Characterisation and modelling of cement treated soil column used as earth retaining structure

The first objective of this study is to understand the material properties of cement-treated Singapore marine clay in terms of compression and tension behavior.. Figure 2.7 Effect of cem

CHARACTERIZATION AND MODELING OF TREATED SOIL COLUMN USED AS CANTILEVER

CEMENT-EARTH RETAINING STRUCTURE

SAW AY LEE

NATIONAL UNIVERSITY OF SINGAPORE

2014

CHARACTERIZATION AND MODELING OF TREATED SOIL COLUMN USED AS CANTILEVER

CEMENT-EARTH RETAINING STRUCTURE

SAW AY LEE

(B Eng (Hons.), UTM; M Sc., NUS)

A THESIS SUBMITTED FOR THE DEGREE OF DOCTOR OF PHILOSOPHY DEPARTMENT OF CIVIL AND ENVIRONMENTAL

ENGINEERING NATIONAL UNIVERSITY OF SINGAPORE

2014

Special thanks to Mr Ang Beng Oon and Mr Foo Hee Ann who patiently helped me in my laboratory tasks I am also grateful to Muhammad Faizal and Dr Xiao Huawen for their willingness to share the material and equipment during the time I conducted my experimental works A special thank you is also extended to Mr Ann Kee Tong and Mr Edward Lim for their help in offering me the resources Grateful acknowledgement is expressed to Dr Namikawa for enlightening me on the tension-softening behavior of cement-treated soil I would also like to thank the fellow colleagues, in particular Hartono, Kok Shien, Yang Yu, Zongrui, Junhui, Xuguang, Sun Jie, Sandi for their company through this journey

Also, I would like to thank my good friends for their support and continuing belief in me Last but not the least, an honorable mention goes to my family for always being there for me

Table of Contents

Acknowledgement i

Table of Contents ……… ii

Summary v

List of Tables vii

List of Figures ix

List of Symbols xvii

Abbreviations xxi

Chapter 1 Introduction 1.1 Background 1

1.2 Issues Related to the Use of Cement-treated Soil Columns 2

1.3 Objective and Scope of Study 4

1.4 Structure of Thesis 5

Chapter 2 Literature Review 2.1 Introduction 9

2.2 General Aspects of Cement-treated Soil 9

2.2.1 Physical Properties of Cement-Treated Soil 11

2.2.2 Mechanical Properties of Cement-Treated Soil 13

2.3 Existing design approaches 18

2.4 Numerical Modeling on Behavior of Cement-treated Soil 23

2.4.1 2-D and 3-D Finite Element Analysis 24

2.4.2 Constitutive Models for Cement-treated Soil 26

2.5 Summary 29

Chapter 3 Fracture Behavior of Cement-treated Singapore Marine Clay 3.1 Introduction 46

3.2 Properties of Base Materials 47

3.2.1 Untreated Marine Clay 47

3.2.2 Ordinary Portland Cement 48

3.3 Sample Preparation Procedure 48

3.4 Testing Procedure and Apparatus 50

3.4.1 Uniaxial Compression Test 51

3.4.2 Split Tension Test 51

3.4.3 Three-point Bending Notched Beam Test 52

3.5 Compressive Fracture Behavior 53

3.6 Tensile Fracture Behavior 55

3.6.1 Split Tensile Strength 56

3.6.2 Fracture Energy, Gf 58

3.6.3 Tension-softening Relationship 62

3.6.4 Parametric Studies 67

3.6.5 Outstanding Issues 70

3.7 Summary 74

Chapter 4 Constitutive Model for Cement-treated Soil 4.1 Introduction 94

4.2 Finite Element Method 94

4.3 Constitutive Models for Cement-treated Soils 96

4.3.1 Elastic Perfectly-Plastic Tresca Model 96

4.3.2 Isotropic Model 99

4.3.3 Concrete Damage Plastic Model 100

4.4 Evaluation of Constitutive Model Prediction for Behavior of Cement-treated Toyoura Sand 105

4.4.1 Drained Triaxial Compression Test by Namikawa (2006) 106

4.4.2 Direct Tension Test by Koseki et al (2005) 109

4.4.3 Three-point Bending Notched Beam Test by Namikawa (2006) 112

4.5 Evaluation of Constitutive Model Prediction for Behavior of Cement-treated Singapore Marine Clay 116

4.5.1 Uniaxial Compression Test (UCT) 117

4.5.2 Three-point Bending Notched Beam Test 119

4.6 Conclusions 121

Chapter 5 Numerical Approaches in Simulating Cemented Soil Mass 5.1 Introduction 138

5.2 Weighted Average Simulation (WAS) and Real Allocation Simulation (RAS) Approaches 139

5.3 Evaluation of WAS and RAS Approaches in Simulating Cement-treated Soil Mass ……….141

5.3.1 Test description 141

5.3.2 Numerical Analyses 142

5.3.3 Results and Discussion 147

5.4 Study of Cement-treated Ground Improvement Pattern 151

5.4.1 Hypothetical Cases 153

5.4.2 Numerical Analyses 155

5.4.3 Results and Discussions 161

5.5 Conclusion 167

Chapter 6 Field Studies 6.1 Introduction 192

6.2 Field Case Study 1: Lateral Load Test by Babasaki et al (1997) 193

6.2.1 Test Description 193

6.2.2 Numerical Analyses 195

6.2.3 Results and Discussions 198

6.3 Field Case Study 2: Waterway Construction in Northeastern Singapore 201

6.3.1 Characteristics of Site 201

6.3.2 Construction Method 202

6.3.3 Numerical Analyses 204

6.3.4 Results and Discussions 208

6.4 Field Case Study 3: Basement Construction in Central Singapore 210

6.4.1 Characteristics of Site 211

6.4.2 Construction Method 212

6.4.3 Numerical Analyses 213

6.4.4 Results and Discussions 215

6.5 Conclusion 217

Chapter 7 Conclusions 7.1 Summary of Findings 238

7.2 Recommendations for Further Study 242

References ……… …….243

Summary

Owing to presence of soft soil which covers at least one quarter of land area of Singapore, it is often necessary to improve the soft soil for various construction purposes particularly for excavation works One such ground improvement technique is to improve the soft soil with cement to increase the in-situ strength and stiffness The treatment may be conducted at great depth as embedded struts to support deep excavation or at shallower depth not far below ground level to support an open excavation work However, relatively little studies had been performed to study the behavior of cement-treated wall in an open excavation This research covers experimental studies to investigate behavior of cement-treated marine clay and numerical studies to examine the behavior of cement-treated soil columns used as a retaining system in an open cut excavation

