoilStab,
2002; Kitazume et al. 2009).

Figure 1. Molding techniques used: a) N.C., b) TA., c) RO.,


d) S.C.25, e) S.C.50, f) D.C.

2.2 Laboratory procedures and testing methods


A Hobart type mixer apparatus was adopted. After placing the
natural soil in the mixer, the water content was adjusted to the
desired value by adding water. Before adding the binder the soil
was homogenised by mixing. The grout made of Portland
cement (PC) and water or the PC in dry form was then added to
the soil and mixed for ten minutes according to JGS 0821
(2000).
Using a commercial device applied directly on the kitchen
mixer apparatus, it was measured the torque required to turn the
impeller in a soil-binder mixture just before the molding phase.
According to the proposed method the workability was
expressed as a torque (Mt) applied to mix certain amount of soilbinder mixture (Vm) with set impeller shape (Sh) and rotational
speed (Rs). In the study were assumed the following parameters:
Vm0 = 3dm3; Sh0 = K shape; Rs0 = 10rpm.
Since the kitchen mixer has a planetary motion, the test was
undertaken continuously on the whole mixture therefore giving
more reliable outputs. For each completed revolution it was
possible to measure the Torque, and for the test it was decided
to set the number or revolution to 10, to obtain more accurate
measures.

To prevent water evaporation from the specimen each mold


was covered with the sealant and stored in special curing tanks
at 95% relative humidity. To reduce the effect of the time of rest
between the hydration of binder and completion of molding on
the specimens properties, according to Kitazume et al. (2009),
all the stabilized soil was molded in less than 45 minutes from
binder hydration. After curing times of 28 days, the specimens
were removed from the molds and then subjected to unconfined
compression tests at a rate of 1.0 mm/min. Unconfined
compression tests were conducted on triplicate samples for each
case (mixture type and molding technique) analyzed.
3

RESULTS

The applicability of a molding technique was evaluated by the


Applicability index, which is related to densest specimens
with the highest strength and results repetitiveness.
For the same mixture a well made specimen has low
cavities/bubbles/voids amount and therefore higher wet density
() if compared to a bad made one. Furthermore it was indeed
observed that the specimens produced by different molding
techniques could have similar wet density but very different
unconfined compressive strength (UCS) values. For that reason
the N-parameter was introduced to condense the indications
given by both parameters, and UCS, into one. N is defined as
the mean of the normalized unconfined compressive strength
(UCSN) and the normalized unit weight (N) as reported in the
Eq. 1. The UCSN and N values for a given molding technique

2482

Technical Committee 211 / Comit technique 211

and mixture were calculated by dividing the UCS and by the


maximum values (UCSmax and max) obtained from all the
employed molding techniques. The normalizations are
necessary to allow direct comparison between two parameters
with different unit of measurements. According to the Eq. 1 the
N parameter values range between 0 and 1. In order to define a
criteria for the choice of the applicable techniques, it was set the
acceptable limit of 0.9N considering a variation of 10 % from
the maximum N value.

(1)
The Figure 2 shows as example the N parameters vs. torque
values obtained for all the analyzed soil-binder mixtures molded
by the Rodding technique. From the figure clearly appears that
for all the measured mixtures workability the N values are
above the set limit as an expression of the high quality of the
specimens realized by Rodding.
SD
AT

SS
KC

SG
SD

PC
SS

(3)
According to the Eq. 3 and to the N and E parameters
definitions, also the IA values range between 0 and 1. To obtain
a target value for the choice of the applicable techniques, the
limit values given for the two different parameters N and E were
introduced in the Eq. 3. A target value of IA = 0.82 was then
obtained.
The Figure 3 shows an example of the IA vs. mixtures
workability graph obtained for all the analyzed soil-binder
mixtures molded by the Rodding technique.
KC
BP

BP
SG

1.0
0.9

N parameter
limit = 0.90

N, E parameter

0.8
0.7

SD
RP

SS
AT

SG
limit

PC

1.0

Applicability index, IA

KC
RP

In order to take into account the different aspects of a well


made specimen, expressed by the N and E parameter, an index
of applicability IA defined in Eq. 3 was introduced.

0.6
0.5
0.4

0.8
0.6
0.4
0.2
0.0

0.3

E parameter
limit = 0.10

0.2
0.1

0.0
RODDING

12

18

mixture workability, Torque Mt(Nm)

28 days

Figure 2. Applicability of Rodding technique considering the N and E


parameters.

18.0

28 days

Despite the N-parameter is a good indicator of the


applicability of a molding technique, to take into account that
the applicability should be also related to the repeatability of
the tests results the E-parameter was introduced. Repeatability
means that the results related to the specimens produced by a
specific molding technique should have a low scatter or
relative error. Unconfined compression tests was conducted on
triplicate samples for each case analyzed, therefore it was
possible to evaluate the relative error on the unconfined
compressive strength and wet density values for the different
mixtures types and molding techniques. E is defined as the
mean of the relative error on the UCS and values as reported
from Eq. 2. According to its definition also this parameter
ranges between 0 and 1. To set a criteria to select the applicable
techniques, some literature works were taken into account. For
the accuracy or repeatability of the experiments, Richards and
Reddy (2010) claimed that a standard deviation of 10 % was not
unheard of in geotechnical testing. Al-Tabbaa et al. (2012) also
reported an error of 5 15% for laboratory mixed specimens
tested with unconfined compression tests. Therefore even for
the E parameter the acceptable limit was set equal to 10% of E.

The figure show that Rodding is applicable for all the


measured mixtures workability since IA values are all above the
set target limit. From the results it is possible to see a very good
trend of the IA despite the fact that data were obtained from
mixtures based on different types of soil (cohesive and granular
types), with different grout dosage and water contents. The
results obtained also from other techniques show that the IA is
strictly dependent on the workability of the mixture among
other factors.
The results related to the No Compaction technique are
shown in Figure 4. It clearly appears that this technique is
applicable for Mt < 3Nm and not applicable for Mt > 6Nm. In
the range Mt = 36Nm it is not possible to obtain univocal
indication from the data, therefore this technique have been
considered marginally applicable in this workability interval.
KC
BP

SD
RP

1.0

Applicability index, IA

RODDING

12.0

Figure 3. Applicability index of Rodding technique.

0.0
0

6.0

mixture's workability - Torque, Mt (Nm)

(2)

SS
AT

SG
limit

PC

0.8
0.6
0.4
0.2
0.0
0.0

The results obtained from the specimens molded by the


Rodding technique in terms of E vs. torque values are shown in
Figure 2 as example. It can be clearly seen that also for the E
parameter all the obtained values are below the set limit,
expression once again of the high repeatability of the tests
results obtained from the specimens molded by the Rodding
technique.

6.0

12.0

18.0

mixture's workability - Torque, Mt (Nm)

NO COMPACTION

28 days

Figure 4. Applicability index of No Compaction technique.

Similar graphs to the ones shown in Figures 3 and 4 were


also obtained for the other molding techniques used in the
study. From these graphs it was possible to determine for each
molding technique the ranges of workability in which they are

2483

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

applicable, marginally applicable and not applicable. The results


are summarized in the operational abacus of Figure 5.
Applicable

Marginally Applicable

Not Applicable

3 6

Molding technique

No Compaction
65

75

Tapping

Static
Compaction
25kPa

10 15

Static
Compaction
50kPa

30

Dynamic
Compaction

40

6
0

10

20

30

40

50

60

70

80

120
90

...
100

mixture's workability, Torque Mt (Nm)


High
workability,
liquid

Low
workability,
consistent

Figure 5. Ranges of applicability of the different molding techniques.

To allow the standardisation and the use of the method a


calibration curve was elaborated by drawing the torque versus
the water content (w) of an easily available kind of soil such as
kaolin clay using the set of mixer related parameter Vm0, Sh0 and
Rs0 (Figure 6).
calibration curve

experimental data

example curve

100

water content, w [%]

90
80
70
60
50
40
Vm0, Sh0 , Rs0

30

Vm, Sh, Rs

20
10
0

10 20 30 40 50 60 70 80 90 100 110 120


workability, torque Mt [Nm]

Figure 6. Calibration curve.

By using different set of parameters (Vm, Sh and Rs),


function of the mixer type and torque evaluation procedure,
other curves can be drawn in the same graph. Therefore for each
molding technique the range of applicability (expressed by
torque values) corresponding to the used set of parameter can be
graphically obtained from the calibration curve, as shown in
Figure 6.
By mean of the abaci of Figures 5 and 6, it would be possible
to select for every kind of mixable soil-binder mixture the
molding technique that gives high quality specimens in a very
quick and easy way only by measuring the workability of the
material.
4

ACKNOWLEDGEMENTS

The authors wish to thank Prof. Masaki Kitazume for his help
and suggestions throughout the study. The Soil Stabilization
Group of the Port and Airport Research Institute (JP) is also
acknowledged for the helps during the tests and for providing
part of the data.

Rodding

The results obtained represent a useful data set for the


correct selection of the molding technique for different kind of
soils and mixing conditions headed for the international
standardisation. This study represents a significant step forward
towards the definition of highly required guidelines for the
molding procedures of stabilised soil specimens as used in
QC/QA processes for Deep Mixing applications.

CONCLUSIONS

The results of the large laboratory study performed on eight


types of natural soils confirm that the mixtures workability
has a great influence on the mechanical and physical properties
of the stabilised soil specimens. The results provide very useful
operational abaci to select the molding technique that produces
high quality specimens in function of the soil-binder mixtures
workability.

2484

REFERENCES

Al-Tabbaa. A., Liska, M., McGall, R., Critchlow, C. 2012. Soil Mix
Technology for Integrated Remediation and Ground Improvement:
Field Trials. IS-GI Brussels 2012, Belgium.
Bruce, D.A., Bruce, M.E., Di Millio, A.F. 2000. Deep Mixing: QA/QC
and Verification Methods. Grouting Sol Improvement, Geosystems
Including Reinforcement. 4th Intnl Conf. Ground Improvement
Geosystems. Helsinki, Finland, pp. 11-22.
CDIT, Coastal Development Institute of Technology 2002. The Deep
Mixing Method - principle, design and construction. A.A. Balkema
Publishers, Lisse, Abingdon, Exton (PA), Tokyo, p. 123.
EuroSoilStab, 2002. Development of design and construction methods
to stabilise soft organic soils. Design Guide Soft Soil Stabilization.
EC project BE96-3177, 94p.
Filz, G., Adams, T., Navin, M., Templeton, A.E. 2012. Design of Deep
Mixing for Support of Leeves and Floodwalls. 4th Intnl. Conf. on
Grouting and Deep Mixing, Marriott New Orleans, LA, US.
Grisolia M., Kitazume M., Leder E., Marzano I.P., Morikawa Y. 2012.
Laboratory study on the applicability of molding procedures for the
preparation of cement stabilised specimens. IS-GI Brussels 2012,
Belgium
JGS 0821-00 2000. Practice for Making and Curing Stabilised Soil
Specimens Without Compaction (Translated version). Geotechnical
Test Procedure and Commentary, Japanese Geotechnical Society.
Kitazume, M., Nishimura, S., Terashi, M., Ohishi, K. 2009.
International Collaborative Study Task 1: Investigation into
Practice of Laboratory Mix Tests as Means of QC/QA for Deep
Mixing Method. International Symposium on Deep Mixing &
Admixture Stabilization, Okinawa, Japan.
Koehler, E.P., and Fowler, D.W. 2003. Summary of Concrete
Workability Test Methods. Research Report, International Center
for Aggregates Research The University of Texas at Austin.
Larsson, S. 2005. State of Practice Report Execution, monitoring and
quality control. Intnl. Conf. on Deep Mixing - Recent Advances
and Best practice, Stockholm, Sweden.
Marzano I.P., Al-Tabbaa A., Grisolia M. 2009. Influence of sample
preparation on the strength of cement-stabilised clays. Intnl. Symp.
on Deep Mixing & Admixture Stabilization, Okinawa, Japan.
Marzano I.P., Leder E., Grisolia M., Danisi C. 2012. Laboratory study
on the molding techniques for QC/QA process of a Deep Mixing
work. 3rd International Conference on New Developments in Soil
Mechanics and Geotechnical Engineering, Near East University,
Nicosia, North Cyprus. ISBN 975-8359-28-2.
Richards, K. S., Reddy, K. R., 2010. True triaxial piping test apparatus
for evaluation of piping potential in earth structures. Geotechnical
Testing Journal, 33(1): 1-13.
Terashi, M. 1997. Deep mixing method Brief state of the art. 14th
International. Conference on Soil Mechanics and Foundation
Engineering. Tokyo
Terashi, M., and Kitazume, M. 2011. QA/QC for deep-mixed ground:
current practice and future research needs. Proc. of the Institution of
Civil Engineers Ground Improvement 164 (3), 161-177.

Analysis
Analysis of
of Floating
Floating Pile
Pile Capacity
Capacity in
in Improved
Improved Ground
Ground for
for Thi
Thi Vai
Vai Port,
Port, Vietnam
Vietnam
Analyse
Thi Vaidu
Amlioration
deVietnam
Port, Vietnam
Analyse de
de la
la capacit
capacit de
de Pile
pile flottant
flottantedans
dansun
unterrain
sol amlior
port Thi Vai,
Hai
Hai N.M.,
N.M., Tuong
Tuong N.K.
N.K.

Faculty of Civil Engineering, Thu Dau Mot University, Vietnam


Faculty of Civil Engineering, Thu Dau Mot University, Vietnam

Long
Long P.D.,
P.D., Nhon
Nhon P.V.
P.V.

Vietnamese Society of Soil Mechanics and Geotechnical Engineering, Vietnam


Vietnamese Society of Soil Mechanics and Geotechnical Engineering, Vietnam

ABSTRACT: A static loading programme was performed to respond to total and effective stress analysis of floating pile capacity in
90 km southeast
of Ho
City, Vietnam.
soil profile
consists
an about
improved ground
at Thi
Vai Port
approximately
ABSTRACT:
A static
loading
programme
was performed
to respond
to Chi
totalMinh
and effective
stress The
analysis
of floating
pile of
capacity
in
15 to 23 ground
m thickatdeposit
soft,
normally consolidated,
compressible
clay
deposited
on denseThe
to soil
compact
Scheme
soil
improved
Thi VaiofPort
approximately
90 km southeast
of Ho Chi
Minh
City, Vietnam.
profilesand.
consists
of anofabout
improvement
was deposit
imposed,ofconsisting
of wick
drains installed
at a spacing
of 1.5
m through
the clay
the sandsand.
and placing
m
15
to 23 m thick
soft, normally
consolidated,
compressible
clay
deposited
on dense
to to
compact
Schemetoof4.7
soil
thick surcharge.
removal
of the surcharge,
two square
precast
concrete
mm diameter,
to placing
depths of
m
improvement
wasAfter
imposed,
consisting
of wick drains
installed
at a spacing
ofpiles,
1.5 m400
through
the clay towere
the driven
sand and
to16
4.7and
22 m surcharge.
and the static
loading
testsofwere
performedtwo
reaching
pile capacities
driven
about 14were
and driven
23 days,
The
thick
After
removal
the surcharge,
squarethe
precast
concrete after
piles,piles
400 mm
diameter,
to respectively.
depths of 16 and
capacity
16 and
22 mtests
pilewere
measured
are about
450 the
andpile
1,100
KN at maximum
5 and
6 mm,
respectively.
22
m and of
thethe
static
loading
performed
reaching
capacities
after piles movements
driven aboutof14about
and 23
days,
respectively.
The
The testsofindicate
a good
between
measurements
and KN
analysis
of total movements
stresses, and
correlation
coefficient,
NKT,
capacity
the 16 and
22 magreement
pile measured
are about
450 and 1,100
at maximum
of the
about
5 and 6 mm,
respectively.
between
porea pressure
adjusted cone
stressmeasurements
and vane shearand
stress
is about
instead
of 12and
through
16 as used for
the Project.
The
testsCPTU
indicate
good agreement
between
analysis
of 18
total
stresses,
the correlation
coefficient,
NKT,
between CPTU pore pressure adjusted cone stress and vane shear stress is about 18 instead of 12 through 16 as used for the Project.
RSUM : Un programme de chargement statique a t ralis pour rpondre l'analyse des contraintes totale et effective de la
capacit du: Unprogramme
pieu flottant dans
sol amlior au port
Vaipourrpondre
environ 90 l'analyse
km au sud-est
de Ho Chi Minh-Ville,
Vietnam. Le
profil
RSUM
de un
chargementstatiquea
tdeThi
ralis
des contraintestotale
et effectivedela
capacit
de pieuflottantdans
sol est constitu un
d'un
dpt de 15
23 deThiVai
m dpaisseur
d'argile
compressible
normalement
consolid,
sur de
unsolest
sable
du
solamlior
au port
environ
90 souple
kmau sud-estde
Ho Chi
Minh-Ville,
Vietnam.dpos
Leprofil
dense et compact.
Le schma
d'amlioration
des sols a t
impos, compos de drains verticaux
installs
une
de1,5 m
constitud'un
dptde
15 23m
dpaisseurd'argile
souplecompressiblenormalement
consolid,
dpos
surdistance
un sabledenseet
travers l'argile
au sable
et la mise des
4,7 sols
m d'paisseur
en supplment.
l'limination
de la surcharge,
deux pilesde1,5
carres
bton
compact.Le
schma
d'amlioration
a t impos,
compos Aprs
de drains
verticauxinstalls
une distance
mentravers
prfabriqu
de 400
de4,7md'paisseur
diamtre ont teninstalls
desAprs
profondeurs
de 16 22
m et les essais
de chargement
statique
ont t
l'argileau
sableet
lamm
mise
supplment.
l'liminationde
la surcharge,
deux pilescarresen
bton
prfabriqu
raliss
pour de
atteindre
les capacits
de pieux
ce que des 16
pieux
en place
14et 23
jours, respectivement.
La
de
400 mm
diamtreont
t installs
desaprs
profondeursde
22sont
metmis
lesessais
de environ
chargement
statiqueont
t ralisspour
capacit de capacitsde
la pile16 et pieuxaprs
22 m mesure
est denviron
450et 1100
kN aux mouvements
maximum
d'environ 5et
6 mm, respectivement.
atteindreles
ce quedes
pieux sontmis
en placeenviron
14et 23jours,
respectivement.
La capacitde
la pile16 et
Les tests montrent
un bon 450et
accord1100KNaux
entre les mesures
et l'analyse des contraintes
querespectivement.
le coefficient deLestests
corrlation,
NKT, entre
22mmesureest
denviron
mouvementsmaximum
d'environ totales,et
5et 6 mm,
montrentun
bon
la pression interstitielle
CPTUl'analyse
ajuste la descontraintes
contrainte cne totales,et
et la contrainte
cisaillementdeest corrlation,
d'environ 18au
lieu de
12 16comme
accordentre
les mesureset
quele decoefficient
NKT,
entrela
pression
utiliss pour le projet.
interstitielleCPTUajustelacontraintecne
et la contrainte de cisaillement est d'environ 18au lieu de 12 16comme utiliss pourle
projet.
KEYWORDS: soil improvement, static loading test, total stress analysis, effective stress analysis.
KEYWORDS: soil improvement, static loading test, total stress analysis, effective stress analysis.

10

15

20

Sleeve Friction, fs (KPa)


0 20 40 60 80 100
0

10

15

20

Pore Pressure (KPa)


0 250 500 750 1,000

10
U2

15

20

DEPT H (m)

5
DEPT H (m)

The Thi Vai Container Port is built on an improved ground over


a 470 m by 600 m area along the Thi Vai River approximately
90 km southeast of Ho Chi Minh City. The soil profile consists
of deltaic sediments of about 15 to 23 m of soft, normally
consolidated, highly compressible clay on a thick layer of dense
to compact sand. The highest water level is at Elev. +4.0 m,
which requires raising the ground elevation by about 2 m to
Elev. +5.0 m to avoid flooding. To reduce settlement after
construction, wick drains were installed with a spacing of 1.5 m
through the clay to the sand and a temporary surcharge was
added raising the surface to Elev. +7.6 m, an additional 2.0 m of
fill height. The surcharge was removed after 80 % to 90 % of
the consolidation settlements had developed and the expected
future ground settlement shall not exceed 200 mm over a period
of 20 years, which included secondary compression. After
removal of surcharge, two 400 mm square precast piles driven
into 16 and 22 m depth to serve for test and design of building
foundations.
This paper presents the methods used for analysis of the
pile capacity based on total and effective stress approaches,
details of the results from the tests on 16 and 22 m piles, and
discusses the merits of each method.
2 SOIL PROFILE
The soil profile is indicated in Figures 1 and 2. Figure 1 shows
the results of a typical CPTU sounding pushed at the site before
driving test piles.

Cone Resistance, qt (MPa)


0 2 4 6 8 10
0

DEPT H (m)

INTRODUCTION

DEPT H (m)

Friction Ratio, fR(%)


1 2 3 4 5

10

15

20
U0

25

25

25

25

Figure 1. Diagram of CPTU sounding pushed before


construction start
Figure 2 presents distribution of the grain size, water content,
consistency limits, and the distribution of the undrained shear
strength in the clay from a field vane. The field vane
demonstrates the clay to be very soft above 10 m depth and soft
below. Total saturated density is about 1,600 kg/m3 throughout
the clay (from wn = 61 %). The density of the sand above and
below the clay is estimated to 2,100 kg/m3 (from wn = 24 %)
and 1,800 kg/m3 (from wn = 22 %), respectively.

2485

WATER CONTENT (%)


20

40

60

80

GRAIN SIZE DISTRIBUTION


100

20

80

100

FIELD VANE STRENGTH


SU (KPa)
0 20 40 60 80 100
0

CLAY
SAND

SILT

wP
15

20

wn wL

SILT

SAND
10

10

DEPTH (m)

10

DEPTH (m)

CLAY

DEPTH (m)

60

+ 4.0 m GWL

wn

20

20
SILT
25

CLAY

SAND GRAVEL
25

Figure 2. Water content and Atterberg Limits, grain size


distribution, and field vane strength
3

200

Temporary
stockpile

150

March 12, 2009 - June 24, 2012

100

SS8

50
0
0

300

300

600
600 (day)
TIME

0.0

15

15

wn
25

40

SETLLEMENT (m)

0
+ 5.0 m
0

FILL STRESS (KPa)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

900

1,200

900

1,200

0.5
1.0

22 mTest
Pile Driving

SS8

1.5

16 mTest
Pile Driving

2.0
2.5
3.0

SOIL IMPROVEMENT AND MEASUREMENTS

Figure 4. Fill stress versus settlement


+ 5.0 m, Elevation of top sensor
0.0
0.5
1.0
0

SETTLEMENT (m)
1.5

2.0

2.5

5
June 30, 2011
DEPTH (m)

Figure 3 shows the location of SS8 settlement benchmark was


installed on the original ground surface before the placing of
the fill. Pore pressure measurement was performed by
piezometer tips installed at depths of 6.5 m and 15.2 m.
Settlement distribution with elevations was measured by means
of extensometer gages placed at elevations of +5 m, -1.6 m, 5.9 m, -10.3 m and -17 m in the clay. Figure 4 also indicates
location of field soil tests and test piles.

10
August 29, 2009 June 30, 2011

15

Zero Reading was taken on August


29, 2009. Total settlement of top sensor
was about 1.95 m. The dark blue curves
show readings spaced about two
months apart.

20

25

Figure 5. Distribution of settlement versus depth

BHs
Test Piles
Settlement Plates
CPTU and VST
Extensometer, Piezometer, and Standpipe

Figure 4 shows the settlements measured at SS8 plate near


the tested piles. Day 0 is March 12, 2009 and the total
settlement measured after completed removal of surcharge
was about 2.15 m. Removal of surcharge at the 22 and 16 m pile
was completed on August 22, 2011 and September 9, 2011,
respectively. 15 and 60 days after removal of surcharge, the 22
and 16 m test pile was driven on September 6, 2011 and
November 8, 2011, respectively, to serve for test and design of
building foundation.
Figure 5 shows the settlement distribution with depth as
measured at extensometer station, next to the SS8 at 4.6, 8.9,
13.3, and 20 m depths below the original ground surface from
August 29 through June 30, 2011. The extensometer station had
to be removed on June 30, 2011 before the test pile driven. The
four settlements anchors were referenced to the presumed zero
for the fifth anchor point placed at 20 m depth. The settlement
distribution is almost linear from the fill surface to zero at 15 m
depth.

2486

200

Temporary
stockpile

150

SS8

100
50
0
0

250

PORE WATER PRESSURE (KPa)

Figure 3. Locations of test piles, borehole, CPTU, VST, and


field instrumentation

FILL STRESS (KPa)

Figure 6 shows the pore pressures measured at Elev. -1.5 m


and -10.2 m from August 29, 2009, through September 23,
2011. The Piezometer had to be removed before driving the test
piles. As shown in Figure 6, the pore water pressures seem to be
equal to the hydrostatic pressures after removing the surcharge.

200

400

600

800

1,000

August 29, 2009 - September 23, 2011

200

150

-10.2 m

100
-1.5 m
50
0

200

400

600
TIME (Day)

800

1,000

Figure 6. Pore water pressure versus fill stress

Technical Committee 211 / Comit technique 211

ANALYSIS OF PILE CAPACITY

4.1

Total stress approach ( Method)

The so-called alpha method is a most common method to


calculate shaft resistance in total stress approach, which
correlates the shaft resistance, rs, to the undrained shear strength
of the clay, Su, via an adhesion reduction coefficient, , as the
undrained shear strength of the soil increased:
rs = Su

The Nkt values in Figure 7 determined correlation between cone


stresses and field vane strengths are used to determine the Su of
CPTU test and calculate the unit shaft resistance according to
Eq. 1 and Eqs. 4 - 6.
The results are plotted in Figure 8 versus the accumulated unit
shaft resistance for the full dissipation of excess pore pressures.
For 5 m thick fill sand layer above the clay layer surface, the
effective stress analysis is applied with a coefficient of 0.3
and indicates the bearing capacity of this layer is about 97 KN.

(1)

SHAFT RESISTANCE (KN)


0

For field CPTU tests, Lunne et al. (1985) proposed a method


to indirectly estimate the Su versus corrected cone resistance
(qt) as:
Su = (qt v0)/Nkt

12

qt

16

DEPTH (m)

DEPTH (m)

12

Su (kPa) & Nkt


30
60
90

1.1

Sand
Clay

Eq. 4
Eq. 6

P
I
L
E
1
6
m

2,000
P
I
L
E
2
2
m

20
Eq. 1
Eq. 5
Sandy silt and silt

Effective stress approach ( Method)

The effective stress approach for evaluating the pile capacity,


Burland (1973) developed a simple equation written as:

Su

rs = Ktanv0 = v0

Figure 7. Correlation between cone stress and field vane


strength
To improve total stress approach, Randolph and Murphy
(1985) considered ratio of the undrained strength to the
effective overburden stress, v0 (stress history) and proposed a
reduction coefficient incorporated into the API 1987 edition
(excluding the effects of the pile length) as:
(3a)
(3b)

The effects of the pile slenderness (ratio of the embedment pile


length, L, to the pile width, B) also were considered by Murff
(1980), Kraft et al. (1981), Semple and Rigden(1984), Randolph
and Murphy (1985). The unit shaft resistance proposed from
two alternative combinations of undrained shear strength and
effective stress was refined in the API 1993 edition as shown in
Eq. (4) and (5).
(4)
(5)

Kolk and van der Velde (1996) suggested an updated version


incorporated directly length effects and the unit shaft resistance
was determined in Eq. (6):
rs = 0.55(Su)0.7 (v0)0.3(40B/L)0.2

= 0.3

1,500

Figure 8. The accumulated shaft resistances versus depth

Nkt

20

rs = 0.5(Su)0.50 (v0)0.50
rs = 0.5(Su)0.75 (v0)0.25

15

30

16

= 0.5(Su/v0)-0.50 for (Su/v0) 1


= 0.5(Su/v0)-0.25 for (Su/v0) > 1

10

25

v0
20

DEPTH (m)

Where, Nkt is known as an empirical cone factor and v0 is


the total overburden stress.
The records of undrained shear strength from the field vane test
are substituted to Eq. (2) to determine Nkt as shown in Figure 7.
(kPa)
2,000 4,000 6,000

1,000

(2)

qt & v0

500

(6)

2487

(7)

Where, K is the lateral earth-pressure coefficient, is the


constant volume friction angle, and = Ktan is BjerrumBurland coefficient.
Two direct CPTU methods typical of effective stress
approach are method of Eslami and Fellenius (1997) and
Takesue et al. (1998). In the Eslami and Fellenius CPTU
method, the cone stress is transferred to an apparent effective
cone stress, qE, by subtracting the measured pore pressure, U2,
from the measured total cone stress, qt, the unit shaft and toe
resistance is obtained from:
rs = CsqE

(8)

Where, Cs is the side correlation coefficient determined from


the soil profile chart which uses both cone stress and sleeve
friction.
For method of Takesue et al. (1998), the unit pile shaft
resistance, rs, is estimated from the measured sleeve friction, fs,
which is scaled up or down depending on the magnitude of the
measured excess porewater pressures during penetration, U
(U = U2 U0). The data used to derive the correlation were
obtained from both bored and driven pile foundations in clays,
sands, and mixed ground conditions.
rs = fs(U/1250+0.768) for U<300 kPa
rs = fs(U/200-0.5) for 300<U<1,250 kPa

(9a)
(9b)

The results of the effective stress analysis are presented in


Figure 9. For fill layer above clay surface, the analysis is made
the same as above total stress analysis.

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

SHAFT RESISTANCE (KN)


0

500

1,000
= 0.3

1,500
Sand

DEPTH (m)

5
Eq. 9

10

Clay

Eq. 8

15

2,000

P
I
L
E
1
6
m

The total and effective stress analysis on two tested piles driven
in improved ground of Thi Vai Port is performed and
summarized as:
For total stress approach, the analysis indicates the
alpha method with a reduction coefficient
incorporated into the API 1987 edition gave the best
agreement.
For effective stress approach, the analysis shows the
Eslami and Fellenius method (1997) is approximately
the same as the measurements.
For two approaches, the total stress approach is the
better agreement.
Correlation coefficient, NKT, between CPTU pore
pressure adjusted cone stress and vane shear stress is
about 18.

P
I
L
E
2
2
m

Sandy silt and silt

20
25
30

Figure 9. The accumulated shaft resistances versus depth


1.2

Pile test results and evaluation of approaches

The static loading tests for the 22 and 16 m Pile were performed
after piles driven about 23 and 14 days, respectively. The 22
and 16 m were tested on September 29, 2011 and November 22,
2011. The load-movement curves of the two tested piles in
Figure 12 indicate the ultimate loads of the plunging failures at
1,100 and 459 KN, respectively.
1,500

Pile 22 m

LOAD (KN)

1,200

1,100 KN

900
Pile 16
m

600
459 KN

300
0
0

10
15
20
25
MOVEMENT (mm)

30

Figure 10. Pile-head load-movement curves of two static


loading tests performed on 22 and 16 m pile.
Normally, the plunging failures of the test piles are found
when the pile toe is in soft clay as in the subject case and the
pile toe resistance can be disregarded when evaluating the
analysis methods. Performance of two approaches are evaluated
basing on ratio of the estimated capacities, Qp, to the measured
capacities, Qm. Table 1 shows the performance evaluation of
two approaches for the measured pile capacities.
approach for predicting the pile capacity.

Total stress approach


Eq. 1
Eq. 4
Eq. 5
Eq. 6
Effective stress approach
Eq. 8
Eq. 9

Pile 22 m
(Qp/Qm)

0.967
1.422
1.048
1.203

0.949
1.416
1.280
1.423

0.915
0.624

0.884
0.688

REFERENCES

API (1987) Recommended practice for planning, designing, and


constructing fixed offshore platforms, API RP2A, 17th edn.
American Petroleum Institute, Washington.
API (1993) Recommended practice for planning, designing and
constructing fixed offshore platformsworking stress design, API
RP2A, 20th edn. American Petroleum Institute, Washington.
Cai G., Songyu L., Liyun T., Guangyin D. 2009. Assessment of direct
CPT and CPTU methods in predicting the ultimate bearing capacity
of single piles. Eng Geol 104:211222,.
Eslami A., Fellenius B.H. 1997. Pile capacity by direct CPT and CPTu
methods applied to 102 case histories. Can Geotech J 34:886904.
Fellenius B.H. 2008. Effective stress analysis and set-up for shaft
capacity of piles in clay. ASCE Geotechnical Special Publication,
GSP180:384-406.
Karlsrud K., Haugen T. 1981. Cyclic loading of piles and pile anchors,
field model tests at Haga. Norwegian Geotechnical Institute
Research Report.
Kolk H.J., van der Velde E. 1996. A reliable method to determine the
friction capacity of piles driven into clays. In: Proceedings of the
28th annual offshore technology conference, Houston, pp 337346.
Lunne, T., Christoffersen, H.P. and Tjelta, T.I. 1985. Engineering use of
Piezocone data in North Sea Clays. Proceedings XI ICSMFE, San
Francisco.
Murff D. 1980. Pile capacity in a softening soil. Int J Numer Anal
Methods Geomech 4:185189.
Randolph M.F., Murphy B.S. 1985. Shaft capacity of driven piles in
clay. In: Proceedings of the 17th annual offshore conference,
Houston, pp 371378.
Semple R.M., Rigden W.J. 1984. Shaft capacity of driven pipe piles in
clay. In: Proceedings of the on analysis and design of deep
foundations, San Francisco, pp 5979.
Takesue, K., Sasao, H., Matsumoto, T. 1998. Correlation between
ultimate pile skin friction and CPT data. Geotechnical Site
Characterization (2): 1177-1182. Rotterdam: Balkema.

Table 1. Evaluation on performance of the total and effective


Pile 16 m
(Qp/Qm)

SUMMARY AND CONCLUSION

2488

Carbonate Cementation via Plant Derived Urease


Cimentation carbonate par lutilisation durase issue de plantes
Hamdan N., Kavazanjian Jr. E., ODonnell S.

School of Sustainable Engineering and the Built Environment, Arizona State University, Tempe, AZ 85287-5306; PH:
(480) 965-3997

ABSTRACT: The use of plant-derived urease enzyme to induce calcium carbonate (CaCO3) cementation has been demonstrated
through laboratory column tests. Benefits of the use of plant-derived urease over the use of microbially-generated urease to induce
carbonate cementation include the small size of the enzyme, which permits penetration into finer grained soils and makes the process
less sensitive to bioplugging, and the availability of 100% of the carbon in the substrate for conversion to CaCO3. The laboratory
column tests employed both Ottawa 20-30 silica sand and finer-grained F-60 silica sand. The laboratory column specimens were
prepared in a variety of manners and showed varying degrees of cementation and carbonate yield. Triaxial tests performed on
cemented specimens showed significant strength increases over non-cemented specimens. These tests confirm the feasibility of using
plant-derived urease to induce carbonate cementation in sand and provide valuable insight into the factors that must be considered in
developing practical applications for ureolytic carbonate precipitation using plant-derived urease enzyme.
RSUM : La cimentation de sable par du carbonate de calcium (CaCO3) produit par lenzyme urase obtenue partir de plantes a
t ralise en laboratoire. Les avantages dutiliser de lurase obtenue de plantes plutt que de lurase produite microbilogiquement
pour produire la cimentation carbonate sont la petite taille de lenzyme qui permet la pntration dans les sols fins et rend le
processus moins sujet au colmatage biologique et la disponibilit 100% du carbone prsent dans le substratum pour conversion en
CaCO3. Des essais en colonnes ont t raliss sur deux sables de silice dits Ottawa 20-30 et F-60 (plus fin). Les chantillons ont t
prpars de diffrentes manires et ont atteint des degrs de cimentation varis et des productions de carbonate diffrentes. Les
rsultats des essais de compression triaxiale sur des chantillons ciments et des chantillons non-ciments indiquent que les premiers
sont beaucoup plus rsistants. Ces essais confirment que lurase obtenue partir de plantes peut tre utilise pour induire une
cimentation carbonate dans les sables. De plus ces essais ont permis de didentifier les facteurs considrer pour dvelopper des
applications pratiques pour lutilisation de la prcipitation carbonate urolytique en utilisant lurase issue de plantes.
KEYWORDS: carbonate, cementation, urease, calcite, soil improvement
1. INTRODUCTION
1.1 Background
The potential for using plant-derived urease enzyme to cement
sands by inducing calcium carbonate (CaCO3) precipitation has
been demonstrated through a series of laboratory column tests
on two different gradations of silica sand. The use of
microbially induced carbonate precipitation (MICP) to cement
cohesionless soils has recently received substantial attention
from geotechnical researchers (Burbank et al. 2012, Chou et al.
2011, Dejong et al. 2010, Harkes et al. 2010, van Paassen et al.
2010). The MICP mechanism most often discussed in the
literature and most advanced in terms of field application is
hydrolysis of urea (ureolytic hydrolysis). MICP via ureolytic
hydrolysis relies on microbes to generate urease enzyme, which
then serves as a catalyst for the precipitation reaction. The use
of plant-derived urease (enzymatic ureolytic hydrolysis) to
induce CaCO3 precipitation eliminates the need for microbes in
the CaCO3 precipitation process.
Besides eliminating the need to nurture urease-producing
microbes, enzymatic ureolytic hydrolysis offers several other
advantages over ureolytic MICP. Applications of ureolytic
MICP on clean sands in laboratory column tests and limited
field tests have encountered significant practical difficulties,
including bioplugging (permeability reduction accompanying
induced mineral precipitation) and generation of a toxic waste
product (ammonium salt) (Harkes et al. 2010, van Paassen et al.
2008). Bioplugging not only limits the distribution of

precipitation agents within the soil but also makes flushing of


the waste product from the soil a difficult, energy intensive task.
Due to these limitations, mass stabilization of soil using
ureolytic MICP remains problematic. Furthermore, the microbes
that produce the urease enzyme cannot readily penetrate the
pores of soils smaller than medium to fine sand, limiting the
minimum grains size of soils amenable to ureolytic MICP to
clean fine sands or coarser graded soils. The small size (on the
order of 12 nm) of the urease enzyme suggests that CaCO3
precipitation by enzymatic ureolytic hydrolysis will be less
susceptible to bio-plugging and will be able to penetrate finer
grained soils, perhaps into the silt-sized particle range,
compared to MICP processes.
1.2 Sustainability of Ground Improvement Practices
Finding effective solutions to ground improvement challenges is
becoming increasingly complex due to sustainability
considerations. Established materials and methods often need
to be either replaced or supplemented by innovative materials
and environmentally-friendly practices to address sustainability
considerations. One example of a common building material
that poses significant sustainability concerns is Portland cement.
Portland cement is widely used in ground improvement
applications. Unfortunately, Portland cement production is
extremely energy intensive and a major source of emissions of
carbon dioxide (CO2), as well as of sulfur and nitrogen oxides.
MICP has been explored recently as an alternative to Portland
cement for ground improvement. Reductions in the use of

2489

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Portland cement through either direct substitution or


complementary use of MICP could contribute considerably
towards reduction in CO2 emissions. Research suggests that
cementation using MICP can address a number of important
geotechnical problems in granular soils, including slope
stability, erosion and scour, under-seepage of levees, the
bearing capacity of shallow foundations, tunneling, and seismic
settlement and liquefaction (Dejong et al. 2010, Harkes et al.
2010, Kavazanjian and Karatas 2008, van Paassen et al. 2010).
1.3 Ureolytic MICP
MICP attempts to create a cemented soil mass by precipitating
calcium carbonate from the pore fluid such to form cementation
bonds at the interparticle contacts (van Paassen et al. 2010,
DeJong et al. 2006). Karatas et al. (2008) have identified several
mechanisms for MICP. The MICP mechanism that has garnered
the most attention and is most advanced in terms of
development is ureolytic hydrolysis, or ureolysis (Chou et al.
2011, DeJong et al. 2006, van Paassen et al. 2010, Whiffin et al.
2007). Ureolytic MICP has typically been accomplished using a
technique best described as biogrouting (Harkes et al. 2010, van
Paassen et al. 2010), wherein bacteria and nutrients are mixed in
a tank ex-situ and then injected into the soil followed by a
fixation fluid to foster microbial attachment to soil particles and,
finally, by a calcium-laden cementation fluid. Ureolytic MICP
by stimulation of indigenous bacteria has also been reported in
the literature (Burbank et al. 2012).
1.4 Agricultural Urease
Urease is a widely occurring hexameric protein found in many
microorganisms, higher order plants, and some invertebrates.
The enzyme is approximately 12 nm in dimension (Blakely &
Zerner 1984). The small size of a solubilized urease enzyme
affords it a distinct advantage over carbonate cementation
methods that employ ureolytic microbes in cases that require
penetration into very small pore spaces as nearly all known
bacteria are greater than 300 nm in diameter, with the majority
in the range of 500-5000 nm. Several families of common
plants are very rich in urease, including some varieties of beans,
melons and squash, and the pine family (Das et al. 2002).
Extraction of urease enzyme from most urease containing plants
has been shown to be very simple (Srivastava et al. 2001) and
the enzyme is readily available from laboratory suppliers.
It is well-established that urease can occur as both an intraand extra-cellular enzyme (Ciurli et al. 1996, Marzadori et al.
1998). Free soil urease (i.e. urease not bound to any living
organism), generally derived from dead and decaying
microorganisms and possibly from plant sources, readily occurs
apart from the host microorganism and, upon absorptive
association with soil particles, can persist for long periods of
time without degradation or loss of function (Pettit et al. 1976).
By contrast, exogenously added urease (i.e. urease added as a
free enzyme) has a limited lifespan and its activity and function
decrease with time (Marzadori et al. 1998, Pettit et al. 1976).
This limited lifespan is potentially advantageous in some
engineering applications as the enzyme can naturally degrade
thereby eliminating long term impacts to the ecosystem.
2. METHODS
2.1 Ottawa 20-30 Sand
Laboratory column tests were conducted using plant derived
urease to induce CaCO3 precipitation in Ottawa 20-30 sand
These tests were carried out in 6x 2 (152 mm x 51 mm)
acrylic tubes and membrane-lined 2.8 x 6 (71 mm x 152 mm)
split molds (for creating specimens for triaxial testing). Three
acrylic tubes and two columns for triaxial testing were filled
with 20-30 Ottawa silica sand (mean grain size 0.6 mm,

2490

coefficient of uniformity 1.1) and treated as follows: tube #1:


the sand was dry pluviated via funnel at 3 (76 mm) drop
height and then received 5 applications of a cementation
solution containing urea and calcium chloride mixed with
1.4g/L enzyme (total solution volume 300 ml); tube #2: sand
was added in same manner as tube #1 and then received 2
applications ( 150 ml total) of the same cementation solution
mixed with 1.4g/L enzyme; tube #3: the lower-third of tube was
filled with sand and dry enzyme ( 3g), the remainder of the
tube contained dry pluviated sand without enzyme, and the tube
then received 2 applications ( 150 ml) of the cementation fluid
with no enzyme added. The cementation fluid composition was
based upon stochiometry and experience with microbial urease
cementation, e.g. DeJong et al. (2007), Whiffin et al. (2008).
Approximately 100 mL of a pH=7.8 solution containing 383
mM urea (reagent grade, Sigma-Aldrich), 272 mM CaCl2-2H2O
(laboratory grade, Alfa Aesar) was used for the first application
in each acrylic tube. Subsequent applications employed
approximately 50 mL of a pH=7.6 solution containing 416 mM
urea and 289 mM CaCl2-2H2O. Solution concentrations, while
variable, were formulated within a reasonably similar range as a
matter of convenience. In each application, the cementation
fluid was poured into the top of the acrylic tube with the bottom
closed off. The cementation fluid was allowed to stand, loosely
covered, in the acrylic tube for at least 24 hours and then
drained out the bottom of the cylinder. The next application
followed immediately after drainage was complete. Drainage
was accomplished by puncturing the base of the cylinder with a
20-gauge needle. When drainage was complete, the needle was
removed and the puncture was plugged with a dab of silicone.
Occasionally, the needle became plugged and an additional
needle was inserted through the base. The triaxial columns
were filled with sand in the same manner as tube 1 and then
received 2 applications (each application 250 ml) of
cementation solution with 1.4g/L enzyme.
In each application of cementation fluid, the fluid was
added until it rose to approximately -inch (12-mm) above the
soil line. After 2 applications, tubes #2 and #3 were allowed to
air dry for several days and then analyzed. Experimentation
with tube #1 was continued for several more days as three more
batches of cementation fluid were applied. The last 2
applications of cementation fluid were allowed to slowly drain
through the needle in the base immediately after application
rather than sit for 24 hours (drainage rate 10-25ml/hour). The
triaxial columns were allowed to stand for at least a week after
the second cementation fluid application and then drained.
After drainage was complete, the triaxial columns were
moved to a triaxial testing device. After draining the specimens
from the acrylic tubes and after the completion of the triaxial
tests, all samples were triple washed with de-ionized water.
Tubes #2 and #3 were separated in 3 layers, while tube #1 was
separated into six layers (for better resolution). Each layer from
the specimens in the acrylic tubes and the entire mass of the
triaxial specimens were acid washed to determine CaCO3
content by oven drying for 48 hours, weighing, digesting with
warm 1M HCl, washing, drying, and reweighing to determine
carbonate mineral content.
Several of the cemented specimens were analyzed for
mineral identification using X-Ray Diffraction (XRD). Samples
were ground in an agate mortar and pestle and powdered onto a
standard glass slide for analysis. Scanning electron microscopy
(SEM) imaging was performed on intact cemented chunks of
material with an Agilent 8500 Low-Voltage SEM (LV-SEM). A
LV-SEM is a field emission scanning electron microscope
capable of imaging insulating materials, such as organic and
biological substances without the need for a metal coating and
without causing radiation damage to samples.
2.2 Ottawa F-60 Sand
A triaxial column was prepared using Ottawa F-60 silica sand
(mean grain size 0.275 mm, coefficient of uniformity 1.74) to

Technical Committee 211 / Comit technique 211

investigate enzymatic ureolytic CaCO3 precipitation in a finer


grained material. The specimen was prepared in the same
manner as described for the triaxial columns for the Ottawa 2030 sand. The cementation fluid for the first of the two
applications contained approximately 2.0 g/L enzyme, 400 mM
urea (reagent grade, Sigma-Aldrich), 300 mM CaCl2-2H2O
(laboratory grade, BDH) at pH=7.7. The fluid for the second
application contained 1 M urea-CaCl2-2H2O solution at pH=7.8
without any enzyme. After the test, the triaxial specimen was
washed and subject to acid digestion in the same manner as the
Ottawa 20-30 triaxial specimens.

3 during sample preparation. In the top layer of tube 3, where no


urease was mixed with the sand, carbonate precipitation was
nearly undetectable. There was no visual evidence of
precipitation and practically no measurable change in weight of
this layer after acidification (weight change = 0.23%). In the
bottom layer of tube 3, where 3 g of dry enzyme was mixed
with the soil, there was a weight change of 1.7% following acid
washing. The middle layer of this specimen had a minor change
in weight (0.58%), possibly due to uneven distribution of the
layers during preparation or splitting of the specimen or to
upward migration of urease from the bottom layer.

3. RESULTS

XRD analysis, presented in Figure 1, confirms that calcite is the


mineral phase present in the cemented soil chunks. LV-SEM
images, presented in Figure 2, show silica (quartz) sand
particles cemented with calcium carbonate and various
morphological features associated with the cementation process
on the silica surface.

3.1 Acrylic Tubes


Approximately 100 ml of cementation solution was delivered
per application for the first application in each acrylic tube.
However, the amount of solution the tube would accept was
notably reduced in subsequent applications, when less than 75
ml was typically required to fill the tubes to inch (12 mm)
above soil line. At the conclusion of the experiment,
precipitation was visible along the entire length of tubes 1 and
2. Internally the cementation was variable, with some highly
cemented zones and other zones with little to no cementation.
Tube 1 yielded mostly small, loose chunks of sand with
strong effervescence upon digestion. Most of this column
appeared un-cemented and exhibited unusually viscous behavior
when wet. A fairly large (compared to column diameter) piece
of strongly cemented sand (not breakable without tools) formed
in the deepest layer of tube 1. Tube 2 had many small chunks
of weakly cemented sand with strong effervescence upon
digestion. Tube 3 had little to no precipitation in the top layer
(i.e. this layer did not show any indication of carbonate upon
acid digestion.) The deepest layer of tube 3 contained many
pieces of weakly cemented sand that effervesced strongly upon
digestion. The middle layer of tube 3 contained a few pieces of
cemented sand that effervesced moderately upon digestion. The
results from the acid washing are presented in Table 1.

2
3

Layer

Tube #

Table 1. Results from Experiment Set 1 using 20-30 Ottawa silica sand

1
2
3
4
5
6
1
2
3
1
2
3

Summary of Results
Amt.
Total
Weight
of
Amt.
Change via
CaCO3 CaCO3
Digestion
(g)
(g)
11%
3.57
3.8%
1.67
2.7%
1.73
11.8
2.1%
1.40
2.3%
1.74
2.0%
1.64
0.76%
0.63
2.07
0.65%
0.69
0.49%
0.75
0.23%
0.31
0.58%
0.63
3.57
1.7%
2.63

Figure 1. XRD results from cemented sand sample (top plot). Quartz &
calcite standards (middle & bottom plot, respectively).

Theor.
Max
CaCO3
(g)

14.5

4.35
4.35

The theoretical maximum CaCO3 content is the stoichiometric


maximum balanced on initial concentrations. The primary
experimental differences between the tests are (1) the number of
applications of cementation fluid and (2) the manner in which
the urease was delivered. The results indicate that there is
greater carbonate precipitation with increasing number of
applications, as expected. The data show more precipitation in
(or on) the top layer of tubes 1 and 2 but not in tube 3, as the
enzyme was physically confined to the lower-third layer in tube

2491

Figure 2. LV-SEM images a.) Well-grown and cementing calcite crystals;


b.) Cementing calcite crystals at inter-particle contact; c.) Indention of quartz
surface (blue arrows) and nucleation of calcite crystals (red arrows); d.) Calcite
crystal growing on quartz surface.

3.2 Triaxial Columns


The three triaxial sand columns (2 Ottawa 20-30 sand columns
and 1 Ottawa F-60 sand column) were tested in drained triaxial
compression prior to acid digestion. All three columns were
able to stand upright after removal of the split mold. The results

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

of the triaxial compression tests performed on the 20-30 Ottawa


sand are presented in Figure 3 and the results for the F-60
Ottawa sand are presented in Figure 4. The carbonate cement
content for one of the 20-30 silica sand columns was 2.0%
CaCO3 (by weight). The carbonate content of the other 20-30
Ottawa sand column could not be quantified due to unintended
sample loss. The carbonate cement content for the finer grained
F-60 Ottawa sand was 1.6% CaCO3 (by weight). The results
show substantial strength increase for all 3 sand columns tested.

2.0% CaCO3
CaCO3 not quantified

Figure 3. p-q plot failure envelopes for 20-30 silica sand: Cemented
(Dr = 60%); Uncemented (Dr = 60%)

1.6% CaCO3

Figure 4. p-q plot failure envelopes for F-60 silica sand: Cemented (Dr
= 35%); Uncemented (Dr = 37%);

4. CONCLUSION
Sand column tests at Arizona State University have shown that
agriculturally-derived urease can be used to induce calcium
carbonate precipitation in sand. Sand columns were developed
using Ottawa 20-30 and F-60 sand and three different
preparation methods: dry pluviation followed by percolation of
a calcium-urease-urea cementation solution, pluviation into a
calcium-urease-urea cementation solution, and mixing the sand
with urease prior to pluviation with a calcium-urea solution.
Cementation was observed in all of the columns. XRD and
SEM testing confirmed that calcium carbonate (specifically
calcite) was the cementing agent. Acid digestion showed that
increased applications yielded correspondingly greater

carbonate precipitation. The quality of cementation, as


determined by the effort needed to break apart cemented chunks
of sand, varied depending on the sampling location within the
column. Triaxial test results on cemented columns showed
substantial strength increase over non-cemented columns at the
same relative density.
5. REFERENCES
Baumert, K.A., Herzog, T., Pershing, J., 2005. Navigating the Numbers
Greenhouse Gas Data & International Climate Policy World
Resources Institute
Blakely, R.L. and Zerner, B., 1984. Jack Bean Urease: The First Nickel
Enzyme. Journal of Molecular Catalysis 23, 263292.
Burbank, M., Weaver, T., Lewis, R., Williams, T., Williams, B. and
Crawford, R. 2012. Geotechnical Tests of Sands Following
Bio-Induced Calcite Precipitation Catalysed by Indigenous Bacteria
ASCE JGGE 132, DOI: 10.1061/(ASCE)GT.1943-5606.0000781.
Chou, C.W., Seagren, E.A., Aydilek, A.H., and Lai, M. 2011.
Biocalcification of Sand through Ureolysis. JGGE 137, 11791189.
Ciurli S., Marzadori C., Benini S., Deiana S. and Gessa C., 1996.
Urease from the soil bacterium Bacillus pasteurii: Immobilization on
Ca-polygalacturonate. Soil Biol. & Biochem. 28, 811-817.
Das, N., Kayastha, A.M. and Srivastava, P.K., 2002. Purification and
characterization of urease from dehusked pigeonpea (Cajanus cajan
L.) seeds. Phytochemistry 61 (5), 513-521.
DeJong, J.T., Fritzges, M.B., and Nusslein, K. 2006. Microbially
Induced Cementation to Control Sand Response to Undrained
Shear. ASCE JGGE 132, 13811392.
DeJong, J.T., Mortensen, B.M., Martinez, B.C., and Nelson, D.C., 2010.
Bio-mediated soil improvement. Ecological Engineering 197-210.
Harkes, M. P., van Paassen, L. A., Booster, J. L., Whiffin, V. S., and
van Loosdrecht, M. C. M. 2010. Fixation and distribution of
bacterial activity in sand to induce carbonate precipitation for
ground reinforcement. Ecological Engineering 36 (2), 112117.
Karatas, I., 2008. Microbiological Improvement of the Physical
Properties of Soils. PhD. Dissertation, Department of Civil,
Environmental, and Sustainable Engineering Arizona State
University, Tempe, AZ.
Kavazanjian, E. and Karatas, I., 2008. Microbiological Improvement of
the Physical Properties of Soil, 6th International Conference on
Case Histories in Geotech. Eng., Arlington, VA August 11-16.
Krogmeier, M.J., McCarty, G.W. and Bremner, J.M., 1989.
Phytotoxicity of foliar-applied urea. Proc. Natl. Acad. Sci., USA
86, 81898191.
Marzadori, C., Miletti, S., Gessa, C. and Ciurli, S., 1998.
Immobilization of jack bean urease on hydroxyapatite: urease
immobilization in alkaline soils. Soil Biol. & Biochem. 30 (12),
1485-1490.
Pettit N. M., Smith A. R. J., Freedman R. B. and Burns R. G., 1976.
Soil urease: activity, stability, and kinetic properties. Soil Biol. &
Biochem. 8, 479-484.
Srivastava, P.K. and Kayastha, A.M., 2001. Characterization of gelatinimmobilized pigeon pea urease and preparation of a new urea
biosensor. Biotechnology and Applied Biochemistry 34, 55-62.
van Paassen, L.A., Ghose, R., van der Linden, T.J.M., van der Star,
W.R.L., and van Loosdrecht, M.C., 2010. Quantifying Biomediated
Ground Improvement by Ureolysis: Large-Scale Biogrout
Experiment. ASCE JGGE 136, 17211728.
van Paassen, L.A., Daza, C.M., Staal, M., Sorokin, D.Y. and van
Loosdrecht, M.C., 2008. In situ soil reinforcement by microbial
denitrification. 1st Int. Conf. on Bio-Geo-Civil Engineering,
Netherlands, 124-133, June 23-25.
Whiffin, V.S., van Paassen, L.A., and Harkes, M.P., 2007. Microbial
Carbonate Precipitation a Soil Improvement Technique.
Geomicrobiology 24, 1-7.

2492

Experimental investigation on bearing capacity of geosynthetic encapsulated stone


columns
tude exprimentale sur la capacit portante des colonnes de pierre gosynthtiques encapsules
Hataf N., Nabipour N.

Department of Civil and Environmental Engineering, Shiraz University, Shiraz, Iran

aBstract: civil engineers have developed different soil improvement techniques in recent decades to improve the bearing
capacity of soft soils loaded by foundations and reduce soil settlement. a method for increasing the bearing capacity of foundation
soil is the use of stone columns. however, one of the major weaknesses in use of stone columns in loose soils is lack of confinement.
Using geosynthetic reinforcement to compensate low confinement pressure in these soils, is a solution to this problem. this paper
presents the results of an experimental study on the improvement of the bearing capacity of stone columns reinforced by
geosynthetics. in this study the influences of three variables have been investigated, including: surrounding soil types (i.e. clay and
sand), stone column aggregate size and length of reinforcement. having mentioned these variables, the results showed that
encapsulating stone column with geosynthetic is more effective in cohesive soil compared to granular soil. the results of the
experiments revealed that the coarser the aggregate the better behavior is expected for the stone column. the results also showed that,
reinforcing half height of stone columns is the optimal encapsulating length.
rsUm: ces dernires dcennies, les ingnieurs civils ont dvelopp diffrentes techniques pour l'amlioration de la capacit
portante du sol mou ainsi que celles des fondations. Une des techniques couramment utilise permettant l'augmentation de la capacit
portante des sols et des fondations est l'utilisation des colonnes de pierre. cependant, l'une des grandes faiblesses de l'utilisation de
colonnes de pierre dans les sols mous est le manque de confinement. l'utilisation de renforts gosynthtiques permet de compenser
pour la faible pression de confinement. cet article prsente les rsultats d'une tude exprimentale sur l'amlioration de la capacit
portante des colonnes de pierre renforces par des mthodes gosynthtiques. dans cette tude, l'influence de trois variables ont t
tudies, notamment: le type de sol environnant (i.e. argile et sable) ainsi que la longueur de l'armature de renforcement. les rsultats
dmontrent que l'emploi de la colonne en pierre avec encapsulation gosynthtique est plus efficace dans un sol consistant compar
aux sols granulaires. de plus, les rsultats de ces expriences ont rvl que plus la rugosit de l'agrgat augmente, plus le
comportement des colonnes de pierres est amlior. finalement, les rsultats indiquent que la longueur d'encapsulation optimale est
atteinte en renforant la hauteur mdiane des colonnes de pierre.
KeYWords: stone column, bearing capacity, geosynthetic, reinforcement.
1

2.1

introdUction.

in recent years with increasing in population density in specific


locations,the value of land has increased signifacntly. this has
made the use of areas with soft soils inevitable.
due to the lack of bearing resistance in these soils, different
methods of soil improvement techniques, including stone
columns as a method of strengthening the loose soil are used.
stone columns behavior has been studied experimentally,
theoretically and numerically by many researchers (Bergado and

teerawattanasuk2008, Guetif et al. 2007, castro and sagaseta2011)

however, one of the major weaknesses in use of


stonecolumns in loose soils is lack of confinement. this lead
researchers and practioners to use geosentitics to increase
confinement of column, compansating the scarcity of studding
around reinforced stone columns (malarvizhi and ilamparuthi2007,
Gniel and Bouazza2009, Gniel and Bouazza2010). in this study the
parameters affecting the behavior of reinforced stone columns
have been investigated. these parameters are reinforced length,
column material and surrounding soil type.
2

laBoratorY settinGs

since the focus of this research was on the laboratory results,


the phyical model, is described, firstly.

Test apparatus

a cylindrical tank (height=1.0 m and dia.=1.0m) filled with soil


was used as the soilenvironment. stone column run in the
middle of the tank.the static loading system consists of a
loading arm and weights were used(razavi and hataf, 2003) to
determine the bearing capacity of a circular foundations resting
on stone column, figure 1.
2.2

Soil tested

to test and evaluate the behavior of reinforced stone columns in


loose soil, two soil types were used, a clay soil as cohesive soil
and a sandy soil as granular soil.
physical properties of the soils are listed in table 1.
2.3

Specimens preparation

to prepare the soil and column, first two 10 cm soil layers has
been poured in the tank and compacted using 20 strokes caused
by dropping a 50 n weight attached to a wooden handle from a
distance of 40 cm as the substrate layer.the next layerswere
compacted with 10 strokes from 10cm distance to provide loose
soil.
toprepare the stone column an open ended hollow
cylindrical pipe with a diameter of a little more than the
diameter of the stone column was used.after that the cylinder
was placed at its position and the surrounding was filled slowly
with soil.

2493

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

figure 1.laboratory setting for model testing.


table 1. physical properties of the soils tested.
parameter

Clay

Sand

friction angle

26.0

35.0

cohesion (Kn/m2)

5.0

0.0

Unit weight(Kn/m3)

15.0

16.0

liquid limit(%)

44.5

plasticity index (%)

20.0

then until reaching up to the surface level the stone column


filled with stone aggregates in 10 cm layers.after filling each
layer the cylinder pulled out about 10 cm and aggregates were
poured in and compacted with 40cm length rod.
three types of aggregates were used to fill the stone
columns. these are shown in figure 2.
stone columns with no reinforcement, half-length
reinforcement and full-length reinforcement were prepared for
testing. a commercially available geogrid was used for
reinforcement.
r3

r2

reinforcementfor optimal strength. this was achieved by


changing the length of reinforcement compared to the column
length as full-length, half-length and non-reinforced.
test results indicated that in both types of surrounding soils
and for all sizes of column aggregate materials, it is enough to
reinforce only half length of the column to achieve desired
bearing improvement. however the improvement ratein
cohesive soils is more noticeable. this can be related to the fact
that the confining pressure in the bottom of the column is higher
than that in the upper parts of the column due to higher
overburden pressure. By increasing the confining pressure in the
upper parts of the column by installing reinforcement, the radial
strain reduces and as the result, it reduces the side contact
pressure between the soil and stone column. this in turn causes
just vertical distribution of the stresses to the layer below the
column and not distributing of stresses to the surrounding soil.
this obviously causes more vertical deflections in the below
layers of soil and less in the upper layers.
in the half-length reinforced column by increasing in
stresses,a small amount of inflation on the side layers are
observed which results in increase in lateral soil friction and so
the stresses spreads over a larger surface of the soil and it results
the deflection not to increase below the column but spread in
larger area homogeneously.

F
Fmax(no)

3-a fine aggregate material

r1

F
Fmax(no)

figure 2. different aggregates used as stone columns materials.

test resUlts

test results as load settlement curves for stone columns


embedded in cohesive and granular soils are illustrated in
figures 3 an 4. in this figures f-ri, h-ri and no-ri stand for
full-length, half-length and no reinforced column, respectively.
loads were normalized to maximum load obtained for
unreinforced column in each case and settlements were
normalized to radius of stone column.
as it can be seen from these figures it is obvious that
reinforcement improve the bearing capacity of stone columns in
both cohesive and granular soils. the reinforcement however is
more effective in cohesive soil than in granular soil.
further numerical studies (not presented here) showed that
the effect of viscosity is reduced with the increase in cohesion
of soil which in turn caused increase in the confining pressure
of surrounding soil. therefore this results in decrease in stone
column material to spread out within the surrounding soil.
the most important variable in this study was to
experimentally and practically examine the optimal lengthof

2494

3-b medium aggregate material

Technical Committee 211 / Comit technique 211

F
Fmax(no)

as it was mentioned earlier one of the variables in this study


was the size of the column aggregate materials. the results of
tests on the same stone column conditions but with different
size of stone column materials are illustrated in figure 5 and 6
for cohesive and granular surrounding soils, respectively.it can
be seen that keeping all conditions constant, there was an
increase in bearing resistance of the column with increasing
grain size dimension of column material. however the
improvement due to the use of geosynthetic reinforcement was
the same for all column material sizes.

F
Fmax(no)

3-c coarse aggregate material


figure 3. test results for stone columns embedded in cohesive soil.

F
Fmax(no)

5-a non-reinforced stone column

F
Fmax(no)

4-a fine aggregate material

F
Fmax(no)

5-b half-length reinforced column

F
Fmax(no)

4-b medium aggregate material

F
Fmax(no)

5-c full-length reinforced column


figure 5. test results for different stone columns materials embedded in
cohesive soil.

3-c coarse aggregate material


figure 4.test results for stone columns embedded in granular soil.

2495

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

behavior of stone column encapsulated by geosynthetic in its


entire length was compared to partially encapsulated stone
column behavior. the results showed that, reinforcing half
height of stone columns in both types of soils, especially in
clay, is the optimal encapsulating length. this finding is
significant regarding the economical and efficiency of use of
stone columns as a soil improvement technique.

F
Fmax(no)

Bergado, d.t. and teerawattanasuk, c. 2008. 2d and 3d numerical


simulations of reinforced embankments on soft ground. Geotextiles
and Geomembranes, 26, 3955.
castro, J., sagaseta, c. 2011. consolidation and deformation around
stone columns: numerical evaluation of analytical solution.
Computers and Geotechnics 38,354362.
Guetif, a. Z., Bouassida, m., debats, J. m. 2007.improved soft clay
characteristics due to stone column installation. Computers and
Geotechnics 34,104111.
Gniel, J., and Bouazza, a. 2009. improvement of soft soils using
geogrid encased stone columns. Geotextiles and Geomembranes27,
167175 .
Gniel, J., and Bouazza, a. 2010. construction of geogrid encased stone
columns: a new proposal based on laboratory testing. Computers
and Geotechnics, 28, 108118 .
hataf, n. and razavi, m. r., 2003. model test and finite element
analysis of ring footings on loose sand. Iranian J. of Science &
Technology, 27(B1), transaction B, 1-11.
malarvizhi, s. n. and ilamparuthi, K. 2007. comparative study on the
behaviour of encased stone column and conventional stone
column.Soils and Foundations, 47 (5), 873885.

6-a non-reinforced stone column

F
Fmax(no)

6-b half-length reinforced column

F
Fmax(no)

6-c full-length reinforced column


figure 6. test results for different stone columns materials embedded in
granular soil.

references

conclUsion

one of the recent methods for increasing the bearing capacity of


foundation soil is the use of vertical stone columns. stone
columns consist of a stiffer material or aggregates, compared to
the surrounding soils, which are usually vibrocompacted into
the soil. these columns increase the bearing capacity of the soil
significantly. compared to concrete or steel piles inclusion, for
soil improvement, this technique is more economical and needs
to be studied further. however, one of the major weaknesses in
use of stone columns in loose soils is lack of confinement. this
leads to use reinforcement to compensate low confinement
pressure in these soils. Because of the lack of experimental
studies on the behavior of reinforced stone columns, an
experimental study has been performed. it was shown that the
use of stone columns improves the soil bearing capacity,
significantly. the results showed that encapsulating stone
column with geosynthetic is more effective in cohesive soils
compared to granular soils. three types of stone column
materials were used with different aggregate dimensions. the
results of the experiments revealed that the coarser the
aggregate the better behavior is expected for the stone column.
although the increase in grain size should not be more than two
percent of stone column diameter. on the other hand, the

2496

Performance and Prediction of Vacuum Consolidation Behavior at Port of Brisbane


Avantages et prdictions de comportement due a la consolidation sous vide au port de Brisbane
Indraratna B., Rujikiatkamjorn C., Geng X.

Centre for Geomechanics and Railway Engineering, University of Wollongong, Wollongong City, NSW Australia, ARC
Centre of Excellence in Geotechnical Science and Engineering, Australia

Ameratunga J.

Coffey Geotechnics, 47 Doggett Street, Newstead, QLD. 4006, Australia

aBstract: due to a projected increase in trade activities at the port of Brisbane, new berths on fisherman islands at the mouth of
the Brisbane river will be constructed in the outer area (235ha) close to the existing port facilities via land reclamation. a vacuum
assisted surcharge load in conjunction with prefabricated vertical drains was choosen to reduce the required consolidation time. the
features of the combined vacuum and surcharge fill system and the construction of the embankment are described in this paper. a
comparison of the performance of the vacuum combined surcharge loading system with a standard surcharge fill emphasizes the
obvious advantages of vacuum consolidation. field data is presented to show how the embankment performed during construction.
an analytical solution for radial consolidation incorporating both time-dependent surcharge loading and vacuum pressure is
employed to calculate the settlements and associated excess pore pressures of the soft holocene clay deposits.
rsUm : laugmentation des activits de commerce au port de Brisbane ncessite la construction, proximit des terminaux
existants, de nouveaux postes de quais dans les iles fisherman a lembouchure de la rivire de Brisbane sur une superficie de 235 ha
gagne sur la mer. Un chargement sous vide contrle, associe a des drains prfabriqus, a t appliqu pour rduire le temps de
consolidation. larticle dcrit les caractristiques de la technique de consolidation sous vide associe au chargement par remblaiement
et la construction du remblai. Une comparaison entre la consolidation sous vide associe au remblaiement et le pr chargement
classique montre clairement les avantages en faveur de la consolidation sous vide. les donnes enregistres sur le site illustrent le
comportement du remblai durant la consolidation. Une solution de consolidation horizontale tenant compte du chargement et de la
pression sous vide est prsente en vue de prdire le tassement et lexcs de la surpression interstitielle du dpt dargile molle de
lholocne.
KeYWords: consolidation, soil improvement, vertical drains, vacuum.
1

introdUction

the port of Brisbane is one of the australias largest


commercial ports located at the entrance of the Brisbane river
at fisherman islands. With demand in commercial activities, a
new outer area (235ha) is being reclaimed for major expansion
to maximise the available land, and to provide the maximum
number of berths suitable for container handling for servicing
regional importers and exporters. in this area, the soil profile
mainly consists of compressible clay deposits over 30m in
thickness with very low undrained shear strength (<15 kpa at
shallow depth). the strength of dredged mud had a much lower
strength depending on the placement time and the thickness of
capping material. Without surcharge preloading, it is estimated
that the consolidation time could be more than 50 years with
overall settlements of 2.5-4.0m. therefore, vacuum
consolidation with prefabricated vertical drains (pVds) was
suggested to accelerate the consolidation process and to
minimise lateral deformation adjacent to the moreton Bay
marine park (indraratna et al. 2011).
the effectiveness of the vacuum preloading assisted by
pVds has been illustrated by chu et al. (2000) and chai et al.
(2005). in this technique, vacuum pressure can propagate to a
greater depth of the subsoil via pVd length. also, extended
consolidation time due to stage construction can be minimized
(indraratna et al. 2005). the surcharge fill height can be reduced
by several metres, if a vacuum pressure (at least 70 kpa) is
applied and sustained (rujikiatkamjorn et al. 2008). the
embankment construction rate can be increased and the number
of construction stages can be reduced (Yan and chu 2003).
once the soil has increased its stiffness and shear strength due
to consolidation, the post-construction settlement will be
significantly less, thereby eliminating any risk of differential
settlement of the overlying infrastructure (shang et al. 1998).
to the authors knowledge, there is no comprehensively

reported case history where both the conventional surcharge


preloading and vacuum technique have been applied in the same
area with distinct variation of drain types and spacing.
in this paper, the performance between the vacuum and
non-vacuum areas has been compared based on the measured
settlements, excess pore pressures and lateral displacements.
the influences of drain spacing, drain types and type of soil
improvement are discussed based on the observed degree of
consolidation. the analytical solutions for radial consolidation
considering both time dependent surcharge loading and vacuum
pressure are proposed to predict the settlement and associated
excess pore pressure.
2

General description of emBanKment


characteristics and site conditions

at the port of Brisbane, to evaluate the performance of the


vacuum consolidation system with the non-vacuum system
(pVd and surcharge load), a trial area (s3a) shown in fig. 1
was partitioned into Wd1-Wd5 (non-vacuum areas) and Vc1Vc2 (Vacuum areas). after placing the dredged fill, the mud
was capped off with a 2-3m layer of dredged sand, which acted
as a working platform for pVd installation machine, whilst
serving as a drainage layer.
the upper holocene sand beneath the reclaimed dredged
mud was about 2m thick, followed by the holocene clay layer
with different in thickness from 6m to 25m. a pleistocene
deposit containing highly over-consolidated clay underlies the
softer holocene clay layer. site investigation techniques
including cone penetration/piezocone tests, dissipation tests,
boreholes, field vane shear tests and oedometer tests were
carried out to assess the relevant consolidation and stability
design parameters. the water contents of the soil layers were
similar to or exceed their liquid limits. the vane tests show that
the undrained shear strength of the reclaimed dredged mud and

2497

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

the holocene clays were from 5 to 60 kpa. the compression


index changed from 0.1 to 1.0. the coefficient of consolidation
in vertical direction is similar to that in horizontal direction (ch)
for the remoulded dredged mud layer, while cv/ch is about 2 for
the holocene clay layer.
70m

WD5A
35m

84.5m

41m

WD5B

MS22-1

WD1 VWP2-WD1

WD5B

70m

MS15-1

WD2

MS19VWP5

WD3
MS18-1
MS27WD3

VWP1-WD2

70m

MS20-VWP5

VC1

VC1-2

interpretation of field resUlts

the embankment performances including settlements and


excess pore pressures together with the staged construction of
the embankments are depicted in fig. 2. it would be observed
that the trends are very comparable where the settlement
occured more quickly at the early stage of consolidation. the
amount of final settlement depends on the clay thickness and
embankment height. the highest settlement was measured in
the Wd4 area having the greatest clay thickness (19-26m),
whereas the lowest settlement was in the Wd5a area in which
the clay layer was relatively thin (8-12m).

VWP4-WD4

MS17-1

155m

WD4
MS16-1

VC2
VC2-1

84.5m

pipes and the pumps that represent the vacuum system. the
horizontally pipes offered the desired uniform distribution of
suction beneath the membrane. the measured suction varied
from 60 kpa to 75 kpa, and no air leaks were observed during
vacuum application that ensured the intact seal provided by the
membrane. a vacuum pressure of 70kpa was applied after 40
days.

50m

MS28-VC1
169m
210m

Embankment Height (m)

Surface settlement plates


Piezometers
Inclinometers

figure 1. site layout for s3a with instrumentation plan

the surcharge preloading system was adopted for the inner


areas (Wd1-Wd5) while, in the outer area (Vc1 and Vc2)
close to the marine park, the technique of vacuum combined
preloading was selected to control lateral displacement in order
to minimise disturbance of the nearby marine habitats. stringent
design criteria were adopted for the design and construction of
emabankment over the soft holocene deposits: (a) service load
of 15-25 kpa, (b) maximum residual settlement less than 250
mm over 20 years after treatment. the surcharge embankment
heights varied from 3.0m to 9.0m. Based on the design criteria,
table 1 presents the pVd characteristics and treatment types
applied to each section. in the non-vacuum areas, both circular
and band shape drains were established in a square pattern at a
spacing in the range of 1.1-1.3m. the length of pVds changed
from 6m to 27.5m across the site as shown in the table 1.

Vertical displacement
(m)
Excess pore pressure (kPa)

drain type
Wd1
Wd2
Wd3
Wd4
Wd5a
Wd5B
Vc1
Vc2

circular
drains
circular
drains
Band drain
type -a
Band drains
type -a
Band drains
type -B
Band drains
type -B
circular
drains
circular
drains

fill height
(m)

treatment
scheme

1.1

5.2

surcharge

1.3

7-7.2

surcharge

1.1

4.3-4.6

surcharge

1.3

6.1

surcharge

1.2

3.3

surcharge

1.1

5.5

surcharge

1.2

3.2

1.2

2.8

4
2
0
100

table 1. pVd characteristics and improvement scheme


drain
spacing

(a)
6

200

300

400

100

200

300

4 00

100

200
300
Time (days)

400

0.5
1
1.5
2

200

(b)

WD3
WD5A
WD5B
VC1
VC2

(c)

100

-100

figure 2. embankment responses (a) staged construction, (b)


settlements and (c) excess pore pressures

the measured lateral displacement normalized to total


change in applied stress (vacuum plus surcharge load) for two
inclinometer locations is shown in fig. 3. for Wd3 area, the
total surcharge height was 90 kpa, whereas for Vc1 area the
reduced surcharge pressure of 40 kpa was complemented with a
vacuum pressure of 65 kpa. the lateral displacements clearly
lessen in the holocene sand due to its greater stiffness. fig. 3
indicates that the lateral movements are effectively controlled to
minimise the disturbance in the adjacent moreton Bay marine
park, due to the isotropic consolidation by vacuum pressure.

surcharge+
vacuum
surcharge+
vacuum

the inevitable variation in drain lengths was attributed to


the non-uniform clay thickness. Wick drains (Band drain typea and Band drain type-B) had dimensions of 100mm x 4mm,
and the circular drains had an internal diameter of 34mm. the
authors have deliberately omitted the commercial brand names
of all pVds used.
to monitor the ground behaviour,
comprehensive instruments were installed e.g. settlement plates,
vibrating wire piezometers, magnetic extensometers, and
inclinometers. in the vacuum area, only circular drains were
employed at a spacing of 1.2m in conjunction with a high
density polyethylene (hdp) membrane, horizontal perforated

2498

settlement and eXcess pore pressUre


predictions

in order to analyse the radial consolidation caused by vertical


drains, the unit cell theory has been employed to predict the
settlement and excess pore pressure. a unit cell theory was
introduced by Barron (1948) and richart (1957) for surcharge
preloading alone. lekha et al. (1998) further extended the radial
consolidation by including time-dependent surcharge loading.
indraratna et al. (2005) introduced the unit cell analysis for
vacuum preloading under instantaneous loading while Geng et
al. (2012) proposed analytical solutions under time-dependent

Technical Committee 211 / Comit technique 211

surcharge preloading. during embankment construction, the


surcharge fill is increased at a prescribed rate to reach the
desired height. therefore, the time-dependent loading due to the
filling would be more realistic than an instantaneous loading,
especially during the stages of embankment construction. in this
section, the embankment load is assumed to be a ramp loading:
i.e., the embankment load (t) increases linearly with time up to
a maximum value (within time t0 and is constant thereafter
(fig. 4a). the vacuum is applied at t=tvac. figure 4b shows the
unit cell adopted for analytical solutions with boundary
conditions (fig. 4c).
10

Platform

Platform

4
2

Dredged mud

Holocene sand

HS

-2

Upper Holocene Clay

UHC

-4

Depth (m)

Dredged mud

LHC

-10
-12
-14
-16

Lower Holocene Clay

-18
-20
-22
-24

Section/Plate No.
VC1/MS28
WD3/MS27

-26
-28
-30
0

1
2
Lateral displacement/Total change in applied stress
(mm/kPa)

figure 3. comparison of lateral displacements at the embankment toe in


vacuum and non-vacuum area after 400 days (indraratna et al. 2011)

(a)

(b)
(c)
figure 4. (a) time-dependent surcharge loading, (b) unit cell including
smear zone, and (c) boundary conditions with vacuum distribution (after
indraratna et al. 2011)

the excess pore pressure due to radial consolidation


considering smear effect under time-dependent surcharge can be
expressed by (indraratna et al. 2011):

uL

8ch t
1 exp
2

8ch t0
d e

d e2

for

8ch t t0
8ch t0
exp

uL
1 exp
2
d e2
8ch t0
d e

d e2

0 t t0

for t t0

(1)
(2)

recently, indraratna et al. (2005) proposed that the excess


pore pressure dissipation due to vacuum pressure alone could be
determined from:
(3)
u vac 0,
t t vac
8ch t tvac
u vac p0 exp
d e2

p0 ,

t t vac

(4)

2499

2k
kh
3
lns h l 2
4
3q w
ks

(5)

n de dw

(6)

s ds dw

(7)

where, de = the diameter of soil cylinder dewatered by a drain,


ds= the diameter of the smear zone, dw = the equivalent diameter
of the drain, ks= horizontal soil permeability in the smear zone
and qw = drain discharge capacity.
the excess pore pressure at a given time t can calculated
based on the equations (2) to (7). for normally consolidated
clay, the settlement () can now be determined by the following
equation:
(8)
'
HCc

-6
-8

n
s

ln

1 e0

log
'i

where, = settlement at a given time, Cc = compression index,


and H = compressible soil thickness.
in order to calculate excess pore pressures and associated
settlements, equations (1)-(8) are employed using parameters in
table 2. for the completely remoulded dredged mud that was
reclaimed from the seabed and the Upper holocene sand the
ratio kh/ks was assumed to be unity. for the upper and lower
holocene clay, the ratios of kh/ks and ds/dw were 2 and 3,
respectively, in accordance with the laboratory obsevrvation
decribed by indraratna and redana (1998).
the embankment load was applied according to a staged
construction (unit weight of 20 kn/m3). settlement and
associated excess pore pressure predictions were calculated at
the embankment centreline using eqs. 1-8. it is noted that, at the
beginning of each subsequent stage, the initial in-situ effective
stress was calculated based on the final degree of consolidation
of the previous stage. in vacuum areas, a suction pressure of 65
kpa was employed.
figures 5 and 6 present the predicted settlement and
associated excess pore pressure with the measured data in areas
Wd1 and Vc1, where the total applied load (vacuum and
surcharge =120-130kpa) and clay thickness (20-23m) are
comparable. overall, the comparisons between prediction and
field observation show that the settlement and associated pore
water pressure can be predicted very well. in vacuum areas, the
degree of consolidation was more than 90% after 13 months,
whereas that in the non-vacuum area was less than 85%. this
confirms that, at a given time, the vacuum combined preloading
would speed up consolidation compared to a surcharge
preloading alone. this is because in non-vacuum areas, a staged
embankment construction had to be adopted to avoid any
undrained failure in the remoulded dredged layer.
table 2. soil properties for each layer
Soil
Cc/(1+e0)
layer Soil type
1
dredged mud
0.235
Upper holocene
2
0.01
sand
Upper holocene
3
0.18
clay
lower
4
0.2
holocene clay

ch
(m2/yr)
1

kh/ks
1

s=ds/dw
1

1.9

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

analysed and discussed. the dredged materials from the seabed


were placed in the reclaimed area. a total of 8 areas were
selected to examine the performance of vacuum consolidation,
and the vertical drain spacing varied from 1-1.3m for 3 different
drain types. the vacuum application induces an inward lateral
movement, whereas the conventional surcharge fill creates
outward movement. When the vacuum pressure combined with
surcharge fill is employed, the overall lateral movement is
decreased due to the isotropic consolidation induced by vacuum
pressure. from a stability point of view, vacuum pressure
reduces the ratio of lateral displacement to surcharge fill height
at any given time.
the unit cell theory considering time-dependent surcharge
load and vacuum application was employed to predict the
settlement and associated excess pore pressure, which provided
a good agreement with the field measurements. after 1 year, the
degree of consolidation in the vacuum areas was much higher
than the non-vacuum areas for the same total stress.

Embankment Height (m)

5
(a)

4
3
2
1
00

100

Settlement (m)

0.4

200
300
Time (days)

400

500
(b)
Field
Prediction

0.8
1.2

Excess pore pressure (kPa)

1.6
0

100

80

200
300
Time (days)

400

500

(c)

60
40
20
0

Field
Prediction
0

100

200
300
Time (days)

400

500

figure 5. Wd1 area: (a) stages of loading, (b) surface settlements at the
embankment centreline and (c) excess pore pressures at 9.2m deep

Embankment Height (m)

(a)
3

2
1
Vacuum application of 70 kPa

Settlement (m)

00

Excess pore pressure (kPa)

100

200
300
Time (days)

400

500
(b)
Field
Prediction

0.4

0.8

1.2
80

100

200
300
Time (days)

400

500
(c)
Field
Prediction

40

-40

100

200
300
Time (days)

400

500

figure 6. Vc1 area: (a) stages of loading, (b) surface settlements at the
embankment centreline and (c) excess pore pressures at 14.1m deep
(indraratna et al. 2011)

Writers acknowledge the support of the port of Brisbane


corporation, coffey Geotechnics and austress menard. the
research funding from the australia research council is
acknowledged. the assistance of prof. a.s. Balasubramaniam
of Griffith University, daniel Berthier of austress menard, prof
harry poulos, cynthia de Bok, tine Birkemose and chamari
Bamunawita of coffey Geotechnics is appreciated. most of the
contents reported in this paper are also described in greater
detail in a number of and asce Journal of Geotechnical and
Geoenvironmental engineering.
7

conclUsions

a system of vertical drains with vacuum preloading is an


effective method for speeding up soil consolidation. the
performance of 2 treatment schemes at the port of Brisbane was

2500

acKnoWledGements

references

Barron, r. a. 1948. the influence of drain wells on the consolidation of


fine-grained soils. diss., providence, U s eng. office.
chai, J.c., carter, J.p., and hayashi, s. 2005. Ground deformation
induced by vacuum consolidation. Journal of Geotechnical and
Geoenvironmental Engineering, 131(12):1552-1561.
chu, J. Yan, s.W., and Yang, h. 2000. soil improvement by the
vacuum preloading method for an oil storage station. Geotechnique,
50(6): 625-632.
Geng, X. Y., indraratna, B. and rujikiatkamjorn, c. (2012). analytical
solutions for a single vertical drain with vacuum and timedependent surcharge preloading in membrane and membraneless
systems. International Journal of Geomechanics, asce, 12(1), 2742.
indraratna, B., and redana, i. W. 1998. laboratory determination of
smear zone due to vertical drain installation. J. Geotech. Eng.,
asce, 125(1): 96-99.
indraratna, B., sathananthan, i., rujikiatkamjorn c. and
Balasubramaniam, a. s. 2005. analytical and numerical modelling
of soft soil stabilized by pVd incorporating vacuum preloading.
International Journal of Geomechanics, 5(2). 114-124.
indraratna, B., rujikiatkamjorn, c., ameratunga, J., and Boyle, p. 2011
performance and prediction of Vacuum combined surcharge
consolidation at port of Brisbane. J. of Geotechnical &
Geoenvironmental Engineering, asce, 137 (11), 1009-1018.
richart, f.e. 1957. a review of the theories for sand drains. Journal of
the Soil Mechanics and Foundations Division, asce, 83(3): 1-38.
rujikiatkamjorn, c., indraratna, B. and chu, J. 2008. 2d and 3d
numerical modeling of combined surcharge and vacuum preloading
with vertical drains. International Journal of Geomechanics, 8(2):
144-156.
shang, J.Q., tang, m., and miao, Z. 1998. Vacuum preloading
consolidation of reclaimed land: a case study. Canadian
Geotechnical Journal, 35: 740-749.
Yan, s.W. and chu, J. 2003. soil improvement for a road using a
vacuum preloading method. Ground Improvement, 7(4): 165-172.

Improvement of a Clay Deposit using Prefabricated Vertical Drains and Pre-loading.


A Case Study
Amlioration d'unofmassif
d'argile
l'aide
de drains
verticaux prfabriqus
et de pr-chargement.
Improvement
a Clay
Deposit
using
Prefabricated
Vertical Drains
and Pre-loading
Une
tude
de
cas
- A Case Study

Amlioration
d'unS.J.M.
massif d'argile l'aide de drains verticaux prfabriqus et de pr-chargement
Islam M.S., Yasin
-Bangladesh
Une tude
de casof Engineering and Technology, Dhaka, Bangladesh
University
Islam M.S., Yasin S.J.M.

Bangladesh University of Engineering and Technology, Dhaka, Bangladesh


aBstract: construction of a container terminal covering an area of of 153,000 m2 is underway at chittagong sea port in
Bangladesh, situated on the right bank of the Karnaphuli river at its confluence with the Bay of Bengal. the sub-soil at the site
consisted of a 4 to 6 m thick clay layer with random zones of soft to stiff clay (cl). this was underlain by a 10 m thick loose to
medium dense silt/fine sand (sm) layer below which a loose clayey silt layer existed beyond 30 m depth. the unconfined
compressive strength, void ratio and compression index of the soft clay zones varied in the ranges of 12~16 kpa, 0.8~1.2 and 0.2~0.3,
respectively. the targeted use of the land required improvement of the sub-soil. this paper presents the design considerations,
comparison of required time and cost of alternative options, effectiveness of the adopted measure and the achieved improvement of
the engineering properties. actual consolidation settlements were up to 600 mm over a period of about 30 days with pVd and preload. the effectiveness of the available theories of consolidation settlement under vertical and radial drainage, in the design of the
ground improvement measures, has been demonstrated.
rsUm : . la construction d'un terminal conteneurs d'une superficie de 153.000 m2 est en cours en ville portuaire de chittagong
au Bangladesh, situe sur la rive droite de la rivire Karnaphuli sa confluence avec le golfe du Bengale. le sous-sol sur le site se
composait d'une couche de 4 6 m d'paisseur d'argile avec des zones alatoires d'argile molle raide (cl). cela a t superpose
une paisseur de 10 m limon / sable fin (sm) lche dense, une couche en dessous de laquelle une couche limon argileux lche
existait au-del de 30 m de profondeur. la rsistance la compression, lindice des vides et l'indice de compression des zones d'argile
molle varient dans les plages de 12 16 kpa, 0.8 ~ 1.2 ~ 0.3 et 0.2, respectivement. l'utilisation obligatoire de ces terres requiert
l'amlioration du sous-sol. cet article prsente la conception, la comparaison du temps ncessaire et le cot des options alternatives,
l'efficacit de la mesure adopte et de l'amlioration obtenue des proprits mcaniques. les tassements de consolidation rels ont t
de 600 mm sur une priode d'environ 30 jours avec pVd et pr-chargement. dans la conception de l'amlioration des sols, l'efficacit
des thories existantes de tassement de consolidation en vertu de drainage vertical et radial a t dmontr.
KeYWords: clay, Ground improvement, pre-loading, pVd
1

introdUction

chittagong sea port, the largest sea port of Bangladesh is


situated on the right bank of the Karnaphuli river at its
confluence with the Bay of Bengal. the port, that once handled
mostly bulk cargo is gradually shifting its operational mode to
handle increasing volume of container traffic. in this regard,
chittagong port authority (cpa) is implementing a project for
construction of backup facilities at new mooring behind berths
4 and 5. the site is locally known as nct (new-mooring
container terminal). the project area is about 153,000 m2
which is planned to accommodate stacking yard for containers,
passage for truck and trailer movement, tracks for Gantry
crane, electrical substation etc.
a comprehensive geotechnical investigation was carried out
at the site to assess the sub-soil condition, decide on the
necessity of improvement and determine relevant design
parameters for the envisaged improvement methodology. the
soil profile in the project area, consisted of a 4 to 6 m thick soft
to medium stiff clay layer, underlain by a 10 m thick loose to
medium dense silt/fine sand layer below which a loose clayey
silt layer existed to more than 30 m depth.
to keep conformity with the earlier constructed adjacent
yard, it is considered that the area will be paved with
interlocking block (ilB) except the rmG (rail mounted
Gantry) and rtG (rubber tyred Gantry) tracks which will be
pile founded. on the basis of design requirements and
geotechnical characteristics, improvement of the upper soft clay
layer was considered essential to eliminate the possibility of
differential settlement within the yard as well as between pile
founded structures (i.e. jetty and rmG, rtG tracks) and yard.

from an study of several alternatives, prefabricated Vertical


drain (pVd) with pre-loading was adopted as the ground
improvement measure for the site. improvement measures have
been completed on a part of the project area. the settlement
under preloading with pVd has been monitored using
settlement plates. field and laboratory tests have also been
conducted to evaluate the effectiveness of the adopted measures
in terms of change of soil properties. this paper presents the
geotechnical characteristics of the sub-soil in the area, the
design considerations and a comparison of cost of several
alternative improvement methods. from the limited data, that
has so far been available, comparison of some engineering
properties before and after preloading has also been made.
2

site location and topoGraphY

the site for the container yard is in a tidal plain at a narrow strip
between chittagong hilly uplands and the Bay of Bengal.
Geologically it is a recent alluvium formed by the material
carried by the river Karnaphuli and its tributaries from the upper
tertiary hills. figure 1 shows the site map with grid lines. about
half of the land, the eastern side (segments marked 1, 2, 3 and 4)
had been used as jetty yard for more than 50 years and housed
storage sheds for general cargo, road and railway tracks. the
other half (western part, segments marked 1a, 2a, 3a and 4a)
contained a city road, a residential area of cpa containing
one/two storey building, ponds, play ground, open land, village
dwellings etc. different parts of this western side had different
elevations with 1~3 m ditches. Because of earlier diverse use of
the land, there was little possibility of homogeneity of the upper
soil layer in the area.

2501

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

sUB-soil characterisctics

a total of 67 exploratory boreholes were drilled in the area to


gather information about sub-soil type and characteristics. the
borehole locations were carefully decided to make them
distributed over the entire area as well as to cover zones of
different land use (i.e. pond, road, houses etc.). the borehole
that are referred here are marked by grid points as shown in
figure 1. the boreholes, approximately 100 mm in diameter,
were drilled using water flush aided by chiselling. twelve
boreholes were of 30.5 m depth and the rest were of 15.0 m
depth below the existing ground level. standard penetration
tests (spt). were made at 1.5 m interval. Undisturbed samples
were retrieved from cohesive layers by pushing conventional 76
mm external diameter thin-walled shelby tubes. disturbed
samples were also collected from the spt spoon (conventional
split spoon) from cohesive and cohesionless soil layers at
different depths for visual-manual identification of the layers as
well as for laboratory testing.
50

SPT N-value (uncorrected)


16
24
32

-8
Fine
sand/silt
8~10 m

-12

-16
Claye silt
16m~

-20

Distance in meter

Grid points

-24

C-0 BL
D-6 BL
F-5 BL
H-7 BL
I-1 BL
L-8 BL
B-9 BL

3
4A

11
12
13

-32

Fine content, %
0

16

LL, PL, PI
24 0

15
16
17
18

20

40

60

LL/BL
PL/BL
PI/BL

-1

14

D-8 BL
E-4 BL
H-9 BL
I-5 BL
J-7 BL
L-6 BL

figure 2 General ground profile and variation of spt with depth.

Reduced Level, m

Jetty deck

2
3A

9
10

-28

KarnafUli riVer

2A

1
2
3
4

7
8

48

Silty clay
4~6 m

-4

100

5
6

40

N
M
1A
L
K
J
I
H
G
F
E
D
C

silt', the field spt-n values ranged between 5 and 27 except in


one borehole where a 'sandy silt' layer existed.

Reduced Level, m

Void ratio, e 0

80 0.4

LL/AL
PL/AL
PI/AL

0.8

1.2
BL

1.6

AL

-2
-3
-4
-5

19

-6

figure 1 site map showing grids and loading blocks.