The first objective of this study is to understand the material properties of cement-treated Singapore marine clay in terms of compression and tension behavior A series of samples with different mix proportions and curing periods was tested by different means in the laboratory The experimental results show that the material strength increases rapidly to a peak value and then decreases abruptly to a small value upon further straining The tensile strength of this material is found to be 11% of its unconfined compressive strength in the cement content tested in the present study This material becomes brittle when 20% of cement content is added and cured for 14 days The post-peak softening of the treated clay was derived based on fracture mechanics concept Numerical calibration analyses were carried out

to evaluate the appropriateness of three available constitutive models based on laboratory test results and published data The calibration results show that the concrete damage plasticity (CDP) model is superior to Tresca and isotropic models in simulating the behaviors of cement-treated soil in compression, tension and bending

This thesis further examines the significance of modeling the cement-treated soil column configurations in finite element analysis This is of importance to account for the localized overlap areas of treated soil columns and the interaction between treated and untreated soil In practice, cement-treated soil mass with certain columnar shaped treated soil are often analyzed using weighted average simulation (WAS) approach with properties that are averaged over the treated area In this thesis, the limitations of generalizing the properties based on comparison with the results of a published laboratory test were discussed The analysis results show that the shortcomings of this approach can be overcome by the real allocation simulation (RAS) approach with CDP model

The established numerical approach was then adopted to study hypothetical cases of a vertical cut with three different ground improvement geometrical patterns The study demonstrates that tensile damage in the cement-treated soil columns is the trigger for failure The ability of this recommended numerical simulation method – RAS with CDP model is examined by back-analyzing three field case studies on cement-treated soil columns failure The numerical approach provides a fair prediction of the ground response and failure pattern compared to field observations

List of Tables

Table 2.1 Summary of selected E-qu relationships for cement-treated clay

Table 2.2 Summary of some t – qu relationships of cemented soil

Table 3.1 Basic properties of Singapore marine clay from Marine Bouvelard site

Table 3.2 Chemical composition and physical properties of Ordinary Portland Cement

provided by supplier

Table 3.3 Mixture proportions and tests conducted for cement-treated Singapore marine

clay in the present study

Table 3.4 Summary of fracture energies, Gf from BNT tests

Table 3.5 Nominal size for materials used in the present study

Table 3.6 Mixture proportions and experimental tests with addition of filter sand in clay Table 3.7 Chemical compositions of filter sand provided by supplier

Table 3.8 Summary fracture energy Gf for cemented clay and cemented clay+sand

Table 4.1 Mixing proportions of cement-treated Toyoura sand

Table 4.2 Design parameters for Tresca model

Table 4.3 Design parameters for Isotropic model

Table 4.4 Design parameters for CDP model

Table 4.5 Summary results of three-point bending notched beam test by Namikawa (2006) Table 4.6 Tensile design parameters for CDP model in Case T2

Table 4.7 Mixing proportions of cement-treated Singapore marine clay

Table 4.8 Calibration parameters for cemented Singapore marine clay using CDP model Table 4.9 Damage variables used in CDP model for cemented Singapore marine clay Test

B1

Table 5.1 Tresca design parameters for WAS approach in Test 2

Table 5.2 CDP design parameters of lime-cement-treated soil columns for RAS approach

in Test 2

Table 5.3 CDP damage variables for lime-cement-treated soil columns in Test 2

Table 5.4 Summary table of hypothetical cases

Table 5.5 Design parameters for Mohr Coulomb and Tresca models in hypothetical cases Table 5.6 CDP model design parameters for cement-treated soil columns used in

hypothetical cases

Table 5.7 CDP model damage variables for cement-treated soil columns used in

hypothetical cases

Table 6.1 Model parameters for steel plate, concrete cap and untreated soil (reproduced

from Namikawa et al., 2008)

Table 6.2 Summary of case analyses based on the in-situ strength of cemented soil column Table 6.3 Model parameters for CDP model of cemented column

Table 6.4 Summary table of analysis cases for case study 2

Table 6.5 Design parameters for Mohr Coulomb and Tresca models used in case study 2 Table 6.6 Design parameters for case study 3

List of Figures

Figure 1.1 Cement-treated soil columns used in Lexington, Virgina, USA to facilitate an

open excavation (after Ruffing et al (2012)

Figure 1.2 Failure case of adopting cement-treated soil columns as a retaining system for

an open excavation in Northeastern Singapore

Figure 1.3 Examples of soil improvement pattern

Figure 2.1 Change of water content by in-situ cement treatment in Tokyo Port (after

Figure 2.4 Effect of cement content on the permeability of cement-treated clay (after

Kawasaki et al., 1981a)

Figure 2.5 Comparison of stress-strain curve obtained by the conventional method

(external strain) and local transducer approach (local strain)

Figure 2.6 Variation of initial modulus at small strain with confining pressure (after

Shibuya et al., 1992)

Figure 2.7 Effect of cement content on stress-strain behavior of treated clay by unconfined

compression test (after Kamruzzaman, 2002)

Figure 2.8 Comparison of stress-strain curves for (a) with and without curing stress; (b)

CIU and UCT test (after Chin, 2006)

Figure 2.9 Peak and residual strengths of cemented soil under different confining

pressures (after Cement Deep Mixing Association of Japan, 1994)

Figure 2.10 Relationship between DST and UCT results (after Saitoh et al., 1980)

Figure 2.11 Effect of cement content and curing time on effective shear strength parameters

of treated clay (after Kamruzzaman, 2002)

Figure 2.12 Effective stress path of triaxial test for treated soil (after Kamruzzaman, 2002) Figure 2.13 Stress-strain curves of direct tension test Left: light cemented sand by Das and

Dass, 1995; Right: cemented Toyoura sand by Koseki et al (2005)

Figure 2.14 Tension-softening relation for cemented Toyoura sand (after Namikawa 2006)

Left: BNT setup; Right: Tensile stress-opening crack displacement relation Figure 2.15 Arrangement of typical improvement pattern for columnar treated soil (after

CDIT, 2002)

Figure 2.16 Schematic for computing the area replacement ratio, (after CDIT, 2002)

Figure 2.17 Mobilized shear strength of treated column and soil (after CDIT, 2002) Figure 2.18 Failure modes of single columns suggested by Kivelo (1998)