10

20

MC, %
30

40

0
BL

AL

50 0

qu , kPa

100

BL

200
AL

300 0

Com p. Index, C c
0.1 0.2 0.3 0.4 0.5

0
BL

AL

-1
Reduced Level, m

in general, the sub-soil at the site is found to consist of a


layer of soft to medium stiff silty clay extending from the
ground surface to about 4 to 6 m depth. this layer is underlain
by a 10 m thick loose to medium dense fine sand/silt layer. A
clayey silt layer is encountered below the fine sand/silt layer
which extend beyond the maximum depth of investigation (i.e.,
about 30 m from surface). thus, up to a depth of 30 m the subsoil at the site is idealized to have three distinct layers (top silty
clay layer, intermediate fine sand/silt layer and bottom clayey
silt layer). in a small number of boreholes medium dense sand
was encountered near the ground surface instead of the clay
layer, which was probably a fill during past use of the land.
figure 2 presents the field spt-n values at different depths for
the explored borehole locations and the stratigraphy. in the top
silty clay layer, field spt-n values (not corrected for
overburden) ranged between 2 to 10. spt-n values greater than
10 in the top layer are for locations where pockets of sand/silt
exist. in the second layer (fine sand/silt), field spt-n values
ranged between 5 and 44. in the bottom layer, which is 'clayey

-2
-3
-4
-5
-6

figure 3 Variation of index, strength and deformation properties of the


upper clay deposit with rl (Bl=Before loading, al=after loading).

extensive laboratory tests have been conducted on samples


of top silty clay layer (Brtc, BUet report, 2009) and some of
the results are presented in figure 3. the layer may be

2502

Technical Committee 211 / Comit technique 211

characterized as having ll=50~60, pl=20~30, pi =20~30,


nmc=20~35%. according to Unified soil classification
system (Uscs), the soil in this layer is mostly plastic-silty clay
of low plasticity (cl), though a few samples were found to be
clay of high plasticity (ch). on the Uscs chart, the data points
lie just above a-line. the dry unit weight varied in the range of
13~15 kn/m3 and the range of void ratio was 0.80~1.20. the
variation of these properties can be seen from figure 3. the
layer also has some organic content is about 1.4~4.0%. the
values of the coefficient of consolidation in the vertical
direction, cv were mostly within 3.1 to 25.2 m2/year and at
some location as low as 0.79 m2/year.
4

desiGn of GroUnd improVement method

it was decided by cpa, that the project will be carried out by


local contractors. therefore, capacity, experience, equipments
etc. of local contractors were to be considered in the design of
the yard. furthermore, ground improvement was to be
completed for the entire project site within one year. hence, the
area was divided into four blocks and time for improvement for
each block was 3 months. an area adjacent to the north
boundary of the site was earlier developed for similar purpose,
by a foreign contractor, where dynamic temping was used for
ground improvement and interlocking block pavement was
made. to keep similarity with the earlier part, interlocking
block pavement was decided for this yard too.
the presence of very soft to medium stiff silty clay at
various locations within the site indicated strong possibility of
substantial total and differential settlement unless effective
measures for improvement of sub-soil are undertaken before the
construction of pavement for the container Yard. therefore,
effective measures for improvement of sub-soil before the
construction of pavement were considered essential in order to
avoid/minimize future problems.
the necessity and extent of the ground improvement
measures are judged with an objective to reduce the differential
settlement and maintenance operations considering the
maximum load from stacking of containers on the entire area
(i.e. p= 52 kn/m2). it should be understood that a solution, for
which there will be no future settlement, will lead to high cost
and time for completion and thus may not be practical. the load
on the rtG tracks from the gantry is estimated to be 77.5
kn/m2. the extent of improvement and design of pavement
system at the site is targeted to keep maintenance option with
minimal disruption. for rtG and rmG tracks and other
facilities, suitable deep/shallow foundations will be considered
so that they do not undergo relative settlement with respect to
jetty top.
five alternatives, that appeared to be feasible for local
contractors, were assessed. these are- (i) preloading (ii) sand
drain with surcharge (iii) pVd with surcharge and (iv) dynamic
Working surface for pVd installation

compacted backfilling. table 1 presents the comparison of cost


and completion time for different methods. Both time and cost
depends to some extent on the number of equipments mobilized
and source of material, particularly the surcharge (max. 5 m of
soil considered). considering the capacity of local contractors
minimal engagement of equipments and dredge sand from the
Karnaphuli river were considered. though dynamic
temping/compaction appeared to be very prospective in terms of
time and cost, it posed the risk of damaging the adjacent yard
and structures. finally, pVd with surcharge was adopted as the
ground improvement measures, mainly because of reduced time
in pVd driving compared to sand drain installation, though
pVd is an imported material. also this method was considered
advantageous over other methods in bringing the clay layer to a
state where differential settlement potential will be reduced as it
will automatically take care of soft zones and bring the soft and
stiff zones to closer soil properties in terms of deformation and
strength.
since, from e~log(p) curves, most of the samples of the
upper clay layer was found to be normally consolidated, the
total consolidation settlement under the working loads (52 kpa)
without improvement was calculated using sc=e.h,
e=cc/(1+e0)log(p+p'0)/p'0 and p'0 ='h where, e0= initial
void ratio, p'0 = effective past maximum overburden pressure,
'=effective unit weight of soil, h=thickness of the compressible
layer. the estimated settlement for different borehole locations
varied from about 140 mm to 570 mm. this variation is due to
difference in e0, cc and layer thickness. in these estimations, p
is calculated as p = 0[1-{1+(r/z)2}-1.5] where 0=intensity of
stress applied on the surface, r = radius of the loaded area,
p=increase in stress at depth z from the centre of the loaded
area. this expression for p is obtained by integration of
Boussinesque's equation that gives the stress at a point within a
semi-infinite, homogeneous, isotropic, weightless, elastic halfspace for a point load on the surface (Bowels,1988). estimated
time to achieve this consolidation (Uav 99%) varied from
about 50 days to more than 700 days for different borehole
locations. the time was determined using terzaghi's one
dimensional consolidation theory with double drainage and
constant initial pore pressure distribution using the equations
(das, 1983):
m 2

U av 1

m 0 m

Ur 1 e

h2

8T r
)
m

where Tr

C vr t

d e2
k
n
n 3 S
m 2
ln 2 h
2
S
4
ks
n S
4n

de
ds
and S
n
dw
dw

200 mm
300 mm
150 mm

n2 S 2

n2

ln S

the equivalent diameter of pVd was calculated following


hansbo (1979) as

5m

figure 4 details of the ground improvement work.

cvt
2

and m 2m 1
e m t v , tv

it was intended to apply a surcharge with pVd such that a


maximum of 25 mm of total settlement remains to occur in
future under the working loads expecting a differential
settlement of not more than 12 mm. estimation of required time
to achieve this level of consolidation was made considering
both vertical and radial drainage (carillo,1942) as U=1-(1Uv)(1-Ur) where Uv and Ur are the average degree of
consolidation respectively for vertical and radial drainage. the
average degree of consolidation for radial drainage was
calculated using the following as

surcharge, 5m of soil

pVd

dw

temping and (v) soft pocket identification, removal and

2(b t )

where, b is the width and t is the thickness of pVd. considering


smear effect s was chosen between 1.0 and 1.2. the effective
diameter of soil column around the pVd was taken as de=1.06s,
s=pVd spacing in triangular pattern.
for all the calculations horizontal permeability is taken as

2503

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

the pVd used were of 100 mm width, 3 mm thick placed 1m


c/c in triangular pattern. other properties - drain: Water
discharge capacity 9010-6 m3/s, and 6010-6 m3/s respectively
at 10 and 350 kpa (i=0.5); core: tensile strength 700 n; filter
jacket: apparent opening size (aos) 90 m, Grab tensile
strength 400 n, elongation at break 50%, puncture resistance
130 n, Burst strength 800 kn/m2, permeability 210-4 m/s.
details of the ground improvement work is shown in figure 4.
5

assessment of GroUnd improVement

monitoring of settlement has been made using settlement


plates placed at 25 m grid as shown in figure 1. after
preloading exploratory boreholes were made at selected
locations with field spt and laboratory tests were conducted on
collected undisturbed samples.
SPT N-value (uncorrected)

16

24

32

40

48

0
E-4 BL
D-8 BL
F-5 BL
L-6 BL
I-8 BL
D-8 AL
F-5 AL
L-6 AL
I-8 AL

-2

-4

Reduced level, m

-6

J-2 BL
D-6 BL
J-7 BL
K-8 BL
I-8 BL
D-6 AL
J-7 AL
K-8 AL
Afte r loading
Data points

600

350

500

300
250
200
150
100

400
300
200
100

(a)

50

(b)

0
0

50

100 150 200 250 300


qu, kPa (Before loading)

350

400

100

200
300
400
500
Estimated settlement, mm

600

figure 6 comparison of (a) unconfined compressive strength and


(b) observed and estimated settlement .

the unconfined compressive strength of the upper clay layer


before and after preloading can be seen on figure 6(a) for
different locations and depths. in general the unconfined
compression strength has increased at most of the spots.
however, the magnitude of increase is not same. a few data
points lie below the 45 degree line, that apparently shows to
have reduction in strength but quite unlikely. the reason for
these discrepancies may be the variation of non-plastic silt
content in the layer.
in figure 6(b) the recorded settlements are plotted against
the estimated settlement for some of the grid points. out of the
six locations (for which estimates were made) four appear to
match reasonably well. for one location the observed settlement
is about five times the estimated value, which may be due to
presence of localized sand lenses. on the other hand for another
location the observed settlement is one-fifth of the estimated
value, which may be due to clogging or disturbance of the clay
during drain installation. conclusive comments regarding the
variation may be made when data from the remaining project
work become available.
6

conclUsions

the following conclusions can be made based on the design and


field monitoring of the ground improvement work:
1) due to the application of the surcharge with pVd the
consolidation settlement could be achieved within the stipulated
time.
2) Both the spt-n value and unconfined compressive strength
were found to increase satisfactorily due to application of
preload with pVd.
3) the available theories of 1-d consolidation and combined
vertical and radial consolidation used in the design of ground
improvement for the project site using pVd and preload
appeared to have been fairly applicable. predicted and observed
settlement matched reasonably.
7

-8

acKnoWledGements

this research work was carried out in connection with the consultancy
services provided to cpa through Brtc, BUet. the authors would
like to express their sincere thanks to cpa and contractor's personnel
involved in the project.

-10

-12

-14

400

Observed settlement, mm

table 1 comparison of estimated cost and completion time for different


ground improvement methods.
method
time
cost comments
(month) (million
Usd)
reliable
Better assessment of improvement
preloading
36
1.96
long time required for
improvement
relatively less reliable
sand drain
installation of drains takes long
20
3.29
with surcharge
time
pVd with
reliable
14
3.24
surcharge
Good control of field operation
Vibration may damage adjacent
facilities
dynamic
high noise pollution
11
2.04
compaction
assessment of improvement
needs lot of field tests
relatively less reliable
soft pocket
assessment of improvement is
identification,
difficult
removal and
12
7.25
highly dependent on field
improvement,
monitoring and control
compacted
backfilling

changed. the field spt-n values are found to increase


significantly in the 'fine sand/silt' layer up to about 12 m.

qu, kPa (After loading)

the same as the vertical permeability. from the calculations it


appeared that for 5 m surcharge and pVd the target settlement
would occur within 10 to 50 days in different locations.

8
Be fore loading
Data points

-16

figure 5. Variation of spt-n value before and after loading.

figure 5. compares the field spt-n values at several spots


before and after preloading. it can be observed that in the upper
silty clay layer the spt-n values has become twice or more up
to about 3m depth. at about 4~5m, which is the boundary
between the clay and sand/silt layer the spt-n values have not

2504

references

Bowels, J.e. (1988) foundation analysis and design, 4th edn.,


mcGraw-hill Book company.
Brtc, BUet report (2009) sub-soil investigation and ground
improvement measures for construction of backup facilities behind
berth 4 & 5 at new mooring container terminal (nct), cpa,
chittagong.
carillo, n.J. (1942) simple two- and three- dimensional cases in the
theory of consolidation of soils.Journ. math. and phy.(21),pp.1-5.
das, B.m. (1983) advanced soil mechanics, international edn.,
mcGraw hill
hansbo, s.(1979) consolidation of clay by band-shaped prefabricated
drains, Ground engineering,(12),5.pp.16-25.

Importance et applications des inclusions de grande inertie


Importance and practical examples of inertial soil improvement
Jeanty J.M., Mathieu F., Benhamou L.

Soletanche-Bachy, Rueil-Malmaison, France.

Berthelot P.

Bureau Veritas, Paris, France.

rsUm : les techniques traditionnelles damlioration des sols par inclusions visent le plus souvent renforcer laptitude du massif
reprendre les charges verticales auxquelles il est soumis, moyennant des dformations acceptables par les ouvrages. elles peuvent
galement avoir pour objectif damliorer la rsistance aux efforts horizontaux, et plus gnralement aux sollicitations dominante
dviatorique, dans le cadre de problmatiques lies aux sismes (traitements anti-liqufaction), la stabilit gnrale (remblais sur
sols compressibles), la rduction des pousses sur des ouvrages de soutnement, au poinonnement du sol sous de fortes surcharges
(effets de bord), la ralisation dinclusions forte inertie, sous forme de tranches parallles ou de rseaux de tranches, est une
rponse particulirement bien adapte ces problmatiques damlioration des sols. Une manire lgante de construire ces inclusions
sans perturber les structures existantes consiste traiter le sol en place en y incorporant un liant hydraulique, par application des
techniques de soil mixing les plus rcentes. aprs une prsentation de ces mthodes, la communication propose met en vidence le
rle fondamental jou par linertie des inclusions, par des considrations fonctionnelles et par divers exemples dapplication.
aBstract: conventional soil reinforcement techniques as rigid inclusions mainly report vertical loadings to the substratum layer
with an induced settlement. they can also improve soil resistance regarding lateral forces as those related to earthquakes (liquefaction
hazard), or slope stability for embankments on soft soil foundation. they can reduce active pressure on retaining walls.
the appropriate design answer to those issues is to create strong inertia inclusions based on a trench geometry with either a parallel or
a crossed frame arrangement. last but not least, an even better technique to build those inclusions with a reduced environment impact
consists in treating in situ soil by adding cement. this article presents different soil mixing projects and explains how strong inertia
trenches are relevant.
mots-cls: amlioration de sols, sol-mixing, sol-ciment, inclusions rigides, inertie, liqufaction, soutnement, pousse des terres,
tassement, tranches de sol-mixing.
KeYWords : soil reinforcement , soil mixing, rigid inclusions, slope stability, retaining wall, liquefaction, seismic, active pressure,
Bouassida approach, settlement, soil mixing caissons, soil mixing trench, crosswalls.
1

introdUction

le soil mixing profond est une technique dveloppe dans les


annes 1970 en europe du nord et au Japon, initialement pour
rsoudre des problmatiques lies au comportement des sols
compressibles. cette technique consiste amliorer les
caractristiques dun sol en le mlangeant en place avec un liant
hydraulique.
la dstructuration des terrains et lincorporation du liant
seffectue par des moyens mcaniques, en utilisant un outil dont
la gomtrie et le mouvement dans le terrain dfinissent les
dimensions des lments de sol trait.
la mise en uvre de cette technique a longtemps fait appel
lutilisation doutillages simples ou multiples, rotatifs axe
vertical, munis de pales latrales de gomtries trs varies. des
outils de type tarire simple ou tarires multiples avec inversion
des sens de rotation entre forage descendant et malaxage /
compactage en remontant sont aussi utiliss. des colonnes
ralises par jet-grouting peuvent aussi tre apparentes au solmixing.
plus rcemment, durant la dernire dcennie, de nouveaux
procds de soil mixing avec des outils axe horizontal ont fait
leur apparition : haveuses et trancheuses. ces procds ont
permis de repousser les limites du soil mixing en largissant la
mthode au traitement dune plus large gamme de sols et
prsentant des atouts en termes de caractristiques et
dhomognit.

le soil mixing se distingue depuis l'origine des autres procds


d'amliorations des sols en ce sens quil permet de raliser,
l'extrme, des rseaux d'inclusions isoles (pour l'excution
desquelles des outillages de type tarire peuvent tre suffisants)
ou des traitements dans la masse (utilisation d'outillages de type
multitarire ou haveuse pour raliser des "pavs" de sol trait),
mais c'est dans la ralisation d'lments linaires de grande
inertie, de type refend ou cran continu (susceptible la fois de
jouer un rle porteur et d'assurer une fonction soutnement
moyennant l'incorporation d'armatures mtalliques), voire de
rseaux orthogonaux, permettant de confiner le sol en place, que
le soil mixing trouve sa vritable originalit et ses
dveloppements les plus prometteurs.
Gnralement apparent aux rseaux d'inclusions plutt qu'aux
traitements dans la masse, il s'en distingue pourtant de faon
fondamentale, non seulement par les aspects gomtriques
prcdemment voqus, mais encore par ses proprits
mcaniques. cela ne permet de le classer ni dans la catgorie
des inclusions souples (colonnes ballastes, dont la rsistance en
compression est nulle en l'absence d'treinte latrale de la part
du sol avoisinant), ni dans la catgorie des inclusions rigides,
dont la rsistance en compression, qu'il s'agisse de mtal, de
bton ou de mortier, est au contraire indpendante de l'treinte
exerce par le sol et pour lesquelles la rsistance en
compression simple est par consquent le paramtre
dterminant pour le dimensionnement.
au contraire, les proprits du soil mixing sont bel et bien celles
d'un sol amlior, mme s'il ne s'agit gnralement pas d'un

2505

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

traitement uniforme de l'ensemble du massif de fondation, dont


le critre de rupture n'est autre que le critre de coulomb, qui
lui permet de bnficier la fois :
.d'une rsistance intrinsque indpendante de l'treinte (terme de
cohsion) ;
.d'une augmentation de rsistance avec la profondeur en
fonction de l'treinte latrale du sol (terme de frottement).
en termes de dformations, ses proprits sont galement
intermdiaires entre celles gnralement attribues aux
inclusions rigides et souples ce qui conduit naturellement le
ranger dans la catgorie indite des "inclusions semi-rigides".
le tableau qui suit rsume les ordres de grandeurs usuels de la
rsistance la compression fc et du module de dformation ey.
renforcement

inclusion
rigide

inclusion
semi-rigide

inclusion
souple

fc (mpa)

5 500

0.5 5

ey (Gpa)

3 200

0.2 3

0.04 0.08

1.1

Mlange par haveuse

le procd met en uvre un outillage appel cutter soil


mixing (csm), driv de la technologie utilises sur des
machines de type haveuse ou hydrofraise. deux tambours
spcialement conus pour cette application et entrains en
rotation par des moteurs hydrauliques de fort couple, sont
utiliss pour dstructurer et mlanger le sol en place. Une buse
dinjection situe entre ces moteurs permet lincorporation dun
fluide. le mlange ainsi ralis est ensuite dplac soit au
dessus des tambours dans le sens descente ; soit au dessous dans
le sens remonte.
les inclusions construites avec cet outillage jusqu des
profondeurs de quelques dizaines de mtres, sont de section
rectangulaire (longueur courante 2,80 m ; paisseurs 500 1000
mm). elles peuvent tre utilises unitairement (on parle alors de
barrettes) ou disposes de manire contiges afin de constituer
des parois continues (figure 1).
certains outillages plus perfectionns sont munis dun dispositif
de mesure inclinomtrique, permettant de mesurer et de corriger
en temps rel la position de loutil lors de la construction dune
inclusion.
le fluide inject peut tre de diffrents types : il peut sagir dun
fluide facilitant le forage, remplac en phase remonte par un
mlange eau-ciment (appel aussi coulis de ciment). il est
galement possible dinjecter directement un coulis pendant le
forage, la phase de remonte tant alors utilise pour parfaire le
mlange.
le type et la quantit de liant utilise permettent datteindre une
large gamme de caractristiques (rsistance, permabilit,
cohsion, module de dformation), en fonction de la nature des
sols en place.
pour les outillages les plus perfectionns, les volumes injects
ainsi que les nergies de malaxage sont contrls et ajusts en
temps rel grce un systme informatique embarqu dans la
cabine de la machine. tous les paramtres opratoires sont
enregistrs afin dtre restitus sous forme de rapports.

figure 1 : principe de construction de parois avec le procd csm.

enfin, divers types darmatures peuvent tre mis en place dans


le matriau encore frais, permettant ainsi de raliser des
ouvrages de soutnement provisoire ou caractre permanent.
1.2

Mlange par trancheuse

les machines de type trancheuse mettent en uvre une chaine


avec outils de coupe et de malaxage. la chaine est guide par
une lame travaillant dans un plan vertical dans le sol.
lensemble de loutillage prsente une certaine similitude
visuelle avec une trononneuse. laction de cette lame,
combine lincorporation dun fluide, permet de construire des
tranches de sol trait en place (figure 2).
suivant les cas, le liant hydraulique peut tre inject sous la
forme dun coulis ou incorpor au mlange sous forme
pulvrulente, auquel cas une adjonction deau est gnralement
ralise afin de faciliter laction des outils.
les tranches construites de cette manire sont continues, et
toutes les couches de sol sont uniformment mlanges.
lpaisseur de linclusion est de lordre de 0.3 0.5 m pour une
profondeur maximale denviron 10 mtres.
de manire analogue au procd par havage, le procd par
trancheuse saccompagne dun dispositif de contrle-qualit
embarqu, permettant le pilotage de la machine, le respect des
paramtres de traitement ainsi que les enregistrements
ncessaires pour lmission de rapports.

figure 2. principe de construction dune tranche.

2506

Technical Committee 211 / Comit technique 211

domaines dapplication

les tranches de sol trait en place peuvent tre parallles une


seule direction ou constituer une trame selon deux directions en
gnral perpendiculaires. le premier cas correspond une
problmatique avec une direction privilgie de sollicitation, le
second avec deux directions de sollicitation dgales
probabilits.
diffrents domaines dapplication selon le type de renforcement
recherch mritent dtre distingus.
.domaine 1 : reprise de charges surfaciques, permettant un
traitement plus rparti qu'avec des inclusions rigides (ce qui
permet de limiter fortement l'effet de "point dur" qui
conditionne sinon le ferraillage du dallage ou du radier susjacent et l'paisseur du matelas de rpartition).
a noter que le caractre "2d" du procd permet en outre de
procder des calculs plus rigoureux que les approches
traditionnelles, utilisant par exemple les coefficients de capacit
portante de Bouassida (Bouassida, 2002) (qui permettent de
traiter le cas d'une semelle de fondation sur sol renforc par une
tranche), ainsi que des modles de calcul numrique en
dformation plane pour le calcul du tassement (ce qui ne serait
pas acceptable pour un rseau d'inclusions isoles).
Un exemple d'application rcent est donn par le chantier de
saint-roch (06), consistant renforcer le sol sous un remblai
sncf par un rseau de tranches longitudinales.

figure 3. coupe type sncf st roch (06).

.domaine 2 : renforcement du sol dans les zones o le champ


de contraintes est dominante dviatorique.
la justification des rseaux d'inclusions repose en gnral sur
des justifications de portance effectues en partie courante de la
surface charge, l o le risque de rupture du massif de sol est
insignifiant dans la mesure o le champ de contraintes est
gnralement de type plutt oedomtrique. le plus souvent,
aucune justification particulire n'est demande dans les zones
soumises de fortes contraintes dviatoires mobilisant de faon
significative la rsistance au cisaillement du sol (priphrie des
zones de stockage par exemple), et corrlativement susceptibles
de mobiliser les inclusions en flexion.
Un exemple de ce type est donn par les zones latrales des
remblais de forte hauteur reposant sur des sols compressibles.
alors, une combinaison judicieuse d'inclusions isoles en partie
centrale et de refends sous la partie latrale a pu tre prconise
et mise en uvre par diffrents auteurs (rf. filz G. & al,
Kitazume m.).

.domaine 3 : rduction des pousses exerces sur les crans de


soutnement.
les inclusions isoles agissent par limitation du tassement
derrire le soutnement et allgement des contraintes verticales
dans le massif de sol par transfert partiel sur les inclusions,
tandis que les inclusions de forte inertie ajoutent cet effet celui
d'une diminution "directe" des contraintes de pousse par
mobilisation du frottement sur les refends autostables.
Un exemple d'application significatif est celui du rempitement
du quai poste 7 du transmanche (calais), pour lequel le premier
mode de comportement s'avrait inefficace en raison d'un
contraste de rigidits verticales insuffisant entre le sol en place
et les inclusions, et o seul l'effet inertiel permettait donc de
rduire la pousse de faon significative (cf chapitre 3.1).
.domaine 4 : traitement anti-liqufaction des sols sous
sollicitation sismiques.
il a t montr (rf. seed) que certaines mthodes de
justification des rseaux d'inclusions disjointes reposaient tort
sur un effet de transfert sur ces dernires des contraintes de
cisaillement induites par le sisme.
ces mthodes reliaient en effet l'efficacit des inclusions leur
module de cisaillement, en ignorant le fait que leur lancement
induit en gnral un mode de dformation en flexion largement
prpondrant par rapport au mode de dformation par
cisaillement, l'instar des poutres sur appuis lastiques de la
rsistance des matriaux.
c'est donc bien l'inertie et non la rigidit qui constitue le facteur
cl dans l'efficacit de ce type de traitement, ce qui conduit l
encore privilgier les rseaux d'inclusions de forte inertie. le
traitement de type "quadrillage" est souvent qualifi tort de
"confinement", alors que c'est bien l'effet "inertiel" qui est
recherch, le double rseau d'crans orthogonaux permettant
avant tout de raliser un traitement isotrope dont l'efficacit est
ainsi rendue indpendante de la direction des ondes sismiques.
Un exemple particulirement reprsentatif est donn par le
chantier de l'extension de la prfecture de fort-de-france, pour
lequel s'ajoutait, la problmatique du traitement de terrains
fortement liqufiables sur une grande hauteur, celle de
l'coulement post-liqufaction induit par le pendage significatif
du substratum, ce qui rendait ncessaire la mise en uvre d'un
rseau autostable (cf chapitre 3.2).

Ntan

figure 5. stabilit. principaux efforts dans le plan dun refend.

3
3.1

figure 4. diagramme de grand glissement.

pp

eXemples dapplication
Renforcement de sol derrire un soutnement

la mise en place de tranches de soil-mixing de grande inertie


larrire dun cran de soutnement permet la rduction des
pousses du terrain sur lcran lui-mme.
Une optimisation de la reprise des efforts de pousse conduit
installer des lments dinertie maximale (concentration des
efforts pour des dplacements limits). des lments isols sont
bien moins efficaces en terme dinertie globale.

2507

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

cest ce principe qui a t propos et retenu pour les travaux de


reconstruction du poste transmanche n7 calais.
la mise en place de refends de Geomix (mthode par havage
spcifique lentreprise) perpendiculairement laxe du quai
(figure 6) a permis de rduire la section des palplanches
mettre en uvre dans le cadre des travaux dapprofondissement
du quai.

longueur de seulement 40 m). les inclusions isoles de faible


inertie ne peuvent rsister ce phnomne.
cette premire technique, mise en uvre entre octobre 2010 et
janvier 2011 reprsente un vrai pas en avant dans lapproche des
fondations pour les dpartements doutre mer.
plus largement, il s'agit d'une solution technique innovante,
reposant sur un procd propre soltanche Bachy et qui
rpond efficacement aux problmatiques des clients en sites
sismiques.

figure 6. Vue 3d du renforcement derrire le rideau du quai existant.

le massif de sol trait reprend la pousse des terres et les


palplanches lavant ne reprennent plus que la pousse deau.
la stabilit du massif renforc par les perons de soil mixing est
vrifie vis--vis du non-basculement, du non-glissement, et de
la non extrusion entre tranches.
lautre partie des vrifications a consist tablir la cohsion
homognise du massif de sol, variable de 45 85 kpa pour ce
projet.
la solution de base en barrettes isoles apportait une cohsion
homognise moyenne de lordre de 24 kpa, attestant la bien
moindre performance des renforcements disjoints par rapport
la disposition en refends de grande inertie.
3.2

Traitement anti-liqufaction

les dgts engendrs par le dernier sisme significatif de


novembre 2007 en martinique ont ncessit la reconstruction de
la prfecture de fort de france (btiments type r+4).
le contexte gotechnique du site montre un fort potentiel de
liqufaction des alluvions dominante sablo-limoneuse de
faible caractristique (pl*~0,3 mpa, em~2,2 mpa), sur une
paisseur variable de 9 18 m correspondant la pente du
substratum.
en rponse lappel doffres, lentreprise a propos une
solution variante pour rpondre la fois aux problmatiques de
liqufaction et d'coulement post-liqufaction des sols sur la
pente du substratum.
Un nouveau type de fondations bas sur un quadrillage en sol
mixing sous lemprise totale des btiments (environ 36 m x 40
m) a t conu (figure 7).
les tranches Geomix, dpaisseur 0,50 m, sont espaces
denviron 4.5 m entre axes. par leur forte inertie (par
comparaison aux inclusions rigides) et leur gomtrie, les
dformations des panneaux sont limites pendant lpisode
sismique. le cisaillement additionnel du sol et les efforts
horizontaux provenant de la structure se concentrent sur les
bandes Geomix. le traitement de confinement permet ainsi de
limiter le cisaillement et le dveloppement de pressions
interstitielles dans le sol confin non trait. le risque de
liqufaction est vit.
les caissons anti-liqufaction servent galement en phase
service de fondation aux btiments par lintermdiaire dune
dalle de transfert. ils ont donc aussi le rle de rduction des
tassements sous la structure en situation statique.
le phnomne d'coulement post-liqufaction a provoqu de
nombreux dgts lors du sisme de Kob au Japon en 1995. sur
le site de la prfecture de fort-de-france, ce risque est accru par
une pente importante du substratum (dnivel de 9 m sur une

figure 7. Vue en 3d des fondations en caisson.

conclUsion

les quelques exemples prsents dans l'article, reprsentatifs de


diffrents domaines d'application, ont permis de mettre en
vidence la ncessit de privilgier non pas tant la rigidit que
l'inertie des rseaux d'inclusions : la mise en uvre de refends,
de parois continues ou de caissons "semi-rigides", raliss au
moyen de techniques rcemment dveloppes pour tendre le
domaine d'application du procd "soil mixing", permet ainsi
dans bien des cas d'apporter des problmes complexes une
rponse particulirement pertinente.
5

references

the deep mixing method. Coastal Development Institute Tokyo, June


2002.
lebon s.p.. new methods in european deep mixing a contractors
perspective on the developing challenges of execution. Deep
Mixing 2005, Stockholm.
Benhamou l. and mathieu f.. Geomix caissons against liquefaction.
ISSMGE - TC 211 International Symposium on Ground
Improvement IS-GI Brussels, 2012.
Bouassida mounir et Belgacem Jellali. capacit portante ultime dun sol
renforc par une tranche. revue franaise de gnie civil volume 6.
no 7 8/2002.
corneille s. and r a.. trenchmix : une technique damlioration de
sols qui contribut au dveloppement durable. Revue Travaux n854,
Juillet 2008.
filz G and al.. design of deep mixing for support of levees and
floodwalls. 4th International Conference of Grouting and Deep
Mixing, New Orleans, 2012.
Kitazume m. application of physical modelling for investigating
ground failure pattern. physical modelling in Geotechnics 6th
icpmG 2006. london isBn 0-415-41586-1.
Gueguin m and al. a homogenization approach for evaluating the
longitudinal shear stiffness of reinforced soils: column vs. cross
trench configuration. International Journal of Solids and
Structures, November 8th 2011.
r B seed & al. recent advances in soil liquefaction engineeering : a
unified and consistent framework. 26th annual asce los angeles
Geotechnical spring seminar. Queen mary presentation 2003.
shinkawa n. and Bessho n. application examples of deep mixing
method as aseismic measures. International Symposium on Deep
Mixing & Admixture Stabilization, Okinawa 2009..

2508

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Assessment of Carillos Theory for Improved Tunis Soft Soil by Geodrains

Assessment of Carillos theory for improved tunis soft soil by geodrains

valuation de la thorie de Carillo pour les sols mous de Tunis amliors par godrains
halima Jebali1, Wissem frikha2 & mounir Bouassida3
Jebali
H., Prikha
Bouassida
M.
Universit
TunisW.,
El Manar;
Geotechnical
Engineering Research Team, Ecole Nationale dIngnieurs de

Universit Tunis El Manar, Geotechnical Engineering Research Team, cole nationale dingnieurs de Tunis (ENIT)

tunis, enit, Bp 37 le Belvdre 1002, tunis. tunisia.

email: (1) hlm.jebali@gmail.com; (2) frikha_wissem@yahoo.fr; (3) mounir.bouassida@fulbrightmail.org

aBstract: this paper presents an experimental study carried out on undisturbed cored samples of tunis soft soil extracted
at 17.25 m depth at the lagoon of sejoumi. three types of oedometer tests had been performed: first type was a standard test
on tunis soft soil, the second was an oedometer test on the same soil improved by a prefabricated vertical drain mebradrain
88 (mb88) type and the third test is similar to the second test in which vertical drainage was prevented. then, the assessment
of Carrillos theory is studied by quantifying the effect of radial and vertical consolidation from the observed global degree of
consolidation of improved tunis soft soil specimens by geodrains.

rsum: ce papier prsente une tude exprimentale ralise sur des chantillons intacts du sol mou de tunis prlevs
17,25 m de profondeur de la marcage de Sjoumi. Trois types dessais oedomtriques ont t effectus : le premier est un
test standard sur le sol mou de tunis, le second tait un essai oedomtrique sur le mme sol amlior par un drain vertical de
type mebradrain 88 (mB88); le troisime test est similaire au deuxime test dans lequel seulement le drainage radial a t
favoris . ensuite, l'valuation de la thorie de carrillo est tudie en quantifiant l'effet de la consolidation radiale et verticale
sur le degr de consolidation global.

1.

introdUction

does not occur in 1D condition. Carrillos theoretical


solution (1942) is used to combine the vertical and radial
drainage effects to predict the global degree of
consolidation U:

considerable attention has been recently devoted


worldwide to the problem of building structures on highly
compressible saturated soils and to the development of soil
improvement techniques for increasing stability, reducing
settlements, and accelerating consolidation of soft soils.
prefabricated verticals drains (pVd) with preloading
method was considered the most used improvement
technique to accelerate the consolidation of soft soils and,
consequently, to increase their bearing capacity.
the commonly used consolidation theory for designing
PVDs is the unit cell model, e.g., Barron (1948), hansbo
(1981) and terzaghi (1943). Because the solutions
considering both vertical and radial drainage are
complicated, those most used in practice ignore the effect
of vertical drainage, such as Barrons theory. Barron
(1948) developed solutions for two types of boundary
conditions at the surface of improved soil such as: (i) free
vertical strain, resulting from a uniform distribution of
vertical load, and (ii) equal vertical strain, which results
from imposing the same vertical deformation. however, in
some cases, the vertical drainage by pdV has a
considerable effect on the degree of consolidation of
improved soil; terzaghi (1943) suggested the well known
simple method for one-dimensional (1d) vertical
consolidation condition.
furthermore, for most cases in practice, the soil is not
homogeneous, and the deformation of pVd improved soil

(1 U) = (1 Ur) (1 Uv)

(1)

Ur and Uv are respectively the radial and the vertical


average degree of consolidation.
Theoretically speaking, Carillos formula (Eq 1) is only
valid for instantaneously applied loading.
the consolidation of soft soil is related to the dissipation
of excess pore pressure generated by the surcharge load.
for radial consolidation problem with centered vertical
drain in oedometer cell, the governing differential
equation of excess pore pressure is (parakash et al, 1996):

2 u r 1 u r
u r
c r

2
t
r r
r

(2)

cr is the coefficient of radial consolidation of soft soil


r

and (u ) =u(r, t) is the excess of pore pressure at


radius r and time t.
solution of equation (2) that uses the condition of equal
vertical strain without smear effect is given by (Barron,
1948):

8t

U r 1 exp r
fn

1
2509

(3)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

The smear zone is defined as the remolded zone of soil


immediately adjacent to the drain. F(n) is a Barrons
function given by :
n
3n 1
Fn
ln n

n 1
4n

with dimension less than 80 m, it also includes a high


fraction of silt.

(4)

n is the drain spacing ratio given by:


(5)

D and dw denote the equivalent diameters of unit cell and


of PVD, respectively.
Tr is the dimensionless time factor of consolidation due to
radial drainage is written in function of time t:
Cr t
D

(6)

.
For vertical consolidation problem, the differential
equation of one-dimensional consolidation for the excess
pore pressure is written (Terzaghi, 1943):

2 u z
u z

CV
z2
t

- Uv < 50 %:
TV

(8)

- Uv > 50%:
U
V 1

T
8
exp V

(9)

Tv denotes the time factor of vertical drainage:


Cv t
H

60
50
40
30

(10)

H is the drainage distance that is equal half of the


thickness of specimen.

STUDIED SOIL

Tunis soft soil specimens used in this study were obtained


from the Sejoumis lagoon at depth of 17.25 m. The
extracted sample is grey coloured, it has a characteristic
smell and contains shell debris. From grain size analysis
performed by hydrometer and sieving in accordance with
standards NFP 94-056 and NFP 94-057, (AFNOR, 1995),
it was found that Tunis soft soil presents 85 % of particles

0,1

0,01

1E-3

sieve's diameter (mm)

(7)

Cv is the coefficient of vertical consolidation.


Solution of the differential equation (7) is the vertical
degree of consolidation U v as follows:

2.

Clay

70

10

Figure 1. Gradation curve of Tunis soft soil

3.

depending of the depth z and time t;

Tr

Silt

20

(u z ) = u (z, t) is the excess of pore pressure

UV 2

Fine sand

80

D
dw

Tr

Coarse sand

90

Percentage passing, (%)

Gravel

100

CONSOLIDATION TESTS

Three series of consolidation tests were carried out on the


Tunis soft soil in oedometer cells . These tests involved
applied increments of vertical load to the specimen and
measurements of the settlement. For each increment of
loading, the decrease of the thickness of the sample versus
time is recorded. Duration of the applied increment of load
depends on the soil and its consolidation characteristics.
The range of applied stress depends on the range of
effective stress which is needed in the consolidation
analysis of the studied. When the primary consolidation at
prescribed load level is completed (200 kPa) the sample is
unloaded in one or several steps until the increment
(9) of
load of 25 kPa is dismounted and the swelling of specimen
can be measured. The applied vertical load is doubled at
each increment until reaching the maximum required load
(50, 100, 200,400,800 kPa). The specimen is again
unloaded. At the end of the test, the sample is careful
removed and its thickness and water content are measured.
Series 1 (VD): It corresponds to a standard oedometer test
performed according to NF P94-90-1 standard (French
Standard, 1997). This test is carried out on a cylindrical
sample of saturated soil with 70 mm diameter and 19 mm
thickness. The soil sample is enclosed in a metal ring and
is placed on a porous stone. The loading cap has also a
porous stone, so the sample is sandwiched between two
porous stones at the top and bottom of the sample to allow
vertical drainage (VD).When preparing the sample, filter
papers are placed between the soil and the porous stones.
The sample is then placed in the consolidation cell and the
unit cell. Water is added into the cell around the sample,
so the sample remains saturated during the test.
Series 2 (RD): It corresponds to an oedometer test
performed on Tunis soft soil improved by a single
geodrain (Mebradrain 88) of sizes (thickness = 0.5 cm,
width = 1cm and length= 19 mm). In these tests only

2
2510

Technical Committee 211 / Comit technique 211


th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

radial drainage (rd) is allowed, vertical drainage is


prevented by mean of an impervious membrane which
covers the porous stone at the top and the bottom levels of
the specimen.
series 3 (V&rd): it corresponds to an oedometer test
performed on tunis soft soil improved by a single of
geodrain (mebradrain 88) type sized as that used in series
2. in these tests the vertical drainage radial drainage are
allowed.
results of the three series of tests presented in figure 2
show the variation of void ratio e in function of the
effective stress plotted in the logarithmic scale for loadingunloading reloading sequences for three series of tests.

Cs(Serie 3)

1,0

Cs(Serie 2)

0,9

Cs(Serie 1)

0,8

void ratio,e

4.2
coefficient of permeability
Vertical
and
radial
hydraulic
conductivities
(permeability coefficients kv and kr) are determined by
the variable head permeability test. in fact, oedometer
apparatus (in series 1 and 2) is equipped with a
conventional measuring device (tubes connected to the
base of the specimen). the measurements are performed
for different levels of applied load form 100 kpa to 800
kpa (100, 200, 400 and 800 kpa).
figure 9 shows opposite variations of the ratios cr/cv
and kr/kv when the consolidation stress varies from 100
to 800 kpa. in this range, ratio cr/cv varies from 36 to
12 and ratio kr/kv varies from 4 to 12. obtained results
show that the assumption made e.g. cr/cv = kr/kv is
only valid at high levels of consolidation stress (Jia and
chai, 2010).

0,7

Cc

0,6

40

Cc
Cc

0,5
Serie 1

Serie 2

35

Serie 3

0,4
1

10

100

Variation of Cr/Cv and Kr/Kv

log(kPa)

figure 2. oedometer curves obtained from three


experimental series

compression cc and the swelling cs indices, obtained


from the three series of tests (Vd, rd and Vr&d) were
determined from oedometer curves and summarized in
table 1. notice that the compression index obtained from
series 3 (Vr&d) is roughly the double of that recoded in
series 1 (Vd) and 2 (rd). this can be explained by the
allowed vertical and radial drainage paths from which
follows enhanced consolidation of the compressible soil.
from table 1, it is understood the swelling is only
attributed to vertical infiltration of water with sample
unloaded.

25
20
15
10

coefficients of consolidation:

200

300

400

500

600

700

800

900

figure 3. ratios kr/kv and cr/cv versus consolidation stress

degree of consolidation:

in this paper, the global degree of consolidation U(t) is


predicted by two methods. the first one uses the
measured settlement at different levels of applied load in
series 3 (case of vertical and radial drainage, Vr&d):

s( t )
(11)
s
s(t) and s denote respectively the settlements at given time
and at the end of consolidation.
the second method consists in calculating U by the
Carillos equation (1). The radial degree of consolidation
Ur is estimated from the experimental results of series 2
(case of radial consolidation rc) and equations (7) and
(9). the vertical consolidation Uv is obtained from
recorded results in series 1 (case of vertical
consolidation Vc) by using equation (2).
figures 4a and 4b illustrate the variation of global
degree of consolidation U in function of time for vertical
consolidation stress of 400kpa and 800 kpa.
U( t )

4.1

100

verical consolidation stress (kPa)

4.3

stUdY of three dimensional


consolidation:

kr/Kv

table 1: Values of compression and swelling indices


serie of
1 : Vd
2 : rd
3 : V&rd
tests
cc
0.16
0.16
0.30
cs
0.022
0.022
0.023

4.

Cr/Cv

30

coefficients of vertical and radial consolidation cv


and cr are determined from the evolution in time of
settlement for each increment of loading (from 50 to
800 kpa). from the results obtained for series 1 and 2:
cv and cr were determined by the logarithmic
method; which use the plot of thickness of sample
versus the logarithmic of time: log (t) (casagrande,
1938).

3
2511

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

from the observed global consolidation of improved


tunis soft soils was discussed.

1,0

Degree of consolidation

0,8

(400 kPa)

Ucarillo

0,6

Acknowledgement:
authors gratefully appreciate the help provided by mrs.
s. Boussetta during the experimental work carried out at
the soil mechanics laboratory of the national
engineering school of tunis.

0,4

0,2

0,0
-200

200

400

600

800

1000

1200

1400

1600

time (min)

References

(a)

1,0

Degree of consolidation

0,8

(800 kPa)

Carillo

0,6

0,4

0,2

0,0
-200

200

400

600

800

1000

1200

1400

1600

time (min)

(b)
figure 4. Variation of global degree of consolidation

for applied loads (100, 200, 400 and 800 kpa), it is


noted that the degree of consolidation as predicted by
the Carillos theory reaches 100% for a time less than 24
hours, while the degree of consolidation U, estimated
from equation (11) by using measurements of serie 3
results, reaches 100% in 24 hours.
one can also remarks that by using the Carillos theory a
lower degree of consolidation which starts from 10% is
obtained, however when using measurements of in series
3 simple approximate methods, higher degrees of
consolidation starting from 70% are obtained.
comparing between recorded and predicted global
degree of consolidation U, it follows that the evolution
of U predicted by the Carrillos theory are overestimated
with respect to that deduced from recorded settlement
from series 3. the final global consolidation degree U is
identical by using the two methods.
conclUsion
this paper presented an experimental study conducted
on tunis soft clay, in which three types of oedometer
tests were executed: a standard oedometer test; an
oedometer test on specimen soil improved by an element
of geodrain and test a similar test to the second one by
preventing the vertical drainage. from measurements
coefficients of permeability kv and kr were determined
by the variable head permeability test. in addition,
coefficients of vertical and radial consolidation cv and
cr were determined from the evolution in time of
settlement at different levels of consolidation stress.
comparison between the ratios kr/ kv
and cr/cv
demonstrated that equality between the two ratios only
happens at high level of stress consolidation, contrarily
to the common assumption made in previous studies.
predictions of the global degree of consolidation showed
that the Carillos theory leads to overestimated results
with respect to predictions from recorded settlements.
further, the effect of vertical and radial consolidations

Barron a., (1948). consolidation of fine grained soils


by drains wells, american society of civil engineers,
Journal of soil mechanics, Volume 73, pp 718-743.
carillo n., (1942). simple two-and three-dimensional
cases in the theory of consolidation of soils, Journal of
mathematics and physics, Volume, n1, pp 1-5.
casagrande a. (1938) notes on soil mechanics-first
semester. harvard University 1-29.
Guofu.Z and Jian-Hua, (2001). Design charts for
vertical drains considering construction time. Canadian
Geotechnical Journal. 38: 11421148.
hansbo, s. (1981). consolidation of fine-grained soils
by prefabricated drains. proceedings of 10th
international conference on soil mechanics and
foundation
engineering,
stockholm,
Balkema,
rotterdam, 3, pp. 677-682.
indraratna. B, and rujikiatkamjorn .c. (2006)
Predictions and Performances of prefabricated Vertical
drain stabilized Soft Clay Foundations. proceedings
of the symposium on rigid inclusions in difficult soft
soil conditions international society for soil mechanics
and Geotechnical engineering (issmGe tc36).
Jia, r. and chai, J. c. (2010). effect of strain
distribution pattern on interpreting crs consolidation
test results. proceedings of the fourth Japan-china
Geotechnical symposium, okinawa, Japan, 29-36.
Kjellman W. (1948). accelerating construction of finegrained soils by means of card board wicks. in
proceedings of the 2nd international conference on soil
mechanics and foundation engineering,rotterdam. Vol.
2, pp. 302305.
Kjellman W. (1948). Discussion: Consolidation of finegrained soils by drain wells by R.A. Barron.
transactions, asce,113 (2346): 748751.
parakash ,.K, sridharan . a and asha, sr (1996).
consolidation behavior of clayey soils under radial
drainage. Geotech.test. J, astm, 19 (4), 421-431.
sridhar, G and robinson, G (2011) determination of
radial coefficient of consolidation using log t method.
international Journal of Geotechnical engineering (373381).
terzaghi, K. 1943. theoretical soil mechanics. Wiley,
new York.
tyn myint-U. 1980. partial differential equations of
mathematical physics. elsevier, north holland.

4
2512

Improvement of soft fat clay using rigid inclusions and vertical drains
Amlioration dune argile plastique molle par inclusions rigides et drains verticaux
Kirstein J.F.

BVT DYNIV GmbH; Germany

Wittorf N.

Ingenieurbro Dr. Lehners und Wittorf; Germany

ABSTRACT: In the case of a new road crossing in Germany with 1.5 to 7.0 m high embankments nearby the Danish border
particularly soft clays were found 13 to 20 m deep below sea level. The undrained shear strength of the clay varied between 7 and 20
kN/m. The water content was almost 100 % and the organic matter below 6 %. The consolidation coefficient Cv < 0.3 m/year is
characteristics of a fat clay which requires a long time or tight spacing of vertical drains to consolidate. Due to stability risks, vertical
wick drains were installed at a 0.5 m spacing in the part of the highest embankments, which were built in three load steps, each time
waiting for 60 to 80 % consolidation degree before loading the next step. Even using 600 kN/m woven geotextiles, a total vertical
settlement of around 1.5 m and up to 27 cm horizontal deformation were measured throughout one year of monitoring. These
deformations were too high for the existing and running highway in the middle of the new projects. Therefore, full displacement
concrete columns (rigid inclusions system CMC) were installed up to 22 m deep with load transfer platforms installed on top the
inlcusions. In order to improve the installation process of the rigid inclusions, additional vertical drains were installed in the soft soil
before the inclusions. Within the first two years, the area supported by the rigid inclusion experienced less than 2 cm of deformation, a
proportionally small amount compared to the deformations recorded in the wick drain consolidation parts of the project.
RSUM : Pour un projet dune nouvelle route sur des remblais de 1,5 7,0 m de hauteur en Allemagne prs de la frontire danoise,
des argiles particulirement molles ont t trouvs de 13 20 m de profondeur sous le niveau de la mer. La rsistance au cisaillement
de l'argile varie entre 7 et 20 kN / m. La teneur en eau est proche de 100% et la matire organique infrieure 6%. Le coefficient de
consolidation Cv <0,3 m / an montre une argile plastique qui ncessite un long temps ou un rseau de drains verticaux trs serre
pour la consolidation. En raison de calculs de stabilit, les drains verticaux ont t installs avec un espacement de 50 cm dans la
partie des remblais les plus hauts, qui ont t construits en trois tapes de chargement, avec pour chaque tape des priodes d'attente
de 60 80% degr de consolidation avant de la prochaine tape de chargement. Mme avec lutilisation de geotextiles de 600 kN/m,
des tassements verticaux de 1.5 m et des dformations horizontales jusqu 27 cm ont t mesurs pendant une anne de surveillance.
Ces dformations sont trop importantes pour l'autoroute existante en exploitation prs du nouveau projet. Des inclusions rigides
(systeme CMC) ont t installes jusqu' 22 m de profondeur avec diffrents matelas de rpartition placs au dessus des colonnes.
Afin damliorer le processus d'installation des inclusions rigides supplmentaires, des drains verticaux ont t installs dans le sol
mou avant linstallation des colonnes. Au cours des deux premires annes de construction, la zone supporte par les inclusions
rigides a eu moins de 2 cm de dformation, une dformation relativement petite compar avec celles enregistres dans des zones du
projet consolides par des drains verticaux.
KEYWORDS: soil improvement, Controlled modulus columns (CMC), vertical drains
1
INTRODUCTION AND DESCRIPTION OF THE
PROJECT
Large areas nearby the northern sea are nearly flat with
elevations slightly above or under the sea level. Soft soil of silt,
clay, mud and peat reach between five and twenty meters from
the surface, before glacial sands are encountered.
The existing west coast highway B5 near the German city of
Husum will be widened from two to three lanes in the future in
order to improve traffic. The crossing between B5 and B202
was designed as a bridge project with high embankments
located on the unconsolidated soft soils, typical at the flat costal
region near the North Sea.
All traffic constructions bring new loads in form of deadand live-loads to these soft soils. Without soil improvement
methods large long-time settlements will occur, which often
causes damages to the road during the construction or later on.
The traffic on the highways B5 and B202 in the site had to
be maintained during the construction period and the existing
road could not tolerate additional stability risks or settlements,
especially when the 1.5 to 7 m high embankments are built
directly beside the traffic. There are different stages to look at,
but we will focus only on the western part with the highest dam
nearby the bridge.

Figure 1. detail of the highest embankment west with the bridge


abutment over the highway B5 (CMC close to bridge and coloured areas
with vertical drains and preloading)

2513

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Due to stability and settlement calculations the foundation


works took place according to the following sequence of works
and according to the figure 2 below:
1. Installation of vertical drains in different spacings from
a one meter thick sand working platform.
2. Preloading with three load steps with a distance of 30 m
security and working space from the bridge and
existing highway B5. (A)
3. The measured consolidation settlements shown in figure
8 fit with the given predictions according to figure 7.
An additional strong woven geotextile layer of 600
kN/m tensile strength between the embankment and
vertical drains had very little influence on the vertical
inclinometer results with 27 cm of deformation as
shown in figure 9.
4. After waiting for 1.3 m settlement (figure 8) a part of
the embankment and preload was temporarily rebuilt
in order to install the controlled modulus Columns
CMC. (B)
5. The preload was brought back to the edge of the
foundation systems between CMC and vertical drains
area in order to optimize the settlement behaviour.(C)
6. Installtion of deep foundations for the bridge took place
on driven concrete piles with additional sleeves
sockets in the soft soils.
7. The CMC were installed between the driven piles
afterwards, free of vibrations.

4094-4, Part 4 (Deutsche Institut fr Normung 2002). In additon


to borings, several laboratory testing ( water content, organic
matter and plasticity index ) as well as several load-settlement
tests were performed.

F
igure 3. boring, cone penetration test and shear vane test in the detail
area bridge west

The vane tests showed an undrained shear strength of cu = 6


to 8 kN/m near the bridge and an undrained shear strength of cu
= 12 to 20 kN/m in other parts of the project. This was one
more reason to select a CMC foundation nearby the bridge in
the area of the lowest undrained shear strength.
Following this decision and the results of soil investigation
and laboratory the geotechnical engineers assumed an undrained
shear strength of cu = 12 kN/m in vertical drain areas. The
representative soil parameters for the calculation of
consolidation and stability in the project are given in the
following table.
Table 1 . soil parameters for the calculation of consolidation and
stability in the coloured drain areas
density
k/k
[kN/m3
]

shear
strenght
k
[grade]

fill sand

18/10

30,0

---

soft soil, clay


[top level]

14/4

17,5

15

soft soil, silt


top level

15/5

20,0

10

sand

18/10

27,5

---

soft soil, silt


(Bottom
level)

16/6

20,0

10

soil
properties /
soil

Cohesion
C,k Cu,k
[kN/m]

50-100
kN/m

Figure 2. steps of consolidation and construction

The working sequence with different steps was necessary


because of stability calculations and the wide influence of the
settlements during the consolidation. The CMC brought the
following advantages:
- short installation period to complete the project on time
- the vibration free technique allows to work close to the
piles of the bridge
- The settlements of the embankment support on CMC
with a stiff load transfer platform are compatible with
the bridge abutment
2

SOIL-PARAMETERS

After the first part of the soil investigations with several borings
(BS) and cone penetration tests (CPT) it was clear that there
was a problem of stability and consolidation time due to the
presence of fat clay in the upper soft soil layer. The project can
be modelled with two layers of soft soils divided by a loose
sand layer in between. This reaches 13 m up to 22 m in the
deepest parts from the surface.
The undrained shear strength cu in the soft soil from the
results of shear vane tests multiplied with factors of 0.5 to 0.65
are linked to the plasticity according to Bjerum standard DIN

2514

3
3.1

Modulus
Es,k
[MN/m]

Consolidation
coefficient
cv [m/s]

60

6,0*10-1

12

0,8

8,0*10-9

12

0,8

2,0*10-8

25

2,5*10-3

2,0

1,0*10-7

20

SOILIMPROVEMENT TECHNIQUES
Vertical drains

Prefabricated vertical drains were installed in different spacings


with lengths between 15 m (corresponding to the conditions in
figure 3) and 22 m in other parts of the project. It was necessary

Technical Committee 211 / Comit technique 211

to pass the intermediate sandlayer in order to place the vertical


drains in the glacial sand below the second layer of soft soil.
The small spacings in this project were justified by the step
loading and the presence of fat clay in the upper layer of soft
soil with special low permeability and corresponding primary
consolidation coefficient.
3.2

Controlled Modulus Columns CMC

The controlled modulus columns CMC are well adapted to


installation in soft soils. The full displacement auger acts as a
casing and maitain the right borehole diameter over more than
two meter length. Concrete pressure and adequate volume are
monitored and maintained throughout the concreting phase,
which is very critical in very soft soils. The typical piling
standards give a minimum limit of 15 kN/m undrained shear
strength to use for cast-in-place-concrete.
By the standard DIN EN 12699 (Deutsche Institut fr
Normung 2001) above cu = 15 kN/m the minimal distance
between full displacing elements is linked to the undrained
shear strength of the soils. Critical distance is only relevant
during the concrete curing period.
Compared to vibrating techniques, CMC are usually faster to
install and can be performed in softer soils with lower undrained
shear strength. There are several references with CMCinstallation directly adjacent to freshly grouted CMC under
cu < 15 kN/m conditions. In this project the CMC have been
first successfully checked under conditions with the lowest cuvalues by integrity tests and dynamic pile tests. Loads larger
than 500 kN could be tested with a factor of safety larger than 2
FOS on the CMC, drilled into the glacial sand layer.
On part of the project, the process of installing additional
CMCs close to nearby fresh CMC was improved through the
installation of vertical drains in-between the CMC. Immediately
after the CMC installation the water starts to flow out of the
vertical drain even at the top of the sandy working platform. A
continuous flow for several hours up to one day and the volume
of water collected show an efficient fast additional
consolidation.
Compared with other CMC areas the heave of the working
platform and the excessive over-consumption of concrete,
normally increasing with the thickness of softsoil, could be
reduced by the additional intermediate vertical drains.

Figure 5. 129 cm of settlements within year of primary consolidation


with vertical drains spacing of 0.75 m

The stability calculations are based on undrained shear


strength cu and required to build the embankment in three steps
of loading with berms and twice waiting for the sufficient
degree of consolidation necessary. According to (Chaumeny,
Kirstein and Varaksin 2008) the shear strength was calculated
using the following relation to the degree of consolidation:
= U ( tan '+c) + (1-U) cu
U:
:
':
c:
cu:

(1)

degree of consolidation
total load at a given depth
internal friction angle
final drained cohesion
undrained shear strength

Figure 6. stability calculation of three loading steps and control


calculation of the final situation

For this project c = cu in formula (1) as improvement cu


was added to the basic c u value in the stability calculations.
cu = U tan '

(2)

Figure 7. settlement calculations with the three load steps

Figure 4. installation of CMC combined with vertical drains and porewater on the platform

Field measurements and the stability analysis in final


configuration based on ', c and porewater pressure were in
good agreement with the calculations using the improved
undrained shear strength.

4.2

4.1

CALCULATIONS AND PREDICTIONS


Consolidation and stability calculations in the areas
receiving vertical drains

Initially, a total settlement of 1.29 m was calculated in the area


west of the bridge. The time-settlement curves for both primary
and secondary consolidation are shown below on figure 5.

Controlled Modulus Columns CMC

Due to the presence of very soft soils, the CMC are designed to
take the full load of the embankment, neglecting the small load
bearing capacity of the soil in between the inclusions. With
500 kN characteristic load per CMC, the calculated settlement
at the top of each CMC is very similar to the settlement of the
piles under the bridge.

2515

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Nevertheless, below the embankment, there is no concrete slab


or rigid structure like for the foundation of the bridge.
Reinforced earth with galvanized steel was designed to hold the
large horizontal forces of active earth pressure. Because of the
large geotextile deformations during the consolidation period, as
shown in the following monitoring results, the decision was
made to use a stronger more rigid construction with nearly no
deformation. Compared with plastic geotextiles, the steel grid
material has only very small elastic deformations, and as a
result limiting the horizontal deformations of the embankment.
Through the addition of some gravel in parts of the sandy load
transfer platform LTP, the friction between LTP and CMC was
greatly increase and nearly no deformation was necessary to
mobilize the friction of the LTP.
5
5.1

5.2

Controlled Modulus Columns CMC

Several measurement systems were installed between the CMC


and the reinforced earth in the load transfer platform. The
instruments show an almost perfect full stress concentration of
the load on the CMC and less than one centimetre of horizontal
deformation. Figure 10 shows the cross section and the 5
vertical deformation measurements over a period of 2 years.
The horizontal inclinometer was laid across six marked CMCcolumns (figure 1 and figure 9). A settlement of one centimetre
of the top of the CMCs and two centimetres in-between CMC in
the reinforced earth steel construction were measured. There
was a good agreement between the calculated values of the
settlement and the results of the monitoring.

MONITORING RESULTS
Wick drains

F
igure 10. horizontal inclinometer results with arround 1 cm of CMC
settlements and 2 cm of reinforced earth settlements
Figure 8. measurement at the settlement plates SP 9 und SP10.

SUMMARY AND CONCLUSIONS

Soft and fat clay were found at the B5 / B202 road crossing.
Additional soil investigations and laboratory tests were
performed to be able to complete a proper design, regarding
stability and consolidation time.
Oedometer consolidation tests allowed to precisely predict the
movements during the consolidation processes that were
accelerated by the use of vertical drains at different spacings.
Large deformations of up to 1.5 m of settlements and 27 cm of
horizontal displacement were experience and closely match the
calculations and show that it was the right decision not to place
the highest embankment directly on the softest soil beside the
bridge over the running traffic on the highway B5.
Vibration free CMC in combination with reinforced earth
allowed to construct this high embankment with less than two
centimetre differential settlements to the piled bridge.
With a careful planning of the work within the overall
construction schedule, detailed design combined with an
extensive monitoring program, economic soil improvement
techniques can be combined with deep foundations in one
project even on very soft soil can be treated successfully.
7

igure 9. vertical inclinometer results at the 7 m high damm with drains


and 600 kN/m vowen geotextile

The measured settlements during the consolidation process in


figure 8 follow very closely the predictions shown in figure 7.
An additional strong geotextile layer of 600 kN/m tensile
strength between the embankment and the vertical drains had
27 cm of deformation measured with vertical inclinometers.

REFERENCES

DIN Deutsches Institut fr Normung, 2002, DIN 4094-4: Subsoil


field testing part 4 : Field vane test.
DIN Deutsches Institut fr Normung, 2001, DIN EN 12699: Execution
of special geotechnical work - Displacement piles; German version
J-L Chaumeny, J.F.Kirstein, S. Varaksin, 2008, An experience of
consolidation of extremly soft mud for one of Europes largest
projects The AIRBUS A-380 factory in Hamburg, Glasgow.

2516

Interaction of stone column and surrounding soil during its construction: 3D


numerical analysis
Interaction dune colonne ballaste et du sol environnant pendant sa construction : analyse
numrique 3D
Klimis N.S.

Civil Engineering Department, Democritus University of Thrace (DUTh), Xanthi, Greece

Sarigiannis D.D.
AUTh, MSc DIC

ABSTRACT: This work deals with a simulation of a construction sequence of a stone column in two distinct stages: a) a one stage
excavation and b) a multi-stage backfilling of the column stone excavation with crushed gravel at ascending steps of 1m. Simulation
of this procedure is attempted using a 3D model which represents the stone column and the surrounding soil. Analysis is carried out
using a numerical code, called FLAC3D, based on finite differences. The mathematical model incorporates geometry and boundary
conditions of the problem, profile of soil layers with their physical, deformational and mechanical properties and their constitutive
laws, as well as, initial conditions of stresses and deformations of subsoil stratums of the examined area. Special emphasis is given to
simulation of an harmonically imposed vertical loading of the vibrating column, into an equivalent static vertical loading and
subsequently into an equivalent radial pressure against internal wall of the cylindrical excavation of the constructed stone column.
Results clearly denote that there is a strong interaction of the complex system in the kinematical and stress field, which satisfactorily
justifies modification of the final diameter of the constructed stone column compared to the theoretical proposed diameter.
RSUM : Ce travail se rfre une simulation numrique de la squence de construction dune colonne ballaste, en deux tapes
spares : a) une tape unique dexcavation, et b) plusieurs pas successifs de remblayage de lexcavation cylindrique de la colonne
ballaste, avec du matriau granulaire cras, des pas montants de 1m. La simulation est effectue laide dun modle 3D qui
reprsente la colonne ballaste et le sol environnant. Le code numrique utilis est FLAC3D et il est bas sur le modle des
diffrences finies. Le modle mathmatique intgre la gometrie et les conditions limites du problme, le profil du sol avec leurs
proprits physiques, mcaniques et de dformation, ainsi que leurs lois de comportement et les conditions initiales de la rgion
examine. Une attention particulire est donne la simulation dun chargement harmonique vertical impos la colonne vibrante,
un chargement quivalent vertical statique, et par la suite, une pression quivalente radiale exerce sur lintrieur de lexcavation
cylindrique de la colonne ballaste construite. Les rsultats dmontrent clairement linteraction prononce du systme complexe, qui
justifie aisment le grossissement du diamtre construit par rapport au diamtre thorique conu lors du dimensionnement du projet.
KEYWORDS: stone column, excavation, multi-stage backfilling, Flac3D, interaction, complex system, diameter.
5

INTRODUCTION SCOPE OF THE WORK

The present work focuses on the investigation of kinematic and


strain interaction of a complex system consisting of a single
column stone and the surrounding soil, during the excavation
stage and the backfilling stage with crushed gravel.
The scope of this work is the investigation and a possible
explanation of the problem concerning modification of the
constructed stone column diameter, versus the theoretical
(design) one, taking into account the procedure of the stone
column construction, its geometrical characteristics and the
geotechnical model representing the surrounding soil and its
physical, deformational and mechanical properties.
In the framework of this work, a summary of geological,
geophysical, geotechnical and seismological data are presented
in a succinct way in the following chapters, for the examined
area, based on a number of corresponding projects performed in
the recent past. After a short technical description of the stone
column constructing procedure adopted for this project, the
numerical model is determined and numerical analyses results
are presented, in an attempt to explain the deduced discrepancy
between constructed and designed stone column diameter.
The examined area is located in the wide bed of a river in
northern Greece, prone to liquefy, where a bridge is founded.
2. GEOLOGICAL AND SEISMOLOGICAL
2. GeoloGical and seismioloGical
DESCRIPTION OF THE SITE
description of the ste
According to geological and geotechnical data, resulting from
preceding investigation projects on this area, the surface is

covered by deposits that belong to the Quaternary and is


subdivided into: a) river deposits (RD) consisting of silty sands,
clay-silty sands, gravels and locally cobbles of gneiss or marble,
and b) alluvial deposits (AL), consisting mainly of sands with a
largely fluctuating percentage of clays, silts and gravels, of a
thickness ranging from 12 to almost 55m.
The geological bedrock of the examined site consists of
rocks of the alpic age and belongs to the Rodopic Mass,
consisting mainly of biotitic gneisses (gn) interpolated by
amphibolites and marbles green-gray coloured. The upper part
of the gneissic rockmass appears intensively weathered to
totally weathered, consisting thus the weathering zone of 2 to
4m of thickness. The permeability of different geological
formations is quite heterogeneous: the riverbed deposits, mainly
gravel consisting (RDg) are a rather permeable soil formation
(k 10 3 m / sec ), whilst alluvial deposits present a rather low
permeability ( 10 7 k 10 5 m / sec).
As for the seismological data, the examined site belongs to
zone I of low seismic hazard, with a horizontal free-field peak
ground acceleration value: amax=0.16g, according to the most
recent Hellenic map of seismic zones, valid from 1/1/2004.
3.

GEOTECHNICAL CHARACTERIZATION

According to the entity of the geotechnical and geophysical


investigation programs performed on the broad area
(geotechnical boreholes, CPTs and Cross-Hole tests), it results
that the prevailing soil formation are alluvial deposits consisting

2517

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

of sands to silty sands, with a high degree of heterogeneity,


characterized by USCS as SP, SW, SM, SM-SP, SM-SW. In
some cases they appear as clayey sand (SC) to sandy clay (CL),
whereas in other cases, they turn out to be gravel layers, such
as: GP, GW, GM, GP-GM. According to the almost 200 SPTs
performed, the mean value of blows was calculated about 23,
with a standard deviation of +11. The whole area, where the
bridge is founded, has been initially divided into three subregions represented each by a different geotechnical design
section (ITSAK & Gazetas 2003), and finally a design
geotechnical section has been attributed to each bridge pier
(Edafomichaniki 2007) used for dynamic analyses purposes.
From various simplified design geotechnical sections, each
per bridge pier, it has been chosen one, for the needs of the
present project, corresponding to a precise pier of the bridge, as
being the most representative of the area, but not the most
conservative one. The soil profile used in the present work, can
be described as follows:
Layer S1A (0 2m): loose to medium dense gravels with sand
and sand or silty sand with local presence of gravels (GP,
SW-SM, SP): NSPT 22, =20.5kN/m3, =360, c=3kPa,
Es=10MPa, =0.33
Layer S1B (2 to 5m): medium dense gravels with sand and sand
to silty sand with local presence of gravels (GP, SW-SM, SP):
NSPT 23, =20.5kN/m3, =370, c=5kPa, Es=12MPa, =0.32
Layer S2A (5 to 12m): medium dense gravels with silt and sand
to silty sand with presence of gravels (GM-GP, SP-SM, SM):
NSPT 25, =21.0kN/m3, =390, c=6kPa, Es=16MPa, =0.31
Layer S2B (12 to 19m): medium dense silty gravels, silty sand
with presence of gravels to silty sand (GM-GP, SP-SM, SM):
NSPT 28, =21.0kN/m3, =400, c=8kPa, Es=20MPa, =0.30
Layer S3A (19 to 23m) and layer S3B (23 to 35m): medium dense
clayey sand-gravels mixture to sandy clay with gravels, or silty
sand-gravels mixture (GC-GM, SM, CL): NSPT 26,
=21.2kN/m3, =370, c=12kPa, Es=15MPa, =0.31.
From 35 to almost 48m the weathering zone of the gneissic
bedrock or highly weathered gneiss is met.
4.

METHODOLOGICAL APPROACH

The analysis was carried out with FLAC 3D numerical code of


finite differences.
4.1

Modeling Procedure

By considering the construction of a stone column in the above


soil profile, simulation of two distinct stages of the construction
of a stone column is attempted using a three-dimensional (3D)
model which represents the stone column and the surrounding
soil. Simulation of soil materials is realized by a 3-diamensional
polyhedral grid with use of the finite difference method. The
mathematical model adopted, incorporates geometry and
boundary conditions of the problem, the profile of soil layers,
physical, deformational and mechanical properties, constitutive
laws for the geomaterials, as well as, initial conditions of
stresses and deformations of the subsoil stratums of the area
under study.
Geometry of the problem is simplified to axial symmetry. A
vertical plane through stone column axis is a plane of symmetry
for the analysis. Model grid is shown in figure (1). Coordinate
axes are located with origin at the base of the grid, whereas yaxis is oriented along vertical column axis and upward. The
initial grid is assigned by 5.0m and 50 units in x-direction, by
5.0m and 50 units in z-direction and by 28.0m and 56 units of in
y-direction. A Mohr-Coulomb constitutive model elastoplastic
behavior is assigned to all zones of soil surrounding stone
column, whilst linear elastic one is assigned to stone column
backfilling crushed material. Boundary conditions consist of
roller boundaries along the external grid sides of column axis
and a fixed base. Equilibrium conditions for initial stresses are

based on earth pressure coefficient at rest Ko=v/(1-), where :


Poissons ratio.
The modeling sequence consists of the following stages:
Stage I
: Initial stresses
Establish equilibrium conditions to initialize stresses
Stage II
: Excavation
Stone column excavation at full penetration depth was decided
to be numerically simulated in one and only stage, since in
reality, excavation was accomplished in about 30 min for a
typical stone column of the project, and also, because no steps
of excavation during its construction, could be discretized.
Stage III : Stone Column Construction
In reality, construction of cylindrical stone columns of the
project with a theoretical diameter D=0.8m and a length
L=23.0m, is realized by ascending steps of 0.5m; at each step,
the crushed geomaterials are driven through the top of the stone
column downwards (top feed method), and then, the vibrational
torpedo is sinked into the excavated cyclic area, reaches the top
of the crushed material and starts vibrating harmonically at a
frequence of 30Hz, in order to achieve an harmonically applied
normal stress of 30 to 35MPa. However, our choice of
computational ascending steps to simulate stone column
construction was of 1.0m, since an initial comparative study
between 0.5m and 1.0m ascending steps, revealed no significant
differences, whereas computational time difference was
important. Therefore, Stage III is sub-divided in two distinct
calculation steps, ever after named as Sub-stage IIIa and IIIb
Sub-stage IIIa : Simulation of Vibration and Compaction
Based on the construction procedure concerning the one stage
of excavation of the stone column to be realized, which affects
significantly the mechanical properties of the surrounding zone,
a weak zone boundary has been created, by reducing & c, in
a distance of 0.60m surrounding column lateral sides, in order to
simulate relaxation due to excavation. The width of the weak
zone, the reduced values of the mechanical parameters and the
elastic deformation modulus, resulted from a trial and error
back calculating procedure, based on the quantity of the crushed
material measured in situ, during the construction of a stone
column of the project. Namely, we tried to match the increase of
the as built diameter of the examined stone column, in
agreement with the quantity of the crushed material used for the
construction of the stone column, by adjusting the values of
mechanical and deformational parameters of the disturbed zone.
Vertical normal stress, harmonically applied on top of filling
crushed material in order to compact the crushed fill material,
per numerical ascending step of the stone column construction,
is transferred as a lateral pressure p to simulate subjected
compressive lateral loads of material due to gravel compaction,
in terms of an equivalent static lateral (radial) pressure, as
explained in the following paragraph.
Sub-stage IIIb : Simulation of Crushed Stone Material filling
This sub-stage simulates filling of the stone column crushed
material taking under consideration the preceding compaction
procedure. In order to maintain the shape of the deformed
diameter per constructed step of the stone column, crushed fill
material, considered as a linear elastic one, it has been attributed
a very high modulus of elasticity, avoiding thus a rebound of
the plastic lateral displacements obtained from sub-stage IIIa.
4.2

Assessment of equivalent lateral static loading

It is widely known in Mechanics, that a dynamic system


responds to an harmonic external loading, according to the
following equation:
u f ust

1
2

1 f / f 2 4 2
1

(1)

where, u(f): dynamic displacement, ust : equivalent static


displacement (=P/K), : frequency of the input motion, 1:

2518

Technical Committee 211 / Comit technique 211

predominant frequency of the system (herein: the soil column


overlying gneissic bedrock), and : damping ratio of the system.
FLAC3D 3.10

Pcyclic qcyclic

2006 Itasca Consulting Group, Inc.


Settings: Model Perspective
13:40:35 Sat Sep 27 2008
Center:
X: 6.836e+000
Y: 9.322e+000
Z: 5.757e+000
Dist: 7.957e+001

column of a diameter d=0.40m. The vertical harmonic loading,


is calculated, as follows:

Rotation:
X: 140.000
Y: 130.000
Z: 360.000
Mag.: 1.25
Ang.: 22.500

Magfac = 0.000e+000
Live & unassigned mech zones shown

Axes

z ,st

Linestyle

Itasca Consulting Group, Inc.


Minneapolis, MN USA

Figure 1. Model grid used for 3D numerical analyses.

From equation (1), it results that ratio u(f)/ust is greater than


1.0 when f/f1<1.0, and vice versa, when f/f1>>1.0. In this last
case, it results:

f / f1

2 1

(2)

Based on the aforementioned, in order to use an equivalent


static loading instead of a dynamic or harmonic one, we need
to use a coefficient b(f), defined as in equation 2. As b(f) is
proportional to u(f)/ust, it is evident that it will be inversely
proportional to loadings, i.e. the ratio Pst/P(f). Therefore:

b f

P f
Pst

2
1 f / f 2 4 2
1

(3)

In the present problem, it can be assumed approximatively, that:


f1

VLa

4H

3VS

8H

(4)

where, VLa: wave velocity according to Lysmer (VLa1.5VS),


VS: shear wave velocity, and H: depth of the soil column
overlying the gneissic bedrock.
Consequently, for the examined case , where a mean depth of
the soil column is admitted as: H=30m and VS30250m/sec, the
predominant frequency of the system for vertically induced
harmonic external loading, can be roughly approximated, as:

f1

30 x

3.14 x0.4 2
3.768MN
4

(6)

providing thus an equivalent static vertical loading


Pst30%Pcyclic=0.3 x 3.7681.13MN, and an equivalent vertical
normal stress that is estimated to compact vertically the crushed
fill material of the stone column at every step of construction:

Surface

u ( f ) / u st 1

d 2

3 250m / s

3 Hz
8 30m

(5)

For input motion frequencies ranging from 20 to 35Hz (mean


estimated value of 30Hz) and mean estimated value of damping
ratio =20% (Mylonakis et al 2006), equation (3) results b0.15,
which represents a reductional coefficient due to the frequency
of the input motion. It is estimated that due to a large number of
uncertainties of the system, and also because the examined
system is not a single degree freedom oscillator, it would be
wiser to impose a factor of safety of 2.0, resulting thus to a
design coefficient bdesign=b x 2 = 0.3. Accordingly, it results that
Pst30%Pcyclic.
Based on the above, vertical harmonic loading imposed by a
hydraulic vibrating torpedo, can be calculated via cyclic normal
stress (30 to 35MPa) applied through the edge of the vibrating

1130 x 4
1777 kPa
3.14 x 0.8 2

(7)

According to linear elastic theory, earth pressure coefficient at


rest, equals to: k0 = /(1-) = 0.3/(1.0-0.3)0.429, and then the
equivalent radial (horizontal) static normal stress is estimated
h = 0.429 x 1777762kPa.
For the numerical analyses performed, for the deeper part of the
stone column it was adopted a radial pressure of 750 to 800kPa,
whereas, it has been progressively reduced as ascending steps of
stone column construction were getting close to the head of the
stone column at free surface until it has almost been nullified in
the last step.
5.

NUMERICAL ANALYSIS IMPLEMENTATION &


RESULTS

Developing a step by step simulation of a stone column


construction (excavation, filling & compaction), analysis results
are mainly concentrated to the plasticity limits of soil strength
and to the outwards lateral displacement of the stone column
excavated sides due to gravel compaction. Plasticity indicators
for shear or tension are divided at a present plastic yield
indicator with symbol (n) or a past plastic yield indicator with
symbol (p). Outwards lateral displacement are being recorded
at every depth level of the stone column, in different grid points
with distance of 0, 30cm, 60cm and 100cm of the excavated
sides of the stone column.
Figure (2) shows plasticity indicators generated due to the
excavation at full penetration depth. It can be seen that one step
column excavation, has no remarkable effect at inwards
horizontal displacements. At this case, plasticity limits of soil
strength developed in a distance of 0.20-0.40m surrounded
excavated sides. Inwards horizontal displacements of the
excavation are limited in a range of 4-5mm with maximum
values appearing at deeper levels of excavation.
Sub-stages IIIa & IIIb simulate the compaction/filling of
crushed stone material and interaction of the above to
surrounding soil. Figures (3) and (4) exhibit plasticity indicators
for two different construction depths from 16m to 15m and from
1m up to the head of the stone column (free soil surface)
respectively. Although, most of plastic indicators, reveal a past
plastic yield (indicator p) in shear or tension, plasticity
disturbance of the soil is generated in a remarkable distance of
1.0 to 1.2m surrounding column sides for the first example and
in almost the entire surface area of the surrounding soil at the
second one. Low initial stress state at free soil surface, leads to a
remarkable plastic yield over limit close to the stone column
head, even though equivalent static normal radial stress is very
low. Concerning lateral outwards displacement of stone column
excavated sides, due to gravel compaction/filling, shows that
values between 10 and 20cm keep well at a distance of 100cm
of the excavated sides. Indicatively, outwards radial
displacement values (at excavated sides) for depths at 22.5m,
11.0m and 1.0m are in a size of 23cm, 12cm and 20cm
respectively. In general terms, outwards horizontal
displacements are eliminated at distances more than 60cm of
excavated sides.

2519

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

FLAC3D 3.10

2006 Itasca Consulting Group, Inc.


Step 3676 Model Perspective
10:49:27 Mon Sep 29 2008
Center:
X: 1.126e+001
Y: 2.046e+001
Z: 9.320e+000
Dist: 7.957e+001

Rotation:
X: 110.000
Y: 130.000
Z: 360.000
Mag.: 3.05
Ang.: 22.500

1716
13
11
1007
1006
1003
1001

Surface

Magfac = 0.000e+000
Exaggerated Grid Distortion
Live mech zones shown

2726
23
21

Axes

37
36
33
31

Linestyle

Block State

Live mech zones shown


None
shear-n shear-p
shear-p
shear-p tension-p
tension-p

47
46
43
41
57
56
53
51

History Location

67
66
63
61
77
76
73
71

Itasca Consulting Group, Inc.


Minneapolis, MN USA

Figure 2. Plasticity zones during the one stage excavation of the


examined stone column
126
123
121

FLAC3D 3.10

2006 Itasca Consulting Group, Inc.


Step 42971 Model Perspective
07:52:29 Fri Sep 26 2008
Center:
X: 6.840e+000
Y: 8.394e+000
Z: 5.784e+000
Dist: 7.957e+001

Rotation:
X: 120.000
Y: 130.000
Z: 360.000
Mag.: 5.96
Ang.: 22.500

137
136
133
131

147
146
143
141

Surface

Magfac = 0.000e+000
Exaggerated Grid Distortion
Live mech zones shown

Axes

Linestyle

Block State

Live mech zones shown


None
shear-n shear-p
shear-n shear-p tension-p
shear-n tension-n shear-p tension-p
shear-p
shear-p tension-p
tension-n shear-p tension-p

History Location
Itasca Consulting Group, Inc.
Minneapolis, MN USA

157
156
153
151

167
166
163
161

177
176
173
171

187

Figure 3. Plasticity zones during multi-stage filling of the stone column


with crushed geomaterial at depth of 16 to 15m simulated by an
equivalent static radial pressure (sub-stage IIIa, 8th ascending step of
construction of the examined stone column)

and the surrounding soil is numerically analyzed with FLAC3D


numerical code based on finite differences.
The numerical code used considered the procedure of
construction, as well as, its effects on the surrounding soil, and
simulated at its best, the physical procedure of the stone column
construction, in a rational and well documented way.
Excavation stage is simulated in one and unique stage,
whereas, construction of a stone column is simulated by a multistage complex procedure divided in two distinct calculating
steps. Those are identified as two sub-stages per ascending step
of construction: a) vibration and compaction, materialized by
application of an equivalent radial pressure against the internal
wall of the cylindrical excavation and b) stone column filling
with a linear elastic geomaterial assigned a high elastic modulus
of compressibility, due to the compaction procedure, preventing
a rebound of the induced radial displacements of the first substage.
Commenting the outcome of numerical analyses performed,
the following points can be outlined:
1. after completion of excavation stage, the plastic zones
developed around the cylidrical excavation are limited,
same as horizontal displacements, ranging from some
millimeters to only a few centimeters,
2. once excavation procedure is completed, it has been
documented via a trial and error back calculating
procedure, that a zone of about 60cm is seriously disturbed,
affecting notably the mechanical and deformational
parameters of the surrounding soil,
3. the stage of constuction of the stone column has been
simulated by a multi-stage procedure of ascending steps of
1m and application of an equivalent static radial pressure,
as defined in 4.2, progressively reduced as ascending
construction steps approached the head of the stone column
at the free surface,
4. horizontal inelastic displacements in the limit of the side
wall of the cylidrical excavation range between 10 and
20cm, resulting thus in an expansion of the constructed
diameter, compared to the theoretical one as designed.

FLAC3D 3.10

Rotation:
X: 120.000
Y: 120.000
Z: 360.000
Mag.: 4.77
Ang.: 22.500

Surface

Magfac = 0.000e+000
Exaggerated Grid Distortion
Live mech zones shown

17
16
13
11

Axes

Linestyle

Block State

Live mech zones shown


None
shear-n shear-p tension-p
shear-p
shear-p tension-p
tension-p

History Location

1007
1006
1003
1001
27
26
23
21
37

Itasca Consulting Group, Inc.


Minneapolis, MN USA

36
33
31

Figure 4. Plasticity zones during multi-stage filling of the stone column


with crushed geomaterial at depth of 1m to head of the stone column,
simulated by an equivalent static radial pressure (sub-stage IIIa, 23rd
final ascending step of construction of the examined stone column)

6.

REFERENCES

Edafomichaniki s.a. 2007. Egnatia Odos s.a., section of Nestos bridge


and road access on it (14.1.2/14.2.1). Geotechnical Final Design
Study (boreholes GT1 to GT5).
Itasca Consulting Group Inc. FLAC3D v3.10 : Fast Lagrangian
Analysis of Continua. Users Manual version 3.10.
Itsak and Gazetas G. 2003. Study of seismic response and evaluation of
liquefaction risk. Issue 1, pp 73.
Mylonakis G., Nikolaou S. and Gazetas G. 2006. Footings under
seismic loading: Analysis and design issues with emphasis on
bridge foundations. Soil Dyn. Earthquake Eng., 26(9), 824-853.

2006 Itasca Consulting Group, Inc.


Step 115079 Model Perspective
20:07:50 Sat Sep 27 2008
Center:
X: 1.265e+001
Y: 1.964e+001
Z: 7.970e+000
Dist: 7.957e+001

7.

CONCLUDING REMARKS

For the needs of the present project it has been decided to adopt
a rather simple, yet representative, soil profile corresponding to
a bridge pier, where typical stone columns of 0.8m diameter and
23m length are constructed, in order to improve foundation soil
behaviour. The complex system consisting of a stone column

2520

Laboratory tests and numerical modeling for embankment foundation on soft chalky
silt using deep-mixing
Essais au laboratoire et modlisation numrique de la fondation dun remblai sur un limon crayeux
mou des sols amliors par malaxage en profondeur
Koch E., Szepeshzi R.

Szchenyi Istvn University, Gyr, Hungary

ABSTRACT: The deep-mixing is nowadays world-wide accepted method as a ground treatment technology to improve the
permeability, strength and deformation properties of soils. Binders, such as lime or cement are mixed in-situ with the soil by rotating
mixing tools. The method is undergoing rapid development, particularly with regard to its range of applicability, cost effectiveness
and environmental advantages. The paper describes the results of laboratory tests on chalky silt samples mixed with cement of
different content. The influence of the different mixing parameters on the unconfined compression strength and deformation modulus
is shown and evaluated. Typical results of the laboratory tests were used in numerical modeling with PLAXIS 3D as input parameters
to study the behavior of a 4 m high embankment constructed on this soil improved by deep mixed columns with different spacing and
diameters. The parameters of the soil improvement technique were analyzed to study their influence on the settlement and the stability
of the embankment. The trends of the calculation outputs are shown and evaluated.
RSUM : Pour lamlioration de la prmabilit, de la rsistance et des caractristiques de dformation des sols mous la malaxage
est considr comme une technique courante, prconise partout. La procdure consiste malaxer, par rotation, les liants: la chaux
ou/et le ciment et le sol in-situ laide de loutil de malaxage par rotation. Grace la diversit technique et aux possibilits
dapplication de lappareillage, ainsi que ses avantages conomiques, tout en respectant les intrts de lenvironnement, cette
technologie approuve un dveloppement continu mme dans nos jours. Ltude a pour but de faire connaitre las rsultats des essais au
laboratoire raliss sur des prauvettes prleves du sol trait avec les liants: la chaux et le ciment, dont la teneur par prouvettes tait
variable. Lanalyse des rsultats de ces essais a mis lvidence linfluence des divers paramtres de malaxage sur la rsistance la
compression simple et sur le modul de dformation du sol trait. Ces rsultats nous ont rendu possible dappliquer le programme
dlments finis PLAXIS 3D, en vue dtudiar une digue de 4m de hauteur, reposant sur des colonnes de sol trait, ayant une
disposition variable et des diamtres diffrents. Le but de cette tude tait de fournir un moyen de calcul qui permat le suivi des
tassements et la stabilit de la digue, en fonction de la variation des paramtres de malaxage.
KEYWORDS: deep mixing, laboratory test, numerical modeling
1

INTRODUCTION

Road and railway embankments have often been constructed on


soft, saturated, organic subsoil. In the future this type of
construction is suspected to increase, due to environmental and
land management considerations. The low strength and the high
compressibility together with the low permeability and the high
creep potential result in stability problems, extremely large
settlements with prolonged consolidation times, and long term
secondary compression. One of the solutions to avoid these
problems is deep-mixing stabilization of the subsoil.
The development of deep-mixing was started in Sweden and
Japan in the late 1960s with the application of a single mixing
tool to produce column-type elements (Figure 1.). Since then,
new technologies using different mixing tools or binder types
have been introduced. Lately, another technology; mass
stabilization, based on Finnish research is gaining acceptance,
where the whole soil mass is treated normally to a depths of 2 to
4 m (Figure 2.).
The goal of deep-mixing is to improve the soil
characteristics, e.g. increase the shear strength and/or reduce the
compressibility, by mixing the soil with some type of chemical
additives that react with the soil. The improvement occurs due
to ion exchange at the clay surface, bonding of soil particles
and/or filling of voids by chemical reaction products.
Mass stabilization is preferred if the subsoil is very poor e.g.
peat, organic clay or soft clay deposits, and the thickness of the
mass to treat is less than 5 m, the height of the embankment is
low, and the main purpose of the treatment is to increase
stability (Allu Stabilisation System). If the main purpose is to
reduce settlements and the weak soil is thicker than 5 m;

approximately 60 cm diameter single columns are used. With


this technology the treated depths can be increased up to 40 m
(Moseley and Kirsch, 2004, Logar, 2012).
Recently, the use of deep-mixing technology has been
planned on several Hungarian railway projects. The Srrt
railway line rehabilitation is one of these projects; the railroad
crosses an area where the subsoil is soft chalky silt. Both deepmixing technologies could be applied on this site. This paper
describes the preparation of their use at this project. Firstly, the
mechanical properties of the improved soil were investigated in
the laboratory, then, the effectiveness of the technology as
embankment foundation was evaluated with the PLAXIS 3D
finite element program using the laboratory test results.

2521

Figure 1.
Column-type deep-mixing

Figure 2.
Mass stabilization

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

LABORATORY TEST RESULTS

2.1

Parameters of deep-mixing technologies

The quality of the mixed depends on the applied binder type and
quantity as well as the ratio of water to binder in the mixture.
These quantites can be expressed by volume or weight. It is
essential that the water content of the original soil is considered
when calculating the water content of the slurry.
The binder quantity is described with the cement factor (
and the in-place cement factor (inpl):
= mcement / Vsoil = binder weight / soil volume [kg/m3],
inpl = mcement / Vmix = binder weight / mixture volume [kg/m3].
The water content of the soil is described with
wT / c = mw,mix / mcement = the total water-cement ratio [-].
The quality of the mixture is generally described with two
parameters:
qu = the unconfined compressive strength [MPa],
E = the Youngs modulus [MPa]).
These mechanical properties are generally measured at 7, 14,
28, 42 and 90 days after mixing, because the stregthening of the
improved soft fine grained soils is a long process, but the qualifying parameter is generally the 28 day unconfined compression strength [Filz et. al., 2003].
2.2

the 28-day unconfined compressive strength of mixtures 2, 3


and 8 (cement content = 150-175 kg/m3 ) was about 330 kPa,
and for 90 days it increased to 500 kPa (50 %). These
mixtures could be accepted for mass stabilization,
the 28-day unconfined compressive strength of the rest of the
samples with cement contents of 200-300 kg/m3 were 5002000 kPa with a 90-day to 1000-3000 kPa (50-100 %).
These would be acceptable for column-type deep-mixing.
Table 2. Parameters of the mixtures

mix- wT / c
ture
kg/m3
P1
6.8
102
P2 13.5
51
P3
3.4
204
1
5.3
127
2
4.4
153
3
3.8
178
4
2.7
254
5
2.2
305
6
6.8
108
7
5.5
134
8
4.5
162
9
3.4
214
10
2.7
268

Properties of the chalky silt soil before treatment

Based on the laboratory tests, the main parameters of the


original chalky silt are listed in Table 1.
The soil changes its color if its water content changes: the
in-situ moist soil is pale yellow, while it turns light grey when
drying. It has high lime content; the texture has small roots and
organic threads, and high sensitivity. Based on laboratory tests,
it is classified as highly plastic silt (MH).

unconfined compressive strength qu [kPa]

IP
%
17.7

w
%
71.1

e
Es Es,ur
c
*
MPa MPa
2.08 2.1 15 0.0015 0.038

Data of chalky silt mixtures

In the testing program the use of both deep-mixing


technologies was investigated. Thirteen different mixtures were
prepared by varying and wT / c parameters (Table 2).
Mixtures P1-P3 were made with low water contents and with
slightly-varying cement contents. Mixtures 1-5 were prepared
with lower water contents but highly varying cement contents.
The mixtures 6-10 were made with a little bit greater water
contents and with cement contents varied in the similar range.
Since the water content of the original soil was high the
addition of water was less significant in comparison to cement.
The cement content dominated the behavior of the mixture.
2.4

70
12
303
58
196
301
655
1037
81
81
196
370
508

day
70
11
567
92
343
312
1351
2125
94
165
334
910
1162

80
17
418
69
235
297
878
1487
71
92
231
542
670

93
18
727
88
334
380
1384
2853
82
174
424
1024
1458

90
104
17
980
93
430
598
1900
2991
117
246
508
1559
1952

2500
2000
1500
1000
500
0

P1

20
P2

P3

40
1

60

80
100
hardening time t [days]
6

10

Figure 4 shows the relationship between the 28-day


unconfined compressive strength (qu) and in-place cement
factor (inpl). The exponential trendline fits the points well with
R2=0.97. The chalky silt responded well to cement addition.

Evaluation of stabilized soil parameters

In Figure 3, the increase of unconfined compressive strength


with time is shown. As expected, the strength increases with
time, but the hardening/strengthening rate is different from that
of the concrete. The strength is less than 2.0 MPa for cement
content of 50-300 kg/m3. Generally, 0.5-2.0 MPa 28-day
unconfined compressive strength is required for column-type
deep-mixing, and somewhat lower strength for mass
stabilization (Moseley and Kirsch, 2004). The data presented in
figure three indicate that:
4 tested mixtures (P2, 1, 6 and 7) which have a cement content of 125 kg/m3 or less did not reach 200 kPa unconfined
compressive strength, but 3 of them would be acceptable for
a mass stabilization, only P2 with a cement content of 50
kg/m3 should be considered as too weak,

kg/m3
96
49
187
120
144
166
231
274
97
120
144
188
231

qu kPa
28
42

14

Figure 3. Measured hardening/strengthening of chalky silt mixtures

28 days unconfined compressive strength q u [kPa]

2.3

wP
%
54.4

3000

Table 1. Soil properties of the chalky silt soil in Srrt

wL
%
72.1

inpl

2500
wT/c total water-cement ratio

2000

1500

2,22

2,67

2,74

3,37

3,43

3,81

4,43

4,54

5,33

5,48

6,75

6,81

13,47

q u = 4.810-5 inpl3.1

1000

R = 0.97

500

50

100

150

200

250

300

cement factor in-place inpl [kg/m3]

Figure 4. Measured relationship between qu inpl

Figure 5 shows how the 28-day unconfined strength depends


on total water-cement ratio. Samples with high water content

2522

Technical Committee 211 / Comit technique 211

show very low strength, and improvement of soil with wT / c > 8


is not possible. When wT / c < 4 the strength increased rapidly
with decreasing water-cement ratio, but this also means the soil
is very sensitive to changes in its properties. Since the total
water-cement ratio hardly changes, it is clear that the role of
cement factor is significant.

model partial mass stabilization, 1.8-m diameter equivalent


columns were placed in 2.42.4-m, 3.63.6-m and 5.45.4-m
square grids (Figure 9.). Total mass stabilization has been
analyzed by modeling the treated soil as a homogeneous
composite of mixed and in-situ soils with averaged strength
properties.

28 days unconfined compressive strength q u [kPa]

railtrack

inpl cement factor in-place

49

2000

1500

96

97

144

144

166

231

231

274

120
187

0.6

120

embankment = 20 kN/m
2
E = 50 MN/m = 40 = 10

3.4

188

5.0

q u = 24000/(w T /c) 3.0

1000

R = 0.97

soft soil

Es = 2000 kN/m

gravel

E = 25 kN/m

10

cref = 10 kN/m

unsat=15 kN/m3

= 6 = 22 kN/m3

= 36

distance of equivalent columns


L = 2.4; 3.6 and 5.4 m

column distance

L = 2.0 and 3.0 m


0

1:1.5

Figure 7. The model geometry and soil properties

500

3.0

3.0

3.0

2500

12

14

16

total water-cement ratio wT/c [-]

Figure 5. Measured relationship between qu wT/c


column diameter
d = 0.6 m

Figure 6 shows the relationship between the unconfined


strength and the Youngs modulus. It can be seen that the trendline fits very well. In this respect, the chalky silt of Srrt
behaves as expected: the modulus is proportional to unconfined
strength. The equation from the figure can be simplified to
Eu = 70qu

Young modulus E u [MPa]

250
200

Eu = 0.068 qu
R 2 = 0,97

MohrCoulomb

500

1000

1500

2000

2500

3000

unconfined compressive strength q u [kPa]

Figure 6. Measured relationship between qu E

3.1

Eref
qu
cref

50

Figure 9.
Partial mass stabilization

Table 2. Mechanical parameters of the mixtures

100

Figure 8.
Column-type deep-mixing

The unconfined compressive strength of the 5 mixtures was


used for modeling as base parameters (Table 2). Strength
assigned to the column material in the analysis was assumed to
be half of the unconfined compressive strength measured in the
laboratory. In PLAXIS, this strength is represented by the
cohesion (cref). Based on laboratory tests, Youngs modulus for
the columns was 70 times the unconfined compressive strength.
The value of Poissons ratio was = 0.2 .

(1)

where the units are both in kPa.

150

diameter of equivalent diameter of single


column d = 0.6 m
column d = 1.8 m

MODELING OF DEEP-MIXING TECHNOLOGIES


Site evaluation

The second part of our research program was to apply a


calculation method and give some guidelines for design. Both
technologies (column-type and mass stabilization) were studied
for expected design conditions at the Srrt site. Variation in
soil layering, soil strength and compression parameters, and
embankment height will dictate the choice of technology. The
PLAXIS 3D program was used to assess the effect of
stabilization on stability and settlement.
The geometry of the embankment and the parameters of the
untreated soil are shown in figure 7. Groundwater level was
assumed to be even with the ground surface. Sandy-gravel,
suitable for structural fill, was used for embankment material. A
3-m wide, 52.5 kPa distributed load was placed on top of the
ballast during the stability analysis.
Column diameters were 60 cm, with a 5.0-m uniform length
extending into the gravel layer. The columns were placed in
2.02.0-m and 3.03.0-m square grids (Figure 8). In order to

kN/m2
kN/m2
kN/m2

1.
7000
100
50

2.
1500
200
100

mixture
3.
20000
300
150

4.
40000
600
300

5.
70000
1000
500

The analysis modeled the construction and load stages in


five steps:
placement of deep-mixing soil material,
construction of initial 2-m high embankment in 30 days,
construction of final embankment height in 30 days,
final state (consolidation up to 5 kPa pore pressure).
stability analysis considering traffic load.
3.2

Analysis of settlement reduction

The results were evaluated by plotting the calculated settlements


versus the unconfined strength of improved soil elements
(Figure 10.). The following conclusions can be drawn:
with increasing strength all technologies reduce settlement,
but the effectiveness depends significantly on column
diameter and spacing,
there is a relation between column spacing and qu. If qu is too
small, the column spacing is no longer effective, no matter
how close. At a higher qu, the column spacing scheme is
efficient,
for partial and total mass stabilization settlements reduce
rapidly as qu, increases, up to 0.4 MPa. Beyond this value,
the improvement is much less significant,

2523

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

60-cm diameter column-type improvements reduce the


settlements linearly with increasing unconfined strength, but
not very markedly,
60-cm diameter columns are more effective in 2.02.0-m
grid spacing than in 3.03.0-m, although the settlements are
halved at qu = 1 MPa for the larger grid as well,
there is little difference between the reduction curves of the
60-cm diameter columns in 2.02.0-m grid spacing and of
the 1,8-m diameter equivalent columns made in 5.45.4-m
grid spacing,
the improvement with 1,8 m diameter equivalent columns in
3.63.6 m grid spacing is dramatic, the settlements are
halved at about qu = 0,2 MPa,
total mass stabilization can be the most effective technology. Even for very small unconfined strengths (qu = 0.1
MPa) the settlements are reduced to one-fourth.
40

settlement s [cm]

35
30

without treatment

mass stabilization

5.45.4 m square grid - d=1.8 m

3.63.6 m square grid - d=1.8 m

3.03.0 m square grid - d=0.6 m

2.02.0 m square grid - d=0.6 m

25

E 70 qu

20
15
10
5
0

0,0

0,2

0,4

0,6
0,8
1,0
1,2
unconfined compressive strength q u [MPa]

Figure 10. Calculated relationship between s - qu

3.3

Stability analysis

The influence on sliding stability was evaluated by plotting


safety factor as a function of unconfined strength (Figure 11.).
For untreated soil, SF = 1.18 and it could be significantly
increased with even a slight amount of treatment.

strength values, but quickly reaches a plateau. Beyond this


point the mechanics of the stability failure changes with the
failure surface travelling through the embankment slope
only.
4

For road and rail embankment foundations, soil improvement is


a frequently-used technique. Column-type deep-mixing and
mass stabilization are effective soil improvement technologies
to reduce settlements increase safety against slope failure. To
prepare new railway rehabilitation projects the usability of both
methods was investigated on a special soil type: chalky silt in
Srrt (Hungary).
While the underlying chemistry may be complex, the
performance of the mixed material can be evaluated by standard
laboratory and field tests. Laboratory tests have clearly
demonstrated that the Srrt chalky silt is suitable for
improvement by cement. While it cures relatively slowly, an
adequate strength is reached in about 40 days. Unconfined
strengths up to 1,0 MPa can be reached by adding relatively
small amounts of cement. Its uniform and predictable response
to treatment allows the engineer to design the field
improvement. For example, the relationship between
unconfined strength and total water-cement ratio can be
described with simple equations. The Youngs modulus of the
chalky silt can be calculated as 70 times the unconfined
strength.
Finite element modeling was used to study the effectiveness of
the mixing improvement. Column-type and mass stabilization
scenarios were analyzed using strength and compressibility
values from laboratory test results. Both technologies showed
reductions in settlement and increase in stability. Based on the
figures presented, the effectiveness of various solutions can be
evaluated at the first design stages easily and rapidly. Using the
trends from the figures, an optimal solution can then be arrived at
during the detailed design phase by making only some
calculations with PLAXIS for the actual design conditions.
In the future, a further refinement of the proposed method can
be achieved by assessing and involving the cost-effectiveness of
the alternatives in the design.
5

3,0

CONCLUSIONS

REFERENCES

safety factor SF [-]

2,8
2,6

without treatment

mass stabilization

2,4

5.45.4 m square grid - d=1.8 m

3.63.6 m square grid - d=1.8 m

2,2

3.03.0 m square grid - d=0.6 m

2.02.0 m square grid - d=0.6 m

2,0
1,8
1,6
1,4
1,2
1,0

E 70 qu
0,0

0,2

0,4

0,6
0,8
1,0
1,2
unconfined compressive strength q u [MPa]

Figure 11. Calculated relationship between SF qu

The results can be summarized in the following:


the lines for different diameters and grid spacing are very
similar (except for mass stabilization),
the four lines show that for qu > 0.5 MPa, improvement is not
necessary,
the 0.6-m diameter columns with 3-m grid spacing is the
least effective just reaching SF = 1.4 value with the
maximum strength investigated,
the most effective technology to insure stability is the partial
mass stabilization. With 1.8-m diameter equivalent columns
and 3.6-m grid spacing, the required SF = 1.35 value can be
achieved with even small unconfined strength,
the line for total mass stabilization shows a very different
behavior. It generates a high safety factor even for small

Allu Stabilisation System, http://www.allu.net/products/stabilisationsystem


Brinkgreve R.B.J., Vermeer P.A. (2010): PLAXIS-Finite element code for
soil and rock analyses, Plaxis 3D. Manuals, Delft University of Technology Plaxis bv, The Netherlands.
Dumas, C. et. al. (2003): Innovative Technology for Accelerated Const-ruction of Bridge and Embankment Foundations in Europe, FHWAPL-03-014, 2003, pp. 6-13.
Filz, G.M., Hodges, D.K., Weatherby, D.E. and Marr, W.A. (2005):
Standardized Definitions and Laboratory Procedures for Soil-Cement
Specimens Applicable to the Wet Method of Deep Mixing, GSP 136
Innovations in Grouting and Soil Improvement, ASCE Geo-frontiers,
Reston, Virginia, pp.1-13.
Hayward Baker, (2010) Geotechnical Construction, Construction Techniques, http://www.haywardbaker.com/WhatWeDo/Techniques/default.aspx
Logar, J. (2012), Ground Improvement State of the Art in South Eastern
Europe, 2. Symposium Baugrundverbesserung in der Geotechnik am
13. und 14. September 2012 an der TU Wien, pp. 19-46.
Moseley, M.P., Kirsch, K. (2004): Ground Improvement, Taylor and
Francis, London, pp. 57-92, 331-428.

2524

Assessment of bio-mechanical reinforcement materials influencing slope stability,


based on numerical analyses
valuation des matriaux de renforcement bio-mcaniques qui influencent la stabilit des pentes
par des analyses numriques
Koda E., Osinski P.

Department of Geotechnical Engineering, Warsaw University of Life Sciences - SGGW, Warsaw, Poland

ABSTRACT: The article is an answer proposal for the conclusion stated in European regulations regarding the environment friendly
and more sustainable development, which among others includes utilising secondary and recycled material in order to obtain durable
and stable cuttings and embankments. Bearing in mind that the slope stability and erosion control on embankments are the issues
rising the nowadays geotechnics awareness through all around the world, the paper content provides the alternative engineering
solutions to such problems. The techniques proposed in the paper mainly consist of the proper vegetation cover implementation on
embankment slopes, the reinforcement of earth structures by utilising geotextiles and a combination of those two. Additionally, it is
presented how secondary materials could be used as a vegetation development accelerating and enhancing material. In order to prove
the reliability and efficiency of such activities the laboratory material tests and numerical modeling of slope failures were conducted.
RSUM : L'article est une proposition de rponse la conclusion nonce dans les rglements europens concernant
l'environnement de dveloppement favorable et plus durable, ce qui comprend entre autres l'utilisation des matriaux de rcupration
et de recyclage afin d'obtenir des dblais et remblais stables et durables. Il est admis que la stabilit des pentes et le contrle de
lrosion sur les remblais sont des problmes qui apparaisse comme des priorits pour la gotechnique actuelle. Le papier propose des
solutions dingnierie ces problmes. Les techniques proposes dans le document se composent principalement de la mise en uvre
de couverture vgtale sur les talus, le renforcement des structures en terre en utilisant par gotextiles et une combinaison des deux.
En outre, il est prsent comment les matriaux secondaires pourraient tre utiliss comme un dveloppement de la vgtation
acclrant et en amliorant le matriau. Afin de prouver la fiabilit et l'efficacit de telles activits, les essais de matriaux en
laboratoire et la modlisation numrique des ruptures de pente ont t effectues
KEYWORDS: slope stability, reinforcement, vegetation cover, recycled materials, landfill.

The most significant element of the embankment type landfill


reclamation process is the reinforcement and biological
stabilisation of slopes, which are very sensitive to several
destabilisation processes like i.e surface erosion. The landfill
stability improvement activities are divided into phase 1 technical reclamation (implementation of civil engineering
techniques), and phase 2 - biological restoration (establishment
of the vegetation cover). For both of them it is highly
recommended to use such recyclable materials as sewage sludge
and fly ash as a landfill reinforcement filling (CEN/BT, 2009).
In 2012, in Poland the production of fly-ashes from the coal
combustion was 18.5 mln tones. The amount of slag and ashes
disposed on Polish landfills and usable for the road
embankment construction and land reclamation is 261.8 mln
tons. Furthermore, the annual production of sewage sludge in
Poland is also significant - 500 000 tones, and could be
successfully utilised in landfill reclamation process, as a rich in
nutrients fertilizer (Koda et al 2012). The combination of
carefully selected types of fly-ash, sewage sludge, soil and
vegetation cover can be excellent alternative for the heavy
engineering activities for the landfill slopes reinforcement. All
the presented solutions are based on the analyses conducted at
the Radiowo landfill site located near Warsaw.

The municipal solid waste was disposed there up to the early


90s. The local landslides treatment, changes in further
exploitation, and the reinforcement treatment were required.
Since 1993 only non-composted waste from the compostory
plant has been disposed there (approximately 300 tons/day).
The remediation works on the landfill have been carried out
since 1994. They include: slopes forming and planting, stability
improvement solution, mineral capping, bentonite cut-off wall
as a limitation of the groundwater pollution and a peripheral
drainage.
The in situ and laboratory tests for Radiowo landfill has been
performed since 1993. The field investigation consists of
settlement measurements, geotechnical tests of waste, back
analysis (as well as slope failure tests), quality tests of sealing
(capping layer and vertical barrier) and filter materials. In the
Radiowo landfill case, the morphological composition of waste
creates an additional factor influencing the mechanical
parameters. The organic matter content for non-composted
waste is ca. 5 %. A location map involving cross sections
selected for slope stability analyses is presented in Figure 1.
Nowadays, the landfill site is planning to be adopted as a
winter sports activity complex. The construction plan has
already been accomplished and accepted by a legal body, which
is a requirement when considering new development plan for
contaminated sites (for more detail please refer to Koda 2012).

INTRODUCTION.

SITE DESCRIPTION

The Radiowo landfill (embankment type) was established in


1962. It covers approximately 15 ha and the altitude is 60 m
high. No protection system against the environmental pollution
was introduced into the surrounding area at the start of the
landfill operation.

UTILISATION OF ANTHROPOGENIC MATERIALS

One of the elements of the landfill reclamation process is the


construction of capping system. It is a landfill surface cover
protecting against the rainfall infiltration (limitation of leachate
penetration). It provides good establishment conditions for the
vegetation cover, and significantly enhances slopes stability.

2525

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

I
II
II

III
III

retaining wall

Legend:
designed ski slope

I- slope with grass carpets


II- designed slope with grass
III- slope with trees

Figure 1. Current development plan of Radiowo landfill, and location of cross-section for slope stability analyses.

Because of its appropriate geotechnical properties like


impermeability and good compaction conditions, flay-ashes
mixed with cohesive soil are a great material for the capping
system. Recently it was recommended to use geomembrane
instead of mineral barriers to insulate the surface of the landfill,
however there are lots of disadvantages like decreasing slope
stability or slowing down the bio-chemical decomposition of
waste. Applying mineral capping systems (made of ashes and
sewage sludge), in many cases resulted in enhanced
fermentation processes (Koda 2011).
The flay-ash is basically a by product of the coal combustion
process in power plants. The mineral and chemical composition
is determined by mineral elements present in coal. These
minerals are: iron oxides, carbonates and clayey minerals. The
properties of the flay-ash mainly depend on shape and size
distribution of its particles. The bulk density of ashes contains
in the range of 2000-2500 kg/m3.

2526

The reason for this is some of the particles are filled with
gas. The chemical reactions proceeding during the coal
combustion process produce mineral phases stated in Table 1.
Table 1. Mineral phases of the fly-ash (Koda and Osinski 2011).

Mineral phases
Glass
Millite
Quartz
Hematite
Magnetite
Coke

% content of total mass


60-83
4-25
4-18
0.5-2
1-7
0.5-5

The additional anthropogenic component which also


presents high usability in terms of slope surface reinforcement
is sewage sludge (Katsumi et al. 2010). The mineral elements of

Technical Committee 211 / Comit technique 211

G E O T E X T IL E

G E O G R ID

w i d th 2 .0 m

2 .0m

2 .0 m

w id th 1 .3 m

2.0m

2 .0 m

1 .3 m

1 .3 m

1 .3 m

1. 3 m

1.3m

1.3m

1 .3m

1 2 .5 m

1 2 /2
5 /3

1 3 /2
1 5.0 m

4 /2

MP

1 4 /1

2 1.0 m

3 /2

MT
1 4 /2

4 8 . 0m

2 /2

1 2 .5 m

1 /2

1 5 .0 m

2 1.0 m

2 5 .0 m

5 /4

2.0m

1 .3 m

1 2 .5 m

2 .0 m

5 /2

1 2 .5 m

2 5 .0 m

the sewage sludge are developing slowly and are not exposed to
the erosion processes. This kind of material is hazardous when
disposed but when treated by vegetation and additives it is
safely absorbed and utilized by plants. Additionally it has to be
mentioned that sewage sludge supply is free of charge. The
mixture is applied by hydraulic seeders supplied with high
pressure pumps, which enables spraying on different
soil/material types. The advantage of using sewage sludge is
that seeds are protected from the erosion and excessive drying.
The viscosity of the sludge and its mixing ability with other
components, assure even and smooth protection cover, and
moreover, high adhesion to the sprayed surface. The most
significant advantage of using the sewage sludge is the nutrition
content, essential for the vegetation cover establishment.
Especially the undrained sludge is rich in microelement,
nitrogen, phosphorus, potassium and organic matter. Some of
them are highly valuable for plants (Koda 2011), however
cannot exceed normative values of dry mass. The usability of
ashes and sewage sludge for the geotechnical purpose is
determined by several physical and mechanical properties such
as: capacity index in saturated conditions, grain-size
distribution, maximum dry bulk density, swelling, internal
friction angle, and passive capillarity.

1 /1
MT

2 /1
MT

3 /1

4 /1

5 /1

MP

MP

1 P /1 K /1

10

11

MT

MT

MT

MT

MP

MP

1 2 /1

1 3 /1

V o lu m e tric c o n te n t o f su b s tra te :
1 s a n d m e a s u re
1 , 2 c o m p o st m e a su re
1 ,2 l a y e rs o f s u b s tra te

K /2

1 P /1 K /2

M T - L a w n t yp e m ix tu re

1 P /2 K /1

M P - P a stu re ty p e m ix tu re

1 P /2 K /2

G - G a z o n t y p e m ix tu re

1 P / 1 K /1
1 P / 2 K /2

1 /1 1 /2 3 /2 8 1 1 1 4 /2 - N u m b e rs o f v a ria n ts

Figure 2. Scheme of the experimental plot established at Radiowo


landfill slopes (Koda 2011).

4
VEGETATION COVER AS A RELIABLE METHOD
OF SLOPE STABILITY IMPROVEMENT
Beyond described activities for the slope stability and erosion
control improvement purpose on the Radiowo landfill, there
were also bio-engineering techniques applied with additional
use of geosythetics. Due to the usability assessment of the
compost, from organic waste as an enhancing material for the
grass carpets, an experimental plot was established within the
compostory plant area (Koda, 2012). A composite (grass carpet)
consisting of three elements was constructed: reinforcing
material, substrate and grass seeds mixture was prepared. As a
reinforcing material the geotextile (G) and geogrid (Gs) was
used. A reinforcing material task was to connect particular
elements of the carpet, improving the shear strength and
hydraulics conditions, and also an increase of erosion control on
landfill slopes. A porous structure of geotextile and geogrid
enhances establishment of the root zone deeper into the surface.
During the selection of reinforcing material the mechanical
properties and the stock was considered. The polypropylene
materials guarantee long term durability and resistance to
aggressive environmental conditions. A seeding suspension
consisted of a mixture of three types of grass seeds: lawn type
(MT), pasture type (MP) and gazon type grass seeds (G). A
substrate consisted of sand and compost mixture in three
different volumetric proportions: 1P/1K- 1:1 (1 measure of sand
+ 1 measure of compost), 1P/2K- 1:2 (1 measure of sand + 2
measures of compost), and K- pure substrate (100% compost).
The scheme of experimental plot is presented in Figure 2.
Additionally an application of already described fly-ash and
sewage sludge suspension on such slopes to accelerate the
establishment of a green cover was also provided. The grass
carpets were introduced in order to maintain the observation and
to conduct further research on how does such solution influence
conditions of slopes. The assessment of the effectiveness of bioengineering activity on landfill slopes were undertaken after 1,
2, 6, and 10 years of the experiment duration. The result of the
observation confirms the reinforcing purpose of the method, as
even after 10 years of grass carpets establishment the slopes are
evenly covered with plants, while on the slopes where only
traditional method of planting was applied, the slope conditions
are significantly worse. Additionally, the numerical analyses
involving the influence of reinforcing layer also proved the
correctness of applied method on slope of section I-I marked on
Figure 1 where location map is provided. For the results please
refer to Table 3.

The additional solution improving the slope stability is a


proper establishment of high trees and shrubs on slopes (Coppin
and Richards 1990, Norris and Greenwood 2003, Clark et al.
2003). Such activity was also conducted for Radiowo landfill
site. Comprehensively analysed plant species were selected in
terms of root system characteristics and assimilation ability in
such specific ecosystem as contaminated land (Coppin and
Richards 1990, Greenwood 2006).
In the present study, slopes where the vegetation cover was
applied, have been assessed to see whether implementation of
plants affected the resulting stability significantly. Firstly,
however the geotechnical parameters of waste had to be
determined. For such purpose back analyses, CPT and WST
tests were conducted on site. The back-stability analysis by the
Bishops', Swedish (GEO-SLOPE program) and FEM (Z-SOIL
numerical program) methods were performed for three chosen
cross-sections of Radiowo landfill slopes and were applied for
the shear strength parameters verification. The results are listed
in Table 2.
Table 2. Shear strength parameters for municipal solid waste (Koda,
2011)

Material
non-composted
waste
non-composted
waste + sand
old municipal
waste

[kN/m3]

[]

11.0

20

12.0

25

14.0

26

c
[kPa]

Method

failure tests,
CPT, WST
failure tests,
23
CPT, WST
back-analysis
20
CPT, WST
25

The computations of factor of safety including vegetation


cover influencing slope stability were conducted with use of
General Greenwood Method. Greenwood (2006) developed an
equation, based on the limit equilibrium method, where
parameters of plants existing on the slope are considered. These
parameters are: root reinforcement forces, wind forces, or the
mass of vegetation, or related to these, changes in the pore
water pressure. In Slip4EX the Factor of Safety can be
calculated by using several equations developed by Greenwood
(2006), however in this study the Greenwood General Method
was used, as it presents similar characteristics to other methods
used in this study. A powerful equation of FOS concerning a
vegetation influence, proposed by Greenwood is as follows:

2527

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

(c cv )l (W Wv ) cos (U U v )l (U 2 U 2 v) (U1 U1v ) sin Dw sin ( ) T sin tan


F

(W
) T cos

(c cv )l (W Wv ) cos (U
(U 2D
U v)W
l v )sin
Uw2cos(
v) (U 1 U 1v ) sin Dw sin ( ) T sin tan
F
(W Wv ) sin Dw cos( ) T cos

(1)

(1)

1.32
1.32

Figure
includingthe
theinfluence
influenceofofvegetation
vegetationcover
cover
(cross-section
III-III
please
refer
to Figure
Figure3.3.Numerical
Numericalanalysis
analysis of
of slope
slope stability
stability including
(cross-section
III-III
please
refer
to Figure
1). 1).

where
with vv mean
mean changes
changes
where all
all the
the parameters
parameters indexed
indexed with
according
suchparameters
parameters
accordingtotovegetation
vegetation influence.
influence. Additionally
Additionally such
like
strength) are
arealso
alsoincluded
included
likeDDww(wind
(windforce),
force), TT (tensile
(tensile rooth
rooth strength)
ininthe
basically developed
developedtotoassess
assess
theequation.
equation.This
This method
method was
was basically
the
soil reinforcement
reinforcement by
by
thestability
stability of
of slope
slope according
according to the soil
anchors
effects. By
By using
using the
the
anchors oror geotextiles,
geotextiles, or
or vegetation effects.
Slip4Exspreadsheet
spreadsheet (Greenwood
(Greenwood 2006), itit is
Slip4Ex
is possible
possible totoassess
assess
howthe
thedistribution
distribution and
and type
type of vegetation
vegetation can
how
can influences
influencesthe
the
FactorofofSafety.
Safety. After
After full
full establishment
establishment and
Factor
and grow
growof
ofproposed
proposed
plantsthe
thenumerical
numerical analyses
analyses of
of slope
slope stability
plants
stability were
wereconducted.
conducted.
distribution of
of high
high vegetation
vegetation cover
AA distribution
cover on
on analysed
analysed slope
slope
(cross-sectionIII-III)
III-III)isis presented
presented in
in Figure
Figure 3.
(cross-section
3.
Firstly the
the numerical
numerical analyses
analyses were
Firstly
were conducted
conducted for
for bare
bare
slopes.
The
computations
were
based
on
Bishop
method
slopes. The computations were based on Bishop methodwhich
which
wasemployed
employed during
during analyses
analyses performed
performed in
was
in GeoStudio2007
GeoStudio2007
software.
The
second
step
was
to
determine
software. The second step was to determine factor
factor ofof safety
safety
influenced by plants.
influenced
by plants.
The results obtained proved that the factor of safety for the
The results obtained proved that the factor of safety for the
slopes covered with plant was improved as much as 20%. The
slopes
covered with plant was improved as much as 20%. The
initial results of numerical modeling for bare slopes has
initial
results
of numerical
for term
bare monitoring
slopes has
presented
unstable
condition, modeling
however long
presented
unstable
long The
termonly
monitoring
proved that
no singscondition,
of failure however
were noticed.
reliable
proved
that no
of failure
noticed.ofThe
reliable
explanation
forsings
such state
could were
be a presence
wellonly
developed
explanation
stateThe
could
be a presence
of well
developed
vegetation for
on such
slopes.
example
of results
of numerical
vegetation
on slope
slopes.
The example
of results
of slopes
numerical
analyses for
stability
for bare and
vegetated
is
analyses
slope 3.stability for bare and vegetated slopes is
presentedfor
in Table
presented in Table 3.
Table 3. Comparison of numerical analyses of factor of safety for bare
and vegetated
slopes of numerical analyses of factor of safety for bare
Table
3. Comparison
and vegetated slopes

Factor of safety
Factor of Vegetated
safety
Cross Section
Bare slope
slope
Cross Section
Bishop
method
Greenwood
Bare slope
Vegetatedmethod
slope
I-I
1.30 Greenwood method
1.38
Bishop method
1.35
1.42
I-III-II
1.30
1.38
III-III
1.15
1.32
II-II
1.35
1.42
III-III
1.15
1.32
5 CONCLUSIONS
5
CONCLUSIONS
The instability of slopes is one of the most significant problem
concerning
reclamation
of landfill
sites. Theproblem
partial
The instability
of slopes processes
is one of the
most significant
solution forreclamation
this issue processes
is presented
in the paper.
Therepartial
are
concerning
of landfill
sites. The
availableformethods,
which
relatively
efficient,
solution
this issue
is are
presented
in simple,
the paper.
Thereand
are
available methods, which are relatively simple, efficient, and
cost effective. The use of fly ash and sewage sludge for the
reclamation of the surface of landfill is an alternative.. It also
solves the problem of the ash storage which, from the
economical and environmental point of view, is very positive.
4

The reinforcement
doessewage
not require
heavy
cost effective.
The use ofofflyslope
ash and
sludge only
for the
engineeringofmethods,
basic
solutionis as
proper selection
reclamation
the surface
of landfill
an aalternative..
It also and
implementation
of plants
always
worthwhich,
consideration.
solves
the problem
of theis ash
storage
from theIt is
definitely and
costenvironmental
effective, environment
friendly
reasonable
economical
point of view,
is veryand
positive.
technique
acceleratingoflandfill
works.only
Furthermore,
The reinforcement
slope reclamation
does not require
heavy
other wastemethods,
material basic
like compost
be a great
substitute
engineering
solution could
as a proper
selection
and of
implementation
plants isreclamation
always worth
consideration.
It is The
humus for theofsurface
layer
establishment.
definitely
and reasonable
compost cost
couldeffective,
be used environment
for reinforcedfriendly
grass carpets
production,
technique
accelerating
landfill reclamation
Furthermore,
which positively
influences
the erosionworks.
control
on slopes, a
other
waste
material
like compostprocesses
could be of
a great
factor
which
often determines
slopesubstitute
failure. of
humus for the surface reclamation layer establishment. The
compost
could be used for reinforced grass carpets production,
6
REFERENCES
which positively influences the erosion control on slopes, a
CEN/BT
2009. Earthworks.
Final
Feb. 2009.
factor
whichWG2003
often determines
processes of
slopereport
failure.
Clark L.J., Whalley1 W.R, and Barraclough P.B. 2003. How do
roots penetrate strong soil? Plant and Soil , 255, 93104.
6 REFERENCES
Coppin, N. J., and Richards, I. G. 1990. Use of vegetation in
civilWG2003
engineering.
Butterworths,
London.
CEN/BT
2009.CIRIA
Earthworks.
Final report
Feb. 2009.
Greenwood,
J.R. 2006.
SLIP4EX
A P.B.
Program
Clark
L.J., Whalley1
W.R, and
Barraclough
2003. for
HowRoutine
do
Slope
Stability
Effects of
roots
penetrate
strongAnalysis
soil? PlanttoandInclude
Soil , 255,the
93104.
Vegetation,
and Hydrological
Changes.
Coppin,
N. J., and Reinforcement
Richards, I. G. 1990.
Use of vegetation
in
civil
engineering.
CIRIA
Butterworths,
London.
Geotech.
and Geol.
Eng.,
No. 24 (3), 449-465.
Greenwood,
J.R.Inui
2006.T.,SLIP4EX
A Program
for Routine
Katsumi T.,
and Kamon
M. 2010.
Sustainable
Slope
Stability
Analysis
to Include Proc.
the ofEffects
geotechnics
for reuse
of by-products.
the 6th of
Inter.
Vegetation,
Changes.
Congress onReinforcement
Env. Geotech.,and
NewHydrological
Delhi Vol. 1, 302-317.
Geotech.
Geol.Stability
Eng., No.and
24 (3),pollutant
449-465. transport from
Koda
E. and
2011.
Katsumi
T., Inuilandfills
T., and
2010. Sustainable
remediated
withKamon
the use M.
of Observational
Method.
geotechnics
for reuse
of by-products.
Proc.
the [In
6th Polish].
Inter.
SGGW Press,
Warsaw,
Scient. public.
Noof384
Congress
on
Env.
Geotech.,
New
Delhi
Vol.
1,
302-317.
Koda E. 2012. Development plan of Radiowo landfill site, a ski
Koda slope
E. 2011.
StabilityX Conf.
and pollutant
from
construction.
For city transport
and environmentremediated
landfills with
the use
of Observational
waste management
issues,
Warsaw,
110-117 [InMethod.
Polish].
SGGW
Press,
Warsaw,
Scient.
public.
No 384
[In Polish].
Koda
E. and
Osinski
P. 2011.
Slope
erosion
control
with the use
KodaofE. fly-ash
2012. Development
of Radiowo
a ski Life
and sewageplan
sludge.
Ann. landfill
Warsawsite,
Univ.
slope construction. X Conf. For city and environmentScien.,
Land
Reclam.
No.
43
(1),
1-12.
waste management issues, Warsaw, 110-117 [In Polish].
KodaE.E.,
Gaewski
M. 2012.
Usewith
of fly-ash
Koda
andOsiski
OsinskiP.,
P.and
2011.
Slope erosion
control
the use and
for thesludge.
erosionAnn.
control
on sanitary
landfill
ofsewage
fly-ashsludge
and sewage
Warsaw
Univ. Life
slopes.
GeoCongress
- State
Scien.,
Land
Reclam. No.2012,
43 (1),
1-12.of the Art and Practice in
ASCE
GSPUse
No of
225,
3873-and
3890.
Koda Geotechnical
E., Osiski P.,Engineering.
and Gaewski
M. 2012.
fly-ash
Norris,
J.E.sludge
and Greenwood,
J.R.control
2003 Root
reinforcement
sewage
for the erosion
on sanitary
landfill on
unstable
slopes in Northern
Greece
Italy. in
Inter.
slopes.
GeoCongress
2012, - State
of theand
ArtCentral
and Practice
Conf. on Problematic
Soils,
Nottingham,
414-418.
Geotechnical
Engineering.
ASCE
GSP No 225,
3873- 3890.
Norris, J.E. and Greenwood, J.R. 2003 Root reinforcement on
unstable slopes in Northern Greece and Central Italy. Inter.
Conf. on Problematic Soils, Nottingham, 414-418.

2528

Evaluation of Vertical Drain-enhanced Radial Consolidation with Modified Analytical


Solution
valuation de la consolidation radiale amliore par des drains verticaux par une solution
analytique modifie
Lee C., Choi Y., Lee W.

School of Civil, Environmental, and Architectural Engineering, Korea University

Hong S.J.

Technology Research & Development Institute, Daelim Industrial Co., Ltd.


aBstract: the installation of vertical drains accelerates the consolidation process by reducing the drainage path and
predominating horizontal flow within the soft deposits. however, the radial consolidation of the vertical drain installed into the soft
ground is governed by the permeability of a smear zone. modification of hansbo's analysis is proposed to analyze the degree of
consolidation on a horizontal plane by considering the properties of the soil within the smear zone in this study. a parametric study is
carried out to investigate the effects of the soil properties on the proposed analysis. the proposed equation is observed to be relatively
insensitive to the uncertainty of the horizontal permeability ratio between the undisturbed and smear zones. the validity of the
proposed analysis is examined by comparison with the settlement data from a field measurement. it is revealed that the proposed
analysis provides a reliable prediction on the consolidation rate of soft ground installed pVd.
rsUm : l'installation de drains verticaux acclre le processus de consolidation en rduisant les chemins de drainage et
d'coulements horizontaux prdominant dans les dpts mous. toutefois, la permabilit de la zone dinfluence dtermine le degr de
la consolidation radiale induite par les drains verticaux installs dans les sols mous. Une modification de l'analyse de hansbo est
propose dans cette tude pour analyser le degr de consolidation dans un plan horizontal en considrant les proprits du sol dans la
zone dinfluence. Une tude paramtrique est notamment mene pour tudier les effets des proprits du sol sur l'analyse propose.
l'observation montre que lquation propose est relativement insensible l'incertitude du rapport entre permabilits horizontales
entre les zones non perturbes et les zones dinfluence. la validit de l'analyse propose est examine par comparaison avec les
donnes de tassement d'une mesure sur site. il sest rvl que l'analyse propose fournit une prvision fiable sur le taux de
consolidation des drains verticaux installs dans les sols mous.
KeYWords: permeability, radial consolidation, smear zone, Vertical drain
1

introdUction

the radial consolidation flow into the vertical drain induces


a reduction in the flow channel and an increase in flow rate
approaching the drain. it causes that the hydraulic head is
dramatically decreased as the distance to the drain decreases.
therefore, the permeability of drain and soil near the drain
control the rate of consolidation by the vertical drain. since the
permeability of drain is generally designed to be larger enough
than that of soil, it is known that the well resistance is negligible
if the discharge capacity exceeds the required discharge
capacity (holtz et al. 1987, lo 1991). on the other hand, the
installation of vertical drains induces a soil disturbance in the
vicinity of the mandrel. the disturbed zone, called smear zone,
is an area where has reduced permeability and increased
compressibility comparing with an undisturbed soil. reduced
permeability in the disturbed zone governs the rate of
consolidation, because the hydraulic head loss in soil near the
drain further increases when the permeability decreases.
many researchers insisted that the soil adjacent to the drain
is remolded, and several researches were investigated the smear
effect by obtaining the permeability of the distrubed zone from
the permeability of remolded clay (tavenas et al. 1983, Bergado
et al. 1991, hird and moseley 2000, sathananthan and
indraratna 2006). in this study, modification of hansbos
solution is proposed to evaluate the degree of radial
consolidation, considering the consolidation characteristics of
remolded clay. characterisitcs of the modified solution are
discussed, in comparison with hansbos solution. and the
consolidation settlement predicted by the modified solution is
comprared with measured settlement data in the field.

modified analYtical solUtion

hansbos (1981) solution has been widely used to evaluate the


consolidation behavior with vertical drain. it is simple and
accurate as compared with other rigorous solutions and
numerical analysis (onoue 1988, lo 1991). according to
hansbo (1981), the average degree of radial consolidation ( U r )
by vertical drain is
U r 1 exp(8Th / s )

(1)

where, s = ln(de/dw)+(kh/ks-1)ln(ds/dw)+z(2l-z)(kh/qw)-0.75,
th is a time factor (= cht/de2), ch is the coefficient of horizontal
consolidation in the field, de is the circular diameter influenced
by the drain, dw is the drain diameter, kh is the coefficient of
horizontal permeability in the undisturbed zone, ks is the
coefficient of horizontal permeability in the disturbed zone, ds is
the disturbed zone diameter, l is the drainage path length, and
qw is the drain discharge capacity.
hansbos analysis is based on the horizontal flow
characteristics of the undisturbed zone (kh or ch) to evaluate the
radial consolidation. however, it is difficult to obtain due to the
anisotropy of permeability and difference between laboratory
and field measurement values (Bergardo et al. 1991, chai and
miura 1999).
the consolidation characteristics of the disturbed zone are
homogeneous and isotropic due to the disturbance (lo 1991).
to analyze the radial consolidation based on the consolidation
characteristics of disturbed zone, hansbos solution is modified
in this study. since hansbos solution assumes an equal vertical
strain, the ratio of horizontal permeability between the
undisturbed and disturbed zones (kh/ks).

2529

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

table 1. Void ratio and compression index of each clay layer


property

layer 1
U

el
kh/ks=1
kh/ks=2

layer 2

1.81

r
2.22

kh/ks=20

r
1.35

e0

1.65

1.32

1.95

1.46

0.94

0.78

cc

0.84

0.46

1.04

0.57

0.57

0.34

4.8

22.0

5.0

cv
7.1
3.8
13.0
(10-4 cm2/sec)
note. U: undisturbed clay, r: remolded clay

kh/ks=5
kh/ks=10

layer 3
U

2
40
60
0

kh/ks=1
kh/ks=2
kh/ks=5

kh/ks=10
kh/ks=20

figure 2. profiles of soil properties


figure 1. effect of kh/ks on the degree of consolidation: (a) hansbos
method, (b) proposed method

is the same as the consolidation coefficients ratio between the


undisturbed and disturbed zones (ch/chs). By using this
condition, rearranged the average degree of radial consolidation
( U ' r ) is
U ' r 1 exp(8Ths / ' s )

(2)

where, s = ln(ds/dw)+[ln(de/ds)+z(2l-z)(kh/qw)-0.75]/(kh/ks),
ths is a time factor based on chs.
the permeability reduction in the disturbed zone, frequently
represented as kh/ks, is important factor for the vertical drainenhanced consolidation. the effect of kh/ks on the analysis
results is investigated both hansbos solution and modified
solution. figure 1 shows the effect of kh/ks on U r Th and
U 'r Ths curves. other factors are maintained as a constant
value (de/dw=25, ds/dw=5), and well resistance is ignored.
as shown in figure 1(a), the rate of consolidation by
hansbos method is continuously retarded with increasing in
kh/ks. however, the consolidation rate by proposed method
(figure 1(b)) is slightly speeded up and finally converged with
increasing in kh/ks, because the rate of vertical drain-enhanced
consolidation is governed by the permeability of disturbed zone
(Basu and prezzi 2007).
3

application (BUsan neW-port site)

the consolidation behavior in Busan new-port site is analyzed


to verify the proposed analysis. the rate of consolidation
settlement is evaluated by hansbos solution and modified
solution, and these results are compared with observed
settlements in field.
3.1

Soil properties of clay layers

the profiles of clay layer properties (Busan new-port) are


shown in figure 2. the natural water content (wn) and liquid
limit (wl) vary 35~75% and 40~80%, respectively. the plastic
limit (wp) exists in relatively narrow range from 20 to 30%.

2530

although ocr at shallow depth is slightly larger than 1, the


clay layers can be presumed normally consolidated. Busan clay
can be divided into upper and lower clay layers based on el 30m.
3.2

Consolidation properties of each clay layer

in this study, the clay layers of Busan new-port are divided into
3 layers for the consolidation analysis based on the soil
properties. the consolidation tests were carried out for 50
samples for the natural clay and 3 samples for the remolded clay
to figure out the consolidation characteristics of clay layers.
table 1 shows the void ratio, compression index (cc) and
coefficient of consolidation (cv) representing the each clay
layer.
3.3

Extent of the disturbed zone

in this study, several assumptions are made to evaluate the


extent of the disturbed zone: 1) the soil adjacent to the drain is
completely remolded. therefore, the void ratio of the clay
adjacent to the drain is the same as that of the remolded clay at
the same effective stress level; 2) the void ratio reduction due to
the disturbance around the drains occurs faster than the
consolidation settlement under a surcharge load. therefore,
ground settlement that occurred without applying the surcharge
load is mainly caused by the void ratio reduction due to the
disturbance; 3) the extent of the disturbed zone and the variation
of the void ratio within the disturbed zone are a invariable
property with depth; 4) the shape of disturbed zone is a circular
cross section. With these assumptions, the extent of the
disturbed zone is evaluated from measured ground settlement,
which occurred in the interval between pVd installation and a
surcharge loading.
Burland (1990) suggested that the e-log 'v relation for the
remolded clay:
e r e L ( A B log ' v )

(3)

where, er is the void ratio of the remolded clay, el is the void


ratio at the liquid limit, 'v is a vertical effective stress (kpa),
and a and B are constants. for Busan clay, the values of a and
B are 1.224 and 0.256, respectively (hong 2011). therefore, the

Technical Committee 211 / Comit technique 211

void ratio reduction (e) due to the disturbance can be


evaluated from the liquid limit and the natural water content.

table 2. analysis conditions


case
1
2
3
4
5

figure 3. Variations of void ratio in disturbed zone

figure 3 shows two possible variations of the void ratio with


radial distance from the center of the drain. case a assumes a
constant permeability or void ratio within the disturbed zone
(rs). however, most studies (onoue et al. 1991, indraratna and
redana 1998, shin et al. 2009) were consistently insisted a
decrease in the permeability or the void ratio within the
disturbed zone, although there are some differences in shape of
variation (e.g. linear, bilinear, and parabolic). to consider
variation of the permeability or the void ratio within the
disturbed zone, case b assumes that the void ratio linearly
increases from the value equal to er at the outer boundary of the
fully disturbed zone (rf) to the initial void ratio (e0) of the
undisturbed soil at the outer boundary of the transition zone (rt).
for cases a and b, the volume changes due to the disturbance
induced by pVd installation can be expressed as:

V rs 2 H e /(1 e0 )
V [(r f

Case a (4)

r f rt rt ) / 3] H e /(1 e0 ) Case b (5)

where, e is the void ratio reduction due to the disturbance, and


h is the thickness of the target clay layer.
figure 4 shows the ground elevation and total ground
settlement during the entire period of the improvement. the
measured settlement that occurred between the pVd installation
and surcharge loading is 85.6 cm. this ground settlement could
be occurred by the two reasons: 1) void ratio reduction within
the disturbed zone; and 2) consolidation settlement in the
undisturbed zone due to the sand mat. the consolidation
settlement in the undisturbed zone is calculated as 13.8 cm by
using Zeng and Xie's solution (1989). therefore, the ground
settlement caused by the void reduction within the disturbed
zone is 71.8 cm.

figure 4. Ground level and settlement of Busan new-port site

2531

analytical condition
hansbos method

proposed method

no disturbance, ch = cv

no disturbance, ch = cv

ch = cv in disturbed zone,
ch = 2cv in undisturbed zone

no disturbance, ch = chs

ch = cv in undisturbed zone,

ch = chs in disturbed zone,

kh/ks = 2.5

kh/ks or kh/kf = 2.5

ch = cv in undisturbed zone,

ch = chs in disturbed zone,

kh/ks = 5.0

kh/ks or kh/kf = 5.0

ch = cv in undisturbed zone,

ch = chs in disturbed zone,

kh/ks = 10.0

kh/ks or kh/kf = 10.0

the extent of the disturbed zone (rs) for case a is easily


calculated as 21.6 cm based on 71.8 cm of the ground
settlement. however, it is hard to calculate the values of rf and rt
for case b because both rf and rt values are variables. for the
linear spatial variation, previous studies suggested that the rf is
approximately 1.0~1.6rm (onoue et al. 1991, hird and moseley
2000, sharma and Xiao 2000), where rm is the equivalent radius
of the mandrel. in this study, since the rf is assumed to be 1.0rm
(8.0 cm), calculated value of rt is 4.1rm.
3.4

Consolidation analysis

the consolidation rate of Busan new-port is predicted using


both hansbos method and proposed method. to evaluate effect
of consolidation properties, parametric study is performed for a
set of different conditions, as shown in table 2. in case of
proposed method, two possible permeability variations within
disturbed zone are considered. Based on the pVd property,
dw=6, de=135cm, and qw=15cm3/sec are used for analysis.
figure 5 shows the rate of consolidation settlement predicted
by both hansbo's method and proposed method, and the
measured settlement for the layer located above el -30 m. the
average degree of consolidation ( U ) is calculated by using
carillo's suggestion (1942), and then the consolidation
settlement is calculated by considering the non-linear
relationship between the consolidation settlement and the
degree of consolidation.
as shown in figure 5(a), hansbo's analysis for cases 1 and 2
overestimate the settlement rate compared with the measured
one because the coefficient of horizontal consolidation in the
disturbed zone is assumed to be the same as cv. all cases do not
fit well with the measured settlement. to obtain the best result
by hansbo's analysis, it is necessary to know proper values of ch
and kh/ks. however, the suitable kh/ks ratio appears to vary with
the assumed ch value.
proposed analysis (case a) results show in figure 5(b). the
settlement rate at a certain time is underestimated to compare
with the measured settlement, since the extent of disturbed zone
is evaluated relatively large compared with the real condition
due to an assumption for a constant permeability or void ratio
within the disturbed zone. Basu et al. (2006) suggested the
simplified s for the linear spatial variation in disturbed zone
(case b). Using this suggestion, the settlement rate for case b is
calculated by the proposed method, as shown in figure 5(c).
case 4 (kh/kf = 5.0) is well matched with measured settlement
within 100 days, and then case 3 (kh/kf = 2.5) shows good
agreement with the measured settlement after 100 days. since
the typical value of ch/cv could be larger than 1.0 in nature, the
kh/kf is presumed larger than 3.0, based on the consolidation test
results. the slightly underestimation of the settlement rate
predicted with the presumed kh/kf value may occur due to the
difference in the surcharge schedule. in the analytical solution,

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

within the disturbed zone. the proposed method has advantages


to evaluate the extent of disturbed zone and it is less influenced
by the disturbance effect than hansbos method.
5

figure 5. measured and predicted settlement rate for the layer above el
-30m: (a) hansbos method, (b) proposed method with case a, (c)
proposed method with case b

the surcharge load is assumed to be applied all at once, while, in


the field, the surcharge load is applied incrementally.
4

conclUsion

in this study, the radial consolidation enhanced by the vertical


drain is discussed with the analytical method existed, and the
modified solution is suggested. through parametric study and
comparison between the calculated and measured settlement
rates, the results are summarized as follows.
as the degree of disturbance increases, hansbos analysis
shows that the time factor th increases for a certain degree of
radial consolidation. however, the time factor for proposed
analysis (ths), which corresponds to a certain degree of radial
consolidation, slightly decreases as the degree of disturbance
increases. furthermore, proposed analysis gives the almost
identical U 'r Ths curves when the kh/ks value becomes larger
than 20.
for Busan new-port site, the extent of the disturbed zone is
evaluated using two possible void ratio variations within the
disturbed zone. When a constant permeability or void ratio
within the disturbed zone is assumed, the extent of the disturbed
zone rs is estimated to be 2.7rm. for the linear spatial variation
within the disturbed zone, the extent of the transition zone rt is
estimated to be 4.1rm with the same equivalent radius between
fully disturbed zone and mandrel (rf = 1.0rm).
the settlement rate predicted by the proposed analysis is
well matched with the measured field settlement when the kh/kf
ratio is 2.5 with a linear spatial distribution of the permeability

2532

references

Basu d., Basu p., and prezzi m. 2006. analytical solutions for
consolidation aided by vertical drains. Geomechanics and
Geoengineering: An International Journal 1(1), 63-71.
Basu d. and prezzi m. 2007. effect of the smear and transition zones
around prefabricated vertical drains installed in a triangular pattern
on the rate of soil consolidation. Journal of Geomechanics 7(1), 3443.
Bergado d.t., asakami h., alfaro m.c., and Balasubramaniam a.s.
1991. smear effects of vertical drains on soft Bangkok clay.
Journal of Geotechnical Engineering 117(10), 1509-1530.
Burland J.G. 1990. on compressibility and shear strength of natural
clay. Geotechnique 40(3), 329-378.
carillo n. 1942. simple two and three dimensional cases in the theory
of consolidation of soils. Journal of Mathematics and Physics
21(1), 11-18.
chai J.c. and miura n. 1999. investigation of factors affecting vertical
drain behavior. Journal of Geotechnical and Geoenvironmental
Engineering 125(3), 216-226.
hansbo s. 1981. consolidation of fine-grained soils by prefabricated
drains. Proceedings of 10th International Conference on Soil
Mechanics and Foundation Engineering, stockholm,sweden,
Vol.3, 677-682.
hird c.c. and moseley V.J. 2000. model study of seepage in smear
zones around vertical drains in layered soil. Geotechnique 50(1),
89-97.
holtz r.d., Jamiolkowski m.B., lancellotta r., and pedroni s. 1987.
Performance of prefabricated band-shaped drains. construction
industry research and information association (ciria) report,
research project 364.
hong s.J. 2011. Evaluation of geotechnical properties of Busan
Newport clay, doctoral thesis, Korea University.
indraratna B. and redana i.W. 1998. laboratory determination of smear
zone due to vertical drain installation. Journal of Geotechnical and
Geoenvironmental Engineering 124(2), 180-184.
lo d.o.K. 1991. Soil improvement by vertical drains, doctoral thesis,
University of illinois at Urbana-champaign.
onoue a. 1988. consolidation by vertical drains taking well resistance
and smear into consideration. Soils and Foundation 28(4), 165-174.
onoue a., ting n.h., Germaine, J.t., and Whitman, r.V. 1991.
permeability of disturbed zone around vertical drains. Proceedings
of 1991 ASCE Geotechnical Engineering Congress, Boulder,
colorado, Vol. 2, 879-890.
sathananthan i. and indraratna B. 2006. laboratory evaluation of smear
zone and correlation between permeability and moisture content.
Journal of Geotechnical and Geoenvironmental Engineering
132(7), 942-945.
sharma J.s. and Xiao d. 2000. characterization of a smear zone around
vertical drains by large-scale laboratory tests. Canadian
Geotechnical Journal 37(6),1265-1271.
shin d.h., lee c., lee J.s., and lee W. 2009. detection of smear zone
using micro-cone and electrical resistance probe. Canadian
Geotechnical Journal 46(6),719-726.
tavenas f., Jean p., leblond p., and leroueil s. 1983. the permeability
of natural soft clays. part ii: permeability characteristics. Canadian
Geotechnical Journal 20(4), 645-660.
Zeng G.X. and Xie K.h. 1989. new development of the vertical drain
theories. Proceedings of 12th International Conference on Soil
Mechanics and Foundation Engineering, rio de Janeiro, Brazil,
Vol.2, 1435-1438.

Adjusting the soil stiffness with stabilisation to minimize vibration at Maxlab IV


Asynchrotron radiation facility in Sweden
Ajustement de la rigidit du sol par stabilisation pour minimiser les vibrations Maxlab IV, un
centre de rayonnement synchrotron en Sude
Lindh P.

Peab Anlggning AB, Sweden

Rydn N.

Lund University, Sweden

aBstract: in lund a new next-generation synchrotron radiation facility are under construction, maX iV. this facility requires
extraordinary techniques for the earthworks at site. the vibration requirements are very stringent compared to traditional earthwork
standard. the tolerance is 26 nm (1 s rms above 5 hz) and this requires a very good damping from external and internal vibrations.
different solutions were discussed and simulated during the design phase and the best performance was achieved with a four meter
thick layer of stabilised soil below the concrete foundation. the soil consists of clay till with high clay content. during the design
phase many different binder combinations were tested to meet the design criteria regarding seismic modulus. in order to achieve a
monolith the binders setting time was critical since the soil is stabilised in 0.35 meter layers were the next layer are mixed into the
layer below. the binder to best meet both design and construction requirements were a combination of quicklime and ground
granulated blast furnace slag (GGBfs).
rsUm : Un nouveau centre de rayonnement synchrotron de dernire gnration, maX iV, est en cours de construction lund. ce
centre ncessite des techniques exceptionnelles pour les travaux de terrassement sur le chantier. les exigences de vibrations sont trs
strictes par rapport la norme de terrassement traditionnel. la tolrance est de 26 nm (valeur efficace 1 s rms au-dessus de 5 hz), ce
qui ncessite un trs bon amortissement des vibrations internes et externes. des solutions diffrentes ont t discutes et simules au
cours de la phase de conception et la meilleure performance a t ralise avec une couche paisse de quatre mtres de sol stabilis en
dessous de la fondation en bton. le sol se compose de till argileux forte teneur en argile. au cours de la phase de conception, de
nombreuses combinaisons de liants diffrents ont t testes pour rpondre aux critres de conception concernant le module sismique.
en raison de la ralisation d'un monolithe, le temps de durcissememt tait critique puisque le sol est stabilis en couches de 0,35 mtre
dont la couche suivante est mlange dans la couche de dessous. le liant qui rpondait le mieux aux exigences la fois de conception
et de construction tait une combinaison de chaux vive et de laitier granul de haut fourneau (slGhf).
KeYWords:soil stabilisation, sesmic testing, vibration, p-wave.
1

introdUction

max-lab is a swedish facility for materials research based on


synchrotron radiation. the new version, max iV, will be 100
times more efficient than any now existing comparable
synchrotron radiation facility in the world. the location of the
new max-lab is placed just outside the city of lund in southern
sweden. the geology consists of 12 to 16 meters of soil (clay
till) on top of the bedrock. close to max iV runs a major
highway which will introduce ground vibrations. since the
facilitiesare sensitive to vibrations an extensive measurmet
program of background vibrations vere executed.
several foundation alternatives vere discussed and some of
them were tested with fem-simulations to determine which
alternative that fullfilled the requrement of damping both
external and internal vibrations. the alternative that best
fullfilled external and internal damping was a 4 meter thick
stabilised layer underneath the concrete slab.
1.1

Geotechnical testing

the pre-investigation of the geology included geotechnical


sounding as well as geophysical measurements as well as core
drilling through the soil layers down into the bedrock. after the
in-situ investigation and evaluation a geological model for the
site was developed. from this model minor excavations were
performed for soil sampling. the soils were classified and an
extensive testing was performed to evaluate which binder or
binder combination that was optimal for the soils. the major
parts of the soils were clay till with layers of silty sand till.

three different binders were tested, lime; cement and slag.


the slag was ground granulated blast furnace slag (GGBfs).
the clay till contained up to 40% clay and the sandy silt till has
low clay content. the high clay content indicated that lime
should be used to break up the clay. however, lime alone would
not work with the sandy silt till. the clay till from this area have
been tested in a earlier study and the combination of lime and
slag was discovered to be efficient in this type of soil (lindh,
2004).two different binder recipes were chosen from the initial
laboratory testing;
cement/slag (80/20)
lime/slag (50/50)
during the construction phase of the mock-up, cement and
slag were chosen due to the current weather conditions and the
time schedule for the mock-up. during the seismic testing of the
mock-up cracks were found in the stabilized material. the
results indicated that the cracks were introduced during
construction of the stabilised layers. the layer in question was
milled 50 mm down into the layer below to ensure interaction
between layers. the binders working period was not sufficient
to guarantee that the next layer could be milled into the
stabilised bottom layer without causing cracks. this resulted in
a change of binder to a combination of lime and slag (50/50).
fe-calculations as well as seismic measurements performed
on the mock-up showed that a shear wave velocity needed to be
at least 900 m/s in the stabilised soil.in this case it corresponds
to a compression wave velocity (p-wave) of 1430 m/s. the
seismic velocity testing was performed according to a
methodology developed at lund University and tested on

2533

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

cement stabilised soils (rydnet al., 2006).new samples were


prepared to measure the early strength development, see
figure 1.after more than 1200 hours the samples were removed
from the plastic mould that supported the samples during
compaction and in the beginning of the curingperiod of the
samples. this resulted in a small drop in p-wave velocity.

in order to study the development of p-wave at different


temperatures two pairs of specimens were manufactured. one
pair was stored in room temperature and the other pair was
stored at outside temperature. the difference in p-wave
development is shown in figure 3. the different samples are
denoted ps9 and ps11 in the figure. the sample ps 9 was
stored at room temperature and the sample ps 11 was stored at
outside air temperature.

1800

1600

P-wave velocity (m/s)

the causes of this variation were a combination of different


parameters such as;
Variation in water content
Variation in density
Variation in grading (clay content)
Variation in the degree of pulverization

1400

figure 3.development of compressive wave velocity with time for

1200

1000

2200

Sample 1
Sample 2
Limit

800

2000
1800

200

400

600

800

1000

1200

1400

Curing time (h)

figure 1.the figure shows the development of compressive wave


velocity with time. the drop in velocity after 1240 hours is caused by
removing the samples from the plastic mould.

the seismic measurements of the prepared samples were


performed several times every 24 hours for the first 400 hours.
in figure 2 the measured frequency is shown together with
higher frequency modes.

P-wave velocity (m/s)

600

1600
PS 6
PS 5

1400

800

PS 6
PS 5

PS 6
PS 5

PSPS
11

PS 2
PSPS
22

PS 1

1200
1000

PS 6
PS 6 PS 5
PS 5

PS 6
PS 5
PS 1

PS 6
PS 5

PS 6
PS 5

PS 6 PS 1
PSPS
15
PS 6
PS 5
PS 2

PS 2

PS 9
PS 9
PS 11
PS 9 PS 11
600 PS 1 PS 9
PS 11
11
PSPS
2 PS
9
PS 11

PS 9

PS 9

PS 9
PS 11

PS 9

PS 9

PS 9

PS 11
PS 11 PS 11

PS 9
PS 9PS 9

PS 9PS 9
PS 9
PS 11
PS 11
PS 11

PS 11
PS 11
PS 11

PS 11
PS 11

400 PS 9

PS 11

200
0

200

400

600

800

1000

1200

1400

Curing time (h)

reference and production samples.

figure 2.different frequency modes for sample number 2 after 56


hourscuring. the lowest frequency peak corresponds to a fundamental
mode longitudinal resonance frequency of 2074 hz which corresponds
to a p-wave velocity of 1043 m/s.

the production sample denoted ps9 required more than


2400 hours achieving the limit value of 1430 m/s and the
sample ps 11 did not achieve the required limit. however,
storing a specimen in an outside air temperature is not fully
correct compared to the in situ conditions due to larger volume
of stabilised soil and in the in-situ case the heat transfer is one
dimensional. it does however give an idea of how low
temperature will affect the stabilised soil in-situ.
an example of the in situ seismic measurements is shown in
figure 4.

the longitudinal resonance frequency of 2074


hzcorresponds to a p-wave velocity of 1043 m/s for sample 2.
the samples needed approximately 500 hours of curing in 20
degree celsius to meet the requirements regarding p-wave
velocity.
2

in-sitU measUrements

the quality testing in-situ was done both as ordinary testing


with binder content, mcV, pulverization and evib measurement
with the compaction roller. the testing procedure also included
sampling from the stabilised soil when the mixer had made two
mixing passes. the stabilised soil that would be tested was
excavated and transported to a field laboratory for compaction
in plastic moulds. after compaction the p-wave velocity was
measured and compared with the laboratory mixed samples.
most of the production samples (ps) were stored in room
temperature to ensure the same conditions compared with the
laboratory compacted samples.
there was a great variation in the development of p-wave
velocity versus curing time for different samples, see figure 3.

figure 4.a result from seismic in-situ measurement along the surface of
the stabilised layer is presented in the figure. the top layer has almost
reached a surface wave velocity (~0.92Vs) of 800 m/s at the time for
measurement. the target shear wave velocity after curing is 900 m/s.

the in-situ measurement of the stabilised soil is performed


with the same equipment as used for sample testing. however,
the in-situ testing involves the whole volume and gives a true
value of the stabilised soils performance. the seismic testing
will be followed up in future with testing on the concrete slab.

2534

Technical Committee 211 / Comit technique 211

conclUsions

soil stabilisation with binders increases the stiffness (emodulus) of the soil and thereby changes the resonance
frecuency of the soil. in this project the soilstabilisation has
been a key issue to meet the requrements regarding vibrationsin
a cost effective way.it has been proven that it is possible to
achieve a homogenius stabilised monolitic ring with a
circumference on 528 meters and a depth of 4 meters.
the homogenity of the stabilised material is a result of an
extensive testing program in both laboratory and full scale. the
binders working period has also been an important issue to
ensure a crac-proof construkton.
it has also been shown that seismic testing works very well
for both laboratory and in-situ testing of stabilised soils.

acKnoWledGements

the authors acknowledgepeabanlggning


opportunity to publish this data.
5

aB

for

the

references

lindh p. 2004. compaction- and strength properties of stabilised and


unstabilised fine-grained tills.lund University. department of
Building and environmental technology. division of soil
mechanics and foundation engineering. lUtVdG/tVGt-1013 /
swedish Geotechnical institute, sGi. report 66
rydn n., ekdahl U. and lindh p. 2006. Quality control of cement
stabilised soil Using non-destructive seismic tests. advanced
testing of fresh cementitious materials, stuttgart, august 3-4, 2006 /
dGZfp - proceedings BB102-cd (deutsche Gesellschaft fr
Zerstrungsfreie prfung e.V.) lecture 34.

2535

Construction and Performance of Containment Bund Using Geotextile Tubes Filled


With Cement Mixed Soil in Singapore
La construction et la performance de la digue de confinement utilisant des tubes gotextiles remplis
de terre mlange au ciment Singapour
Loh C.K.

Maritime and Port Authority of Singapore (MPA), Singapore

Chew S.H., Tan C.Y.

Department of Civil Engineering, National University of Singapore (NUS), Singapore

Lim S.K.

Housing and Development Board (HDB), Singapore

Lam J.P.W.

Surbana Corporation Pte Ltd, Singapore


ABSTRACT: In a major port development project in Singapore, a containment bund using modified geotextile tubes (M-GT) filled
with cement mixed soil has been constructed. The main purpose of this bund is to create a containment area to contain any sediment
plumes due to construction activities (i.e. dredging activities, dumping activities and sand-filling activities). The containment bund
also serves as a retaining structure to retain dredged materials during the sand-key construction and other port expansion works. This
paper presents the key consideration in the innovative design and construction of a geotextile containment bund. In addition, instead
of usual sand fill, the dredged soil mixed with cement was used as the fill material in this bund. Among the challenges faced in this
project were the great water depth of this containment bund location (>25m) and high traffic volume in Singapore water course as the
site is next to the existing operating port terminal. Extensive field instrumentation and monitoring were carried out during and post
construction phase to verify the design, as well as ascertain the performance of the geotextiles containment bund system.
RSUM : Dans un important projet de dveloppement portuaire Singapour, une digue de confinement utilisant des tubes
gotextiles modifis (M-GT) et remplis de terre mlange au ciment, a t construite. Le but principal de cette digue est de crer une
zone de confinement afin de contenir les dblais de sdiments provenant des travaux de construction (cest--dire travaux de dragage,
de dversement et de remplissage au sable). La digue de confinement sert galement de structure de retenue pour retenir les matriaux
de dragage lors de la construction de la tranche d'tanchit et d'autres travaux d'extension de port. Cette tude prsente le facteur cl
dans la conception innovatrice et la construction d'une digue de confinement en gotextiles. En outre, au lieu du remplissage au sable
habituel, le sol dragu est mlang avec du ciment avant dtre utilis comme matire de remplissage dans cette digue. Parmi les dfis
relevs durant ce projet taient la grande profondeur des eaux lemplacement de la digue de confinement (> 25m) et le volume du
trafic maritime dans les eaux de Singapour vue que le site se trouve proximit du terminal portuaire existant. Des instrumentations et
mesures approfondies ont t menes pendant et aprs la phase de construction pour vrifier la conception, de mme que la
performance du systme de digue de confinement en gotextiles.
KEYWORDS: Geotextile tubes, containment bund, cement mixed soil.
1

INTRODUCTION

A containment bund consisting of modified geotextile tubes (MGT) filled with cement mixed soil has been constructed for a
major port development project in Singapore. This containment
bund forms part of the Pasir Panjang Terminal Phase 3 & 4
Expansion Project, which is located at the Southern part of
Singapore water (Figure 1). During the project construction
phase, this bund serves as a retaining structure to retain dredged
materials and at the same time contains any sediment plumes
arises from construction activities from being transported
towards the nearby forest reserve area by currents. This
containment bund, termed as a geotextile containment bund, is
being constructed by systematically stacking of modified
geotextile tubes (M-GT) and filling of cement mixed dredged
soil. A typical cross section of the geotextile containment bund
is shown in Figure 2.
A geotextile tube is a tubular container (diameter 1m to
10m) that is formed in-situ, on land or in water, by hydraulically
filling the tube with sand or dredged material (Pilarcyzk, 2000)
and Lawson, 2006). On the other hand, geotextile container is
made of geotextile sheet laid onto a split-bottom barge, filled
mechanically with sand or other fill material, and sewn the top
opening to form into a closed container. The barge will then

move to the desired position, and the bottom of the barge will
open allowing the containers to sink into the sea at the intended
location. The volume of these containers can range from 100m3
to 800m3.

Figure 1 Location of project site in Singapore (Google image)

2537

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

The modified geotextile tube (M-GT) introduced in this


paper is an innovative application, which combines the structure
/ shape of a geotextile tube and the method of installation of a
geotextile container. The diameter of M-GT is 5m and the
length is 25m (limited by barge length). The theoretical
maximum volume of M-GT is 490m3. However, for practical
reasons, the filled volume is only about 290m3 or 60% filled in
this project.

Figure 2 Typical cross section of geotextile containment bund

3) Free-falling of M-GTs onto the seabed Air pockets


inside the tube or container during free-falling would exert
certain forces onto the geotextile and cause higher strain
(Pilarcyzk 2000). Tensions are generated in the tube due to the
balancing of these forces, fill weight, buoyancy, drag, etc.
(Lawson, 2006).
4) Impacting onto the seabed At the point of impact, the
kinetic energy of the falling tube is converted to elastic energy,
which will reshape the tube, from a cone shape into a
transitional cylindrical shape and eventually into a semi-oval
shape or rectangular shape (Pilarcyzk, 2000).
5) Stabilized phase of the M-GTs The final shape of the
tube attained depends on a number of interrelated factors such
as the volume of fill, internal shear resistance of the fill material
and the stiffness of the geotextile material (Lawson, 2006).
There are a number of equations and formulas available for
the determination of the tension development in some of the
stages mentioned above. The equations used in the design of MGT in this project can be found in Chew et al. (2010).
The construction sequence of the bund is illustrated in five
steps (Figure 4(a) to (e)).
a)

Two geotextile containment bunds were constructed in this


project. The length of bund 1 is 500 m and bund 2 is 1800 m.
Bund 1 was constructed first in order to provide a staging
ground for other construction activities at the site. The layout
and length of the bunds are shown in Figure 3.

d)

b)

e)
c)

Figure 4 (a) to (e) Construction sequence of geotextile containment


bund (cross-section view)

Figure 3 Length of geotextile containment bund 1 and 2 (Plan view)

DESIGN AND CONSTRUCTION OF GEOTEXTILE


CONTAINMENT BUND

In the design of this bund, there are a few stability criteria that
have to be fulfilled: Stability against hydraulic force of waves
and current, local stability against sliding failure, local stability
against slip failure, settlement and deformation. The tensile
strength of the geotextile material is one of the major design
parameters. This is because the installation of the tubes at water
depth of 25m is deemed to be extreme in the field installation
of geotextile tubes and containers.
The installation process of the M-GT consists of five (5)
main phases, namely:1) Filling of the M-GT The dredged material mixed with
cement, known as cement mixed soil, is being pumped into the
modified geotextile tube via the inlet ports that are available at
the top face of the M-GTs.
2) Opening of split-hopper barge the bottom of the splithopper barge opens slowly to allow the exit of the filled M-GTs
through its opening. High tension in geotextile is expected to be
experienced at this stage.

2538

USE OF CEMENT MIXED SOIL (CMS) AS IN-FILL


MATERIAL

Discarded soil from other excavation projects on land or sea in


Singapore, and dredged materials from port extension works
have been mixed with cement to form into Cement Mixed Soil
(CMS), and was used as in-fill material in the M-GTs and as the
core of the geotextile containment bund as shown in Figure 2. In
order to satisfy the stability criteria of the geotextile
containment bund, the cement mixed soil has to achieve a
design value of unconfined compressive strength qu of
200kN/m2. After taking into account of soil variability and the
factor between the laboratory test result and in-situ achieved
results, the targeted in-situ unconfined compressive strength is
state as 1.3x200, which is 260kN/m2.
4

PERFORMANCE OF CONTAINMENT BUND

The performance of the bund has been monitored during and


after the construction through an extensive instrumentation plan.

Technical Committee 211 / Comit technique 211

A total of 28 numbers of M-GTs were instrumented and


monitoring at various stages of the installation process. One of
the key parameters was that monitored closely is the strain
development of the M-GT at different stages. The results of the
monitoring during the installation process were presented by
Chew et al. (2011) and it showed that high tensile forces of
about 180kN/m were recorded at the bottom of M-GT during
the impact onto the seabed.
Hydrographic survey was used to monitor the shape of the
installed M-GTs, which is one of the performance indicators of
this design. The accuracy of the installation was determined by
using survey results conducted before and after the dumping of
the instrumented M-GTs. The overall construction progress of
the bund was also tracked using hydrographic surveys that were
conducted every 5 days. The profile of the bund can be plotted
using the survey results as shown in Figure 5, which shows one
of the completed bund.

Figure 5 Profile of completed bund 1

After the completion of the bund, a total of 11 instrument


clusters have been installed to monitor the performance of the
containment bund during other construction activities such as
the filling of dredged soil behind the containment bund and soil
improvement works for the dumped material within the
containment bund.
Out of the 11 instrumented clusters, 6 of them were placed at
the top of the bund and the remaining was installed to monitor
the slope of the bund by using a staging. The cross section of
the instrument clusters is given in Figure 6.

Figure 7 Location of inclinometers at CH. 1370 of the bund (Plan view)

The inclinometer readings show that the maximum


deformation of the bund centre is 10mm at elevation of -8.5m
(Figure 8a). The measurement was taken at 911 days after the
completion of the bund at that location. This shows that the
containment bund has remained stable throughout the period of
other construction activities that occurred during this period.
Figure 8b shows the lateral deformation at the sides of the
bund, which was also found to be within 10mm, where the
maximum deflection occurred close to the bottom of the bund.
Higher lateral deformation of up to 30mm was also recorded by
the inclinometer at elevation above the bund (i.e. -5m to +10m).
The lateral movement above the surface of the bund (side
inclinometer) indicates that the dredged filled material has been
placed onto the sides of the bund and at the same time being
treated.
The settlement measured by extensometers installed on the
top and side instrumentation clusters and settlement plates at the
top of the bund are given in Table 1. The settlement readings
showed that the geotextile containment bund filled with cement
mixed soil has remained stable and performed as expected
throughout the construction period of this project.
Table 1 Settlement of containment bund

Figure 6 Instrumentation cluster installed in the containment bund

The results from the inclinometers installed at the top and


side faces of the bund are discussed here. The location of the
inclinometer is at CH. 1370 (Figure 7). The lateral deformation
in the section perpendicular to the centre line of the bund is
plotted in Figure 8 for both top and side inclinometer.

2539

Settlement(mm)

Elevation

Extensometer

Extensometer

(Center)

(Side)

Settlement
plate

Top

11

3.0CD

24

6.0CD

26

40

9.0CD

25

29

12.0CD

17

23

15.0CD

18

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

(b)

(a)

Figure 8 Lateral deformations measured from inclinometer at: (a) top of the bund. (b) side of the bund

CONCLUSION

The construction of the geotextile containment bund using


modified geotextile tubes (M-GT) filled with cement mixed soil
(CMS) has been completed successfully over a total length of
2.3 kilometres in Singapore. Field measurements of lateral
deformations and settlements showed that the bund has
performed well within the design limits and expectations. The
innovative use of discarded soil from other excavation projects
on land or sea via mixing with cement is proven to be a good fill
material. This CMS material was shown to be able to achieve
highly uniform and well controlled properties, and deemed to be
suitable as in-fill material for geotextile tubes and the core
portion of a containment bund.

2540

REFERENCES

Chew, S.H., Tan, C.Y., Loh, C.K., Lim, S.K., Lam, J.P.W. 2011. Design
and Construction of Containment Bund using Geotextile Tubes in
Singapore. The 14th Asian Regional Conference on Soil Mechanics
and Geotechnical Engineering, Hong Kong
Chew, S.H., Tan, C.Y., and Tan, H.W.A. 2010. Application of
Geotextile Containment System in Coastal, Beach and River
Restoration Projects. The 1st International GSI-Asia Geosynthetics
Conference, Taiwan.
Lawson, C. 2006. Geotextile containment for hydraulic and
environmental engineering. The 8th International Conference on
Geosynthetics, Yokohama, Japan.
Pilarczyk, K.W 2000. Geosynthetics and Geosystems in Hydraulic and
Coastal Engineering. Balkema, Rotterdam.

Reinforcement of completely decomposed granite with discrete fibres


Renforcement de granite compltement dcompos avec des morceaux fibres
Madhusudhan B.N., Baudet B.A.

The University of Hong Kong, Hong Kong


ABSTRACT :The use of discrete fibres as reinforcing material for soils has been researched by many, e.g. Gray and Al-Refeai
(1986), Maher and Ho (1994), Crockford et al. (1993), Santoni et al. (2001), Consoli et al. (2009a), but these studies have been
generally done independently and have not always been consistent. Silva dos Santos et al. (2010) used data gathered through many
years of study to develop a framework of behaviour for a poorly graded quartzitic sand reinforced with polypropylene fibres. In Hong
Kong, the construction industry has used reinforcement with continuous fibres for some time, but it is mainly applied to landscaping
of otherwise stabilised slopes, for example as a green cover on an existing shotcreted slope. Using randomly distributed short discrete
fibres in Hong Kong completely decomposed granite (CDG) could help stabilise the soil while keeping the density low enough to
allow growth of vegetation. It is not guaranteed, however, that a well graded residual soil like CDG would behave in the same way as
sands used by previous researchers. Laboratory tests have been carried out on completely decomposed granite using short discrete
polypropylene fibres as a reinforcing material. The fibres are randomly distributed in the soil. It was found that the fibres increase the
unconfined compressive strength of the CDG prepared at its maximum dry density by up to tenfold for fibre contents less than 1%.
The behaviour of the fibre-CDG mixture during drained triaxial compression changed from dilative to compressive, with more effects
at low confining pressures. These tests seem to indicate that discrete fibres could be considered for improving the performance of
CDG.
RSUM : Lutilisation de fibres pour renforcer les sols ont dj fait lobjet de nombreux travaux de recherche e.g. Gray and AlRefeai (1986), Maher and Ho (1994), Crockford et al. (1993), Santoni et al. (2001), Consoli et al. (2009a), mais ces tudes ont t
gnralement faites indpendamment et elles nont pas toujours t synthtises. Silva dos Santos et al. (2010) ont utilis les donnes
obtenues au cours dannes de recherche pour dvelopper un modle de comportement pour un sable quartzitique uniforme renforc
avec des fibres en polypropylne. A Hong Kong, lindustrie de la construction a utilis des fibres continues comme moyen de
renforcement depuis longtemps, mais lapplication se limite laspect paysager de pentes dj stabilises, par exemple pour la
plantation de surfaces de pentes recouvertes de bton projet. Lutilisation de fibres courtes distribues de faon alatoire dans le
granite compltement dcompos de Hong Kong (CDG) pourrait aider stabiliser le sol tout en gardant sa densit assez basse pour
permettre la vgtation de pousser. Il nest pas garanti cependant quun sol rsiduel a la distribution granulomtrique bien calibre
comme le CDG se comportera de la mme faon que les sables utilises par les chercheurs prcdents. Des essais de laboratoire ont t
faits sur du granite compltement dcompos en utilisant du bton projet et des fibres courtes en polypropylne comme matriau de
renforcement. Les fibres sont distribues de faon alatoire dans le sol. On a trouv que les fibres ont pour effet de multiplier par
presque dix fois la rsistance en compression simple du CDG prpar sa densit sche optimale, pour une teneur en fibres de moins
de 1%. Le comportement du mlange CDG-fibres lors de lessai triaxial drain en compression es tpass de dilatant a contractant avec
plus deffet aux pressions faibles. Les essais paraissent indiquer que lutilisation de fibres courtes pourrait tre considre pour
amliorer la performance du CDG.
KEYWORDS: laboratory tests ; reinforced soils ; residual soil

INTRODUCTION

Adding fibres to soil can be an effective way of strengthening it,


by providing tensile strength at high strains. The factors
influencing the effectiveness of the fibre-reinforced soils are a)
the type of soil and its deformation behaviour; b) the type of
fibre and its specifications (fibre length, fibre content and its
aspect ratio). A careful study of the mechanics of the fibrereinforced soil will help practising and design engineers to
understand better its behaviour under different loading
conditions.
Hong Kong is a modern city with growing population, so
that engineers are pressed to optimise land utilisation. The
topology of Hong Kong has led to urban development on
natural or man-made slopes. Conventional methods of
stabilising slopes such as shotcreting the whole face of the slope
(current practice) are neither cost effective nor environmentally
friendly and alternative sustainable methods are being sought
after.
Many researchers have produced a large body of research
on the performance of discrete fibres with soils (Gray and AlRefeai, 1986; Maher and Ho, 1994; Crockford et al., 1993;
Santoni et al., 2001; Consoli et al., 2009a), but these studies

have been generally done independently and they have not


always been consistent (Silva dos Santos et al., 2010). This
paper presents initial results from laboratory tests performed on
completely decomposed soil reinforced with discrete fibres.
2

MATERIALS AND METHODS

Completely decomposed granite was used as the host soil. It


originates from in-situ weathering of the parent igneous rock,
and is one of the most common geo-materials in Hong Kong.
The short discrete fibres used in the tests presented here were
similar to those used by Silva dos Santos et al. (2010).
2.1

Materials tested

The completely decomposed granite (CDG) host soil was


obtained from a construction site near Beacon Hill, Hong Kong.
Completely decomposed residual soils are well-graded in nature
as the tropical climate has weathered the parent rock to a
material comprising gravel and sand grains down to silt and
clay-sized particles. Coarser particles are usually of quartz
origin owing to its high chemical resistance while finer particles
are most likely other primary hydrous minerals, such as

2541

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

kaolinite and feldspar (Yan and Li, 2012). The grain size
distribution (shown in figure 1) reflects that the soil has 16%
particles finer than 63. The specific gravity of the soil was
found to be 2.65. From Standard Proctor compaction tests, the
maximum dry density of the soil was determined as 1.93Mg/m3
with an optimum moisture content of 12.3%. Tests on particles
finer than 425 indicated the plastic and liquid limits to be
25.6% and 35.6% respectively. Using the Unified Soil
Classification System (USCS) the soil can be classified as
clayey sand of low plasticity (SC-CL).
The fibres used are short filaments made of polypropylene
similar to those used by Silva dos Santos et al. (2010). They are
chemically inert and have uniform characteristics, with a
relative density of 0.91, a tensile resistance of 120MPa, an
elastic modulus of 3GPa and a range of linear deformation at
rupture between 80% and 170%. The dimensions of the fibres
used in the tests were 0.023mm in diameter and 24mm long
(Silva dos Santos et al., 2010). After performing a series of
unconfined compression tests on CDG reinforced with a range
of fibre contents (0.3 1%), it was decided to continue the
study with 0.3% of fibre per weight in the triaxial tests.

3
3.1

TEST RESULTS
Unconfined compressive strength

Representative unconfined compression test results on pure


CDG and CDG + 0.5% fibre are presented in figure 2. The plot
clearly shows that the specimens of reinforced CDG yielded at
very high strain, contributing an additional tenfold strength to
the soil. On the other hand unreinforced CDG yielded at very
low strength (131kPa) and low strain.

Figure 2. Unconfined compression of CDG and fibre-reinforced


CDG.

Figure 1. Particle size distribution of CDG.


2.2

axial strain rate of 0.01% per minute to ensure no excess pore


pressure development within the sample (this was checked by
measurement at the opposite end of specimen). The membrane
and area corrections were made as per the recommendations
proposed by La Rochelle et al. (1988). The void ratios are
calculated averaging from that obtained by the initial density of
the sample and the final moisture content, taking account of the
measured volume change in all the stages. In all tests the
difference in specific volume compiled was less than 0.02.

Testing apparatus, methods and sample preparation

3.2

2.2.1 Uniaxial Compression Test


Unconfined compression tests on CDG and CDG+fibre soils
were performed in a uniaxial compressive testing machine. The
particle sizes passing 2mm diameter sieve were used for
preparing specimens in a 38mm diameter; 76mm height mould
at maximum dry density and optimum moisture content. The
compression tests were performed at 0.5mm/min in all cases.
2.2.2 Triaxial Testing
Drained triaxial tests were performed using a conventional
triaxial apparatus with a computer controlled GDS cell and back
pressure controllers. The shearing tests were performed with a
constant effective stress on specimens of both unreinforced and
reinforced (with polypropylene fibres) CDG soil.
The soil was first soaked in water with a deflocculating
agent and left for air drying, then it was sieved to constituent
particle sizes so that the samples could be prepared in exact
proportion as shown in figure 1, discarding particles above
5mm. Loose specimens were prepared, avoiding macro-voids
and taking care of minimising membrane penetration. The
specimens of 76mm diameter and 152mm height were prepared
in a sample preparation mould.
The samples were saturated under back pressure and the
effective confining pressures ranged from 100 to 500kPa.
Saturation was monitored in each test, ensuring Skempton B
values of at least 0.92 throughout the testing programme. The
axial strains were measured outside the cell using a standard
displacement transducer. The triaxial tests were run at a low

2542

Triaxial shearing

Triaxial drained tests were performed on isotropically


consolidated specimens of pure CDG and reinforced CDG
(Table 1). Some specimens were over-consolidated by a ratio of
OCR=5 before being sheared. Details of the tests are shown in
Table 1 (UR and R refer to unreinforced and reinforced
specimens respectively).
Table 1. Summary of the triaxial tests.
Test
v0
vc
UR 100
1.42
1.32
UR 200
1.42
1.37
UR 500
1.48
1.32
R 100
1.58
1.36
R 200
1.43
1.32
R 500
1.58
1.41

pc'(kPa)
112.0
210.4
499.6
98.9
202.9
499.4

OCR
5
1
1
5
1
1

The void ratios determined after consolidation (before shearing)


were found to vary between 0.32 and 0.37 for pure CDG
specimens and 0.32 and 0.42 for CDG-fibre specimens. Only
dense specimens were prepared for the test programme. Looser
specimens were difficult to prepare due to the presence of
macro-voids which caused an initial collapse of the specimen,
resulting in void ratios after consolidation within the same range
as those for the dense specimens.
The stress-strain and volumetric responses during shearing
are shown in figure 3. The stress-strain response (figure 3a)
shows that the reinforced specimens generally have higher

Technical Committee 211 / Comit technique 211

strength and higher initial stiffness at the beginning of shearing,


when compared to their unreinforced counterparts. However the
reinforced specimens that mobilised their full strength only did
so at shear strains in excess of 20%. The unreinforced
specimens on the other hand either reached a constant stress by
20% strain or they showed strain-softening, depending on their
consolidation history. Unlike the other reinforced specimens,
R100 kept gaining strength and never reached critical state even
at large strains (about 50%). This may be due to the overconsolidation history of the specimen, which may have released
some of the tension in the fibres prior to shearing. The peak
strengths of reinforced CDG were calculated to be 1.76
(100kPa), 1.29 (200kPa) and 1.26 (500kPa) times that of the
pure CDG.
Similar behaviour on fibre-reinforced sands is reported by
Consoli et al., 2007, Consoli et. al (2005) and Silva dos Santos
et al. (2010). For example Silva dos Santos et al. (2010) found
that the effect of fibres depends on the effective stress at which
they are sheared, reducing marginally with increasing effective
stress. For sands, it is already reported that at low effective
stress, adding fibres contributes to reducing the degree of
dilation in the reinforced specimens (Silva dos Santos et al.,
2010). The persistent strain hardening behaviour (figure 3a,
R100) was also observed by Consoli et. al (2005) on Botucatu
residual soil, however their data were limited to strains of about
25%. In the present study, the specimens were sheared to strains
up to 50% and it is clear that the strain hardening behaviour of
specimens R200 and R500 stopped beyond s>35% to reach a
critical state. The governing mechanism for the strain hardening
behaviour of R100 specimen might therefore be due either to
the effect of low effective stress or to the effect of overconsolidation, or a combined effect.
The effect of the fibres on the volumetric response of the
reinforced CDG in comparison seems to be that of restricting
the degree of dilation in the specimen sheared at lower effective
stresses, while at higher effective stress this effect is not so
evident (figure 3b). The over-consolidated specimen of
reinforced CDG shows a different volumetric response i.e. it
tends to dilate after 20% shear strain even though it is expected
that reinforcement will impede dilation. This behaviour is again
either due to over-consolidation or to low effective confining
stress. Previous findings on Botucatu residual soils (Consoli et.
al, 2005) and other pure sands may be extrapolated to normally
consolidated CDG, but the effect of over-consolidation is new
and more test results are required to explain it within the critical
state framework.
The stress-dilatancy behaviour of CDG (black symbols) and
reinforced CDG (grey symbols) samples tested at different
effective stress are shown in figure 4. All normally consolidated
specimens, reinforced and unreinforced, show a typical
frictional behaviour. The pure CDG specimens converge to a
unique critical state stress ratio ranging from M=1.57 to
M=1.61. The reinforced CDG specimens tested at effective
stresses of 200 and 500kPa converged to a critical stress ratio of
M = 1.83. For the lower effective stress of 100kPa (R100), the
specimen reached a higher stress ratio of M = 2.14, which is
similar to what was found by Silva dos Santos et al. (2010) on
fibre-reinforced sand. The over-consolidated specimens, UR100
and R100, did not follow the frictional trend but showed much
less volumetric deformation up to critical state, which was also
observed in the stress-strain behaviour. This may have been
caused by locking of the fibres during compression and swelling
prior to shearing.

(a)

(b)
Figure 3. Stress-strain-volumetric response of CDG and fibrereinforced CDG sheared at different effective confining stresses.

Figure 4. Stress-dilatancy response of CDG and fibre-reinforced


CDG.
The deviatoric stress and corresponding mean effective
stress in the test that reached a stable critical state are plotted in
a q-p' plane in figure 5. These points form a critical state
envelope for the pure CDG with a critical state gradient
M=1.57. This is found to be consistent with critical stress ratio
M = 1.57 - 1.61, obtained from the stress-dilatancy plot (figure
4). The end of test points are also plotted for the reinforced
specimens but no attempt has been made in this paper to define
the critical state envelop for fibre-reinforced CDG because at
low stresses, the deviatoric stress does not stabilise (figure 3a).
More tests are required over a larger range of stresses to do so,
as was done by Silva dos Santos et al. (2010) who found that
the critical state lines of the unreinforced and reinforced
specimens converge at large stresses of the order of 5MPa.

2543

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

REFERENCES

Altuhafi F.N. and Coop M.R. (2011). Changes to particle


characteristics associated with the compression of sands.
Gtechnique 61, No. 6, 459471.

Figure 5. Critical states and end of test points for CDG and
fibre-reinforced CDG in q-p' plane.
4

CONCLUDING REMARKS

The results presented indicate that using discrete fibres can be


an effective means of reinforcing CDG, specifically at low
effective stresses. The unconfined compressive strength tests
showed a tenfold increase in strength with 0.5% fibres content
in the soil prepared at maximum dry density and optimum
moisture content. In triaxial drained tests, adding fibres seems
to increase the shear strength by up to two times the strength of
the unreinforced specimens, as well as its initial stiffness.
Dilation was also found to be reduced. Unique critical states
were reached for the unreinforced CDG and reinforced CDG
tested at high effective stress. The stress-dilatancy was found to
be frictional for all normally consolidated specimens, but with
different critical state stress ratios (M) for the fibre-reinforced
specimens depending on their effective confining stress. Initial
results also seem to indicate that the over-consolidation ratio
affects the performance of the reinforced CDG, noticeably in
the stress-dilatancy response, but more work is needed to
confirm it.
5

ACKNOWLEDGEMENTS

The authors wish to acknowledge Ku Hei Man, Gloria (final


year project student 2012, The University of Hong Kong) for
providing the Unconfined compressive test results. The
financial support provided by Hong Kong Research Grant
Council GRF No.70211 is gratefully acknowledged.

2544

BS 1377:1990. Methods of test for soils for civil engineering


purposes. British Standards Institution, London
Consoli N.C., Casagrande M.D.T. and Coop M.R. (2005).
Effect of fiber reinforcement on the isotropic compression
behavior of a sand. J. Geotech. Geoenviron. Engng, ASCE
131, No. 11, 14341436.
Consoli N.C., Heineck K.S., Casagrande M.D.T. and Coop
M.R. (2007). Shear strength behavior of fiber-reinforced sand
considering triaxial tests under distinct stress paths. J.
Geotech Geoenviron. Engng , ASCE 133, No. 11, 14661469.
Consoli N.C., Casagrande M.D.T., Thom A., Dalla Rosa F.
and Fahey M. (2009a). Effect of relative density on plate tests
on fibre-reinforced sand. Gotechnique, 59, No. 5, 471476.
Consoli N.C., Festugato L. and Heineck K.S. (2009b).
Strainhardening behaviour of fibre-reinforced sand in view of
filament geometry. Geosynthetics Int. 16, No. 2, 109115.
Crockford W.W., Grogan W.P. and Chill D.S. (1993). Strength
and life of stabilized pavement layers containing fibrillated
polypropylene. Transpn Res. Rec. 1418, 6066.
Gray D.H. and Al-Refeai T. (1986). Behavior of fabric versus
fiber reinforced sand. J. Geotech. Engng, ASCE 112, No. 8,
804826.
Lee I.K. and Coop M.R. (1995). The intrinsic behaviour of a
decomposed granite soil. Geotechnique 45 (1), 117130.
Maher M.H. and Ho Y.C. (1994). Mechanical properties of
kaolinite/fiber soil composite. J. Geotech. Engng, ASCE 120,
No. 8, 13811393.
Santoni R.L., Tingle J.S. and Webster S.L. (2001). Engineering
properties of sandfiber mixtures for road construction. J.
Geotech. Geoenviron. Engng, ASCE 127, No. 3, 258268.
Silva Dos Santos A.P., Consoli N.C. and Baudet B.A. (2010).
The mechanics of fiber-reinforced sand. Gotechnique, 60,
No. 10, 791799.
Yan W.M. and Li X.S. (2012). Mechanical response of
medium-fine-grained decomposed granite in Hong Kong.
Engineering Geology 129-130 (2012) 18.

Hybrid Application of Deep Mixing Columns Combined with Walls as a Soft Ground
Improvement Method Under Embankments
Application hybride de la mthode de Deep Mixing sur des colonnes combines des murs en
tant que mthode damlioration des sols mous sous remblais
Matsui H., Ishii H., Horikoshi K.

Technology Center, Taisei Corporation, Yokohama, JAPAN

ABSTRACT: In this paper, we introduce theconcepts and general functions of a hybrid application of deep mixing columns combined
with walls. This newmethod for improving the soft ground under embankments helps control ground deformation. We brieflydescribe
a case in which the method was applied under an embankment 7m in height. The method effectively restricted the induced
deformation of the ground surface to a target level, not only under the embankment but also adjacent to the embankment toes. Twodimensional finite element analysis was adopted to the case and foundeffective for simulating the performance. Also proposed is a
design flow for the new methodto efficiently determine the best arrangement of deep mixing columns and walls. Numerical
parametric studies were carried out to compare the new method with conventional methods.
RSUM : Dans cet article, nous prsentons les concepts et les fonctions gnrales de lapplication hybride de mthode de Deep
Mixing sur des colonnes combines des murs. Cette nouvelle mthode damlioration de sols mous sous remblais aide contrler
la dformation du terrain. Nous dcrivons brivement un cas dans lequel la mthode a t applique sous un remblai dune hauteur de
7 m. La mthode a permis de limiter efficacement la dformation induite de la surface du sol un niveau cible, non seulement sous le
remblai, mais aussi dans les zones adjacentes aux pieds de talus. Une analyse par lments finis en deux dimensions a t applique
ce cas et sest avre efficace pour simuler les performances. Une mthode doptimisation est galement propose en vue de
dterminer de manire efficiente la meilleure disposition des colonnes et des murs. Des tudes paramtriques numriques ont t
menes pour comparer la nouvelle mthode avec les mthodes classiques.
KEYWORDS:soft ground improvement method, finite element analysis, deep mixing method
12.0m

2 CONCEPTS AND GENERAL FUNCTIONS OF A


HYBRID APPLICATION OF DEEP MIXING COLUMNS
COMBINED WITH WALLS

1:1.8

Ac1- 2
As2

Dvc

5.2m

Inside piles

Ac2- 3

Walls

Dvs

9.8m

39.2m

Ac2- 2

36.2m

1.9m

37.2m

Section view

6.7m 4.5m

7.0m

Deep mixing methods have been widely used in Japan for the
foundation systems of embankments constructed on soft clayey
ground, and various low improvement ratio arrangements have
been proposed (Miki and Nozu 2004, Ishikura et al. 2009, Miki
et al. 2011). Typical of recent applications is to achieve limited
soil improvement around 10-20%through an arrangement
of soil improvement columns. This reduces the volume of soil
that must be improved and limits the ground settlement under
the embankments. Moreover, embankment construction in
urban areas requires strict control of ground deformation,
especially in the areas adjacent to the embankment toes.
The authors propose a new hybrid application of deep
mixing columns combined with walls (Tsutsumi et al. 2009) as
a method of improving the soft ground under embankments to
control ground deformation. In this paper, the concepts and
general functions of the method are introduced. The paper then
describes a case in which the method was applied under a tall
embankment 7 m in height. Two-dimensional finite element
analysis was adopted to simulate the performance. Also
proposed is a design flow for the new method that efficiently
determines the best arrangement of deep mixing columns and
walls. Finally, numerical parametric studies were carried out to
compare the new method with conventional methods.