Figure 2.19 Failure modes of treated soil columns proposed by Kitazume (2008)

Figure 2.20 Estimated tilting pattern of treated soil columns in collapse failure mode (after

Kitazume, 2008)

Figure 2.21 Pressure distribution diagram used for moment equilibrium check in collapse

failure mode (after Kitazume, 2008)

Figure 2.22 Induced tensile stress in the treated column (after Kitazume, 2008)

Figure 2.23 Pressure distribution diagram used for moment equilibrium check in bending

failure mode (after Kitazume, 2008)

Figure 2.24 Hypothetical excavation model to compare RAS and WAS by Ou and Wu

Figure 2.27 Numerical calibrated model for axial stress-strain results of uniaxial

compression tests on lime-cement columns using CDP model (after Larsson et al., 2012)

Figure 3.1 Soil-cement and water-cement ratios from previous studies (reproduced from

Lee et al., 2005) and the present study

Figure 3.2 Working range of cement-clay mixes from previous studies (reproduced from

Lee et al., 2005) and mix proportion for the present study

Figure 3.3 Left: Experimental test setup for uniaxial compression test; Right:

Experimental test setup for split tension test

Figure 3.4 Experimental test setup for three-point bending notched beam test

Figure 3.5 The UCT stress-strain results of cement-treated Singapore marine clay for 5

samples with Aw = 30% at 28 days curing period

Figure 3.6 The UCT stress-strain curves of cement-treated Singapore marine clay at 14

days curing period for various cement contents

Figure 3.7 The UCT stress-strain curves of cement-treated Singapore marine clay at 28

days curing period for various cement contents

Figure 3.8 The UCT stress-strain curves of cement-treated Singapore marine clay with

Aw=20% at three curing periods

Figure 3.9 Failure modes of specimen after uniaxial compression test

Figure 3.10 Unconfined compressive strength of cement-treated Singapore marine clay for

various cement contents and curing periods

Figure 3.11 Variation of split tensile strength versus cement content for three curing

periods for cement-treated Singapore marine clay

Figure 3.12 Typical tested specimens of cement-treated Singapore marine clay from split

tension test Left: Front view of testing specimen when failure occurred; Right: Specimen split into two halves

Figure 3.13 Relationship between unconfined compressive strength and split tensile

strength for the present cement-treated Singapore marine clay

Figure 3.14 A stress-strain example from direct tension test for concrete

Figure 3.15 a) Deformation properties of the material outside the fracture zone: - ; b)

Deformation properties of the fracture zone: - c

Figure 3.16 Schematic load-deflection curve from a BNT and the corresponding complete

curve when the specimen weight is taken into account

Figure 3.17 Specimen dimension used in three-point bending notched beam test

Figure 3.18 BNT test results for Aw = 30% at 14 days curing period: (Left) Load-deflection

curves and (Right) Load-crack mouth opening displacement curves

Figure 3.19 BNT load-deflection curves of cement-treated Singapore marine clay at 14

days curing period

Figure 3.20 BNT load-crack mouth opening displacement curves of cement-treated

Singapore marine clay at 14 days curing period

Figure 3.21 BNT load-deflection relationship of cement-treated Singapore marine clay at

28 days curing period

Figure 3.22 BNT load-crack mouth opening displacement relationship of cement-treated

Singapore marine clay at 28 days curing period

Figure 3.23 Fracture zone and the hypothesis stress distribution (modified from Petersson,

1981) in front of the notch tip

Figure 3.24 J-integral contours around crack tip (left) and a typical tension-softening curve

(right) (after Li and Ward, 1989)

Figure 3.25 Li’s J-integral method (after Rokugo et al., 1989)

Figure 3.26 F - v and c0 - v curves for modified J-integral method by Rokugo et al

(1989)

Figure 3.27 Fictitious crack width distribution in the modified J-integral method by Uchida

et al (1991)

Figure 3.28 Tensile stress-crack mouth opening displacement relationship of

cement-treated Singapore marine clay for Test A1 (Aw = 25%)

Figure 3.29 Tensile stress-crack opening displacement relationship of cement-treated

Singapore marine clay for Tests A1 and B1

Figure 3.30 BNT results for specimen notch widths 0.7 mm and 2.5 mm Left:

Load-deflection curves; Right: Load-crack mouth opening displacement curves Figure 3.31 BNT results for specimen size db = 50 mm and db = 40 mm Left: Load-

deflection curves; Right: Load-crack mouth opening displacement curves Figure 3.32 Stability criterion in a load-deflection test corresponding to the stiffness of

testing machine (after Hillerborg, 1989)

Figure 3.33 UCT stress-strain curves of cement-treated Singapore marine clay with and

without sand (at 28 days curing periods unless otherwise stated)

Figure 3.34 BNT load-deflection curves of cement-treated Singapore marine clay with and

without sand (at 28 days curing periods unless otherwise stated)

Figure 3.35 BNT load-crack mouth opening displacement curves of cement-treated

Singapore marine clay with and without sand (at 28 days curing periods unless otherwise stated)

Figure 4.1 Typical monotonic stress-strain behavior assumed in Tresca model

Figure 4.2 Tresca failure criterion in a 3-D stress space

Figure 4.3 Isotropic model: evolution of the yield surface in 2-D (left) and 3-D (right)

principal stress space

Figure 4.4 Two ways of modeling crack in finite element analysis (after Pankaj, 1990) Figure 4.5 Yield surfaces for CDP model in the deviatoric plane, corresponding to

different values of (modified from Abaqus 6.11, 2011)

Figure 4.6 Yield surface for CDP model in plane stress (after Abaqus 6.11, 2011)

Figure 4.7 Definition of cracking strain ̃ used in tension data (after Abaqus 6.11, 2011) Figure 4.8 Definition of compressive inelastic (or crushing) strain ̃ used in compression

data (after Abaqus 6.11, 2011)

Figure 4.9 Drained triaxial compression test for cemented Toyoura sand by Namikawa

(2006)

Figure 4.10 Loading and boundary conditions simulated in finite element models

Figure 4.11 Calibrated stress-strain curves for three constitutive models and laboratory

measurement for drained triaxial compression test of cemented Toyoura sand Figure 4.12 Direct tension test for cemented Toyoura sand by Koseki et al (2005)

Figure 4.13 Classical and tension truncated for Tresca criteria (after Antão et al., 2007) Figure 4.14 Projection of Rankine surface in deviatoric plane