10.0m

INTRODUCTION

21.2m

Outside piles
16.48m

2.6m 1.76m
3.0m 3.0m

Plain view

2.5m

2.5m

1.76m 2.6m
3.0m 3.0m

2.81@8=22.48m
37.2m

The basic concept of this method is to place deep mixing walls


in the ground directly under the embankment slopes, which

Figure 1.Geological profile and arrangement of the deep mixing


columns and walls at theconstruction site where the method was applied.

2545

Heightof
embankment (m)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Out l i ne of embankment

Settlementof
embankment(mm)

I nsi de pi l es

Out si de pi l es
Wal l s

8
6
4

500

1500

3.2

Notreatment

2546

1000

Settlementafterconstruction
ofembankment(mm)

100
Walls
200
300

Improvement

400

Notreatment

500
0

10
20
30
40
50
Distancefromcenterlineofembankment(m)

60

Figure 4.Distribution of ground surface settlement after construction of


the embankment.

3000
300

Table 1.Material properties used for numerical model

800

Limitvalue 20mm

10m

Embankment

7000

Figure 3 shows the settlement history of the ground surface at


the center of the embankment. The same figure also shows a
similar settlement history, observed at a trial embankment
nearby with no subsoil treatment. In the improvement case,
200mm of settlement occurred one year after embankment
construction. Subsequently, settlement converged in both cases.

400
600
Elapsed days(day)

100

18600
6000
12600
4@2810 2600 3000
1:1 1760
.8

Steelwire
Hsteel
Geotextile
Ac12
Outsidepiles

21200

Work outline

Result of construction

200

Figure 3.History of ground surface settlement in at the center of the


embankment.

TRIAL EMBANKMENT

The effectiveness of this method was demonstrated in a road


construction project along the Ariake Sea in Kumamoto
Prefecture. The soft clay at the construction site was about 40 m
thick, so a large volume of settlement could be expected after
constructingan embankment 7m in height. Some parts of the
proposed road were close to residential buildings. Therefore, a
limit value for deformation was set not only for the
embankment but also for the area adjacent to the embankment,
as described below.
Embankment: Settlement sincethe start of service is equal to
or lower than 300 mm.
Adjacent area: Lateral and horizontal displacement since the
start of construction is equal to or lower than 20 mm.
During the design stage, many of the arrangements were
compared using two- and three-dimensional effective stress
analysis. After considering all of the above, the arrangement
shown in Fig. 1 was determinedto be optimal. Each column had
a design strength of 1.0MN/m2, and the arrangement hadan
improvement ratio of 18.5%.
Before the embankment was constructed, settlement plates
and pressure gauges were installedfor the purpose of taking
measurements. The ground surface after soil improvement is
shown in Fig. 2.

Startingdateof
service

2000

GL1.90m
GL4.50m

As2
GL11.20m

Insidepiles

39200

3.1

Improvement

1000

bear the embankment loads as well as the lateral movement of


the soft ground. Deep mixing piles are placed inside and outside
the walls to restrict vertical and horizontal deformation caused
by the embankment.
Figure 1 shows an example of the arrangement of deep
mixing columns and walls at a site. The function and placement
of each pile and wall are explained below.
Inside piles: Columns placed in the ground directly under
the crown of the embankment. This part transfers the load from
the center part of the embankment to the deep layer.
Walls: Walls are placed in the ground under the edges of the
embankment crown. This part bears a large part of the
embankment load and prevents the soil from moving.
Outside piles: Columns placed in the ground directly under
the embankment slopes. This part transfers the load of the
embankment slopes.
This method is designed to economically satisfy the limit
value of settlement by optimizing and minimizing these parts in
the design.

Starting dateof
service

Improvement

0
0

Figure 2.Ground surface after soil improvement.

Notreatment

Ac22
GL21.20m

Walls

Ac23
GL31.00m
Dvc
GL36.20m
Dvs

GL41.60m

Figure 5.Section view of the numerical model.

Figure 4 shows the settlement history fora one-year period


after the construction road was removed.In the improvement
case, large the walls prevented deformation under the
embankment, keeping the settlement around the embankment
below the limit value. The vertical strain measured in the walls
is shown in Fig. 7; this, too, was kept below the fracture strain
value.
3.3

Back-analysis

To investigate the applicability of two-dimensional effective


stress analysisunder actual construction conditions, the
geological profile and mechanical properties of the deep mixing
columns were analyzed using Plaxis 2D Ver.9.02.
The numerical model is shown in Fig. 5. Due to the
symmetry of the embankment, only half of the geometry was
considered for the model. The distance from the embankment

Technical Committee 211 / Comit technique 211

=H/quck

(1)

Settlementof
Embankment (mm)

6
4
2
0
0

50
100

Measurement

150
200
250

Analysis

300
100

200

300
400
500
Elapseddays(day)

600

700

Figure 6.Settlement history of ground surface in center of the


embankment
100

Embankment

Limit value
20mm

10m

100
Walls

Measurement
Analysis

200
0

10
20
30
40
50
Distancefromcenterlineofembankment(m)

60

Figure 7.Distribution of ground surface settlement after construction of


the embankment
Horizontaldisplacementofwalls after
constructionofembankment(mm)
40 20
0
20
40
0
(a)

4 DETERMINING THE OPTIMUM ARRANGEMENT OF


DEEP MIXING COLUMNS

Verticalstrainofwalls after
constructionofembankment()
500
0
500 1000 1500
(b)

10
Depth(m)

In this method, the piles and walls are effectively arranged


according to the limit values of deformation in the embankment
and the adjacent area. Due to the countless combinations of
planar arrangements and improvement depths, arbitrary
parametric studies require considerable time to identify
optimum arrangement. Therefore,the following3-step method is
proposed fordetermining the optimum arrangement.
1) Determine the planar arrangement: First, walls are placed
in the ground under the edges of the embankment crown. Next,
inside and outside piles are arrangedequidistantly by an amount
not less than the necessary improvement ratio , defined as

Settlement afterconstruction
ofembankment(mm)

toe to the lateral boundary is 80m. As a boundary condition of


deformation, the bottom surface was fixed. The side surface was
free verticallyand fixed horizontally.As a drainage condition,
excess pore water pressures at the ground surface and bottom
surface were set to zero.
The soil layer is modeled as an elasto-plastic material using
the Sekiguchi-Ohta model (Sekiguchi and Ohta 1977). The sand
layers and deep mixing columns are modeled as a linear elastic
material. The embankment is modeled as an elasto-plastic
material using the Mohr-Coulomb model. Table 1 lists the
model parameters used for the analysis.
The history of the embankment construction was modeled
bybuilding up the elements. In converting from actual threedimensional ground to the two-dimensional numerical model,
the deformation modulus of the deep mixing columns was
reduced according tothe improvement ratio and the coefficient
on permeability for deep mixing columns was set to thevalue for
each layer of ground.
The following figures are for the sake of comparison and
analysis: Figure 6 shows the history of ground-surface
settlement at the center of the embankment; Fig.7 shows the
distribution of ground-surface settlement after construction of
the embankment; Fig. 8 shows the horizontal displacement and
vertical strain of the walls. The settlement history and
displacement of the ground surface and walls are quantitatively
evaluated using two-dimensional analysis. However, a clear
difference in the vertical strain exists at greater depths. In the
numerical models, the deformation modulus of wallsless than 21
m in height is lower than that of walls greater than 21m in
heightas perthe arrangement of the deep mixing columns. This
is thought to be the cause of the difference in vertical strain.
Individual material propertiesare effective for evaluating the
strain distribution of walls.

Heightof
Embankment (m)

Effective
Coefficient
Unit Effective
Critical
Deformation Initial Consolidation
Compression Expansion
angle of
Poissons
of
weight cohesion
void
yield stress
state
modulus
friction
permeability
index
index
ratio
t
c'
ratio
pc
parameter
E

'

k
e0
M
(kN/m2)
(kN/m3) (kN/m2)
(kN/m2)
(deg.)
(cm/sec)
Embankment 19.0
10.0
35.0
28,000

0.25
1.0010-3
Ac1-2
14.6
10.0
36.4
1,720 2.13
36.8
0.289
0.029
1.48
0.35
1.3010-6
As2
18.7

28,000

0.25
1.0010-3
Ac2-2
14.3
10.0
36.2
6,380 2.53
146.1
0.665
0.067
1.47
0.35
3.0010-7
Ac2-3
15.1
10.0
33.0
7,130 2.00
178.5
0.408
0.041
1.33
0.35
2.3010-7
Dvc
15.8
10.0
33.0
6,510 1.21
215.7
0.149
0.015
1.33
0.35
1.4010-9
Dvs
19.0

70,000

0.35
1.0010-3
1.4010-9
367,000

0.20
Columns
19.0

1.0010-3*2
718,000*1
*1 The deformation modulus of thedeep mixing columns was derived from quality verification tests, which reduced dependence on the improvement
ratio.
*2 The coefficients of permeability of the deep mixing columns are same as those for each layer.

Measurement
20

Analysis

Analysis

Measurement
30

40

Figure 8.(a) Horizontal displacement of walls (b) Vertical strain


in walls

2547

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Optimumarrangement

Floating

1000

Result forSaga(Example)
ResultforKumamoto
:Optimumarrangement

200

150
Walls reached
bearinglayer

100

(a)

Settlementofembankment(mm)

250

Allcolumnsreached
bearinglayer

600
Newtechnique
400
200

0
100
200
300
400
500
Improvementvolumeperonemeterinalongitudinaldirection(m3/m)

100
0

Limitvalue

Settlementofapoint10maway
fromembankmenttoes(mm)

Confinedrange

50
100
150
200
250
300
350
Improvementvolumeperonemeterinalongitudinaldirection(m3/m)

Figure 9.Example of confining the range of consideration and the result


of the consideration for in-situ construction in Kumamoto

in which is the unit weight of the embankment, H is the height


of the embankment and quck is the design strength of the deep
mixing columns.
2) Confine the range of consideration: For the planar
arrangement noted above, the deformation of three
arrangements with different improvement depths(as shown in
Fig.9) is calculated. The relation between the improvement
volume and the deformation of the three arrangements is
illustrated in Fig. 9. The range of consideration is narrowed by
comparing with the limit value of deformation in the adjacent
area.
3) Identify the optimum arrangement: The optimum
arrangement in the range noted above is the arrangement with
the lowest improvement volume that satisfies the limit value.
Figure 9 shows the results of a search for the optimum
arrangement in areas along the Ariake Sea in Saga Prefecture.
Figure 9 also showsthe results of a search in Kumamoto as an
example of an arbitrary parametric study. The positional relation
between both cases is fitted and the results indicate the
effectiveness of the search method.
COMPARISON WITH CONVENTIONAL METHODS

To confirm the effect of displacement suppression, a hybrid


arrangement is compared with conventional columns
arrangements as well asan arrangement in which the columns
are equidistant and narrowly spaced.
Under the same geological conditions and embankment
height as in the Kumamoto case, the settlement of the
embankment and at a point 10 m from the embankment toes of
each arrangement were calculated using two-dimensional
analysis.
Figure 10 shows the relation between individual settlement
values and improvement volumes per meter in the longitudinal
direction. Regarding settlement of the embankment, the
settlement of the hybrid arrangement and the equidistant
arrangement are lower than the arrangement under the slopes,
confirming the effect of displacement suppression. For the
settlement ata point 10m from the embankment toes, the hybrid
arrangement is the lowest among same improvement volumes.
When the limit value of settlement in the adjacent area is 20mm,
the hybrid arrangement is more effective than conventional
methods in reducing the improvement volume.
6

Equallyarrenged
inlowratio

50

Arrangement
underslopes

800

CONCLUSIONS

On-site measurements confirmed the methods effectiveness in


suppressing displacement. The validity of deformational
estimation using two-dimensional effective stress analysis also

(b)

80

Equallyarrenged
inlowratio

Arrangement
underslopes

60
40
20
Newtechnique

Limitvalue

0
0
100
200
300
400
500
Improvementvolumeperonemeterinalongitudinaldirection(m3/m)

Figure 10.(a) Settlement of embankment (b) Settlement at a point 10


meters from the embankment toes.

was confirmed. However, little difference was seen in the


estimation of stress and strain distribution in the walls.Using
individual material properties for the walls, however, is
effective.The two examples of searching for the optimum
arrangement using the method proposed in this paper confirmed
the methods effectiveness. Analytical comparison ofthe new
method with conventional methods also confirmed the
economic efficiency of the new method.
7

ACKNOWLEDGEMENTS

The development of this column link method is the result of


collaborative research involvingthe Public Works Research
Institute, Japan, and thirteen private corporations in Japan. The
authors are particularly gratefulfor the kind assistance of
ShouichiTsutsumi (PWRI), Hirotaka Kawasaki (Shimizu Corp.),
ShouichiTsukuni (Takenaka Civil Eng. & Const. Co., Ltd.), and
NaotoshiShinkawa (Fudo Corp.).
8

REFERENCES

Miki, H. and Nozu, M. 2004. Design and numerical analysis of road


embankment with low improvement ratio Deep Mixing
method,Geotechnical Engineering for Transportation Projects, Vol.
2,1935-1402.
Ishikura, R. Ochiai, H. and Matsui, H. 2009. Estimation of settlement of
in-situ improved ground using shallow stabilization and floatingtype columns, Proceedings of 17th International Conference on Soil
Mechanics and Geotechnical Engineering, 2394-2398.
Miki, H. Okochi, Y. and Makino, M. 2011. Evaluation of constraint
effect of DMM with varied shape and arrangement of stabilized
bodies using centrifuge model test, Proceedings of Indian
Geotechnical Conference, 501-504.
Tsutsumi, S. Sawamatsu, T. Iso, Y. and Oshita, T. 2009. Centrifuge
model experiment of new improvement type in deep mixing method
with steel tied by cable for lateral flow, Deep mixing 2009 Okinawa
symposium.
Sekiguchi, H. and Ohta, H. 1977. Induced anisotropy and time
dependency in clays, Constitutive equations of soils, Proceedings of,
9th International Conference on Soil Mechanics and Foundation
Engineering, 229-238.

2548

Application of cement deep mixing method for underpinning


Application de colonnes de sol-ciment pour travaux de reprise en sous uvre
Melentijevic S., Arcos J.L.

Grupo Rodio-Kronsa, Madrid, Spain

Oteo C.

Universidad de A Corua, A Corua, Spain

aBstract: this paper presents a case history of the application of wet deep soil mixing columns for underpinning of the existing
floor slab of an industrial building, which settled due to different encountered post-constructive pathologies related to ground
conditions. the soil-cement columns were constructed with the application of the new developed springsol tool that permits the
underpinning of existing foundations, infrastructure transport platforms and embankments, as well as working in limited spaces and
under low headroom conditions. the quality control regarding laboratory testing of core and wet grab samples is reported. design
procedure and the finite element analysis that verify settlement calculations are described. the fem is focused on the axisymmetric
numerical modeling in plaxis.
rsUm : cet article prsente une tude de cas de ralisation de colonnes de soil mixing par voie humide pour la reprise en sous
uvre du dallage dun btiment industriel, ayant tass aprs sa construction cause de pathologies du sol. les colonnes de sol-ciment
ont t ralises avec la mthode springsol, qui permet la reprise en sous oeuvre de fondations existantes, dinfrastructures de
transports et de remblais, partir demprises troites et sous faible gabarit. les contrles le qualit raliss en laboratoire sur des
prouvettes carottes et sur des prlvements frais y sont prsents. le mode de dimensionnement ainsi que les analyses par lments
finis pour estimer les tassements sont galement dcrits. les calculs ef ont t raliss avec le code plaxis en axi-symtriei.
KeYWords: deep mixing, soil-cement columns, springsol, underpinning, fem.
1

introdUction

in order to reduce settlements, increase bearing capacity of


natural ground and improve the overall stability, different
ground improvement techniques can be put into practice, but not
all of them can be applied for underpinning projects. the
limitations for the applications are mainly related to capacity of
the machinery to pass existing foundation structures as
reinforced slabs or footings, and insufficient working spaces
and/or low headroom conditions.
the soil-cement deep mixed columns for ground
improvement of soft soils have an extensive application for
different geotechnical projects due to their higher strength and
lower compressibility than the untreated natural soft soil. the
application of traditional deep mixing methods, both wet and
dry, was very restricted for the underpinning of existing
foundations, improvement of existing embankments and
infrastructure platforms, due to the form and dimensions of the
mixing tool.
With objective to present new wet deep soil mixing system
called springsol a case history with its application in
underpinning project is reported in this paper. to prevent
further settlements and guarantee bearing capacity of the
foundation of the industrial building that presented various postconstructive pathologies, the springsol deep mixing columns
were proposed as an alternative method to basic project
underpinning solution comprising jet-grouting, traditional tubea-manchette grouting and micropiles for different areas of the
building. due to its technical, economic and environmental
advantages, soil-cement columns were accepted and executed as
a global solution. in the following chapters the main
characteristics of the springsol system will be described as
well as the analysis of the solution adopted and performed for
this project. some recent applications of the springsol
technique are given in melentijevic et al., 2012.

sprinGsol soil cement colUmns

springsol device was originally developed for improvement of


soils under existing railways due to its spreadable form
(innotrack 2009, le couby 2010). the folded tool is introduced
through the casing to the required depth at the beginning of the
column head. once it reaches the end of the casing and
penetrates the underlying soft soil, the blades spread out
forming the soil-cement column down to the required depth.

figure 1. the springsol spreadable tool: (a) original and (b) modified.

at present, springsol columns permit an application in


ground improvement for underpinning of existing foundations
(both slabs and footings), paving, embankments and subbase
below infrastructures (both highways and railways). originally
it was developed to form columns of 400 mm diameter. due to
continuous necessity for construction of soil-cement columns of
larger diameters the springsol soil mixing tool has technically
evolved into the new modified version, permitting achievement
of different column diameters ranging from 400 to 700 mm. the
modified tool also includes the automatic system for opening
and closing blades thus having the possibility to form variable

2549

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

diameter along the column depth. figure 1 shows the nowadays


available original and modified improved springsol tool. the
folded tool is of a diameter of 150 and 165 mm for the original
and modified version respectively.
some of the advantages of the method are:
no pollution of the subgrade layer with the cement
slurry, due to insertion of the casing that enables the
recovery of spoil.
the spoil collection with the system installed at the
base of the mast of the drilling rig, connected to the
peristaltic pump drawing the spoil directly to the
container.
the high production rate.
Working under difficult execution conditions and
limitations, i.e. under low headroom conditions and
within reduced spaces.
execution with small batching plants and small drilling
rigs in reduced limited spaces, etc.
the quality of soil-cement columns regarding their
homogeneity and strength is influenced by two parameters:
im (rev/m) - blade rotation number determining the
mixing efficiency defined as a total number of mixing
blades passing along one meter of tool penetration, and
ii (kg/m3) - cement quantity introduced per m3 of the
treated soil.
table 1. springsol columns performance and geo-mechnaical
parameters.
parameter
diameter (mm)
Water / cement ratio
penetration velocity (cm/min)

0.6-1.2
15-50
min 350

ii (kg/m3)

150-350

Ucs (mpa)
e50
shear strength
Bending strength

Grouting gap slab-fill


Slab

see Figure 3

Man made fill

Springsol columns 400mm


Length = 5.508.00m
Grid=1.50x1.50m 2.00x2.00m
Natural soil-clays with
gravel and boulders

figure 2. cross section of the ground treatment solution.

400-700

im (rev/m)

same material for construction of a fill without its appropriate


compaction.
the affected area included more than 8000 m2 with the
installation of more than 2500 soil-cement columns. the length
of soil-cement columns ranged from 5.50 m to 8.00 m in
function of the thickness of the man-made fill, with the total
length of columns of more than 15000 meters. due to the form
of the springsol tool, the columns were embedded
approximately 20 cm in the natural ground. the columns of a
400 mm diameter, performed with the originally developed tool,
were distributed in a square grid pattern ranging from 1.50 m to
2.00 m in function of the surcharge to be transmitted from the
slab. the performed solution is schematically presented in
figure 2.

0.5-6.0
(50-500) Ucs
20-40% Ucs
8-15% Ucs

the general execution parameters and geo-mechanical


characteristics (unconfined compressive strength - Ucs,
stiffness modulus e50, shear and bending strength) of the soilcement columns executed by the springsol device are given in
table 1. these data are established on experiences gained on
different projects and field tests carried out recently in spain
(melentijevic et al 2012, melentijevic et al 2013). these
findings on geo-material properties are in agreement with
worldwide published information on deep mixed columns
(Bruce 2001, cdit 2002, etc.).

the post pathology site investigation consisted of 46


dynamic penetration tests and 5 drilled boreholes with standard
penetration tests, executed from the working platform, i.e. the
existing floor slab level. the natural ground, detected at the
depth of 5.50 to 8.00 m from the surface, consisted of clays of
high consistency with gravels and boulders, with the n20>40
(dpsh). the overlaying treated loose man made fill was
formed of clays with gravels (n20<10) proceeding from the
natural ground after a massive excavation for the foundation of
the main structure elements.
the soil treatment solution included following steps:
coring of the existing slab (diameter = 62 mm) for
grouting of the gap between slabs and fill.
contact grouting between the slab and the fill in order
to fill gaps due to settlement of badly compacted man
made fill.
coring of the existing slab and contact grouted gap
(diameter = 182 mm) for the passage of the spreadable
tool.
execution of springsol columns (diameter = 400 mm).
filing the gaps of coring the existing slab.
Visual description of the executed steps previously
mentioned is shown in figure 3.
Coring 182 mm

Coring 62 mm

Grouting gap slab-fill

3 proJect details and adopted GroUnd


improVement solUtion
in this chapter an example of application of the springsol
technique for underpinning is presented. the industrial building
in the central spain presented different post-constructive
pathologies regarding differential settlements of floor slabs and
pavements as a consequence of poorly compacted anthropic fill
material. the main structure (walls and columns) were founded
on a natural ground, and due to its adequate geotechnical
characteristics did not present any pathology. the shallow
foundation on a natural ground was performed after a massive
excavation of superficial layers of natural soil, applying the

2550

Springsol columns 400 mm

figure 3. Visual control of the excavated treatment area.

Slab

Technical Committee 211 / Comit technique 211

4 Geomaterial soil-cement colUmn


characteriZation
the cement used for the construction of soil-cement columns
was of the portland type cem i 52.5 sr. the slurry mix was of
a cement / Water type with the relation 1/1. the average
penetration rate for the construction of columns was 30 cm/min
with the rotation velocity of 50 to 55 rpm and the average
cement consumption of 350 kg/m3.
the unconfined compressive strength tests (Ucs) were
performed both on drilled core samples and wet grab samples
(cylinder dimensions height / diameter > 2), both of them
usually being the main mean of the quality control of wet deep
mixing methods. three core samples were taken from different
soil-cement trial test columns, 21 days after the completion of
the soil-cement columns. the samples were cored at a distance
of 110 mm to 120 mm from the centre of columns. the overall
average total core recovery was more than 97% for all soilcement columns. Wet grab samples were taken in the half an
hour after execution of columns and were tested at same age as
core samples. the Ucs tests were also used to determine the
stiffness modulus e50 (secant value of Youngs modulus of
elasticity determined at 50% of Ucs).
the Ucs values of wet grab samples after 7 days varied
from 1.4 to 3.9 mpa, while Ucs values for drilled core samples
on 28 days ranged from 2.2 to 4.4 mpa and axial failure strain
values varied from 1 to 1.2 %. stiffness modulus values
determined from Ucs tests varied from 270 to 330 mpa, with
the average relationship between e50 and Ucs resulting in
approximately 100.
some of the drilled core samples extracted from soil-cement
trial test columns is presented in figure 4. it can be observed
uniformly treated springsol columns.

by the elastic law. the load transfer layer formed by grouting


the gap within the contact gravel layer below the existing slab
of the approximate thickness of 20 cm is modelled by the mohrcoulomb law. Geotechnical parameters of each material (ltlload transfer layer, cU-upper clay layer, cm-medium clay
layer, nsc-natural soil clay layer, sc-soil cement column) used
in the numerical analysis are given in table 2.
the cross section of the fe model is presented in figure 6
showing the geometry and soil layers used in analysis, as well
as the finite element mesh.
table 2. material parameters.
parameter

LTL

CU

CM

NSC

thickness
(m)

0.2

4.06.5

1.5

>4.5

density
(kn/m2)

22

16

17

18

20

500

10

20

500

40

18

20

22

35

300

2.5

5.0

50

300

0.2

0.3

0.3

0.3

0.2

cohesion
(kpa)
friction
angle ()
Youngs
modulus (mpa)
poisson ratio

5.2

SC

Results

figure 6 present the employed mesh in the fem model, and the
results regarding vertical displacements and axial stresses for
the case of the grid spacing of 1.50 m corresponding to the
surcharge of 20 kn/m2.
Req

LTL
5-8
mm

CU
figure 4. drilled core samples of soil-cement columns.

5
5.1

SC

nUmerical model

2-5
mm

max 385.6 kN/m 2

NSC

0-2
mm

4.5m

(1)

where: req is the radius of the unit cell and s is the grid
spacing.
in this project different square grid patterns (grid spacing
varying from 1.50 to 2.00 m) depending on the surcharge of the
slab (ranging from 10.00 to 20.00 kn/m2) are taken into
account.
in this study the commercial finite element code used for 2d
modelling is plaxis (version 8.6). Both the soft soil and the soilcement column behaviour are modelled by the elastic-plastic
mohr-coulomb failure criterion, while the slab is characterized

50-150
kN/m 2

1.5m

When using finite element analysis to model deep mixed


columns installed in a periodic pattern, the problem is usually
modelled in a 2d axisymmetric model, referred as a unit cell
model. the homogenization equivalent model is usually not
used due to lack of access to column stresses.
the radius of the unit cell depends on the grid spacing:

Req s

4.0m
6.5m max 7.85 mm

CM

General data

0-50
kN/m 2

150-200
kN/m 2

figure 6. numerical modeling results for reinforced soil. (a) Geometry


of the unit cell mesh and model dimensions. (b) Vertical
displacements. (c) axial stresses.

the homogenized settlements as well as negligible


differential settlements due to high rigidity of the load transfer
layer, formed by grouting the layer of gravel below the slab, and
soil-cement columns can be observed in figure 6-b.

2551

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

the maximum allowable axial stresses in soil-cement


columns defined by the Ucs value and verified by the fem
analysis was not exceeded for different cases of grid spacing,
surcharge and column lengths taken into consideration in the
project.
the comparative study of the maximum vertical
displacement, in cases without and with the soil improvement
for different grid meshes is shown in figure 7. the significant
settlement reduction with the applied soil improvement can be
observed in function of grid spacing and for different
thicknesses of the soft soil layer.
50
45

Settlements (cm)

40
35
30

25
20
15

10

5
0
1,5

1,6

1,7

1,8

Grid spacing (m)

1,9

With improvement SC col length 5.5m

With improvement SC col length 8m

Without improvement thickness 5.5m

Without improvement thickness 8m

figure 7. maximum vertical displacements.

the numerical calculations were also analyzed in terms of


load and settlement efficiency (asiri 2012) in order to
determine the effectiveness of the soil improvement method.
the load efficiency (el) is defined as a ratio of a transmitted
load to the head of a soil-cement column and the total load
acting on the unit cell. the settlement efficiency (eset)
represents the reduction of a settlement by a soil-cement column
compared to the settlement of the unit grid without ground
improvement. they are represented by the following equations:

EL

QP

W Q

E
set 1

SM

(2)

where: Qp is the load acting on the head of soil-cement


column, W is the dead load of the load transfer platform and Q
is the force of the surcharge applied to the slab. sm is the
settlement of the soil reinforced by soil-cement columns
measured at the surface of the load transfer platform and s0 is
the settlement of the natural soil without ground improvement.
80
75

Efficiency (%)

70

65
60
55
50

45

40
35
1,5

1,6

1,7

1,8

1,9

Grid spacing (m)


2

Load efficiency - SC col length 5.5m

Settlement efficiency - SC col length 5.5m 4

Load efficiency - SC col length 8m


Settlement efficiency - SC col length 8m

figure 8. load and settlement efficiency.

the results of the load and settlement efficiency are given in


figure 8 in terms of different grids adopted in the project for
different surcharge loads and for different column length in
function of the thickness of the soft layer. Both terms of

2552

conclUsions

the general application of the wet deep mixing method by the


springsol system for ground improvement of existing paving,
embankments and subbase below railways and roads, and
underpinning of existing structures is presented. the case
history of underpinning of existing industrial building, which
settled due to poorly compacted antrophic fill, with the use of
the springsol system is reported.
the design procedure and the estimation of settlements by
the finite element method (plaxis commercial code) based on
axisymmetric model is in concordance with the monitored
settlements. the general homogenized nature of settlements, as
well as insignificant differential settlements are achieved by the
good interaction of the performed soil-cement column and load
transfer layer formed by grouting of the layer of gravel below
the slab. the evaluation of efficiency of the soil treatment by
soil-cement columns, in terms of load and stress efficiency, is
determined confirming its effectiveness.
7

acKnoWledGements

the authors wish to thank to the personnel of Grupo rodioKronsa for their technical assistance, especially to Juan ignacio
lpez, Juan manuel dimas, francisco martn and esteban
casado. also, the collaboration and provision of all necessary
information by the proprietary of the industrial building is
highly appreciated.
8

(3)

S0

efficiency have the same tendency. the load efficiency ranges


from 36 to 50 %, while the settlement efficiency varies between
64 and 78 %. the difference of approximately 25 to 30 %
between load and settlement efficacy relationships is observed.
it is important to emphasize that the estimated settlements
obtained by the analysis by fem were in accordance with the
observed settlements after ground improvement by the
performance of soil-cement columns and re-loading of the slab
of the industrial building.

references

asiri national project 2012. recommendations for the design,


construction and control of rigid inclusion ground improvements.
Bruce d.a. 2001. an introduction to deep mixing methods as used in
geotechnical applications, Volume iii: the verification and
properties of treated ground. U.s. department of transportation,
federal highway administration, report fhWa rd-99-167.
cdit (coastal development institute of technology), Japan. 2002. the
deep mixing method, a.a. Balkema.
innotrack. project n tip5-ct-2006-031415. 2009 subgrade
reinforceemnt with columns. part 1 Vertical columns, part 2
inclined columns.
le Kouby a., Bourgeois e. & rocher-lacoste f. 2010. subgrade
improvement method for existing railway lines an
experimental and numerical study. EJGE Vol. 15: 461-494
melentijevic s., martin f. & prieto l. 2013. execution of springsol
deep mixed columns: field trials. International Conference
Installation Effects in Geotechnical Engineering. Rotterdam. The
Netherlands. 24-27 March 2013: accepted for publishing.
melentijevic s., prieto l. & arcos J.l. 2012. aplicaciones de columnas
suelo-cemento tipo springsol. 9 Simposio Nacional de Ingenieria
Geotcnica. Cimentaciones y Excavaciones Profundas. Proc. Symp.
Sevilla. 17-19 October 2012: 255-268.
plaxis BV. 2008. plaxis 2d manual - version 8. www.plaxis.nl.

Lime Remediation of Reactivated Landslides


Traitement la chaux pour la stabilisation des glissements ractivs
Mesri G.

University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, U.S.A.

Moridzadeh M.

Montgomery Watson Harza, Chicago, Illinois, U.S.A


ABSTRACT: Lime improvement of frictional resistance was examined using samples of Brenna Clay Formation from North Dakota.
The montmorillonitic stiff clay samples had a natural water content, plastic limit, liquid limit, clay size fraction, fully softened friction
angle, and residual friction angle, respectively, in the range of 42 to 85%, 20 to 40%, 62 to 154%, 60 to 95%, 14 to 24, and 7 to 9.
Immediately after introduction of hydrated lime, pH increased to a range of 12.2 to 12.7; within hours, however, pH began to
decrease. Whereas there was a large increase in plastic limit, the liquid limit response to lime treatment was dependent on the
effective confining pressure. Lime treatment increased fully softened friction angle by 5 to 10 at effective normal stress of 100 kPa
and by 3 to 5 at 300 kPa. Lime treatment increased the residual friction angle by 3 to 6 at both 100 kPa and 300 kPa.
RSUM: Lamlioration par addition de chaux de la rsistance en frottement est examine sur des chantillons de la formation
dargile de Brenna dans le Dakota du Nord. Les chantillons dargile raide montmorillonitique ont une teneur en eau, une limite
plastique, une limite liquide, une fraction de dimension argileuse, un angle de frottement aprs remaniement et un angle de frottement
rsiduel respectivement de lordre de 42 85%, de 20 40%, de 62 154%, de 60 95%, de 14 24, et de 7 9. Immdiatement
aprs laddition de chaux hydrate, le pH augmente des valeurs de 12,2 12,7 mais commence ensuite dcrotre dans les heures
qui suivent. Laugmentation de la limite de plasticit suite au traitement la chaux est importante, laugmentation de la limite de
liquidit dpend cependant de la pression de confinement. Le traitement la chaux augmente langle de frottement aprs
remaniement de 5 10 sous une contrainte effective normale de 100 kPa et de 3 5 sous 300 kPa. Le traitement la chaux
augmente langle de frottement rsiduel de 3 6% autant sous une pression de 300 kPa plutt que de 100 kPa.
KEYWORDS: Brenna clay, frictional resistance, lime treatment, landslides.
1

INTRODUCTION

The effectiveness of lime treatment of soils has been commonly


evaluated in terms of improved workability and increased
undrained unconfined stiffness and compressive strength, in
connection to road and airfield construction (Bell 1996). Soil
improvement is expected to result from the flocculation of clay
minerals and cementing action of lime-soil chemical reactions.
On the other hand if the objective of lime treatment is to
improve long-term stability of first-time or reactivated
landslides in stiff clays and shales, permanent changes in the
size and shape of clay particles must be realized to increase
drained frictional resistance. Lime-soil interactions that may
produce less platey and larger soil particles begin and continue
with time under the highly alkaline pH environment. For
Brenna clay samples treated with lime, measurements of pH as
an indicator of chemical environment, Atterberg plastic limit
and liquid limit as indirect measures of changes in particle size
and shape, and fully softened friction angle and residual friction
angle, were used to examine possible mechanisms of lime-soil
interactions. The main variables, in addition to soil mineralogy,
are soil water content, lime content, and duration of lime-soil
interactions.
2

LIME-SOIL INTERACTION

When dry hydrated lime is thoroughly mixed with a wet soil,


lime is consumed, in the absence of carbonation, through two
mechanisms: (a) part of the lime particles is adsorbed on soil
particles during the mixing process, and (b) part of the
remaining lime is dissolved in the soil porewater. The solubility
of calcium hydroxide in water is rather small (0.75 g/).
Therefore, the maximum lime content as percent of dry weight

of soil that can dissolve in the porewater during the mixing


process is quite small and a function of soil water content (only
1.5% of lime for 5% lime content at soil water content of
100%). Dissociation of hydrated lime to (OH)- and Ca2+ leads
to a rise in the pH. If enough lime is left, after satisfying the
adsorption, soil porewater becomes saturated and pH increases
to approximately 12.3 to 12.4. Under the strong alkaline
condition, soil mineral particle surfaces become unstable and
begin to dissolve in the porewater. Simultaneously, under the
elevated pH condition, adsorbed lime particles begin to attack
the soil particle surfaces at the points of contact.
Dissolved silica and alumina react with the dissociated
calcium hydroxide and form new compounds. As the dissolved
hydrated lime is used up in the chemical reactions with silica
and alumina, the remaining free lime, if any, dissolves in the
porewater and pH is maintained at 12.3-12.4. The dissolution of
soil particles and local attack of adsorbed lime on the particle
surfaces continue at the initial rate until all free lime is
completely consumed. Thereafter, pH begins to decrease as the
dissociated calcium hydroxide is used up in the chemical
reactions with dissolved silica and alumina. This has been
confirmed by pH measurements and chemical analyses
conducted by Clare and Cruchley (1957) and Diamond et al.
(1964). Dissolution of soil particle surfaces continues at a
decreasing rate, becoming insignificant as pH drops to values
probably less than around 9 (Eades and Grim 1960, Eades et al.
1962, Hunter 1988). The reaction products begin to harden or
crystallize as pH decreases. A calcium hydroxide particle is
attached to more than one soil particle, connecting them
together and producing silt- and sand-sized flocs and
agglomerates (Diamond et al. 1964, Verhasselt 1990). The
Atterberg plastic limit increases, often dramatically, because
large amount of water is enclosed within the flocs and

2553

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

agglomerates. In other words, only part of the porewater


contributes to plasticity. This is similar to diatoms with
poriferous particles in soils such as the Mexico City clay, and
andosols containing allophane in which water is trapped within
soil aggregates (Mesri et al. 1975, Terzaghi et al. 1996). Both
soils display unusually high plastic limits. In summary, total
lime content, lc, is used up through adsorption, lca, and
dissolution, lcd.
The time-dependent manifestation of adsorbed lime is a
gradual chemical reaction of calcium hydroxide with soil
particle surfaces. As the reaction products continue to form and
later harden or crystallize at the reaction sites of adsorbed lime
particles, they improve soil particle connections within the flocs
and agglomerates that may mature into porous soil aggregates
(Baver 1956). The proposed concept of lime particle adsorption
on soil particles is somewhat similar to physical adsorption of
calcium hydroxide molecules proposed by Diamond and Kinter
(1965). However, considering that a clay-sized hydrated lime
particle may contain 1011 molecules of Ca(OH)2, a more
significant time-dependent chemical reaction of adsorbed lime
with soil particle surfaces is expected for adsorbed lime particles
than for adsorbed lime molecules. Richardson et al. (1994) have
mentioned layers of Ca(OH)2 sandwiched between silicate
layers.
140
130
120

Untreated

110

Liquid Limit

Water Content, %

100
80
Lime Content, %

60

0
6.6

Plastic Limit

Untreated

20
100
Untreated
80
60
40
20

Plasticity Index
0

Figure 1.
pressure

10

20

30

40

50

BRENNA FORMATION

The highly plastic lacustrine clays of Lake Agassiz lead to slope


instability along the banks of the Red River that separates Grand
Forks, North Dakota from East Grand Forks, Minnesota, as it
flows north to Lake Winnipeg in Manitoba, Canada (Mesri and
Huvaj 2004). The clays of the Red River slopes are the glaciolacustrine deposits of glacial Lake Agassiz that is believed to
have existed from 13,000 to 8,500 years before present, during
the Late Wisconsin Glacial Episode of the Pleistocene Epoch
(Quigley 1980).
The Brenna Formation, which is characterized as a uniform,
soft to firm, dark grey, glacio-lacustrine clay with little or no
visible stratification, is full of slickensided surfaces. The major
source of sediment for the Brenna Formation was the highly
plastic montmorillonitic Pierre Shale bedrock (Quigley 1968,
Baracos 1977). The clay size fraction of Brenna Formation
ranges from 60 to 95% (Arndt 1977). This unit is divided into
Lower Brenna and Upper Brenna members. The natural water
content, plastic limit and liquid limit of Lower Brenna are in the
range of 42 to 69%, 20 to 40%, and 62 to 103%, respectively,
and the corresponding range for Upper Brenna are 60 to 85%,
27 to 38%, and 107 to 154%, respectively. Samples of both
Lower Brenna and Upper Brenna were used in the present
investigation.

100

40

particle size decreases and therefore, surface area increases, lca


increases. Lime content consumed through adsorption is
probably also related to the soil water content as it influences
dispersion of soil particles and facilitates thorough mixing to
allow full distribution and intimate contact between lime and
soil particles, degree of pulverization of hydrated lime, and the
intensity of mixing.
Because the solubility of calcium hydroxide in water is very
small, for typical soil water contents a very small lime content is
required to saturate the porewater. However, experience
indicates that pH remains below 12.3-12.4 for lime contents far
in excess of that required for the saturation of porewater. This
behavior appears to suggest that lime adsorption must be
satisfied before lime is dissolved in the porewater to increase the
pH. Zolkov (1962) considered it as remarkable that in spite of
the very small solubility of lime in water, large amount of lime
was required "to bring the pH of the soil slurry to 12.6."
Most of the chemical reaction products have a layer
structure, have high surface area, and a particle morphology that
has been described as thin plates, foils, and rolled up sheets
(Diamond et al. 1964; however sometimes fibers or laths occur
which could contribute to particle interlocking, Richardson et al.
1994). On the other hand, adequate but not excessive lime attack
may improve morphology of existing soil particles by producing
ragged, irregular, frosted or serrated particles and following
proper compaction connect them by the new reaction products.
These features are expected to improve mechanical behavior of
soils.
Because some of the reaction products during the
stabilization process are amorphous and hydrated, drying of
lime-treated soils during stabilization is likely to result in some
irreversible dehydration as well as irreversible aggregation.

60

Curing Time, days

Lime-Brenna clay interaction under effective confining

The lime content required to fully satisfy adsorption is


mainly related to soil particle size and shape and therefore, the
mineralogy of soil solids (Goldberg and Klein 1952, Eades and
Grim 1960) and degree of dispersion or aggregation. As soil

TESTS ON LIME-TREATED BRENNA CLAY

Drained direct shear tests on lime-treated Brenna clay were


performed using reconstituted specimens. Drained multiple
reversal direct shear tests on precut specimens were used to
measure residual shear strength, and drained direct shear tests
on uncut specimens were used to measure fully softened shear
strength. Air dry Brenna clay was pulverized until all of a
representative sample passed the no. 200 US standard sieve.

2554

Technical Committee 211 / Comit technique 211

the range of 100 to 450 kPa in the direct shear tests, there is a
minor increase in liquid limit (Fig. 1). The interpretation of this
significant observed behavior appears to be that when
aluminosilicates form in unconfined condition, they hydrate
fully, thus holding significant amount of water that contributes
to the high liquid limit. On the other hand when lime-clay
reaction products form under effective confining pressure, either
the resulting aluminosilicates do not hydrate much or they
experience irreversible dehydration through consolidation, thus
resulting in little change in the liquid limit. The implication of
this behavior, which is under more detailed examination, is
significant for both laboratory study of lime-soil interaction to
improve frictional resistance as well as field application of lime
to remediate reactivated landslides.
13
12
11
PH

The pulverized clay was mixed with dry hydrated lime, and was
thoroughly rehydrated using distilled water. Two halves of the
pre-cut specimen were formed by remolding or compaction and
separately consolidated inside the top and bottom halves of the
shear box using the procedure described by Mesri and CepedaDiaz (1986) and Mesri and Huvaj-Sarihan (2012). The
consolidation pressure ranged from 100 to 450 kPa, and shear
displacement rate was in the range of 3.3x10-4 to 5x10-4
mm/min.
Lime content as a percent of dry weight of clay ranged from
0 to 10%, and water content was in the range of 30 to 274%. In
a few direct shear tests, dry hydrated lime was sprinkled on the
exposed shear surface or on the top and bottom, of the direct
shear specimen to examine lime diffusion.
For one series of direct shear specimens with lime content
of 6.6%, liquid limit and plastic limit were determined at the end
of the test. These data are shown in Fig. 1. For another series
of lime-treated Brenna clay samples with lime content of 2, 5
and 9% and water content of 80, 100, 150 and 230%, pH and
Atterberg limits were measured as a function of time. These
samples were sealed; however, they were not subjected to
confining pressure. The pH measurements are shown in Fig. 2,
and the liquid limit and plastic limit at lime content of 5% and
water contents of 80 and 100% are shown in Fig. 3.
The data on residual friction angle and fully softened
friction angle from drained direct shear tests are summarized in
Table 1. All index tests and direct shear tests reported here were
performed at laboratory temperature of 20 2C.

10
9
8

Untreated

7
140
130

13

120

12

110
100

10
Lime Content, %
2
5
9

9
8
7

Untreated

80
100

Untreated

90

10

20

30
40
50
Curing Time, days

60

70

Figure 2. pH measurements of lime-Brenna clay


5

Liquid Limit

90

Water Content, %

PH

11

INTERPRETATION OF THE MEASUREMENTS

The pH measurements on lime-treated Brenna clay, such as


those in Fig. 2 as well as others, show that immediately after
introduction of lime, pH increases to a range of 12.2 to 12.7;
shortly thereafter, however, pH begins to decrease. This
observed behavior suggests that either within hours no free lime
is left to dissociate to maintain pH above 12, or dissociated
(OH)- is simultaneously consumed by the silica and alumina
dissolved from Brenna minerals. Nevertheless, during the 60
day observation period, pH remained above 9 suggesting
continued lime-clay chemical reactions.
The rather immediate large increase in plastic limit above
that of the untreated Brenna clay, such as observed in Figs. 1
and 3 results from flocculation and agglomeration of limetreated clay, especially as the water content is reduced during
the plastic limit measurements. Rapid chemical attack of
adsorbed lime on clay particles contributes to the production of
porous flocs and agglomerates that entrap water.
When the curing of lime-Brenna clay takes place
unconfined, liquid limit dramatically increases above the liquid
limit of untreated clay (Fig. 3); whereas when curing takes place
under an imposed effective stress condition, such as the n in

80
70
60

Plastic Limit

50

Untreated

40
30
50

Untreated
40

30

Lime Content, %

20

10

0
5

Plasticity Index
0

10

20

30

40

50

60

70

Curing Time, days

Figure 3. Lime-Brenna clay interaction under unconfined condition

The fully softened friction angle and residual friction angle


of stiff clays and shales decrease with the increase in effective
normal stress (Mesri and Shahien 2003, Mesri and HuvajSarihan 2012 ). The secant friction angles of Brenna clay in

2555

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Table 1 correspond to effective normal stresses of 100 kPa and


300 kPa. Lime treatment of Brenna clay increased fully
softened friction angle by 5 to 10 at effective normal stress of
100 kPa, and by 3 to 5 at 300 kPa. Lime treatment increased
the residual friction angle by 3 to 6 at both 100 kPa and 300
kPa. These results suggest formation of stable clay aggregates
through the lime-clay chemical reactions. These increases in
frictional resistance were realized with lime contents in the
range of 3 to 8% and treatment periods of 2 to 8 weeks. The
detailed correlation between improvement in frictional
resistance of Brenna clay as well as other stiff clays and shales,
with lime content and with duration of treatment, is under
further investigation with additional index and direct shear tests,
including scanning electron observations of reaction products.

effective normal stress range of 100 to 300 kPa. The increase in


drained frictional resistance suggests formation of stable clay
aggregates through lime-clay chemical reactions under the
highly alkaline pH 12.5 to 9.8 environment measured over a
period of 8 weeks. The measurements of liquid limit as an
indicator of changes in particle size and shape resulting from
lime treatment must be carried out on samples cured under an
effective stress condition rather than sealed but unconfined.
Unconfined lime treatment results in a significant increase in
liquid limit, thus underestimating the decrease in plasticity
index and associated increase in frictional resistance resulting
from lime remediation.
7

Table 1. Frictional resistance of lime-treated Brenna clay


Sample Ic (%)
1
2a
3b
4
5a
6b
7c
8
9a
10 a,d
11
12
13
14
15
16
17
18
19 c
20 a

0.0
0.0
3.0
3.0
3.0
3.0
4.0
5.0
5.0
5.0
5.0
5.0
6.6
6.6
6.6
6.6
6.6
6.6
8.0
10.0

Curing
(days)
0
0
7
11
28
54
1
11
7
120
40
56
3
8
14
26
26
0.4
180

w0 (%) [fs]s100 [fs]s300 [r]s100 [r]s300


67
111
74
98
30
74
74
109
111
111
274
274
75
97
75
64
77
105
74
111

15
24
17
-

14
-

20
29
34
20
18
20
36

17
18
21
18
-

9
9
11
11
11
11
15
13
12
13
15
13
11
16
-

7
6
7
9
8
9
11
10
12
12
13
13
11
16
-

Notes:
a- Lower Brenna was used for these specimens.
b- Lime was sprinkled on top and bottom of the sample to
investigate the treatment caused by lime diffusion.
c- Lime was sprinkled on the shear surface.
d- Lime-treated sample was stored for 120 days before being
placed in the shear box.
For the 27th Avenue slide in Grand Forks, North Dakota
(Mesri and Huvaj 2004), with entire slip surface in Brenna clay
at residual condition, 5% lime content treatment of fifty percent
of the slip surface increases computed factor of safety from 1.00
to the range of 1.26 to 1.37 (r = 7 to 8 increases to r = 12).
This level of lime remediation effort is expected to have a
significant effect on rate of movement of the slide.
A combination of horizontal directional drilling (HDD),
mechanical deep mixing (MDM) with augers and paddles, and
dry jet mixing (DJM), together with signal receivers at the
ground surface, is being investigated for introducing lime into
clay along a pre-existing slip surface. The longest crossing of
HDD to date has been 2000 m and borehole diameter of up to
160 mm.
6 CONCLUSIONS
Remediation of the montmorillonitic Brenna clay from North
Dakota using lime contents of 3 to 8% and treatment periods of
2 to 8 weeks increased drained fully softened friction angle by 3
to 10 and drained residual friction angle by 3 to 6, in the

2556

REFERENCES

Arndt B.M. 1977. Stratigraphy of offshore sediment Lake AgassizNorth Dakota, Report of Investigation No. 60. North Dakota
Geological Survey.
Baracos A. 1977.
Compositional and structural anisotropy of
Winnipeg soils a study based on scanning electron microscopy
and X-ray diffraction analysis. Can. Geotech. .J., 14 (1), 125-143.
Baver L.D. 1956. Soil Physics, 3rd edition. John Wiley & Sons,
London.
Bell F.G. 1996. Lime stabilization of clay minerals and soils.
Engineering Geology 42, 223-237.
Clare K.E. and Cruchley A.E. 1957. Laboratory experiments in the
stabilization of clays with hydrated lime. Gotechnique 7 (2), 97111.
Diamond S. and Kinter E.B. 1965. Mechanisms of soil-lime
stabilization: an interpretive review. Highway Research Record
92, 83-102.
Diamond A, White J.L. and Dolch W.L. 1964. Transformation of clay
minerals by calcium hydroxide attack. In: Bradley, W.F.(Ed.),
Proc. 12th Int. Conf. Clays and Clay Minerals. Pergamon Press,
New York, 359-379.
Eades J.L. and Grim R.E. 1960. Reaction of hydrate lime with pure
clay minerals in soil stabilization. Highway Research Record 262,
51-63.
Eades J.L., Nichols F.P. and Grim R.E. 1962. Formation of new
minerals with lime stabilization as proven by field experiments in
Virginia. Highway Research Board 335, 31,-39.
Goldberg I. and Klein A. 1952. Some effects of treating expansive
clays with calcium hydroxide. ASTM Special Publication 142,
Symp. on Exchange Phenomenon in Soils, 112-128.
Hunter D. 1988. Lime-induced heave in sulphate-bearing clay soils. J.
Geotech. Engrg. 1114 (2), 150-167.
Mesri G. and Cepeda-Diaz A.F. 1986. Residual shear strength of clays
and shales. Gotechnique 36 (2), 269-274.
Mesri G. and Huvaj N. 2004. Residual shear strength mobilized in Red
River slope failures. Proc. 9th Int. Symp. on Landslides, Brazil,
925-931.
Mesri G. and Huvaj-Sarihan N. 2012. Residual shear strength
measured by laboratory tests and mobilized in landslides. J.
Geotech. and Geoenviron. Engrg. 138 (5), 585-593.
Mesri G., Rokhsar A. and Bohor B.F. 1975. Composition and
compressibility of typical samples of Mexico City clay.
Gotechnique 25 (3), 527-554.
Mesri G. and Shahien M. 2003. Residual shear strength mobilized in
first-time slope failures. J. Geotech. and Geoenviron. Engrg. 129
(1), 12-31.
Quigley R.M. 1968. Soil mineralogy, Winnipeg swelling clays.
Can.Geotech. J., 5 (2), 120-122.
Richardson I.G., Brough A.R., Groves G.W. and Dobson C.M. 1994.
The characterization of hardened alkali-activated blast-furnace
slag pastes and the nature of the calcium silicate hydrate (C-S-H)
phase. Cement Concrete Res. 24 (5), 813-829.
Terzaghi K., Peck R.B. and Mesri G. 1996. Soil Mechanics in
Engineering Practice, 3rd edition. John Wiley & Sons, New York,
549 p.
Verhasselt A. 1990. Lime-cement stabilization of wet cohesive soils.
Proc. 6th Int. Symp. on Concrete Roads, Madrid, 67-76.
Zolkov E. 1962. Influence of chlorides and hydroxides of calcium and
sodium on consistency limits of fat clay. Highway Research
Record 309, 109-115.

Improvement of the Soil under the Concrete Pavement of a Plants Hall


Amlioration du terrain dassise sous la dalle en bton dune halle dusine
Mihova L.

University of Architecture, Civil Engineering and Geodesy, Sofia, Bulgaria,

Kolev Ch.

Todor Kableshkov University of Transport, Sofia, Bulgaria,

aBstract: the geological profile of the ground for the construction of a hall of the stilmet plant in sofia includes soft saturated
soils. the improvement is developed of the natural ground by constructing a geosynthetic reinforced pad of crushed stone. to
determine the mechanical parameters of the improved soil ground, in situ tests have been performed and settlement/load relationships
and e modulus values have been obtained. a numerical model is made of the ground by the finite element method. the undrained
short term stability and the consolidation long term stress-strain process of the improved soil ground are investigated.
rsUm : le profil gologique du terrain dassise, prvu pour la construction dune halle de lusine stilmet sofia, contient des
sols peu solides, imbibs deau. on a effectu une amlioration du terrain dassise naturel par la mise en place dune semelle en
pierres concasses, arme de matriaux gosynthtiques. pour dfinir les paramtres mcaniques de la fondation consolide, on a
excut des essais in situ et lon a obtenu la relation affaissement-charge, ainsi que le module e. on a tabli un modle numrique
suivant la mthode des lments finis. la stabilit court terme (non drain) et lvolution des contraintes et dformations
(consolidation) des sols amliors sont tudis.
KeYWords: soft saturated soil, geosynthetics, reinforced foundation pad, fem

introdUction

the design of reinforced earth structures to replace natural soft


soils is a modern practice in geotechnical engineering of
improving the foundation ground. high bearing capacity and
low ground deformation values are obtained by applying a
foundation pad constructed of layers of hard soil, like
compacted crushed stone, and of geosynthetic reinforcement.
the required thickness of the reinforced pad is much smaller
compared to unreinforced soil replacements. some projects
based on this way of soil improvement are realized in Bulgaria
in the recent years (mihov Y. and mihova l. 2012, Kolev ch.
and mihova l. 2012).
this paper presents some investigations of the improvement
of soft saturated ground under the hall of the stilmet plant in
sofia, which specializes in producing aluminum elements. the
geological profile includes uncompacted non-homogeneous
fillings at a depth of up to 4 m and soft clays at a depth of up to
10 m. the design of the ground improvement by the reinforced
pad involves the following steps: (1) choosing the thickness of
the pad and the number of reinforced layers, based on fe
analysis of various configurations of reinforced soil
replacement; (2) construction of an experimental improved
ground area and realization of in situ settlement/load tests,
using a circular steel plate with a diameter of 300 mm; (3) fe
modeling using the actual mechanical parameters, and analyzing
the stress-strain behavior and the stability of the improved
ground; (4) realization of the improvement of the halls ground,
and verifications of its deformation behavior using plate
settlement/load tests.

deep strip excavation in three longitudinal axes. the pavement


of the hall is made of fibre concrete with a thickness of 20 cm.
the equipment of the hall is composed of steel shelves, each
being supported at 8 points, and each being 12 m high and
weighing 12 tons. longitudinal beams on the concrete
pavement of the hall transform the point loading into striped.
the seismic loads on the pavement are obtained by performing
a dynamic analysis of the shelf structures.

(a)

(b)

figure 1. steel hall structure (a) and equipment shelf (b).

2.1 Geological profile


2

Geotechnical considerations

the stilmet plant hall, whose area is 3000 m2, is being


constructed near halls of the same kind (fig. 1). it has a frame
steel structure with spread footings constructed after a 4-meter-

the geological profile is shown in figure 2, and the properties of


the different layers are summarized in table 1. the water level is
1.5 m under the surface. the high water level requires analysis
of both the undrained short-term stability and the consolidated
long-term stress-strain behavior of the soil ground. the ground

2557

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

is being examined at a depth of up to 10 m, where solid clay


lies.

900/600 kn/m at 0.5% strain and coefficient of friction


soil/geogrid 1.2 (www.tenax.net).
3

to determine the e-modulus of the improved ground, a field


test program is performed. it includes the construction of the
reinforced pad of area 150 m2 and an application of a static
loading by rigid plate of dimension 30 cm at the following four
stages of construction: (1) after compaction of the natural
ground; (2) after building the first layer of crushed stone with a
thickness of 30 cm; (3) after placing the first geogrid layer and
building the second crushed stone layer with a thickness of 50
cm; (4) at the end of the pad construction. at each stage three
loading/unloading cycles are applied by steps of 0.05 mpa and
settlement/load curves are obtained. the E-modulus of total
settlement and the Ee-modulus of their elastic part are estimated,
and the results are shown in table 2. the settlement/load curves
for the first and the last stage of pad construction are shown in
figure 6. the moduli values increase more than five times after
the soil ground improvement.

figure 2. Geological profile


table 1. average values of the soil layers characteristics.
no.

soil type

e
-

kn/m3

c'
kpa

'
deg

E
mpa

top soil

1,35

1,45

11,0

8,0

3,5

Black clay

1,30

1,62

15,0

5,0

3,5

Brown clay

0,95

1,86

32,5

7,0

8,5

silty clay

1,41

1,70

11,0

5,0

6,0

sandy clay

0,82

1,89

32,5

18,5

15,0

field testinG procedUre

2.2 Structure of the reinforced crushed stone pad


investigations about the stress-strain behavior of the improved
soil ground with various thickness values of the crushed stone
pad, various numbers and various stiffness values of the
geosynthetic layers have been carried out in advance by fem
models. the optimal structure of reinforced pad with regard to
mechanical behavior of improved soil is obtained (fig. 3).

figure 5. construction of the experimental reinforced crushed stone pad

figure 3. structure of the reinforced crushed stone pad


figure 6. settlement/load curves for the plate loading tests
table 2. Values of the E-moduli of the soil ground at field testing
no.

figure 4. tenaX 3d Geogrid Xl (www.tenax.net)

the pad should be built of stone particles sized 085 mm,


and should be 1.3 m thick. the reinforcement is composed of
two polypropylene tenaX 3d geogrid Xl layers which have
particularly large concaved shaped ribs that enhance the
interaction mechanism between grids and stone particles by
restricting the horizontal movement of particles (fig. 4).
technical characteristics of the geogrids are: bi-axial stiffness

2558

stage of the pad construction

E
mpa

Ee
mpa

compaction of the natural ground

10,0

33,0

the first 30-cm-thick stone layer

25,7

60,0

the first geogrid layer and the second 50cm-thick stone layer

44,3

121,0

the end of the pad construction

57,7

181,0

Technical Committee 211 / Comit technique 211

nUmerical analYses

4.1 Finite element model


plane-strain finite element model of the improved ground is
made (fig. 7). the behavior of soil is modeled as mohrcoulomb material. linear bar elements that only have tensile
strength are used for the geogrids. the concrete pavement is
modeled by using linear beam elements. interface elements are
included for modeling the interaction between the soil and the
structure elements.
the loading of the pavement is assumed as uniformly
distributed with a value of 30 kpa for combination of dead and
live static loads and with a value of 45 kpa for seismic load
combination. Before the pavement loading calculations, the
initial condition of gravity loading is formed by the k0procedure. the construction stages of consecutive excavation
and the replacement of the soil are simulated by means of
phases of calculation with various fe meshes. an impermeable
bottom boundary of the fe model is assumed in consolidation
analysis.

the maximum pore pressure values are obtained


immediately after the load application, and its distribution is
represented in figure 9. it is evident that in all clayed soils under
the pavement the pore pressure increases up to the value of the
applied load. the 29.6 kpa maximum value of pore pressure is
calculated at point B situated at the bottom of the field. the
consolidation curves pore pressure vs. distance at 18 time steps
are shown in figure 10 for the cross section a B. step number
6 is related to the loading completion.

figure 10. the curves pore pressure vs. distance

the membrane forces of geogrids, caused by vertical


loading, reduce the normal stresses under it. the maximum
value of the normal stresses on the soft subsoil at the bottom of
the crushed pad is 52.3 kpa. figure 11 presents the tensile
forces in geogrids.

figure 7. finite element model

4.2 Results from FE analyses


4.2.1 Consolidation of the ground at dead and live loads
the consolidation process is investigated, and 3 years and 4
months is the time of the pore pressure dissipation. the
maximum value of the pavement settlement is 2.62 cm at point
a (fig. 7) and this value corresponds to the end of the
consolidation process. the distribution of the vertical
displacements is shown in figure 8.

( i layer )

( ii layer )

figure 11. the forces in geogrids at time moments: (a) at a pavement


loading; (b) at the end of consolidation

4.2.2 Stability of the ground at seismic load combination


the undrained analysis is performed and the lateral
displacements are estimated. the vectors of the total
displacements are shown in figure 12. the maximum horizontal
displacement is 1.4 cm and it occurs at depth of 4 m under the
crushed stone filling of the foundation excavation. the zones of
lateral displacements are located, as shown in figure 13, and the
stability of the ground is provided.
figure 8. the vertical displacements

figure 12. the vectors of the total displacements

figure 9. pore pressure distribution

2559

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

6 references
mihov Y. and mihova l. 2012. finite element analysis of reinforced
foundation soil. proc. 2nd international scientific meeting, tuzla.
Kolev ch. and mihova l. 2012. project for improvement of a soft soil
under the foundation slab of a building in haskovo. (unpublished).
tenaX 3d Geogrid Xl, www.tenax.net.

figure 13. the horizontal displacements

the stability of the ground has been estimated using the

, c -reduction method. the coefficient of stability has a value

fs = 3.42 for deep slide surface.


5 conclUsion

the required thickness of reinforced crushed stone pad is about


two times smaller compared to the unreinforced pad. the
improvement of the ground by replacing of the soft foundation
soil by the reinforced crushed stone pad is an effective modern
technology which decreases excavation works and increases the
heartedness of the foundation soil.

2560

Effect of Smear on Strength Behavior of SCP-Reinforced Soft Ground


Effet de comportement de ltalement de force du SCP- Sol mou renforc
Mir B.A.

Deptt. of Civil Engineering, National Institute of Technology Srinagar- 190006, Kashmir, India

Juneja A.

Deptt. of Civil Engineering, Indian Institute of Technology Bombay, Mumbai-400076, MH, India

ABSTRACT: Sand columns traditionally known as sand compaction piles-(SCPs) have been used to increase the load carrying
capacity of soft clays and accentuate consolidation during preloading. Installation of SCPs is known to cause disturbance due to
smear in a limited zone of the soil surrounding the SCP. In this study, conventional triaxial tests have been performed on 200mm
long and 100mm-diameter clay samples installed with SCP to simulate the strength behaviour of composite ground under different
confining pressures ranging from 50kpa to 575kpa. The SCPs were prepared using area replacement ratio of 6.25 to 64% and
compacted using pneumatic compactor. The smear zone was created by using a rough casing to drill the hole. The results seem to
suggest that the stress-strain behaviour of the clay was influenced by the presence of smear zone. The effect of smear zone on SCP
was investigated by observing the change in pore pressure during undrained shear strength of the composite ground. The natural
fabric of the soil was destroyed adjacent to the SCPs and the shear-induced pore pressures were less in composite specimens with
smear-effect. In addition, as the area replacement ratio was increased, both the stiffness and the strength of the specimen increased.
RSUM : Des colonnes de sable traditionnellement connues comme piles de compactage (SCPs) de sable ont t utilises pour
augmenter la capacit portante des argiles molles et accentuer la consolidation au cours du prchargement. Linstallation de MCS est
connue pour causer des perturbations dues au frottis dans une zone limite du sol entourant le SCP. Dans cette tude, les essais
triaxiaux conventionnels ont t raliss sur les chantillons dargile, de 100 mm de long et de 200 mm de diamtre, installs avec
SCP pour simuler le comportement de la rsistance du terrain composite sous diffrentes pressions de confinement allant de 50 kPa
575 kPa. Les MCS ont t prpars laide du coefficient de remplacement de 6,25 64 %, et compactes au pneumatique. La zone
de souillure a t cre en utilisant une enveloppe rugueuse pour percer le trou. Les rsultats donnent penser que le comportement de
contrainte-dformation de largile a t influence par la prsence de la zone de souillure. Leffet de zone de souillure sur SCP a t
examin en observant le changement de pression interstitielle au cours de la consolidation et de la rsistance au cisaillement du sol
composite. Le tissu naturel du sol a t dtruit adjacent la SCP et les pressions interstitielles induites par cisaillement taient
infrieures dans les chantillons composites avec un effet de maculage. En outre, lorsquon augmente le ratio de remplacement du
frottis, la rigidit et la rsistance de lchantillon augmentent.
KEYWORDS: Sand compaction pile, installation effects, smear, soft ground
MOTS-CLS : Sable tas de compactage, les effets de l'installation, les frottis, sol mou
1

INTRODUCTION

Soft ground is widely distributed especially along the coastal


area, having large potential for settlement with low inherent
shear strength. In the recent years, improvement of soft soils
has been extensively implemented for the various development
projects all over the world due to extremely limited stable
construction sites. Granular piles such as sand compaction piles
(SCPs) are considered as cost-effective and alternative solution
to the problem of stability and settlement posed by construction
on soft ground. The insertion of SCPs into soft clay has been
shown to have a positive effect on the load carrying capacity of
the clay, resulting in a composite soil mass that has greater
shear strength and improved stiffness compared to the
unreinforced clay. Sand compaction pile (SCP) is a method of
constructing large diameter sand column in the ground. This
method of ground improvement has been widely used for rapid
improvement of soft ground, and also in near-shore regions for
land reclamation works (e.g. Aboshi and Suematsu 1985,
Bergado and Balasubramniam 1994). In India, the granular
columns have been used to improve ground for container freight
station at Navi Mumbai and the construction of dry dock at
Pipavav shipyard (Raj and Dikshith 2009). Many researchers
(e.g. Bergado et al. 1991, Juneja and Mir 2011) have
investigated the effect of SCP installation on disturbance to the
surrounding soil. The extent of the disturbed or smear zone can

affect the engineering behaviour of the composite ground. The


disturbance in this zone depends upon the column diameter and
the tools used in the installation (e.g. Singh and Hattab 1979,
Madhav et al. 1993). Shear induced pore pressures were found
to be less in specimens which had the smear zone surrounding
the sand column. However, pore pressures began to increase
close to failure due to rearrangement of soil particles (Mir
2010). Laboratory and field tests previously conducted to
determine the extent of the disturbance caused by pile driving
into soft clay deposits have demonstrated that the natural
structure of the clay around the pile is excessively disturbed
(Randolph et al. 1979, Xu et al. 2006). It was observed that the
diameter of the severely disturbed or remoulded ground around
a driven closed-ended casing was about 1.4 times the diameter
of the casing. Recently, Weber et al. (2010) compared the
smear zone around model SCPs to that observed around driven
piles. It was observed that the smear zone around SCPs
installed on the centrifuge extended up to 1.2 to 1.4 times the
SCP diameter. Dissipation of the excess pore pressures often
results in increase in the shear strength. Aboshi et al. (1979)
observed up to 50% increase in the undrained strength in about
one month after the SCP installation at test sites in Japan.
Matsuda et al. (1997) also reported an increase in strength of the
composite SCP ground within three months of the SCPs
installation.

2561

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

In this study, the effect of smear zone on strength of model


SCP installed in 100mm diameter and 200mm long clay
specimens is investigated using conventional triaxial
compression tests under different confining pressures ranging
from 50kPa to 575kPa. The composite specimen were prepared
by driving a small diameter PVC casing into the sample and
then backfilling the cavity with sand column after removing the
casing. The casing was roughened using sand glued to its outer
walls prior to insertion to replicate the smearing effect. The
SCPs were prepared using area replacement ratio of 6.25 to
64% and compacted using pneumatic compactor. SCPs of
different diameters (25-80mm) were used to investigate the
improvement in the load-carrying capacity of the specimens.
The effect of smear zone on SCP was investigated by observing
the change in pore pressure during consolidation and undrained
shear strength of the composite ground. The test results suggest
that, tress-strain behavior of the clay was influenced by the
presence of smear zone. The natural fabric of the soil was
destroyed adjacent to the SCPs with smear zone which in turn
affected pore pressure response of the composite soil sample.
Shear induced pore pressures were less in soil specimens with
smear-effect, but this difference was not apparent when 80mm
diameter SCP with smear zone was used. In addition, as the
reinforcement area ratio increased, both the stiffness and the
shear strength of the specimen increased. Thus, sand
compaction piles currently stand as one of the most viable and
practical techniques for improving the mechanical properties of
soft clays.
2. EXPERIMENTAL WORK
2.1. Materials and methods of sample preparation

Soil
specimen

The test specimens were prepared in 450mm long and 250mm


diameter stainless steel cylindrical mould. Deaired clay slurry
was consolidated on the laboratory floor, first under its own
self-weight and later under surcharge of 211- to 404 kN/m2
applied in stages on top of the clay surface using a custom
designed pneumatic load frame (Fig.1).

Slurry consolidation Specimen trimming Final specimen size


Figure 1. Consolidation set-up on the laboratory floor

PVC casing
pushed into
the sample

Pneumatic
compactor

Table 1 Properties of kaolin clay


Clay
Silt
Liquid
(%)
(%)
limit (%)
75
25
49

Plastic
limit (%)
23

Shrinkage
limit (%)
16

Gs
2.64

3. RESULTS AND DISCUSSIONS


Consolidated undrained triaxial tests were performed on 200mm
long and 100mm diameter cylindrical samples prepared from
remoulded and reconsolidated commercially available kaolin
clay installed with SCP. Table 2 shows the details of the soil
specimens prepared for testing. In the table, OCR is defined as
the ratio of the isotropic preconsolidation pressure, p0' to p'. p0'
was taken equal to the higher of either p' or the mean effective
stress after 1D consolidation, p' estimated using the equation
(Wroth 1984):

p' 'v 1 0.67 sin '

(1)

where is the effective angle of friction (e.g. Schofield and


Wroth 1968). The load-deformation data was analyzed using the
unit cell arrangement proposed by Balaam et al. (1977). In this
method, the column and surrounding clay are assumed to act as
a single element with equivalent distributions of stresses and
strains in composite specimens. Figures 3a-b show results of
deviator stress, q plotted against axial strain, a. As can be
seen, all samples reached peak deviator stress (qmax) at 6 to 10%
axial strain. Figures 3a-b also show that the ultimate strength
exhibit transient peaks in some tests. This was expected since
these soil samples were overconsolidated prior to the shearing.
In few tests on normally consolidated clays, q decreased after
passing qmax because of instability of the failed samples at high
confining pressure.

Soil sample

Upon completion of the 1-D consolidation, the block of clay


was extruded and trimmed into three 100mm diameter
cylindrical specimens using soil lathe. Up to 3 specimens could
together be prepared using this mould. The experimental
program consisted of 20 tests on composite clay with sand
column. The specimens were held in split cylindrical moulds
and a smooth PVC casing slowly pushed along its length to form
a cylindrical hole at the centre. The hole was backfilled with
fine sand (d50=0.3mm) compacted in layers at 90% relative
density using a pneumatic compactor (Fig. 2).
Sand column

(Aboshi et al. 1979) that ranges between 6.25- and 64%. The
smear zone was created by using a rough casing painted with a
paste of coarse sand (d50 = 1.3mm) to drill the hole. Thickness
of the smear zone was taken equal to the thickness of the paste.
The effect of smear beyond this zone was ignored. After
preparing the sand column, the ends of the specimen were
covered with a thin circular rubber sheets having a central hole.
Diameter of the hole was slightly less than that of the sand
column so as to only permit radial drainage. Two deaired
porous stones were then placed at the two ends of the specimen
and the entire assembly mounted on the triaxial chamber.Table
1 shows properties of the clay used in this study. The ratio of
the diameter of sand column with smear zone to the diameter of
sand column without smear zone (ds/d) was about 1.1 to 1.2 in
all tests, which compares well with the values reported by the
previous researchers (e.g. Indraratna and Redana, 1998;
amongst others). The specimen was enclosed in a rubber
membrane and the chamber filled with water. The soil samples
were then isotropically consolidated under mean effective
stress, p' which varied between 50 and 575 kN/m2.

Figure 2. Preparation of composite specimen

Diameter of the sand column varied between 25- and 80mm in


the specimens. This corresponds to an area replacement ratio, as

2562

Technical Committee 211 / Comit technique 211

Table 2 Experimental program


Test
v*
ds#
Smear
po $
p**
No
(kPa)
(mm)
zone
(kPa)
(kPa)

S1
404
24.6
285
95

S1
404
24.6
285
146

S1
404
24.6
300
299

S1
264
29
187
95

S1
264
29
187
145

S1
264
29
300
289

S2
211
31.7
450
450

S2
211
31.7
200
197

S2
211
31.7
149
50

S2
211
35.7
450
434

S2
211
35.7
200
195

S2 211
35.7
149
49

S3
211
40
375
374

S3
211
40
575
575

S3
211
40
149
71

S3
211
45.2
575
572

S3
211
45.2
375
372

S3
211
45.2
149
68

S4
211
80
149
144

S4
211
80
149
142
*:
v =Vertical stress at end of 1D loading,
#:
ds = Equivalent diameter of sand column,
$:
po =Preconsolidation pressure,
**:
p = Mean effective stress at end of consolidation,
800
600
q (kN/m2)

3
2
1
2
1.3
1
1
1
3
1
1
3
1
1
2.1
1
1
2.2
1
1

500
300
100
0

10

15

20

25

q (kN/m2)

600

p ' 0.33OCR 0.90 - with smear effect

30

1.2

400
200
100
15

20

25

A-factor

0.4

S2:50kPa
S3:575kPa
S3:375kPa
S3:75kPa

10

15

20

30

Figure 4. Variation of Skempton`s parameter A with axial strain for


samples with and without smear zone.

30

900

CSL

Without smear
With smear

800

S1:100kPa
S1:150kPa

q (kN/m2)

700

Figure 3a-b Deviator stress versus axial strain relationship for: (a)
Samples without smear zone; and
(b) Samples with
smear zone.

The undrained shear strength (su) of composite specimen was

600

500

400

was

S2:450kPa
S2:200kPa

S3:375kPa
S3:75kPa
S4:150kPa

100
0

S1:300kPa

S2:50kPa
S3:575kPa

300
200

q
taken equal to max , and the effect of stress history induced
2

su
ratio '
p

25

Axial strain (%)

(b)

over consolidation on undrained strength

S1:150kPa
S1:300kPa
S2:450kPa

0.6

-0.2

Axial strain (%)

expressed as:

S1:100kPa

S2:50kPa
S3:575kPa
S3:375kPa
S3:75kPa
S4:150kPa

300

(4)

Without smear
With smear

0.2

S1:300kPa
S2:450kPa

500

an empirical

(3)

0.8

700

10

for OCR=1, and m is

The evidence of the smear zone was not significant on the


ultimate undrained shear strength when 25mm and 30mm
diameter sand columns were used. What was surprising is that
the effect of smear was apparent on the ultimate shear strength
when 80mm diameter sand columns were used. It seems clear
that the presence of smear zone has reduced the ultimate
undrained shear strength by 25%. In addition, as the area
replacement ratio was increased, both the stiffness and the shear
strength of the composite samples also increased. Figure 4
show results of Skempton's (1954) pore pressure parameter, a,
plotted against axial strain, a. As seen, the A-factor in
specimens with smear effect was less from early stage of
shearing even when the specimens had the same OCR. This
was expected because the smear zone did not permit the pore
pressures to dissipate within the SCP. However, there was a
marginal increase in a-factor after passing qmax. af of these
samples was between 0.7 and 1.1 which is typical for normally
consolidated clays.

With smear
Test
S1:100kPa
S2:150kPa

NC

(a)

800

p'

p' 0.44OCR 0.99 - without smear

Axial strain (%)


900


exponent equal to 1 , , are soil model parameters

obtained from triaxial testing. Using test data, undrained shear
strength (su) of composite ground was expressed in the
following form:

S3:575kPa
S3:375kPa
S3:75kPa
S4:150kPa

200

equal to

S2:450kPa
S2:50kPa

400

(2)

where a is the normalized undrained shear strength of NC soil

Without smear
Test
S1:100kPa
S1:150kPa
S1:300kPa

700

s p'
aOCR m
u

OC

OCR

100 200 300 400 500 600 700 800 900


p' (kN/m2)

Figure 5. Effective stress paths for samples without smear and with
smear zone

Figure 5 show the effective stress path in p'-q stress space.

2563

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

The figures show that the shear induced pore pressures were
found to be less in specimens which had the smear zone
surrounding the sand column. Due to the smear zone, these
pore pressures within the SCP were reduced because the water
was not permitted to flow towards the column during shearing.
This was also evident from Scanning Electron Microscope
(SEM) images (Fig. 6a-b) taken on post shear tests of specimens
with and without smear. The images of samples with and
without the smear zone show differences in the microstructure.
The clay minerals in the smear zone appear to be closely packed
with reduced pore space. 7.5 mm x 7.5 mm x 7.5 mm air dried
samples were prepared at room temperature for the SEM
images.
Pores

(a)

(b)

Figure 6. SEM images: (a) Composite samples without smear zone; and
(b) Composite samples with smear zone.

4. CONCLUSIONS
The strength behavior of composite ground reinforced with sand
compaction piles has been studied using 20 consolidated
undrained triaxial tests. Vertical stress of the sand column was
examined when the composited specimens were tested to failure
in conventional triaxial tests. The test results suggest that the
stress-strain behavior of the clay was influenced by the presence
of smear zone. It seems clear that the presence of smear zone
has reduced the ultimate undrained shear strength by 25%.
SEM images indicated that the natural fabric of the soil was
destroyed adjacent to the SCPs with smear zone which in turn
affected pore pressure response of the composite soil sample.
Specimens sheared with smear effect appear closely packed and
more homogeneous with partly discernible particle systems,
while specimen without smear condition indicate a distinct
division between smaller intra-aggregate pore spaces and the
larger inter-aggregate voids. Shear induced pore pressures were
less in soil specimens with smear-effect because the water was
not permitted to flow towards the column during shearing. It
was postulated that the difference in the behavior of smear and
non-smear specimens was because that the pore pressure
measurements were taken within the sand column. This
difference was not apparent when 80mm diameter SCP with
smear zone was used. These changes affect the effective
horizontal stress in the clay and hence the load carried by the
individual sand columns.

6. REFERENCES
Aboshi, H. and Suematsu, N. (1985). Sand compaction pile method:
State-of-the art paper. Proceedings of the 3rd International
Geotechnical Seminar on Soil Improvement Methods, Narryang
Technological Institute, Singapore.
Aboshi, H., Ichimoto, E., Enoki M. and Harada, K. (1979). Composer:
method to improve characteristics of soft clays by inclusions of
large diameter sand column. Proceedings of the International
Conference on Soil Reinforcement: Reinforced Earth and other
Technique, Paris, Vol. 1, 211-216 (1979).
Asaoka, A., Kodaka, T. and Nozu, M. (1994). Undrained shear strength
of clay improved with sand compaction piles.
Soils and
Foundations, Vol. 34, No. 4, 23-32.
Balaam, N.P., Brown, P.T. and Poulos, H.G. (1977). Settlement
analysis of soft clays reinforced with granular piles. Proceedings
of the 5th Southeast Asian Conference on Soil Engineering,
Bangkok, 8191.
Bergado, D. T. and Balasubramniam, A. S. (1994). Laboratory testing
of prefabricated vertical drains (PVD). Geotechnical Engineering
Journal, Vol. 25, No. 1, pp.1.
Indraratna, B. and Redana, I. W. (1998). Laboratory determination of
smear zone due to vertical drain installation.
Journal of
Geotechnical Engineering, ASCE, Vol. 124, No. 2, 180-184 (1998).
Juneja, A. and Mir, B. A. (2011). Effects of Sand Compaction Pile
Installation on Surrounding Soft Soil. Procc. of National
Conference on Recent Advances in Ground Improvement
Techniques (RAGIT-2011), pp. 1-10, CBRI Roorkee, India.
Madhav, M. R., Park, Y. M. and Miura, N. (1993). Modelling and Study
of Smear Zones around Band Shaped Drains.
Soils and
Foundations, Vol. 33, No. 4, 137-149.
Matsuda, H., Fujiwara, Takahasi, S. and Kitayama, M. (1997).
Influence of SCP driving on the behaviour of clay. Ground
improvement geosystems: Densification and reinforcement,
Thomas Telford London, 233-238.
Mir, B. A. (2010). Study of the influence of smear zone around sand
compaction pile on properties of composite ground. Ph.D. Thesis,
IIT Bombay.
Singh, G. and Hattab, T. N. (1979). A laboratory study of efficiency of
sand drains in relation to methods of installation and spacing.
Geotechnique 29(4), 395-422.
Skempton, A. W. (1954). The pore-pressure coefficients A and B.
Geotechnique 4(4), 143-147
Raj, D. and Dikshith, C.V. (2009). Vibro replacement columns for
shipyard infrastructure at Pipavav, Gujarat, India. Proceedings of
the International Symposium on Ground Improvement
Technologies and Case Histories, GeoSS, Singapore, 763-769.
Randolph, M. F., Carter, J. P. and Wroth, C. P. (1979). Driven piles in
clay-the effects of installation and subsequent consolidation.
Geotechnique 29(4), 361-393.
Schofield, A. N. and Wroth, C. P. (1968). Critical State Soil
Mechanics. MacGraw-Hill, London, p. 218.
Xu, X. T., Liu, H. L. and Lehane, B. M. (2006). Pipe pile installation
effects in soft clay. Proceedings of the Institution of Civil
Engineers, Geotechnical Engineering 159, Issue GE4 pp. 285-296.
Weber, T. M., Plotze, M., Laue, J., Peschke, G. and Springman, S. M.
(2010). Smear zone identification and soil properties around stone
columns constructed in-flight in centrifuge model tests.
Geotechnique 59(3), 197-206.
Wroth, C. P. (1984).
The interpretation of in-situ soil tests.
Geotechnique 34(4), 449-489.

5. ACKNOWLEDGEMENT
The first author would like to acknowledge the financial support
from the Indian Institute of Technology Bombay (IITB) in the
form of research scholarships.

2564

Bio-mediated soil improvement utilized to strengthen coastal deposits


Amlioration du sol biologiquement ngocie utilise pour renforcer les dpts ctiers
Montoya B.M., Feng K., Shanahan C.

North Carolina State University, Raleigh, NC, USA

aBstract - Vital coastal lifelines can be vulnerable during large storm events. large wave action and high sea levels erode the
sandy soil that supports coastal infrastructure, including highways, structures, pipelines, and other utilities. damage from these
events can result in severe property damage, loss of revenue, and large repair costs. natural bio-geochemical methods can be used
to reinforce the erodible sandy soil to help prevent damage to the infrastructure. Utilizing naturally-occurring biological metabolic
activity, calcium carbonate cementation can be induced in situ to bind the sand grains together, thereby improving the strength and
stiffness of the soil and in turn preventing erosion of the coastal deposits. microbial induced carbonate precipitation (micp) has
been shown to be an effective method to improve the soil behavior in saturated conditions subjected to undrained monotonic and
seismic loading in both laboratory and centrifuge tests. applying this proven natural treatment technique to unsaturated coastal soils
can improve the soils resiliency during large storm events. results indicate that the strength of the unsaturated soils increase from
intermittent surficial treatments. rigid-walled soil column tests were conducted to evaluate the effectiveness of treating sandy soils
by flooding the surface of the soil with the appropriate microbes and nutrients and allowing free drainage. clean fine sand, typical
of coastal dune deposits, was used in the soil column tests. the strength of the cemented sand was evaluated using unconfined
compression tests. a discussion of upscaling the results from the laboratory tests to application in situ to improve the resiliency of
coastal infrastructure is also presented herein.
1. introdUction
Vital coastal lifelines can be vulnerable during large storm
events. large wave action and high sea levels erode the sandy
soil that supports coastal infrastructure, including highways,
structures, pipelines, and other utilities. damage from these
events can result in severe property damage, loss of revenue,
and large repair costs. the outer banks of north carolina have
seen several hurricanes in recent years (irene in 2011, and
sandy in 2012) which have inflicted damage to vital coastal
lifelines as illustrated in figures 1 and 2.
natural bio-geochemical methods can be used to reinforce the
erodible sandy soil to help prevent damage to the
infrastructure.
Utilizing naturally-occurring biological
metabolic activity, calcium carbonate cementation can be
induced in situ to bind the sand grains together, thereby
improving the strength and stiffness of the soil and in turn
preventing erosion of the coastal deposits. microbial induced
carbonate precipitation (micp) has been shown to be an
effective method to improve the soil behavior in saturated
conditions subjected to undrained monotonic and seismic
loading in both laboratory and centrifuge tests (montoya et al.,
2013, mortensen and deJong, 2011, deJong et al., 2006).
applying this proven natural treatment technique to
unsaturated coastal soils can improve the soils resiliency
during large storm events. results indicate that the strength of
the unsaturated soils significantly increase from intermittent
surficial treatments. rigid-walled soil column tests were
conducted to evaluate the effectiveness of treating sandy soils
by flooding the surface of the soil with the appropriate
microbes and nutrients and allowing free drainage. clean fine
sand, typical of coastal dune deposits, was used in the soil
column tests. changes in the strength of the sand from the
unsaturated cementation treatments was evaluated using
unconfined compression tests. a discussion of upscaling the
results from the laboratory tests to application in situ to
improve the resiliency of coastal infrastructure is also
presented herein.

2565

figure 1. a section of highway 12 at the edge of rodanthe, n.c.


undermined by erosion due to the storm surge and wave action during
hurricane irene. (photo: news & observer, aug. 31, 2011)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

cementation treatments were performed by flooding the


top surface of the soil column, and allowing the cementation
media to freely drain through the sand (figure 3). Bacteria
were introduced into the soil during the initial cementation
flush. calcium chloride was not included in the initial
treatment with the bacteria to prevent precipitation during
inoculation. cementation treatments were repeated every 3 to
6 hours. two pore volumes of nutrients were used in each
treatment flush (concentrations of nutrients presented in table
ii). the cementation treatments were repeated for a total of 40
times.

figure 2. high ocean waves from hurricane sandy lap against


highway 12 and erode the underlying sand. (photo: news & observer,
nov. 14, 2012)

2. materials and methods


2.1. Sand and Specimen Preparation
four soil column specimens were prepared by dry pluviation
to a target relative density of 40%. the soil column specimens
had a 50.8 mm (2 in) diameter and an aspect ratio of 2:1.
ottawa 50-70 sand was used for the initial cementation trials,
because of the published results with the sand (deJong et al.,
2006, mortensen and deJong, 2011, montoya et al., 2013,
martinez et al., 2013). a summary of the sand characteristics
is listed in table i.
figure 3. initial biological flush through soil columns. cementation
flushes allowed to free
drain through pluviated soil.

table i. ottawa 50-70 sand characteristics

d50 (mm)
cu
cc
Gmax
emin
emax
mineralogy
shape

ottawa 50-70
sand characteristics
0.22
1.4
0.9
2.65
0.55
0.87
Quartz
round

2.3. Specimen Shearing


after cementation was completed, the specimen was flushed
with water to remove residual chemicals from the cementation
treatments. the cemented sand was removed from the soil
columns. the cemented soil columns were then subjected to
unconfined compression tests until failure.
a GeoJac
automated load actuator was used to perform the unconfined
compression test.

2.2. Biological Treatment Process


Sporosarcina pasteurii (atcc 11859), a urea hydrolyzing
bacterium, was grown at 30 c in an ammonium-Yeast
extract medium (atcc 1376: 0.13 mol l-1 tris buffer
(ph=9.0), 10 g l-1 (nh4)2so4, and 20 g l-1 yeast extract).
individual ingredients were autoclaved separately and mixed
together post-sterilization.
the growth medium was
inoculated with the S. pasteurii stock culture and incubated
aerobically at 30 c in a shaking water bath (200 rpm) for
approximately 40 hrs before harvesting at a final optical
density (od600) of 0.8-1.0. cultures were centrifuged at 4000
g for 10 min in 15 ml volumes and washed in fresh growth
medium. harvested bacteria were stored in the centrifuge vials
at 4 c for a maximum of 2 days.
Urea-calcium cementation media was used to induce
ureolytic-driven calcite precipitation. a summary of the
components and concentrations are presented in table ii.
table ii. chemical recipe for cementation media

chemical

chemical concentration (m)

Urea

1.0

cacl2

0.25

2.4 Mass of Calcium Carbonate Measurements


the mass of calcium carbonate was determined post-test using
methods outlined in astm d4373, Standard Test Method for
Rapid Determination of Carbonate Content in Soils. at the
end of the cementation treatments, oven dried cemented sands
are dissolved in hydrochloric acid and the resulting pressure
generated from the dissolution of calcium carbonate is
measured. the generated pressure is equated to an equivalent
mass of calcium carbonate. the percentage of mass of calcium
carbonate is expressed as the mass of calcium carbonate
divided by the mass of soil (not including calcium carbonate).
3. resUlts
3.1 Calcium Carbonate Content in Soil Columns
mass of calcium carbonate was taken in the top and bottom of
the cemented soil column, which are reported in table iii. as
indicated in table iii, the mass of calcite in the four soil
columns is relatively small compared to published results from
other micp treatment studies (Weil et al., 2011). however,
the cementation within the soil column was extremely uniform,
as indicated by the mass of calcium carbonate of the top and
bottom samples (table iii).

2566

Technical Committee 211 / Comit technique 211

table iii. summary of mass of calcium carbonate

soil column
1 (top)
1 (bottom)
2 (top)
2 (bottom)
3 (top)
3 (bottom)
4 (top)
4 (bottom)

mass of caco3 (%)


0.082
0.081
0.102
0.103
0.070
0.068
0.070
0.068

3.2 Cemented Sand Compressive Strength


Unconfined compression tests were performed on the
cemented soil columns (figure 4 and figure 5). as mentioned,
the mass of precipitated calcium carbonate is relatively small
compared to published results from other micp treatment
studies; however even at low mass of calcium carbonate levels,
unconfined compression tests were able to be performed on the
cemented sand columns. a summary of the compression test
results are listed in table iV.

figure 5. failed cemented soil column at the


end of the unconfined compression test.
table iV. summary of compressive strength

soil column
1
4

figure 4. cemented soil column mid-test


during the unconfined compression test.

Unconfined compressive
strength (kpa)
5.2
5.4

as indicated in table iV, unconfined compression tests were


performed on only two of the four columns. two of the soil
columns were not able to be tested because they were
disturbed during extraction from the soil column walls. the
cemented soils were especially vulnerable because of the low
levels of cementation. to rectify the tendency for disturbance,
the soil in the remaining columns was extracted from the soil
column walls by creating vertical slices through the acrylic
walls and allowing the soil to be removed through the sliced
opening.
the unconfined compressive strength of the cemented sand
columns was about 5 kpa. other studies found that micp
treated ottawa 50-70 sand could get compressive strengths of
about 170 to 350 kpa at higher levels of cementation (faison
and mahin, 2012).
the angle of the failure plane in soil columns 1 and 4 was
about 63 degrees from the horizontal. this failure plan angle
is representative of soil with a friction angle of 36 degrees.
Untreated ottawa 50-70 sand has a friction angle of about 33
degrees (montoya, 2012). Based on previous work, micp
treated sand with a friction angle of 36 degrees is typical of
sand treated to a shear wave velocity of 400 m/s (montoya,
2012). an approximate shear wave velocity of 400 m/s
corresponds to the strength data, indicating the cemented soil
columns represent lightly cemented sand.
for use as a treatment process for costal sand deposits, an
appropriate level of micp cementation should be used. a high
enough level of cementation should be used to resist induced
shear loads from waves and storm surges, and a low enough
level of cementation so that native wildlife, such as birds,
burrowing animals, and dune grass, can still interact with the
coastal deposits. further work will include upscaling the

2567

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

treatment process using a wave tank to identify optimum


ranges of cementation for treatment of coastal deposits.
4. conclUsions
micp can be used to reinforce sandy coastal deposits to
improve the resiliency of vital lifelines during large storm
events. soil columns of clean fine sand were treated with
micp, and resulted in lightly cemented sand. the lightly
cemented sand had an increase in strength, as demonstrated
with the unconfined compression tests, and increase in friction
angle. the free-draining treatment process was designed to be
similar to likely treatment processes of unsaturated surficial
sands in situ. this treatment process provided uniform levels
of cementation throughout the height of the soil column. the
light levels of cementation achieved in the soil columns
provide an increase in shear strength while still allowing for
birds, burrowing animals, and dune grass to interact with the
coastal deposits. future work involves investigating the
optimal range of micp cementation to provide enough
strength to resist the loads from large storm events while
continuing to support the coastal ecology.

5. references
astm d4373, Standard Test Method for Rapid Determination
of Carbonate Content in Soils
deJong, J.t., fritzges, m.B., and nsslein, K. (2006)
microbial induced cementation to control sand
response to Undrained shear, ASCE Journal of
Geotechnical and Geoenvironmental Engineering, Vol.
132, no. 11, pp. 1381-1392.
faison, h., and mahin, s.a. (2012). microbial induced
calcite precipitation in partially saturated soils. PEER
2011/10 Earthquake Engineering for Resilient
Communities: 2011 PEER Internship Program Research
Report Collection, december.
martinez, B.c., deJong, J.t., Ginn, t.r., mortensen, B.m.,
Barkouki, t.h., hunt, c., tanyu, B., major, d. (2013)
experimental optimization of microbial induced
carbonate precipitation for soil improvement, ASCE
Journal of Geotechnical and Geoenvironmental
Engineering, in press.
montoya, B.m. (2012) Bio-Mediated Soil Improvement and
the Effect of Cementation on the Behavior, Improvement,
and Performance of Sand, doctoral dissertation,
University of california, davis, pp. 238.
montoya, B.m., deJong, J.t., and Boulanger, r.W. (2013)
seismic response of liquefiable sand improved by
microbial induced calcite precipitation, Geotechnique. in
review.
mortensen, B.m., and deJong, J.t. (2011). strength and
stiffness of micp treated sand subjected to Various
stress paths, ASCE GeoFrontiers 2011: Advances in
Geotechnical
Engineering,
Geotechnical
special
publication 211, pp. 4012-4020.
Weil, m.h., deJong, J.t., martinez, B.c., mortensen, B.m.,
Waller, J.t. (2012). seismic and resistivity
measurements for real-time monitoring of microbially
induced calcite precipitation in sand.
ASTM
Geotechnical Testing Journal, Vol. 35, no.2.