Figure 4.15 Post-failure stress-fracture energy curve simulated by CDP model (after

Abaqus 6.11, 2011)

Figure 4.16 Calibrated tensile stress-strain curves for three constitutive models and

laboratory measurement for direct tension test of cement-treated Toyoura sand

Figure 4.17 Three-point bending notched beam test for cement-treated Toyoura sand by

Namikawa (2006)

Figure 4.18 Load-deflection curves of three-point bending notched beam test results for

cement-treated Toyoura sand by Namikawa (2006)

Figure 4.19 Tension-softening relation for cement-treated Toyoura sand by Namikawa

(2006)

Figure 4.20 Finite element meshes for simulating the three-point bending notched beam test:

(a) coarser mesh with 5,353 elements; and (b) finer mesh with 9,866 elements Figure 4.21 Load-deflection curves for two different mesh sizes in simulating the three-

point bending notched beam test for cement-treated Toyoura sand

Figure 4.22 FEM input tension-softening relation in Case T2 for cement-treated Toyoura

sand

Figure 4.23 Calibrated load-deflection curves and laboratory measurement for three-point

bending notched beam test of cement-treated Toyoura sand

Figure 4.24 Tensile damage process happened around the notch tip observed in FE analysis

Case T2

Figure 4.25 Schematic experiment setup and stress-strain curves for UCT of cement-treated

Singapore marine clay

Figure 4.26 FEM model and stress-strain curves for UCT of cement-treated Singapore

marine clay

Figure 4.27 Schematic diagram of three-point bending notched beam test B1 of cemented

Singapore marine clay (unit: mm)

Figure 4.28 Load-deflection curves of three-point bending notched beam test B1 for

cement-treated Singapore marine clay

Figure 4.29 Finite element mesh for simulating the three-point bending notched beam Test

B1

Figure 4.30 Calibrated load-deflection curves and laboratory measurement for three-point

bending notched beam Test B1 for cement-treated Singapore marine clay Figure 4.31 Post-test damage observed in numerical (Top) and laboratory (Bottom) for

three-point bending notched beam Test B1

Figure 5.1 Schematic of lime-cement columns in shear box test (after Larsson, 1999) Figure 5.2 (a) Mobilized shear stresses-lateral deformations for Tests 1 and 2; and (b)

damage pattern of the lime-cement-treated soil columns observed in Test 2 (after Larsson et al., 2012)

Figure 5.3 UCT results for lime-cement-treated columns in Test 2 (after Larsson et al.,

2012)

Figure 5.4 Finite element model for Test 1 Left: Finite element mesh; Right: Boundary

and loading conditions

Figure 5.5 Finite element meshes for Refined Mass (RM) model in WAS approach Figure 5.6 Finite element meshes for RAS approach

Figure 5.7 Design parameters based on UCT results for lime-cement-treated columns in

Test 2

Figure 5.8 Two configurations of treated mass for WAS approach in Test 2

Figure 5.9 Calibrated analysis result of mobilized shear stress and lateral deformation for

Test 1 without treated soil columns

Figure 5.10a Analysis result of mobilized shear stress and lateral deformation for Test 2 with

WAS approach

Figure 5.10b Yielding zone appeared in the models for Test 2 with WAS approach

Figure 5.11 Analysis result of mobilized shear stress and lateral deformation for Test 2 with

WAS and RAS approaches

Figure 5.12 Schematic failure in treated soil columns for RAS approaches using Tresca

model

Figure 5.13 Schematic failure in treated soil columns for RAS approaches using CDP

model

Figure 5.14 Plan view of improvement patterns: a) Grid type (GD); b) Tangential buttress

type (TN); and Double-wall type (DW)

Figure 5.15 Two configurations of WAS approach for grid type ground improvement

pattern

Figure 5.16 Typical boundary condition and geometry for hypothetical case

Figure 5.17 Finite element mesh for vertical cut without improvement

Figure 5.18 Finite element mesh for Whole Mass (WM) model in WAS approach

Figure 5.19 Finite element mesh for Refined Mass (RM) model in WAS approach

Figure 5.20 Finite element mesh for grid type (GD) improvement pattern, L1

Figure 5.21 Finite element mesh for tangential buttress type (TN) improvement pattern, L2 Figure 5.22 Finite element mesh for double-wall type (DW) improvement pattern, L3 Figure 5.23 Progressive ground movement, U (in m) for vertical cut without soil

improvement (Case Li)

Figure 5.24 Progressive development of plastic strain (PE) for vertical cut without soil

improvement (Case Li)

Figure 5.25 Ground response for vertical cut with grid type soil improvement pattern with

WAS approach (Left: Case La; Right: Case Lb)

Figure 5.26 Plastic strain (PE) for vertical cut with grid type soil improvement pattern with

WAS approach (Left: Case La; Right: Case Lb)

Figure 5.27 Ground response for GD type with RAS approach: Case L1-1

Figure 5.28 Ground response for GD type with RAS approach: Case L1-2

Figure 5.29 A close-up plastic strain developed in soil and column for Case L1-2

Figure 5.30 Ground response for GD type with RAS approach: Case L1-3

Figure 5.31 A close-up plastic strain developed in soil and damage occurred in cemented

column for Case L1-3

Figure 5.32 Progressive tensile crack development in cemented column for Case L1-3

Figure 5.33 Ground response for TN type with RAS approach: Case L2-1

Figure 5.34 Ground response for TN type with RAS approach: Case L2-2

Figure 5.35 A close-up plastic strain developed in soil and column for Case L2-2

Figure 5.36 Ground response for TN type with RAS approach: Case L2-3

Figure 5.37 A close-up plastic strain developed in soil and damage occurred in cemented

column for Case L2-3

Figure 5.38 Progressive tensile crack development in cemented column for Case L2-3

Figure 5.39 Ground response for DW type with RAS approach: Case L3-1

Figure 5.40 Ground response for DW type with RAS approach: Case L3-2

Figure 5.41 A close-up plastic strain developed in soil and column for Case L3-2

Figure 5.42 Ground response for DW type with RAS approach: Case L3-3

Figure 5.43 A close-up plastic strain developed in soil and damage occurred in cemented

column for Case L3-3

Figure 5.44 Progressive tensile crack development in cemented column for Case L3-3

Figure 6.1 (a) Soil profile and (b) schematic of test setup (after Namikawa et al., 2008) for

case study 1

Figure 6.2 Field measurement data (a) Load-displacement at applied point (after

Namikawa et al., 2008); (b) Post-test damage at columns (after Babasaki et al., 1997)

Figure 6.3 Laboratory test results for cemented soil Left: UCT results; Right: Split tensile

strength results (reproduced from Namikawa et al., 2008)

Figure 6.4 Finite element model for lateral load test of cemented soil column in case study

1

Figure 6.5 The range of qu and st used in finite element analyses for case study 1

Figure 6.6 FE model with boundary and loading conditions for case study 1

Figure 6.7 Processed field data for lateral load-displacement at top of the cemented soil

column in field

Figure 6.8 Comparison of lateral load-displacement relations for field test and analysis

results

Figure 6.9 Lateral load-displacement relations for Case 2 and Case 3 with tensile strength

varies from 0.1qu to 0.2qu

Figure 6.10 Damage at cemented column observed in numerical analysis Case 3b: (a)

compressive damage; (b) tensile damage

Figure 6.11 Lateral load-displacement relations modeled by classic Tresca and CDP models Figure 6.12 Yielding zone at cemented column observed in classical Tresca model analysis Figure 6.13 Soil profile sketch for case study 2

Figure 6.14 Proposed construction supported by diaphragm wall and cement-treated soil for

case study 2

Figure 6.15 On-site excavation profile for case study 2

Figure 6.16 Collapse of front row cement-treated soil columns for waterway construction in

northeastern Singapore

Figure 6.17 Typical geometry and boundary condition for case study 2

Figure 6.18 Finite element mesh for Case C-W (case study 2)

Figure 6.19 Finite element mesh for Case C-R (case study 2)

Figure 6.20 Ground response for Case C-W Left: Ground movement (m); Right: Plastic

Figure 6.24 Soil profile and proposed construction method for case study 3

Figure 6.25 On-site excavation profile and visible cracks at the cement-treated soil columns

appeared immediate after the excavation for case study 3

Figure 6.26 Localized collapse of cement-treated soil columns for basement construction in

central Singapore (case study 3)

Figure 6.27 Geometry and boundary condition for case study 3

Figure 6.28 Finite element mesh for case study 3

Figure 6.29 Numerical analysis results for case study 3

Figure 6.30 Numerical comparison of crack development in cement-treated soil columns

with steel I-beam (left) and without steel I-beam (right)

Figure 6.31 Comparison of field failure observation and predicted damage by numerical

analysis for case study 3

List of Symbols

Sectional area of treated column

Alig Area of the ligament

Al2O3 Aluminium oxide

Crack length

Ligament depth

Area of replacement ratio

Difference of notch lengths

CaO Calcium oxide

Cu Undrained shear strength

Elasticity matrix

Degraded elasticity matrix

d Scalar stiffness degradation variable

d c Compression damage variable

d t Tension damage variable

E 0 Initial elastic Young’s modulus

E 50 Secant modulus of elasticity

Fe2O3 Iron oxide

Yield function

Gf Tensile facture energy

Gfs Shear fracture energy

Gs Specific gravity

g Gravitational acceleration

Plastic potential function

First principal stress invariant

J-integral / Potential energy

Rate of energy absorption in the cohesive zone

Second stress invariant

Third stress invariant

K2O Potassium oxide

Ratio of on the TM to that on the CM at initial yield

k Stiffness of the machine in BNT

LL Liquid limit

MnO Manganese oxide

Equivalent parameter index

Effective mean stress

q Von Mises equivalent stress

Effective von Mises equivalent stress

δc0 Crack mouth opening displacement

δc Crack opening displacement

δcr Maximum crack opening displacement

̃ Equivalent plastic strain

̃ Compression crushing strain

̃ Tensile cracking strain

b Total unit weight

Strain mobilization factor

Lode angle

Bulk density

Stress

Compressive strength in the biaxial state

Compressive strength in the uniaxial state

Normal stress

st Split tensile strength

dt Direct tensile strength

t Tensile strength

Uniaxial tensile stress at failure

Effective stress

Abbreviations

BNT Three-point Bending Notched Beam Test

CASH Calcium aluminium silicate hydrate

CAX4R 4-node Bilinear Axisymmetric Quadrilateral Element

CDIT Coastal Development of Institute Technology, Japan

CDP Concrete Damage Plasticity

CID Isotropically Consolidated Undrained Triaxial Compression Test

CIU Isotropically Consolidated Drained Triaxial Compression Test

CM Compressive Meridian

CPS3 Linear Triangular Element

CPS4R Linear Quadrilateral Element

CSH Calcium silicate hydrate

C3D8R 8-node Linear Brick Element, with Reduced Integration

DW Double-wall Type

FEM Finite Element Method

HITEC US High Innovative Technology Evaluation Center

LDT Local Displacement Transducer

LOI Loss on Ignition

OPC Ordinary Portland Cement

RAS Real Allocation Simulation

RILEM French acronym for the International Union of Laboratories and Experts in

Construction Materials Systems and Structures

RM Refined Mass Model

STT Split Tension Test

TM Tensile Meridian

TN Tangential Buttress Type

UCT Uniaxial Compression Test

VERT Vertical Earth Reinforced Technology

WAS Weighted Average Simulation

Chapter 1 Introduction

1.1 Background

To effectively utilize very limited land in urban area, underground space is commonly exploited for development Various construction methods are employed to facilitate underground construction in different ground conditions For ground where bedrock is encountered at shallow depth, installation of conventional temporary retaining wall such as sheet pile or soldier pile requires pre-boring process to anchor the wall into bedrock Pre-boring work increases time and cost; moreover, the fixity at the bedrock level might induce large bending moment which compromises the capacity of sheet pile or soldier pile wall Construction cost increases tremendously if higher capacity wall such as contiguous bored pile or diaphragm wall is adopted to facilitate this shallow depth of excavation

One quarter of Singapore is covered by sedimentary deposit known locally as the Kallang Formation (Tan et al., 2002; Pitts, 1992), so the development in this formation is inevitable This formation consists of deposits of marine, alluvial, littoral and estuarine origins (Lee et al., 2005) Marine clay is the main constituent of this formation and the thickness is usually between 10 m to 15 m but in some instance, it can be more than 40 m (Tan et al., 2003) Deep excavation in this ground condition is always facilitated with robust system such as contiguous bored pile wall, steel tubular pipe wall or diaphragm wall with a layer of embedded treated soil together with strutting support (Gaba, 1990; Hsi and Yu, 2005; Shirlaw

et al., 2005) However, this system is too expensive for a small scale development involving shallow depth of excavation In such a thick soft soil, ensuring the toe stability of conventional retaining wall and controlling maximum wall deflection at the base of excavation (Tanaka, 1994) are always a challenge As a solution, the designers tend to form

an embedded treated soil layer just beneath the excavation level to increase the stability as well as provide lateral support to the wall Nevertheless, for large area development with

shallow excavation depth, construction of this embedded treated soil becomes costly and less practical