2568

Effect of Grout Bleed Capacity on the Engineering Properties


of Cement Grouted Sands
Effet de la capacit de ressuage de coulis de ciment sur les proprits mcaniques
des sables injects
Pantazopoulos I.A., Atmatzidis D.K., Basas V.G., Papageorgopoulou S.K.

Geotechnical Engineering Laboratory, Department of Civil Engineering, University of Patras, Greece


ABSTRACT: Grouts of three different cement types, each at four different cement gradations, with W/C ratios ranging from 0.6 to
3.0 and bleed capacities ranging up to 70% were injected into two different sands. Permeability, unconfined and triaxial compression
and resonant column tests were conducted to investigate the influence of grout bleed capacity on the engineering properties of cement
grouted sands. Cement grouting resulted in (a) permeability coefficient values as low as 10-8cm/s, (b) unconfined compressive
strength in the range of 1MPa to 35MPa, (c) cohesion in the range of 100kPa to 1400kPa, (d) improvement of the internal friction
angle by up to 5, (e) higher shear modulus by up to 25 times and (f) improved damping ratio by up to 10 times. Bleed capacity is an
indicator of sand void volume filled with solidified grout but its degree of correlation with the static and dynamic properties of the
grouted sands ranges from very good to negligible.
RSUM : On a inject des coulis de trois types de ciments diffrents, chacun avec quatre dosages en ciment diffrents, avec un
rapport eau/ciment variant de 0.6 3.0, et une capacit de ressuage se situant jusqu 70% lorsque injects, dans deux sables
diffrents. On a effectu des essais de permabilit, de compression simple et triaxiale et de colonne rsonnante pour tudier
l'influence de la capacit de ressuage des coulis sur les proprits mcaniques des sables injects. Linjection du ciment a rsult en:
a) des valeurs de coefficients de permabilit aussi faible que 10-8cm/s, b) une compression simple de 1MPa 35MPa, c) une
cohsion de 100kPa 1400kPa, d) une augmentation de langle de frottement jusqu 5, e) un module de cisaillement jusqu 25 fois
plus lev, f) une augmentation du coefficient damortissement jusqu 10 fois plus lev. Le ressuage des coulis est un indicateur du
volume des vides du sable remplis de coulis solidifi mais son degr de corrlation avec les proprits statiques et dynamiques des
sables ciments varie de trs bonnes ngligeables.
KEYWORDS: cement grout, bleed capacity, permeability, strength, shear modulus, damping ratio

INTRODUCTION

Improvement of the mechanical properties and behavior of soils


by permeation grouting using cement suspensions is frequently
required in order to assure the safe construction and operation of
many structures. The grout water-to-cement ratio (W/C) and the
maximum cement grain size (dmax) are two important
parameters controlling the cement grout bleed capacity and,
consequently, the effectiveness of cement grouts in terms of the
percentage of soil voids volume filled by grouting. Although the
bleed capacity of cement grouts has been frequently quantified,
its correlation with the engineering properties of the grouted
sand has not been investigated so far.
Scope of this presentation is to provide some insights on the
effect of grout bleed capacity on permeability, unconfined
compressive strength, shear strength parameters and dynamic
properties of ordinary and microfine cement grouted sands, in
conjunction with the effect of the grout W/C ratio.
2

preparing cement-based suspension grouts. The W/C ratio of the


suspensions was set equal to 0.6, 0.8, 1.0, 2.0 and 3.0 by weight,
in order to test both stable and unstable suspensions in terms of
bleed capacity. A superplasticizer (patented new generation of
admixture based on polycarboxylate chemistry) at a dosage of
1.4 % by weight of dry cement was used to improve grout
properties. All suspensions were prepared using high speed
mixers. As recommended by the superplasticizer producer,
100

80

MATERIALS AND PROCEDURES

For the purposes of this investigation, a Portland, a Portlandcomposite and a pozzolanic cement (CEM I, CEM II/B-M and
CEM IV/B according to Standard EN 197-1) were used. Each
cement was pulverized to produce three additional cements with
nominal maximum grain sizes (dmax) of 40m, 20m and 10m
and average Blain specific surface values of 567, 720 and
928m2/kg, respectively. Cements with dmax=10m can be
considered as microfine according to Standard EN 12715
(d95<20m and specific surface over 800m2/kg). Also, cements
with dmax=20m have adequately small characteristic grain sizes
to be considered, marginally, as microfine. Typical gradations
of these cements are presented in Figure 1.
All suspensions tested during this investigation were prepared
using potable water since it is considered appropriate for

Cement : CEM II/B-M


nominal dmax

90

Finer by weight (%)

10m
20m
40m

70

100m

60
50
40
30
20
10
0
100

10

Figure 1. Typical cement gradations.

2569

Cement grain size (m)

0.1

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Table 1. Bleed capacity values (%) of all cement grouts


3.0

100m

5-10

17-19

16-39

44-60

60-70

40m

5-7

11-15

10-26

42-55

56-68

20m

N/T

N/T

0-4

25-37

43-49

10m

N/T

N/T

0-2

7-26

38-42

N/T: Not tested


the total amount of cement, 70 % of the water and the
superplasticizer dosage were mixed for 5 min. Then, the rest of
the water was added and mixing continued for another 5 min.
Bleed capacity measurements were conducted for all cement
suspensions used and the results are summarized in Table 1.
According to Standard EN 12715, a suspension is stable
when it has a bleed capacity of not more than 5 % after 120 min
from preparation. It can be observed that a W/C ratio of about
0.6 was required to obtain stable suspension of the coarse
cements (dmax=100m and 40m) while microfine cement
suspensions were stable for a W/C ratio of 1.0
The soils used were clean, uniform, limestone sands with
angular grains and were grouted at a dense (relative density
approximately 90%) and dry state. Two different sand
gradations were used with grain sizes limited between sieve
sizes (ASTM E11) Nos. 10-14 and 14-25 (d15 size of 1.5mm
and 0.8mm, respectively) in order to allow grouting by both the
coarse- and fine-grained suspensions. The angle of internal
friction of the sands was 42.2 and 42.6, respectively.
Laboratory equipment, similar to the arrangement described
in ASTM D4320-84, was used to produce small-size grouted
sand specimens, with a height of 112mm and a diameter of
50mm, ready for testing (Pantazopoulos et al. 2012). Injection
was stopped when the volume of the injected grout was equal to
two void volumes of the sand in the molds. After 24 h, the
specimens were extracted from the split molds and cured in a
humid room for 28 days before testing.
Grouted specimens were tested in unconfined compression at
a displacement rate equal to 0.1%/min. Hydraulic conductivity
tests were performed according to the procedure described by
Head (1986) for permeability testing in a triaxial cell with two
back-pressure systems. Drained triaxial compression tests were
conducted under confining pressures of 100, 200 and 400kPa
and axial strain rate equal to 0.1%/min, without initial saturation
and consolidation. The dynamic properties of the grouted sands
were investigated at confining pressures up to 400kPa by
conducting torsional resonant column tests for a shear strain
range, , of approximately 5*10-5 % to 5*10-2 %. Testing
procedures and interpretation of raw data complied with well
established methods (Pantazopoulos and Atmatzidis 2012). For
comparison, similar tests were conducted on clean sands.
3

Coefficient of permeability, k 20 (cm/s)

0.6

dmax
100m - 40m
20m - 10m

1.00E-03
1.00E-04
1.00E-05
1.00E-06
1.00E-07
1.00E-08
1.00E-09
0.5

1.5

2.5

Water-to-cement ratio, W/C

3.5

1.00E-02

Coefficient of permeability, k 20 (cm/s)

dmax

dmax
1.00E-03

100m - 40m
20m - 10m

1.00E-04
1.00E-05
1.00E-06
1.00E-07
1.00E-08
1.00E-09
0

10

20

30

40

Bleed capacity (%)

50

60

70

Figure 2. Effect of grout W/C ratio and bleed capacity on the


permeability of cement grouted sands.
40

Unconfined compr. strength, q u (MPa)

2.0

1.00E-02

dmax

35

100m - 40m
20m - 10m

30
25
20
15
10
5
0
0.5

1.5

2.5

3.5

Water-to-cement ratio, W/C


40

COEFFICIENT OF PERMEABILITY

The coefficient of permeability values of all grouted sands


tested are presented in Figure 2 with respect to W/C ratio, bleed
capacity and maximum cement grain size of the suspensions.
The coefficient of permeability decreases considerably (by
about 5 orders of magnitude) as the W/C ratio decreases from 3
to 0.6 and attains a value of about 10-7 to10-8cm/s indicating
practically impermeable materials. The permeability of the
grouted sands appears not to be affected by the cement grain
size. Evaluation of the permeability of the grouted sands in
terms of grout bleed capacity indicates a similar trend as with
the W/C, but allows some observations to be made in terms of
the effect of cement grain size. For cement grouts with dmax
equal to 100m and 40m, the coefficient of permeability of the
grouted sands attained values in the range of 10-7 to 10-8cm/s
and 10-3 to 10-4cm/s, for grout bleed capacity ranging from 6%

2570

Unconfined compr. strength, qu (MPa)

0.8

W/C
1.0

d max

35

100m - 40m
20m - 10m

30
25
20
15
10
5
0
0

10

20

30

40

50

60

70

Bleed capacity (%)

Figure 3. Effect of grout W/C ratio and bleed capacity on the


unconfined compression strength of cement grouted sands.

Technical Committee 211 / Comit technique 211

to 30% and from 48% to 68%, respectively. Sands injected with


microfine cement grouts (dmax=20m and 10m) obtained,
generally, higher coefficients of permeability, by half to one
order of magnitude, compared to sands grouted with the coarser
cement suspensions, for similar bleed capacities. This is
reasonable and can be attributed to the increased amount of
coarse-grained cement needed to obtain the same bleed capacity
with suspensions of microfine cements. It should also be noted
that (a) similar coefficient of permeability values (10-7 to 10-8
cm/s) are obtained when injecting with stable or unstable
suspensions for bleed capacity values up to 30% and (b) for
higher bleed capacity values, the coefficient of permeability of
the grouted sand decreases dramatically but remains in the
range of 10-4 to 10-3cm/s.
4

60

dmax
Internal friction angle, ( )

45

40

3 (kPa)
100 - 200 - 400

30
0.5

UNCONFINED COMPRESSION STRENGTH

1.5

2.5

3.5

Water-to-cement ratio, W/C


60

dmax
100m - 40m
20m - 10m

Internal friction angle, ( )

55

50

45

40

35

3 (kPa)
100 - 200 - 400

30
0

10

20

30

40

50

Bleed capacity (%)

60

70

Figure 4. Effect of grout W/C ratio and bleed capacity on the internal
friction angle of cement grouted sands.

SHEAR STRENGTH

The shear strength of the grouted sand specimens is expressed


in terms of internal friction angle and cohesion, by applying the
Mohr-Coulomb failure criterion. As indicated in Figure 4, the
internal friction angle ranged from 40 to 50 and the effect of
W/C ratio, bleed capacity and cement grain size appear to be
insignificant. In general, the internal friction angle of the
grouted sands was up to 5 higher than the value obtained for
clean sands. The cohesion of the grouted sands is strongly
affected both by the W/C ratio and by the bleed capacity of the
grouts. As shown in Figure 5, the cohesion values of the grouted
sands ranged from 600kPa to 1450kPa, from 300kPa to 500kPa
and from 50kPa to 250kPa, for W/C ratios equal to 1, 2 and 3,
respectively. Furthermore, grouted sands injected with stable
grouts (bleed capacity values less than 5%) obtained the highest
cohesion values ranging from 1200kPa to 1450kPa. Increased
bleed capacity values (unstable suspensions) in the range of
15% to 65%, leads to an almost linear decrease of the cohesion
values from 800kPa to 100kPa. The effect of cement grain size
on grouted sand cohesion, as shown in Figure 5, where the
microfine cements exhibit higher values of cohesion than the
coarse-grained cements, by 40% to 150%, is misleading since
the suspensions used had different bleed capacities for the same
W/C ratio. For example, at W/C ratio equal to 1, the microfine
cement suspensions are stable (bleed capacity < 4%) and fill the
sand voids with cement more completely and uniformily than
the coarse cement suspensions with W/C=1 (bleed capacity
>16%).
6

50

35

The results presented in Figure 3 indicate that the unconfined


compression strength of the grouted sands increases
significantly with decreasing W/C ratio of the grouts, as verified
by other research efforts (i.e. Dano et al. 2004) and seems not to
affected by cement grain size. However, the effect of cement
grain size can be clearly demonstrated in terms of grout bleed
capacity. The unconfined compression strength of the grouted
sands is very well correlated with grout bleed capacity of both
the coarse-grained cements (dmax=100 and 40m) and the
microfine cements (dmax=20 and 10m) but, definitely,
microfine cement grouts with the same bleed capacity as cement
grouts yield significantly lower grouted sand strength. As with
permeability, this can be attributed to the increased amount of
coarse-grained cement needed to obtain the same bleed capacity
as microfine cement suspensions.
5

100m - 40m
20m - 10m

55

1600

d max

1400

100m - 40m
20m - 10m

Cohesion, c (kPa)

1200
1000
800
600
400

3 (kPa)
100 - 200 - 400

200
0
0.5

1.5

2.5

3.5

Water-to-cement ratio, W/C


1600

d max

1400

100m - 40m
20m - 10m

Cohesion, c (kPa)

1200
1000
800
600
400

SHEAR MODULUS

Presented in Figure 6 are typical results obtained for the


shear modulus, G, of grouted sands at a confining pressure
equal to 50kPa and shear strain equal to 10-5%. The effect of
confining pressure is not pronounced for the grouted sands
tested (Pantazopoulos and Atmatzidis 2012). As shown in
Figure 6, the shear modulus values decrease, from 4.1GPa to

3 (kPa)
100 - 200 - 400

200
0
0

10

20

30

40

50

60

70

Bleed capacity (%)

Figure 5. Effect of grout W/C ratio and bleed capacity on the cohesion
of cement grouted sands.

2571

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

4.5
4

Shear modulus, G (GPa)

1.5 GPa, with increasing W/C ratio, from 0.6 to 3. The shear
modulus values of the clean sands did not exceed 170MPa,
indicating an improvement up to 25 times by grouting. The
effect of grout bleed capacity on the shear modulus of grouted
sand is clearly depicted in Figure 6, where it can be observed
that above a bleed capacity value of about 30%, the shear
modulus of the grouted sand decreases sharply by about 40%.
Cement grain size seems to have a measurable effect on the
shear modulus values of the grouted sands. For similar bleed
capacity values, the sands grouted with microfine cement grouts
have lower shear modulus values, by 15% to 30%, compared to
sands grouted with coarse-grained cement grouts.

dmax
100m - 40m
20m - 10m

3.5
3
2.5
2
1.5
1

3 = 50 kPa

0.5

= 10 -5 %

0
0.5

1.5

2.5

3.5

Water-to-cement ratio, W/C


4.5

dmax

Shear modulus, G (GPa )

100m - 40m
20m - 10m

3.5
3
2.5
2
1.5
1

3 = 50 kPa

0.5

= 10 -5 %

0
0

10

20

30

40

50

Bleed capacity (%)

60

70

Figure 6. Effect of grout W/C ratio and bleed capacity on the shear
modulus of cement grouted sands.
4

dmax

Damping ratio, D t (%)

20m - 10m

2.5
2
1.5
1

3 = 50 kPa

0.5

= 10 -3 %

0
0.5

1.5

2.5

3.5

Water-to-cement ratio, W/C

Damping ratio, D t (%)

100m - 40m
20m - 10m

ACKNOWLEDGEMENTS

The information reported herein is part of research project


PENED-03ED527, co-financed by the European Social Fund
(75%) and the Greek Ministry of Development (25%).

d max
3.5

CONCLUSIONS

Based on the results obtained and the observations made, the


following major conclusions may be advanced:
1. Bleed capacity is an indicator of grout effectiveness, since it
is representative of the soil void volume filled by cement.
2. The distinction between stable and unstable grouts may not
be an indicator of grout effectiveness since similar effects may
be produced by both stable and unstable grouts.
3. Bleed capacity values correlate very well with some grouted
sand properties (i.e. unconfined compression strength, cohesion)
and not at all with other properties (i.e. damping ratio, internal
friction angle) .

The damping ratio values of the grouted sands are presented in


Figure 7 for a confining pressure equal to 50kPa and shear
strain equal to 10-3%. The effect of shear strain and confining
pressure on the grouted sand damping ratio has been presented
elsewhere (Pantazopoulos and Atmatzidis 2012). In general, the
values obtained ranged from 0.5% to 8.0%, increased with
increasing shear rate (from 5*10-5% to 5*10-2%) and decreased
with increasing confining pressure (from 50kPa to 400kPa). The
grout W/C ratio has a measurable effect on the damping ratio
values of the grouted sands, which have a tendency to increase
with increasing W/C ratio. The effect of grout bleed capacity on
the damping ratio of the grouted sand appears to be less
dominant, mainly for coarse-grained cements. For microfine
cements there is a tendency for the damping ratio of the grouted
sands to increase with increasing bleed capacity of the grouts.
Even though the available data are limited, grouted sands
injected with stable grouts (bleed capacity less than 5%) of
microfine cements indicated damping ratios lower by 50% than
those for grouting with unstable grouts. The damping ratios of
the clean sands (for confining pressure and shear strain equal to
50kPa and 10-3%, respectively) did not exceed 0.5%, indicating
an improvement up to 10 times by grouting.
8

100m - 40m

3.5

DAMPING RATIO

10 REFERENCES

2.5
2
1.5
1

3 = 50 kPa

0.5

= 10 -3 %

0
0

10

20

30

40

Bleed capacity (%)

50

60

70

Figure 7. Effect of grout W/C ratio and bleed capacity on the damping
ratio of cement grouted sands.

Pantazopoulos I.A., Markou I.N., Christodoulou D.N., Droudakis A.I.,


Atmatzidis D.K., Antiohos S.K., Chaniotakis E. 2012.
Development of microfine cement grouts by pulverizing ordinary
cements. Cement and Concrete Composites 34, 593-603.
Head K.H. 1986. Manual of soil laboratory testing. vol.3. Pentech Press
Ltd, London.
Pantazopoulos I.A. and Atmatzidis D.K. 2012. Dynamic properties of
microfine cement grouted sands. Soil Dynamics and Earthquake
Engineering 42, 17-31.
Dano C., Hicher P-Y, Tailliez S. 2004. Engineering properties of
grouted sands, Journal of Geotechnical and Geoenvironmental
Engineering 130, 328-338.

2572

Numerical Analysis to Quantify the Influence of Smear Zone Characteristics on


Preloading Design in Soft Clay
Analyses numriques pour quantifier linfluence des caractristiques de la zone endommage sur
la conception de prchargement dans les argiles molles
Parsa-Pajouh A., Fatahi H., Khabbaz B.

School of Civil and Environmental Engineering, University of Technology Sydney (UTS), Sydney, Australia
ABSTRACT: In this paper, the effects of uncertainties of smear zone characteristics induced by installation of prefabricated vertical
drains on the preloading design are numerically investigated. FLAC 2D finite difference software with additional developed
subroutines has been employed to conduct the numerical simulations. The finite difference analyses have been verified using a case
study. Furthermore, a comprehensive parametric study is conducted to investigate the influence of smear zone permeability and extent
on the model predictions. Results of this study indicate that the assumptive properties for smear zone characteristics may result in
inaccurate predictions of ground deformations and pore water pressures. This may lead to early removal of the surcharge in the
construction process causing excessive post construction settlement. It is recommended to practising engineers to use results of trial
preloading to back calculate the required smear zone characteristics in the early stages of embankment construction to optimize the
design.
RSUM : Dans cet article, les effets des incertitudes des caractristiques de la zone endommage induites par linstallation des
drains verticaux prfabriqus sur la conception du prchargement sont tudis par une mthode numrique. Le logiciel de diffrences
finis FLAC2D avec sous-programmes additionnels a t utilis afin de raliser les simulations numriques. Les analyses de
diffrences finis ont t vrifies laide dune tude de cas. Par ailleurs, une tude paramtrique approfondie est effectue afin
dinvestiguer linfluence de la permabilit de la zone endommag sur les prdictions du modle. Les rsultats de cette tude
montrent que les proprits supposes pour les caractristiques de la zone endommag peuvent entrainer des prdictions incorrectes
de dformations du sol et de pressions interstitielles. Cela peut conduire un retrait prcoce de la surcharge dans le processus de
construction engendrant un tassement post-construction excessive. Il est recommand aux ingnieurs d'utiliser les rsultats de lessai
de prchargement afin de calculer les caractristiques requises de la zone endommage pour optimiser la conception.
KEYWORDS: FLAC, numerical analysis, preloading, smear zone, vertical drain
1

INRODUCTION

Intact Zone

Finding efficient ground improvement techniques to modify the


soft soil properties, considering the project time limitation and
the construction cost has been a continuous challenge for the
construction companies. Various ground improvement methods
have been proposed to improve the strength properties of the
soft soil. In the last two decades, employing prefabricated
vertical drain (PVD) assisted preloading has been recognised as
a very efficient ground improvement method for sites with deep
soft soil deposits (Holtz et al. 1991; Shang et al. 1998;
Indraratna et al. 2005). Installation of the prefabricated vertical
drains using mandrel, induces disturbance of the soil
surrounding the drain, resulting in a smear zone of reduced
permeability adversely affecting the consolidation process.
Predicting the soil behaviour surrounding the drain requires an
accurate estimation of the smear zone properties. Generally, two
major parameters are proposed to characterise the smear zone;
the permeability (ks), and the extent (rs) of the smear zone.
Figure 1 illustrates the cross section of prefabricated vertical
drains surrounded by smear zone, which are installed in
rectangular pattern. Determining both the smear zone extent
and its permeability is a challenging task. According to
literature, very diverse values are reported for the permeability
ratio (kh/ks) and extent ratio (rs/rm), which are illustrated in
Figure 2. The proposed range shows that the extent of the smear
zone (rs) may vary between 1.6 to 7 times of the drain radius
(rw) or, 1.0 to 6 times of mandrel equivalent diameter (rm). The
proposed range for the permeability ratio (kh/ks) is 1.3 to 10,
where kh is the horizontal permeability of the intact soil.

Intact Zone

Smear
Zone

kh

rs

kv

rw

Vertical
drain

S R

kh

rs
Vertical
drain

rw

kh
ks k
v

Intact Zone

Intact Zone

Smear
Zone

Smear
Zone

rs

kv

Smear
Zone

ks

rw

ks
Vertical
drain

rs
Vertical
drain

rw

ks

kh
kv

Figure 1. Cross section of PVD surrounding by smear zone


Extent ratio (rs/r m)
Barron (1948)
Casagrande and Poulos (1969)
Holtz and Holm (1973)
Akagi (1976)
Hansbo (1981)
Hansbo et al.(1981)
Jamiolkowski et al. (1983)
Bergado et al. (1991)
Onoue et al. (1991)
Bergado et al. (1993)
Almedia et al. (1993)
Hansbo (1994)
Mesri et al. (1994)
Hansbo (1997)
Indraratna and Redana (1998)
Chai and Miura (1999)
Eriksson et al. (2000)
Sharma and Xiao (2000)
Hird and Moseley (2000)
Bo et al. (2003)
Indraratna et al. (2005b)
Indraratna et al. (2005c)
Sathananthan and Inraratna (2006)
Sathananthan et al. (2008)
Ghandeharioon et al. (2009)
Kim et al. (2010)
Tran-Nguyen and Edil (2011)
Ghandeharioon et al. (2012)

Extent ratio (r s/r m)

Lower bound

Barron (1948)
Upper bound Casagrande and Poulos (1969)
Holtz and Holm (1973)
Akagi (1976)
Hansbo (1981)
Hansbo et al.(1981)
Jamiolkowski et al. (1983)
Bergado et al. (1991)
Onoue et al. (1991)
Bergado et al. (1993)
Almedia et al. (1993)
Hansbo (1994)
Mesri et al. (1994)
Hansbo (1997)
Indraratna and Redana (1998)
Chai and Miura (1999)
Eriksson et al. (2000)
Sharma and Xiao (2000)
Hird and Moseley (2000)
Bo et al. (2003)
Indraratna et al. (2005b)
Indraratna et al. (2005c)
Sathananthan and Inraratna
Sathananthan et al. (2008)
Ghandeharioon et al. (2009)
Kim et al. (2010)
Tran-Nguyen and Edil (2011)
Ghandeharioon et al. (2012)
4 5 6 7 8
0

Applied method: Back-Analysis

Experimental

Analytical

Lower bound
Upper bound

FEM

Assumed

Figure 2. Proposed values for smear zone characteristics

It can be observed that wide ranges are proposed for kh/ks and
rs/rm and there is no definite method to predict these parameters
precisely to be used by practising engineers. The assumptive
properties for smear zone characteristics may result in

2573

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

inaccurate predictions of the ground behaviour. This can lead to


early removal of surcharge in construction process resulting in
excessive post construction settlement. Therefore, it is essential
to study the influence of the uncertainties in the smear zone size
and its permeability on the preloading design to improve the
performance of soft deposits. Thus, a numerical code using
FLAC 2D has been developed in this study to investigate the
uncertainties of PVD smear zone characteristics on the
preloading design which can be used to back calculate smear
zone characteristics for actual preloading projects.
2

Embankment Height (m)

In the present study, FLAC 2D v6.0 has been employed to


model the PVD assisted preloading process focusing on smear
zone uncertainties. Required new subroutines have been written
using the built-in programming language FISH (FLACish) to
tailor analyses to suit specific needs for the parametric study,
giving the following unique advantages to the developed code
for this study; (i) automatic mesh generation process by entering
the required parameters to modify the grid pattern inside and
outside the smear zone; (ii) ability to change different
parameters such as the model dimensions, vertical drain
properties, subsoil profile, smear zone characteristics and
preloading conditions; (iii) the option to define the exact
location of desired points to generate and plot any future history
graphs; and (iv) automatic solving process based on the
modified input data. Chittagong Sea Port in Bangladesh with
3.0 m high embankment on 9 m deep soft clay, has been
selected for the numerical simulations and verification of the
developed code and subroutines.
According to Dhar et al. (2011), a container yard has been
constructed at Chittagong Port, the largest sea port in
Bangladesh, for handling loaded containers. The site is located
on the bank of Karnafully river beside the Bay of Bengal in the
Indian Ocean. The yard covers an area of 60,700 m2 and was
designed to support a container load producing a contact
pressure of approximately 56 kPa. Geotechnical investigations
revealed the presence of a soft to very soft clayey silt/silty clay
deposit with a thickness of approximately 7 m (Figure 3).
Preloading with prefabricated vertical drains was adopted to
preconsolidate the compressible soft deposits, which was
followed by the field monitoring. Vertical drains were installed
down to the depth of approximately 9 m below the ground level
in square patter to cover the full depth of the soft clay. A
surcharge load consisting of 3.0 m high fill of sand was placed
for preloading. Surcharge material was placed in two layers of
approximately equal thickness. The sides of the surcharge load
were kept vertical along the boundaries of the area using sand
bags and brick stacks. Figure 3 shows a profile detailing the
ground improvement work schematically. In addition, Figure 4
shows the construction history of the embankment.
Surcharge Material

LL=45
PI=18
Gs=2.74
Silty
sand

0m

10

20

30

40
50
60
Time (days)

70

80

90

FLAC 2D numerical code incorporating modified Cam-Clay


constitutive soil model has been employed to simulate
Chittagong Port preloading process applying plane strain
conditions. The zero excess pore water pressure has been
considered along the vertical drains and the ground surface
boundary to model the PVD and surface drainage, respectively.
Adopted soil properties in the numerical analysis are
summerised in Table 1.
Table 1. Adopted soil properties (after Dhar et al. 2011)
Layer

Soil
type

s
kN/m3

kh
10-9m/s

kh/
kv

Clayey
Silt

Soft
soil

0.94

0.13

0.026

0.3

1.28

14.0

2.31

1.5

The equivalent plane-strain permeability (khp) proposed by


Indraratna and Redana (2000) has been used in the numerical
analysis.
(khp/kh) = 0.67 / [(ln(n)-0.75]

(1)

(ksp/khp) = / [(khp/kh) [(ln(n/s)+(kh/ks) ln(s)-0.75]-]

(2)

= 2(n-s) / [3(n-1)n ]

(3)

= [2(s-1) / (n-1)n ] * [n(n -s-1) +1/3 (s +s+1)]

(4)

where, kh and khp are axisymmetric and plane-strain horizontal


permeability values of intact zone respectively, ks and ksp are
axisymmetric and plane-strain permeability values of smear
zone, respectively, and are geometric coefficients, n is the
spacing ratio equal to B/bw where B and bw are equivalent planestrain radius of the influence zone and radius of the drain
respectively, and s=rs/rw. The value of kh needs to be
determined first (laboratory or field), then khp can be calculated
using Equation (1). When khp is known, ksp can be obtained
from Equation (2). The discretised plane- strain finite-difference
mesh composed of quadrilateral elements is shown in Figure 5,
where only half of the trial embankment is considered by
exploiting symmetry.
Settlement
gauge (G1)

Intact zone
Smear zone

60 kPa

PVD
20 m

Sand bags

t=20 kN/m 3
10 m

3.0m
Clayey
silt/Silty
clay
(CL/ML)

Commencement of recording
settlement plate readings

1
0

2.1. Case Study: Chittagong Sea Port in Bangladesh

PVDs

Figure 4. Construction history (Chittagong Port embankment)

NUMERICAL MODELLING

CL

G1

4m
9m

G2

1m
7m
10m

Figure 3. Cross section of constructed embankment

1m

Figure 5. Sample of mesh grid pattern for Chittagong Port embankment


considering the smear

Numerical results are compared with the field measurements in


Figure 6. According to Figure 6, FLAC predictions are in a
good agreement with the field measurements considering
kh/ks=2 and rs/rm=3. The primary consolidation settlement is

2574

Technical Committee 211 / Comit technique 211

predicted to be approximately 258 mm. As illustrated in Figures


4 and 6, the field settlement is measured immediately after
placing the surcharge to the full height of 3 m (after 12 days).
300
=90%

Settlement (mm)

250

(Kh /Ks=2 & rs/rm=3)

200

Field measurements (Dhar et al. 2011)

150

Numerical predictions (Point G1)

100
50
0
0

10

20

30
40
Time (Days)

50

60

70

Figure 6.Comparison of numerical results with filed data

PARAMETRIC STUDY AND DISCUSSION

Parametric studies have been conducted to investigate the


influence of the smear zone characteristics on the preloading
design simulating Chittagong Port case study with the details
presented in the previous section. For this purpose, kh/ks
(permeability ratio) and rs/rm (extent ratio) have been changed
from 2 to 5. Figure 7 illustrates the parametric study results for
settlement-time relationships.
300

(a)

Settlement (mm)

250

rs/rm=2

= 90%

200

t = 53.0

t = 47.5

100

t = 41.0

t = 34.0

150
Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

50
0
20

80

(b)

250

100

rs/r m=3
= 90%

200

Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

t = 57.0

50

t = 48.0

100

t = 42.0

150

t = 35.0

Settlement (mm)

40
60
Time (Days)

Excess pore pressure (kPa)

0
300

zone properties are kh/ks=5 and rs/rm=5, the required time would
be the maximum and equal to 67 days, which is approximately
twice longer than the minimum (see Figure 7d). According to
the settlement curves in Figure 7, the influence of smear zone
permeability variations is more critical when the smear zone
extent ratio is larger. For instance the required time to obtain
90% degree of consolidation has been increased by 56% (from
34 days to 53 days) changing the permeability ratio from 2 to 5
considering the extent ratio equal to 2, while this boost is 80%
(from 37 days to 67 days) for extent ratio of 5.
The general trend in Figures 7(a)-7(d) shows that changing
the permeability ratio in a smaller range results in large
variations of the required time to obtain 90% degree of
consolidation considering a constant extent ratio. According to
Figure 7(a), the consolidation time is increased by 23% by
varying the permeability ratio from 2 to 3, while this change is
17% and 12% when the permeability ratio is changed from 3 to
4 and 4 to 5, respectively.
Figure 8 illustrates the numerical parametric study results
investigating the influence of the smear zone properties on the
excess pore water pressure (EPWP) dissipation. Graphs are
plotted for point G2 located at the depth of 4 m (see Figure 3).
Figure 8 confirms that increasing the permeability and extent
ratios prolongs the pore water pressure dissipation process
considerably. According to Figure 8, the permeability ratio is
more critical parameter than the extent ratio, although the
influence of extent ratio variation on the consolidation time can
not be neglected. For example, according to Figure 8b, there is
160% difference between the predicted excess pore pressure
values after 34 days (90% of the field degree of consolidation)
for kh/ks=2 (EPWP=13 kPa) and kh/ks=5 (EPWP=34 kPa), while
keeping rs/rm=3.

0
20

80

(c)

250

t = 63.0

t = 53.0

t = 45.0

150
Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

30
20
10
20

40

60
Time (Days)

80

(b)

70

100

r s/rm=3
Full height of
embankment reached

60
50

Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

40
30
20

= 90%

10
0

0
20

40

60
Time (Days)

80

(d)

250

100

80

rs/rm=5

= 90%

Excess pore pressure (kPa)

0
300

200

t = 47.0

50

t = 67.0

100

t = 60.0

150

t = 37.0

Settlement (mm)

Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

40

200

50

rs/rm=2

Full height of
embankment reached

50

80

rs/rm=4

100

(a)

60

100

= 90%

t = 36.0

Settlement (mm)

40
60
Time (Days)

Excess pore pressure (kPa)

300

70

Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

10

20

30

40

50
60
Time (Days)

(c)

70

70

80

90

100

rs/rm=4
Full height of
embankment reached

60
50

Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

40
30
20
10
0

0
0

20

40

60
Time (Days)

80

100

80

According to Figure 7, the settlement curves are converged to a


unique value of approximately 258 mm, which is the primary
consolidation settlement. The required time to obtain 90% of
primary consolidation settlement (232 mm) has been considered
to investigate the effect of smear zone properties on
consolidation process. According to Figure 7a, the minimum
time of 34 days is needed to achieve 90% degree of
consolidation, considering kh/ks=2 and rs/rm=2. When smear

Excess pore pressure (kPa)

Figure 7. Parametric study results for Chittagong port case history at


point G1; (a) rs/rm=2; (b) rs/rm=3; (c) rs/rm=4; and (d) rs/rm=5

20

40

60
Time (Days)

(d)

70

80

100

rs/rm=5
Full height of
embankment reached

60
50

Kh/Ks=2
Kh/Ks=3
Kh/Ks=4
Kh/Ks=5

40
30
20
10
0
0

20

40

60
Time (Days)

80

100

Figure 8. Effect of smear zone properties on excess pore water pressure


dissipation for Chittagong port case history at point G2

2575

The required time to obtain 90% degree of consolidation for


different smear zone properties is illustrated in Figure 9 using
parametric study results, which presents a better interpretation
of the effects of the smear zone properties on consolidation
time. According to Figure 9, the consolidation time significantly
depends on the smear zone permeability and extent. For
example, assuming rs/rm=2, for the case with kh/ks=2 and
kh/ks=5, the required times to obtain 90% degree of
consolidation are approximately 33 days and 53 days,
respectively, indicating 60% difference. It can be noted that the
difference is more significant for larger values of rs/rm.
rs/rm=2

Permeability ratio (k h/ks)

5.0

rs/r m=3

r s/r m=4

rs/rm=5

3.0

S4 S3

1.0
32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70
Time (Days)

Figure 9. Predicted time to obtain 90% degree of consolidation

Figure 9 clearly indicates that the smear zone extent ratio (rs/rm)
is an important parameter influencing the consolidation time
and cannot be neglected. Varying rs/rm in the range of 2 to 5,
assuming kh/ks as a constant parameter can influence the
required consolidation time by more than 25%. Combined
effects of uncertainties in the smear zone extent and
permeability will result in momentous changes of consolidation
time. Results presented in Figure 9 indicate that the influence of
uncertainties in rs/rm becomes more important when
permeability of smear zone decreases.
According to the back calculation results presented in
Figure 6, the predicted settlement curve is in the best agreement
with the field measurements considering smear zone properties
of kh/ks=2 and rs/rm=3. The required time to obtain 90% degree
of consolidation for this condition is equal to 34 days, which is
highlighted as point S2 in Figure 9. A vertical line is plotted
from t90%= 34 days, which intersects the set of lines at points
S1, S2, S3 and S4. Smear zone properties at these points are
summarised in Table 2.
Table 2. Back calculated smear zone properties to achieve t 90%= 34 days
S1

S2

S3

S4

kh/ks

2.10

2.0

1.85

1.75

rs/rm

Numerical analyses applying developed FLAC code have been


conducted to compare the settlement and excess pore water
pressure variations against the consolidation time. Different
combinations of smear zone extent and permeability may result
in the same t90%= 34 days and predictions are presented in
Figure 10.
300

(a)

Settlement (mm)

250
200
150

Kh/Ks=2.1 & rs/rm=2 (S1)

100

Kh/Ks=2 & rs/rm=3 (S2)


Kh/Ks=1.85 & rs/rm=4 (S3)

50

Kh/Ks=1.75 & rs/rm=5 (S4)

0
0

20

40
Time (Days)

60

80

Kh/Ks=2.10 & rs/rm=2 (S1)


Kh/Ks=2 & rs/rm=3 (S2)
Kh/Ks=1.85 & rs/rm=4 (S3)
Kh/Ks=1.75 & rs/rm=5 (S4)

50
40
30
20
10
0
0

20

40
Time (Days)

60

80

Figure 10. FLAC analysis results for points in Table 2 (a) Settlement
variation; (b) Excess pore water pressure dissipation

S2 S1

Point

(b)

60

Figure 10 shows that the curves for the settlement variations


and the excess pore water pressure dissipations with time follow
the same trend for points S1, S2, S3 and S4. Therefore, smear
zone properties of any of these points can be adopted for the
practical design purposes.

4.0

2.0

Excess pore water pressure (kPa)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

CONCLUSIONS

Preloading time during consolidation process can significantly


be affected by formation of the smear zone in the vicinity of the
prefabricated vertical drains (PVDs). Smear zone is a reduced
permeability area induced by mandrel insertion that halts the
consolidation process. Available literature proposes a wide
range for the smear zone extent and permeability and yet there
is no definite prediction method that can be used to estimate the
extent of smear zone and its permeability to be used in the
design procedure. In this study, numerical analyses have been
employed to investigate the effects of uncertainties of smear
zone characteristics on the preloading design. FLAC 2D
software has been employed to develop a numerical code
assisting with the parametric study and back calculating smear
zone properties. The verification exercise on Chittagong port
case history confirms the validity of the developed numerical
code. According to the parametric study results the properties of
the smear zone have key roles on the required consolidation
time to achieve a certain soil strength and stiffness satisfying
both bearing capacity and settlement design criteria. Therefore,
accurate estimation of the properties of smear zone based on the
soil type and the installation method is vital for the ground
improvement projects adopting PVD assisted preloading.
Results of this study indicate that assumptive properties for
smear zone characteristics may result in inaccurate predictions
of ground deformations and pore water pressures. This can lead
to early removal of surcharge in construction process resulting
excessive post construction settlement. Thus, it is recommended
to practising geotechnical engineers to back calculate the smear
zone properties using a trial construction similar to the future
construction procedure.
5

REFERENCES

Dhar, A. S., Siddique, A. and Ameen, S. F. 2011. Ground improvement


using pre-loading with prefabricated vertical drains. International
Journal of Geoengineering Case Histories, 2(2), 86-104.
Holtz, R.D., Jamiolkowski, M.B., Lancellotta, R. and Pedroni, R. 1991.
Prefabricated vertical drains: design and performance. CIRIA,
Butterworth-Heinemann, London.
Indraratna, B. and Redana, I. W. 2000. Numerical modeling of vertical
drains with smear and well resistance installed in soft clay.
Canadian Geotechnical Journal, 37(1), 132-145.
Indraratna, B., Rujikiatkamjorn, C. and Sathananthan, I. 2005.
Analytical and numerical solutions for a single vertical drain
including the effects of vacuum preloading. Canadian Geotechnical
Journal, 42(4), 994-1014.
Shang, J.Q., Tang, M. and Miao, Z. 1998. Vacuum preloading
consolidation of reclaimed land: a case study. Canadian Geotechnical
Journal, 35(5), 740-749.

2576

Construction of virtual sites for reliability-based design


Construction de sites virtuels des fins de conception fiabiliste
Phoon K.K.

National University of Singapore, Singapore

Ching J.

National Taiwan University, Chinese Taipei

ABSTRACT: This paper presents the construction of virtual sites using multivariate normal distributions calibrated from actual soil
property databases. By doing so, the actual magnitude of uncertainty reduction from conducting better/more soil tests can be
estimated realistically, rather than theoretically.
RSUM: Cet article prsente la construction de "sites virtuels" en utilisant des distributions normales plusieurs variables calibres
partir de bases de donnes de proprits de sols rels. Par cette mthode, la rduction relle de l'incertitude que l'on peut obtenir en
augmentant le nombre et/ou la qualit des essais de sol peut tre estime de manire raliste, et non plus seulement thorique.
KEYWORDS: virtual site; uncertainties; soil properties; correlation; site investigation; reliability-based design.
1

INTRODUCTION

This paper presents the concept of a virtual site; the purpose


is to emulate site investigation efforts as realistically as
possible. It is not possible to emulate every aspect of a real site
at present. In this paper, the scope is to reproduce the
information content arising from a typical mix of laboratory and
field tests conducted in a site for the purpose of estimating a
design undrained shear strength (su) for clays and friction angle
() for sands. The critical feature here is the consistent and
realistic coupling of different test data, which is achieved using
multivariate normal distributions. Data from different tests will
be correlated, because they are measuring the same mass of soil,
although they could be measuring different aspects of soil
behavior under different boundary conditions and over different
volumes. The purpose of developing a virtual site is not to
replace actual site investigation. The purpose is to quantify the
uncertainty reduction in su and by incorporating the test
results from better and/or more tests.
The idea of simulating a virtual site is not new. For
example, Jaksa et al. (2005) and Goldsworthy et al. (2007) used
three dimensional random fields and Monte Carlo simulation to
simulate the spatially variable elastic modulus of a virtual
site. Each spatially variable realization constitutes a plausible
full information scenario. Site investigation is then carried out
numerically by sampling the continuous random field at discrete
locations. The site investigation data so obtained constitute the
typical partial information scenario commonly encountered in
practice. The goal of these studies was to quantify the
difference in the designs based on these full and partial
information scenarios. In this paper, the virtual site simulation
is based on multivariate normal distributions that couple soil
parameters such as su, overconsolidation ratio, standard
penetration test N-value, cone tip resistance, and Atterberg
limits. The distinct features of this paper are: (a) a more
realistic bag of multivariate information containing both
laboratory and field data and (b) the probability model is
constructed from an actual database of clays and sands. These
features are critical to the objective of this paper, which is to
quantify the uncertainty reduction in su and by incorporating
the test results from better and/or more tests. This objective is
only achievable if the information contained in the virtual site is
comparable to that contained in a real site, not merely pertaining

2577

to a single laboratory/field parameter, but to a group of


parameters that are correlated in a realistic way. By doing so, it
is possible to evaluate the actual merits of reliability-based
design approximately, rather than elaborate on the theoretical
merits widely discussed in previous studies. This paper
summarizes the current development of such virtual sites.
2

MULTIVARIATE GEOTECHNICAL DATA

Multivariate information is usually available in a typical site


investigation. For instance, when undisturbed samples are
extracted for oedometer and triaxial tests, SPT and/or piezocone
test (CPTU) may be conducted in close proximity. Moreover,
data sources such as the unit weight, plastic limit (PL), liquid
limit (LL), and liquidity index (LI) are commonly determined
from relatively simple laboratory tests on disturbed samples.
These data could be correlated, and these correlations can be
exploited to reduce the coefficient of variation of a design
parameter. The impact on RBD is obvious. This section
presents statistical characterization of multivariate geotechnical
data.
Most soil parameters are not normally distributed, because
they are positive valued. Let Y denote a non-normally
distributed soil parameter.
One well known cumulative
distribution function (CDF) transform approach can be applied
to convert Y into a standard normal variable X: X = -1[F(Y)],
where (.) is the CDF of a standard normal random variable,
and F(.) is the CDF of Y. A set of multivariate soil parameters
Y = (Y1, Y2, Yn) can be transformed into X = (X1, X2, Xn).
By definition, X1, X2, Xn are individually standard normal
random variables. It is crucial to note here that collectively (X1,
X2, Xn) does not necessarily follow a multivariate normal
distribution even if each component is normally distributed.
Even so, recent studies by Ching et al. (2010) and Ching and
Phoon (2012a) showed that the multivariate normal distribution
is an acceptable approximation for selected parameters of clays,
and Ching et al. (2012b) arrived at the same observation for
selected parameters of sands.
The multivariate normal probability density function for X =
(X1, X2, Xn) can be defined uniquely by a correlation matrix:

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

f (X) C

1
2

(2 ) 2 e

1
X ' C-1 X
2

(1)

where C is the correlation matrix. For n = 3, the correlation


matrix is given by:

1 12

C
1
12
13 23

13
23

(2)

between Xi and Xj (not equal to the correlation between the


original physical variable Yi and Yj). It is clear that the full
multivariate dependency structure of a normal random vector
only depends on a correlation matrix (C) containing bivariate
correlations between all possible pairs of components, namely
X1 and X2, X1 and X3, and X2 and X3. It is not necessary to
measure X1, X2, and X3 simultaneously. The practical advantage
of capturing multivariate dependencies in any dimension (i.e.,
any number of random variables) using only bivariate
dependency information is obvious.
It is simple to obtain realizations of independent standard
normal random variables U = (U1, U2, U3) using library
functions in many softwares. Realizations of correlated
standard normal random variables X = (X1, X2, X3) can be
obtained using X = LU, in which L is the lower triangular
Cholesky factor satisfying C = LL. Finally, each soil parameter
is obtained using Yi = F-1[(Xi)].
2.1

Y1 = LI
Y2 = su
Y3 = sure
Y4 = p
Y 5 = v

Lognormal
Lognormal
Lognormal
Lognormal
Lognormal

1.25
31.01kN/m2
2.51kN/m2
105.82kN/m2
66.63kN/m2

0.49
0.95
1.52
0.98
0.80

ln(Yi), i
0.122
3.051
0.226
4.311
3.891

ln(Yi), i
0.459
0.898
1.191
0.835
0.823

Table 2. Correlation matrix C for (X1, X2, X5) for the five
selected parameters of structured clays (Source: Ching & Phoon
2012a).
X1 (LI)
X2 (su)
X3 (sure)
X4 (p)
X5 (v)

X1 (LI)
1.000
-0.083
-0.824
-0.176
0.280

X2 (su)
-0.083
1.000
0.276
0.915
0.801

X3 (sure)
-0.824
0.276
1.000
0.365
0.453

X4 (p)
-0.176
0.915
0.365
1.000
0.850

X5 (v)
0.280
0.801
0.453
0.850
1.000

Complete multivariate information (structured clays)

A multivariate database of Y1 = LI (liquidity index), Y2 = su, Y3


= sure (remolded undrained shear strength), Y4 = p
(preconsolidation stress), and Y5 = v (effective vertical stress)
is complied in Ching & Phoon (2012a). There are 345 data
points of structured clays from 37 sites worldwide, covering a
wide range of sensitivity, LI, and clay types, with simultaneous
knowledge of (Y1,Y2, Y5). The OCR values of the data
points are generally small, mostly less than 4. Fissured and
organic clays are mostly left out of the database. Because su
values depend on stress state, strain rate, sampling disturbance,
etc., all su values are converted into mobilized su values
following the recommendations made by Mesri and Huvaj
(1997). The marginal probability density functions (PDF) for
(Y1,Y2, Y5) and their statistics (mean of Yi = i, COV of Yi =
Vi, mean of ln(Yi) = i, standard deviation of ln(Yi) = i) are
summarized in Table 1.
For lognormal Y, the CDF transform is:

X i ln Yi i i

(3)

The transformed (X1, X2, , X5) are individually standard


normal random variables. The correlation matrix C for (X1, X2,
X5) is shown in Table 2, and (X1, X2, X5) is assumed to be
multivariate normal with the correlation matrix listed in the
table.
The multivariate normal distribution is employed to
simulate samples of (LI, su, sure, p, v), shown in Figure 1
together with the calibration database. Not only the correlations
among the original random variables (LI, su, sure, p, v) are
shown but the correlations among their derived (normalized)
quantities, including St = su/sure, OCR = p/v, su/v, are also
shown.
The multivariate normal distribution performs
adequately, as the simulated samples closely mimic the
correlation behaviors of the calibration database, even for those
with nonlinear trends, e.g. LI-sure and LI-St correlations.
Table 1. Distributions and statistics of (Y1, Y2, Y5) for
structured clays (Source: Ching & Phoon 2012a).
Distribution

Mean

COV

Mean of

stdev of

2578

Figure 1. Comparisons between the calibration database and the


simulated data points (Source: Ching & Phoon 2012a).
2.2

Incomplete multivariate information (unstructured clays)

Ching et al. (2010) presented another clay database containing


four soil parameters: Y1 = OCR, Y2 = su from CIUC test, Y3 =
qT - v (net cone resistance), and Y4 = N60 (SPT N corrected for
energy efficiency). The range of OCR of this database is
wider from 1 to 50. However, only bivariate data on (Y1, Y2)
= (OCR, su), (Y3, Y2) = (qT - v, su), and (Y4, Y2) = (N60, su) are
available. Bivariate data on (Y1, Y3) = (OCR, qT - v), (Y1, Y4)
= (OCR, N60), and (Y3, Y4) = (qT - v, N60) are missing, i.e., the
bivariate correlations ij are only partially known. Given that
complete bivariate information is not available, it is not possible
to apply the aforementioned CDF transform approach directly.
It is accurate to say that although it is common to measure more
than two soil parameters in a site investigation, it is uncommon
to establish correlations between all possible pairs of soil
parameters.
To deal with this difficulty of incomplete bivariate
correlations, Ching et al. (2010) constructed a multivariate
normal distribution using a Bayes net model which prescribed a
dependency structure based on some postulated but reasonable
conditional relationships between the soil parameters. They
considered Y1 = OCR as a given number and the remaining soil
parameters (Y2, Y3, Y4) are lognormally distributed random
variables. Hence, ln(Y2) = ln(su) = 2+ 2X2, ln(Y3) = ln(qT - v)
= 3+ 3X3, and ln(Y4) = ln(N60) = 4+ 4X4, in which Xi are
standard normal random variables. The simulation of (Y1, Y2,
Y3, Y4) starts from OCR. The undrained shear strength, Y2, is
next simulated using this OCR sample and the SHANSEP
model (Ladd and Foott 1974):

Technical Committee 211 / Comit technique 211

ln s u 0.64 ln OCR ln v 0 0.874 0.237U1

(4)

where 0.237 is the standard deviation of the transformation


uncertainty, and U1 is standard normal. The third step is to
simulate N60 and qT - v using the su sample:

X1 (cv)
X2 (IR)
X3 (p)
X4 (qc1)
X5 [(N1)60]

X1 (cv)
1.000
0.000
0.642
0.491
0.536

X2 (IR)
0.000
1.000
0.642
0.491
0.536

X3 (p)
0.642
0.642
1.000
0.764
0.835

X4 (qc1)
0.491
0.491
0.764
1.000
0.638

X5 [(N1)60]
0.536
0.536
0.835
0.638
1.000

ln

N 60 1.633ln s u 0.403ln v 0 3.845 0.456U 2

(5a)

ln q T
v ln s u 2.54 0.34U 3

2.4

(5b)

The undrained shear strength (su) of a clay is not a constant. In


particular, su of a clay evaluated by different test procedures are
different because these tests may have different stress states,
stress histories, degrees of sampling disturbance, and strain
rates. Ching & Phoon (2013) constructs the multivariate normal
distribution of the su values from seven su tests (CIUC, CK0UC,
CK0UE, DSS, VST, UU, UC) based on a large clay database
consisting data points from 146 studies. Many su data points are
associated with a known test mode (6310 points), a known OCR
(4584 points), and a known plasticity index (PI) (4541 points).
The geographical regions cover Australia, Austria, Brazil,
Canada, China, England, Finland, France, Germany, Hong
Kong, Iraq, Italy, Japan, Korea, Malaysia, Mexico, New
Zealand, Norway, Northern Ireland, Poland, Singapore, South
African, Spain, Sweden, Thailand, Taiwan, United Kingdom,
United States, and Venezuela. The clay properties cover a wide
range of OCR (mostly 1~10, few studies OCR > 10, but nearly
all studies are with OCR < 50) and a wide range of sensitivity St
(sites with St = 1~ tens or hundreds are fairly typical).
An important step for the construction of the multivariate
distribution is to convert all su data points in the database into
the following standardized form:

where 0.456 and 0.34 are the standard deviations of the


transformation uncertainties, and U2 and U3 are standard
normal. Figure 2 shows the correlation plots for the simulated
{OCR, su, N60, qT - v} for a case where OCR is uniformly
distributed over [5, 24].

Figure 2. Correlation plots for {OCR, su, N60, qT - v} samples.


Based on the results of Ching et al. (2010), Phoon et al.
(2012) further assumed OCR to be lognormal with a reasonable
COV = 0.25. Under this assumption, they showed that the
underlying standard normal variables (X1, X2, X3, X4) have the
correlation matrix shown in Table 3. The correlation matrix in
Table 3 should be suitable for unstructured clays covering a
fairly wide range of OCR.
Table 3. Correlation matrix C for (X1, X2, X3, X4) for the four
selected parameters of unstructured clays (Source: Phoon et al.
2012).
X1 (OCR)
X2 (su)
X3 (qT - v)
X4 (N60)

2.3

X1 (OCR)
1.000
0.554
0.355
0.395

X2 (su)
0.554
1.000
0.642
0.714

X3 (qT - v)
0.355
0.642
1.000
0.458

X4 (N60)
0.395
0.714
0.458
1.000

Incomplete multivariate information (clean sands)

Ching et al. (2012b) presented a study that is very similar to


Ching et al. (2010) but for clean sands. The study was based on
a database containing five selected parameters of normally
consolidated clean sands: Y1 = cv (critical state friction angle),
Y2 = IR (dilatancy index, see Bolton 1986), Y3 = p (peak secant
friction angle), Y4 = (qc/Pa)/(v/Pa)0.5 = qc1 (corrected cone
resistance), and Y5 = (N1)60 (SPT N corrected for energy
efficiency and overburden stress). They considered Y1 = cv
and Y2 = IR as given numbers and the remaining soil parameters
(Y3, Y4, Y5) are random variables: Y3 is normal, while Y4 and
Y5 are lognormal. Hence, Y3 = p = 3+ 3V3X3, ln(Y4) =
ln(qc1) = 4+ 4X4, and ln(Y5) = ln[(N1)60] = 5+ 5X5, in which
Xi are standard normal random variables. If we further assume
cv and IR are normal with reasonable standard deviations of 3o
and 1o, respectively, i.e., Y1 = cv = 1+ 3X1 and Y2 = IR = 2+
X2, and also assume independence between cv and IR, it can be
shown that the underlying standard normal variables (X1, X2,
X3, X4, X5) has the correlation matrix shown in Table 4. The
correlation matrix in Table 4 should be suitable for normally
consolidated clean sands.
Table 4. Correlation matrix C for (X1, X2, X3, X4, X5) for the
five selected parameters of clean sands (Source: Ching et al.
2012b).

Undrained shear strengths under various test procedures

s u,NC,1%,PI20 v = s u v bOCR c rate d PI

(6)

where su,NC,1%,P20 is the undrained shear strength of a NC clay


with PI = 20 subjected to a 1% per hour strain rate; bOCR, crate,
and dPI are modifier factors that adjust the reference normalized
undrained shear strength for overconsolidation ratio, strain rate,
and plasticity. Table 5 shows these factors (Ching et al. 2013;
Ching & Phoon 2013). The standardized su,NC,1%,P20/v is be
denoted by Ytest mode index. The test mode indices are respectively
1, 2, 3, 4, 5, 6, and 7 for CIUC, CAUC, CAUE, DSS, FV (field
vane), UU, and UC. Hence, there are seven random variables
(Y1, Y2 Y7). Table 6 shows the statistics of Yi. The Y data
points for each test mode are roughly lognormally distributed,
i.e., Xi = [ln(Yi)-i]/i is roughly standard normal. Given a test
mode i, the scatter in the Yi data points, quantified by the COV
in Table 6, may be due to measurement errors in su and global
inherent variability in su (su from different geographic locales)
as well as the transformation uncertainties associated with the
standardization steps for PI, strain rate, and OCR.
The Yi data points are converted to standard normal
variables Xi = [ln(Yi)-i]/i. Table 7 shows the correlation
matrix C for (X1, X2, , X7). The estimated correlation
coefficients ij are quite sensible. The four triaxial compression
(TC) test modes (X1, X2, X6, X7) seem mutually highly
correlated (ij > 0.8), with the exception of (X6, X7) having ij =
0.59. The CAUE test mode (X3) has weak correlation with TC
test modes (ij < 0.5), probably because it imposes a different
stress state from TC tests. The correlation coefficients between
FV and TC are relatively weak as well (ij 0.63). Such
relatively low correlation between FV and TC may be due to the
fact that the FV test has several distinct aspects (stress state,
drainage boundaries, strain rate, and failure mode). It is
interesting that the correlation between FV and DSS is high (ij
= 0.73).
Table 5 bOCR, crate, and dPI factors (Source: Ching et al. 2013).

2579

Factor
bOCR

Test type
CIUC
CAUC

Formula
OCR0.602
OCR0.681

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

CAUE
DSS
VST
UU
UC

crate

CIUC
CAUC
CAUE
DSS
VST
UU
UC

dPI

shown in the right plot of Figure 3. It is clear that the


uncertainty in su is significantly reduced, given the information
from better and/or more tests.

OCR0.898
OCR0.749
OCR0.902
OCR0.800
OCR0.932
1.0+0.1log10(strain rate/1%)
(PI/20)0 = 1
(PI/20)0 = 1
(PI/20)0.178
(PI/20)0.0655
(PI/20)0.124
(PI/20)0 = 1
(PI/20)0 = 1

Table 6 Statistics of Y data points (Source: Ching et al. 2013).


Y1 (CIUC)
Y2 (CAUC)
Y3 (CAUE)
Y4 (DSS)
Y5 (FV)
Y6 (UU)
Y7 (UC)

# pts.

Mean

COV

637
555
224
573
1057
435
387

0.404
0.350
0.184
0.241
0.275
0.243
0.223

0.316
0.318
0.324
0.399
0.416
0.504
0.611

Stdev of
ln(Yi), i
0.315
0.280
0.355
0.277
0.372
0.463
0.523

Mean of
ln(Yi), i
-0.955
-1.090
-1.748
-1.468
-1.363
-1.523
-1.640

Figure 3. Histograms of the (conditional) su samples.


4

Table 7 Correlation matrix C for (X1, X2, , X7) (Source:


Ching & Phoon 2013).
X1 (CIUC)
X2 (CAUC)
X3 (CAUE)
X4 (DSS)
X5 (VST)
X6 (UU)
X7 (UC)

X2
X3
X4
X1
(CIUC) (CAUC) (CAUE) (DSS)
1.00
0.84
0.47
0.72
0.84
1.00
0.39
0.78
0.47
0.39
1.00
0.45
0.72
0.78
0.45
1.00
0.63
0.35
0.41
0.73
0.88
0.7*
0.4*
0.6*
0.85
0.6*
0.3*
0.5*

X5
(FV)
0.63
0.35
0.41
0.73
1.00
0.64
0.46

X6
(UU)
0.88
0.7*
0.4*
0.6*
0.64
1.00
0.68

* insufficient data pairs, estimated based on judgments

X7
(UC)
0.85
0.6*
0.3*
0.5*
0.46
0.68
1.00

3 REDUCING UNCERTAINTY IN DESIGN PARAMETER


WITH BETTER AND/OR MORE TESTS
As mentioned earlier, it is simple to simulate virtual site
investigation data (Y1, Y2, , Yn). First, obtain realizations of
independent standard normal random variables U = (U1, U2, ,
Un) using library functions in many softwares. Realizations of
correlated standard normal random variables X = (X1, X2, ,
Xn) can be obtained using X = LU, in which L is the lower
triangular Cholesky factor satisfying C = LL. Finally, each soil
parameter is obtained using Yi = F-1[(Xi)]. For lognormal
distribution, Yi = exp(i+iXi). Figures 1 & 2 already showed
the simulated data (Y1, Y2, , Yn). This section will further
discuss how to use the simulated data to quantify the
uncertainty reduction in su and by incorporating the test
results from better and/or more tests.
This is illustrated below using results presented in Figure 2.
The histogram of the simulated su data for the same virtual site
is given in the left plot of Figure 3, showing the simulated su
data when no site-specific tests are conducted. Let us consider a
site investigation program consisting oedometer, CPTU, and
SPT N tests. Suppose the test results show that OCR is within
within
[7,9],
and
qT-v
within
[9.5,13.1],
N60
[1100kN/m2,1350kN/m2]. Based on the above information, the
conditional samples of su can be easily obtained by filtering out
samples satisfying OCR[9.5,13.1], N60[7,9], and qTsimultaneously
from
the
v[1100kN/m2,1350kN/m2]
population at large. The su values associated with this filtered
set of (OCR, su, N60, qT - v) values are therefore the conditional
su samples. The histogram of these conditional samples is

CONCLUSION

The construction of virtual sites are demonstrated in this


paper using multivariate normal distributions calibrated from
actual soil property databases. By doing so, it is possible to
evaluate the reduction in the uncertainties associated with
design parameters as a function of better and/or more tests. The
practical goal is to establish an actual (not theoretical) link
between the cost of a site investigation program and the
potential design savings accrued from reliability-based design.
5

REFERENCES

Bolton, M.D. 1986. The strength and dilatancy of sands. Geotechnique


36(1), 65-78.
Ching, J., Phoon, K. K. and Chen, Y.C. 2010. Reducing shear strength
uncertainties in clays by multivariate correlations. Canadian
Geotechnical Journal 47(1), 16-33.
Ching, J. and Phoon, K.K. 2012a. Modeling parameters of structured
clays as a multivariate normal distribution, Canadian Geotechnical
Journal 49(5), 522-545.
Ching, J., Chen, J.R., Yeh, J.Y., and Phoon, K.K. 2012b. Updating
uncertainties in friction angles of clean sands. ASCE Journal of
Geotechnical and Geoenvironmental Engineering 138(2), 217-229.
Ching, J. and Phoon, K.K. 2013. Multivariate distribution for undrained
shear strengths under various test procedures, submitted to
Canadian Geotechnical Journal.
Ching, J., Phoon, K.K., and Lee, W.T. 2013. Second-moment
characterization of undrained shear strengths from different test
modes, to appear in Geotechnical Special Publication honoring
Professor F. H. Kulhawy.
Goldsworthy, J.S., Jaksa, M.B., Fenton, G.A., Griffiths, D.V., Kaggwa
W.S., and Poulos, H.G. 2007. Measuring the risk of geotechnical
site investigations. Proc. Geo-Denver 2007, Denver.
Jaksa, M.B., Goldsworthy, J.S., Fenton, G.A., Kaggwa, W.S., Griffiths,
D.V., Kuo, Y.L., and Poulos, H.G. 2005. Towards reliable and
effective site investigations. Gotechnique 55(2), 109-121.
Ladd, C.C. and Foott, R. 1974. New design procedure for stability in
soft clays. ASCE Journal of Geotechnical Engineering Division
100(7), 763-786.
Mesri, G. and Huvaj, N. 2007. Shear strength mobilized in undrained
failure of soft clay and silt deposits. Geotechnical Special
Publication 173, ASCE, Reston.
Phoon, K.K., Ching, J., and Huang, H.W. 2012. Examination of
multivariate dependency structure in soil parameters. GeoCongress
2012 State of the Art and Practice in Geotechnical Engineering
(GSP 225), ASCE, Reston, 2012, 2952-2960.

2580

Technique of reinforced soil base calculation under fall initiation in ground mass
Technique du compte arm les raisons du sol l'apparition des checs le massif du sol
Ponomaryov A., Zolotozubov D.

Perm national research polytechnical university, e-mail: spstf@pstu.ac.ru

ABSTRACT: On the basis of carried out investigations, the authors obtained stress and strain development mechanisms of the
reinforced ground mass depending on the properties of soils, the characteristics of the reinforcing elements, the depth of their location
and their number. The obtained mechanisms allowed to propose the calculation methodology of the reinforced base surface settlement
in the territories expose to deformation. To evaluate the proposed method, it was compared with previously obtained results of the
experiments and calculations carried out with the help of BS8006, Giroud, Perrier, R.A.F.A.E.L methods and PLAXIS and Sofistik
programs.
RSUM : la base des tudes accomplies les auteurs ont reu les mcanismes du dveloppement de l'effort et l'effort de la masse
affermie de la raison en fonction des proprits des sols, les particularits des lments du renforcement, la profondeur de leur
situation et leur nombre. Les rgularits reues ont permis de proposer la mthode du compte le dpt de la surface des raisons armes
sur les territoires exposs aux dformations. Pour estimer la mthode propose, c'tait en comparaison d'auparavant rsultats acquis
des expriences et calculs accomplis avec l'aide de BS8006, Giroud, Perrier, les mthodes R.A.F.A.E.L et PLAXIS et les programmes
Sofistik.
KEYWORDS: reinforced soils, fall in ground mass, technique of calculation.
1

INTRODUCTION

When laying foundations of buildings and structures in areas


prone to possible vertical deformations (for instance, karstic and
technogenic dolines), it is necessary to provide measures to
prevent emergency situations.
The choice of measures depends on the type of security
perfect or partial. When it is sufficient to provide only partial
security, geosynthetic material reinforcement of a ground base
is most commonly used. It is connected with the fact that
ground reinforcement is more economical as compared with
other methods. In the majority of case reinforcing of the bases
by geosynthetic materials apply at building automobile and
railways. Besides at building on karstic territories it is expedient
to reinforce geosynthetic materials of the bases low-charged
constructions, for example, low-rise buildings.
In Russia ground base reinforcement has not been used
widely so far due to various factors, including both the increase
in the cost of construction connected with the use of
geosynthetics and sufficiently large amount of excavation
works.
High quality geosynthetic materials themselves are not
cheap, and large volumes of excavation arise from the need of a
sufficiently deep placement of reinforcing layers. However, the
use of local materials and a well-tried technology of reinforced
base laying, as well as the increase in the safe upkeep of
buildings, give a good economic effect.
2

EXPERIMENTAL INVESTIGATIONS

Effective use of geosynthetics for reinforcement under ground


mass collapse is possible with the joint account of such factors
as physical and mechanical properties of foundation soil, tensile
properties of geosynthetics, the depth of reinforcing layers and
their number. The optimal choice of these parameters requires
rather complex calculations taking into account load
elongation dependences. The existing methods (the method
outlined in the British Standard BS 8006, section 8.4 (BSI, 1995
Fig. 1), the method of Giroud et al (1990), the method of
Perrier (1985); R.A.F.A.E.L. method (Blivet et al, 2002) do
not consider the actual tensile force relative deformation ratio.

They are applied for single-layer reinforcement. Being used to


solve geotechnical problems, software packages that implement
numerical methods give great inaccuracy, but at the same time
they allow to calculate more quickly and check more types of
reinforcement including those of multi-layer reinforcement.

Figure 1. BS 8006: Parameters used to determine reinforcement.

The study of reinforced bases under ground mass collapse


conducted by the authors allowed to obtain the mechanisms of
stress strain development in the reinforced ground mass
depending on the foundation soil properties, the characteristics
of the reinforcing elements, their depth and quantity. The results
of S. Schwerdts investigations were also used in the study.
On the basis of the mechanisms obtained we proposed the
technique for calculating the reinforced base surface settlement
in areas prone to deformation. As in the above-mentioned
methods, the calculations were carried out for single-layer
reinforcement, but at the same time the change in elongation of
the geosynthetic reinforcing material depending on the load was
taken into account. To do calculations using this method it is
necessary to have load elongation dependences which are
obtained when testing geosynthetics at rupture, in accordance
with ISO 10319:2008 (Fig. 2).
In the course of our studies we were doing experiments with
account of the current Russian regulations enabling to apply
tensile-testing machines to ensure the constant rate of bottom

2581

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

clamp sinking, the constant strain rate or the fixed rate of load
increase (similar to ISO 10319:2008) with relative error
indications of breaking load 1.0 %, with absolute error
indications of elongation 1.0 mm, with an average rupture
duration regulated from (3015) to (6015) sec.

Figure 2. Typical load-elongation curve.

In practice, we had to build these relationships using the


results obtained with the help of the tensile-testing machine that
provided the constant rate of bottom clamp sinking (Fig. 3, 4).
Figure 4. Example of elongation-load curve according to the test results.

The obtained dependences were used in the calculations


done with the help of both numerical methods (PLAXIS
program) and the developed technique.
3 TECHNIQUE OF REINFORCED SOIL BASE
CALCULATION
The design scheme of the proposed method is shown in
Figure 5.

Figure 5. Design diagram of reinforced ground settlement under earth


collapse.
Figure 3. Example of elongation-load curve according to the test results.

In this method the following assumptions allowing to use


formulas well-known in soil mechanics for the calculation of
stresses in ground bases were made:
the reinforced ground mass is in an equilibrium (stabilized)
state before the ground collapse formation;
the reinforcing layer is located in the homogeneous
ground;
the stress-strain state is considered at that moment when
the marginal state of the ground mass is reached;
the deformation form of the ground mass above the
reinforcing interlayer has a sectional view of a trapezoid;
the geosynthetic material does not stretch beyond the
collapse region;
the arch effect is not taken into account.

2582

Technical Committee 211 / Comit technique 211

The algorithm for calculating the proposed technique is


based on the tensile force dependence in the geosynthetics Fa on
the size and shape of the collapse, the reinforcing interlayer
depth, the surface load and geosynthetic material elongation .
Our studies showed that the tensile force elongation
dependence is determined most accurately with the use of the
formula similar to that in the BS 8006 standard
F 0,5 k ( h q ) D 1

1
6 a

(1)

where k is the coefficient taking into account the supposed form


of the collapse (for the rectangular collapse plane problem it
is equal to 1). In the British standard BS 8006 the value of 0.67
for the axisymmetric case is given , but studies showed that
higher convergence with the experimental results was obtained
when k = 0.78; a is the specific elongation of the geosynthetic
material depending on the tensile force in the reinforcing
interlayer, which is determined according to the graphs (for
example, as shown in Fig. 2); h is the depth of the reinforcing
layer, m; is the specific weight of the ground, which is located
above the reinforcing interlayer, kN/m3. If there are ground
layers having different specific weight values and located above
the reinforcing interlayer, it is necessary to make the following
replacement in the formula

project. If the condition s s is not satisfied, then a


geosynthetic material with different characteristics is selected
and the calculation is done again.
To evaluate the proposed method, its comparison with the
results of the experiments and calculations performed by other
methods (BS8006, Giroud, Perrier, R.A.F.A.E.L.) as well as
PLAXIS and Sofistik programs was carried out. Due to the fact
that we were not able to do model experiments in Russia, the
data for comparison were taken from Schwerdts works.
Table 1. Results of calculations

sa 3 / 8 a D 2 .

(3)

To calculate the maximum ground surface settlement, the


following formula is used,
2

D
s
sa ,
2h tan

(4)

where is the inclination angle of the slip plane to the vertical.


The values depend on the characteristics of the backfill soil.
Since it is necessary to determine the maximum surface
settlement by the current Russian regulations, in practical
calculations = is taken. In case the ground layers located
above the reinforcing interlayer have different values, the
following value is used.
n

h
i

i 1

(5)

i 1

Surface
settlement,
mm

105

90

30

64

240

30*

Giroud

215.5

90*

30*

Perrier

120

90

90

R.A.F.A.E.L.

95.8

120

**

PLAXIS

103

160

Sofistik

113

130

***

114.5

107

34

Experiment
BS8006

i 1

where n is the number of ground layers above the reinforcing


interlayer, i is the specific weight of the i-th ground layer, hi is
the height of the i-th ground layer; q is the equivalent surface
load on the reinforcing layer, /. Its calculation depends on
the surface load amount by analogy with the calculation of the
additional pressure (tension) in the ground mass, as well as on
the load type, the load area-to-collapse region ratio, the surface
load location with respect to the ground collapse; D is the
collapse length (diameter), m.
The main problem when calculating by this method is that at
the initial calculation stage we are aware of neither the tensile
force Fa, nor the specific elongation of the geosynthetic material
a because the actual dependence of the elongation on the
tensile load is not taken into account in the formula (1). Thats
why, we used the successive approximations method accurate to
5 %. The received value a is used to determine the maximum
deflection sa of the reinforcing material

Deflection
of
geosynthetic
material,
mm

Calculation method

(2)

hi

Tensile force
in
geosynthetic
material ,
kN/m

Proposed method

*-These are initial data according to the indicated methods.


**-Negative values are received.
***-The Sofistik program does not allow to determine the surface
settlement.

REFERENCES

Blivet et. al. 2002. Design method for geosynthetics as reinforcement


for embankment subjected to localized subsidence. Delmas; Gourc;
Girard (ed): Geosynthetics 7. ICG. Swets & Zeitlinger.
BS 8006: 1995. Code of Practice for Strengthened/Reinforced soil and
Other Fill, British Standart Institution. Section 8. Design of
embankment with reinforced soil foundation on poor ground, 98
121.
Giroud J.P., Bonaparte R.; Beech J.F. 1990. Design of Soil LayerGeosynthetic Systems overlying Voids. Geotextiles und
Geomembranes. 9Jg, H. 1., 11-50.
Schwerdt, S. 2003. Die Ueberbrueckung von Erdeinbruechen unter
Verwendung von einlagig verlegten Geogittern - Vergleich
zwischen Versuchsergebnissen und den Ergebnissen von
analytischen und numerischen Berechnungen. Geotechnik, 26, 95105.
Schwerdt, S., Naciri, O., Jenner, C.G. 2004. Performance of aggregates
in geogrid-reinforced soils used for protection against surface
collapse into underground voids. EuroGeo 3: Geosyntetics
conference, Munich, Germany, 483-488.
Pappiau Ch., Baraize E.; Perrier H. 1995. Motorway level fortification
above carstic cavities. Geotextiles Geomembranes rencontres.
Tome 1, 93-99.
Paul A., Schwerdt S. 2001. Untersuchungen zur Uberbrueckung von
Tagesbruechen und Erdfaellen durch Einbau einer einlagigen
Geokunststoffbewehrung.
Proc.
7.
Informationsund
Vortragstaguung Kunststoffe in der Geotechnik. Muenchen, 251257.
Ponomaryov, A.B., Zolotozubov, D.G. 2010. Effect of reinforcing
material depth on bearing capacity of foundation under ground
collapse. Herald of Civil Engineers, 2 (23), 100-104.
Zolotozubov, D.G., Ponomaryov, A.B. 2009. Structural protection of
ground bases under collapse initiation in karst areas. Herald of
Volgograd State University of Architecture and Civil Engineering.
Section: Construction and architecture 15 (34), 15-18.

The obtained value of the surface settlement s is compared


with the normative or design values for this construction

2583

Stress Concentration Ratio and Design Method for Stone Columns using 2D FEA with
Equivalent Strips
Ratio de concentration de contraintes et mthode de conception pour les colonnes ballastes en
utilisant une analyse aux lments finis 2D avec des bandes quivalentes
Poon B., Chan K.

GHD Geotechnics, 57-63, Herbert Street, Artarmon, NSW, Australia


ABSTRACT: This paper presents an approach for the prediction of vertical and horizontal displacements of soft ground treated with
stone columns in a 2D finite element analysis (FEA). This involved modeling the columns as strips with appropriate strip width,
spacing and smeared properties based on stress concentration ratio. Charts to assess the equivalent 2D column stress concentration
ratio are provided for the design of full depth and floating columns under the influences of various key parameters. The accuracy of
the proposed 2D strip model is investigated by comparing the results with a baseline 3D and axi-symmetric FEA. It is found that the
proposed strip model is preferable over the conventional approach using composite block properties to represent the improved soil.
RSUM : Cet article reprsente une approche pour la prdiction des dplacements verticaux et horizontaux de sols mous traits avec
des colonnes ballastes par une analyse aux lments finis (FEA) en 2D. Cela implique la modlisation des colonnes en tant que
bandes avec une largeur de bande approprie ainsi que lespacement et les proprits des zones dinfluence bases sur le ratio de
concentration de contrainte. Les graphiques pour valuer le ratio 2D quivalent de concentration de contraintes sont donns pour la
conception des colonnes profondes et flottantes sous l'influence de divers paramtres. La prcision du modle de bande 2D propos
est tudie en comparant les rsultats avec une base en 3D et dune analyse aux lments finis axisymtrique. Il se trouve que le
modle de bande propos est prfrable l'approche conventionnelle qui utilise les proprits dun bloc composite pour reprsenter le
sol amlior.
KEYWORDS: Stone column, stress concentration, ground improvement, numerical analysis.
1

b cos30

INTRODUCTION

Conventionally, the design of stone columns involves the


prediction of settlements using a composite material approach in
which equivalent strength and deformation parameters are
derived using semi-empirical correlation to represent the entire
improved soil. While these approaches have been accepted as
reasonable methods for settlement prediction, they are less
certain for the prediction of horizontal displacement. This paper
presents a design approach where stone columns are idealised as
equivalent strips in 2D finite element analysis (FEA). The stress
distribution between the stone column and surrounding soil is
essential for determining the strength parameter of the
equivalent strips. A series of design curves for the stress
concentration are presented to facilitate parameter derivation in
practice. The accuracy of the 2D strip model is investigated by
comparing the results with the 3D and axi-symmetric FEA.
2

d = diameter of
a = width of equivalent
stone column
strip in 2D FEA

2D strip

Asoil

d
Acolumn

Figure 1. 2D stone column strips

The equivalent friction angle eq of the strips can be derived


based on force equilibrium approach as given by
(2)

IDEALISED 2D MODELLING APPROACH

For the modeling of stone columns in 2D FEA, the width of the


stone column strips can be made to be equal to the width of an
equivalent square for the cross-sectional area (Figure 1). The
spacing of the strips is equal to the actual spacing, b, for square
column arrangement and
for equilateral triangular
arrangement. Mohr-Coulomb model is used for the stone
columns with Poissons ratio of 0.3, which is taken to be the
same as the soil itself. The equivalent Youngs modulus Eeq and
the cohesion ceq of the strips can be calculated based on
weighted average approach as given by Eq 1.
(1)

where Asoil and Acolumn are the areas of the soil and column
inside a unit cell within the 2D strip as shown in Figure 1.

The determination of eq requires a presumption of stress


concentration, n, which is defined as the ratio of the average
applied vertical stress within stone column to the average
applied vertical stress of the surrounding soil at the same level.
Section 3 presents an appraisal for this parameter. Note that the
present 2D FEA is an elasto-plastic analysis in which the decay
of excess pore pressure with time was not taken into account.
3

STRESS CONCENTRATION OF STONE COLUMN

This section presents a series of elasto-plastic solutions in charts


for the stress concentration (n) of stone columns founded on (i)
rigid boundary and (ii) infinite compressible soil materials. The
solutions were obtained based on axisymmetric FEA using
PLAXIS software programme for a unit cell consisting of a
stone column and the surrounding soil within a columns zone
of influence. Interface elements were introduced at the soilcolumn contact to allow for slippage. The interface strength was
assumed to be 70% of the original soil strength (Rint= 0.7 in

2585

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Column: E= 60MPa, = 40,


c= 0kPa, total = 22kN/m3
clay = column = 0.3; b/a = 2;
Ko = 0.5

Figure 2. Stone column on rigid base

Soil
Rigid
boundary

Stress Concentration ratio n


0
-2
-4
-6
-8
-10
-12
-14
-16
-18
-20
-22
-24

0 2 4

6 8 10 12 14 16

20kPa

60kPa
80kPa
100kPa
140kPa 120kPa

Elastic
solution 1

Stress Concentration n
0 2 4 6 8 10 12 14 16

nmax

40kPa
Fill
Load

(c)

B
C

z/ qa

Depth z belowtop of stone column (m)

Plastic
stress
point

(b)

4
5
6
7

120kPa

10

b/a = 2; Ec/Es = 30
b/a = 2.5; Ec/Es = 30
b/a = 3; Ec/Es = 30

(for y/a )
(3)
where is the influenced zone (also normalized by the column
diameter a) that is measured from the base of the column to the
equal settlement plane (where r = 1). The magnitude of m
controls the rate of reduction of r with y/d. The higher the m the
more rapid reduction of r would be towards the column tip.
nmax

140kPa

Figure 3. Stone column with rigid base (elasto-plastic solution)

Figure 3c shows a normalised plot in which the depth of the


column, z, was normalised by qa/, where qa is the applied fill
stress and is the total unit weight of the soil. It is found that
the normalised stress concentration curves for the different load
levels ( 40kPa) lie on a single curve. The turning point of the
normalized curve corresponds to the transition from the upper
yielding zone to the lower non-yielding zone, which occurs at
different z for the different qa. For example, point A in Figure 3c
occurs at z/qa = 4. When qa=40kPa and =17kN/m3, z = 9.5m
(B in Fig 3b). Conversely, when qa = 60kPa, z 14m (Point C).
Figure 4 presents a series of normalised curves for the n
value under different modulus ratios, column spacing and
friction angles of the stone column. For a given column spacing
ratio and friction angle, the stress concentration is higher for
higher modulus ratio Ec/Es. Conversely, for the columns with a
given modulus ratio, the extent of the yielding zone, and hence
the reduction of stress concentration, is greater as the spacing
ratio increases even though the maximum stress ratio in the
columns is ultimately similar. This occurs because there is less

b/a = 2; Ec/Es = 20
b/a = 2.5; Ec/Es = 20
b/a = 3; Ec/Es = 20

For stone columns founded on compressible soil, the elastic FE


solution has indicated that there exists a lower equal settlement
plane, below which the columns move more than the soil to
mobilise positive skin resistance of the soil. More load is
transferred from the column to the surrounding soil and
therefore the stress concentration n reduces (see Figure 5a).
Figure 5b shows a plot of normalised distance from the
column base y/a (y and a defined in inset in Figure 5b) versus
stress concentration reduction ratio r (= n / nmax) for the
corresponding elastic FEA results given in Figure 5a. The nmax
is the maximum computed n value based on elasticity as shown
in Figure 5a. The FEA results for r near the column base can be
approximated by the following logarithmic relationship.

80kPa

column = 35

Stone columns on compressible soils (elastic appraisal)

3.2

60kPa
100kPa

0
2
4
6
8
10
12
14
16
18

column = 40

Stress Concentration n
0 2 4 6 8 1012141618202224

Figure 4. Stone column with rigid base

40kPa

0 2 4 6 8 1012141618202224

b/a = 2; Ec/Es = 10
b/a = 2.5; Ec/Es = 10
b/a = 3; Ec/Es = 10

If the column and soil were appraised as elastic materials,


the calculated n (dash line in Figure 3b) increases from 5 at the
top of column, which is consistent with design chart solution
provided in FHWA (1983) for embankment supporting
columns, to about 14 at depth, which is commensurate with the
equal strain solution (soil and column settle at the same rate at
depth) given by Balaam and Poulos (1982).
When the column and soil are modeled as Mohr-Coulomb
materials, yielding elements begin to form at the column top
after a small load (~20kPa) is applied, leading to a reduction in
stress concentration. The yielding of the column (hence the
reduction of n) progresses downwards through the column as
the applied load level increases (see the solid curves in Figure
3b). Figure 3a shows the stress state of the unit-cell model after
the application of maximum embankment load. It indicates that
most yielding elements are confined within the column
periphery. The soil is generally elastic and therefore the soil
friction angle has little influence on the solution.
(a)

0
2
4
6
8
10
12
14
16
18

Embank.
fill
Stonecolumn

Clay:
E=3MPa,
=22,
c=1kPa, total = 17kN/m3

(b)

Stress Concentration n

(a)

z / qa

Figure 3b presents the calculated n with depth for a particular


case where embankment load is applied on stone columns that
are founded on rigid base. The selected column configuration
and parameters are shown in Figure 2. Note that the
embankment fill was modeled as soil elements and the arching
stresses developed above the column have been accounted for in
the FEA model.
b

-2

Ecolumn=60MPa

-4 Esoil=3MPa
b/a=2
-6
-8

Lower equal
settlement plane

-10
Ebase=50MPa
-12
-14

(a)

Ebase=200MPa
0 2 4 6 8 10 12 14 16
Stress concentration n

normalised y/a from column base

Stone columns on rigid base

Depth z below top of column

3.1

confinement for the spaced columns, leading to greater yielding


zone and stress reduction within columns. A comparison of the
corresponding curves in Figures 4a and 4b shows that the loss
of stress concentration due to yielding is more severe for
column material having a lower angle of internal friction.

z / qa

PLAXIS). Note that the interface properties have minimal effect


on the results as the stone column is in triaxial state.

Ecolumn
Esoil

4
3

Ebase

Eq.3:=2.45, m = 9
Ebase = 50MPa

1
0

(b)

FEA
results

Eq3: =1.2, m = 27
Ebase = 200MPa
0

0.2

0.4 0.6 0.8


r = n / nmax

Figure 5. Stone column with compressible base -elastic solution

Figure 5b indicates that as the Youngs modulus Ebase of the


soil beneath the columns increases, the extent of reduces.
Also, the ratio r reduces more rapidly towards the column tip
(i.e. m increases) as Ebase increases. Figure 6 presents the
computed and m for the different Ebase/Ecolumn and Ebase/Esoil
ratios based on elastic FEA. The following points can be drawn:
The influenced zone at the column base reduces as
Ebase/Ecolumn increases. The reduction may be approximated by
a straight line in vs. log(Ebase/Ecolumn) plot. Curves 1 and 4 in
Figure 6a delineate such relationships for column spacing b/a
of 3 and 2, respectively. A curve in between representing b/a =

2586

Technical Committee 211 / Comit technique 211

(a)

60
50
40
m
30
20
10
0

5
6
4

0.01

0.1

10

(b)
Ebase/Ecolumn
2 b/a = 3, Ebase/Esoil =2
5 b/a = 2, Ebase/Esoil =2
4 b/a = 2, Ebase/Esoil 10
8 b/a 3, Ebase/Esoil =2
7 b/a 3,Ebase/Esoil 10
1 b/a = 3, Ebase/Esoil 10

9
8

Elastic
solution

-2

20kPa

-4

40kPa

-6
-8Fill
-10

Load

(a)

-12

60kPa 140kPa

Stress Concentration ratio n


0 2 4 6 8 10 12 14 16

0
-2
-4
-6
-8

-10
-12

(b)

-14

Figure 7. Stone column on compressible (a) c- and (b) su soil

0 1 2 3 4 5 6 7 8
Ebase/Ecolumn

3 b/a = 3, Ebase/Esoil =1
6 b/a = 2, Ebase/Esoil =1
9 b/a 3, Ebase/Esoil =1

Stone columns on compressible soils - elasto-plasticity

The effect of compressible base on stress concentration n is now


discussed based on Mohr Coulomb model. In particular, the
soils surrounding and below the stone columns have been
appraised alternatively using (i) effective shear strength (c, )
and (ii) undrained shear strength su.
Figure 7a shows the computed n under different fill loads for
the same case as in Figure 2,except that the column is founded
on compressible soil that is represented by c- materials. The
stress concentration curves initially follow identical paths as
those shown in Figure 3b until they intercept the lower equal
settlement plane and thereafter trace along the curve of the
elastic solution at the column base. To explain this stress
transfer mechanism, the material stress state of the model at the
end of simulation (under 140kPa fill stress) is presented (inset
in Figure 7a). As before, yielding of the column follows a topdown process. While there is significant yielding of the column
due to high stress ratio, there is little yield in the surrounding
soil especially towards the column base because of sufficient
confinement even with an adopted soil friction angle as low as
22. Since the soil is elastic, the reduction of n due to the
compressible elastic base soil can be superimposed directly onto
the aforementioned reduction due to yielding of column.
Figure 7b presents the results for the case where su = 30kPa
has been adopted for the soils surrounding and below the
column. Significant yielding occurs in the soils, which has
altered the shape of the stress concentration curves towards the
column base as compared to that of the c- soils. However the
differences are not great and for the purpose of assessing n, the
problem can be idealised by assuming that there is no failure in
the surrounding soil so that its behavior is essentially elastic.
3.4

0 2 4 6 8 10121416

-14

Figure 6. Stone column with compressible base -elastic solution

3.3

Stress Concentration ratio n

Depth z below top of stone column (m)

7.5
6.5
5.5
4.5
3.5
2.5
1.5
0.5

where r is the stress concentration reduction ratio given in Eq.


3, which is a function of and m given in Figure 6. nmax is the
maximum elastic n value below the turning point of each
normalised z/qa n curve in Figure 4.
Step 3 Superimpose the solution from Step 2 onto that of
Step 1. Thereby, the final n along the depth of the column is the
lower of the two solutions at the same depth.
Depth z below top of stone column (m)

Influenced zone y/a

2.5 has not been shown for clarity of the figure. Note that
these curves can apply to cases where Ebase/Esoil 10 as Esoil
has negligible effect on the shape of r under this condition.
For a particular b/a ratio, the shows a lower value as
Ebase/Esoil reduces to less than 10, although the trend of
reduction with log(Ebase/Ecolumn) remains linear and parallel
with that for Ebase/Esoil 10 (curves 2 & 3, 5 & 6 in Figure 6a).
The rate of reduction of r towards the column tip, represented
by the m, has been found to increase linearly with Ebase/Ecolumn.
Curve 7 in Figure 6b shows such relationship and is applicable
for cases with different b/a ratio up to 3 (limit of parametric
range) and with Ebase/Esoil 10. Curves 8 and 9 delineate the
corresponding curves for cases with Ebase/Esoil = 2 and 1.

Procedure for assessing stress concentration

The following procedure for assessing the stress concentration


of the stone columns under fill embankment may be proposed:
Step 1 Assessing the stress concentration n along column
depth by using charts such as Figure 4, which have accounted
for the influence of load level, column spacing, modulus ratio of
column and surrounding soil, and yield of the stone column.
Step 2 Assess the influence of the compressible base soil
on n based on elasticity by the following equation:
(4)
n = nmax r

COMPARISON WITH FULL 3-D MODELLING

The accuracy of plane strain idealisation of stone columns using


equivalent strips in 2D FEA was investigated under self-weight
load imparted by a 6m high embankment with 2H:1V batter.
The analyses undertaken for the investigation include: Analysis
1 - Full 3D FEA of embankment over stone columns modeled
by solid elements; Analysis 2 - Axisymmetric FEA of a unit cell
consisting stone column; Analysis 3 - 2D plane strain FEA with
the stone columns modeled as strips; and Analysis 4 - 2D FEA
with the soil and columns modeled as equivalent block. The 2D
and 3D FEA were carried out using software programme
PLAXIS 2D and PLAXIS 3D, respectively.
Table 1 summarises the adopted parameters for all analyses.
The 3D FEA is considered a baseline model that comprises a
13m long segment of embankment over soft clay treated with
stone columns which are founded on compressible soil. The
analysis was repeated with the 0.9m diameter stone columns
spaced at 1.7m, 2m and 2.5m in triangular pattern. The 3D FE
mesh is shown in Figure 8. The stone columns are modeled
using 15 nodes wedge element with interface elements at the
column-soil contact. Two cases of interface strength of 100%
and 67% of the surrounding soil strengths have been considered.
Table 1. FEA Model Parameters
Analysis
1,2
3D FEA
3 - 2D
FEA
(strips)

b/d
2.0,
2.3,
2.9

ar
0.26,
0.19,
0.12

2.0

0.26

2.3

0.19

2.9

0.12

Stone Column Parameters


Ecol=50MPa, ccol=0kPa, col = 40
Estrip=26MPa, cstrip~1kPa, strip = 36.5 38 along shaft , = 35.5 near base
Estrip=22MPa, cstrip~1kPa, strip = 35.537 along shaft ; =34 near base
Estrip=18MPa, cstrip~1kPa, strip = 34.535.5 along shaft; = 33 near base
Eblock=6MPa, cblock~1kPa,block = 30

4 - 2D
2.0
0.26
FEA
2.3
0.19
Eblock=6MPa, cblock~1kPa, block = 30
(equiv.
2.9
0.12
Eblock=6MPa, cblock~1kPa, block = 30
block)
Soil surrounding columns are Esoil = 3MPa, csoil = 2kPa, soil = 26;
Soil beneath columns are Ebase = 3MPa, csoil = 5kPa, soil = 28

In Analysis 3, a 2D plane strain idealisation of the stone


columns using equivalent strips was investigated. The strips are
divided into several segments, each of which has different
strength properties that correspond to the varying stress
concentration along the column depth. The dimension and
spacing of the 2D strips are as per those outlined in Figure 1.
Analysis 4 presents a conventional 2D approach in which the
entire treated soil is represented by a single block with the

2587

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

equivalent properties, block, c block and E block derived based on


the semi-empirical relationships given by Madhav, 1996.
Embankment with 2H:1V batter

6m
10m

~13m

Soft Soil without


treatment

Figure 8. 3D FEA mesh

The baseline 3D model (Analysis 1) and the 2D strip model


(Analysis 3) show similar deformation mechanisms of the stone
columns, which can be broadly divided into three zones (see Fig
9a, 9b): Zone 1 away from the fill batter where columns
underwent vertical deformation by bulging; Zone 2 just
behind the crest of the fill batter where columns underwent both
vertical and horizontal deformation by bulging and leaning;
and Zone 3 beneath the fill batter where columns underwent
mainly leaning. This numerical prediction of the deformation
appears to be consistent with the results of the centrifuge model
test carried out by Stewart and Fahey (1994). The maximum
settlement of the embankment occurs in Zone 2 just before the
crest of the fill batter (more than that in Zone 1). This is
presumably due to the concurrence of bulging and leaning
deformation mechanisms of the stone columns. Conversely, the
columns in Zone 3 exhibit the maximum horizontal
displacement and are likely due to the prevailing leaning
deformation of the stone columns.

Zone 1
Bulging

Leaning
(a) 3D FEA (Baseline Analysis 1) with cylindrical stone columns
Max vert.
Max vert. disp.
disp.
Q

(b) 2D FEA (Analysis 3) with


equivalent stone column strips

230
210
190
170
150

10
15
20
25
30
Area replacement ratio ar

(d)

Hori. displ. (mm) at point Q

Axi-symmetric

250

5
3
1
2

0
10
20
30
Area replacement ratio ar

(e)

Settlement at Point P

3D (100% )
2D_Strip (100%)

110
100
90
80
70
60
50
40
30
20

3D (67%)
2D_Strip (67%)

CONCLUSIONS

This paper presents a 2D FEA approach for analysing the


response of stone columns under embankment loading. The
stone columns are modeled as equivalent strips with the ceq and
Eeq of the strips calculated based on weighted average area
approach, and the eq derived based on force equilibrium
method, which requires a presumption of stress concentration
ratio of the stone column. For convenience, charts to assess the
stress concentration ratio have been generated for full depth and
floating stone columns. The solutions cover key parameters
including load levels, column spacing ratio, Ecolumn/Esoil ratio
Ebase/Ecolumn ratio, Ebase/Esoil ratio and column friction angles.
The accuracy of the proposed 2D strip model has been
investigated by comparing the results of the 3D baseline FEA
and the conventional composite approach. It has been shown
that the proposed strip model is preferable over the conventional
approach for the prediction of horizontal displacement.
However, further research is needed to develop a regime to
determine equivalent interface strength in the 2D strip method.

(c) 2D FEA (Analysis 4) with


equivalent composite block

270

Settlement (mm) at point P

Zone 3

Zone 2
Bulging & Leaning

Figure 9e presents the predicted horizontal displacement at


point Q. The following points are drawn from the results:

hen original soil strengths are used for the interface


properties, the result of the 2D strip model (curve 1) compares
well with that of the 3D baseline model (curve 2). Both results
show a trend of reducing horizontal displacement with ar.

hen the interface strength of the columns in the 3D model are


reduced to 67% of the soil strengths, the result (curve 3)
indicates an initial drop off in horizontal displacement with ar,
but increases again once ar > 20%. This is due to increasing
proportion of yielding elements in the remolded soil as the
columns draw closer to each other.

he application of the same interface strength reduction (67%


of surrounding soil strength) in the 2D equivalent strip model
has caused excessive yield in the remolded soil and led to
increased horizontal displacement with ar (curve 4). A better
fit to the 3D solution is by changing the interface strength to
80% of the surrounding soil strength (curve 5). Evidently,
there needs a regime to determine an equivalent interface
strength for the strip model. This merits further research.

he 2D block model result (curve 6) under-predicts the


horizontal displacement when compared with the 3D baseline
model predictions. This indicates that the use of isotropic soil
properties in the 2D block model, which were derived based
on semi-empirical relationships originally for settlement
prediction under axially loading condition, have
overestimated the reduction in lateral spreading underneath
the embankment batter. The use of equivalent strips in the 2D
strip model is able to capture the interaction between the soil
and the stone column, leading to a better agreement for the
lateral deformation with the 3D baseline solution.

Hori. Disp.at Point Q

2D_Block
2D_Strip (80%)

Figure 9. Comparison of FEA results

Figure 9c presents the deformation predicted by the


conventional 2D FEA using composite block material (Analysis
4). This method is unable to capture the bulging and leaning
deformation of the stone columns. The maximum settlement
occurs at the centre of the embankment (i.e. in Zone 1) as
opposed to in Zone 2 as predicted by the baseline 3D FEA and
the proposed 2D FEA using equivalent strips.
Figure 9d shows a plot of predicted settlements at points P
versus area replacement ratio ar. All analyses give comparable
results, indicating that all the different FE methods are
commensurable in terms of settlement prediction under axially
symmetric load condition.

REFERENCES

Balaam, N.P. and Poulos, H.G. 1982. The behavior of foundations


supported by clay stabilized by stone columns. Proc. 8th European
Conf. on Soil Mechanics and Foundation Engineering, Helsinki.
FHWA. 1983. U.S. Department of Transportation Federal Highway
Administration (Dec, 1983) Design and Construction of Stone
Columns, Vol 1. Report No. FHWA/RD-83/026.
Madhav,M.R. and Nagpure,D.D. 1996. Design of granular piles for
embankments on soft ground. Proc. 12th SE Asian Geot.Conf.,
Kaula Lumpur. 1: 285-290
Stewart, D.P. and Fahey, M. (1994). Centrifuge modelling of a stone
column foundation system, Seminar on ground improvement
techniques , Perth, Curtin Printing Services, 1: pp 101-111.

2588

Porosity/cement index to evaluate geomechanical properties of an artificial


cemented soil
Le paramtre porosit/ciment pour lvaluation des proprits gomcaniques d'un sol ciment
artificiellement
Rios S., Viana da Fonseca A.

Faculty of Engineering of the University of Porto

ABSTRACT: This paper highlights the importance of the porosity/cement index on the evaluation of the geomechanical properties of
soil-cement mixtures as a contribution to analyse these materials. This index is defined as the ratio between porosity and volumetric
cement content combining the degree of compaction with the cement content. The relevance of these two parameters is defined by an
exponent to the volumetric cement content which changes with the type of soil. This paper results from a broad experimental program
with unconfined compression tests, indirect tensile tests, triaxial tests and oedometer tests, which were all analysed by this index
adjusted by a specific exponent value. The (tensile and compression) strength, the (elastic and initial tangent) stiffness, as well as the
compressional behaviour are conveniently represented by this index and a different behaviour is observed when this index is changed.
RSUM : Limportance du paramtre porosit/ciment dans lvaluation des proprits gomchaniques des mlanges sol-ciment est
prsente dans cet article comme une contribution pour lanalyse de ces matriaux. Ce paramtre est dfini comme le rapport entre la
porosit et la teneur volumique en ciment. Limportance relative entre la porosit et la teneur en ciment est introduite en introduisant
un exposant la teneur volumique en ciment dpendant du type de sol. Les rsultats dun vaste programme exprimental incluant
essais de compression simples, essais de traction indirect, essais triaxiaux et essais domtriques sont prsents et analyss par ce
paramtre ajust par un exposant spcifique. La rsistance la compression et la traction, la rigidit lastique et tangente initiale,
ainsi que le comportement en compression sont bien reprsents par lintermdiaire de ce paramtre et un comportement diffrent est
observ si le paramtre est modifi.
KEYWORDS:soil-cement, porosity/cement index, tensile strength, compression strength, compressional behaviour.
1

INTRODUCTION

Soil-cement mixtures are very interesting for the construction of


road and railway platforms, especially in the noble layers of
subgrade as well as in transition zones between embankment
and concrete structures, where good mechanical properties are
required. This solution, not only concurs to improve those
characteristics, but also leads to a significant reduction in the
economic and environmental costs of these works. Despite these
advantages this method has not a generalized application in
Portugal due to the lack of design methodologies based on
mechanical parameters.
There are several factors affecting the behaviour of cemented
soils, such as the type of cement and cement content, the curing
time and stress, the water content and porosity. Seeking for a
ratio that would reflect the influence of some of these
parameters Consoli et al. (2007) presented an index property
defined as the ratio of porosity to the volumetric cement
content, called porosity/cement ratio (n/Civ). Some previous
attempts have been made, such as the degree of cementation
proposed by Chang and Woods (1992) that concerns the
percentage of voids filled with cement, being this parameter
developed for sands. Lorenzo and Bergado (2004) have also
presented the ratio of the after curing void ratio to the cement
content (eot/Aw) proving to be quite interesting for clay mixtures
with high values of water and cement content.
Another available parameter is the water/cement ratio used
for concrete. However, soil-cement mixtures for road or railway
platforms are usually cured in a non saturated condition, which
makes the previous ratio inadequate in the analysis of these
mixtures behaviour. The main difference between soil-cement
mixtures and concrete (besides the cement content) is that
during the curing of concrete all voids are completely full of

water and therefore concrete stress-strain behaviour is not


dependent on the void ratio but on the water content. In
opposition, soil-cement mixtures currently executed in
embankments and transport platforms have curing water content
lower than the saturation water content and so their
compressibility will be related to its porosity. Moreover, while
concrete has an almost linear behaviour for a wide range of
deformations, soil-cement mixtures have a clear non-linear
behaviour since very small strains as a result of the progressive
degradation of the cemented structure. Therefore, even if the
soil-cement mixture is saturated after the maximum strength has
been achieved (i.e. after curing) the curing void ratio still has a
very important role on the mechanical behaviour of the mixture.
The influence of the porosity/cement ratio on strength and
stiffness parameters is described in Consoli et al. (2012)
providing the comparison between two different materials
mixed with Portland cement: well graded Porto silty sand and
uniform Osorio sand. An advance analysis on the compression
and shearing behaviour of cemented Porto silty sand through
this parameter is described in Rios et al. (2012).
This paper summarizes some geomechanical properties of
cemented Porto silty sand through this index in terms of
strength (unconfined, tensile and triaxial), stiffness (initial
tangent and unload-reload) and one-dimensional compression.
2

MATERIALS AND SPECIMEN PREPARATION

A well graded soil, classified as silty sand (SM) in the unified


classification system (ASTM, 1998) was used in this study. The
soil is derived from weathered Porto granite which is abundant
in Northern Portugal (Viana da Fonseca et al., 2006). Its particle
specific gravity is 2.72, and it contains around 30% fines,

2589

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

although a low plasticity index was obtained (IP=wL-wP=34%31%=3%). From the particle size distribution curve presented in
Figure 1 an average diameter D50 equal to 0.25 mm was
obtained, as well as uniformity and curvature coefficients of 113
and 2.7 respectively.A high strength Portland cement (CEM I
52,5R) of grain density equal to 3.15 was used as the cementing
agent in order to speed up the laboratory tests.
The experimental program is performed with specimens
made by the mixture of silty sand, Portland cement and tap
water that is compacted statically in three layers in a stainless
steel mould. For each specimen, a quantity of fines equal to the
weight of cement to be introduced was removed from the soil,
in order to have the same grain size distribution curve in the
mixture of soil-cement as in the soil itself. Following this
procedure the dry density of the soil was also constant
throughout the study even though the cement content changed.
The specific gravity of the cement-soil mixture was calculated
as a weighted average of those of the soil (Gs=2.72) and of the
cement (Gs=3.15), and thus it was different for different cement
contents.
100

ASTM sieves series (mm)

90

10

80

20

70

30

60

40

50

50

40

60

30

70

20

80

10

90

0
0.0001

0.001

0.01
0.1
Grain size (mm)

Taking into account the possibility of shrinkage in cemented


materials, the evaluation of the tensile strength is of utmost
importance. In that sense, indirect tensile tests following the
standard EN 13286-42 (CEN, 2003) were performed on similar
specimens whose results were plotted against n/Civ0.21 for which
Eq. (2) was obtained,
Rtb (kPa) = 2E+09 (n/Civ0.21)-4.719

(2)

The results showed that the indirect tensile strength (Rtb) was
about 11% of the UCS. In Figure 2 both Rtb and UCS are
plotted against n/Civ0.21 in different scales for comparison. It is
clear that both trends are very similar (except for the absolute
values) corroborating the convenience of the adjusted
porosity/cement ratio.
In Consoli et al. (2011), where the data from these tests is
plotted together with data from other two soils, it is shown that
for the three soils a decrease in porosity promotes an increase in
the tensile strength as a consequence of the higher number of
contact points between particles which improves the
cementation. Also for the other two soils, a unique correlation
was found between the adjusted porosity/cement ratio and the
indirect tensile strength, the exponent of the ratio depending on
the soil.

Figure 2. Indirect tensile strength and unconfined compression strength


against the adjusted porosity/cement ratio

STRENGTH PARAMETERS

3.2 Triaxial tests

3.1 Unconfined compression strength


Strength properties of the cemented sand were evaluated in
different ways by means of unconfined compression tests,
indirect tensile tests, as well as triaxial tests. First, several
specimens moulded to have different cement contents (2%, 3%,
5% and 7%) and dry unit weights (16.4, 17.2, 18.0 and
18.8 kN/m3) were tested in unconfined compression in a total of
16 tests. In these tests, the water content was kept equal to 12%.
The representation of the unconfined compression strength
(UCS) and the ratio of porosity to the volumetric cement
content (n/Civ) revealed that some adjustment was needed and
therefore, an exponent was added to Civ. This exponent was
defined as the value that provides the best correlation
coefficient with the data, which, for this material, was found to
be 0.21 Eq. (1).
UCS (kPa) = 4E+09 (n/Civ0.21)-4.296

Tensile strength

100
10

Figure 1. Grain size distribution curve

1.1

(1)

This exponent seems to depend on the type of soil as other


authors have found different coefficients when working with
different soils (Consoli et al., 2007, 2011): an exponent of 0.28
was found in a residual soil from sandstone (Botucatu soil),
while a value of 1.0 was found in an uniform sand (Osorio
sand). Based on this parameter, named adjusted porosity/cement
ratio (n/Civ0.21), the results of different tests were analysed.

2590

Drained triaxial compression tests were performed over soilcement specimens, which were moulded to have two different
adjusted porosity/cement ratios (n/Civ0.21=36 and 29)
corresponding respectively to UCS of 800 kPa and 2000 kPa.
For these ratios, two moulding conditions were defined
characterized by cement content and dry unit weight while the
water content remains constant and equal to 12%. For the first
ratio (n/Civ0.21= 36), 2 and 4% cement contents were considered
which lead to dry unit weights of 16.7 and 15.4 kN/m3,
respectively. For the second ratio (n/Civ0.21 = 29) higher strength
was needed, so 5 and 7% of cement contents were assumed with
17.0 and 16.4 kN/m3 of dry unit weight. The tests were
performed at three different effective confining pressures (30,
80 and 250 kPa) over specimens moulded in four moulding
points, comprising 12 tests (Table 1).
The stress-strain curves (see Figure 3 as an example) clearly
evidence that the specimens with n/Civ0.21=29 have higher peak
deviator stresses than the specimens with n/Civ0.21=36
independently of the cement content. Adding cement to the sand
had the effect of increasing the shear strength by up to five
times for the adjusted porosity/cement ratio of 36 and tenfold
for the adjusted porosity/cement ratio of 29.All specimens
initially compressed, followed by significant dilation, which
was associated to a peak strength, before strain softening. This
is typical of cemented soils, with the maximum rate of dilation
taking place right after the peak strength (Viana da Fonseca
1998). The peak strength corresponds to the onset of significant

Technical Committee 211 / Comit technique 211

breakage in the cement, while dilation involves particle


rearrangement that is only possible after bonding breakage.
Assuming only compressive volumetric deformations up to the
point of zero dilation, beyond this point yielding exists, which
indicates that the onset of cement breakage is progressive
starting even before peak. However, being the peak strength not
frictional but controlled by the cement yielding, then most
destructuration may take place only at peak.
Table 1.Moulding conditions of the specimens for the triaxial tests
Moulding
Point

%C
(%)
2

d
(kN/m3)
16.6

2
2
4
4
4
5
5
5
7
7
7

16.5
16.7
15.4
15.7
15.5
16.9
17.0
17.0
16.3
16.5
16.7

n/Civ0.21

0.61

w
(%)
12

0.62
0.60
0.74
0.71
0.73
0.59
0.58
0.58
0.66
0.63
0.61

12
12
12
12
12
12
12
12
12
12
12

36
36
36
36
36
29
29
29
29
29
29

e0

36

UCS
(kPa)
800
800
800
800
800
800
2000
2000
2000
2000
2000
2000

c
(kPa)
30
80
250
30
80
250
30
80
250
30
80
250

In Figure 4 the stresses acting on the plane are plotted on a


(, ) graph for peak conditions from which the correspondent
strength parameters were obtained. The points are assigned to
each adjusted porosity/cement ratio (n/Civ0.21) expressed before.
It is interesting to notice that for peak conditions two strength
envelopes were obtained depending on the index ratio. The
adjusted porosity/cement ratio influences the peak angle of
friction and cohesion intercept, being the peak envelope for the
index n/Civ0.21=29 higher than that for the index equal to 36.
This could have been predicted from moulding characteristics as
each ratio corresponds to different UCS.
4

STIFFNESS PARAMETERS

4.1 Initial tangent stiffness


The unconfined compression tests were performed with local
measurement of deformation using LDTs and so the stiffness
modulus could be evaluated. An initial tangent modulus (Eti)
was then calculated based on the linear part of the stress-strain
curve. Plotting this modulus against the adjusted
porosity/cement ratio for the 16 tests presented above, as
Figure 5 shows, it can be concluded that the general adjustment
of the data is quite reasonable.
6000
Eti = 2E+09 (n/Civ0.21)-4.169
R = 0.85

5000

2%C
3%C
5%C

Eti (MPa)

4000
3000

7%C

2000
1000
0

20

25

n/Civ0.21

30

35

40

Figure 5. Indirect tensile strength and unconfined compression strength


against the adjusted porosity/cement ratio
Figure 3. Stress-strain curves for the confining pressure of 80 kPa.

All the cemented specimens tested in triaxial tests suffered


strain localisation. Therefore, it becomes difficult to rely on the
local instrumentation at strain levels close and after the peak,
but especially at ultimate conditions. In that sense, the stress
invariants such as the deviator stress (q) and the mean effective
stress (p) were not considered representative of the stress state
and thus, the stresses acting on the shear plane were calculated
by the procedure used by Gasparre (2005) based on the Mohr
circles and taking into account the post rupture analysis
described by Burland (1990).

Figure 4.Peak strength envelopes by adjusted porosity/cement ratio.

As expected, stiffness parameters are more scattered than


strength parameters because strain measurements are always
more sensitive to non-homogeneities of the specimen and
anchors are introduced in the specimen in single reference
points. On the contrary, strength measurements capture more
easily an average value of the whole specimen.
4.2 Unload-reload moduli
In the triaxial compression tests reported above, a small
static cycle was performed during shearing, between 30% and
15% of the expected peak deviatoric stress. These loads were
selected to avoid soil yielding before the cycles so the modulus
could be assumed elastic. Figure 6 summarizes the results of the
unload-reload moduli (Eur) obtained for the two adjusted
porosity/cement ratio. The values of Eur obtained from these
triaxial tests are higher than the initial tangent modulus (Eti)
obtained in the unconfined compression tests presented in
Figure 5. This can be considered expected because Eur is usually
assumed to follow an elastic pattern, if performed at low ranges
of cyclic stress, while in the initial monotonic loading path
some compliance errors of strain gauges may be presented. This
graph also evidences a clear and almost discrete increase in the
stiffness modulus values for the specimens with n/Civ0.21=29
(5% and 7% cement contents) in comparison with the
specimens with distinct ratio n/Civ0.21=36 (2% and 4% cement
contents). This could be even clearer if the instrumentation

2591

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

would perform in a completely satisfactory way for the highest


cemented mixtures (for 5% and 7% of cement content).
6000

1
2
3
4

Test condition
2%C; 'c =30 kPa
2%C; 'c =80 kPa
2%C; 'c =250 kPa
4%C; 'c =30 kPa

3000

5
6

4%C; 'c =80 kPa


4%C; 'c =250 kPa

2000

7
8
9

5%C; 'c =30 kPa


5%C; 'c =80 kPa
5%C; 'c =250 kPa

Eur (MPa)

n/Civ0.21=29

n/Civ0.21=36

5000
4000

1000
0

5 6 7 8
Test condition

10 11 12

10 7%C; 'c =30 kPa


11 7%C; 'c =80 kPa
12 7%C; 'c =250 kPa

Figure 6. Stiffness modulus obtained in the unload-reload cycles.

COMPRESSIBILITY PARAMETERS

One-dimension compression tests in oedometer cells with


constant rate of deformation (CRD) were performed over soilcement specimens in the four moulding conditions presented in
Table 1. The preparation of the different mixtures for these tests
followed the same procedure of the other tests, as expressed
briefly in section 2. Due to the size of the mould, the static
compaction was performed in one layer, although the soil was
placed in several stages followed by tapping. For the calculation
of the mean effective stress (p) in each test the value of the
coefficient of earth pressure at rest (k0) was considered equal to
1 due to the high compaction degree that the specimens were
subjected during moulding (>80% of the Modified Proctor test).
Figure 7 shows two of those tests, corresponding to two
different porosity cement ratios, indicating that these
compressibility curves do not seem to converge.
0.80

n/Civ0.21= 36

0.70
0.60

0.50

n/Civ0.21= 29

0.40
0.30
0.20
0.10
0.00
0.001

0.010

0.100
1.000
p' (MPa)

10.000

100.000

Figure 7.Void ratio against mean effective stress for two different
porosity/cement ratios

However, plotting the tests performed over specimens with


the same porosity cement ratio, a unique compressibility line
was obtained. The same has happened for the other two tests
with the other porosity cement ratio of 36. On the contrary,
when specimens with the same cement content but different
void ratio were represented no unique line was obtained. These
results indicate that this ratio can better reproduce the behaviour
in one-dimensional compression than the cement content or
initial void ratio alone.
6

CONCLUSIONS

This paper presented a great number of data from different tests.


Together, they allowed a better understanding of the artificially
cemented soil used in this work. Compressive and tensile
strength, strength envelopes, stiffness parameters and one
dimensional behaviour were some of the most important issues
studied. The adjusted porosity/cement ratio revealed to be very
consistent and useful for the analysis of the unconfined

2592

compression strength since a unique trend was obtained


between this variable and n/Civ0.21. A similar trend was obtained
for the indirect tensile strength performing tests over specimens
moulded in the same conditions. The comparison of the two
curves provided a relationship between indirect and
compressive strength of about 11%.
The strength envelope values of the cemented specimens
tested in triaxial compression were obtained through a
procedure based on the Mohrs circles analysis to solve the lack
of representativeness of principal stress analysis due to non
correspondence of the real localised shear locus. In fact, strain
localisation is unavoidable in these very stiff materials, and
consequently, the global stress-strain measurements are no
longer representative of the conditions throughout the shearing
process. Two peak strength envelopes were obtained for each
n/Civ0.21 (n/Civ0.21=36: =30 and c= 253 kPa; n/Civ0.21=29:
=39 and c= 589 kPa), showing once again the convenience
of this ratio for the analysis of these mixtures behaviour.
This ratio also seems to be very useful to reproduce the one
dimensional compression behaviour of the mixture, since for
each n/Civ0.21 a single line was obtained for higher stresses.
7

ACKNOWLEDGEMENTS

This research was developed under the activities of FCT


(Portuguese Foundation for Science and Technology) research
unit CEC, in FEUP [PTDC/ECM/ 099475/2008], and financed
by the European Community (QREN/UE/FEDER), Operational
Program for Competitive Factors "COMPETE".
8

REFERENCES

ASTM 1998. D 2487-98 Standard practice for classification of soils for


engineering purposes (Unified Classification System)
Burland J. B. 1990. On the compressibility and shear strength of natural
clays. Gotechnique 40(3), 329-378.
CEN 2003.EN 13286-42 - Unbound and hydraulic bound mixtures. Test
method for the determination of the indirect tensile strength of
hydraulically bound mixtures, ComitEuropen de Normalisation
Chang T. S. and Woods R.D. 1992. Effect of particle contact bond on
shear modulus.Journal of Geotechnical and Geoenvironment
Engineering118(GT8), 1216-1233.
Consoli N.C., Viana da Fonseca A., Cruz R., Rios S. 2011.
Voids/Cement ratio controlling tensile strength of cement treated
soils. Journal of Geotechnical and Environmental Engineering,
137(11), 1126-1131 (doi:10.1061/(ASCE)GT.1943-5606.0000524)
Consoli N.C., Viana da Fonseca A., Rios S., Cruz R. and Fonini A.
2012.Parameters controlling stiffness and strength of artificially
cemented
soils.Gotechnique
62(2),
177-183
(doi:
10.1680/geot.8.P.084)
Consoli N.C., Foppa D., Festugato L. and Heineck K. 2007.Key
Parameters for Strength Control of Artificially Cemented Soils,
Journal of Geotechnical and Geoenvironmental Engineering
133(2), 197-205.
Gasparre A. 2005. Advanced laboratory characterisation of London
Clay. PhD dissertation presented to the University of London
Lorenzo G. and Bergado D. 2004.Fundamental parameters of cementadmixed clay - New Approach.Journal of Geotechnical and
Geoenvironmental Engineering 130(10), 1042-1050.
Rios S., Viana da Fonseca A. andBaudet B. 2012. The effect of the
porosity/cement ratio on the shearing behaviour of cemented
soil.ActaGeotechnica (accepted)
Viana da Fonseca, A. 1998. Identifying the reserve of strength and
stiffness characteristics due to cemented structure of a saprolitic
soil from granite. The Geotechnics of Hard Soils - Soft Rocks,
Evagelista and Picarelli (eds), Balkema, Roterdam
Viana da Fonseca A., Carvalho J., Ferreira C., Santos J.A., Almeida F.,
Pereira E., Feliciano J., Grade J. and Oliveira A. 2006.
Characterization of a profile of residual soil from granite combining
geological,
geophysical
and
mechanical
testing
techniques.Geotechnical and Geological Engineering 24, 13071348

Compressive Strength of Fiber-Reinforced Lightly-Cement Stabilized Sand


Rsistance la compression des sables renforces par fibres et ciment
Sadek S., Najjar S., Abboud A.

Department of Civil and Environmental Engineering, American University of Beirut

ABSTRACT: The stabilization of soils using cementing agents has long gained acceptance and is well established in geotechnical
engineering practice. Furthermore, adding discrete fibers to the cement-treated soil has been shown to improve the soils response to
loading and its overall engineering behavior. Limited studies of the behavior of fiber-reinforced cemented sand in the laboratory
indicated that the addition of cement and fibers results in an increase in the strength of the composite, especially at high fiber contents
and lengths. Cemented sands were found to be brittle compared to un-cemented sands, with the brittleness decreasing with the
inclusion of fibers. The objective of this paper is to investigate the effect of randomly distributed fiber reinforcements and cement
addition on the response of sandy soils. To achieve this objective, the behavior of cement/fiber-reinforced sands was studied using
unconfined compression tests. The parameters that were varied are [1] the cement content (0.5% and 1%), [2] the fiber content (0%,
0.25%, 0.5%, and 1.0%), and [3] fiber lengths (6, 12 and 20 mm). The results of the tests were used to quantify the degree of
improvement in strength and stiffness due to the addition of fibers and cement to the cohesionless sand.
RSUM: La stabilisation des sols en utilisant des agents de cimentation a longtemps t accept et est bien tablie dans la pratique
de la gotechnique. De plus, l'ajout de fibres discrtes au sol-ciment a t demontr effectif pour amliorer la rponse du sol au
chargement et son comportement mcanique global. Seules quelques tudes xistent o le comportement des sols renforcs par des
fibres et ciment a t tudie au laboratoire. Elles ont indiqu que l'addition de ciment et de fibres engendrent une augmentation de la
rsistance du composite, en particulier des teneurs en fibres et des longueurs leves. Les sables ciments ont t jugs fragiles par
rapport aux sables non ciments; cette fragilit diminue avec l'inclusion de fibres. L'objectif de cet article est d'tudier l'effet de
renforts en fibres distribues de faon alatoire et l'ajout de ciment sur la rponse des sols sableux. Pour atteindre cet objectif, le
comportement de sables enforcs par ajout de ciment / fibres r a t tudie par des essais de compression non confine. Les
paramtres qui ont t varies sont [1] la teneur en ciment (0,5% et 1%), [2] la teneur en fibres (0%, 0,25%, 0,5%, et 1,0%), et la
longueur des fibres [3] (6, 12 et 20 mm). Les rsultats des tests ont t utiliss pour quantifier le degr d'amlioration de la rsistance
et de rigidit due l'ajout de fibres et de ciment au sable pulvrulent.
KEYWORDS: fiber-reinforced sand, cement stabilized sands, fibers, cement, unconfined compressive strength.
1

INTRODUCTION

The geotechnical and materials/pavement engineering fields are


witnessing an increasing interest in exploring soil improvement
schemes that are based on the addition of stabilizing agents such
as synthetic or natural fibers and/or cementing agents for
various applications. The objective is to produce a composite
material with improved engineering properties that could be
used in lieu of good quality construction material that is
typically obtained through non-sustainable and environmentally
problematic activities such as quarrying. The composite
material with its improved engineering properties could be used
to replace conventional base and sub-base material under
pavements, or to support foundations of light structures or
infrastructure, which otherwise could not be adequately
supported by the natural soil. The improved material could also
be used as backfill behind earth retaining walls and reinforced
or stabilized slopes.
The experimental data that is available in the literature for
fiber/cement reinforced sands is relatively limited (Maher and
Ho 1993, Consoli et al. 1998, Kaniraj and Havanagi 2001,
Sobhan and Mashnad 2002, and Consoli et al. 2002). There is a
need for designing and implementing a comprehensive
experimental testing program that is aimed at investigating the
behavior of fiber/cement reinforced sands systematically. To
achieve this objective, the behavior of cement/fiber reinforced
sands was studied in the laboratory using unconfined
compression tests. The parameters that were varied in this study
are [1] the cement content (0.5% and 1%), [2] the fiber content
(0%, 0.25%, 0.5%, and 1.0%), and [3] fiber lengths (6, 12 and
20 mm). The results of the tests were used to quantify the
degree of improvement in strength, stiffness, and ductility due
to the addition of fibers and cement to the cohesionless sand.

EXPERIMENTAL PROGRAM

Twenty unconfined compression tests on fiber/cement


reinforced sands were conducted as part of this study.
2.1

Material Properties

The sand used in this study is Ottawa Sand with the properties
shown in Table 1. The sand classifies as a poorly graded sand
(SP) according to the Unified Soil Classification System.
The fibers (Fig. 1) chosen for the reinforcement are
polypropylene fibers, typically used as secondary reinforcement
of lightweight concrete and mortar mix designs. They were
adopted because they are available in several lengths, they can
be mixed with soil-cement mixtures and satisfy efficiently the
intended role of reinforcement. The fibers have a specific
gravity of 0.91 g/ml, a tensile strength of 0.38 kN/mm2 and a
young modulus of 3.5 kN/mm2. Fiber lengths of 6 mm 1, 12
mm 1 and 20 mm 1 were used in the testing program. The
nominal diameter of the fibers was determined in the lab under
an electronic microscope to be in the order of 0.1mm.
Table 1. Table caption (TNR 8), numbered consecutively. Tables placed
below caption. TNR 8 for text and numbers in Table.
Soil Property

Value

D10 (mm)

0.22

D30 (mm)

0.31

D60 (mm)

0.42

Coefficient of uniformity (D60/D10)

1.95

Coefficient of curvature (D30)2/(D60*D10)

1.04

Maximum and minimum void ratios (emax,emin)


Specific gravity

(0.75,0.49)
2.65

The cement used in this study is normal Portland cement type I.


The same sources of cement and sand were used for all the
2593 specimens to eliminate all risk of material discrepancy.
2.2

Sample Preparation

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Specific gravity

2.65

The cement used in this study is normal Portland cement type I.


The same sources of cement and sand were used for all the
specimens to eliminate all risk of material discrepancy.
2.2

Sample Preparation

The specimens used in the UCS tests were prepared in


cylindrical PVC split molds to facilitate the extraction of the
sample after formation. For a given test, the material quantities
were determined based on the target fiber content, cement
content, and sand density. Initially, the sand and cement were
mixed in dry conditions before adding 5% by weight of water
necessary for the hydration of the cement and blending of the
mixture. The fibers were then mixed thoroughly with the sandcement to obtain a final homogenous mix with well-distributed
and untangled fibers. It is to be noted that all the mixing was
done manually since the use of a mechanical mixer could result
in tangling and clodding of the fibers and their segregation from
the soil mixture. Each layer was then compacted into the mold
to the required height under the effect of a compaction tool
which was specifically designed for the purpose. The top
surfaces of the 1st and 2nd layers were scratched prior to putting
the new material for the subsequent layer in order to obtain, to
the extent possible, a homogenous specimen and eliminate the
risk of weak shear planes at the contact surface between two
layers.
A curing time of 8 1 days was chosen to allow the cement
enough to time to set. Since the curing time is not a parameter
which was studied in the testing program, the period of 8 days
was chosen as an average time which provides a significant
period for curing without unduly prolonging the overall time
needed for each test.
2.3

Unconfined Compression Strength Tests

The UCS tests were performed according to ASTM D2166 with


specimens having a diameter of 5.5 cm and a height of 11 cm
giving an acceptable height to diameter ratio of 2. The machine
used in the tests is a HUMBOLDT HM-3000 loading frame
fully automated and computer software-controlled. The vertical
deformation is recorded by an LVDT, while the resisting axial
load is recorded by load cells of different capacities. The rate of
application of the strain is 0.05 cm/min. The data were recorded
automatically every 4 seconds and the test was continued until
failure occurred or when the axial strain exceeded 15%.
3

TEST RESULTS AND ANALYSIS

Results from twenty unconfined compression tests on


fiber/cement reinforced sands are presented in this paper. The
tests were restricted to lightly cemented sands (cement content
= 0.5% and 1.0%) that were reinforced with fibers of different
lengths (6mm, 12mm, and 20mm) at different fiber contents
(0%, 0.25%, 0.5%, and 1.0%). The analysis of the tests includes
an assessment of the stress-strain behavior and the dependency
of the unconfined compressive strength on the reinforcement
parameters (fiber content, fiber length, and cement content).

2594

Figure 1. Polypropylene fibers used in the experiments.

3.1

Stress-Strain Response

The stress-strain response of specimens that were stabilized


with a cement content of 0.5% is presented in Figs. 2a, 2b, and
2c for fiber contents of 0.25%, 0.5%, and 1.0%, respectively.
The response of specimens that were reinforced with a cement
content of 1.0% is similarly presented in Figs. 3a, 3b, and 3c.
On each of the plots, stress-strain curves are presented for
different fiber lengths (6mm, 12mm, and 20mm) and for the
specimen that was prepared with no fibers.
For specimens that were reinforced with a cement content of
0.5% (Fig. 2), the stress strain curves indicate a consistent
increase in stress with strain up to a maximum peak stress value
at which failure occurs. The value of the peak and the post peak
behavior are a function of the cement content, fiber content, and
fiber length. The failure mode as indicated by the value of the
strain at failure and by the post peak response is found to be
more ductile as the fiber content increased from 0% to 1.0%. In
addition, for a given fiber content, ductility was found to
improve as the length of fibers increased from 6mm to 20mm.
For the higher cement content of 1.0% (Fig. 3), the behavior
of the composite specimens was found to be more brittle
compared to their lightly cemented counterparts. The inclusion
of fibers added some ductility to the mode of failure, but this
effect was minor for the smaller fiber contents (0.25% and
0.50%). The improvement in the mode of failure was only
evident in the higher fiber content of 1.0% at all fiber lengths
and for the intermediate fiber content of 0.5%, but only at the
larger fiber length of 20mm.
3.2

Effect of Fiber/Cement on Stiffness

The stress-strain response at the onset of loading in Figs. 2 and


3 could be used as a measure of stiffness for the fiber/cement
reinforced specimens. For the smaller cement content, results on
Fig. 2 indicate that the stiffness of the specimens was not
affected by the addition of fibers except for cases involving the
longest fibers (20mm) with fiber contents of 0.25% and 0.50%
where the stiffness was found to be improved. For cases
involving fibers with a high fiber content of 1.0%, no
improvements were observed in the stiffness, irrespective of the
fiber length.
A slightly different behavior was observed for the higher
cement content of 1.0% where slight improvement in stiffness
were observed for the shorter fibers at the smaller fiber
contents, with the improvements in stiffness vanishing for the
longest fiber and the highest fiber contents, where slight
reduction in stiffness was actually observed. This indicates that
fibers could result in a softer initial response for higher cement
contents, higher fiber content, and longer fibers.

Technical Committee 211 / Comit technique 211

200

0.50% Cement
0.25%Fibers

160

Compressive Stress (kPa)

Compressive Stress (kPa)

200

120
80
40

L=20mm

L=6mm
L=12mm

0
3.0

4.0

5.0

6.0

7.0

8.0

Compressive Stress (kPa)

2.0

0.50% Cement Axial Strain (%)


0.50%Fibers

160
120
80

L=20mm

40
0
200 0.0

Compressive Stress (kPa)

1.0

160

NoFibers
1.0

2.0

L=6mm
3.0

4.0

L=12mm
5.0

120

6.0

7.0

8.0

L=20mm
L=12mm

80

L=6mm
NoFibers

0
0.0

1.0

2.0

80
40
0
2000.0
160

3.0

4.0

L=12mm

NoFibers
1.0

2.0

1.0% Cement
0.50%Fibers

5.0

6.0

7.0

8.0

4.0

5.0

6.0

7.0

8.0

5.0

6.0

7.0

8.0

7.0

8.0

Axial Strain (%)

80

L=12mm

40

L=6mm

160

1.0

2.0

1.0% Cement
1.0%Fibers

120

3.0

4.0

Axial Strain
(%)
L=20mm

L=6mm
L=12mm

80
40

NoFibers
0.0

Figure 2. Stress-strain response for cement content of 0.5%.

3.0

L=20mm

120

Axial Strain (%)

3.3

L=20mm

NoFibers

0.50% Cement Axial Strain (%)


1.0%Fibers

40

120

0
200 0.0

Compressive Stress (kPa)

Compressive Stress (kPa)

200 0.0

160

1.0% Cement
0.25%Fibers

1.0

2.0

3.0

4.0

5.0

6.0

Axial Strain (%)

Figure 3. Stress-strain response for cement content of 1.0%.

Effect of Fiber Content on UCS

3.4

For a given cement content, the stress-strain curves in Figs. 2


and 3 indicate that the unconfined compressive strength
increases as the fiber content increases. The unconfined
compressive strength for each test was computed and plotted as
a function of the fiber content in Figs. 4a and 5a for cement
contents of 0.5% and 1.0%, respectively. For the two cement
contents and for all fiber lengths, results indicate a consistent
increase in the unconfined compressive strength with fiber
content. For a cement content of 0.5%, the UCS increased from
about 15 kPa (no fibers) to about 40 kPa (1.0% fibers) for the
shortest fiber length of 6mm, and from 15 kPa (no fibers) to
about 160 kPa (1.0% fibers) for the longest fiber length of
20mm. For the larger cement content of 1.0%, the UCS
increased from about 50 kPa (no fibers) to about 112 kPa and
178 kPa, for the shortest and longest fibers at 1.0% fiber
content, respectively.
In order to obtain a quantitative measure of the degree of
improvement in the unconfined compressive strength, the ratio
of the UCS with fibers to the UCS without fibers was computed
and plotted versus the fiber content in Figs. 4b and 5b. These
results indicate that the cement content played a significant role
in defining the improvement ratio, with the ratio varying from 2
(smallest fiber content and fiber length) to 10 (largest fiber
content and fiber length) for a cement ratio of 0.5%, and from
about 2 to 4 for the larger cement ratio of 1.0%.

Effect of Fiber Length on UCS

The effect of the fiber length on the stress-strain response and


on the improvement in the UCS is evident in Figs. 2 to 5 and is
found to be dependent on the cement content. For a cement
content of 0.5%, as the fiber length increases, the unconfined
compressive strength increases and the strain at failure
increases, indicating improved ductility. The effect of fiber
length was found to be more evident at higher fiber contents
compared to lower fiber contents. For example, for the small
fiber content of 0.25%, the improvement ratio in the UCS
increased slightly from 1.7 to 2.6 (for fiber length of 6mm and
20mm) compared to the dramatic increase from 2.6 to 10.6 (for
fiber length of 6mm and 20mm) for the larger fiber content of
1.0%.
For the larger cement content of 1.0%, the effect of fiber
length on the unconfined compressive strength becomes
smaller. For the smaller fiber contents of 0.25% and 0.5%, the
difference in the measured values of the UCS is relatively
insignificant, with improvement ratios varying in the narrow
range of 1.6 to 2.0 (fiber content of 0.25%) and 2.1 to 2.15
(fiber content of 0.50%) for the shortest and longest fibers,
respectively. For the largest fiber content of 1.0%, the
improvement ratio increases from 2.2 to 3.7, as the fiber length
increases from 6mm to 20mm.

2595

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Cement Content = 1.0%

160

Unconfined Compressive Strength (kPa)

200

Fiber Length = 6mm


Fiber Length =12mm
Fiber Length =20mm

120
80
40
0
12 0.00
10

0.25
0.50
0.75
Fiber Length = 6mm
Fiber
Length
=12mm
Fiber Content, FC (%)
Fiber Length =20mm

1.00

UCS (Fibers) / UCS (No Fibers)

UCS (Fibers) / UCS (No Fibers) Unconfined Compressive Strength (kPa)

Cement Content = 0.5%

8
6
4
2

0.25

0.50

0.75

120
80
40
0
12 0.00
10

0.25
0.50
0.75
Fiber Length = 6mm
Fiber Length
= 12mmFC (%)
Fiber Content,
Fiber Length = 20mm

1.00

8
6
4
2

0.00

1.00

0.25

0.50

0.75

1.00

Fiber Content, FC (%)

Fiber Content, FC (%)

Figure 5. Improvement in UCS for cement content of 1.0%.

Figure 4. Improvement in UCS for cement content of 0.5%.

of 0.25% and 0.50%, where the stiffness was found to be


improved. For the higher cement content fibers could
result in a softer initial response particularly for higher
fiber contents and longer fibers.
3. For the both cement contents used, results indicated a
consistent increase in the unconfined compressive
strength with fiber content. The cement content played a
significant role in defining the improvement ratio of the
UCS, with the ratio varying from 2 (smallest fiber
content and fiber length) to 10 (largest fiber content and
fiber length) for a cement ratio of 0.5%, and from about
2 to 4 for the larger cement ratio of 1.0%.
4. For a cement content of 0.5%, as the fiber length
increased, the UCS increased and the strain at failure
increased, indicating improved ductility, with the effect
of fiber length being evident at higher fiber contents
compared to lower fiber contents. For the larger cement
content of 1.0%, the effect of fiber length on the
unconfined compressive strength was less significant.

Effect of Cement Content on UCS

The range of the cement content that was chosen in this study
(0.5% to 1.0%) is indicative of lightly-cemented sands.
However, the results presented in Figs. 2 to 5 indicate a clear
difference in the performance of the composite specimens that
were stabilized with 0.5% cement and specimens stabilized with
1.0%, particularly with regards to the contribution of the fibers
to the improved compressive strength.
For specimens that were not reinforced with fibers, the
increase in cement content from 0.5% to 1.0% increased the
unconfined compressive strength from 15 kPa to 50 kPa. With
the addition of fibers, results showed that the UCS could be
improved by more than 10 times for a cement content of 0.5%
but only to 3.7 times for the cement content of 1.0%, indicating
a decreased relative efficiency of the fibers at improving the
compressive strength as the cement ratio increases.
It should be noted however that the actual maximum value
(largest fiber content and fiber length) of the unconfined
compressive strength was still higher (about 185 kPa) for the
cement content of 1.0% compared to the maximum value (159
kPa) measured for the cement content of 0.5%. For the smaller
fiber contents and fiber lengths, the values of the UCS for the
cement content of 1% were all higher than those of the 0.5% at
the same fiber content and fiber length, indicating that the
magnitude of the improved UCS was larger for the higher
cement content.
4

160

Fiber Length = 6mm


Fiber Length = 12mm
Fiber Length = 20mm

0
0.00

3.5

200

CONCLUSION

Based on the results of 20 unconfined compression tests that


were conducted in this study on fiber-reinforced lightlycemented sands, the following conclusions can be drawn:
1. The behavior of specimens with higher cement contents is
more brittle compared to specimens with lower cement
contents. However, brittleness decreased with the
inclusion of fibers and the energy absorption capacity
increased as the fiber content and length increased.
2. For the smaller cement content, the stiffness of the
specimens was not affected by the addition of fibers,
except for the cases of 20mm fibers with fiber contents

2596

ACKNOWLEDGEMENTS

The authors would like to acknowledge the support of the


Lebanese National Council for Scientific Research (LNCSR)
for funding this research program.
6

REFERENCES

Maher M.H. and Ho Y.C. 1993. Behavior of fiber-reinforced cemented


sand under static and cyclic loads. American Society for Testing
and Materials 16, 330-338.
Kaniraj S.R. and Havanagi V.G. 2001. Behavior of cement-stabilized
fiber-reinforced fly ash-soil mixtures. Journal of Geotechnical
Engineering 127, 574-584.
Consoli et. al. 1998. Influence of fiber and cement addition on behavior
of sandy soil. Journal of Geotechnical Engineering124, 1211-1214.
Consoli et. al. 2002. Engineering behavior of a sand reinforced with
plastic waste. Journal of Geotechnical Engineering 128, 462- 472.
Sobhan K. and Mashnad M. 2002. Tensile strength and toughness of
soil-cement-fly-ash composite reinforced with recycled highdensity polyethylene strips. Journal of Materials in Civil
Engineering 14, 177-184.

Conservatoriumhotel Amsterdam, geotechnical design and monitoring


Conservatoriumhotel Amsterdam, conception gotechnique et instrumentation
Stoel van der A.E.C.

CRUX Engineering BV Amsterdam & Universiteit Twente

Vink D., Bouma J.

CRUX Engineering BV Delft

ABSTRACT: This article deals with the geotechnical design of the building pit and the foundation restoration of the
Conservatoriumhotel Amsterdam. It involves a selection of results of analytical and Plaxis calculations and damage predictions
regarding vibrations and settlements. The second part of the paper deals with the execution of the works and the important role of a
pro-active monitoring-system and careful communication with all stakeholders. It briefly outlines the monitoring plan, type and
results of the monitoring and the risk management during implementation. Some measurement results are compared with predictions
and two calamities that occurred during the construction will be specifically addressed. This will clearly show the added value of
monitoring and active risk management, that eventually has led to the successful completion of this project in 2011.
RSUM : Cet article traite de la conception gotechnique de la tranche couverte du btiment et reprise en sous uvre des
fondations du ConservatoriumhotelAmsterdam. Il s'agit d'une slection de rsultats de calculs analytiques et Plaxis et des prvisions
concernant les dommages dus aux vibrations et les tassements induits. La deuxime partie de l'article traite de l'excution des travaux
et le rle important que la surveillance pro-active et de la communication prudent avec toutes les investisseurs. Le plan
dinstrumentation est brivement dcris ainsi que le type et le suivi, des rsultats et de la gestion des risques lors de la mise en uvre.
De plus les rsultats des mesures sont compars aux prvisions. Deux sinistres qui se sont produits lors de la construction seront
abords. La valeur ajoute de linstrumentataion est montre ainsi que la gestion active des risques qui a finalement conduit la
russite de ce projet en 2011.
KEYWORDS: geotechnical design, building pit, Plaxis, monitoring, restoration

INTRODUCTION

In Amsterdam, at Van Baerlestraat 27, the stately listed building


of the Rijkspostspaarbank (Imperial Bank, see Figure 1) is
situated. It was originally build between 1899 and 1901 by
Imperial architect, D.E.C Knuttel. The building was previously
reassigned to function as Sweelinck Conservatory and was since
2008 radically converted into a luxurious hotel, including over
9000 m of five star hotel and 85 parking facilities. This
conversion has been one of the most expensive (more than 30
million euro) hotel refurbishments ever.
This renovation required a number of radical structural
changes that led to a complex task with regard to the
geotechnical design. The most important of these tasks was
realising a two level parking / basement including a (-3)
swimming pool in the courtyard of the existing building.
Special attention had to be paid to the many historical details in
the buildings that were incorporated in the new design and had
to be preserved. In addition, the building location is flanked by
two tram lines and the Van Gogh Museum and Rijksmuseum
and the Royal Concert Hall, resulting in numerous logistic
restrictions.
2

Figure 1. Aerial photo of the building Stradivarius

SOIL CONDITIONS

At the location, the typical Amsterdam soil profile (Figure 2) is


found. The top layer of the first meters below surface level
consists of Anthropogenic sand. Below this top layer the
Holocene deposits are found until a depth of about 10-15 m
below surface level. The Holocene formation can be divided
(from top to bottom) into peat, clay, silty sand, clay and peat.
The Holocene lies on top of the Pleistocene sands which are
divided by an intermediate silty, clayey sand layer. The phreatic
water level is found about 0.4 m below surface level.

Figure 2. Amsterdam soil conditions

2597

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

CONSTRUCTION AND BUILDING LAYOUT

In Figure 3,4 and 7 the lay out of the building and building
pit are shown. Because of the very deep excavation next to the
pile foundation of the existing building, much precaution had to
be paid to settlements and angular distortion (damage) of this
listed building. Also, the bending moments in the existing
wooden piles were a major concern. This resulted in a staged
excavation as shown in Figure 4. Note the different excavation
levels of the two excavations A and B, which cause an
asymmetrical load situation and displacements. An extra
complication was that the building site could only be accessed
through a narrow entrance in the eastern part.

Figure 6. Vertical deformations in building, section C-C.

Because of the proximity of the existing building, full of


marble stairs and exquisite tiles that had to be preserved, all the
applied foundation system were vibration free: the sheet piles
AZ26 were pushed and the Hek-piles were screwed.
Table 1. Results damage prediction sect. C-C (based on BRE regulation
/ Netzel 2010)
uv
L/H

Eh
/L
tot
Damage class
m
%
Slight (minor
-0.0113
1:1605 0.066 0.0002 0.00088 aesthetic damage)
L/H = Ratio depth/height of the building
= Vertical deflection
= Relative angular distortion
h = Horizontal strain
tot = Total building strain
uv = vertical displacement

Figure 3. Section C-C: new situation (note swimming pool at -3)

5
5.1

General

Because of the sensitive nature of the existing building and the


high complexity of the execution of the works, an extensive
monitoring program was implemented. Figure 7 shows a
general overviewincluding the position of the levelling point
(bolts) and inclinometers. By measuring the inclination of the
sheet piles, an excellent comparison could be made between
predictions and execution for all stages of the works. The
leveling points were mainly used to verify whether the
inclination indeed resulted in the predicted building
deformation. This is particularly useful because some deviation
from the predictions is not uncommon.

Figure 4. Schematic cross-section C-C, NAP = reference level

MONITORING

GEOTECHNICAL DESIGN

Because of the asymmetrical excavation and the need to assess


soil and building deformations, 2D FEM Plaxis calculations
were performed. In Figure 5 an example of the used model for
section C-C is shown(see Figure 7). Based on the deformations
and stresses resulting from this model, the vertical deformations
and inclination of the existing building were determined, see
Figure 6. The sheet piles and struts were designed in such a
iterative way that the damage prediction resulted in an
acceptable damage class(see Table 1).

Figure 5. Plaxis geometry section C-C, shallow excavation NAP -6,1m.

2598

Figure 7. Overview leveling points (bolts) and inclinometers.

Technical Committee 211 / Comit technique 211


th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

The monitoring proved to be particularly valuable because of


the occurrence of two incidents during construction.
5.2

Incident 1

Incident 1 occurred during the excavation of building pit A. The


situation at that time is illustrated by Figure 8 (by a recommenddable, remotely operable, permanent webcam!) and Figure 9.

In order to be able to inspect the girder and struts, the water


table had to be lowered further than calculated (-6,5m without
strut). It was found that, probably because of excavating under
the girder, its consoles were removed, thus causing it to hang
by a tread. Because of the
extensive monitoring and
modelling, an alternative model could be made very quickly
based on actual deformations, from which it could be concluded
that the deformationsresulting from the mitigating measures
stayed within acceptable boundaries, see Figure 11. Thus,
within three weeks and without significant delay in
construction, the strut and girder could be repaired allowing for
further excavationof the pit.

Figure 8 Overview building pit (at the time of incident)

Figure 11. Inclinometer & predictions after incident at girder/strut -5.0m

5.3

Figure 9. Top view building pit (at the time of incident; Google)

To limit the displacements the execution sequence has been


that, after placement of the sheet piles and the excavation for the
NAP -1,5m and -5,0m struts, the water table was first set up
again before wet excavation to NAP -10,5m commenced. After
hardening of the underwater concrete floor, the water table was
lowered again. It was then timely discovered that the -5,0m
struts and girder were not in position, see Figure 10. Pumping
was stopped immediately to access the situation.

Figure 10. Situation at 2nd level girder (collapsed during excavation)

Incident 2

Incident 2 occurred during the excavation of the entrance at


building pit B. The situation is illustrated by Figures 12 (before)
and 13 (after excavation). At the left side of the pictures, a listed
building at Paulus Potterstraat 44 (PP44) is located. During the
excavation of building pit B, this building started to settle, as
can be seen in Figure 15. The main concern however was that
the side at the entrance settled significantly more than the
opposite side, thus potentially causing damage.
In November 2009 the monitoring showed that the
settlement rate increased alarmingly. The frequency of
monitoring was immediately increased and owner, contractor,
consultants, insurer and the municipality intensively discussed a
solution. This was found in a combination of allowing more
deformation as long as no damage resulted from frequent
building inspection and, more importantly, the remedial
measure of pre-stressing the NAP -4,0m strut with 150 kN/m.
The last measure resulted in a stabilisation of deformations that
has held up to now. No damage to PP44 was detected.

Figure 12. Building pit at PP44 before incident

2599

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
th

Proceedings of the 18 International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

led to significant additional costs and/or delays. The


investments that have been necessary to achieve this have been
minor compared to the potential cost of the delay that would
have resulted from the lack of information without such a
system.
8

ACKNOWLEDGEMENTS

The author gratefully acknowledges the willingness of the client


to invest in a sound geotechnical design and a pro-active
approach to monitoring the geotechnical execution of the works.

Figure 15. Finished construction atrium new-old

Figure 13. Building pit at PP44 during incident

Figure 16. Finished inside construction

Figure 14. Displacement of building Paulus Potterstraat 44

5.4

Figure 17. Finished outside construction

Costs

The (interest)cost of a delayed opening of the hotel have been


estimated at least at 50.000,- a week. The additional costs of
the additional monitoring and damage risk assessment have
however, as compared to a basic monitoring program, not
exceeded 100.000. When no information would have been
available after one of the calamities, construction could have
been stopped for months.
6

Figure 18. Finished swimming pool at -3 level

COMPLETED WORKS

The works have been completed in 2011, not only resulting in


one of the most luxurious hotels in Amsterdam, but also in a
construction of both geotechnical as well as aesthetic beauty. It
won the 2012 Dutch Renovation Golden Phoenix award for
the most effective reuse of existing property stocks. An
impression of the completed work is given in Figure 14 to 16.
Figure 19 shows the impressive swimming pool at -3 level, right
next to the over 110 years old existing pile foundations.

ACKNOWLEDGEMENTS

The authors wish to express their gratitude to the client Alrov


Group and IQNN Vastgoed, structural engineer Van
RossumRaadgevendIngenieurs, contractor Strukton NV and
foundation subcontractor Van t Hek for their contributions to
this paper.
10 REFERENCES

CONCLUSIONS

The renovation of the former Rijkspostspaarbank Amsterdam to


luxury five star Conservatoriumhotel has been very successful
from a geotechnical perspective. Through a sophisticated
geotechnical design and a detailed risk analysis of the
distortions of the building before construction, combined with a
meticulous and proactive monitoring and excellent
communication during execution, two major incidents have not

2600

Netzel, H.D., Building Response Due to Ground Movements, Delft


University Press, March 2009, Pages 320, ISBN978-1-58603-995-0
Van der Stoel, A.E.C. D. Vink & J. Bouma, Van
RijkspostspaarbanknaarvijfsterrenConservatoriumhotel; deel 1
Amsterdam,
geotechnischontwerpConservatoriumhotel
Geotechniek, jaargang 16, no. 4, augustus 2012 (in Dutch)
Van der Stoel, A.E.C. D. Vink & J. Bouma, Van
RijkspostspaarbanknaarvijfsterrenConservatoriumhotel; deel 2
Monitoring en uitvoeringConservatoriumhotel Amsterdam,
jaargang 16, no. 5, december 2012 (in Dutch)

Impact of the soil-stabilization with lime


Impact de la stabilisation des sols la chaux
Szendefy J.
BUTE, Budapest, Hungary

ABSTRACT: The protection of environment and economical consideration demands giving the usage of internal soils preference over
external materials. Soil-stabilization with lime increase the bearing capacity and treats the properties of cohesive soils due to chemical
reactions. During a soil-stabilization with lime, the Ca2+ ions introduced into the soil attach to the surface of the clay particles,
replacing the cations located there previously. Due to the high charging of the Ca2+ ions, the clay particles coagulate. The coagulation
exerts a decisive effect on the soil structure, resulting the plasticity index in a shifting towards a certain zone. This effect leads to the
elimination of the silt and clay particles, resulting in an intensive reduction surface of the particle of the stabilizations. Due to the
coagulating particles, the friction angle increases, and also the value of cohesion becomes higher. The addition of lime results
consequently in a growth of the compression strength of the soil, leading to an increased load bearing capacity. In order to define the
internal friction angle () and the cohesion (c), triaxial compression test were done with ten different kind of soils. The soilstabilization with lime were made with 2 %, 4 %, and 6% additional lime, the temporal aspect has been investigated in ages of 1, 7,
and 28 days. In case of admixing an ideal quantity of lime, the soils are shifting towards the value of =40o. The extent of growth of
the cohesion did not show as clear tendency as the friction angle, but give back a significant increasing.Contrary to the literature, I
regard to remark that the positive impacts of soil-stabilization with lime is basically a result of the cation exchange, while the
puzzolan reaction playing only a secondary role. It is also important to remark that the lime stabilization of soils is not only suitable
for the drying up of soaked soils. With the application of a planned and properly executed stabilization, load bearing capacity of
ballast materials can be achieved with local soils regarded or qualified as unsuitable. Accordingly, the ballast and pavement layers can
be effectively reduced, and the quantity of delivered external materials minimized.
RSUM : Les considrations environnementales et conomiques font prfrer lutilisation des matriaux locaux aux matriaux
extrieurs transports sur place. Avec ses processus chimiques, la stabilisation des sols la chaux permet une augmentation de la
portance des sols argileux et une amliore de leurs proprits. Lors des processus de stabilisation, les ions Ca2+ remplacent les cations
qui sattachent la surface des grains dargile. Les grains dargile sagglomrent cause de la haute charge des ions Ca2+.
Lagglomration provoque un changement important dans la structure du sol et, en consquence, lindice de plasticit du sol se dcale
vers une certaine zone. Cet effet limine les grains dargile et de limon diminuant ainsi radicalement la surface des grains du sol
stabilis. cause de lagglomration des grains, les valeurs de langle de frottement interne et de la cohsion augmentent. En ajoutant
de la chaux, la compressibilit des sols diminue, en consquence la portance des sols augmente. Afin dtablir le changement des
valeurs de langle de frottement interne () et de la cohsion (c) nous avons effectu des essais triaxiaux sur diffrents sols. Nous
avons stabilis les sols en ajoutant 2%, 4% et 6% de chaux et nous avons mesur leffet du temps avec des essais effectus lge de
1 jour, 7 jours et 28 jours. Dans le cas dune addition de quantit optimale de chaux la valeur de langle de frottement interne est de
40. Dans le cas de la cohsion, il a t impossible de dmontrer une tendance, mais une augmentation importante a pu tre observe
dans tous les cas. Contrairement la bibliographie technique sur les essais lors de la stabilisation des sols la chaux, lagglomration
cause par lchange de cations a un rle dominant, tandis que les ractions pouzzolanes ont plutt un rle secondaire. Il est important
de souligner galement que la stabilisation des sols la chaux na pas pour seule application le schage des sols mouills ou
humidifis. En utilisant une formule bien tablie, une portance quivalente au gravier peut tre atteinte avec les sols locaux, ainsi les
sols locaux qualifis de non-utilisables peuvent tre finalement convenables pour la construction routire. En consquence lpaisseur
de lassises de chausse et celle de la structure de la chausse peuvent tre diminues efficacement, et la quantit des matriaux
extrieurs transports peut tre minimise.
KEYWORDS: soil stabilization with lime, cation exchange, bearing capacity
1

INTRODUCTION

The road vehicle and railway traffic are increasing worldwide


that require a permanent extension of the infrastructure. In
addition to the extension, the high load bearing and economical
operation are required in new infrastructure and rehabilitation
investments. To satisfy the above demands, the expectation
towards pavements, bedding and subsoil is increasing
continuously. Similar tendencies can also be experienced at
industrial and commercial buildings e.g. their industrial floors,
service-roads and parking area.
According to the Hungarian code (T 2-1-222-2002), the
above high requirements can essentially be satisfied by granular

2601

soils; the application of transient and cohesive soils are limited.


Another argument prevailing against the use of locally transient
and cohesive soils is the fact that their characteristics are
extraordinarily influenced by rainfall: their workability is
decreasing rapidly and their load bearing capacity becomes
practically zero. Consequently, they may upset a strict time
schedule due to the drastically abbreviated execution times, so
their application result significant risks.
The omission of local materials and the delivery and use of
extraneous ones arises questions related not only to costs but
also to environment protection and to the preservation of
national property. The excavation of extraneous materials cause

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

damages on nature, and the transportation of thousands and


millions m3 of materials to infrastructure and industrial plants
also results considerable load to the environment. Beyond the
above effects, this significant transportation also abrades,
damages and amortizes the already existing infrastructure.
2

GENERALLY ABOUT THE SOIL-STABILIZATION


WITH LIME

The soil-stabilization with lime is a successful and widespread


procedure for several decades in the USA and Western Europe,
while this technology is applied in Hungary and in the
neighboring countries only for the drying of sodden earthworks.
The research work was performed with Hungarian soils, with
the aim of supporting the work of designers and contractors to
use local soils as high bearing capacity layer and to reduce the
thickness of bedding and floors.
Mainly cohesive soils are used to be stabilized with lime,
however transient and clayey silty gravely soils can be
stabilized satisfactory based on my research. Therefore I
modified a figure of a generally used literature (Mitchell &,
Hooper, 1961.) about the types of suitable soils for stabilization
with lime with the modern machineries.
My suggestion based on my laboratory tests and field
experience is shown in red in Figure 1.
The stabilized layer can be built of materials mixed in situ or
at a distant site, however the soil is generally not moved but
worked in at the location of its original place, transforming it
into a layer of sufficient load bearing capacity. Bedding and
pavement layers are built on to the stabilized layer to achieve
the full layer system of a road, runway or industrial floor.
However in case of roads exposed to a lower traffic, forestry or
agriculture roads, the stabilized layer may also used as
pavement.
During the construction of the stabilized layer, the required
quantity of lime is spread over the previously executed surface
by a truck provided with a dosing head. The quantity of the lime
is 8-60 kg/m2 depending on the thickness of stabilization and on
the type of soil. Subsequently the soil is mixed with the lime by
a remix machine. In case the process of stabilization requires
water, the modern remix machinery is able to add water through
the mixing head. Following mixing, the layer is compacted, and
the required ground level provided.
The lime improves the soil characteristics through chemical
reactions. Chemical reactions start with the mixing and may
take several decades. The chemical reactions during
the soil stabilization with lime are as follows:

dissolution of the lime reduction of the water


content;

cation exchange-coagulation;

puzzolan reaction-cementation;

carbonatization development of limestone.

During the research 21 soils were measured in the laboratory


of Budapest University of Technology and Economics. The
origin soil properties of the measured soils are listed in the
Table 1. For the verification of the laboratory tests some in-situ
stabilization were analyzed as well.

Figure 2. Red spots show the origin of the soils for laboratory
measurements, blue spots show in-situ bearing measurements

After the literature 2, 4, 6% quick lime was added for the soils.
The quantity of the lime is the percent of the dried soil. Some
measurements was done with hydrate lime as well, but the final
determinations are given in quick lime, because it contains only
the agent (CaO), which is the unit of necessary active agent
My laboratory and field work included soil mechanics
parameters (plasticity, grain size, permeability, Proctor-test,
swelling), bearing capacity measurements (triaxial compression
test, CBR test, durability test) and geological tests (X-ray,
derivatograph, pH).
The soils were dried to the laboratory air humidity, than
were determined their moisture content and were determined the
quantity of the added lime. The lime and the required water
were mixed during 15 minutes than left it rest 45 minutes. After
resting they were remixed and compacted with Proctormachine. After compaction the mixtures were left for curing in
waterproof package.
To measure the time effect of soil-stabilization with lime I
made measurements 1 hour, 1 day, 3 days, 7 days, 14 days and
31 days after compaction.
I can show only a part of my research because the compass
of the article.
4

ALTERATION OF THE SOIL STRUCTURE

The clue of soil-stabilization with lime is an alteration of the


structure of soil caused by lime. Through the admission of lime,
the cohesive soils become crumbly, characterized by a reduced
plasticity.
4.1.

Figure 1. Delimitation of grain size distribution areas from the point of


view of soil-stabilization with lime (Kzdi, 1967., Szendefy, 2008.)

DELINEATION OF MY RESEARCH

Alteration of plasticity

Several researchers pointed out that the plasticity index of


soils decreases even if a very low quantity of lime is added
(Wang, 1963., Jan, & Walker, 1963., Kzdi,1967., Nemesdy,
1983.)
They explain the reduction of the plasticity index with a
slight reduction of the liquid limit and with the drastic increase
of the plastic limit.
As I saw the clue of stabilization in the alteration of the soil
structure, I have performed experiments to find the exact
mechanism of changes of the plasticity index, and processed
literature data.
During the experiments, several soil types were tested
(Ip = 6.4 - 69.2 %), admixing 2 % - 4 % - 6 % of lime at
different ages.

2602

Technical Committee 211 / Comit technique 211

with lime, the soil particles settled very rapidly. The settling
process took 5-15 minutes.

Table 1. Origin soil physics parameters


Sign
of soil

Plasticity

Grain size

index

< 0.1mm

[%]

[%]

Friction
angle

Cohesion

[kPa]

[]

T1

22,0

88,4

10.5

61.7

T2

18,5

76,6

22.4

88.3

T3

50,0

96,4

11.9

117

T4

12,3

87,6

25.6

72.9

T5

33,5

98,7

11.4

37.6

T6

61,5

99,3

18.2

123

T7

43,4

98,8

22.8

114.8

T8

22,9

90,5

T9

36,8

T10

69,2

99,5

T11

34,4

94,1

T12

47,5

T13

70,2

T14

41,2

T15

38,5

T16

26.9

124

14.5

104.5

34,8

92,8

20.9

82.5

T17

24,8

88,2

T18

8,8

76,1

T19

29,1

91,8

T20

6,4

74,2

T21

27,0

Figure 3. Alteration of the plasticity index plotted against lime dosage

This rapid sedimentation process produces the grain size


distribution curve characterizing soils treated with lime. The
section situated above hydrometration of the curve shows a
picture identical with that of an untreated soil, plunging
however at the section below 0.063 mm. The stabilized soils
maintain this character later on, with the only difference that
also the proportion of particles of above 0.063 mm becomes
slightly higher. These alterations are visualized in Figure 4.
As was suspected this intensive coagulation that is not
characteristic for the hydrometration of soils of natural
stratification is triggered by the cation exchange taking place
during lime stabilization. In order to prove it, the
hydrometration was performed with a 15 months old sample, by
means of admixing a significant quantity of dispersive material
(sodium hexametaphosphate) to the mixture prepared in the
measuring glass. During the test, a long lasting, continuous
precipitation could be experienced as usual for natural soils. The
obtained curve of grain size distribution is indicated in Figure 4
in purple color, showing a high conformity with the curve of an
untreated soil.

The analysis of the results shows that the liquid limit (wl) is
shifting towards the zone of wl = 35 45 %, irrespective of the
initial values. The high values of the liquid limit decrease
drastically, those situated near the zone practically stagnate, and
a slight increase was experienced at the low liquid limit values.
Similarly to the plastic limit (wp) a tendency of shifting towards
a certain zone can be experienced, similarly to the behavior of
the liquid limit. This zone is situated at wp = 25 35 %. The
plastic limit is generally increasing, however in case of fat clays
having high plasticity limit stagnation or decrease can be
experienced.
The alteration of Atterberg-limits in the value of the
plasticity index (Ip) also leads to shifting towards a zone (Figure
3). The value of plasticity index is shifting to the zone delimited
by Ip = 5 15 %, corresponding to very fine sand and silt type
soils, according to the description of the bound soils.
An investigation of the temporal aspects of the impact
shows that the above tendencies appear during the first hour
following mixing, however the most extreme alteration of the
parameters could be measured appr. on day 7. The following
period was characterized by stagnation, with slight shifts in case
of certain soils toward the properties of an untreated soil.
4.2.

Alteration of the grain distribution

The process of increased crumbliness and the more granular soil


experienced at the plasticity index were tried to be delimited by
an investigation of the grain size distribution. During the grain
size distribution test and hydrometration of soil samples treated

Figure 4. Alteration of the grain size distribution caused by a lime


treatment of medium clay soil

It can be observed with the lapse of time that the more


frequent presence of particles of above 0.063 mm of particle
diameter can be explained with two other chemical processes of
the lime stabilization: the cementation and carbonatization. The
investigation of particles of above 0.063 mm by means of
microscope and X-ray diffraction demonstrated the presence of
aluminates and silicates leading to cementation, together with
carbonatized limestone particles.
The results obtained from the investigation of plasticity and
the grain size distribution show that the particles of a treated
soil coagulate, cohere, resulting from the cation exchange. This
coagulation creates a strong connection between the soil
particles that cannot be dissolved by water, rendering it durable
in the soil.
The coagulation is able to eliminate one of the biggest
disadvantages of bound soils, i.e. the relatively high particle
surface binding high quantities of water. The high level of water
intake reduces namely the shearing strength and consequently
the load bearing capacity of the soils.

2603

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

ALTERATION OF THE LOAD BEARING CAPACITY


OF THE SOIL

Drying and working in soils are facilitated by the addition of


lime which is very important to constructers. Therewith the
important to designers is an intensive increase of the load
bearing capacity based on the effect of lime exerted on the soil
structure. The load bearing capacity of soils is determined by
their parameters related to shearing strength and deformation.
4.3.

Alteration of the parameters related to shearing strength

In order to define the internal friction angle () and the


cohesion (c), I have performed triaxial compression test over a
long range of the above mentioned soils. Similarly to the
previous procedures added lime was 2 %, 4 %, and 6 %, the
temporal aspect has been investigated in ages of 1, 7, and 31
days.
The evaluation of the measurement results shows a growth
of the values of friction angle and cohesion. At the 1 day old
samples the growth showed a fully disordered picture, then the
alteration of value shift towards a single zone as the
boundaries of consistency at the age of 31 days. In case of
admixing an ideal quantity of lime, the soils are shifting towards
the value of =40o. A growth can also be measured at lower
lime quantities, however its extent does not reach the value of
40o that can be regarded as maximum, while the addition of an
excessive lime quantity results in a stagnation or decrease. The
curve fitted to the measurement results is visualized in Figure 5.

Figure 5. Increase of the friction angle at an age of 28 days, plotted


against the quantity of lime

The equation of the fitted curve based on 10 different soil


can be expressed with the following function:
(1)
0 ,589 M 2 7 , 07 M 18 ,9
where
- internal friction angle
M - added lime quantity
With the use of the function, the fitting to the points could be
expressed with the value of R2=0.77.
The function does not work at M=0% and it is limited to
M=8%, which is the rationality quantity of added quick lime.
The extent of growth of the cohesion did not show a
consistency similar to the previous one. The improvement
amounted generally to several hundreds of per cent, with 150 %
experienced in the worst case (Figure 6).
The above demonstrated alterations of the shearing
parameters have been measured at an age of 28 days, when only
the impact of the cation exchange prevailed to a significant
extent among the chemical processes taking place during
stabilization.
Further alterations triggered by the puzzolan reaction have
been tested on a sample taken at an age of 15 months.
According to the results of measurement, no changes could be
experienced at the friction angle of the stabilization, it
maintained its value of =40o, however the value of the
cohesion increased to c = 787 kPa, from the value of
c = 343 kPa measured at an age of 28 days. The alteration of the
shearing parameters clearly demonstrates the cementation effect
of the puzzolan reaction.

Figure 6. Increase of the cohesion at an age of 28 days, plotted against


lime quantity

CONCLUSIONS

During a soil stabilization with lime, the Ca2+ ions introduced


into the soil attach to the surface of the clay particles, replacing
the cations located there previously. Due to the high charging of
the Ca2+ ions, the clay particles coagulate. The coagulation
exerts a decisive effect on the soil structure, resulting the
plasticity index in a shifting towards a certain zone. This effect
leads to the elimination of the silt and clay particles, resulting in
an intensive reduction surface of the particle of the
stabilizations.
Due to the coagulating particles, the friction angle increases,
and also the value of cohesion becomes higher. The addition of
lime results consequently in a growth of the compression
strength of the soil, leading to an increased load bearing
capacity.
Contrary to the literature, I regard to remark that the positive
impacts of lime stabilization of the soil is basically a result of
the cation exchange, while the puzzolan reaction playing only a
secondary role.
It is also important to remark that the lime stabilization of
soils is not only suitable for the drying up of soaked soils. With
the application of a planned and properly executed stabilization,
load bearing capacity of ballast materials can be achieved with
local soils regarded or qualified as unsuitable. Accordingly, the
ballast and pavement layers can be effectively reduced, and the
quantity of delivered external materials minimized.
7

REFERENCES

Boromissza T. 1999. Hungarian design rules of concrete pavements.


Kzlekedsi s mlyptsi szemle (vfolyam, szm, oldalak) (in
Hungarian)
Bowles XXX: Engineering Properties of Soils and their Measurement
Filep Gy. 1988. Chemistry of soils. Akadmiai kiad, Budapest
Jan M. A., Walker R. D. 1963. Effect of Lime, Moisture and
Compaction on a Clay Soil. Highway Research Record 29, 1-12.
Kzdi . 1967. Stabilized earthroads. Akadmiai Kiad, Budapest
Little D.N. 1999: Evaluation of Structural Contribution of Lime
Stabilized Soils and Aggregates, Summary of Findings, 1999
National Lime Association
Lohmeyer G., Ebeling K. 2001. Building of industrial floors. Magyar
ptanyagipari Szvetsg ptsgyi (in Hungarian)
Mitchell J.K., Hooper D.R. 1961. Influence of Time Between Mixing
and Compaction on Properties of a Lime-Stabilized Expansive
Clay, Highway research Board 304, 14-31.
Nemesdy E. 1983. Building technology of road pavements.
Tanknyvkiad, Budapest (in Hungarian)
Szendefy J. 2008. Impact of soilstabilization with lime to the structure
and bearing capacity of internal soils, Geotechnic conference,
Rckeve (in Hungarian)
T 2-1-222-2002, tgyi Mszaki Elrs:General Rules of
Geotechnical Design of Roads (in Hungarian)
Wang J. W. H. 1963. Comparative Studies on the Effects of Hydraulic,
Calcitic and Dolomitic Limes and Cement in Soil Stabilization,
Highway Research Record 29, 42-54.

2604

Etude paramtrique en laboratoire du matriau Deep Soil-Mixing


Laboratory parametric study of the Deep Mixing material
Szymkiewicz F., Le Kouby A., Reiffsteck P., Mosadegh A., Tacita J.-L.

IFSTTAR, boulevard Newton, Champs-sur-Marne, F-77447 Marne la Valle Cedex 2

RSUM : Le Deep Soil-Mixing consiste mlanger mcaniquement un liant hydraulique avec le sol en place en vue damliorer ses
proprits mcaniques. En France, de nouvelles exigences concernant la mthode et le matriau sont apparues mesure que son
champ dapplications sest tendu. Dans cette communication, le travail de recherche entrepris lIFSTTAR est prsent. Il a consist
mener en laboratoire une tude paramtrique, en ralisant des mlanges dits de rfrence. Trois sables et cinq sols fins ont t
traits ( ltat pur ou mlangs entre eux) en faisant varier chaque fois les quantits de ciment et deau, afin dobserver linfluence
de la granulomtrie, de largilosit et de la quantit deau prsente dans le mlange sur la rsistance. Les retombes de ce travail sont
la proposition dabaques reliant la rsistance au dosage en ciment et au rapport C/E, et des formules permettant pour les sols grenus de
prdire la rsistance 28 jours en fonction du pourcentage de fines dans le sol et du dosage en ciment.
ABSTRACT: The Soil-Mixing consists in mixing a hydraulic binder into the soil mechanically in order to improve its mechanical
properties. In France, these new applications imply new requirements on the method as well as on the material. In this paper, the
research work undertaken at IFSTTAR is presented: it consisted in carrying out a parametric study on reference mixtures. Three sands
and five fine soils were treated with different cement and water contents to observe the influence of particle size, clay content and the
amount of water present in the mixture on its strength. Benefits of this work is the proposal of abaci connecting the strength of the
material to its cement content and cement / water ratio (C / E), and formulas allowing the prediction, for granular soils, of the strength
of the material at 7 and 28 days depending on the percentage of fines in the soil and on the cement content.
MOTS-CLS: Deep Mixing, sol, ciment, rsistance, compression simple, abaque.
KEYWORDS: Deep Mixing, soil, cement, strength, unconfined compression, abacus.
1

INTRODUCTION

La mthode du Deep Mixing prsente de nombreux avantages


(respect de lenvironnement, excution facile et rapide ainsi que
cot faible) qui ont grandement contribu tendre son
domaine dapplications (amlioration des sols, confinement des
pollutions et ralisation dlments structurels). Ces nouvelles
applications ncessitent une meilleure comprhension de la
mthode et du matriau. Les caractristiques du matriau Deep
Mixing (en particulier lhomognit et la continuit) sont de
plus en plus tudies, au fur et mesure que le domaine
dapplication sagrandit. De mme, des mthodes de prdiction
de la rsistance du matriau sont en cours de dveloppement.
Babasaki et al. (1996) citent quatre facteurs qui influencent
la croissance de la rsistance du matriau: les caractristiques
du liant, la nature et l'tat du sol rencontr, le malaxage et les
conditions de cure. Porbaha et al. (2000) affirment qu'il est
communment admis que la rsistance du sol trait au ciment
augmente avec le temps, de la mme faon quun bton. De
nombreuses tudes ont t menes pour proposer des formules
en ce qui concerne la prdiction de la rsistance du matriau.
Bruce (2001) et Topolnicki (2004) ont propos des relations
gnrales reliant le gain de rsistance la nature du sol traiter,
ainsi que des plages de rsistances atteignables pour ces mmes
sols aprs 28 jours de cure. En outre, des formules ont t
proposes pour prdire la rsistance du matriau partir d'un ou
plusieurs facteurs noncs par Babasaki et al. (1996). Pour les
btons, des lois portant sur le dveloppement de la rsistance et
dpendant uniquement du rapport ciment sur eau du mlange
tudi ont t tablies. Ces lois ont t tests sur le Soil-Mixing,
mais se sont avres inadaptes.

Mais d'autres formules ont t proposes, en tenant compte du


ciment, de l'eau et / ou du contenu en fines du sol trait.
Cependant, il n'existe toujours pas de formule largement
applicable pour estimer la rsistance du matriau et intgrant
tous les facteurs qui devraient tre pris en compte (CDIT, 2002),
car au jour d'aujourd'hui, aucune norme internationale n'existe
pour la prparation des chantillons de sol traits en laboratoire
(Kitazume et al., 2009). De plus, certaines formules ne peuvent
tre appliques qu un site particulier, tandis que d'autres ne
peuvent tre appliques qu certains sols.
Cet article prsente les rsultats d'une tude paramtrique
exhaustive ayant mis laccent sur la rsistance du matriau Deep
Mixing. Des relations liant la rsistance aux paramtres du Deep
Mixing ainsi quun abaque pouvant servir au dimensionnement
douvrages sont proposes.
2

PROGRAMME EXPERIMENTAL

2.1 Matriaux
Les sols tests au cours de cette tude vont des sables purs aux
argiles pures, avec des limons et des sols artificiels reconstitus
partir de sable de Fontainebleau mlang avec de largile ou
du Silica Flour, afin de permettre ltude de l'impact de la
nature et de la quantit des fines sur la rsistance du matriau
Deep Mixing. Le tableau 1 prsente les principales
caractristiques des sols tests.

2605

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Tableau 1. Principales caractristiques granulomtriques et valeurs au bleu de mthylne (VBS) des sols utiliss
Soils

d60 (mm)

Fontainebleau sand (SF)


Triel sand

d10 (mm)

Percent passing
2 mm
100

Percent passing
80 m
0,1

0,22

0,18

0,15

0,38

0,23

0,13

89,2

2,33

Percent
passing 2 m
0

VBS
0,01

0,1

0,5

0,23

0,07

94,25

11,8

0,2

0,028

0,011

0,004

100

95,2

0,14

Silt TGV

100

98,9

19

2,3

Silt Vmars

96,8

82

30

100

100

82

1,25

kaolinite de Provins

100

100

92

6,67

illite du Puy

100

95,1

56

5,4

illite Arvel

100

100

76

5,9

100

100

53

28

Silica Flour (SilicaF)

kaolinite Soka (kaoS)

Montmorillonite Arvel

(Fontainebleau C 265). La figure 1b montre que la rptitivit


des tests est assure: le coefficient de variation est gal 7%
pour le Puy du Argile, et gale 9,3% pour les sols sableux.

Les sols artificiels sont nomms par l'abrviation cite entre


parenthses dans le tableau 1 : par exemple, un sol artificiel fait
de 75% de sable de Fontainebleau et 25% de kaolinite Soka sera
nomm SF75-kaoS25. Plus de dtails sur tous ces sols peuvent
tre trouvs dans Szymkiewicz (2011).
Le ciment utilis pour ce programme exprimental est un
ciment de haut fourneau contenant au moins 85% de laitier de
haut fourneau, avec le reste de clinker Portland et un peu de
gypse (Classement Europen: "HRC" CEM III / C 32,5 N CE
PM-ES NF). Ce ciment prsente une rsistance
dveloppement lent, et son temps de prise initial est de 4 heures
aprs hydratation.

12

qu (MPa)

9
Argile du Puy C 140 kg/m3 w 70
Fontainebleau 1 C 320 kg/m3

Fontainebleau 2 C 320 kg/m3


Fontainebleau 1 C 265 kg/m3

2 C 265 k / 3

20

40
60
Curing time (days)

100

80

qu (MPa) (Argile du Puy)

Les sols ont t traits avec des teneurs en ciment diverses


allant de 70 400 kg/m3, couvrant ainsi toute la gamme usuelle
de dosages des applications Deep Mixing. Sol et ciment ont
d'abord t soigneusement mlangs sec et manuellement, afin
d'obtenir une consistance uniforme. Ils sont mis dans le
malaxeur et l'eau est ajoute. La teneur en eau du mlange (wi) a
t choisie afin d'obtenir un matriau auto-plaant, ce qui
signifie qu'il doit tre suffisamment fluide pour s'couler sous
son propre poids. Ainsi, wi doit tre au moins gale la limite
de liquidit du mlange (Szymkiewicz et al., 2012 b).
Le matriau est ensuite malax pendant 5 minutes pour les
sols non cohrents et pendant 10 minutes pour les sols
cohrents. Le mlange est ensuite coul dans des moules
cylindriques de 52 mm de diamtre, et, pour viter les bulles
d'air dans les chantillons, piqu et vibr.
Les moules sont ensuite bouchs et scells dans un sac
hermtique contenant une humidit relative leve. Ceux-ci sont
conservs une temprature de 20 3 C jusqu' la date
dessai (7, 14, 21, 28, 56 et 90 jours).
Le test de compression simple a t choisi principalement
pour sa fiabilit et son utilisation internationale. La charge
verticale statique a t applique une vitesse constante de 0,3
mm min-1. Le dplacement axial externe a t mesur l'aide
d'un LVDT. Plus de dtails peuvent tre trouvs dans
Szymkiewicz et al. (2012 a).

De nombreux essais sur tout type de sol ont t effectus, afin


d'valuer la rptitivit, reproductibilit et la rptabilit des
rsultats (figure 1). La figure 1a montre que la mthode de
ralisation des prouvettes et leurs conditions de cure assurent
une bonne rptabilit et reproductibilit des rsultats, puisque
deux mlanges (Fontainebleau C 265 et C 320) ont t rpts
en suivant exactement la mme procdure, le lendemain, par le
mme oprateur (Fontainebleau C 320) et par un autre oprateur

bl

2.2 Malaxage, coulage et conservation

2.3 Reprsentativit des rsultats

t i

2,0

15

1,6

12

1,2

0,8

6
Argile du Puy C 140 kg/m3 w80
mean Argile du Puy
SF75-SilicaF25 C 320 kg/m3
mean SF75-SilicaF25

0,4

0,0
0

6
8
specimen number

qu (MPa)

Frjus sand

d30 (mm)

0
10

12

14

Figure 1. Reprsentativit des rsultats : rptabilit et reproductibilit


(a) et rptitivit (b).

RSULTATS ET ANALYSES

3.1 Effet du ciment et des fines sur la rsistance des sols


granulaires non plastiques
Consoli et al. (2010) ont propos une fonction puissance (Eq. 1)
comme tant la plus adapte la relation exprimentale entre la
rsistance la compression simple (qu) et le dosage en ciment
(C).

qu

2606

aCb

(1)

o a est un paramtre exprim en kPa et b un paramtre sans


dimension. Les deux sont des paramtres exprimentaux.
Szymkiewicz et al. (2012 a) ont observ la mme relation
pour diffrents sols reconstitus ou naturels et non plastiques.
Ils ont galement observ que les paramtres exprimentaux a et
b sont fonction (au bout de sept jours de cure) du paramtre C63,
qui reprsente le pourcentage de fines dans le sol. Ainsi, ils ont

Technical Committee 211 / Comit technique 211

propos une formule (Eq. 2) permettant de prdire la rsistance


du matriau Deep Mixing constitu d'un sol non plastique et de
ciment, connaissant le C63 du sol, le dosage cible en ciment C
(%), et les teneurs en eau initiale et finale wi et wf du matriau.
Comme les rsultats in situ et obtenus en laboratoire sont trs
comparables dans le cas de sols granulaires, une telle formule
peut tre utilise directement pour des applications sur site.

qu 28

wf
wi

0.544
64.032 C1.777 0.171 ln C 63 C63

(2)

3.2 Effet du ciment, de leau et des fines sur la rsistance des


sols plastiques
Pour les sols cohrents (cest--dire argileux ou limoneux),
cependant, il n'est pas si simple d'analyser l'influence du ciment
et des fines.
Tout dabord, alors que la teneur en eau ncessaire pour
obtenir un matriau auto-plaant ne varie pas beaucoup dans le
cas des sols strictement non plastiques (puisque les particules de
sol sont inertes l'eau), celle-ci est trs variable lorsque des
particules d'argile sont prsentes dans le sol. D'autre part, le
domaine optimal douvrabilit du matriau Deep Mixing (dfini
par Szymkiewicz et al. (2012) comme la teneur en eau comprise
entre les limites de liquidit et de floculation du sol trait (figure
2a)) est trs faible dans le cas dun sol non plastique, tandis
quau contraire pour les sols traits plastiques, il peut tre trs
important. En outre, le domaine douvrabilit des sols volue en
fonction de la quantit de ciment ajoute et le type de sol
rencontr.
Domain of non workability

plastiques d'autre part, en diffrentes proportions, ont t traits,


diffrents dosages en ciment. Les rsultats ont montr que,
pour un dosage en ciment constant, les sols traits faits de sable
75% et de particules fines 25% sont plus rsistants que les
sols purs traits. Ceci peut s'expliquer en partie par une
optimisation de la distribution granulomtrique, ce qui
augmente la densit du matriau. Cependant, les particules
plastiques participent aussi laugmentation de la rsistance du
matriau, via un apport de cohsion non draine supplmentaire.
Les rsultats ont montr que cette rsistance supplmentaire due
cette cohsion diminue avec l'augmentation de la teneur en
ciment. Entre 210 et 320 kg/m3 de ciment, les particules d'argile
cessent davoir un effet bnfique sur la rsistance et peuvent
commencer tre considres comme un agent polluant.
Nanmoins, la fonction reliant la rsistance au ciment pour
les sols plastiques peut encore tre exprime comme une
fonction puissance : le paramtre b, dfini prcdemment,
diminue mesure que le pourcentage de fines dans le sol
augmente (comme pour un sol non plastique, dailleurs) jusqu'
ce qu'il atteigne 1. Ensuite, la relation devient linaire (Eq. 3),
de la forme :

qu c a C b

avec b toujours gal 1, et a et c des paramtres exprimentaux


exprims en kPa.
Les paramtres c et a varient en fonction de la nature du sol
(figure 3a), de mme quen fonction du temps de cure.
Cependant, aucune relation logique ne peut tre propose. La
teneur en eau initiale du matriau semble quant elle
ninfluencer que le paramtre c (figure 3b).

Optimized
domain of
workability

10

Argile du Puy
Silt TGV

SilicaF
kaolinite Soka

qU28 (MPa)

qu

(3)

Domain of workability
2

wL

non treated soil

wL

treated soil

woptimum

wf

0
0

100

200
C (kg/m3)

300

400

300

400

6000
4
w = 80%
w = 70%

4000

qu (kPa)

Artificial silt

qu (MPa)

Argile du Puy

2000

Illite Arvel

0
0

0,4

0,8

w/LL

1,2

1,6

0
0

Figure 2. Reprsentation du domaine douvrabilit (a) et dtermination


de la teneur en eau optimale (b) du matriau Deep Mixing.

Enfin, des essais sur prouvettes ayant 28 jours ont


galement montr l'existence d'une teneur en eau optimale,
indpendante de la teneur en ciment du mlange, mais fonction
de l'indice de plasticit du sol d'origine (figure 2b). Par
consquent, il est presque impossible de dissocier le dosage en
ciment de l'eau lors de l'tude des sols plastiques traits au
ciment.
De plus, les sols plastiques prsentent une cohsion qu'il est
trs important de prendre en compte, puisquelle apporte un
gain de rsistance supplmentaire. Pour dterminer l'effet de
cette cohsion, des sols constitus de sable et de particules fines
non plastiques dune part, et de sable et de particules fines

100

200
C (%)

Figure 3. Relation rsistance dosage en ciment pour 4 sols plastiques


(a) et pour un sol plastique diffrentes teneurs en eau (b).

Dans le cas du Deep Mixing, o le matriau doit tre


suffisamment fluide pour tre auto-plaant, c est toujours nul ou
ngatif. Il existe donc un dosage seuil en ciment qui conditionne
le succs du traitement au ciment dune argile.
On voit bien quil est ncessaire de continuer les travaux en
se focalisant sur les sols plastiques, puisquaucune formulation
gnrale na encore pu tre propose qui permettrait de prdire
la rsistance du matriau.
.

2607

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Triel
Frjus
kaolinite Soka
SF25-kaoS75
SF50-kaoS50
SF75-kaoS25
Fontainebleau
SF75-SilicaF25
SF50-SilicaF50
SilicaF
ADP
limon TGV
Vmars
Montmorillonite

Figure 4. Abaque rsistance dosage en ciment pour un matriau Deep Mixing auto-plaant.

3.3 Proposition dun abaque Rsistance Dosage en ciment


Au cours de cette tude, 57 mlanges partir de 14 sols et 6
dosages en ciment diffrents ont t raliss et tests chaque
ge. Un des objectifs de cette tude tait la ralisation dabaques
afin daider les ingnieurs dterminer le dosage en ciment
ncessaire la ralisation des travaux partir de la rsistance
mcanique ncessaire du matriau. Ces 57 points peuvent donc
tre placs dans le plan C = f(RC28) pour obtenir un premier
abaque (figure 4), qui nintgre pas de manire prcise
linfluence de leau, mais en tient compte tout de mme de
manire gnrale, puisque les mlanges tests ont t raliss
des teneurs en eau stalant de la limite de liquidit la limite
de floculation. Nous avons choisi comme rsistance de
rfrence la RC28, ce qui est toujours trs parlant pour les
industriels car elle est la rfrence utilise pour les btons, et
aussi car cest la rsistance la plus grande que nous puissions
prdire pour les mlanges base de sols grenus. Cet abaque a
t valid en la confrontant des rsultats obtenus sur site par
des entreprises franaises.
Six zones peuvent tre clairement identifies sur cet abaque :
la premire reprsente le domaine des sols trs plastiques et des
sols organiques. La seconde reprsente celui des sols
moyennement ou peu plastiques, quils soient argileux ou
limoneux, tandis que les troisime et quatrime sont ddies aux
sables et graves. Les graves traites nont pas pu tre testes en
laboratoire, du fait de la petite taille des prouvettes, qui
naurait pas assur une bonne reprsentativit du matriau.
Deux zones un peu particulires se dtachent : ce sont les zones
5 et 6 qui sont des zones dites de transition. Elles reprsentent
les rsultats prsents au paragraphe 3.1.
4

CONCLUSION

Dans cette communication, le travail de recherche men


lIFSTTAR depuis plusieurs annes, de concert avec les
differents acteurs du Deep Mixing francais, est present. Une
relation (valide par des retours chantier et des lements
bibliographiques) est propose, qui permet de prdire la
rsistance des sols non plastique traits au ciment selon la
mthode du Deep Mixing, cest--dire avec une teneur en eau
initiale suffisante pour que le matriau cr soit auto-plaant, en
connaissant seulement la granulomtrie du sol, le dosage cible
en ciment, et la teneur en eau initiale du mlange (ce qui sous entend que les paramtres dexcution sont contrls tout au
long du chantier).
Ltude montre que la transition dun sol non plastique un sol
plastique se fait de manire logique, en gardant la mme forme
de relation entre la rsistance finale et le dosage en ciment,
mme si la nature mme des particules argileuse fait que leau

2608

et le temps auront un impact beaucoup plus important sur les


sols plastiques que sur les sols purement grenus. Lexistence
doptimums, en ce qui concerne le couple teneur en eau
dosage en ciment, permet nanmoins daffirmer quil sera
bientt possible de proposer une formulation gnrale, un ge
donn.
Un abaque, ralis partir de cette tude et valid grce
des retours chantier, est propos : il permet, de manire simple,
de dterminer le dosage en ciment utiliser afin de sassurer de
lobtention de la rsistance mcanique requise, en fonction de la
nature du sol rencontr.
Les perspectives de ce travail sont donc la poursuite de
ltude des sols argileux traits, ainsi que le dveloppement
dune mthode de dimensionnement des ouvrages en matriau
Deep Mixing, en mettant laccent sur ltude de lhomognit
du matriau sur site.
5

REFERENCES

Babasaki, R., M. Terashi, T. Suzuki, A. Maekawa, M. Kawamura and E.


Fukazawa (1996). Factors influencing the strength of improved soil,
Grouting and Deep Mixing. 2nd International Conference on
Ground Improvement Geosystems
Bruce, D. A. (2001). "Practitioner's guide to the deep mixing method."
Ground Improvement 5(3): 95-100.
CDIT. (2002). The Deep Mixing Method Principle, Design and
Construction. The Netherlands: A.A. Balkema Publishers.
Consoli, N.C., Caberlon Cruz, R., Floss, M.F., & Festugato, L. (2010).
Parameters controlling tensile and compressive strength of
artificially cemented sand. Journal of Geotechnical and
Geoenvironmental Engineering, 136, 759763.
Kitazume, M., & Nishimura, S. (2009). Influence of specimen
preparation and curing conditions on unconfined compression
behaviour of cement-treated clay. Deep Mixing09, Japan.
Porbaha, A., Shibuya, S., & Kishida, T. (2000). State of the art in deep
mixing technology, Part III: Geomaterial characterization of deep
mixing. Ground Improvement, 4, 91110.
Szymkiewicz, F. (2011). Evaluation des proprits mcaniques du
matriau Soil-Mixing (PhD Thesis). Paris: Universit Paris-Est.
Szymkiewicz, F., Guimond-Barrett, A., Le Kouby, A. & Reiffsteck, P.
(2012 a) Influence of grain size distribution and cement content on
the strength and aging of treated sandy soils. European Journal of
Environmental and Civil Engineering, 16, 7, pp 882902.
Szymkiewicz, F., Tamga, F-S., Le Kouby, A. & Reiffsteck, P. (2012 b),
Optimization of the strength and homogeneity of the deep mixing
material by mean of the determination of the workability limit and
optimum water content, Canadian Geotechnical Journal,
submitted.
Topolnicki, M. (2004). Chapter 9: In situ soil mixing. In M.P.M.K.
Kirsch (Ed.), Ground Improvement (pp. 331423). Abingdon, UK:
Spon Press.