In civil engineering, it is always the particular challenging problems that prompt the adoption

of innovative solutions to optimize the construction efficiency To overcome aforementioned ground conditions, column-shaped cement-treated soil offers viable alternative to conventional earth retaining system The column-shaped cemented soil can be formed by deep cement mixing method or jet grouting method It can serve as a rigid gravity structure when constructed in continuous overlapping column Sometimes, different layouts of column installations are used to achieve the desired effect by utilizing space optimization purpose

There are some documented successful cases and limited published failure cases One of the successful cases was located in downtown Lexington, Virgina, USA (Ruffing et al., 2012) where the site is underlain by limestone and calcareous shale bedrock similar to the subsurface condition discussed earlier Cement-treated soil columns of diameter 2.4 m and length 8.5 m were constructed to facilitate the 4.9 m deep open excavation The cross section

of the retaining wall and site photographs are presented in Figure 1.1 On the other hand, Haque and Bryant (2011) observed substantial soil body movement behind a cement-treated soil retaining wall in Irving-Las Colinas in Texas, USA Failure of cement-treated soil wall did happen in the field but such cases are rarely published Figure 1.2 shows failure happened

in Northeastern Singapore where cement-treated soil columns were adopted as retaining system to facilitate an open excavation The front row of cement-treated soil columns collapsed when excavation reached 7 m depth

1.2 Issues Related to the Use of Cement-treated Soil Columns

To optimize construction cost and efficiency, different layouts of column configurations (refer to samples shown in Figure 1.3) instead of 100% ground improvement are adopted in

design Owing to the short history of employing cement-treated soil columns as earth retaining system, its basic design has yet to be firmly established In current design practice, engineers often regard it as a gravity type structure As such, two stability analyses involving external and internal wall stability are evaluated For external stability analysis, three failure modes of the treated soil mass: sliding, overturning, and bearing capacity are examined For group columns, it is treated as composite material assuming no failure occurred within the treated soil columns and the untreated soil However, averaging the properties for these two materials with very different behavior indeed violates the underlying mechanism In general,

a volume ratio to average the properties is adopted and a simple elastic-perfectly plastic constitutive model is employed to simulate the averaged behavior This can result in errors and put the design in risk as the treated soil behaves as a quasi-brittle material (Kamruzzaman, 2002; Das and Dass, 1995) while the behavior of untreated soil is usually ductile

In addition, the behavior of group columns in the field is unlikely to be that of a composite material assumed in the design, as the bonding between cemented soil columns and between cemented soil column (tensile strength) and soil (cohesive strength) might be weaker than the original ground Weak bonding may cause separation of columns especially the front columns along the intended excavation line from the rest of the row resulting in toppling of these columns, making the composite mass assumption invalid The tendency of bending failure also reveals the limitation on current internal stability assessment where focus has been drawn

on shear failure of the columns and overlooks the tensile forces in the columns It is believed that the disregard of the above considerations in the design results in a good number of failures of the cement-treated soil column retaining system in open excavations

In numerical analyses, the cement-treated soil behavior is very often simulated by a simplistic constitutive model Up to date, elastic-perfectly plastic model is used due to simplicity and parameters that can be readily obtained However, the nature of this treated soil columns is more complex where the strength reaches its peak at a very small strain followed by a sudden

reduction in post-peak stress to a very low residual value Hence, adopting simplistic model may not truly simulate the treated soil behavior Nevertheless, rather than using the most advanced model, Lee (2008) highlighted that it is more important to have a better understanding of the soil behavior in a specified problem This allows one to choose models and parameters which adequately reflect the important behavioral aspects of the soil for a particular problem In spite of this, very few studies were carried out to evaluate the effect of constitutive model in simulating the cement-treated soil columns behavior for open excavation problems

1.3 Objective and Scope of Study

Employing cement-treated soil columns as an alternative for conventional earth retaining system to facilitate a shallow depth excavation can be more effective in terms of time and cost (Ruffing et al., 2012) However, failure of such retaining system may occur if the behavior of this material is not well understood As a result, this study aims to investigate the mechanism

of cement-treated soil columns used as earth retaining system in an open excavation The scope of study is as follows:

a) To carry out laboratory studies to investigate the compressive and tensile behaviors of cement-treated marine clay including interpretation of post-peak softening of the treated clay

b) To evaluate the effect of different constitutive models (namely elastic-perfectly plastic Tresca model, isotropic model and concrete damage plasticity model) in simulating the behavior of cement-treated soil The results are calibrated against those data obtained from different types of laboratory tests conducted on cement-treated sand and cement-treated clay

c) To simulate the behavior of cement-treated soil columns in open excavation using numerical approaches These include evaluating the effectiveness of the commonly

used weighted average simulation approach and modeling the treated mass based on the actual column layout

d) To compare the field observations from case studies with the proposed numerical simulations These field studies would also serve to complement the understanding developed in items (a), (b) and (c)

e) Finally, issues on cement-treated soil columns in an open excavation that are important but overlooked in design practice would be addressed