Investigation of failure analysis of clay reinforced with sand encapsulated


Enqute sur l'analyse des dfaillances d'argile renforc avec du sable enrobe
Tabarsa A.R.

Faculty of Engineering, Golestan University, Gorgan, IRAN

Hajiesmaeilian S.

Islamic Azad University, Science and Research, IRAN.

ABSTRACT: The paper includes discussions on the theory studies and numerical analysis as to stabilityof slopes reinforced with
geotextile encapsulated with lenses of sand. In this study, given the laboratory research conducted on clay reinforced with geotextile
encapsulated with lenses of sand, which showed in the high confining stresses due to the suitable interaction of clay and sand particles
with geotextile, significant improvement is obtained in shear strength. Considering these results, based on the stability importance of
slopesin the engineering projects and optimization, and increasing slopes stability, this technique is used as a reinforcement method in
the clay slopes using various methods of reinforcement and the effect of various conditions such as the number of layers of
reinforcement, confining pressures has been considered .In the study the numerical finite difference using software FLAC 2D has
been applied. The results showed that the slopes with higher height the Sandwich technique to typical reinforcement with geotextile
increases factor of stability safety under different studied scenarios because of improvement of intermediate behavior weakness by
thin sand layers and the amount of factor of stability safety has been also enhanced, increasing the number of reinforcement layers.
RSUM : Le document comprend des discussions sur les tudes thoriques et l'analyse numrique que les pentes stabilityof
renforcs par des gotextiles encapsuls avec des lentilles de sable. Dans cette tude, compte tenu de la recherche en laboratoire
menes sur de l'argile renforce avec gotextile encapsuls avec des lentilles de sable, qui a montr dans les fortes contraintes de
confinement en raison de l'interaction approprie d'argile et de sable avec gotextile, une amlioration significative est obtenue dans la
rsistance au cisaillement. Compte tenu de ces rsultats, bass sur l'importance de la stabilit des slopesin les projets d'ingnierie et
d'optimisation, et d'augmenter la stabilit des pentes, cette technique est utilise comme une mthode de renfort dans les pentes
argileuses en utilisant diverses mthodes de renforcement et de l'effet de diverses conditions telles que le nombre de couches de
renfort, des pressions de confinement a t pris en compte. Dans l'tude de la diffrence finie numrique l'aide du logiciel FLAC 2D
a t appliqu. Les rsultats ont montr que les pentes plus forte hauteur de la Technique du sandwich renfort typique avec un
facteur de scurit augmente gotextile stabilit sous diffrents scnarios tudis en raison de l'amlioration de la faiblesse
comportement intermdiaire par des couches de sable fin et la quantit de facteur de scurit de stabilit a t amliore galement,
l'augmentation du nombre de couches de consolidation.
KEYWORDS: sandwich technique, reinforced soil, factor of safety, stability
1

INTRODUCTION

In non-reinforced backfills are constructed using natural


materials the constraints from poor materials used, methods and
non-favorable environmental andgeotechnical conditions on the
project site created failures or adverse conditions in the backfill.
Intensity and scope of the failures may be increased to some
extent that to impact the general and stability and complete
servicing the intended project. In addition in some exceptional
cases the project needs stimulate the engineers to consider
construction of backfills with special dimensions inslope or
height and stability against particularloadings such as
earthquake. For example construction of road or rail backfills
with more height or backfills with steep (in places due to
limited land, the possibility of Transverse extension of backfill
base missing) can be considered as specific items.Ineach of the
above mentioned cases the Constructionof reinforcedbackfills
with particular materials is felt,to be able to bear external
factors effects as well as to have suitablestability in
environmental and geotechnical conditions.
In the last three decades, Geosynthetics is widely used with
high tension resistance to the soil reinforcement in geotechnical
engineering. Makes using thereinforcesprovidesoil tension
resistance in the soil and thus reduce lateral deformation of soil
and increase overall stability of reinforced soil structures
(Abiera,H.O.1991).

2609

In past decades, mostly the reinforced soil technique has


been applied using coarse materials as backfill (Backfill
materials). Recently, due to economic considerations, the
backfill material with low qualityand localaccess has been used
successfully (Bergado et al. 2008). Although you need to
determine the geotechnical properties for reinforcement
elements and backfill materials to design reinforced soil system,
but the mechanism of interaction of soil - reinforcementelement,
plays an important role.
To consider soil interaction mechanisms reinforcement
element, studies have been done using direct shear and tension
tests by various researchers (Zhenggui Wang and Werner
Richwien 2002).
In the reinforced clay Soil, the contact resistance is low and
consequently, failure in interfaceoccur prior to the
reinforcertension strength to the final extent. Thus, a large
percentage of the tension strength of reinforcernot mobilized at
failure and cannot be used (Jewell.R.A and Wroth,C.P .1987).
Experimental results show that the shear stress around
reinforcementis high and decreases rapidly with distance from
its surface. Thus in the reinforced soil structures with low
quality backfill material it is possible that a thin layer with
granular material containing high resistance around the
reinforcement will be used to control high shear stresses near
the reinforce and in the interface (Ghiassian and Jahannia 2004).
This method improves the stress transfer mechanism due to a

better interface properties which is called sandwich technique


(sandwich technique), In fact, the basis of this method is to
provide a thin layer of sand with high strength on both sides of
Geotextile, in order to improve the shear strength and
deformation
behavior
of th International
the reinforced
clay on
soilSoil Mechanics and Geotechnical Engineering, Paris 2013
Proceedings
of the 18
Conference
(Unikrishnan.N, Rajagopa.K and.Krishnaswamy,N.R. 2001).
1.1 Studies on clay reinforced
The Studies of Unikrishna et al. (2001) on the reinforced clay
better interface properties which is called sandwich technique
with lens of sand (sandwich technique) did show that adding
(sandwich technique), In fact, the basis of this method is to
sand improvesreinforced soil strength properties. Sand lens
provide
a thinproperties
layer of sand
with
on both
sides of
better
interface
which
is high
calledstrength
sandwich
technique
thickness, humidity and type of geotextile was paid attention.
Geotextile,
in order
to improve
shearmethod
strength
(sandwich
technique),
In fact,
the basisthe
of this
is toand
Abdi et al. (2009) during large-scale direct shear tests,
deformation
behavior
of high
the strength
reinforced
provide
a thin layer
of sand with
on both clay
sides ofsoil
studied the resistance improvement due to providethe thin layers
(Unikrishnan.N,
Rajagopa.K
and.Krishnaswamy,N.R.
2001).
Geotextile,
in order
to improve
the shear strength
and
of sand on either side of the geogrid (sandwich technique) in the
deformation behavior of the reinforced clay soil
clay and reported the results of the study as follows.
1.1 Studies onRajagopa.K
clay reinforced
(Unikrishnan.N,
and.Krishnaswamy,N.R. 2001).
- given the Figure (1) one can understand that providing a
thin
layer ofof sand
with ethigh
strength
on reinforced
both sidesclay
of
The Studies
Unikrishna
al. (2001)
on the
1.1 Studies on clay reinforced
reinforceris
to technique)
improve plasticity
andthat
resistance
with lens ofvery
sandeffective
(sandwich
did show
adding
clayimprovesreinforced
soils.
sand
strength
Sand
Theof
Studies
of Unikrishna etsoil
al. (2001)
on properties.
the reinforced
claylens
- using
the
geogrid
system
in did
thewas
sand
ofthat
fine
granular
thickness,
and
typetechnique)
of
geotextile
paid
attention.
with
lens
of humidity
sandburied
(sandwich
show
adding
soilAbdi
(sandwich
technique)
increases
the direct
shear
resistance
et al. (2009)
during
large-scale
shear
tests,
sand
improvesreinforced
soil
strength
properties.
Sand
lens
impressively.
Theand
buried
geogrid
system
in
theattention.
sand
is layers
more
studied humidity
the resistance
improvement
due was
to providethe
thin
thickness,
type
of
geotextile
paid
effective
the
internal
friction
of soil
anddirect
lesstechnique)
effective
on
of
sandetonon
either
side of
the geogrid
(sandwich
in the
Abdi
al.
(2009)
during
large-scale
shear tests,
coefficient
of cohesion.
clay and
reported
the
results of the
study
as follows.thin layers
studied
the resistance
improvement
due
to providethe
- given
the side
Figure
(1)geogrid
one can(sandwich
understand
that providing
of sand
on either
of the
technique)
in the a
thin
of the
sand
withof the
high
strength
on both sides of
clay
andlayer
reported
results
study
as follows.
reinforceris
very effective
plasticity
and resistance
- given the Figure
(1) onetocanimprove
understand
that providing
a
clay soils.
thinof layer
of sand with high strength on both sides of
- usingvery
the buried
geogrid
system plasticity
in the sandand
of fine
granular
reinforceris
effective
to improve
resistance
soil soils.
(sandwich technique) increases the shear resistance
of clay
impressively.
The geogrid
buried system
geogridinsystem
in of
thefine
sand
is more
- using the buried
the sand
granular
on thetechnique)
internal friction
of soilthe
and less
on the
soileffective
(sandwich
increases
sheareffective
resistance
coefficient of
cohesion.
impressively.
The
buried geogrid system in the sand is more
effective on the internal friction of soil and less effective on the
coefficient of cohesion.
Figure 1. The relationship between shear stress - shear displacement
under confining stress 75 kPa with different thicknesses of sand layers.
(Abdi et al., 2009)

Abdi and Arjmand (2011) carried out various experiments on


the pullout testonreinforced clay with geogrid encapsulated with
thin layers of sand. The Samples have been prepared at the
optimum moisture content and maximum dry density weight
which have been obtainedfrom Standarddensity testing Proctor.
One-way
geogridis
usedbetween
with ashear
sandstress
layer,- shear
6, 10displacement
and 14 mm in
Figure 1. The
relationship
thickness.
Experimental
showed
that theofencapsulated
under confining
stress 75 kPa results
with different
thicknesses
sand layers.
geogridin
thin
layersshear
increases
clay pullout
(Abdi
et al.,relationship
2009)sandybetween
Figure
1. The
stress - reinforced
shear displacement
resistance
pullout
conditions.
The results
showed
under
confiningunder
stress the
75 kPa
with different
thicknesses
of sand
layers. that
theet
maximum
resistance
to carried
pullout out
increased
confining
Abdi
and Arjmand
(2011)
various with
experiments
on
(Abdi
al., 2009)
pressure
andtestonreinforced
the optimal thickness
of geogrid
layer ofencapsulated
sand is the same
the pullout
clay with
with
for
alllayers
normal
thin
ofstresses.
sand.(2011)
The carried
Samplesouthave
beenexperiments
prepared atonthe
Abdi
and
Arjmand
various
Studies
of Tabarsa
and
Radmehr
(2011)
on density
the reinforced
moisture
content
maximum
dry
weight
theoptimum
pullout
testonreinforced
clayand
with
geogrid
encapsulated
with
with
sand
(sandwich
technique)
yielded
which
have
been
obtainedfrom
Standarddensity
testing
thinclay
layers
of lens
sand.of
The
Samples
have been
prepared
atProctor.
thethe
following
results.According
tosand
thislayer,
research
One-way
geogridis
used with
10the
and increasing
14
mm in
optimum
moisture
content
and amaximum
dry6, density
weight
reinforcement
layers
caused
to
increase
ultimate
strenght
. Also
thickness.
Experimental
results
showed
thattesting
the
encapsulated
which
have been
obtainedfrom
Standarddensity
Proctor.
improved
the
of samples
increasing
confining
stress
geogridin
thinrecovery
sandy
layers
increases
reinforced
claymm
pullout
One-way
geogridis
used with
a sand
layer,
6, 10 and
14
in
from
100Experimental
to 550
kParesults
and
to confining
stress300
kPathe
resistance
under
the pullout
conditions.
The the
results
showed
that
thickness.
showed
that
encapsulated
improvement
trend
increases
and
thenreinforced
decreases. with
(Figure
2)
the maximum
resistance
pullout
increased
confining
geogridin
thin sandy
layers to
increases
clay
pullout
pressureunder
and the
thickness ofThe
layer
of sand
is thethat
same
resistance
the optimal
pullout conditions.
results
showed
all normalresistance
stresses. to pullout increased with confining
thefor
maximum
Studies
of optimal
Tabarsathickness
and Radmehr
(2011)
on the
reinforced
pressure
and the
of layer
of sand
is the
same
with stresses.
lens of sand (sandwich technique) yielded the
forclay
all normal
following
to this
research
increasing
Studies ofresults.According
Tabarsa and Radmehr
(2011)
on thethereinforced
reinforcement
caused
to increase
ultimate strenght
Also
clay
with lens layers
of sand
(sandwich
technique)
yielded .the
improvedresults.According
the recovery of samples
increasingthe
confining
stress
following
to this research
increasing
from 100 layers
to 550caused
kPa to
and
to confining
stress300. Also
kPathe
reinforcement
increase
ultimate strenght
improvement
trend increases
and then
decreases.
(Figure
2)
improved
the recovery
of samples
increasing
confining
stress
from 100 to 550 kPa and to confining stress300 kPathe
improvement trend increases and then decreases. (Figure 2)

Figure 2. Effect of confining pressure on the reinforced sample, with


geotextile and sand in 4 mm thick.(Tabarsa and Radmehr 2011)

Given the abovementioned forms it can be perceived that


sample plasticity diminishes increasing the thickness of the
layer of sand. Also, recovery percentage of resistance has
increased slightly increasing the thickness of the sand in low
confining stresses.And in highconfining stresses (550 kPa) the
high resistance percentage has increased significantly and
therefore
it canofbe
stated that
the high
impact
Figure 2. Effect
confining
pressure
on theconfining
reinforced stresses
sample, with
on
the sand
the composite
system desired
geotextile
andperformance
sand in 4 mm in
thick.(Tabarsa
and Radmehr
2011) result
is achieved
Because
ofpressure
the angle
of high
internal
friction
Figure
2. Effect of
confining
on the
reinforced
sample,
with of the
sand.
The
Results
andthick.(Tabarsa
research,
suggest
few
studies
on that
the
Given
the
abovementioned
forms
can
be2011)
perceived
geotextile
and sand
in
4 mm
and itRadmehr
subject
of numerical
simulation
techniques
for soil of
slopes
sample plasticity
diminishes
increasing
the thickness
the
stability
analysis
whichrecovery
the forms
research
hasbebeen
done in
layer
sand.
Also,
percentage
ofperceived
resistance
has
Givenofthe
abovementioned
it can
thatthis
direction.
increased
slightly
increasingincreasing
the thickness
of the sand
sample
plasticity
diminishes
the thickness
of inthelow
confining
stresses.And
in highconfining
(550 kPa)
layer
of sand.
Also, recovery
percentagestresses
of resistance
hasthe
high resistance
percentagethehas
increased
significantly
increased
slightly increasing
thickness
of the
sand in lowand
2therefore
MODELING
NUMERICAL
it can AND
be stated
that the highANALYSIS
confining(550
stresses
confining
stresses.And
in
highconfining
stresses
kPa)impact
the
on resistance
the sand performance
the composite
desiredand
result
high
percentage inhas
increased system
significantly
FLAC software is based on the finite differential method. Finite
is achieved
the the
angle
ofconfining
high internal
friction
of the
therefore
it canBecause
be statedofthat
high
stresses
impact
differential methodis used in various engineering issues.Its
Theperformance
Results and
research,
suggest
few desired
studiesresult
on the
on sand.
the sand
in the
composite
system
application in the soil and rock mechanics is common because
subject of
numerical
techniques
for soilof slopes
is achieved
Because
of thesimulation
angle of high
internal friction
the
one can model the big displacements and stresses of soil and
stability
analysisand
which
the research
done
sand.
The Results
research,
suggest has
fewbeen
studies
on inthethis
rock masses.in the research FLAC 2D software has been used.
direction.
subject
of numerical simulation techniques for soil slopes
stability analysis which the research has been done in this
direction.
3 CHARACTERISTICS OF THE STUDIED MATERIALS
2 MODELING AND NUMERICAL ANALYSIS
In this study, to analysis the slope stability, the behavioral
FLAC software is based
on the finite
differential method. Finite
2 MODELING
NUMERICAL
models MohrAND
- Coulomb
is used.ANALYSIS
Also nonwoven geotextile
differential methodis used in various engineering issues.Its
application
soilon
and
mechanics
is common
because
FLAC
softwareinisthe
based
therock
finite
differential
method. Finite
one can model
the big
and stresses issues.Its
of soil and
differential
methodis
useddisplacements
in various engineering
rock masses.in
FLAC
2D software
has been
used.
application
in the the
soilresearch
and rock
mechanics
is common
because
one can model the big displacements and stresses of soil and
rock masses.in the research FLAC 2D software has been used.
3 CHARACTERISTICS OF THE STUDIED MATERIALS
In this study, to analysis
the STUDIED
slope stability,
the behavioral
3 CHARACTERISTICS
OF THE
MATERIALS
models Mohr - Coulomb is used. Also nonwoven geotextile
modeling
slopes.
In have
this been
study,used
to for
analysis
the the
slope
stability, the behavioral
models Mohr - Coulomb is used. Also nonwoven geotextile
Table 1. details the parameters considered in this study

4 GEOMETRIC PROPERTIES OF SLOPES


In the analysis, four types of slope with different heights of 6, 9,
12 and 15 m with the same slope of 56 degrees have been
considered. The method of geotextiles arrangement is shown in
have been used for modeling the slopes.
Figure 3 in the sandwich technique for the slope 12 meters.
Table
1. used
detailsfor
themodeling
parametersthe
considered
have
been
slopes. in this study
Table 1. details the parameters considered in this study

4 GEOMETRIC PROPERTIES OF SLOPES

In the analysis,PROPERTIES
four types of slope
with different heights of 6, 9,
4 GEOMETRIC
OF SLOPES
12 and 15 m with the same slope of 56 degrees have been
considered.
geotextiles
arrangement
shown
In the
analysis,The
fourmethod
types ofofslope
with different
heightsisof
6, 9, in
sandwich
technique
for56
thedegrees
slope 12have
meters.
12 Figure
and 153 in
m the
with
the same
slope of
been
considered. The method of geotextiles arrangement is shown in
Figure 3 in the sandwich technique for the slope 12 meters.

2610

Technical Committee 211 / Comit technique 211

Figure 4. Comparison of factors of safety in 9 m slope

Figure 3.slopes12-meter, reinforced with sandwich technique

Figure 5. Comparison of factors of safety in 12m slope for


geotextiledistances intervals in both 1 and 1.5 meter

5 HOW TO MODEL
First defining the model geometry and thenrequired parameters
are considered for different scenarios of various materials.
Given the importance of the boundary conditions on the both
sidesthe slope is anchored inthe horizontal direction (x) and in
downwardslope horizontal and vertical directions (x, y) are
anchored. Then geotextiles with distances equal to 1.5 m with
the same length have been modeled. And following given the
characteristics of the geotextiles, the suitable interaction
parameters are involved between soil and geotextile. Then thin
layers of sand with equal 30 cm thickness are considered to
simulate the effect of sandwich technique inreinforcement
process on the border between the reinforcement element and
the clay soil (geotextile),given the different distances between
the layers.

stability safety.

Figure 6. Comparison factors of safety of slop different heights in


various states of reinforcement

6 THE RESULTS OF NUMERICAL MODELING


After modeling and static analysis using software it evaluates
and compares the factor of stability safety in slopes for different
heights and different modes of reinforcement as follows.
6.1 Results of static analysis on slopes with different heights
According to the analysis results,
finite differences of
reinforced clay slopes are shown as 6, 9, 12, 15 m, factors of
safety in different states of reinforcement as computed (figures
4 and 5) and includes three modes 1- clay Model (C), 2 - clay
with geotextile (CG), 3 - Sandwich Technique (CGS)
According to the output of numerical modeling analysis and
the results shown in the above graphs, adding layers of sand on
both sides of geotextiles(sandwich technique) increases
significantly factors of safety at different height (6,9,12,15 m).
In Figure 6 this technique has a greater impact on stability of
clay slopes in above heights and further increases the factor of
stability safety.

2611

7 CONCLUSION
- Analyses showed that the sandwich technique is an effective
method of stabilizing clay slopes for engineers and results of the
numerical analysis carried out in vitro studies are consistent
with the sandwich technique.
- Sandwich techniqueis used in clay soils where have the poor
and weak interaction with the geotextiles.
- given the acceptable development of a factor of safety using
the sandwich technique we can make or design slopes with
more height and slope in the engineering projects.
- sandwich technique in the more heights has more impact in the
factor of safety increase , because of confining pressure increase
and consequently getting better interaction condition between
sand and geotextile .
- The factor of stability safety will be increased, increasing
noumber of reinforced layers in the sandwich technique.

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

8 ACKNOWLEDGEMENT
At the end we thank sincerely for cooperation of dear friends
Mr. Amir Gharib and Mohsen Mosivand.
9 REFERENCES
Abiera H.O. 1991. Mechanically stabilized earth using tensar,bamboo
and steel grid reinforcements with weathered Bangkok clay as
backfill , M. Eng.Thesis,No.Gt-90-21 ,Asian institute of technology,
Bangkok,Thailand.
Bergado,D.T,
Sampaco,C.L,
Shivashankar,R,
Alfaro,M.C,
Anderson,L.R. and Balasubramaniam, A.S. 2008. Performance of a
welded wire wall with poor quality backfill on soft clay, In
proceedings of ASCE.
Zhenggui .W and Werner .R . 2002. A study of soil-reinforcement
interaction
friction
.,Journal
of
Geotechnical
and
Geoenvironmental Engineering, 128 (1),92-94.
Jewell. R.A. and Wroth, C.P. 1987. Direct shear test on reinforced
sand., Geotechnique 37 (1), 53-68.

2612

Unnikrishnan.N, Rajagopal.K and Krishnaswamy,N.R. 2002. ,Behavior


of reinforced clay under monotonic and cyclic loading ., Geotextile
and Geomembrances (20),117-133.
Abdi.M.R. and Arjomand M.A. 2011. Pullout tests conducted on clay
reinforced with geogrid encapsulated in thin layers of sand.,
Geotextiles and Geomembranes, 1 8.
Tabarsa,A.R.
and Radmehr.M. 2011. Influence of geotextile
encapsulated with sand on behavior clay reinforced., research
report, Iran.
Abdi.M.R and Sadrnejad. M.R and Arjomand. M.A. 2009. Strenght
enhancement of clay by encapsulating geogrids in thin layers of
sand., Geotextiles and Geomembranes 27 ,447- 455
Ghiassian.H and Jahannia.M. 2004. Influence of encapsulated geogridsand sustem on bearing capacity and settlement characteristics of
reinforced clay. , International Journal of Civil Engineering,
Vol.2,No.1.

Influence of relative density on microbial carbonate precipitation and mechanical


properties of sand
L'influence que la densit relative du sol donne dans prcipitation du carbonate microbienne et
proprits de la mcanique
Tsukamoto M.

Kawasaki Geological Engineering Co., Ltd. (Formerly Public Works Research Institute), Minato-ku, Tokyo, Japan

Inagaki, T. Sasaki, Y.

Public Works Research Institute, Tsukuba, Ibaraki, Japan

Oda K.

Osaka University, Suita, Osaka, Japan


ABSTRACT: There exists a ground improvement technology that uses calcium carbonate precipitated from carbon dioxide generated
by microbial metabolism and calcium sources in the pores of soil. It is known that the mechanical properties of the improved grounds
correlate with the amount of calcium carbonate precipitation, but it is unclear how soil density influences calcium carbonate
precipitation and the mechanical properties of the improved soil. Toyoura sand specimens of three relative densities are used to
precipitate calcium carbonate through microbial metabolism. The injection-improved test and the triaxial test (consolidated-drained
condition) are conducted to investigate calcium carbonate precipitation and the mechanical properties of the soil. The results show
clearly that more calcium carbonate precipitation occurs in soil with lower relative density, but that in soil with higher relative
density, the mechanical properties strengthen as calcium carbonate precipitation increases.
RSUM : Dans la nature, il existe des micro-organismes qui capturent le dioxyde de carbone et les ions calcium prsents dans la
terre pour ensuite rejeter du carbonate de calcium. Ces dernires annes, en sappuyant sur le fonctionnement de ces microorganismes, une technique renforant la rsistance du sol a t dveloppe. Dans le cadre du dveloppement dune technologie qui
renforce le sol l'aide de mtabolisme microbien, la prsente tude a mis en vidence exprimentalement que linfluence de la densit
relative du sol exerce un effet bnfique. Tout dabord, travers une srie de tests de cisaillement, le comportement au cisaillement du
sol renforc par le mtabolisme de ces micro-organismes a t mis en avant. Ensuite, il sest avr que plus la densit relative du sol
tait petite, plus la quantit de carbonate de calcium dpose tait importante. Cependant, nous avons aussi compris que plus la
densit relative du sol tait grande, plus les effets renforant la rsistance du sol taient visibles.
KEYWORDS: micro-organism, ureolysis, soil improvement, mechanical properties, triaxial test, calcium carbonate
1 INTRODUCTION
To ensure the efficient maintenance of civil engineering
structures, which is an issue, ground improvement technologies
can be applied to reinforce existing structures. Ground
improvement technology that uses calcium carbonate
precipitated from carbon dioxide generated by microbial
metabolism and calcium sources in the pores of soil is expected
to be applicable to ground directly under existing structures
because the viscosity of the injected grout is low (Wiffen et al.
2007). It is known that the mechanical properties of grounds
improved through the use of this method correlate with the
amount of calcium carbonate precipitation (Inagaki et al. 2011),
but it is unclear how the relative density of the soil influences
the precipitation and the mechanical properties of the improved
soil.
In this study, we used Sporosarcina pasteurii (ATCC11859)
to stimulate microbial metabolism via the following reactions.
Our aim was to investigate the relationship between the soils
mechanical properties and calcium carbonate precipitation in
three types of Toyoura sands that were compacted by microbial
metabolism.
(Ureolysis)
CO(NH2)2 + 3H2O 2NH4+ + 2OH- + CO2

(1)

(Calcium carbonate precipitation)


CO2 + H2O HCO3- + H+

(2)

HCO3- + Ca+ + OH- CaCO3 + H2O

(3)

2613

In the tests, microbial broth and a nutrient mixture were


injected into the Toyoura sand specimens with three relative
densities to improve compaction. In addition, we carried out a
triaxial test (consolidated-drained [CD] condition) on the
improved specimens. Then, we examined the influence of the
relative density of the soil on calcium carbonate precipitation
and the soils mechanical properties.
2 TEST METHODS
2.1 Method for production of specimens
A half-split mold made of PVC, 15 cm in height and 5 cm in
diameter, was used to create the specimens (Fig. 1 and Photo 1).
Silicone grease was applied on the internal surface of the mold
to prevent the generation of water paths along the wall. Toyoura
sands having the physical properties listed in Table 1 were used
to create the specimens, and the molds were filled with sand
using the air-drop method. The specimens had a relative density
of Dr = 15% (d = 1.372 g/cm3), Dr = 60% (d = 1.504 g/cm3),
and Dr = 85% (d = 1.589 g/cm3). The specimens were checked
for weight and density, fitted with a collar on top, and saturated
with distilled water supplied from the bottom. The surface of
the specimen of Dr = 15% sank significantly during the
hydraulic filling. This settlement was measured using vernier
calipers, and it was confirmed that the relative density after
hydraulic filling was about Dr = 30%.

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

2.2 Curing method (nutrient injection process)


After saturation, 250 ml of microbial broth was injected into
each specimen. The microbial broth was made by planting
Sporosarcina pasteurii in the medium described in Table 2.
After confirming that the broth had completely permeated the
specimen, 200 ml of the nutrient mixture described in Table 3
was injected into the specimen at 12 h intervals for specified
cycles.
Table 1. Physical properties of Toyoura sand
Soil
particle
densitys
(g/cm3)

Water
content
(%)

2.623

0.0

Max.
50%
Min.
Soil
Fine
Max.
grain
diameter fraction
dry
suspension
dry
diameter
on the
pH
content density density
(mm)
grain
dmax dmin
(%)
size
(g/cm3) (g/cm3)
diagram
D50
(mm)

0.425

0.177

0.6

1.645

1.333

6.3

Table 3. Composition of nutrient mixture


Name of reagent
Nutrient broth
NH4Cl
NaHCO3
Urea
CaCl2
Distilled water

Added amount
3g
10 g
2.12 g
0.5 mol = 30.03 g
0.5 mol = 55.49 g
1L

The mold was cured in a room with a constant temperature


set at 22C. The nutrient mixture that had been injected
previously and that remained in the pore was pushed out and
drained to maintain the saturated state of the specimen in the
mold.
Approximately 12 h after the specified cycles of nutrient
injection were completed, 300 ml of distilled water was injected
to wash away the nutrient mixture remaining in the pore.
As Table 4 shows, 15 specimens were made. The nutrient
mixture was injected into these specimens at various
frequencies in order to diversify the amount of CaCO3
precipitation at each relative density. In addition, three
specimens were only saturated with distilled water, and not
injected with the broth and nutrient mixture, in order to examine
the strength of the Toyoura sand itself.
Table 4. Test cases
Case

Dr30

Dr60

Dr85

Figure 1. Schematic diagram of the mold

Photo 1. Photo of the mold


Table 2. Composition of medium
Name of reagent
0.13 M Tris (pH = 9)
Yeast extract
(NH4)2SO4
Distilled water

Added amount
0.13 mol = 15.75 g
20 g
10 g
1L

Frequency of
Specimen nutrient injection
No.
(Total injection
amount ml)
Dr30-N

Dr30-P1
2(400)
Dr30-P2
4(800)
Dr30-P3
8(1600)
Dr30-P4
12(2400)
Dr60-N

Dr60-P1
2(400)
Dr60-P2
4(800)
Dr60-P3
8(1600)
Dr60-P4
12(2400)
Dr85-N

Dr85-P1
2(400)
Dr85-P2
4(800)
Dr85-P3
8(1600)
Dr85-P4
12(2400)

Curing
time
(hours)

24
48
96
144

24
48
96
144

24
48
96
144

Initial dry
density
3

d (g/cm )
1.423
1.428
1.414
1.416
1.414
1.513
1.504
1.504
1.504
1.504
1.589
1.589
1.589
1.589
1.589

Initial
relative
density
Dr (%)
33.2
35.2
30.2
30.8
30.4
62.6
59.9
60.0
59.9
60.0
84.9
85.0
85.0
84.9
85.0

CaCO3
Precipitation
3

(kg/m )

32.24
63.72
141.92
225.53

28.11
51.77
130.61
212.11

27.12
57.38
117.13
198.55

2.3 Triaxial test method


Triaxial tests (CD condition) were conducted using the
specimens produced by the method explained in 2.2. To reduce
disturbance caused by the demolding/shaping of specimens, the
specimens were frozen in the mold after the free water was
removed. The specimens that featured high CaCO3 precipitation
were also frozen to equalize test conditions. The frozen
specimens were removed from the molds and shaped to 10 cm
in height and 5 cm in diameter. The shaped specimens were
measured to check the diameter and height and then placed in a
triaxial cell and defrosted under a negative pressure of 30 kPa.
The defrosting time was set at about 1.5 h, which was the
approximate time needed for stabilization of the axis
displacement caused by contraction in the process of defrosting.
We measured the diameter and the height of the defrosted
specimens, covered them with a cell cover, and saturated them
with degassed distilled water via the double-negative pressure
method. The back pressure was increased to 200 kPa, and
effective consolidation stresses of c = 100 kPa were applied
for isotropic consolidation. After consolidation, we confirmed
that the B values in all the specimens were 0.95 or higher. Then,
axial compression was performed at a strain rate of 0.5%/min.
The axial force was measured by the load cell inside the cell,
axial strain was measured by the displacement gauge outside the
cell, cell pressure and back pressure were measured by the
water pressure gauge, and the volume change was measured by
the low-capacity differential pressure gauge.

2614

Technical Committee 211 / Comit technique 211

Principal stress difference (kN/m2)

After the triaxial tests were completed, the specimens were


dried in a furnace and weighed, and the CaCO3 precipitation in
each specimen was then obtained by measuring the decrease in
mass resulting from CaCO3 decomposition by hydrochloric acid.
3 TEST RESULTS

250
CaCO3 precipitation (kg/m3)

Dr60-N

1000

Dr85-N
Dr30-P2

800

Dr60-P2

600

Dr85-P2

400
200
0
-9

10

15

-6

-3

0
0

5
10
15
Axial strain (%)
Figure 3. Relationships among axial strain, principal stress difference,
and volumetric strain (c = 100 kPa)

Dr = 30 %
200

Dr30-N

1200

Volumetric strain (%)

3.1 Relationship between the amount of nutrient mixture


injected and the CaCO3 precipitation
Figure 2 shows the relationship between the amount of nutrient
mixture injected and the CaCO3 precipitation. The quantity of
CaCO3 precipitation is given per unit volume of the test
specimen.
The tendency for CaCO3 precipitation to increase as the
injections of the nutrient mixture increased can be confirmed for
each relative density. When the total injection of the nutrient
mixture is less than 800 ml, the differences among specimens
with different densities is unclear. When more than 800 ml of
the nutrient mixture is injected, the differences among the
specimens with different densities are observed. It is confirmed
that lower the relative density, the more CaCO3 precipitates.
This is because the low relative density increases the amount of
microbes and nutrients absorbed by the test specimen.

1400

Dr = 60 %
Dr = 85 %

150
100
50
0

500
1000
1500
2000
2500
Total amounts of nutrient injection (ml)

3000

Figure 2. Total amounts of nutrient mixture injected and CaCO3


precipitation

3.2
Relationships among axial strain, principal stress
difference, and volumetric strain
Figure 3 shows the principal stress differenceaxial strain
curves and volumetric strainaxial strain curves. Toyoura sands
of each relative density saturated only with distilled water
(unsolidified) and those injected with 800 ml of the nutrient
mixture and CaCO3 precipitation of 51.863.7 kg/m3 are shown
in Fig. 3. Photo 2 shows the solidified test specimen after shear.
It can be confirmed that the solidification caused by the
CaCO3 precipitation leads to an increase in the maximum
principal stress. In specimens with about the same CaCO3
precipitation, the higher relative density of the soil, the increase
in the maximum principal stress difference is the greater. Strain
softening behavior is observed when the principal stress
difference reaches the maximum in the solidified test specimen.
A residual state occurs when axial stress reaches 5% or more;
then, the principal stress difference is constant at every relative
density, and its value shows no difference at each relative
density. It is thought that the principal stress difference becomes
equality in the residual domain because the test specimen is
sheared along the sliding surface. Photo 2 also shows a shearing
plane along the sliding surface. The increase in the volumetric
strain on the expansion side is confirmed clearly in the
solidified test specimens when the axial strain is small at each
relative density in comparison with the unsolidified test
specimens, and the ratio of increase becomes small around an
axial strain over 5%.

Photo 2. Condition of the test specimen after shear

3.3 Relationship between CaCO3 precipitation and maximum


principal stress difference
Figure 4 shows the relationship between CaCO3 precipitation
and the maximum principal stress difference. There is no
change in the maximum principal stress difference when CaCO3
precipitation is less than 30 kg/m3 at each relative density. At
precipitation levels greater than 30 kg/m3, the maximum
principal stress difference increases monotonically depending
on CaCO3 precipitation. The increase in strength is remarkable
in test specimens have higher relative density but little CaCO3
precipitation. In the case of Dr = 85%, in comparison with Dr =
30% and Dr = 60%, the maximum principal stress difference
increases even as CaCO3 precipitation stays at the same level.
3.4 Relationship between CaCO3 precipitation and secant
modulus
Figure 5 shows the relationship between CaCO3 precipitation
and the secant modulus. The secant modulus was calculated
from the principal stress difference at the axial strain of 0.4%.
At each relative density, there is little change when CaCO3
precipitation is less than 30 kg/m3; however, when CaCO3
precipitation is more than 30 kg/m3, the secant modulus
increases linearly according to the increase in CaCO3
precipitation.

2615

comparison with a case of no precipitation. It is confirmed that


the residual stress increases as the CaCO3 precipitation
increases at precipitation levels of more than 30 kg/m3. The
specimens for each relative density show no difference at
CaCO3 precipitation levels less than 60 kg/m3, and it is unclear
whether a difference is observed at precipitation levels greater
than 60kg/m3.

10000
Dr = 30 %
8000

Dr = 60 %
Dr = 85 %

6000
4000

1000
Dr = 30 %

2000
0

50

100
150
200
CaCO3 precipitation (kg/m3)

250

Figure 4. Relationship between CaCO3 precipitation and maximum


principal stress difference
1400

Secant modulus (MPa)

Dr = 60 %

1000

Dr = 60 %
Dr = 85 %

600
400
200

50

100
150
200
CaCO3 precipitation (kg/m 3)

250

Figure 7. Relationship between CaCO3 precipitation and residual stress

Dr = 85 %

800
600

4 CONCLUSION

400

We investigated the influence of the relative density of soil on


CaCO3 precipitation by microbial metabolism and on the soils
mechanical properties. The results of our experiment are as
follows:
The CaCO3 precipitation tends to increase as the relative
density of the soil decreases.
The maximum principal stress difference increased
monotonically with the CaCO3 precipitation at each relative
density. The secant modulus increased linearly.
The increase in the maximum principal stress difference was
remarkable in soils with high relative density.
The axial strain at the maximum principal stress difference
decreased depending on CaCO3 precipitation in specimens of
all relative densities and became constant regardless of
CaCO3 precipitation when it approached 0.5%.
The residual stress increased monotonically depending on
CaCO3 precipitation, but the differences among the relative
densities are unclear.
A meaningful difference is not seen in the mechanical
properties of the soil among specimens with no precipitation
and those with CaCO3 precipitation up to 30 kg/m3.
These results indicate that when applying this injection
solidification technique in the field, the density of the existing
ground will affect the strength increase. Therefore, like a
conventional compaction method used in construction, this
method would require a combination examination beforehand in
order to confirm extreme expression characteristics.
Because the axial strain at the maximum principal stress
difference becomes constant regardless of density when CaCO3
precipitation becomes constant, we suggest that the approximate
strength of the soil can be estimated using the secant modulus.
We will investigate the influence of soil density on a
permeability change attributable to solidification by microbe
metabolism in the future.

200
0

50

100
150
200
CaCO3 precipitation (kg/m3)

250

Figure 5. Relationship between CaCO3 precipitation and secant modulus


(axial strain 0.4%)

3.5 Relationship between CaCO3 precipitation and axial strain


at maximum principal stress difference
Figure 6 shows the relationship between CaCO3 precipitation
and axial strain at the maximum principal stress difference. For
each relative density, when CaCO3 precipitation is less than 60
kg/m3, the axial strain at the maximum principal stress
difference increases linearly according to the increase in
precipitation. When CaCO3 precipitation is more than 60 kg/m3,
the axial strain at the maximum principal stress difference
remains in the vicinity of 0.5% and shows no difference at each
relative density.
Axial strain at maximum principle stress
difference (%)

800

Dr = 30 %

1200

Residual stress (kN/m2)

Maximum principal stress difference


(kN/m2)

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

10.0
Dr = 30 %

8.0

Dr = 60 %
Dr = 85 %

6.0
4.0
2.0
0.0

50

100
150
200
CaCO3 precipitation (kg/m3)

250

Figure 6. Relationship between CaCO3 precipitation and axial strain at


maximum principal stress difference

3.6 Relationship between CaCO3 precipitation and residual


stress
Figure 7 shows the relationship between CaCO3 precipitation
and residual stress. The residual stress indicates the minimum
principal stress difference, less than axial strain 15%, after the
maximum principal stress difference.
There is little change in the residual stress at each relative
density up to about 30 kg/m3 of CaCO3 precipitation in

2616

5 REFERENCES

Victoria S. W., Leon A. P. and Marien P. H. 2007. Microbial carbonate


precipitation as a soil improvement technique. Geomicrobiology
Journal 24, 417-423.
Y. Inagaki, M. Tsukamoto, H. Mori, S. Nakajima, T. Sasaki and S.
Kawasaki. 2011. A centrifugal model test of microbial carbonate
precipitation
as
liquefaction
countermeasure.
Japanese
Geotechnical Journal 6, No. 2, 157-167.

The reinforcement of soils by dispersed oversized particles


Le renforcement des sols par les particules trop grandes non rparties uniformment
Vallejo L.E., Lobo-Guerrero S., Seminsky L.F.

Department of Civil & Environmental Engineering, University of Pittsburgh, Pittsburgh PA, USA

Caicedo B.

Departamento de Ingenieria Civil & Ambiental, Universidad de los Andes, Bogota, Colombia

ABSTRACT: Soils containing dispersed large particles (greater than # 4 sieve) form part of many engineered fills, glacial tills, debris
flows, and residual soil deposits. Very little is known about the effect that the large particles have on the shear strength of the soillarge particles mixtures. In this study, the influence of the large particles on the shear strength of the mixtures was evaluated
experimentally and numerically. The experimental analysis used direct shear tests on simulated granular materials containing large
dispersed particles. The numerical analysis used the Discrete Element Method (DEM). The laboratory and the DEM simulation
results indicated that the shear strength of the mixtures increased with the concentration (Ca) of the simulated large particles in the
mixtures. Also, this study established that the shear strength of the simulated granular materials with dispersed large particles, Sc, can
be obtained if one uses the following relationship: Sc = Sm (1 + M Ca). In this relationship, Sm is the shear strength of the simulated
soil matrix in which the large particles are dispersed, and M is a constant that varied between 1 and 2 for the numerical and laboratory
analyses.
RSUM :Les sols contenant des particules disperses de grande taille (suprieure tamis # 4) constituent la majorit des remblais,
argiles glacires blocs, des coules dboulis et des dpts rsiduels de sol utiliss dans la construction. Peu de travaux existent sur
l'effet que les grosses particules ont sur la rsistance au cisaillement des mlanges de particules de sol de grande taille. Dans cette
tude, l'influence des grosses particules sur la rsistance au cisaillement des mlanges a t valu exprimentalement et
numriquement. L'analyse exprimentale utilise essais de cisaillement direct sur simules matriaux granulaires contenant de grandes
particules disperses. L'analyse numrique utilis la mthode des lments discrets (DEM). Les essais en laboratoire et les rsultats
des simulations ont indiqu que la rsistance au cisaillement des mlanges augmente avec la concentration (Ca) des particules de
grandes tailles simules. En outre, cette tude a tabli que la rsistance au cisaillement des matriaux granulaires simules avec des
grosses particules disperses, Sc, peut tre obtenu si l'on utilise la relation suivante: Sc = Sm (1 + M Ca). Dans cette relation, Sm c'est
la rsistance au cisaillement de la matrice du sol simul dans lequel les grosses particules sont disperses, et M est constante qui varie
entre 1 et 2 pour les analyses numriques et de laboratoire.

KEYWORDS:granular mixtures, shear strength, laboratory tests, DEM analysis.


1INTRODUCTION.
Materials forming part of natural slopes and engineered fills
have a distinct structure, this consisting of a mixture of a soil
matrix (sand) and large particles of gravel that are dispersed
(fragments do not interact) in the soil matrix . The rock
fragments are composed of materials larger than the No. 4 sieve
(Magier and Ravina, 1982; Poesen and Lavee, 1994; Fragaszy
et al. 1992; Budiman, et al., 1995 and Vallejo 1989, 2001) (Fig
1). Soil Mechanics has dealt mainly with the study of three
main soil types: sands, silts, and clays. However, mixtures of
soils such as those shown in Fig. 1 are more commonly found in
nature and in earth construction projects than pure sands, silts
and clays. Since the determination of the mechanical properties
(i.e. shear strength) of mixtures such as those depicted in Fig. 1
has heretofore received scant attention, such an investigation is
indeed called for. This study reports on the mechanisms
involved with the shear strength of simulated granular materials
with dispersed oversized particles.
2 LABORATORY TESTING PROGRAM
2.1

Equipment and simulated granular materials

For the purpose of understanding the mechanisms involved in


the shear strength of granular materials with dispersed large
particles an open face, two-dimensional direct shear apparatus.

Figure 1.Natural slope in Wisconsin made of a soil-rock mixture

was used (Fig. 2) This apparatus is called the Plane stress Direct
Shear Apparatus (PSDSA) (Vallejo, 1991). The granular matrix
will be simulated by a mixture of wooden sticks. Wooden sticks
are strong and can be easily shaved into polygons as their cross
sectionalareas. These polygons resemble the profiles of actual

2617

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

granular materials (Fig. 3) .The wooden sticks forming the


granularmatrix
matrixwill
willhave
have3 3different
differentaverage
averagediameters.
diameters.These
These
granular
will
equal
and
mm.
Thus,
thegranularmatrix
matrix
will
bebe
equal
ofof6,6,
4,4,
and
2.72.7
mm.
Thus,
thegranular
asas
aa
wholewill
willbebemade
madeofofsticks
stickshaving
havingananaverage
averagediameter
diameterequal
equal
whole
mm.The
Theoversized
oversizedlarge
largeparticles
particleswill
willbebesimulated
simulatedbyby
toto4.24.2mm.
roughwooden
woodencircular
circularcylinders
cylinderswith
witha adiameter
diameterequal
equaltoto1212
rough
mm.
The
irregular
sticks
well
circularcylinders
cylinders
have
mm.
The
irregular
sticks
asas
well
asas
thethecircular
have
aa
lengthequal
equaltoto2525mm.
mm.The
Themixture
mixtureofofwooden
woodensticks
sticksand
and
length
cylinders
were
placed
inside
two
forms
that
comprise
box
cylinders
were
placed
inside
two
UU
forms
that
comprise
thethe
box
Plane
Stress
Direct
Shear
apparatus
(PSDSA)(
Figs.
2 and
inin
thethe
Plane
Stress
Direct
Shear
apparatus
(PSDSA)(
Figs.
2 and
Thearea
areainside
insidethethetwo
twoU Uforms
formsis isa square
a squarearea
areawith
withsides
sides
3).3).The
measuring7.67.6cmcmininlength.
length.The
Theopen
openface
faceofofthetheshear
shear
measuring
apparatusformed
formedbybythethetwo
twoU Uforms
formsallows
allowsthetherecording
recordingofof
apparatus
changestaken
takenplace
placeininthethemixture
mixtureduring
duringshearing.
shearing.Two
Two
thethechanges
provingrings
ringsmeasure
measurethethenormal
normaland
andshear
shearforces
forcesapplied
appliedtoto
proving
mixtures.Dial
Dialgauges
gaugesmeasure
measurethethenormal
normaland
andshear
shear
thethemixtures.
displacements.The
Thechanges
changesininfabric
fabricexperienced
experiencedbybythethe
displacements.
mixtureasaswell
wellasasthetheinteraction
interactionbetween
betweenthethegranular
granularmatrix
matrix
mixture
andthethelarge
largeparticles
particlesduring
duringshear
shearwas
wasrecorded
recordedusing
usingdigital
digital
and
photographs
open
face
PSDSA
photographs
ofof
thethe
open
face
ofof
thethe
PSDSA
. .

(c)

Figure 3. Simulated granular mixture in the PSDSA before shearing: (a)


sample
with no oversized
particles,
(b)PSDSA
samplebefore
with shearing:
one oversized
Figure
3. Simulated
granular mixture
in the
(a)
particle,
(c) no
sample
with two
oversized
sample
with
oversized
particles,
(b)particles.
sample with one oversized
particle, (c) sample with two oversized particles.
2.2 Direct shear testing in the PSDSA
2.2 Direct shear testing in the PSDSA

The simulated granular mixtures depicted in Fig. 3 were


subjected
to shear
in the
PSDSA.depicted
The shear
testing
of the
The
simulated
granular
mixtures
in Fig.
3 were
mixturestowere
using two
stresses.of These
subjected
shearcarried
in theout
PSDSA.
Thenormal
shear testing
the
were equal
99.6 and
rate ofstresses.
shearingThese
of the
mixtures
wereto carried
out199.3
usingkPa.
twoThe
normal
mixtures
equal
2mm/min.
Fig. rate
4 shows
the shear
were
equal was
to 99.6
andto 199.3
kPa. The
of shearing
of stress
the
versus the
displacement
the stress
sample
mixtures
washorizontal
equal to 2mm/min.
Fig.relationships
4 shows thefor
shear
containing
the matrix
alone and the
samples with
one sample
and two
versus
the horizontal
displacement
relationships
for the
12 mm inthediameter
cylinders
representing
the one
largeand
particles
containing
matrix alone
and the
samples with
two
12(Fig.3).
mm in diameter cylinders representing the large particles
(Fig.3).

Matrix (99.6kPa)
One oversized particle (99.6kPa)
Matrix
Two (99.6kPa)
oversized particles (99.6 kPa)
One
oversized
particle
Matrix
(199.3
kPa) (99.6kPa)
Two
oversized
particles
(99.6
kPa)
One
oversized
particle
(199.3
kPa)
Matrix
kPa)particles (199.3kPa)
Two (199.3
oversized
One oversized particle (199.3 kPa)
Two oversized particles (199.3kPa)

350
350

300
300

250
Shear stress (kPa)
Shear stress (kPa)

250

Figure
The
Plane
Stress
Direct
Shear
Apparatus
(PSDSA)
Figure
2. 2.The
Plane
Stress
Direct
Shear
Apparatus
(PSDSA)
(Vallejo,
1991)
(Vallejo,
1991)

200

200

(a)(a)

150

150

100
100

50
50

0
0

0
0

(b)(b)

0.02
0.02

0.04

0.06

0.08

0.1

0.04
0.06
0.08 (cm) 0.1
Horizontal
displacement

0.12
0.12

Horizontal displacement (cm)

Figure 4. Shear stress versus horizontal displacement for the samples


tested4.inShear
the PSDSA
Figure
stress versus horizontal displacement for the samples
tested in the PSDSA

The peak values of the shear stress plots of Fig. 4 have been
used
to plot
theofshear
versus
theofarea
The
peak
values
the strength
shear stress
plots
Fig.concentration
4 have beenof
the tolarge
the sample.
areaconcentration
concentrationof is
used
plot cylinders
the shear in
strength
versus This
the area
the cross insectional
area of
thearea
largeconcentration
cylinders in isthe
theequal
largetocylinders
the sample.
This
mixture
divided
the area
of of
thethe
whole
(7.62
equal
to the
cross by
sectional
area
largemixture
cylinders
in cm
the x
7.62 cm)
(Fig. by
3).the
Thearea
resulting
is shown
Fig. 5.
mixture
divided
of the plot
whole
mixturein(7.62
cmThis
x
figure
the shear
strength
the mixture
increases
7.62
cm)shows
(Fig. that
3). The
resulting
plot of
is shown
in Fig.
5. Thisas
the number
of the
large
cylinders
in theincreases
mixture. asAn
figure
shows that
shear
strengthincreases
of the mixture
this increase
is of in
thethe
form:
theequation
numberthat
of represents
large cylinders
increases
mixture. An
Sc =represents
Sm (1 + 2Cthis
equation that
increase is of the form: (1)
a)

(c)

2618

Technical Committee 211 / Comit technique 211

Sc = Sm (1 + 2Ca)

(1)

where Sc is the shear strength of the mixture, Sm is the shear


strength of the matrix, and Ca is the area concentration of the
large cylinders in the mixture. The results of Fig. 5 and Eq. (1)
indicate that the overall shear strength of the simulated granular
mixtures increases with an increase in the number of the large
cylinders. Thus, in the case of real sand-gravel mixtures, it is
expected that the shear strength of these mixtures will increase
with the volume concentration of the gravel in the mixtures.

The tests ended when the horizontal displacement was equal to


5 mm. Also, using a subroutine available in the PFC2D code, one
can obtain the value of the shear stress in function of the
horizontal deformation. In this study, the peak shear resistance
that was measured in the simulationrepresents the shear strength
of the mixture.

340

Shear strength (kPa)

300
260
220
180
140

Lab. results (99.6 kPa)


Sc = Sm(1+2Ca) ; 99.6kPa
Lab. results (199.3 kPa)
Sc= Sm(1+2Ca) ; 199.3 kPa

100
60
0

0.01

0.02

0.03

0.04

0.05

Area concentration, Ca

Figure 5.Shear strength of the simulated granular mixtures in function


of the area concentration of the large cylinders in the mixture.

3 ANALYSIS OF THE LABORATIRY RESULTS USING


THE DISCRETE ELEMENT METHOD (DEM)
3.1

Configuration of the samples

The PFC2D program produced by Itasca (Itasca Consulting


Group Inc., 2002) was used for the simulation of the direct
shear tests on granular material with dispersed oversized
particles. The first step on the configuration of the sample was
the construction of the shear box. The box had two sections
each with a width of 6 cm and a height of 1.5 cm. The two
sections were placed on top of each other and after the circular
particles were generated inside the box, the gap between the two
sections was maintained at 0.5 mm. The depth of the sample
was assumed to be equal to 1 m. The shear and normal stiffness
of the walls forming the box were set to 1x109 N/m. The
coefficient of friction between the circular particles and the
particles and the walls was set to 0.7.
After the construction of the box, 1000 particles representing
the granular matrix and having a diameter of 0.63 mm were
generated inside the box. The density of the particles was set to
2,500 kg/m3, their normal and shear stiffness were set to 1x108
N/m. Their positions were randomly chosen by the program,
having the limitation of no overlap between particles. A normal
gravity field (9.8 cm/sec2) was used during the simulation. In
order to simulate the dispersed oversized particles, 52 particles
of diameter equal to 0.63 mm were removed and replaced by an
oversize particle measuring 5 mm. If an additional oversize
particle was needed to be placed in the sample, the same
number of smaller particles were removed and replaced by
another large particle of 5 mm in diameter (Fig. 6). The tests
were run under a constant normal compressive load equal to
2x104 N. After the normal compressive force was applied to the
sample, the shearing started by moving the upper section of the
shear box to the left with a constant velocity of 0.44 mm/sec.

Figure 6. Simulated samples using DEM that contained zero, one, and
two large dispersed particles.

3.2

Results of the simulations

The DEM simulations of the direct shear tests were carried out
on mixtures having zero, one, and two oversized particles. Figs
7 shows typical DEM results for the samples with zero, one and
three oversized particles. These figures shows the force chains
and their intensity (the thicker the force chains, the bigger are
the force chain value, their maximum values are shown on top
of the figures) for the samples with 3.5 mm of horizontal
displacement.
An analysis of Fig. 7 indicates that the larger force chains which
were compressive in nature were directed toward the large
particles and were transmitted to them by the smaller
surrounding particles. When the horizontal displacement in the
simulated test reached a 3.5 mm value, the force chains were
inclined at about 45 and 135 degrees with respect to the
horizontal axis of the cross sectional area of the large particles.
It is usually assumed that when samples of granular materials
with oversized particles are subjected to either compressive or
direct shear stress conditions, the smaller particles in the
mixture distribute the loads uniformly around the perimeter of
the bigger particles. This uniform load distribution produces
low compressive stresses on the bigger particles which allows
them to survive without breakage (Fragaszy et al., 1992). The
results shown by Fig.7 indicate that this is not the case. Under
direct shear, the smaller particles concentrate on the oversized
particles, large compressive forces that are exerted on a small
section of the perimeter of the large particles. These high
concentrated compressive forces exerted by the smaller particles

2619

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

on the large particles have also been found by Cheng and Minh
(2009) to be effective in granular mixtures.
The peak shear stress values obtained during the shearing of the
mixtures shown in Figs. 6 and 7 were plotted against the area
concentration
ofof
thethe
large
cylinders
in in
thethe
mixture.
The
result
ofof
concentration
large
cylinders
mixture.
The
result
thethe
plot
is is
shown
in in
Fig.
8. 8.
plot
shown
Fig.

which is very similar to Eq. (1).


It should be noted that the DEM simulations did not represent
exactly the shape of the particles forming part of the laboratory
experiments. Also, the sizes of the particles used in the
laboratory
laboratoryexperiments
experimentswere
weredifferent
differentthan
thanthose
thoseused
usedin inthethe
DEM
DEMsimulations.
simulations.However,
However,thethegeneral
generalresults
resultsof ofthethe
laboratory
laboratorytests
testsarearecorroborated
corroboratedbybythetheDEM
DEMsimulations.
simulations.In In
addition,
thethe
DEM
simulations
help
to to
explain
thethe
way
internal
addition,
DEM
simulations
help
explain
way
internal
forces
forcesarearetransmitted
transmittedthrough
throughthetheparticles
particlesin inthethelaboratory
laboratory
experiments.
Thus,
forfor
thethe
case
ofof
real
sand-gravel
mixtures,
it it
experiments.
Thus,
case
real
sand-gravel
mixtures,
is isexpected
expectedthat
thatthetheshear
shearstrength
strengthofofthese
thesemixtures
mixtureswill
will
increase
increasewith
withthethevolume
volumeconcentration
concentrationofofthethegravel
gravelin inthethe
mixtures.
mixtures.Also,
Also,it itshould
shouldbebenoted
notedthat
thatforforthethecase
caseofof
embankments
embankmentsand
andnatural
naturalslopes,
slopes,thetheeffectiveness
effectivenessof ofthethe
oversized
particles
onon
thethe
shear
strength
ofof
thethe
mixtures
forming
oversized
particles
shear
strength
mixtures
forming
these
thesestructures
structureswill
willdepend
dependupon
uponif ifthethelarge
largeparticles
particlesareare
located
or or
near
thethe
critical
failure
surface
(Fig.1).
locatedonon
near
critical
failure
surface
(Fig.1).
4 4CONCLUSIONS
CONCLUSIONS
In Inthethepresent
presentstudy
studythetheshear
shearstrength
strengthofofsimulated
simulatedgranular
granular
mixtures
mixturesmade
madeof ofgranular
granularmatrix
matrixin inwhich
whichlarge
largeparticles
particlesareare
embedded
embeddedwas
wascarried
carriedoutoutusing
usinglaboratory
laboratoryand
andnumerical
numerical
analyses.
Results
from
using
both
approaches
indicated
that
thethe
analyses.
Results
from
using
both
approaches
indicated
that
presence
presenceof ofthethelarge
largeparticles
particleshashasa areinforced
reinforcedeffect
effectin inthethe
mixtures.
That
is,is,
thethe
greater
thenumber
of of
thethe
large
particles
in in
mixtures.
That
greater
thenumber
large
particles
thethe
mixture,
thethe
greater
is is
thethe
shear
strength
ofof
thethe
mixtures.
mixture,
greater
shear
strength
mixtures.
5 ACKNOWLEDGEMENTS
5 ACKNOWLEDGEMENTS
The
work
described
herein
was
supported
byby
Grants
No.
CMS:
The
work
described
herein
was
supported
Grants
No.
CMS:
0124714
0124714and
andCMS:
CMS:0301815
0301815to tothetheUniversity
Universityof ofPittsburgh
Pittsburgh
from
thethe
National
Science
Foundation,
Washington,
from
National
Science
Foundation,
Washington,D.C.
D.C.This
This
support
is is
gratefully
acknowledged
support
gratefully
acknowledged
6 REFERENCES
6 REFERENCES

Figure
7. 7.Force
chains
in in
thethe
samples
with
zero,
oneone
andand
two
large
Figure
Force
chains
samples
with
zero,
two
large
particles
at aathorizontal
shear
displacement
equal
to to
3.53.5
mm.
particles
a horizontal
shear
displacement
equal
mm.

Figure
Figure8. 8.Shear
Shearstrength
strengthversus
versusthethearea
areaconcentration
concentrationof ofthethelarge
large
cylinders
in in
thethe
simulated
mixture.
cylinders
simulatedgranular
granular
mixture.

AnAnanalysis
analysisof ofFig.
Fig.8 8indicates
indicatesthat
thatthethepresence
presenceofofthethelarge
large
cylinders
in in
thethe
mixture
hashas
a reinforcing
effect.
That
is,is,
as as
thethe
cylinders
mixture
a reinforcing
effect.
That
number
numberofoflarge
largecylinders
cylindersincrease
increasein inthethemixture,
mixture,itsitsshear
shear
strength
also
increases.
The
best
fitfit
line
shown
in in
Fig.
8 has
anan
strength
also
increases.
The
best
line
shown
Fig.
8 has
equation
ofof
thethe
form:
equation
form:
+C)
ScS=c S
=mS(1
m (1 + aCa)

Budiman,
J.S.,
Mohamadi,
J., J.,
andand
Bandi,
S. S.
(1995).
Effect
of of
large
Budiman,
J.S.,
Mohamadi,
Bandi,
(1995).
Effect
large
inclusions
onon
liquefaction
of of
sands.
In:In:
Static
andand
Dynamic
inclusions
liquefaction
sands.
Static
Dynamic
Properties
of of
gravelly
Soils,
Evans,
M.D.,
andand
Fragaszy,
R.J.
(eds),
Properties
gravelly
Soils,
Evans,
M.D.,
Fragaszy,
R.J.
(eds),
ASCEs
Geotechnical
Special
Publication
No.No.
56:56:
48-63.
ASCEs
Geotechnical
Special
Publication
48-63.
Cheng,
Y.P.
Minh,
N.H.
(2009).DEM
investigation
of of
particle
size
Cheng,
Y.P.
Minh,
N.H.
(2009).DEM
investigation
particle
size
distribution
effect
onon
direct
shear
behavior
of of
granular
distribution
effect
direct
shear
behavior
granular
agglomerates.Powders
andand
Grains
2009,
M.M.
Nakagawa
agglomerates.Powders
Grains
2009,
Nakagawa
(Editor),
American
Institute
of of
Physics,
New
York,
401-404.
(Editor),
American
Institute
Physics,
New
York,
401-404.
Fragaszy,
R.J.,
Su,Su,
J., J.,
Sidiqqi,
F.H.,
andand
Ho,Ho,
C.L.
(1992).Modeling
Fragaszy,
R.J.,
Sidiqqi,
F.H.,
C.L.
(1992).Modeling
strength
of of
sandy
gravel.Journal
of of
Geotechnical
Engineering,
strength
sandy
gravel.Journal
Geotechnical
Engineering,
ASCE,
118(6):920-935.
ASCE,
118(6):920-935.
Itasca
Consulting
Group,
Inc.
(2002).PFC2D(Particle
Flow
Code
in in
Itasca
Consulting
Group,
Inc.
(2002).PFC2D(Particle
Flow
Code
Two
Dimensions)
version
3.0.
Minneapolis.
Two
Dimensions)
version
3.0.
Minneapolis.
Magier,
J. and
Ravina,
I. (1982).Rock
fragments
andand
soilsoil
depth
as as
Magier,
J. and
Ravina,
I. (1982).Rock
fragments
depth
factors
in in
land
evaluation
of of
Terra
Rossa.Soil
Science
Society
of of
factors
land
evaluation
Terra
Rossa.Soil
Science
Society
America
(SSSA)
Special
Publication
No.No.
13:13:
13-30.
America
(SSSA)
Special
Publication
13-30.
Poesen,
J., J.,
andand
Lavee,
H. H.
(1994).Rock
fragments
onon
toptop
soil:
Poesen,
Lavee,
(1994).Rock
fragments
soil:
significance
andand
processes.
Catena,
23(1-2):
1-28.
significance
processes.
Catena,
23(1-2):
1-28.
Vallejo,
L.E.
(1989).
AnAn
extension
of of
thethe
particulate
model
of of
stability
Vallejo,
L.E.
(1989).
extension
particulate
model
stability
analysis
forfor
mudflows.
Soils
and
Foundations,
2929
(3):1-13.
analysis
mudflows.
Soils
and
Foundations,
(3):1-13.
Vallejo,
L.E.
(1991).
A plane
stress
direct
shear
apparatus
forfor
testing
Vallejo,
L.E.
(1991).
A plane
stress
direct
shear
apparatus
testing
clays.ASCE
Geotechnical
Special
Publication
No.27
(II):
851clays.ASCE
Geotechnical
Special
Publication
No.27
(II):
851862.
862.
Vallejo,
L.E.
(2001).
Interpretation
of of
thethe
limits
in in
shear
strength
in in
Vallejo,
L.E.
(2001).
Interpretation
limits
shear
strength
binary
granular
mixtures.
Geotechnical
Journal,
binary
granular
mixtures.Canadian
Canadian
Geotechnical
Journal,
38:1097-1104.
38:1097-1104.

(2)(2)

which is very similar to Eq. (1).


It should be noted that the DEM simulations did not represent
2620
exactly the shape of the particles forming part of the laboratory
experiments. Also, the sizes of the particles used in the

Analysis of Displacements of GPA in Normally Consolidated Soft Soil


L'analyse des dplacements des GPA dans le sol mou Normalement consolid
Vidyaranya B.

Research Scholar, Osmania University, Hyderabad, India

Madhav M.R.

Professor Emeritus, JNT University, & I.I.T., Hyderabad, India

ABSTRACT: Granular piles (GP) offer most effective and economical solution for ground improvement due to their drainage,
densification and reinforcement actions, GPs mitigate liquefaction induced damages. An anchor placed at the base of the granular pile
and attached to the footing by a cable or rod transfers the applied pullout force to the bottom of the GP termed as Granular Pile
Anchor (GPA). The effective stresses in a normally consolidated saturated soil increase linearly with depth. Consequently, the
undrained strength and the deformation modulus of the soil increase linearly with depth. Analysis of the displacements of granular
pile anchor is presented considering the influence of the linearly increasing undrained modulus of soil and of the GPA with depth on
the load displacement response of the GPA. A parametric study quantifies effects of the length to diameter ratio of GPA, and the
relative stiffness of the GP with respect to that of in situ soil at ground level, on the variations of tip and top displacements of GPA
with applied load, variation of shear stresses and pullout load with depth, etc.
RSUM : Pieux granulaires (GP) offrent une solution plus efficace et conomique pour amlioration des sols en raison de leur
drainage, la densification et des actions de renforcement, les mdecins gnralistes attnuer les dommages induits par liqufaction. Un
point d'ancrage plac la base du pieu granulaire et fixe la semelle par un cble ou une tige transfre la force de traction applique
sur le fond de la GP qualifi de mouillage pieu granulaire (GPA). Les contraintes effectives dans un sol normalement consolid satur
augmente linairement avec la profondeur. Par consquent, la rsistance non draine et le module de dformation du sol augmente
linairement avec la profondeur. L'analyse des dplacements de l'ancre empilement granulaire est prsente compte tenu de l'influence
de plus en plus le module linaire non draine du sol et de l'GPA avec la profondeur de la rponse force - dplacement du GPA. Une
tude paramtrique de quantifier les effets de la longueur par rapport au diamtre de l'GPA, et la rigidit relative du GP par rapport
celle du sol in situ au niveau du sol, sur les variations de pointe et des dplacements suprieurs de GPA avec la charge applique, la
variation de cisaillement contraintes et la charge de retrait avec la profondeur, etc
KEYWORDS: Granular pile anchor, modulus of deformation, homogenous ground, displacements, load transfer.
1

INTRODUCTION.

Granular piles (GP) offer most effective and economical


solutions in soft marine clays to counter-act low undrained
shear strength and stiffness of the deposits. GPs improve the
performance of ground by reinforcement, densification,
increasing bearing capacity and resistance to liquefaction by
increasing strength and stiffness of ground. GPs are ideally
suited as they form elements of low compressibility and high
shear strength. The effective stresses in a normally consolidated
saturated soil increase linearly with depth. As a result the
undrained shear strength and deformation modulus of the soil
also increase linearly with depth. The increase in modulus of
soil and granular material with depth result in reduced loaddisplacements response and increased confinement pressure.
The functional utility of the GP in compression is extended
by placing an anchor at its base to transfer the pullout load or
uplift forces to the base and the assembly is termed as Granular
Pile Anchor (GPA). Granular pile treated expansive soil adjusts
itself to changes in moisture better than an untreated-soil (Phani
Kumar et al., 2004). White et al. (2001) studied the application
of reinforced geopiers for resisting tensile loads and settlement
control. Lillis et al. (2004) reported results from in situ tests on
pullout response of GPA. Kumar et al. (2004) present results
from laboratory and field tests on pullout response of GPA in
cohesive and cohesionless soils. A linear analysis of
displacements of GPA is presented by Madhav et al. (2008).

PROBLEM DEFINITION

A granular pile of length, L, and diameter, d, with the soil and


pile material characterized by moduli of deformation E s and Egp,
and unit weights of s and gp, respectively is considered (Fig.1).
A force, Po, applied at the base of GPA is resisted by the shear
stress, , acting along the periphery of the pile. The force and
the stresses acting on the GPA are depicted in Figure 2a. The
stresses transferred to the in situ soil are shown in Figure. 2b.
The non-homogeneities of deformation moduli of soil, Es and
granular material, Egp (Fig. 3 & 4)are defined by the parameters,
s & gp, to quantify the rate of increase of deformation moduli
of in-situ soil and granular material with depth. The Poissons
ratio of the soil is s. In order to evaluate the upward
displacements of the elements of the soil adjacent to the GPA

Fig. 1 GPA under Pullout

(a)
(b)
Fig. 2 Forces and Stresses acting
on GPA and Soil.

due to the boundary stresses, , the GPA surface is divided in to


n elements of length, L (=L/n). The stress acting on a typical

2621

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

element, j, is j. The displacement at the centre of an element, i,


due to stresses acting on element, j, are obtained by the method
described by Poulos and Davis (1980).
Integrating numerically, the Mindlins equation for a point
load in the interior of a semi-infinite elastic continuum over the
cylindrical periphery of the element, the displacement, s,ij, of
the soil adjacent to the centre of the ith element due to stress, j,
acting on the element, j, considering deformation modulus, Es,
of the soil, increases linearly with the depth as

Es (z) Eso 1s z is obtained as


L
I S ,ij
(1)
d
S ,ij .
. j
Eso
zi

1 s . L

where Is,ij is the soil displacement influence coefficient.


The total soil displacement, s,i, adjacent to node i due to
stresses on all the elements of the GPA, is obtained by summing
up all the displacements at node i, as

d
Es0

S ,i

I S ,ij

z
j 1
1 s . i L

2.1 DISPLACEMNTS OF GPA


The vertical displacements of GPA are obtained considering
it to be compressible. Figure 5 depicts the stresses on an
infinitesimal element of GPA of thickness, z. Poulos and
Davis (1980) have established that lateral/radial stresses have
negligible effect on the vertical displacements. Considering the
deformation modulus of the granular material, Egp, to increase
linearly with depth, z, as

(4)

The equilibrium of forces in the vertical direction reduces to

d z 4

0
dz d

(5)

where z is the normal stress in to the GPA. The stressstrain relationship for GPA material, is

(2)

Egp z Egpo. 1 gp . z

dgp

Egp. z
z
Egp.

(6)

dz

The vertical soil displacements adjacent to all the nodes are


collected to arrive at

d '
IS
Eso

(3)

where {s} and {} are respectively the soil displacement


and stress vectors of size, n, and elements of the matrix IS'

are

I S' , ij

IS,

ij

1 .z
s

non-dimensional

soil

displacement

Fig. 5 Stresses acting on an Infinitesimal Element

'

influence coefficient of GPA, where Zi=Zi/L- normalized depth


at ith element.
0

Es/Eso

where z and gp are respectively the axial strain and GPA


displacement. Substituting for modified Egp the stress-strain
Equation 6 is modified as

z/L

z d
z Egp. z Egpo1 gp . . gp
L dz

(7)

Differentiating Equation 7 with respect to depth, z,

0.5

dz
z d
Egpo. gp gp Egpo.1gp. . gp
dz
L dz

L dz
2

2
1

s=0
0.25

0.5

On simplification the differential Equation 8 becomes

Fig. 3 Variation of Es with Depth Effect of s


0

(8)

Egp/Egpo

2
gp d gp
d z
z d gp

Egpo
1 gp . .
.
dz
L dz
L dz

(9)

z/L

Combining Equations 5 and 9 simplify


d
Egpo gp gp
L dz

0.5

2
1

gp=0
1

0.25

0.5

Fig. 4 Variation of Egp with Depth Effect of gp

z d gp 4
(10)
0
1 gp . .

L dz d

Equation 10 is solved along with the boundary conditions: at


z =0 (i.e. at the top of GPA) P=0 (Free boundary) and at z=L
(tip of the GPA), P=P0 (the applied load). Equation 10 written in
finite difference form reduces to

2622

Technical Committee 211 / Comit technique 211

(gp,i1 2gp,i gp,i1) 4


gp (gp,i1 gp,i1)
i
0 (11)
(1gp.zi' )

2.(z)
(z)2
L
Egpo.d

I Ed I .{ } E 4..Ln .d .1.{ } 0
'
gp

'
S

so

where z = (L/n) length of differential element of GP.


Rearranging the terms in Eq. 11
(12)

n2 gp
4
'
i
0
(gp,i 1 gp,i 1) (1 gp.zi ).(gp,i 1 2gp,i gp,i 1)
L2 2.n
E
d
.
gpo
Eq. 12 is written as

a .
i

gp,i 1

2.bi . gp,i ci . gp,i1

4L2
i 0
n .Egpo.d

gp

.zi'
bi 1 gp

ai 1 gp .zi'
ci
2
.
n

where

(13)

1 gp . z i' gp
2.n

ai, bi and ci are displacement influence coefficients.


gp,i and i are respectively the displacement at the centre of
node i and the shear stress on the interface of element, i, of
the GPA. Eq. 13 is written for nodes i = 2 to (n-1). Invoking the
first boundary condition, P=0 implies z=0 and hence strain, z=
0, leads to i.e.,
(14)

gp ,1 gp ,1 '

where gp,1displacement at the imaginary node 1 above


the GPA (Fig. 2a). Eqs. 13 and 14 are combined to arrive at the
finite difference equation for node 1, as
2
a1.gp,1' 2.b1. gp,1c1.gp,2 2 4L 1 0
n .Egpo.d

RESULTS AND DISCUSSION

Equation 20 is solved for the displacements in GPA. The


displacements generated along the GPA length are extrapolated
to obtain the top, 0, and the tip, L, displacements considering
the 1st, 2nd and 3rd elements for the top and n-2, n-1and nth
elements for the tip displacements in the GPA, respectively. The
results are presented for the following ranges of parameters.
L/d: 5, 10, 25 and 50; K: 10 to 10,000; Poissons ratio, s: 0.5,
s = 0, 0.25, 0.5, 1and 2; and gp=0, 0.25, 0.5, 1 and 2.
The influence of s on the variation of the shear stresses
with depth is presented in Figure 6 for L/d=10, K=50, s=0.5
and gp=0, 0.5 and 1. The variations of the shear stresses with
depth are magnified at top as shown in Figure 6(b). The
variations of shear stresses with depth are very similar for both
values of s = 0 and 0.5 and decrease with increasing values of
gp. The shear stresses at the tip decrease from 6.65 to 5.63 and
from 5.25 to 4.39 for s = 0 & 0.5 with gp increasing from 0 to
1 respectively. On the contrary, the shear stresses at the top
increase from 0.35 to 0.41 and 0.77 to 0.86 for s = 0 & 0.5
with gp increasing from 0 to 1 respectively.
The variations of shear stresses with depth as a function of
gp are presented in Figure 7 for L/d=10, K=50, s=0.5 and for
s = 0, 0.5 and 1.0. The plots are magnified for the stresses at
the top in Figure 7(b). The variation of shear stresses with depth
for gp =0 & 0.5 are very similar for all s.
*
3.5

(15)

(16)

4L2
1 0
n2 .Egpo.d

s=0
s=0.5

0.5
1

z/L

c1. gp, 2

gp=0

gp,1

(20)

where {1} is the unit vector.

Eq. 15 reduces to

a . 2.b .

gpo

0.5

All the equations for nodes 1 to (n-1) are collated as

I E 4..nL .d . 0
'

gp

where

gp

(17)

gp

I is the displacement coefficient matrix.

'

The pile displacements equations for nodes 1 to (n-1) are


collated and summarized in Eq. 17. The pile displacement
influence coefficients are
c2
2b3
a4
.
.
.
.
.

.
0
c3
2b4

.
.

.
.

.
.

.
.

.
.

.. .
0 .

.
.

.
.

.
.

.
.

.
.

.
.

.
.
.

.
.
.
.
. an 1

.
.

c4
.
.

.
.
.

.
.
.

.
.
2bn 1

0
0
0

0
.

.
..

.
cn 1

(18)

0.5
1

s=0
s=0.5

0.5

Considering the compatibility of displacements in soil and GPA

S gp

gp=0

z/L

gp

c1
( a1 2b1 )
a

2b2
2

0
a3

0
0

.
.

.
.

.
.

.
.

.
.

(a)

(19)

Combining Eqs. 3 and 17 with Eq. 19

2623

(b)
Fig. 6 Normalized shear stress, * vs. Depth, z/L for L/d = 10,
K=50 & s=0.5 (a) Effect of s & gp.(b) Enlarged at top.

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

*
4

0
0

1.95

1
z/L

z/L

s=0

IU
1.35

0.75
0

gp=0
gp=0.5

0.5
1

0.5
gp=0

0.5

s=0
s=0.5

0.5

(a)
0

Fig. 8 Normalized displacement coefficient, IU vs. Depth, z/L for


L/d=10, K=50 & s=0.5 Effects of s & gp.

z/L

0.5
1

z/L

s=0

IU
1.3

0.65
0

gp=0
gp=0.5

1
0.5

0.5

s=0
gp=0
gp=0.5

Fig. 9 Normalized displacement coefficient, IU vs. Depth, z/L for


L/d=10, K=50 & s=0.5 Effect of gp & s.

(b)
Fig. 7 Normalized shear stress, * vs. Depth, z/L for L/d = 10,
K=50 & s=0.5 (a) Effect of s & gp.(b) Enlarged at top.

The variations of normalized tip displacements with depth


for s = 0 (Es constant with depth) and 0.5 are presented in
Figure 8 for L/d=10, K=50, s=0.5 and gp=0, 0.5 and 1. The
displacement coefficients decrease with gp at the tip, and
increase at the top with gp for gp increasing from 0 to 1. The
variation of IU with depth is relatively large for smaller values of
gp compared to that for gp=1. The displacement coefficient at
the tip decreases from 1.91 to 1.47 for s =0 and from 1.732 to
1.34 for s =0.5. IU decreases by about 30% for s =0.5 in
comparison to that for s =0 (homogenous soil).
The influence of gp, the rate of increase of deformation
modulus of granular material with depth on the variations of the
displacement coefficients, IU with depth for varying s from 0
to 1, L/d=10, K=50 and s=0.5 is presented in Figure 9. The
displacement coefficient, IU, for s=0, at the tip decreases from
1.91 to 1.8 and increases from 1.12 to 1.17 for gp = 0 and 0.5
respectively. Similarly for s=0.5 and 1, the displacements at
the tip decrease from 1.47 to 1.39 and 1.14 to 1.10 while at the
top they increase from 0.87 to 0.90 and 0.71 to 0.74 for gp = 0
& 0.5 respectively.
4

0.5

1.95

CONCLUSION

Analysis of the GPA under the influence of the nonhomogeneities of the deformation moduli of the soil and
granular material is presented in this paper. The shear stresses
near the top of GPA are significantly less for s increasing with
depth. Displacements reduce for the deformation moduli
parameters of soil and granular material increasing linearly with
depth (s=0.5 & gp=0.5). The displacements reduce by 30%
with depth for s=0.5 and 6% for gp=0.5 in comparison to
those for s=0 and gp=0 respectively.

2624

REFERENCES

Kumar P., Ranjan G. & Saran S. (2004). Granular Pile System for
Strengthening of Weak Sub-Soils A Field Study. International
Conf. on Geosynthetics and Geoenv. Engg, Bombay, 217-222.
Lillis, C., Lutenegger, A.J & Adams, M. (2004). Compression and
Uplift of Rammed Aggregate Piers in Clay. Geosupport:
ASCE/GEO Geotechnical Special Publication no. 14, 497-507.
Madhav, M.R., Vidyaranya, B. & Sivakumar, V. (2008). Linear
Analysis and Comparison of Displacements Granular Pile Anchors.
J. Ground Improvement, Issue 161, 31- 41.
Phanikumar, Sharma, R.S., Srirama Rao, A. & Madhav M.R. (2004).
Granular Pile Anchor Foundation (GPAF) System for Improving
the Engineering Behavior of Expansive Clay Beds. Geotechnical
Testing J., ASTM, Vol.27(3), 1-9.
Poulos, H.G. & Davis, E.H. (1980). Pile Foundation Analysis and
Design. John Wiley and Sons, New York, 397.
White, D., Wissmann, K., & Lawton, E. (2001). Geopier Reinforcement
for Transportation Application. Geotechnical News : 63-68.

Bridge foundation on very soft alluvia with stone column ground improvement
Fondation de pont sur alluvions trs mous et amlioration du sol avec des colonnes ballastes
Vlavianos G.J., Marinelli A.K.

Ministry of Development, Competitiveness, Infrastructure, Transport and Networks, Greece

Andrianopoulos K.
Foti S.

Geosynolo Ltd.

ABSTRACT: The present paper proposes technical solutions for a road design project comprising both bridges and high
embankments in the Region of Western Greece, where major geotechnical issues had to be dealt with. The very low P-y reaction of
the soft silty clays and the eventual liquefaction of the silty sand layers embedded within the foundation soil, the high ground water
table and the high seismicity of the area, led to the decision to improve the top part of the natural soil given the necessity for an
acceptable solution in terms of both dimensions and cost. Among other possible methods of soil improvement, the application of stone
columns followed by preloading was selected. A comparative parametric stability analysis of the bridge embankments and the pile
foundations for bridge piers, with or without the presence of stone columns, quantified the benefits from the proposed ground
improvement method and verified that the completion of this project is feasible within acceptable performance, safety and cost limits.
RSUM : La communication propose des solutions techniques pour llaboration dun projet de route qui comprend des ponts et des
remblais de grande hauteur dans la rgion de la Grce occidentale o il a fallu faire face des problmes gotechniques importants.
La trs faible rsistance des argiles limoneuses molles aux sollicitations horizontales et la liqufaction ventuelle des couches du
terrain formes de sables limoneux qui sont contenues dans le sol de la fondation, la nappe phratique leve et la haute sismicit de
la rgion, ont conduit la dcision damliorer la partie suprieure du sol naturel en prenant en considration la ncessit de trouver
une solution acceptable en ce qui concerne les dimensions et le cot. Parmi dautres mthodes damlioration du sol, il a t choisie
lutilisation des colonnes ballastes suivie dun prchargement du sol. Une analyse paramtrique comparative de stabilit des remblais
des ponts et des fondations des piliers des ponts avec ou sans la prsence des colonnes ballastes, quantifient les bnfices obtenus par
lutilisation de la mthode damlioration du sol propose et vrifie que lachvement de ce projet est ralisable avec une performance
acceptable en termes de scurit et de cots
KEYWORDS: road project, bridge foundation, soft alluvia, liquefaction, ground improvement, stone column, preloading.

INTRODUCTION

A significant project for road infrastructure is currently under


way in western Greece, perfecture of Aitoloakarnania,
concerning the construction of a 13,1km part of a public
provincial road connecting the municipality of Astakos to the
bridge of Gouria.
Owner of the project is the Greek State and the Supervising
Authority is the Directorate of Studies for Road Works, General
Directorate of Road Works, Ministry of Development,
Competitiveness, Infrastructure, Transport and Networks.
Following the necessary competitive procedure, the design of
the project was assigned to a joint scheme of specialized design
offices, covering the involved scientific areas.
This paper focuses on the technical solutions proposed for
the geotechnical issues that arose with reference to the stability
of embankments and bridge foundation.
2

PROJECT OVERVIEW

The importance of this project lies in its expected contribution


to the improvement of road access towards western
Aitoloakarnania and mainly the touristic zone of AstakosMytikas-Palairos. It is also anticipated to take over some of the
traffic load of other local axis and to support the increase in use
of an existing tunnel nearby. What is more important though, is
the expected traffic load assumption for the shipbuilding and
industrial zone of Astakos, which in the future will be the base
for development in the whole area.

The realization of the project will improve the accessibility


of the area and will facilitate road connection between cities and
existing or planned infrastructure, decreasing time demands and
improving safety and comfort requirements
This road axis under study forms a part of the connection of
Astakos and the port of Platygiali with the major motorway of
Ionia Odos, passing through the bridge of Gouria and the
existing tunnel of Saint Elias. The road section is 11,0m wide (1
lane per direction). From geotechnical point of view, it is to be
mentioned that the whole project comprises 6 bridges (15-105m
long) and a significant length of embankments between 2 and 7
meters high.
Major geological and geotechnical issues that arise for the
last 10km of the road are related to the very low altitude of the
ground and the lack of inclination, the high ground water table,
the insufficient drainage system and the presence of silty clays
and sands, often with high content of organics. The whole
situation is aggravated by the liquefaction potential of the silty
sand layers embedded within the foundation soil, in connection
with the high seismicity of the area.
During the preliminary design stages, it became obvious
that the most significant geotechnical problems for the
realization of the project would be related to the load bearing
capacity of the soil, the expected subsidence under static
loading and the eventual liquefaction phenomena.
3

GEOTECHNICAL CONDITIONS

The area where the bridge foundation will be constructed


consists of soft and compressible saturated alluvial soils, while
the water table is located at ground level. The prevailing

2625

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

geotechnical conditions at these areas can be simplified in two


main profiles.
Soil profile I (Fig. 1) is encountered in the majority of the
bridge sites. Its main characteristic is the surficial layer of finegrained medium plasticity soil. According to the geotechnical
exploration results, this soil layer consists mainly of low to
medium plasticity silts (ML) and clays (CL), with thin layers of
high plasticity silts (MH), fat clays (CH) and organic clays
(OL). The thickness of this layer varies between about 22.5 to
35m. Below this layer, to the depth of 40m, either a medium to
dense non-cohesive soil unit (SC, SM) or a dense cohesive soil
unit (CL) are present. Rock or any other rock-like geological
formation was not encountered at any of the locations explored.

Figure 1: Representative geotechnical profiles

ch=7x10-7 9x10-6 m2/sec based on CPTu dissipation tests). The


lateral resistance of this layer is also considered very low,
leading to large horizontal displacements and structural forces,
especially during seismic loading.
With regard to the seismic response, profile I belongs to
group type S1 according to EC8. The average shear wave
velocity Vs,30 generally ranges between 85 and 140m/sec, as
computed from the CPT recordings:
Gmax,o = (qt v)x0.0188x100.55Ic+1.68
(2)
where Ic is a soil behavior type index (Robertson, 2009). Thus,
special study is required for the definition of the seismic action,
which will take into account the non-linear response of the soil
layers and the dependence of soil moduli and internal damping
on cyclic strain amplitude.
Profile II (Fig. 1) represents the soil conditions prevailing at
one bridge site. The soil conditions resemble those of Profile I,
with the exception of an 8m thick layer of loose silty sand that
interrupts the surficial fine-grained formation. This noncohesive formation (SM according to USCS) is relatively close
to ground surface (at the depth of 7m), while it is classified as
non-plastic, with fines content between 7 and 20% and
potentially liquefiable under seismic conditions.
A preliminary liquefaction analysis with NCEER
methodology (Youd et al. 2001) for CPT recordings revealed
that this non-cohesive formation is liquefiable. As shown in Fig.
4, the factor of safety against liquefaction is well below unity
for the silty sand layer, revealing its high liquefaction potential.
Hence, although this soil layer presents higher stiffness
(Vs,30=140m/s) and bearing capacity for static loading, as
compared to the clay layer, its liquefaction potential deteriorates
its mechanical properties. Thus, during earthquake loading, loss
of bearing capacity, lateral stiffness degradation and settlements
are expected to occur, increasing this way superstructure
displacements and structural forces. Furthermore, Profile II is
now characterized as Group type S2 according to EC8 and
special study is needed to define the seismic action and the
exact liquefaction potential.
4

Figure 2: Distribution of undrained shear strength with depth for profile


I conditions, before and after the improvement

Figure 2 presents an estimation of the undrained shear


strength of the surficial fine-grained soil unit of Profile I, based
on the results of typical CPT & SPT recordings. An estimation
of undrained shear strength for normally consolidated clays is
also presented, based on Jamiolkowski et al. (1985) (see Eq.1):
Su = 0.25 vo
(1)
where vo is the geostatic effective vertical stress. Comparing
these two estimations, it is concluded that the surficial finegrained layer is normally or even at some depths underconsolidated, with low values of undrained shear strength. Thus,
the bearing capacity of this formation is considered low and
significant settlements are expected during loading, with the
necessary consolidation time to exceed the acceptable time
limits (horizontal coefficient of consolidation ranging between

2626

DESIGN CONCEPT

As a result of the existing poor soil conditions, the foundation


of the foreseen bridge piers on surface foundations was
excluded and was replaced by a group of piles with a rigid pile
cap. However, due to the high seismicity of the area, the very
low P-y reaction of the soft silty clays and the eventual
liquefaction of the silty sand layer led to extreme internal forces
of the piles and increased dis-proportionally the cost of the
project. Hence, the necessity of an acceptable solution in terms
of both dimensions and cost, led to the decision to improve the
top part of the natural soil.
Among a number of possible methods of soil improvement
that were examined, it was decided to proceed with the
application of gravel piles followed by preloading. Plastic
drains are also prescribed to act as secondary drainage system
for greater soil depths.
The main aim of pre-loading was to increase the undrained
shear strength of the surficial fine-grained soil unit. The
improved undrained shear strength (when the increase of
effective stress due to surcharge exceeded 10% of its initial
value), was estimated according to Eq. 4:
Su,f = Su,o OCR0.8
(3)
with Su,o reflecting the anticipated undrained shear strength for
normally consolidated clays (see Eq. 2). The increase of
effective vertical stress at each depth was computed according
to the well known Westergaard solutions, taking into account
the increase of soil stiffness at upper layers, where gravel pile
installation accompanies pre-loading. The effect of pre-loading
reduces with depth, while a percentage of the surcharge load is
used for the increase of OCR, due to the distribution of the
external load between gravel piles and original soil. Despite
that, the anticipated increase of undrained shear strength at
upper layers (i.e. at layers that are crucial for the overall safety

Technical Committee 211 / Comit technique 211

of the bridge embankments) is considered substantial, while its


secondary effects such as the acceleration of consolidation at
layers that were found under-consolidated and the reduction of
downdrag forces at piles (i.e. by allowing the consolidating soil
to settle before construction) increase its efficiency. The
prescribed pre-loading embankment were wider from the bridge
embankment / pile cap by 2.5-3.0m at each side, in order to
apply uniform stress at the area of interest, while its height
generally varies between 3 and 7m.
Stage construction of pre-loading embankment was decided
(with height increments between 1.5-2.0m), due to the poor soil
conditions, followed by continuous settlement and porepressure dissipation recordings. Figure 2 presents the
anticipated final (after improvement) distribution of Su with
depth for the CPT recording presented in Section 3.
Gravel pile installment is prescribed ahead of pre-loading,
consisting of 0.80m diameter piles in a 1.80 x 1.80m square
arrangement (denoting replacement percentage equal to as=
0.78x (0.8/1.8)2 =15.4%). Gravel pile length varies between 8
and 13m, depending on soil conditions.
The installation of gravel piles increased the mechanical
properties of the upper cohesive fine-grained layers and
subsequently increased the general stability of bridge & preloading embankments. The following equivalent strength
parameters were used (Van Impe & De Beer, 1983):
(4a)
ceq = (1-as) Su,f
(4b)
taneq = [nas /(nas +1 as)]tan1
where ceq & eq denote the equivalent cohesion & friction angle
of the composite system respectively, 1 denotes the friction
angle of gravels (assumed equal to 42o), as denotes the
replacement ratio (equal to 0.154) and n denotes the ratio of the
load taken by the gravel pile versus the surcharge load. The
contribution of geostatic stresses is omitted; while outside the
embankment limits (where no surcharge is applied) n equals
1.0. The improved shear strength of the composite system,
combined with the increase of the undrained shear strength due
to pre-loading proved adequate for the construction of the
bridge embankments with acceptable factor of safety under both
static and seismic conditions (e.g. the static F.S. increased from
0.64 to 1.51 for a representative height of 4m).
Note that, besides the improvement of shear strength
characteristics, the inclusion of gravel columns combined with
pre-loading has altered the seismic ground response relative to
free-field. In order to take into account this effect, the shear
wave velocity and the spring stiffness in P-y curves of the
relevant soil layers were appropriately increased. Namely, the
formula presented by Baez & Martin (1993) was used for the
estimation of the maximum shear modulus of the composite
system:
(5)
Gmax,eq = Gmax,i as + Gmax,p (1-as)

unimproved soil coefficient k was computed according to


DIN4014 for bored cast-in-place piles.
For the case of Profile II, where a non-cohesive liquefiable
layer is present, the gravel piles are expected to act as a
countermeasure against liquefaction. The gravel piles will be
constructed via bottom-feed vibro-replacement, while a proper
gradation curve range is prescribed in order to ensure the
effective drainage of earthquake-induced excess pore-pressures.
During vibro-replacement, the non-cohesive layer is expected to
be densified and increase its resistance to liquefaction. Based on
Mizuno et al. (1987), the average measured tip resistance is
expected to increase between gravel piles from 4.5MPa to
9.5MPa, providing an adequate liquefaction resistance. Figure 4
compares results from the preliminary (before improvement)
and the detailed (after ground improvement) liquefaction study,
which show the minimization of liquefaction potential. The
densification of the non-cohesive layer due to pre-loading and
the potential dissipation of excess pore pressures were
conservatively ignored. It is noted that even if densification was
ignored, drainage through gravel piles would retain excess pore
pressure ratio ru well below 0.5, as computed according to Seed
& Booker (1977) and Bouckovalas et al. (2011) for the given
characteristics and gravel pile geometry.

Figure 3: Distribution of shear wave velocity with depth for profile I,


before and after the improvement

where Gmax,eq is the maximum equivalent shear modulus, Gmax,i


is the maximum shear modulus of the fine-grained layer after
pre-loading, Gmax,p is the maximum shear modulus of the gravel
pile and as is the replacement ratio (here 0.154). The maximum
shear modulus of the fine-grained layer after pre-loading was
computed as follows (Weiler, 1988):
Gmax,i = Gmax,o OCR0.5
(6)
where Gmax,o is the maximum shear modulus of unimproved
soil, as computed by Eq. 2. The maximum shear modulus of the
gravel pile was computed assuming a dense configuration
(e=0.55). Figure 3 presents the shear wave velocity profile of
the composite system for the CPT recording of Fig. 1. The
average shear wave velocity Vs,30 for this profile increased from
86 to 140m/s, reflecting stiffer ground conditions. This increase
was also implemented to the P-y curves, by increasing the
horizontal subgrade reaction coefficient k. The increase was
assumed proportional to the ratio Gmax,eq/Gmax,o, while for the

Figure 4: Factor of safety against liquefaction for Profile II, before


(preliminary results) and after improvement (detailed study).