1.4 Structure of Thesis

The outline of this thesis after this introductory chapter is presented as follows:

a) Chapter 2 reviews the general aspects of cement-treated soil where the changes in physical and mechanical properties are discussed The current design procedures and methods for cement-treated soil columns used in an open excavation are also evaluated Existing numerical studies of cement-treated soil are reviewed

b) Chapter 3 presents the experiment work to establish the understanding of fracture behavior of cement-treated marine clay in compression and tension At first, the experiment set-up and methodology are presented followed by results and analyses Tensile fracture energy and tension-softening relation of cement-treated marine clay are interpreted from the experimental results Parametric studies and outstanding issues regarding the laboratory studies are discussed

c) Chapter 4 firstly describes the constitutive models (namely elastic-perfectly plastic Tresca, isotropic and concrete damage plasticity models) followed by numerical calibrations against published data and laboratory results presented in Chapter 3 Three types of laboratory tests are involved to evaluate the accuracy of model prediction for cement-treated soil when subject to compression, tension and bending Calibrations are made for cement-treated Toyoura sand and cement-treated marine clay

d) Chapter 5 examines the effectiveness of commonly used weighted average simulation approach in simulating the treated mass as a composite material Back-analysis on a laboratory test using this approach is compared against laboratory measurements Limitations of this common practice are addressed and overcome with the real allocation simulation approach incorporating the findings from Chapters 3 and 4 Thereafter, hypothetical cases of a vertical cut with different ground improvement geometrical patterns are conducted to further examine the behavior of cement-treated soil mass Through this study, the failure mechanisms for cement-treated soil columns

in different layouts are illustrated

e) Chapter 6 presents the back-analyses of large-scale field case studies Comparisons between the field observations and the predicted cement-treated soil columns behaviors using the proposed numerical approach in the present study are made

f) Chapter 7 concludes the findings of the present study and proposes recommendations for future studies

Chapter 2 Literature Review

2.1 Introduction

With increasing popularity in using cement-treated soil in construction works, extensive studies had been conducted by various researchers to understand its behavior Previous studies on the general aspects of cement-treated soil are first reviewed in this chapter The behavior of cement-treated soil through experimental studies is thoroughly reviewed to establish the current state of art in the behavior of cement-treated soil, including its physical and mechanical properties These two properties decide the mix proportion and the performance of the final product; hence it is crucial to understand them well Existing design approaches for assessing cement-treated soil in open excavation are then discussed CDIT (2002) highlighted the importance of adopting numerical analysis to examine cement-treated soil problems involving their current states of stress-strain As such, previous numerical studies for cement-treated soil are reviewed herein

2.2 General Aspects of Cement-treated Soil

Cement-treated soil is a product of mixing cement to in-situ soil through different mechanical methods to increase the strength and stiffness by bonding the particles together The most popular methods are jet grouting method and deep cement mixing method The difference between these two methods is that the former relies on high pressure to erode and mix the soil with cement or cement slurry through rotation while the latter uses blades to cut and mix the soil with cement The product quality formed by the latter is more uniform and promising Mechanical properties of cement treated soils are affected by many factors (Babasaki et al., 1996) The in-situ strength of treated soils varies widely depending on the level of mixing as

in the field, it is difficult to mix cement based agent well with soil The mixing performance can be affected by the efficiency of the equipment used, mixing process and ground condition

Therefore, the inherent heterogeneity of cemented soil has been evaluated in laboratory as well as in the field by some researches such as Larsson et al (2005a); Larsson et al (2005b) and Chen et al (2011)

The increase in strength of cement-treated soil is mainly attributed to the three reactions happening in the process of mixing cement with soil, namely: hydration process, ion exchange (flocculation), and pozzolanic reaction (Diamond and Kinter, 1965; Assarson et al., 1974) In the initial stage of mixing, water in the mixture is consumed by cement and dissociates a product known as calcium hydroxide Dissociation of calcium hydroxide increases the electrolytic concentration as well as the pH of the pore water, and dissolves the SiO2 and Al2O3 (pozzolan) from the clay particles, which then leads to ion exchange, flocculation and pozzolanic reactions Dissociation of calcium ion Ca2+ in the pore water will replace the weaker cations on the surface of clay particles This then leads to decrease in the distance between the diffused double layer and hence the small particles of clay flocculate and coagulate into larger sizes The dissolved dissociated Ca2+ ions react with the dissolved pozzolan from the clay particle’s surface to form hydrated gels, resulting in the combination

of soil particles Hydration reaction dominates the early stage of curing process while the pozzolanic reaction is more significant in prolonged curing period Hence, effective friction angle for the treated soil increases more predominantly at early curing period during the formation of particle interlocking in the soil-cement skeleton (Kamruzzaman, 2002) Increase component of pozzolanic reaction products contributes to the increase of strength parameter c’ Details of engineering properties of cement-treated soils are discussed in the following sections

2.2.1 Physical Properties of Cement-Treated Soil

In addition to the desired improved properties, it is essential to evaluate the overall properties change resulted from the chemical reaction, so that the performance of the cement-treated soil can then be estimated

a) Change in water content

During the mixing process, the cement mineral reacts with water to produce cement hydration products (CDIT, 2002), thus it is expected that some amount of water in the soil will be consumed beyond the amount of water in the mixture slurry itself It was experimentally tested that the consumed water to form the cementitious products will not be expelled by re-heating the products to 105oC (Kamruzzaman, 2002) The water content in the soil will be much more reduced in dry mixing method compared to wet mixing method Figure 2.1 shows the water content after cement treatment decreased about 20% from the original as reported in Tokyo Port clay field It was noted that the mixing water-cement ratio, w/c was 0.6 (Kawasaki et al., 1978) Kamruzzaman (2002)’s study showed that majority of the decrease in water content takes place within the first 7 days of curing (Figure 2.2) It is revealed in some studies (Uddin et al., 1997; Kamruzzaman, 2002) that higher percentage of cement content induces a greater reduction in water content

b) Change in density

There is a concern regarding the effect of cement inclusion into the in-situ soil will induce density change since the water content is decreased For the cement treatment in dry form, the wet density of the treated soil, increases by about 3% to 15% recorded by CDIT (2002) Conversely, the density change due to slurry form treatment is negligible irrespective of water

to cement ratio, w/c Figure 2.3 shows the change in density for in-situ cement treatment in

dry and slurry forms reported by Japan Cement Association (1994) Wen (2005) reported similar density with original Singapore marine clay is obtained for jet grout mixing

c) Change in permeability

Although most of the time, increase in strength and stiffness are the main purposes of mixing cement to in-situ soil, reduce the permeability to form a cut-off wall where problems of groundwater seepage is a concern (Yu et al., 1999) is also one of the purposes The permeability coefficient of the treated soil is highly influenced by the distribution of pore sizes inside the soil mix (Porbaha et al., 2000) Generally, laboratory permeability tests showed that the soil-cement mixture is in the order of 10-10 m/s and is not influenced by the confining pressure and seepage pressure in the specified range (Yu et al., 1999) With the increase of vertical load, the coefficient of permeability increases markedly when the confining pressure is low and changes slightly or even decreases when the confining pressure

is high

The effect of cement content on the permeability of cement-treated clay was studied by Kawasaki et al (1981a) as shown in Figure 2.4 It was found that the permeability reduced with the increment of cement content from 10% to 20% The reduction could be due to the pozzolanic cement substances, which blocked the pores in the soil cement matrix (Broderic and Daniel, 1990)