Finally, consolidation process is expected to be accelerated with


the presence of gravel piles. Excess pore pressures for each
loading stage are expected to diminish within 19 days, assuming
conservatively only radial flow towards the gravel piles and

2627

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

horizontal coefficient of consolidation equal to ch=7x10-7m/s2.


The actual consolidation time is expected to be even lower,
considering the actual 2D water flow, the presence of horizontal
layers of higher permeability and the additional discharge from
the secondary pipe drains that are prescribed.
5

SEISMIC GROUND RESPONSE ANALYSES

Besides ground improvement, detailed ground response


analyses were also crucial for the successful completion of the
project. Since, both Profile I & II belong to group type S1 & S2
according to EC8, special study was necessary to define the
proper seismic action and the exact liquefaction potential. Thus,
1D equivalent linear analyses were performed with the
equivalent-linear frequency domain method (e.g. Schnabel et al.
1972). Modulus reduction and hysteretic damping curves were
used as a function of cyclic strain amplitude (Vucetic & Dobry,
1991), and introduced the non-linear behavior of soil layers in
ground response analyses, according to its layers plasticity
index. According to EC8 provisions, three different
accelerograms were used, which cover a wide range of
frequencies and are representative of the seismic region.
Shear wave velocities of the improved ground were
computed according to Eq. 5, while the peak ground
acceleration at bedrock outcrop was calibrated to 0.24g,
according to the Greek Annex of EC8 for the area under
investigation. Since no bedrock was found, artificial bedrock
was used at the end of each borehole, while the bedrock shear
wave velocity was assumed to range between 300 and 550m/s,
providing a high impedance ratio compared with the soil
column characteristics. Thus, radiation damping was
conservatively minimized. Fig. 5 shows representative results
from ground response analyses conducted in Profile II.
Significant de-amplification of the seismic motion is observed,
due to the flexibility of the soil column but also due to the nonlinearity exhibited by the soil layers. The computed peak ground
acceleration at ground surface ranges between 0.20 to 0.24g,
significantly lower from the 0.32g required by EC8 for the
flexible soil type D. Thus, the structural forces due to seismic
loading were significantly reduced, while the factor of safety
against liquefaction was substantially increased.

Figure 5: Distribution of peak ground acceleration with depth for Profile


II using three different accelerograms.

CONCLUSION

The present paper presents details of the technical solution


proposed for a road design project in Western Greece, where
major geotechnical issues had to be dealt with for the

2628

foundation of bridges and high embankments. Geotechnical


investigations revealed very poor soil conditions consisting of
silty clays and sands, often with high content of organics, and
high ground water table that locally appeared on the ground
surface. As a result, the foundation of foreseen bridge piers on
surface foundations was excluded and was replaced by a group
of piles with a rigid pile cap. Among a number of possible
methods of soil improvement that were examined, it was finally
decided to proceed with the application of stone columns
followed by preloading. This way, the following were
accomplished:
increase of the general stability of the bridge embankments
increase of the bearing capacity of foundation soil layers
reduction of internal forces of piles
acceleration of the stage of primary consolidation of silty
clay-sands and
reduction of the liquefaction potential of sandy layers.
All of the above effects were verified by site-specific
computations and implemented to the design of the relevant
works
7

ACKNOWLEDGEMENTS

Authors acknowledge the assistance of Harris Lamaris, Civil


Engineer M.Sc. on the geotechnical investigations and the
preliminary liquefaction analyses.
8

REFERENCES (TNR 8)

Baez J.I., Martin G.R.1993., Advances in the design of vibro systems


for the improvement of liquefaction resistance, Symposium on
Ground Improvement. Vancouver
Bouckovalas G., Papadimitriou A., Niarchos D., Tsiapas D. 2011, Sand
fabric evolution effects on drain design for liquefaction mitigation,
Soil Dynamics & Earth. Eng., 31, 1426-1439
Jamiolkowski M., Ladd C., Germaine J.T., Lancellota R. 1985. New
developments in field and laboratory testing of soils. IX Intern.
Conf. on SMFE, Vol. 1, 57-154.
Mizuno Y., Suematsu N. & Okuyama K., 1987. Design method of sand
compaction pile for sandy soils containing fines. Journal of
JSSMFE, 53-56
Robertson P.K. 2009. Interpretation of cone penetration tests a unified
approach. Can. Geotech. J., 46, 1337-1355
Schnabel P.B., Lysmer J. and Seed H.B. 1972. SHAKE: a computer
program for earthquake response analysis of horizontally layered
sites. Report EERC 72-12, Earthquake Engineering Research
Center, Univ. of California, Berkeley
Seed HB, Booker JR. 1977. Stabilization of potentially liquefiable sand
deposits using gravel drains. Journ. of Geotech. Eng., ASCE, 103
(GT7). 757-768
Van Impe W., De Beer E. 1983. Improvement of settlement behavior of
soft layers by means of stone columns. Proc. 8th European Conf. of
SMFE, Vol. 1, 309-312
Vucetic M, Dobry R. 1991. Effect of soil plasticity on cyclic response.
Journal of Geotech. Eng. ASCE, 117 (1), 89-107
Weiler W.A. 1988. Small strain shear modulus of clay. Proc. ASCE
Conf. on Earth. Eng. & Soil Dyn. II, New York, 331-335
Youd et al. 2001. Liquefaction resistance of soils: Summary report from
the 1996 NCEER and 1998 NCEER/NSF workshops on evaluation
of liquefaction resistance of soils. Journal of Geotech & Geonv.,
ASCE, 127 (10), 817-833

Subgrade improvement measures for the main rescue roads in the urban
redevelopment area HafenCity in Hamburg
Mesures damlioration du sol de fondation des principales routes de secours dans la zone du
ramnagement urbaine de la HafenCity Hamburg
Weihrauch S., Oehrlein S.
Grundbauingenieure Steinfeld und Partner GbR, Hamburg

Vollmert L.
BBG Bauberatung Geokunststoffe GmbH & Co. KG, Espelkamp

ABSTRACT: The level of many streets in Hamburg's HafenCity is being raised by around 3 m to ensure safety from flooding. The
marshy soils in this area mainly clay (a soft clay) and peat necessitate the use of different foundation-soil improvement concepts for
the limitation of settlement; these have to take account of the surrounding infrastructure (buildings, river-bank structures, services). The
Honkongstrasse serves as a good example of such concepts. Stress and deformation measurements were carried out on the site, in which
the concept of a "pile-supported, geogrid-reinforced sand sub-base layer" was used; this paper reports on the results.
RSUM: Dans le quartier dHafenCity Hambourg, la plupart des routes seront surleves jusqu 3 mtres pour atteindre le niveau de
protection contre les eaux. Les sols en place, dorigine marcageuse, sont constitus dargile et de tourbe. Du fait de lhtrognit des
sols, et en fonction des ouvrages existants (btiments, murs de rive, tuyaux enterrs), diffrentes approches ont t appliques pour
amliorer les sols et limiter les tassements. Dans cette communication, le cas dtude de la rue Hongkong sera prsent. Dans ce
projet, des gogrilles sont utilises en renforcement dune couche de forme sableuse pour la rpartition des contraintes. Une
instrumentation complte a t mise en place. Des mesures de contraintes et de dformation ont t effectues. Les rsultats de cette
instrumentation sont prsents dans cette communication.
KEYWORDS: urban redevelopment, soil improvement, geogrid, piled embankment, expanded clay, pre-loading
1.

INTRODUCTION

At the HafenCity in Hamburg, previously part of the harbour area,


the current inner-city area south of the historic warehouse district
is being extended by some 40 % to a total of 157 hectares (Figure
1). 5,500 dwellings for 12,000 residents are to be constructed,
along with office space for 40,000 employees.
In the course of these infrastructure projects, the trafficked
areas with the exception of the quay and embankment
promenades are being raised from the current MSL +5 m to
MSL +7.5 m to + 8.0 m, to make them safe for flood events.

The soil conditions around today's Honkongstrasse (formerly


Magdeburger Strasse) are typical for the HafenCity Hamburg.
Fill material of low bearing capacity overlies soft organic
layers of clay and peat which in turn overly firm sands (Figure
2). Raising the level of the road embankment by approximately
3.0 m would have resulted in long-term settlements of between
300 and 400 mm, and significant differential settlements would
have been expected.
The following subgrade-improvement methods are used to
limit settlement in the HafenCity:
-

pre-emption of settlement with a sand pre-loading layer


and vertical drains

use of lightweight materials (expanded clay)

construction of a piled, geogrid-reinforced sand layer

The choice of method depends largely on the local boundary


conditions. On the Honkongstrasse site, all three methods were
used on different sections. Measurements of stresses and
deformations were carried out on the section with the piled,
geogrid-reinforced sand layer. The paper therefore concentrates in
greater detail on this method.
Figure1: City-planning concept of the HafenCity
Hamburg (as of January 2008)

2.

METHOD DESCRIPTION AND CONCEPT

2.1

Sand pre-loading layer and vertical drains

If the local boundary conditions e.g. space available, existing and


planned services, local structures (buildings, river-bank structures,
sewers) and the time constraints permit, the use of a sand pre-

2629

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

loading layer and vertical drains to pre-empt settlements is an


economical standard method.
To accelerate consolidation, vertical drains are installed in the
area to be loaded. The planned, elevated cross-section is then
constructed, and an additional 2.0 to 3.0 m thick sand pre-load is
placed.
The consolidation of the weak organic layers is monitored over
time using settlement gauges.
After the time required for settlement (usually at least three
months), during which no further construction takes place, the
sand pre-load is removed to the agreed planned height and the
structure passes out of the responsibility of the earthworks
contractor.

MSL +5.2 m (prev. GL)

reinforced-concrete

0.6

filled
old sewer

2.3.1

MSL +4.8 m

Use of lightweight aggregates (expanded clay)

In areas where pre-emption of settlement is impractical because of


local structures or services, lightweight aggregates are often used.
These materials significantly reduce settlements resulting from the
raising of ground levels.
For forecast residual settlements of around 50 to 70 mm, the
lightweight materials are placed to a depth of around MSL + 2 m.
The expanded-clay layer is wrapped in a nonwoven geotextile
to prevent particle displacement and leaching-in of soil.
The use of expanded clay is regulated in the "Merkblatt ber
die Verwendung von Blhton als Leichtbaustoff im Untergrund
von Straen" of the Forschungsgesellschaft fr Straen- und
Verkehrswesen (FGSV).
2.3

0.5mgeogridreinforcedsandlayer

2.2

of a short-term tensile strength of 400 kN/m is therefore assigned


to two layers, which are wrapped round at the edges of the
structure, and overlapped in the upper layer.

Detail A

Figure 2: Cross section piled sand layer


(forinvestors)
reinforcementinlongitudinaldirection,
Secugrid400/40R6

1:1.5

Construction of a piled, geogrid-reinforced sand layer


Construction method and system chosen

weak layers

The system is characterised by vertical columns (lime-cement


treated gravel, unreinforced) and an overlying sand layer
horizontally reinforced with geogrids. Use is made of the arching
effect of the overburden sandy soil while the foundation soil acts
as a bedding layer. In contrast to concrete slabs on (reinforced
concrete-) piles, in which the individual elements are very stiff in
comparison to the surrounding soils, there is a pronounced
interaction between the columns, geogrid and the foundation soil
in the system described here.
In contrast to the methods described under 2.1 and 2.2, the
foundation soil in this system is only subjected to low additional
stresses. The major part of the vertical stresses is transferred in a
concentrated manner by the vertical columns into the firm
foundation soil. The system settlements remain proportionally
very small both during and after the construction phase.
When correctly designed, the system possesses significant
reserves of bearing capacity, so that subsequent interventions and
changes within limits in the foundation soil have no influence on
the serviceability of the structure.
The placement grid of the supporting elements should be
designed to transfer the geogrid loads in an orthogonal manner.
For Hongkongstrasse, this resulted in a rectangular grid with a
spacing of 2.3 m normal to the embankment axis and 2.5 m in the
axial direction; the diameter of the elements was 0.6 m (Figure 2).
Reinforced-concrete columns with continuous steel reinforcement
were used at the edge of the structure to cope with a bending
moment (e.g. should any excavation be required at a later date) as
a result of lateral pressure.
The geogrid-reinforcement is installed 150 mm above the
columns in order to guarantee adequate safety against shear during
the construction phase, and in case of large settlements. The
reinforcement is placed at right angles to the placement grid, so
that the layers are cross-laid in the longitudinal and transverse
directions of the embankment.
In the transverse direction, the constructive situation leaves
practically no room to anchor the geogrid. The design requirement

2630

reinforcementincrossdirection,
Secugrid200/40R6

Figure 3: Detail A sketch of system showing reinforcement placing


(vertical scale greater than horizontal)

2.3.2

Points of constructive relevance near Hongkongstrasse

To avoid the risk of subsequent construction activity of investors


endangering or destroying sections which have already been built,
an area of 1.0 m of the traffic section must be able to be removed.
Fill which intrudes into investor areas must also be removable.
In order to prevent any damage to the embankment-support
system, the geogrid-reinforced fill layer was built with sufficient
overlap (Figure 3). The outer section can thus be removed in the
course of normal earthworks.
2.3.3

Design and verifications

The design of the support system is performed for the columns


and the geogrid-reinforced fill layer in co-ordination with one
another.
The columns are designed for the total load, and the
contribution of the subsoil to load bearing and subgrade reaction
between the columns is neglected.
The geogrid-reinforced layer was designed according to a
verification concept which has already been used and proven itself
several times in HafenCity Hamburg. The limit bearing capacity
was first verified in accordance with a suggestion from Kempfert
et al. (1997). A conservative value for subgrade reaction was used.
This verification procedure does not enable any deformations to
be inferred.
However, verification of serviceability and of deformation
limitation is compulsory for all construction projects in the
HafenCity Hamburg. A complementary design procedure was

Technical Committee 211 / Comit technique 211

therefore adopted, using a method developed and extended from


EBGEO, which at the time of planning was only available in its
2004 draft stage. This extended design method was verified for
similar subsoil and loading conditions (Vollmert et al., 2006). The
anticipated further settlements at the level of the reinforcement
were estimated at less than 50 mm after termination of
construction.
Comparisons with the current EBGEO (2010), available in its
final form now that construction is complete, show that the design
and verification of the system is sufficiently robust to cater even
for the special case "Loss of Subgrade Reaction" in Load Case 3.
It must however be noted, that extreme changes in subgradereaction conditions such as its total loss will result in load
redistribution in the sub-base layer. The long-term influence of
these on the serviceability can currently not be finally estimated
and is the subject of research. Taking serviceability aspects into
account, a conservative, realistic estimate of foundation conditions
is therefore critical in the design of such subsoil-improvement
projects.

3.
3.1

Use of lightweight aggregate (expanded clay)

Lightweight aggregate in the form of expanded clay was used in


the northern area of Hongkongstrasse.
Various active services crossed the construction area and it had
to be ensured that these were not damaged during the excavation
of trenches, around 3.5 m deep. In the area of the services, the
expanded clay was installed in so-called big bags as a foundation
for the services. In the remaining areas, the expanded clay was
placed, spread, and compacted in layers to the design height (base
of sand protection layer) (Figure 5).

Accompanying measurement

The selection of the subgrade-reaction value at the underside of


the reinforced layer is therefore a critical starting parameter for
design. Values for this are usually derived from the stiffness
modulus of the subgrade and the thickness of the weak soil layers.
The actual reaction is, however, a variable whose value depends
on, among other things, the soil improvement during the
construction phase, consolidation processes, the stiffness of the
geogrid-reinforced foundation layer and the sub-base layer.
In order to obtain further information on the subgrade reaction
and the actual behaviour of the structure, it is necessary to
investigate such structures in-situ. During the development of
Hongkongstrasse, boundary conditions for monitoring with a
measurement system were ideal, as not only the foundation soil
conditions, but also the geometric conditions and the alterations
anticipated in the medium term (development of the investor areas
and resulting encroachments into the stress distribution in the
structure) were seen as exemplary for the particular situation of
HafenCity Hamburg.
The objective of the measurement programme is therefore the
investigation and measurement of
-

3.2

the state of stress in the geogrid


the development of soil arching, taking construction and
operational conditions into account
the proportions of the load allocated to the supports and
the weak layer
the influence of variations in water-table level in the
foundation soil, and
the deformation situation at foundation level.

CONSTRUCTION AND MEASUREMENTS RESULTS


Sand pre-loading layer and vertical drains

After the necessary clearing away of foundations, the removal of


old services and the search for unexploded ordinance, in the
southern part of Hongkongstrasse, vertical drains were hydraulically driven into the foundation soil. Plastic strip drains with a
width b of around 100 mm were used.
In installing the strip drains it was required to ensure that a
depth of soft soil of around d = 1.0 m remained under the drains as
a natural barrier to the lower groundwater aquifer (lower sands).
After this, the sand layer (including any pre-load) was placed.
Sand with a silt content 5 % by weight, a uniformity coefficient
U of approx. 2 and a rating of Z0 in accordance with the LAGA
guidelines was to be used. The sands were compacted in layers to
at least medium compaction density.

Settlements [cm]

2.3.4

Below the level at which the sand was laid, settlementmeasuring rods, each adjustable in length and with a base plate
(1 m x 1 m) were installed, and these were continuously monitored
while the pre-load was in place. A plot of settlement against time
is shown in Figure 4.

Figure 4: Development of settlement during loading phase

3.3

Piled, geogrid-reinforced sand base layer

3.2.1
Construction
For the required design working load of 500 kN, partialdisplacement bored columns with a diameter of 420 mm and an
unreinforced, fresh-in-fresh tapered cast cap C25/30 with a cap
diameter of 600 mm were installed by GKT Spezialtiefbau GmbH
working for Eggers Umwelttechnik GmbH.
High-strength, high-modulus geogrids Secugrid 200/40 R6 and
Secugrid 400/40 R6 were used to reinforce the sand base-layer.
The construction of the geogrid-reinforced sand layer had to be
carried out with overlap. A length of grid from the lower layer,
which was laid across the width of the base layer, was rolled and
stored at the edge of the fill. After the lengthwise reinforcement
and the fill had been placed, the stored roll was pulled up,
wrapped round the fill, and overlapped by the length required in
the design calculations (Figure 6). For a workmanlike placing it is
therefore necessary to ensure that the roll overlaps are arranged in
a staggered pattern in order to avoid distortion and twisting of the
geogrids.
The results of the measurements have been published by
Weihrauch et al. (2010) indicating the expected performance as
well as meeting the deformation requirements as given by the
design.

2631

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

such as pipes is not a problem. Perforation of the geogridreinforced base, e.g. by anchor layers, is possible in principle.
The sandfill material can be assumed to incur the lowest cost as
regards the pure construction costs of the structure. Under the
conditions applying at the HafenCity, the pile-supported, geogridreinforced method is usually somewhat cheaper to construct than
an expanded-clay fill layer. Absolute costs must be ascertained
taking into account project-specific reconstruction, protection and
follow-on measures. In addition to the cost aspect, the time
flexibility for individual measures, and their technical feasibility
under local conditions, are decisive criteria.
The authors wish to thank HafenCity Hamburg, Eggers
Umwelttechnik GmbH and Tilman Kaden Landschaftsbau, Bad
Segeberg for their support and permission for publication.
Thanks are due to NAUE GmbH & Co. KG for providing the
instrumented geogrids and technical support.
Figure 5: Placing of expanded clay

5. REFERENCES
EBGEO Empfehlungen fr den Entwurf und die Berechnung von
Erdkrpern mit Bewehrungseinlagen aus Geokunststoffen.
(Recommendations for Design and Analysis of Earth Structures using
Geosynthetic Reinforcements) 2. Auflage 2010. Hrsg. DGGT. Verlag
Ernst & Sohn, Berlin.
Kempfert, H.-G., Stadel, M. &Zaeske, D. (1997): Berechnung geokunststoffbewehrter Tragschichten ber Pfahlelementen. Bau-technik 74
(1997), Heft 12
Vollmert, L., Kahl, M., Giegerich, G. & Meyer, N. (2007): In-situ
verification of an extended calculation method for geogrid reinforced
load distribution platforms on piled foundations. XIV European
Conference on Soil Mechanics and Geotechnical Engineering
(ECSMGE), Madrid, September 2007
Weihrauch, S., Oehrlein, Vollmert, L. (2010): Baugrundverbesserungsmanahmen in der HafenCity Hamburg am Beispiel des
Stellvertreterprojektes
Hongkongstrae.
Vortragsband
zur
31. Baugrundtagung der DGGT, Mnchen, 3. 6. November 2010

Axis of road

Figure 6: Laying of the geogrid

4.

CONCLUDING REMARKS

Three quite different methods of soil improvement techniques


were used in the area of Hongkongstrasse. The prime objective of
the soil improvement is, in all cases, the reduction of subsequent
settlements in order to guarantee the serviceability of the
trafficked areas under operating conditions.
The decisive factors in the choice of construction method are
the local conditions, the time constraints, and the economic
conditions applicable to the section being planned or built.
Settlement pre-emption with vertical drains and a sand pre-load
requires a relatively long lead time. Significant associated
settlements must be expected in the immediate vicinity; in the case
of existing buildings or services these will usually have to be
classified as unacceptable. On the other hand, the foundation soil
remains more or less undisturbed in its stratification and
composition, and is free of construction elements.
In contrast to this, soil improvement using expanded clay, i.e. a
lightweight aggregate, involves only very small additional
settlements. This method is therefore advantageous mainly at the
interface with existing structures, and with existing roads. Because
of the cohesionless nature and the low compressive strength of
expanded clay, difficulties can be experienced later if services and
anchorage layers are installed, or reconstruction work carried out.
With its columns and the geogrid layers, the pile-supported
base-course method contains the most construction components.
Depending on the construction process of the columns, this
method can be used even close to existing structures. Particularly
in cases of large fill depths and these are typical of the
HafenCity services can be constructed at a later date without
problems. And the structure can be continued without a break in
the adjoining construction section; building over existing elements

2632

Fiber Reinforced Cement Treated Clay


Fibro-ciment renforc argile traite
Xiao H.W., Lee F.H., Zhang M.H., Yeoh S.Y.
National University of Singapore, Singapore

ABSTRACT: Cement-treated soil has been used widely in ground improvement for several decades. However, its behavior, especially
at high cement content, is highly brittle. Previous studies have shown that addition of both fibers and cement in soil improvement
seems to be more efficient and attractive than adding fibers or cement alone. This paper presents an experimental study on fiberreinforced cement-treated marine clay. Two different types of fibers and fiber lengths as well as different fiber contents (0.0%-0.32%)
will be investigated with cement content ranging from 20%-50% and water content ranging from 100%-167%. The experiment results
indicate that the strength and ductility of cement-admixed marine clay improve significantly with increasing fiber content until an
optimum fiber content is reached. It was found that the factors affecting the behavior of fiber-reinforced cement-treated marine clay,
such as fiber content, type and cut length and cement soil mix ratio, are not independent. In general, for water content not higher than
100% and cement content higher than 20%, using 12-mm polyvinyl alcohol fiber was found to give higher strength and better
ductility than polypropylene or shorter fibers.
RSUM : Des tudes prcdentes ont dmontr que lincorporation de fibres et du ciment dans l'amlioration de sol semble tre plus
efficace et attrayante qu'en ajoutant seulement les fibres ou le ciment. Cet crit prsente une tude exprimentale de l'argile marine
traite au ciment et renforce avec fibres. Deux diffrents types et longueurs de fibres de mme que des teneurs en fibre diffrentes
(0.0%-0.32%) seront examins avec une teneur en ciment qui stend de 20%-50% et deau qui stend de 100%-167%. Les rsultats
de lexprience indiquent que la rsistance et la ductilit du ciment-sol sont amliores de manire significative avec le contenu
croissant de fibres jusqu' ce qu'une teneur en fibre optimale soit atteinte. Il a t trouv que les facteurs qui affectent le comportement
de l'argile marine traite au ciment et renforce avec fibres, tel que la teneur en fibres, la longueur et le type de fibre et la proportion
du mlange du sol-ciment, ne sont pas indpendants. En gnral, pour une teneur en eau ne dpassant pas les 100% et une teneur en
ciment de plus de 20%, il a t constat que lutilisation de la fibre d'alcool de polyvinyl de 12mm offre une plus haute rsistance et
une meilleure ductilit que le polypropylne ou des fibres plus courtes.
KEYWORDS: fiber reinforcement, cement treated soil, brittleness index, compressive strength
1

INTRODUCTION

Cement-treated soil has been used widely in ground


improvement during the past forty years and is becoming more
attractive and efficient method for soil treatment due to its
economy, availability and feasibility. However, cement-treated
soil, especially at high cement content, tends to be brittle.
Previous studies have shown that fiber-reinforcement increases
the strength and ductility while decreasing the stiffness of the
soil (e.g., Gray & Ohashi 1983). Recent studies have also
shown that incorporation of both fibers and cement in soil
improvement seems to be more efficient than fibers or cement
alone (e.g., Maher and Ho 1993, Consoli et al. 1998). Maher
and Hos work presented a basic study of the mechanical
behavior of artificially cemented sand reinforced with randomly
distributed glass fibers.
The use of randomly distributed fiber as a new reinforcement
material for cement-treated soil has been receiving increasing
attention in recent years (e.g., Consoli et al. 2003, Khattak and
Alarshidi 2006, Tang et al. 2007, Park 2009; Consoli et al 2011;
Ud-din et al. 2011). Previous research works, however, have
focused mainly on low cement content ( 10%), sand or sandy
soil and particular fiber. In this paper, different fiber types (PP
& PVA), fiber lengths (6 & 12mm) and fiber contents (0.0%0.32% by volume of mixture) will be investigated with different
cement and water contents for Singapore marine clay treatment.
The specimens were made in laboratory and tested with
unconfined compression loading. The strength and ductility of
the FRCT marine clay specimens were then analyzed based on
the experiment results.
2

EXPERIMENT INVESTIGATIONS

The materials used in the study are Singapore upper marine


clay, type I Ordinary Portland cement, and fibers. The
constituents of the clay are 24.13% of colloid, 21.77% of clay,

47.71% of silt and 6.39% of very fine to medium sand. Two


different fibers were used, namely polypropylene (PP) and
polyvinyl alcohol (PVA) fibers. PVA fibers are commonly used
in concrete reinforcement to improve the tensile and flexural
strength of concrete. The properties of the fibers are given in
Table 1. A naphthalene-based superplasticizer (Rheobuilder
1000) was used in some mixtures for workability purpose.
Table 1 Physical and mechanical properties of fibers
Fiber

Length

Type

(mm)

Diameter
(micron)

Aspect
ratio

Tensile
strength
(MPa)

Elastic
modulus
(GPa)

Density
(kg/m3)

PP6

26

231

540

910

PP12

12

26

462

540

910

PVA6

26

231

1600

40

1300

PVA12

12

38

316

1500

40

1300

The cement soil mix ratio will be expressed in the form of


S:C:W wherein S is mass of soil solid, C the mass of cement
and W the mass of water at the point of mixing. The cement
content Aw is defined as the ratio of mass of cement to the mass
of soil solid. The water content Cw is defined as the ratio of
mass of water to the total mass of soil solid and cement. In this
study, the cement content ranges from 20 to 50% by weight of
soil solid while the water content ranges from 100 to 167% by
weight of cement and soil solid. The fiber content is defined as
the ratio of volume of fiber to the total volume of the mixture at
the point of mixing, and ranges from 0 to 0.32%. The ductility
of the fiber-reinforced cement-treated (FRCT) soil is designated
herein by the brittleness index (BI), which is defined as the ratio
of the peak strength to the strength at a prescribed post-peak
strain.

2633

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Figure 1 shows the typical stress strain behavior of FRCT soil


specimens under unconfined compression. Compared to purely
cement-treated soil, fiber reinforcement increases the strength
and ductility significantly, the ductility increasing with fiber
content.

1000

300

fiber content
no fiber
0.08%
0.16%
0.24%
0.32%

500

0
0

0.05 0.1 0.15 0.2


axial strain (%)

0.25

(a)

qu (kPa)

450

qu (kPa)

1500

fiber content
no fiber
0.08%
0.16%
0.24%
0.32%

150

0
0

0.05 0.1 0.15 0.2


axial strain (%)

0.25

(b)

0.08 0.16 0.24


fiber content (%)

peak strength qu (kPa)

peak strength qu (kPa)

no fiber
PVA
PP

250

0.32

(a)

0.08 0.16 0.24


fiber content (%)

(b)

Figures 3a-3b show that the BI, evaluated at four different


axial strain levels between 2% and 20%, reduces significantly
with both PP and PVA fiber content for specimens with mix
ratio 2:1:3. A similar trend is also observed from Figures 3c-3d
for specimens with mix ratio 2:1:5. At higher fiber content, that
is, higher than 0.32% for mix ratio 2:1:3 and 0.24% for mix
ratio 2:1:5, the results were very scattered due to the poor
workability of the mix, which can be observed in Figure 1b.
Hence, with current mixing condition, the optimum fiber
content, taking into account performance and workability, is
0.32% and 0.24% for specimens with mix ratio 2:1:3 and 2:1:5
respectively. For mix ratios 2:1:4, 20:7:27 and 5:1:6, the
optimum fiber content is 0.24%.
2.8

2.7

2% of axial strain
5% of axial strain
10% of axial strain
20% of axial strain

2.6
2.4
2.2
2

2% of axial strain
5% of axial strain
10% of axial strain
15% of axial strain

2.4
2.1
1.8
1.5

1.4
1.2

1.2
0.9

1
0

0.08 0.16 0.24


PVA fiber content (%)

0.32

(a)
3

2% of axial strain
4% of axial strain
8% of axial strain
15% of axial strain

2.7
2.4
2.1

0.32

2.7
2% of axial strain
5% of axial strain
10% of axial strain
15% of axial strain

2.4
2.1
1.8

1.8

1.5

1.5

1.2

0.9

0.9
0

(c)

0.08 0.16 0.24


PP fiber content (%)

(b)

1.2
0.08 0.16 0.24
PVA fiber content (%)

0.32

0.08 0.16 0.24


PP fiber content (%)

0.32

(d)

Figure 3. Effect of fiber content on ductility for FRCT soil specimens.


(a) PVA reinforced specimens with mix ratio 2:1:3 (cement content
50%, water content 100%); (b) PP reinforced specimens with mix ratio
2:1:3 (cement content 50%, water content 100%); (c) PVA reinforced
specimens with mix ratio 2:1:5 (cement content 50%, water content
167%); (d) PP reinforced specimens with mix ratio 2:1:5 (cement
content 50%, water content 167%).

2634

0.32

Figure 2. Effect of fiber content on strength for FRCT soil specimens.


(a) Specimens with mix ratio 2:1:3 (cement content 50%, water content
100%); (b) Specimens with mix ratio 2:1:5 (cement content 50%, water
content 167%).

Effect of fiber content

Figures 2-3 present the effect of fiber content on peak strength


and brittle index of FRCT soil specimens. As Figure 2a shows,
for both PP and PVA reinforced specimens with mix ratio 2:1:3
(cement content 50%, water content 100%), the peak strength
increases with fiber content until a certain fiber content (0.160.24%), after which it decreases slightly. A similar trend is also
observed in Figure 2b for specimens with mix ratio 2:1:5
(cement content 50%, water content 167%).

300

200
0

1.8
1.6

Figure 1. Stress strain behavior for FRCT soil specimens with 6mm
long fiber. (a) 50% cement content and 100% water content; (b) 50%
cement content and 167% water content.

3.2

500

350

Brittleness index

Stress strain behavior

3.1

no fiber
PVA
PP

400

Brittleness index

EXPREIMENT RESULTS AND ANALYSIS

1000

Brittleness index

450

1500

Brittleness index

The natural marine clay was first mixed with the prescribed
amount of water to achieve 100% moisture content and
remoulded. Cement slurry with the water-cement ratio needed
to achieve the desired mix ratio was then added to marine clay
in a Hobart Mixer and mixed at a rotational speed of 125rpm for
around 5 minutes. The fiber was finally added to cement soil
mixing and mixed for another 10 minutes. For mixtures with
water content of 100%, the superplasticizer was used to improve
their workability. The dosage of the superplasticizer was 1.132.27l/100kg soil+cement solids. The mixture was placed into a
50mm (diameter) by 100mm (height) cylindrical polyvinyl
chloride (PVC) split-mould. No compaction was applied during
placement. Specimens were then submerged in distilled water
within their split-moulds without loading for curing. The
specimens were then taken out after 7 days for unconfined
compression testing. The test procedure followed those
prescribed in ISO/TS 17892 (2004). The strain rate used for the
unconfined compression test was 1.32%/min.

Technical Committee 211 / Comit technique 211

3.3

Effect of fiber length

For 12mm fiber reinforcement, 0.08% and 0.16% fiber content


was used. It was found that 0.16% fiber reinforcement is much
better than 0.08% fiber reinforcement (see Figure 4). Hence,
only results of 0.16% fiber content were compared. Figure 5
presents the effect of fiber length on the strength and ductility of
the FRCT soil. It can be seen from Figure 5a that for PVAreinforced specimens, 12mm fiber confers distinctly higher
strength than 6-mm fibers. For PP fibers, the difference in peak
strength is small.
Figure 5b shows that for specimens with different mix ratio
and a given type of fiber (PVA), longer fibers consistently give
a lower BI than shorter fibers; the difference increasing with
strain level. Similar trend is also observed in specimens
reinforced by PP fibers. Therefore, the effect of fiber length on
ductility of FRCT soil is significant for both PP and PVA fiber
type.
1.95

0.08%PVA12
0.16%PVA12
0.08%PP12
0.16%PP12

Brittleness index

1.75
1.55
1.35
1.15

content. As discussed in the previous section, the optimum short


fiber content is 0.32% for mix ratio 2:1:3 while it is 0.24% for
mix ratio 2:1:4, 2:1:5, 20:7:27 and 5:1:6. For long fiber, the
optimum fiber content is 0.16%.
Figures 6-7 present the effect of fiber type on the strength
and BI of the FRCA soil. It can be seen from Table 2 and Figure
6 that 12mm PVA fiber reinforcement always gives higher
strength than 12mm PP fiber reinforcement due to PVAs
higher strength. For water content at and below 133% and
cement content not less than 35%, 6mm PVA fiber
reinforcement gives higher strength than 6mm PP fiber
reinforcement due to the same reason above. For low cement
content (20%) and low water content (100%) or high cement
(50%) and high water content (167%), short PP fiber gives
slightly higher strength than short PVA fiber.
Table 2. Peak strength and BI of specimens with different mix ratios
and fiber types, at optimum fiber content.
Mix
proportion

2:1:3

2:1:4

2:1:5

20:7:27

5:1:6

Aw (%)

50

50

50

35

20

Cw (%)

100

133

167

100

100

catogery

high Aw,
low Cw
(HL)

high Aw,
mid Cw
(HM)

high
Aw &
Cw
(HH)

mid Aw,
low Cw
(ML)

low Aw
& Cw
(LL)

1494

752

386

1053

455

(PVA),

(PVA),

(PVA),

(PVA),

1240

632

405

874

471

(PP)

(PP)

(PP)

(PP)

(PP)

1445

761

403

NA

NA

(PVA),

(PVA),

(PVA),

1354

628

313

(PP)

(PP)

(PP)

SF

PVA>PP

PVA=PP

PVA=PP

PVA=PP

PVA<PP

LF

PVA>PP

PVA=PP

PVA>PP

NA

NA

PVA>PP

PVA>PP

PVA

PVA>PP

PVA<PP

0.95
0

5
Axial strain (%)

10

Figure 4. Ductility of RFCT soil with 0.08% and 0.16% 12mm long
fiber (50% cement and 100% water content).
1600

PVA12
PVA6
PP12
PP6

Peak strength (kPa)

1400
1200

Peak
stren
-gth
qu

LF

(kPa
)

1000

Ductility

800
600

Perfor-

400
Specimen type

200
HL

HL

HM

HM

HH

2.1

(short)

(long)
Note: SF and LF denotes short and long fiber respectively
1600
1400

Peak strength (kPa)

Brittleness index

1.7

1200

1.3
1.1

800
600
400
200

0.9
5

10
15
20
Axial strain (%)

PVA6
PP6
PVA12
PP12

1000

1.5

PP
PVA>PP

HH

2:1:3-PVA12
2:1:3-PVA6
2:1:4-PVA12
2:1:4-PVA6
2:1:5-PVA12
2:1:5-PVA6

1.9

(PVA),

mance

(a) Comparison in peak strength for specimens with 50% cement


content, 100%-167% water content, and 0.16% of 12mm or 6 mm long
fiber (HL, HM, HH denote cement and water content, see Table 2).

25

Specimen type
HL HL HM HM HH HH ML LL

(b) Brittleness versus strain for specimens with different mix ratio and
0.16% of 12mm or 6mm long fiber.
Figure 5. Effect of fiber length on strength and ductility for RFCT soil
specimens.

3.4

SF

Effect of fiber type

For simplicity and convenience, FRCT soil specimens are


categorized in Table 2 according to their cement content and
water content together with their performance at optimum fiber

Figure 6. Effect of fiber type on strength of RFCT soil specimens with


20%-50% cement content, 100%-167% water content, 0.24-0.32% of
6mm long fiber or 0.16% of 12mm long fiber.

For 12mm fiber reinforcement, Figure 7a shows that for


specimens with high cement content (50%) and low or high
water content (100% or 167%), PVA fiber gives lower BI
values at four different strain levels between 5%-25% than PP
fiber. For specimens with high cement content and mid high
water content (133%), no fiber type effect is observed. This may

2635

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

be explained as below. For high cement content and low water


content, there is stronger interaction between PVA fiber and
cement-soil body due to PVAs higher strength, which results in
higher ductility in long PVA fiber reinforced soil specimens.
For high cement and water content, there is high PP fiber
concentration due to PPs higher aspect ratio, which induces
lower ductility in long PP fiber reinforced soil specimens.
For 6mm fiber reinforcement, Figure 7b shows that for low
water content (100%) and high cement content (50%), PVA
fiber gives lower BI values at four different strain levels
between 2%-15% than PP fiber. For low cement content (20%)
and low water content, PVA gives higher BI values than PP
fiber. This suggests that short PP fiber interacts with cementsoil body better than PVA fiber of the same length, at low
cement and water content. Figures 7b-7c also show that for midhigh cement (35%) and low water content or high cement and
mid to high water content, fiber type effect is very small.
Therefore, short PVA fiber generally gives ductility not lower
than PP fiber except for low cement content and low water
content.
2:1:3-PP12
2:1:3-PVA12
2:1:4-PP12
2:1:4-PVA12
2:1:5-PP12

1.35
Brittleness index

1.25
1.15
1.05
0.95

10
15
20
Axial strain (%)

25

(a) Brittleness index versus strain for 12mm long fiber reinforcement
(fiber content, 0.16%).
1.18

2:1:3-PVA6
2:1:3-PP6
20:7:27-PVA6
20:7:27-PP6
5:1:6-PVA6

Brittleness index

1.14
1.1

5:1:6-PP6

1.06
1.02
0.98
0

5
10
Axial strain (%)

15

(b) Brittleness index versus strain for 6mm long fiber reinforcement
(low water content, 100%; cement content, 20%-50%; fiber content
0.24-0.32%).

Brittleness index

1.02
1

2:1:4-PVA6
2:1:4-PP6
2:1:5-PVA6
2:1:5-PP6

0.98
0.96
0.94
0.92
0

5
10
15
Axial strain (%)

20

(c) Brittleness index versus strain for 6mm long fiber reinforcement
(mid to high water content, 133%-167%; high cement content, 50%;
fiber content, 0.24%).
Figure 7. Effect of fiber type on ductility for RFCT soil specimens with
20%-50% cement content, 100%-167% water content, fiber cut length
6-12mm and 0.16-0.32% fiber.

In summary, as Table 2 shows, by considering of the


strength and ductility of FRCT soil specimens, PVA fiber
reinforcement is generally better than PP fiber except for low
cement content and water content.

2636

CONCLUSIONS

Based on the results and analysis, the main conclusions may be


drawn as below.
The strength and ductility of cement-treated marine clay was
improved significantly by fiber reinforcement. There is
optimum fiber content, considering performance and
workability of cement-treated soil specimens. PVA fiber
reinforcement is generally better than PP fiber reinforcement
except for low cement content and water content.
It was observed that the fiber cut length has significant effect
on the ductility of cement-treated soil. However, the cut length
effect on strength for PP fiber reinforcement is much smaller
than that for PVA fiber reinforcement.
5

ACKNOWLEDGEMENTS

This research is supported by the National Research Foundation


Singapore under its Competitive Research Programme (CRP
Award No. NRF-CRP 6-2010-03).
5

2:1:5-PVA12

REFERENCES

Consoli, N.C., Prietto, P.D.M., Ulbrich, L.A. 1998. Influence of fiber


and cement addition on behavior of sandy soil. Journal of
Geotechnical and Geoenvironmental Engineering 124 (12), 1211
1214.
Consoli, N.C., Vendruscolo, M.A., Prietto, P.D.M. 2003. Behavior of
plate load tests on soil layers improved with cement and fiber.
Journal of Geotechnical and Geoenvironmental Engineering, ASCE
129 (1), 96101.
Consoli, N.C., Zorta, F., Souza, M., Festugato, L. 2011. Studies on the
dosage of fiber-reinforced cemented soils. Journal of Materials in
Civil Engineering, ASCE 23 (12), 1624-1632.
Gray, D.H., Ohashi, H. 1983. Mechanics of fiber reinforcement in sand.
Journal of Geotechnical Engineering 109 (3), 335353.
ISO/TS 17892, 2004. Geotechnical investigation and testing Laboratory testing of soil. International Organization for
Standardization, Edition 1, Part 7-9.
Khattak, M. J., & Alrashidi, M. 2006. Durability and mechanistic
characteristics of fiber reinforced soil-cement mixtures.
International Journal of Pavement Engineering 7(1), 53-62.
Maher, M.H., Ho, Y.C. 1993. Behavior of fiber-reinforced cement sand
under static and cyclic loads. Geotechnical Testing Journal 16 (3),
330338.
Park, S. S. 2009. Effect of fiber-reinforcement and distribution on
unconfined compressive strength of fiber-reinforced cemented
sand. Geotextiles and Geomembrane 27, 162-166.
Tang, C., Shi, B., Gao,W., Chen,W. and Cai, Y. 2007. Strength and
mechanical behavior of short polypropylene fiber reinforced and
cement stabilized clayey soil. Geotextiles and Geomembranes 25
(3), 194202.
Ud-din S., Marri A. and Wanatowski D. 2011. Effect Of high confining
pressure on the behaviour of fibre reinforced sand. Geotechnical
Engineering Journal of the SEAGS & AGSSEA 42(4), 69-76.

Large-scale Piled Raft with Grid-Form Deep Mixing Walls on Soft Ground
Comportement en vraie grandeur dune fondation mixte radier-pieux tablie dans un sol meuble
amlior par quadrillage de mlange profond de sol
Yamashita K., Wakai S., Hamada J.

Research & Development Institute, Takenaka Corporation, Chiba, Japan

ABSTRACT: This paper offers a case history of a large-scale piled raft supporting a twelve-story building founded on liquefiable
sand underlain by soft cohesive soil in Tokyo. The building, 55.7 m in height above the ground surface and measuring 120 m by 100
m in plan, is a steel-framed structure with a base isolation system of laminated rubber bearings. An assessment of a potential of
liquefaction during earthquakes indicated that the loose clayey sand between depths of 5 and 15 m had a potential of liquefaction
during earthquakes with the peak horizontal ground acceleration of 3.0 m/s2. Therefore, a piled raft combined with grid-form deep
cement mixing walls was employed to cope with the liquefiable sand and also to reduce settlements of the soft cohesive soil below the
sand. To confirm the validity of the foundation design, field measurements were carried out on the foundation settlements, the axial
loads of the piles, the contact pressures between raft and soil and the pore-water pressure beneath the raft from the beginning of
construction to 11 months after the end of construction.
RSUM : Cet article prsente une tude de cas en vraie grandeur d'une fondation mixte radier-pieux dun btiment douze niveaux
construit Tokyo. Cette fondation est tablie dans une couche de sable liqufiable reposant sur une couche de sol cohrent et meuble.
Le btiment, qui fait 120 m par 100 m dans le plan et 55.7 m en hauteur au-dessus du sol, a une structure mtallique en portiques. Il
est isol sa base par un systme dappareils d'appui en lastomre frett. Le potentiel de liqufaction estim du sable argileux entre 5
et 15 m de profondeur serait atteint sous une acceleration horizontale maximum de 3.0 m/s2. Pour pallier ce phnomne et reduire le
tassement de la couche sous-jacente de sol cohrent et meuble, une fondation mixte radier-pieux a t adopte en combinaison avec
lamlioration, en forme de quadrillage, de la couche de sable par mlange profond. Larticle discute le comportement de cette
fondation sur la base dune srie de mesures sur site, qui se sont poursuivies depuis le dbut jusqu onze mois aprs la fin de la
construction du btiment. Le dimensionnement de cette fondation est estim convenable considrant les mesures de tassement, des
forces axiales sur pieux, des contraintes sur le sol et de la pression de leau interstitielle sous le radier.
KEYWORDS: piled raft foundation, deep cement mixing wall, soft ground, field measurements, settlement, load sharing

INTRODUCTION

In recent years there has been an increasing recognition that the


use of piles to reduce raft settlements can lead to considerable
economy without compromising the safety and performance of
the foundation (Poulos, 2001). Detailed investigations of many
high-rise buildings founded on piled rafts in Germany have
been carried out (Katzenbach et al. 2000). Piled raft foundations
have been used for many buildings in Japan and the settlement
and the load sharing between raft and piles have been carefully
investigated for the selected buildings (Yamashita et al. 2011a;
Yamashita et al. 2011b). It has become necessary to develop
more reliable seismic design methods for piled rafts,
particularly in highly seismic areas such as Japan.
This paper offers a case history of a large-scale piled raft
supporting a twelve-story building founded on liquefiable sand
underlain by soft cohesive soil in Tokyo. To cope with the
liquefiable sand and also to reduce settlements of the soft
cohesive soil below the loose sand, piled raft foundation
combined with grid-form deep cement mixing walls was
employed. To confirm the validity of the foundation design,
field measurements were carried out on the foundation
settlements, the axial loads of the piles, the contact pressures
between the raft and soil and the pore-water pressure beneath
the raft from the beginning of construction to 11 months after
the end of construction. During the construction period, the
2011 off the Pacific coast of Tohoku Earthquake struck the site
of the building. The effects of the earthquake on the settlement

and the load sharing between the raft and the piles are also
discussed.
2

BUILDING AND SOIL CONDITIONS

The twelve-story office building is located in Tokyo, 0.3 km


southeast from the twelve-story residential building (Yamashita
et al., 2011b). Figure 1 shows a schematic view of the building
and the foundation with a soil profile. The building, 55.7 m in
height above the ground surface and measuring 120 m by 100 m
in plan, is a steel-framed structure with a base isolation system
of laminated rubber bearings. The foundation levels were
between depths of 3.6 and 7.2 m.
The subsoil consists of an alluvial stratum to a depth of 44 m
below the ground surface, underlain by a diluvial very dense
sand. The ground water table appears approximately 3 m below
the ground surface. The soil profile down to a depth of 15 m is
made of fill which consists of loose clayey sand, sandy clay and
rubble. Between the depths of 15 to 44 m, there lie very soft to
medium silty clay which is slightly overconsolidated with an
OCR of 1.3 or higher. The shear wave velocities derived from a
P-S logging were 150 m/s at the foundation levels and 290 m/s
in the dense sand below the depth of 44 m.
3

2637

FOUNDATION DESIGN

55.7m

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

SPT
N-Value

GL0
10
Depthm

20
30

40

Fill:

Clayey
sand,
Sandy clay,
R bbl

Laminated rubber bearings

Effective
overburden
pressure

3.6m
8.5m

8.5m
14m

20m

20m

Silty clay
Clayey silt
Clayey silt
Sand

60

Silt
Sand and
gravel

7.2m

20m

Grid-form deep
cement mixing walls

35m

Silty clay

50

70

0 10 20 30 40

Shear
Undrained Consolidatio
wave
shear
n
strength
yield stress velocity
Vs
qu/2 (kPa)
Py
0 100 200 300 0 500 1000 15000 200 400 600

47m

Settlement gauges

Sand

9.6m

50m
60m

49.7m

PHC pile

115.2m

9.6m

21

23

25

Figure 1. Schematic view of the building and foundation with soil profile
Penetration

Improved soil
columns

3.1

Ground improvement

An assessment of a potential of liquefaction during earthquakes


indicated that the loose clayey sand between the depths of 5 to
15 m had a potential of liquefaction during earthquakes with the
peak horizontal ground acceleration of 3.0 m/s2. The foundation
level was between depths of 3.6 and 7.2 m, therefore, grid-form
deep cement mixing walls were introduced to cope with the
liquefiable clayey sand below the raft. Figure 2 shows the gridform deep cement mixing walls constructed by TOFT method.
The high-modulus soil-cement walls confine loose sand so as
not to cause excessive shear deformation to the loose sand
during earthquakes. The effectiveness of the TOFT method for
the prevention of liquefaction was confirmed during the 1995
Hyogoken-Nambu earthquake (Tokimatsu et al., 1996).
3.2

1,000mm
800mm

Liquefiable layer
Improved soil
columns

(a) Grid-form deep cement mixing walls


(b) Construction procedure
Figure 2. Grid-form deep cement mixing method
Grid-form deep cement mixing walls

Design of piled raft

Tributary area

INSTRUMENTATION

The locations of the monitoring devices are shown in Figs. 3


and 4. Four piles, P1, P2, P3 and P4, were provided with a
couple of LVDT-type strain gauges at depths of 8.5 m (near pile
head), 20.0 m and 47.0 m (near pile toe) from the ground

Withdrawing

Couple of soil columns lapped and


continuously arranged

The average contact pressure over the raft is 187 kPa. To


improve bearing capacity of the raft, the grid-form deep cement
mixing walls were extended to the depth of 20 m with the
bottom being embedded in the silty clay with undrained shear
strength of 100 kPa or higher. Furthermore, to reduce the
settlement and the differential settlement to an acceptable level,
180 pre-tensioned spun high-strength concrete (PHC) piles of
0.6 to 1.2 m in diameter were used. The pile toes were
embedded in the very dense sand below the depth of 44 m
enough to ensure the toe resistance as well as the frictional
resistance. The pile was constructed by inserting the precast
piles into a pre-augered borehole filled with mixed-in-place soil
cement. Figure 3 shows a layout of the piles and the grid-form
deep cement mixing walls.
4

Tip treatment
Discharge of soildified
material
Stirring and mixing

Pile diameter
1.2m
1.1m, 1.0m

0.9m, 0.8m
0.7m, 0.6m

Figure 3. Layout of piles and grid-form deep cement mixing walls

2638

Technical Committee 211 / Comit technique 211

surface. In the tributary area of the instrumented piles, six earth


pressure cells and one piezometer were installed beneath the raft
at the depth of 7.2 m. Earth pressure cells E1-E3 were installed
on the intact soil and earth pressure cells D1-D3 were installed
on the deep mixing walls. The vertical ground displacements
below the raft were measured by differential settlement gauges.
LVDT-type transducers were installed beneath the raft at depths
of 8.5 m, 14.0 m, 20.0 m, 35.0 m and 50.0 m to measure the
relative displacements to a reference point at a depth of 60.0 m.
The settlements of the foundation were measured at the
monitoring points on the raft by an optical level. The
measurement of the vertical ground displacements was begun
during the excavation for the foundation construction. The
measurement of the axial loads of the piles, the contact
pressures and the pore-water pressure beneath the raft was
begun just before the casting of the 0.6-m thick foundation slab.

9.6m
9.6m
9.6m

E2

9.6m
5

Pile load and contact pressures

Load sharing between raft and piles

Figure 10 shows the time-dependent load sharing among the


piles, the soil, the deep mixing walls and the buoyancy in the
tributary area of the instrumented piles shown in Fig. 4. The
sum of the measured pile-head loads and the raft load in the
tributary area varied from 61.3 to 62.0 MN after the end of
construction, which was generally consistent with the design
load of 64.0 MN. Figure 11 shows the ratios of the load carried

D2

P2

Monitoring devices
:Settlement gauges
:Instrumented pile
:Earth pressure cell
:Piezometer

9.6m
7

9.6m
9

11

Figure 4. Locations of monitoring devices


Mar.11,2011

Ver t i cal di spl acement ( mm)

End of construction

Nov.3,2012

- 10

Depth
8. 5m
14m
20m
35m
50m

- 20
- 30

200

400

Ti me ( days)

600

800

1000

Figure 5. Measured vertical ground displacements below raft


0

Street
Set t l ement ( mm)

Foundation settlement

D1

S1
W
1

E3

Foundation slab
10
-10

Figure 7 shows the development of the measured axial loads of


piles P1-P4. The axial loads also increased considerably just
before the end of construction due to the water pouring.
Thereafter, the pile-head loads reached 4.7-11.2 MN in
November, 2012. Figure 8 shows the distribution of the
measured axial loads on pile P1. Since the piles were
surrounded by the deep mixing walls to a depth of 20.0 m, the
skin friction of the pile shaft between the depths of 8.5 m and
20.0 m was quite small. The average skin friction between the
depths of 20.0 m and 47.0 m was 76 kPa. The ratio of the piletoe load to the pile-head load was 0.21 in November, 2012.
Figure 9 shows the development of the measured contact
pressures between the raft and the soil and the pore-water
pressure beneath the raft. The measured contact pressures
between the raft and the intact soil seemed to reach a state of
equilibrium in early stage of the construction, while those
between the raft and the deep mixing walls increased with
construction loading in the same way as the axial loads of the
piles. The measured contact pressures between the raft and the
deep mixing walls were 137-180 kPa and those between the raft
and the intact soil were 66-72 kPa just before the end of
construction. The measured pore-water pressure was
approximately 40 kPa..
5.3

P3

P1

RESULTS OF MEASUREMENTS

Figure 5 shows the measured vertical ground displacements


below the raft. The measured ground displacement at the depth
of 8.5 m after the casting of the foundation slab, was
approximately equal to the foundation settlement. The
foundation settlement increased considerably just before the end
of construction (November 15, 2011) due to the water pouring
into the underground pits. Thereafter, the foundation settlement
became stable and reached 21 mm 11 months after the end of
construction (November 3, 2012). Figure 6 shows the settlement
profile of the raft measured by the optical level just before the
end of construction. The measured settlements were 12 to 24
mm and the maximum angular rotation of the raft was 1/1400
radian which satisfied the design requirements.
5.2

E1

D3

- 10

- 20

- 30
A

Figure 6. Measured settlement profile of raft


Foundation slab
15

Mar.11,2011

Nov.3,2012

End of construction

P1
P2
P3
P4

10

Axi al l oad ( MN)

5.1

P4

-5

200

400

Ti me ( days)

600

800

1000

(a) Pile-head loads of piles P1 to P4


Mar.11,2011

Foundation slab
15

End of construction

Nov.3,2012

Depth
8. 5m

10

Axi al l oad ( MN)

Tributary area

20m
47m

-5

200

400

600
Ti me ( days)

(b) Axial loads of pile P1


Figure 7. Measured axial loads of piles

2639

800

1000

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Dept h ( m)

Axi al l oad ( MN)


10

15

20

Just after
casting slab

20
30

40

40

50

50

Mar . 29, 2010


Mar . 1, 2011
Mar . 16, 2011
Nov. 15, 2011
Nov. 3, 2012

Figure 8. Axial load distribution on pile P1


Foundation slab

Mar.11,2011

200

Pr essur e ( kPa)

End of construction

Nov.3,2012

D1
D2
D3

150
100
50

Pore-water pressure

0
- 50

200

400

Ti me ( days)

600

800

1000

(a) Contact pressures between raft and deep mixing walls


Mar.11,2011

Foundation slab
200

End of construction

Nov.3,2012

E1
E2
E3

150
100
50

Pore-water pressure

0
- 50

200

400

Ti me ( days)

600

800

1000

(b) Contact pressures between raft and soil


Figure 9. Measured contact pressures and pore-water pressure beneath raft

ACKNOWLEDGEMENTS

The authors are grateful to Messrs. H. Matsuzaki, H. Nagaoka


of Takenaka Corporation and Mr. N. Nakayama (formerly of
Takenaka Corporation) for their contribution to the foundation
design.

Katzenbach, R., Arslan, U. and Moormann, C. 2000. Piled raft


foundation projects in Germany, Design applications of raft
foundations, Hemsley J.A. Editor, Thomas Telford, 323-392.
Poulos, H.G. 2001. Piled raft foundations: design and applications,
Geotechnique 51, No.2, 95-113.
Tokimatsu, K., Mizuno, H. and Kakurai, M. 1996. Building damage
associated with geotechnical problems, Special Issue of Soils &
Foundations, 219-234.
Yamashita, K., Yamada, T. and Hamada, J. 2011a. Investigation of
settlement and load sharing on piled rafts by monitoring full-scale
structures, Soils & Foundations, Vol.51, No.3, 513-532.
Yamashita, K., Hamada, J. and Yamada, T. 2011b. Field measurements
on piled rafts with grid-form deep mixing walls on soft ground,
Geotechnical Engineering Journal of the SEAGS & AGSSEA,
Vol.42, No.2, 1-10.
Yamashita, K., Hamada, J., Onimaru, S. and Higashino, M. 2012.
Seismic behavior of piled raft with ground improvement supporting
a base-isolated building on soft ground in Tokyo, Soils &
Foundations, Vol.52, No.5.

Mar.11,2011

80

End of construction

70

Nov.3,2012

Design load 64.0 MN

60
Load ( MN)

REFERENCES

Foundation slab

50

Piles

40
30

Deep mixing walls

20

Soil

10
0

Buoyancy

200

400

600
Ti me ( days)

800

1000

Figure 10. Time-dependent load sharing between raft and piles in tributary area
Mar.11,2011

Foundation slab

Rat i o of l oad car r i ed by pi l es

30

CONCLUSIONS

Field measurements were carried out on the foundation


settlement and the load sharing between the raft and the piles
for the large-scale piled raft with the grid-form deep cement
mixing walls on soft ground in Tokyo. The foundation
settlement reached 21 mm and the ratio of the load carried by
the piles to the effective load in the tributary area was estimated
to be 0.71 11 months after the end of construction. During the
construction period, the 2011 off the Pacific coast of Tohoku
Earthquake struck the site of the building. Based on the
measurement results, no significant changes in the foundation
settlement or the load sharing were observed after the
earthquake. Consequently, it is confirmed that a large-scale
piled raft, combined with grid-form deep mixing walls, works
effectively in grounds consisting of liquefiable sand and soft
cohesive soil.
7

0
10

20

Effects of earthquake on settlement and load sharing

On March 11, 2011, nine months before the end of construction,


the 2011 off the Pacific coast of Tohoku Earthquake struck the
site of the building. At the site of the twelve-story residential
building, the peak horizontal ground acceleration of 1.75 m/s2
was observed (Yamashita et al. 2012). Although the contact
pressures between the raft and the deep mixing walls were
increased markedly as shown in Fig. 9(a), no significant
changes in the foundation settlement or the load sharing
between the raft and the piles were observed after the
earthquake, as shown in Figs. 5 and 11.

0 10 20 30 40 50

10

Pr essur e ( kPa)

5.4

SPT N-Value

Depthm

by the piles to the effective load and that to the total load in the
tributary area versus time. The ratio of the load carried by the
piles to the effective load was estimated to be 0.70 just before
the end of construction and increased only slightly to 0.71 in
November, 2012. Meanwhile, the ratio of the effective load
carried by the deep mixing walls to the effective load was 0.14
and the ratio of that carried by the intact soil to the effective
load was 0.15 in November, 2012.

1. 0

End of construction

Nov.3,2012

Effective load

0. 8
0. 6

Total load

0. 4
0. 2
0. 0

200

400

Ti me ( days)

600

800

Figure 11. Ratios of pile load to effective load and total load in tributary area

2640

1000

Initial investigation into the carbonation of MgO for soil stabilisation


Premires investigations sur la carbonatation de MgO utilis pour la stabilisation des sols
Yi Y.L.

Institute of Geotechnical Engineering, Southeast University, China, and Department of Engineering, University of
Cambridge, Cambridge, UK

Liska M., Unluer C., Al-Tabbaa A.

Department of Engineering, University of Cambridge, Cambridge, UK

ABSTRACT: While Portland cement (PC) is the most widely used binder for soil stabilisation, there are significant environmental
impacts associated with its production in terms of high energy consumption and CO2 emission. Hence, the development of new low
carbon foot-print alternative cements has been encouraged. In this paper, reactive magnesia (MgO) was used as a soil stabilisation
binder and the MgO-stabilised soils were carbonated by gaseous CO 2 to improve the mechanical properties of the soil and reduce the
CO2 emission. The mechanical and microstructural properties of the carbonated MgO stabilised soils were investigated by using
unconfined compressive testing, x-ray diffraction (XRD) and scanning electron microscopy (SEM). The results showed that the
strength development rates of carbonated MgO-stabilised soils were much faster than those PC- and MgO-stabilised soils, and the
unconfined compressive strength of highly carbonated MgO-stabilised soils was close to that of 28-day ambient cured PC-stabilised
soils. The XRD and SEM results indicated that nesquehonite (MgCO33H2O) was the main product of the carbonated MgO-stabilised
soils and responsible for the significant strength development.
RSUM : Alors que le ciment Portland est le liant le plus utilis pour la stabilisation des sols, il y a d'importants impacts
environnementaux associs sa production en termes de consommation d'nergie leve et d'mission de CO2. Par consquent, le
dveloppement de nouveaux ciments alternatifs basse teneur en carbone a t encourag. Dans cet article, la magnsie ractive
(MgO) a t utilise comme liant pour la stabilisation des sols et les sols stabiliss la magnsie ont t carbonats par du CO2 gazeux
afin d'amliorer les proprits mcaniques des sols et de rduire les missions de CO2. Les proprits mcaniques et microstructurelles
des sols stabiliss la magnsie et carbonats ont t tudies en utilisant des essais de compression simples, la diffractomtrie de
rayons X (DRX), et la microscopie lectronique balayage (MEB). Les rsultats montrent que le dveloppement de la rsistance des
sols stabiliss la magnsie et carbonats tait beaucoup plus rapide que celui du ciment Portland avec prise l'air ambiant et celui
des sols stabiliss la magnsie. Ils ont galement montr que la rsistance la compression uniaxiale des sols stabiliss la
magnsie et carbonats tait proche de celle des sols stabiliss au ciment Portland avec prise l'air ambiant pendant 28 jours. Les
rsultats des DRX et MEB ont indiqus que la nesquehonite (MgCO33H2O) tait le produit principal des sols stabiliss la magnsie
et carbonats, et responsable de la forte augmentation de la rsistance.
KEYWORDS: soil stabilisation, reactive MgO, carbonation, unconfined compressive strength, microstructure.
1

INTRODUCTION

Soil-cement mix technology is one of the most widely used


ground improvement methods, with Portland cement (PC) being
the most commonly employed binder (Sherwood 1993, Bergado
et al. 1996, Al-Tabbaa, 2003). However, there are significant
environmental impacts associated with the production of PC in
terms of high energy consumption and CO2 emissions (0.85t
CO2/t PC), and hence is responsible for 5-8% of anthropogenic
CO2 emissions worldwide (WBCSD, 2002; IPCC, 2004).
In order to reduce the usage of PC, new alternative
cements have been encouraged. Reactive magnesia (MgO)
cements recently emerged as a more sustainable alternative to
PC (Harrison 2008). Reactive MgO is generally calcinated from
magnesite (MgCO3) at temperatures of ~700-800 and should
not be confused with dead burned MgO manufactured at a
temperature higher than 1400, which is known to cause an
unsoundness problem in PC due to its delayed hydration
behavior (Shand, 2006).
Extensive research has been conducted at the University of
Cambridge since 2004 into the reactive MgO cements, as
detailed in Al-Tabbaa (2013). Reactive MgO hydrates, much
faster than dead burned MgO, to form brucite (Mg(OH)2).
Although brucite has a very limited binding ability, it could
carbonate to form one or more of the hydrated magnesium
carbonates,
namely
nesquehonite
(MgCO33H2O),
hydromagnesite (Mg5(CO3)4(OH)25H2O) and/or dypingite

(Mg5(CO3)4(OH)24H2O). The hydration of MgO and


carbonation of brucite both are expansive reactions, which
significantly fill available pores. These hydrated magnesium
carbonates also form well ramified networks of massive crystals
with a very effective binding ability resulting in substantial and
rapid strength increase. For example, in porous construction
blocks, they significantly outperformed corresponding PC
blocks with strengths of 200-300% higher (Liska 2009; Unluer
2012). High levels of carbonation have been achieved in fullscale porous blocks trial production (Liska et al. 2012a and b),
reabsorbing most of the CO2 generated during the
decomposition of the magnesite.
In this paper, reactive MgO was initially used as a soil
stabilisation binder and the MgO-soil samples were carbonated
by CO2 gas to improve the mechanical properties of the soil and
reduce the CO2 emission. This initial work is thereafter
complemented by investigating the impact of relevant variables
including soil type, soil water content, binder content and
carbonation method (Yi et al., 2012), as well as by using a
laboratory-scale auger to model the installation of carbonated
soil-MgO deep mixed columns (Yi et al. 2013).

2641

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

2.1

MATERIALS AND METHODS


Binders, soils and sample preparations

A model soil was used, namely a slightly clayey silty sand,


consisting of 90% sharp sand, 5% kaolin clay and 5% silica
flour. The sharp sand (obtained from Ridgeons, Cambridge,
UK), had a D50 of 0.8 mm and coefficient of uniformity of 4.3,
and the kaolin clay (obtained from Richard Baker Harrison,
Ilford, UK) had a liquid limit of 51% and plastic limit of 30%.
The silica flour was obtained from David Ball Group,
Cambridge. The water content of the soil was 10%. Reactive
MgO (obtained from Richard Baker Harrison, Ilford, UK) and
PC (obtained from Castle Cement, UK) were applied at 13%
dry content to the soil. The reactive MgO had the following
oxide composition: MgO: 97.2%, CaO: 1.2%, SiO2: 1.2%,
Al2O3: 0.2% and Fe2O3: 0.2%, while the PC had: CaO: 63.6%,
SiO2: 13.6%, Al2O3: 10.2%, Fe2O3: 2.7%, SO3: 6.9%, MgO:
0.6% and K2O: 0.9%.
The sharp sand, kaolin clay and silica flour were initially
mixed and homogenised for 5 minutes in a bench-top food
mixer after which water was added and the mixing continued
for further 5 minutes. The MgO was then applied to the model
soil and the entire system was then mixed twice for 5 minutes
with an inspection inbetween the two intervals. The
homogenised mix was then placed in cylindrical moulds, with
50 mm diameter and 100 mm height, applying consistent
moderate compaction in three layers by rodding and eliminating
any trapped air pockets. The samples were demoulded ~1 hour
later, then placed in the triaxial apparatus and subjected to the
carbonation procedures detailed below. A subset of MgO and
PC stabilised samples was also cured in their moulds under
ambient conditions, of 202 and 953% relative humidity,
for 1, 7, 28 and 90 days.
2.2

In addition, an incubator, with 20% CO2 concentration (1 bar)


at 202 and relative humidity of 953%, was also used to
carbonate MgO-stabilised soils for comparison purposes. The
carbonation periods were 12, 24, 48, 96 and 168 hours (7 days).
A subset of samples carbonated for 7 days using this method
were then cured under ambient conditions for 28 and 90 days.
All the samples were tested in triplicate for their unconfined
compressive strength (UCS) at a constant displacement rate of
1.14 mm/min. X-ray diffraction analysis (XRD) and scanning
electron microscopy (SEM) were conducted for selected mixes.
3

RESULTS AND DISCUSSIONS

3.1 Unconfined compressive strength


Figure 2 shows the UCS of MgO-stabilised soils carbonated
using different carbonation methods. The MgO-stabilised soils
carbonated in an incubator, which is generally used to carbonate
the MgO-based construction blocks (Liska 2009; Unluer 2012),
took ~96 hours of carbonation to reach a maximum UCS value
(~8 MPa). While those carbonated in triaxial cell with 200 kPa
CO2 stationary showed similar strength development behavior
but with much faster rate, whereby after ~3 hours the stabilised
soil also reached a similar maximum UCS value. The UCS of
the MgO-stabilised soils carbonated in triaxial cell with 20 kPa
CO2 flow through for 45 minutes was much lower than those
with 200 kPa CO2 stationary, mainly due to the lower CO2
pressure of the former (Yi et al. 2012). However, the attained
strength is approximately the same regardless of the CO2
pressure applied and its concentration. Figure 2 indicates that
there is no need to keep the CO2 flowing through the sample
during the carbonation process, which causes CO2 leakage.
Carbonated in:
incubator (20% CO2 concentration, 1 bar)

Carbonation procedure and testing

A triaxial apparatus was used to permeate pressurised gaseous


CO2 through the MgO-treated soil as shown in Figure 1. The
samples were subjected to 400 kPa confining pressure and then
followed by upward permeation of the gaseous CO2. First, the
CO2 outflow tap was open during the carbonation process, and
the inflow CO2 pressure was maintained at a low value of 20
kPa to reduce leakage. However, the CO2 leakage was still
serious, and hence only four carbonation periods were
conducted using this method, which were 0.75, 1.5, 3 and 6
hours. Thereafter, the carbonation process was modified: the
CO2 outflow tap was closed two minutes after turning on the
CO2 inlet while keeping the inlet open therefore maintaining the
CO2 pressure at the 200 kPa level for the designated carbonation
periods: 0.75, 1.5, 3, 6, 12, 24, 48 and 96 hours. Besides, a
subset of samples carbonated for 24 hours using this method
was then cured under ambient conditions for 7, 28 and 90 days.

triaxial with 20 kPa CO2 flow through

12

Unconfined compressive strength (MPa)

triaxial with 200 kPa CO2 stationary

10

0
1

10

100

Carbonation period (hours)

Figure 2. UCS of MgO-stabilised soils carbonated in incubator, triaxial


cell with 20 kPa CO2 flow through and 200 kPa CO2 stationary.

Figure 3 presents the UCS of uncarbonated MgO- and PCstabilised soils and carbonated MgO-stabilised soils cured under
ambient conditions. It is evident that the UCS values of the
uncarbonated MgO-stabilised soil are much lower than those
PC-stabilised soils, and both of the two mixes took ~28 days to
finish most of their strength development. Comparing of Figure
2 and Figure 3 indicates that the carbonation significantly
increased the UCS of MgO-stabilised soils in a very short time,
and the UCS of highly carbonated MgO-stabilised soils was
close to that of the 28-day PC-stabilised soils, which was ~10
times that of 28-day uncarbonated MgO-stabilised soils.
However, the ambient curing period did not affect the strength
of carbonated MgO-stabilised soil significantly, indicating the
carbonated MgO-stabilised soil could be used to support a
structure just after the completion of the carbonation procedure.

Figure 1.The triaxial cell used for MgO-stablised soil carbonation.

2642

Technical Committee 211 / Comit technique 211

This is confirmed by the results of laboratory-scale auger


installed soil-MgO deep mixed columns (Yi et al. 2013).
PC: uncarbonated
MgO: uncarbonated
MgO: carbonated in triaxial with 200kPa
CO2 stationary for 24 hours

Unconfined compressive strength (MPa)

12

MgO: carbonated in incubator for 7 days

10

(b)

0
0

20

40

60

80

100

Curing period (days)

Figure 3. UCS of uncarbonated MgO- and PC-stabilised soils and


carbonated MgO-stablised soils versus ambient curing period.

3.2 X -ray diffraction


Figure 4 shows the XRD diffractograms of the MgO-stabilised
soils ambient cured for 28 days (uncarbonated) and carbonated
in incubator for 12 hours and 7 days, the MgO-stabilised soils
carbonated in triaxial cell with 20 kPa CO2 flow through for 45
minutes, 3 hours and 6 hours, as well as the MgO-stabilised
soils carbonated in triaxial cell with 200 kPa CO2 stationary for
45 minutes, 3 hours and 4 days. Two strong brucite (Mg(OH)2)
peaks and a weak MgO peak are detected for the 28-day
ambient cured MgO-stabilised soil (Figure 4a). For the
carbonated samples, the disappearance of the brucite peak
together with the gradually weakening of the MgO peak suggest
the carbonation of the MgO resulting in the formation of
nesquehonite (MgCO33H2O). With no other changes in the
mineralogy, this correlates with the rapid and significant
strength development of carbonated MgO-stabilised soils.
Additionally, the MgO peak was detected in all the
carbonated MgO-stabilised soils, including those carbonated in
an incubator for 7 days and in a triaxial cell with 200 kPa CO2
stationary for 4 days, which were much stronger than that in the
28-day ambient cured sample. These results indicated that the
water content (10%) was not enough for all the MgO (13%) to
carbonate to form nesquehonite. Hence the MgO content should
be linked to the water content of the soil, as well as its porosity.

(c)
Figure 4.XRD diffractograms of MgO-stabilised soils: (a) ambient
cured for 28 days and carbonated in incubator; (b)carbonated in triaxial
cell with 20 kPa CO2 flow through; and (c) carbonated in triaxial cell
with 200 kPaCO2 stationary. The abbreviations stand for: B-Brucite, CCalcite, Mg-MgO, N-Nesquehonite, Q-Quartz.

3.3 Scanning electron microscopy


Typical SEM micrographs of the MgO-stabilised soils ambient
cured for 28 days (uncarbonated), carbonated in triaxial cell
with 200 kPa CO2 stationary for 45 minutes and 4 days, as well
as that carbonated in an incubator for 7 days are shown in
Figure 5.
The 28-day ambient cured sample (Figure 5a) exhibits the
presence of brucite, which has a very limited binding ability due
to its poorly interconnected structure (Liska 2009; Unluer 2012).
All the carbonated MgO-stabilised soils (Figure 5b, c and d)
display formation of nesquehonite, consistent with the XRD
results. Besides, brucite was also observed in Figure 5b
confirming that there was only partial carbonation at 45 minutes,
and which then disappeared at 4 days (Figure 5c) confirming
that the carbonation proceeded to a high degree.

(a)

2643

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

CONCLUSION

The results showed that the carbonation significantly increased


the strength of MgO stabilised soil, and the UCS of highly
carbonated MgO stabilised soil was close to that of the 28-day
ambient cured PC-stabilised soil, which was ~10 times that of
28-day uncarbonated ambient cured MgO-stabilised soil. The
carbonation of MgO stabilised soil in the triaxial cell with high
CO2 pressure happened much faster than in the incubator, the
former could be finished in a few hours while the latter took a
few days. Dissimilar to the uncarbonated ambient cured MgOor PC-stabilised soil, the UCS of highly carbonated MgO
stabilised soil did not increase with ambient curing period. The
XRD and SEM results indicated that nesquehonite, one of the
hydrated magnesium carbonates, was the main product of the
carbonated MgO in soil.
(a)

ACKNOWLEDGEMENTS

The experimental work presented in this paper was carried out


at the Geotechnical & Environmental Laboratory, Department
of Engineering, University of Cambridge, in the academic year
2009-2010 when the first author was a visiting researcher there.
The funding from CSC, NSSFC (51279032) and MOST
(2012BAJ01B02-01) of China is gratefully acknowledged. The
authors sincerely appreciate the French abstract translation by
Marine Deruelle.
6

(b)

(c)

(d)
Figure 5.Scanning electron micrographs of the MgO-stabilised soils: (a)
ambient cured for 28 days (uncarbonated); carbonated in triaxial cell
with 200 kPa CO2 pressure for (b) 0.75 hours and (c) 4 days; (d)
carbonated in incubator for 7 days.

REFERENCES

Al-Tabbaa A. 2003. Soil mixing in the UK 1991-2001: state of practice


report. Ground Improvement 7 (3), 117-126.
Al-Tabbaa A. 2013. Reactive magnesia cements. Chapter 19 in EcoEfficient Concrete (Torgal, Jalali, Labrincha & John, Ed),
Woodhead Publishing, in-press.
Bergado D.T., Anderson L.R., Miura N. and Balasubramaniam A.S.
1996. Soft ground improvement in lowland and other environments.
ASCE, New York.
Harrison A.J.W. 2008. Reactive magnesium oxide cements. United
States Patent, 7347896.
Intergovernmental Panel on Climate Change. 2004. Sources of CO2. In
IPCC Special Report on Carbon Dioxide Capture and Storage.
IPCC, Geneva, Switzerland, 77-103.
Liska M. 2009. Properties and applications of reactive magnesia
cements in porous blocks. PhD Thesis, University of Cambridge,
UK.
Liska M., Al-Tabbaa A., Carter, K. and Fifield J. 2012a. Scaled-up
commercial production of reactive magnesia pressed masonry units.
Part I: Production. Construction Materials 165 (4), 211-223.
Liska M., Al-Tabbaa A., Carter K. and Fifield J. 2012b. Scaled-up
commercial production of reactive magnesia pressed masonry units.
Part II: Performance. Construction Materials 165 (4), 225-243.
Shand M.A. 2006, The Chemistry and Technology of Magnesia. Wiley,
New York.
Sherwood T.P. 1993. Soil stabilization with cement and lime: state of
the art review. HMSO Books, London.
World Business Council for Sustainable Development. 2002. The
cement sustainability initiative-our agenda for action. WBCSD,
Conches-Geneva, Switzerland.
Unluer C. 2012. Enhancing the carbonation of reactive magnesia
cement-based porous blocks. PhD Thesis, University of Cambridge.
UK.
Yi Y. L., Liska M., Unler C. and Al-Tabbaa A. 2012. Carbonating
magnesia for soil stabilisation. Canadian Geotechnical Journal,
submitted.
Yi Y.L., Liska M., Unluer C., Akinyugha, A. and Al-Tabbaa A. 2013.
Preliminary laboratory-scale model auger installation and testing of
carbonated soil-MgO columns. ASTM Geotechnical Testing
Journal, in-press.

2644

Innovative solutions in the field of geotechnical construction and coastal


geotechnical engineering under difficult engineering-geological conditions
of Ukraine
Solutions innovantes dans le domaine de la construction gotechnique et de la gotechnique
ctire dans des conditions gotechniques complexes en Ukraine
Zotsenko M., Vynnykov Y.

Poltava National Technical University, Poltava, Ukraine

Doubrovsky M., Oganesyan V.

Odessa National Maritime University, Odessa

Shokarev V.

State Research Institute of Building Constructions, Zaporog'e, Ukraine

Syedin V., Shapoval

Academy of Civil Engineering and Architecture, Dnipropetrovs'k, Ukraine

Poizner M.

ChernomorNllproekt, Odessa, Ukraine

Krysan V.1, Meshcheryakov G.2


1

"RemBud", Dnipropetrovs'k, Ukraine ; 2 Engineering Center Transzvuk, Odessa, Ukraine

ABSTRACT: This paper contains the research findings of reinforced soil cement properties manufactured by drilling mixing method.
Application of soil cement for solution of specific geotechnical problems is considered herein in terms of specific examples In the
field of coastal geotechnical engineering the hybrid coastal hydraulic engineering structures and sheet pilings using reinforced
concrete semi-shells have been developed. Full-scale physical simulation of pressing-in and removal of steel piles is carried out using
a modular coordination pile-pressing system.
RSUM : Cet article prsente les rsultats de ltude des caractristiques de matriau renforc sol-ciment fabriqu par forage et
mlange en place. Lutilisation de mlange sol-ciment pour la rsolution de certains problmes gotechniques est considr partir
dexemples prcis. Dans le domaine de la gotechnique ctire, des ouvrages mixtes ctiers et maritimes ont t construits, ainsi que
des pieux rainures, fabriqus en utilisant des demies coques en bton arm. La simulation physique grande chelle des processus
denfoncement et dextraction des pieux en acier est ralise par application dun systme de coordination modulaire denfoncement
des pieux.
KEYWORDS: soil cement, drilled pile, roll displacement of buildings, pile-pressing system.
1

DRILLED PILES IN UNSTABLE SOILS

Drilled piles installation in unstable soils is specified by


conditions to secure wellbore stability prior to concrete placing.
We know some well-drilling methods in such soils under
protection of well casing being removed at concrete placing or
under drilling mud protection followed by underwater
concreting. SOLETANCHE METHOD of drilled piles
installation is now very much in evidence as well. It involves
the cored screw dipping into the soil to the projected drilling
depth and feeding of concrete mixture through such screw under
pressure with simultaneous screw lifting.
However the aforesaid methods have their faults,
specifically:
application of well casings results in considerable increasing
of works costs, and in some cases its not possible to
remove well casings. Moreover floated soil may escape
from drill hole into the well casing thus damaging
neighboring buildings and constructions;
the way to install reinforcement cage into the piles
manufactured by SOLETANCHE technology seems rather
problematic, whereas drifting sand penetration to the drill
hole at drilling is not improbable;
drilling under the drilling mud and underwater concreting
prevents floated soil escaping into the well casing, but on
the other hand, makes material quality control more
complicated and increases labor intensity.

Mixed technology of drilled piles manufacturing provides


primarily manufacturing of soil-cement elements (SCE) with
dia. 0.8-1.2 m using drilling mixing method or jet method (Van
Impe 2005, Bruce 2000, Larsson 2003). As soon as the soil
concrete reaches 30% of its rated capacity, a well hole shall be
drilled along the elements center line up to the projected
drilling depth, so that the width of a soil concrete shell around it
amounted to 150 mm. It serves to protect a drill hole against
ingress of water and/or unstable soil for a specific period.
Reinforcement cage is mounted in a dry drill hole and the
concreting is performed.
The soil cement shells also provide increasing of bearing
capacity of composite drilled piles at vertical and horizontal
load actions owing to SCE which provide the contact surface
between a pile and the soil.
The development works for manufacturing of 8 meter long
composite drilled piles were carried out in conditions of
bedding of water-saturated loess clay soils.
The following factors were determined in process of the
research conducted:
prism strength and concrete content in soil cement in the
samples taken from the pile shaft after 28 days of concrete
damp curing (, mPa),
SCE diameter (D, mm),
drill well cross section in SCE (D, mm),
availability of loose soil cement in drill hole bottom
drill hole water flow 2 hours prior to surveillance (h, mm),

2645

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

concrete content in soil cement using method of estimation


of part of hydrated cement as for the time of research
conducting (, %),
visual inspection of reinforcement cage mounting,
testing of concreting sampling.
Four average piles were selected from a pile field consisting of
256 piles, which data are specified in Figure 1 (1 fill-up
ground, 5-6 flow loamy soils, 7 semisolid clays) and in
Table 1.

Figure 1 Composite pile diagram: 1 soil cement shell, 2 cast-inplace reinforced concrete pile, 3 reinforcement cage.
Table 1. Test piles installation indices

,
D,
, %
D, mm
mPa
mm
1
3.4
801
22
512

h, m
0.1

3.7

786

21.4

524

3.2

790

20

518

0.05
-

3.5

793

20.6

510

0.1

Findings of pilot works related to installation of composite


drilled piles in soil cement shells testifies the following:
soil cement elements buried in the clay-pans correspond to
the design size and properties. In 7 days after their
installation they can be lightly drilled and hold wellbore
walls in flow clay-bearing soils;
up to 10 cm of loose soil cement remains in drill holes and
can be compacted using earth rammer and impregnated
with fluid soil cement;
drill hole water flow makes ca. 20l per hour, what just
insignificantly influences the results of drill hole concreting;
reinforcement cage mounting and drill hole concreting can
be easily performed.
2

SOIL CEMENT PILES

Prism strength of soil cement manufactured by drilling mixing


method or jet method without using of reinforcing chemical
additives makes 1,54 mPa depending on water and cement
content (M. Zotsenko, Yu. Vynnykov, 2011). In many instances
such material strength seems insufficient for manufacturing of
underground supporting frames, so there is a necessity to
increase the soil cement strength.
This problem can be solved having applied reinforcing of
soil cement structures with steel reinforcement. Correspondence
of thermal-expansion coefficients of these materials apart from
rather high grip of reinforcement on soil cement as well as its
high waterproofing capacity is deemed to be the ground for
collaboration of soil cement and steel reinforcement.

2646

Effect of reinforcement on soil concrete strength was studied


in vitro by testing of pile models of scale 1:4. Models
dimensions made up 100 100 400 mm (Fig. 2). 4 series each
per 6 samples were investigated. Samples of the first series were
not reinforced, while the samples of the following series were
reinforced 1,13%, 2,03% 3,14% (Pf,%) correspondingly. End
surfaces of reinforcement cages were equipped with supporting
plates with longitudinal reinforcement bars rigidly fastened
thereto. Reinforcement protective coating made 20 mm.

Figure 2. Axial compression testing of test samples

Soil cement was produced in vitro using the drilling mixing


method, i.e. no stabilization of loess soil-water-cement mixture
was carried out. M400 Portland cement content amounted to
20% by weight of soil skeleton. The soil-water-cement ratio
with consideration for soil natural humidity made up W/C
(Water/Cement) = 2.7. At that its slump of concrete cone flowability amounted to 11 cm.
Properties of constructive materials used in the above
experiment are shown in Table 2.
Table 2. Materials mechanical properties
Reinforcement properties
Soil cement properties
Es, mPa
Rsc, mPa
Es, mPa
Rb, mPa
210000
225
2000
1.12

All samples were subject to the axial compression test,


during which the average values of their bearing capacity were
determined, see Table 3.
Table 3. Values of bearing capacity of soil cement prisms
Section
Coefficient of
Average values of
reinforcement
variation, v
Series
bearing capacity
No.
percent
N, kN
, %
1
0.00
11.20
0.21
2
1.13
42.50
0.19
3
2.01
62.70
0.17
4
3.14
84.00
0.18

Definition of bearing capacity of steel soil cement prisms by


materials was performed using two methods to select the most
acceptable one for calculation of structural analysis of steel soil
cement structures.
The first method of testing the axially loaded elements
strength with given dimensions, reinforcement quantity and
loads is equated as

Technical Committee 211 / Comit technique 211

N ( Rb A Rsc As ,tot ),

(1)

where N is a required axial force; coefficient equal to 0,92


in this case; Rb prism strength of soil cement; A normal
prism section area of 0,01 m2; Rsc reinforcement rated
compressing strength equal to 225 mPa ; As,tot total area of
main reinforcement.
Rated values of abovementioned indices, as well as model
bearing capacity and expected bearing capacity of full-scale
piles are shown in Table 4.
Table 4. Bearing capacity of soil cement prisms and piles by materials
(using the first calculation method)
As,
N,
N, kN
Series
f, %
mm2
kN full-scale pile
No.
1
0.00
0
11.20
179
2
1.13
113
33.70
661
3
2.01
201
51.91
1019
4
3.14
314
75.30
1478

The second method constitutes the method of calculation of


oblique loaded reinforced concrete elements in accordance with
deformation model in the stress-strain state in its supercritical
stage. Axial compression figures its special case. Dynamic pile
formula for definition by material of standard cross-section is
given by

R ( K U )
NU Ab ARbU b U

,
1 ( K 2)U

(2)

where A, U, K are bearing capacity coefficients.


Table 5 illustrates the bearing capacity of steel soil cement
prisms and expected bearing capacity of soil cement piles.
Table 5. Bearing capacity of soil cement prisms and full-scale piles
(determined by the second method)
Series

u
K
Nu, kN
N, kN,
Nos.
full-scale piles
1
0
1
1.70
11.20
179
2
1.54
1.40
1.70
32.20
635
3
2.73
1.60
1.70
52.41
1033
4
4.27
1.79
1.70
81.68
1616

Rated bearing capacity of soil cement piles by


material, N, kN

Figure 3 contains relationship: bearing capacity of steel soil


cement samples ratio of standard cross-section reinforcement.
It also contains relationships determined analytically according
to the aforementioned methods.

On basis of the research performed it may be concluded as


follows:
Application of longitudinal reinforcement enables to
considerably increase the soil cement strength;
Comparison of data of analytical calculations of soil cement
strength with in vitro data showed that the described
methods of reinforced concrete structures calculation can be
used for calculation of steel soil cement structures strength;
As for the two above calculation methods the calculation
according to the deformation model shall be given
preference, since it is aimed at calculation of structures at
combined loading. Moreover this method seems to be less
sensitive to soil cement parameter variability.
3

USE OF SOIL CEMENT FOR REGULATION OF


SETTLEMENT SPEED OF BUILDINGS AT
RECTIFYING THEIR ROLL DISPLACEMENT

The method of soil drilling out from bottom of less subsided


foundations for rectifying of roll displacement of buildings is
widely used in Ukraine. For this purpose underworking of the
bottom with help of horizontal drill holes of variable diameters
is usually performed (V. Shokarev, V. Shapoval, 2009).
Technological parameters calculation for underworking of
soil under foundation (diameter and pitch of drill holes) is
carried out by formula

S d t,

where S is the required settlement for rectifying of roll


displacement of a building; d drill hole diameter; t pitch of a
drill hole.
Possessing the research findings related to changing of zeroair dry unit weight (d) under bottom of foundation and
experimental data on critical density of soil (sr.) vertical
pressure relationship we can determine the soil layer depth,
where drilling of horizontal drill holes and soil structure
destruction will be carried out

1 d /

Figure 3. Relationship between bearing capacity of soil cement piles by


material N and ratio of standard cross-section reinforcement f based on
the following data: 1 calculation by the first method, 2 calculation
by the second method, 3 according to in vitro data.

(4)

Time of conditional stabilization of buildings settlement


shall be determined by formula

tg
,
V

(5)

where V is the value of conditional stabilization equal to 0.143


cm per day.
Influence coefficient tg shall be determined by quotation

tg

Ratio of standard cross-section reinforcement f , % %


i f

(3)

S S1
,
nt / t1

(6)

where S, S1 means the value of buildings settlement received


according monitoring data; t, t1 means time interval between
monitoring stages.
If stabilization of buildings settlement during rectifying of
its roll displacement takes place prior to fulfillment of required
settlement, additional breaking up of soil under foundation
bottom shall be necessary. In order to forecast the time of
stabilization of buildings settlement after fulfillment of
required settlement the soil cement mortar using drilling mixing
technology shall be fed to the foundation bottom. It allows to
reduce the buildings settlement speed and to achieve the
required value.

2647

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

INNOVATIVE SOLUTIONS FOR COASTAL


HYDRAULIC ENGINEERING

A new structure of the retaining wall has been worked out to


provide, firstly, resistance to heavy loads stipulated by soil
pressure behind the coastal sheet piled wall and, secondly,
simplifying of construction technology for deep water quay wall
due to absence of traditional anchor devices. Retaining wall
(Fig. 4) includes steel sheet piles installed along the structure,
soil backfilling behind the wall and transversal sheet piled rows
(counterforts) connected with axial wall. Counterforts are made
of sheet piles of different length and have a shape of rectangular
trapezoid. The length of the sheet piles in counterforts is
decreased according to the distance from the axial wall and
width of the transversal rows is increased downwards. Sheet
piles heads are joined by the cap. In such structure the idea of
counterforts is not only to increase a rigidity of the retaining
wall while withstanding the applied loads but also to reduce
effectively soil pressure behind the wall.

The basic research for full-scale physical modeling on pressin and extraction equipment was based on the Modular Piling
System. This multifunctional equipment has been developed
with the intended purpose of implanting prefabricated
construction elements using the press-in method. The piling
system is equipped with the original hydraulic piling machine
(Fig. 6) with wedge-operated clamps (testing devise) and a
modular coordinating skidding system (MKS).

Figure 6. Modular section of the piling system MKS: 1 press-in piling


machine; 2 longitudinal guides (skid tracks); 3 transverse guides
(cross slide).

Figure 4. Sheet piling wall with counterforts: a cross-section; b


plan. 1 axial sheet piling; 2 soil backfilling; 3 sheet piled
counterforts; 4 cap.

A new construction of the bulkhead is worked out to increase


the bearing capacity of the coastal protection wall or quay wall
as well as to reduce their material consumption (Fig. 5). The
structure incorporates the front wall and soil backfilling behind
it. Anchor bearers are made of comb type as several small
anchor plates fixed along the rigid core. Anchor force is taken
by all plates of the comb simultaneously. It provides increasing
of the bearing capacity both of the anchorage and of the
structure in whole (keeping the same material consumption for
the anchorage as at the traditional solutions with one large
anchor plate). On the other hand proposed structure may
provide decreasing of the material consumption (keeping the
same structures bearing capacity as at the known facilities).

Figure 5. Anchored bulkhead with anchor bearer of new comb type:


1 sheet piling; 2 backfilling; 3 tie-rod; 4 rigid core; 5 - anchor
plates.

In terms of its impact capability the piling machine is


completely quiet and vibrations in the ground are at an absolute
minimum allowing for the machine to work on certain highly
sensitive ground areas, extremely small spaces, and in historical
preservation areas. The strategic technological advantages of the
piling system are high productivity, precision and quality
control.
Depending on the features of the project, location of the piles
in terms of engineering and geological conditions of the site,
pressing construction elements into the ground could be
provided with the flow-line and coordinating installation
methods.
REFERENCES
Bruce D.A. An introduction to the deep soil mixing methods as used in
geotechnical applications. Report FHWA-RD-99-138 / D.A. Bruce.
U.S. Dept. of Transportation, Federal Highway Administration,
2000. 143 p.
Characteristics of manmade stiff grounds improved by drill-mixing
method / M. Zotsenko, Yu. Vynnykov, I. Lartseva, V. Shokarev, V.
Krysan // Proc. of the 15th European Conf. on Soil Mechanics and
Geotechnical Engineering Geothechnics of Hard Soils Weak
Rocks. Athens. 2011. P. 1097 1102.
Deep mixing research results in under water conditions / [W.F. Van
Impe, R.D. Verstegui Flores, P.O. Van Impe et. al.] // Proc. of the
16th Intern. Conf. on Soil Mechanics and Geotechnical Engineering
(Osaka, 2005). Millpress Science Publishers Rotterdam, 2005.
V. 3. P. 1275 1278.
Larsson S. Mixing processes for ground improvement by deep mixing.
Doctoral thesis / S. Larsson. Stockholm: Royal Institute of
Technology, 2003. 218 p.
Stress-strain state of the system base-strip foundation at elimination
of excessive tilts of buildings / V. Shokarev, V. Shapoval, V.
Chaplygin, R. Samchenko, D.Volkov // Proceedings of the 17th
International Conference on Soil Mechanics and Geotechnical
Engineering, Alexandria, Egypt, 2009. - Published by IOS Press
under the imprint Millpress, 2009. V. 1. P. 680-683.

2648

S-ar putea să vă placă și