However, it is known that the permeability of soil in field can be higher than those obtained in laboratory test (Tavenas et al., 1986), thus field test is more important in understanding the actual mechanism In Shen (1998)’s study, finite element method (FEM) back-analysis showed that the predicted consolidation degree with elapsed time agreed well with the measured value if the improved column was assigned with high permeability of 10-5 m/s which was in the order of 4 to horizontal permeability of surrounding clay This was contrary

to the studies as mentioned before and it was claimed that the existence of fracture cracks resulted in higher permeability The expansion pressure (pressure acting on the wall of a deep mixing column results from the injection of chemical admixtures under working pressure) increases with the injected volume of cement slurry According to Shen (1998), with a mixed cement content of 2% to 4%, the expanding pressure will be greater than the minimum hydraulic fracturing pressure increment, which indicates that the surrounding clay is conveniently fractured during deep mixing column installation The existence of the shearing force caused by mixing blades in deep mixing makes the surrounding clay more easily to be fractured

Lorenzo and Bergado (2006) acknowledged that the product of cement-treated is a porous but strong matrix and exhibits higher permeability Similar opinion shared with Suzuki et al (1981) where due to cement hydration, increase in pH of the pore fluid and Ca++ ion on clay surface causes shrinkage of the diffused double layer leading to flocculation and hence resulting an increase in permeability This also coincides with the observation by Kamruzzaman (2002) for 7-day permeability for cement-treated Singapore marine clay in laboratory test The permeability of the cemented clay was obtained using falling head permeability test conducted in a modified oedometer consolidation cell

2.2.2 Mechanical Properties of Cement-Treated Soil

a) Modulus of elasticity, E

The Young’s modulus is defined as the ratio of the uniaxial stress over the uniaxial strain in the range of stress in elastic regime Generally the unconfined compression test (UCT) is preferred owing to its easy handling, inexpensive and quick confirmation of in-situ strength The slope of the stress-strain curve at any point to the origin of the axis is called the secant

modulus, Esecant Tan et al (2002) reported that the Esecant for cement-treated soil for the whole

regime of stress-strain curve is nonlinear and decreases with strain increment The adopted

Young’s modulus is the tangent modulus of the initial, linear portion of stress-strain curve

The initial slope is indicated as E0 and secant modulus at 50% of ultimate strength is denoted

as E50 Table 2.1 summarizes selected E-qu relationships suggested for cement-treated clay

Table 2.1: Summary of selected E-qu relationships for cement-treated clay

lab improved Lee et al., 1998 E 0 = 80 – 200qu # Marine clay Lab improved & In-situ

improved Kamruzzaman, 2002 E 0 = 490qu * Marine clay Lab improved

Tan et al., 2002 E 50 = 350 – 800qu *

E 50 = 150 – 400qu # Marine clay Lab improved Lee et al., 2005 E 0 = 80 – 140qu * Marine clay Lab improved

Wong and Goh, 2006 E = 100qu Marine clay In-situ improved Lorenzo and Bergado,

E 0 = Initial elastic Young’s modulus; E 50 = Secant modulus of elasticity (at 50% of ultimate strength);

q u = unconfined compression strength; * denotes local strain measurement method; # denotes conventional strain measurement method

Nevertheless, there were researchers (Jardine et al., 1984; Goto et al., 1991; Tatsuoka et al., 1996; Kohata et al., 1997; Tan et al., 2002) highlighted the limitation of measuring the axial strain using conventional method In the case of unconfined or triaxial compression test, the axial strain in conventional method is usually obtained from the axial displacement of the loading piston or the specimen cap Goto et al (1991) identified the sources of error involved

in the external axial strain measurement as: (1) system compliance due to the deflection of top cap, load cell, cell, piston etc; (2) tilting of the specimen; (3) bending error on bottom and top

of the specimen; (4) strain non-uniformity of the specimen caused by end restraints, leading to bulging of the specimen; and (5) strain localization of the specimen As shown in Figure 2.5, the conventional method tends to underestimate the stiffness compared to the local transducer approach

On the other hand, there were studies (Consoli et al., 1996; Hosoya et al., 1997; Yin and Lai, 1998) revealed that the stiffness and strength obtained from triaxial compression test depend

on confining pressure Typically, the strength achieved a higher value in triaxial compression test with increase in confining pressure However, these tests were carried out using conventional method of strain measurement According to the studies by Tatsuoka et al

(1997), the change in initial elastic modulus at small strain (Emax) corresponding to the

increase in effective confining pressure is negligible when local axial strain measurement

approach is adopted Figure 2.6 shows that the modulus at small strain (Emax) is not a function

of confining pressure (Shibuya et al., 1992) Study by Chin (2006) in comparing the strain behavior from both CIU and UCT tests using conventional method for specimens cured under loaded-drained condition showed that the initial elastic modulus is very similar for both tests Consoli et al (1996) also found that when tensile strength increases to certain value, the ratio of triaxial deviatoric stress over the unconfined compression stress approaches unity for any given confining pressure

stress-b) Strength

In general, the strength is imparted to a soil by virtue of: i) cohesive forces (termed as ‘c’) between particles; ii) frictional resistances ( ) due to particles sliding against one another, or moving from interlocked positions In the past, the cement-treated soil strength has been measured through unconfined compression test (UCT), consolidated isotropic undrained test (CIU) and consolidated isotropic drained test (CID) On top of these, the tensile strength has been studied by uniaxial direct tension test or most of the time split tension test (STT) is preferred due to easiness

From the UCT results, ductile behavior is manifested for very low cement content Conversely, the treated soil becomes more brittle at higher cement content Figure 2.7 shows that the treated soil deviatoric stress increases abruptly associated with a very small strain

followed by a sudden reduction in post-peak stress to a very low residual value According to Chin (2006)’s result shown in Figure 2.8a, the CIU test showed that the difference in deviatoric stress increases significantly with increase in curing stress and resulting in the material turning brittle The curing stress is the isotropically loaded stress applied on the specimen during curing period Comparisons of stress-strain behavior from both CIU and UCT tests for specimens under loaded-drained condition, the maximum deviatoric stress is identical to each other but the post-peak behavior varies (Figure 2.8b) Although both exhibit post peak softening but with the existence of confining stress, the residual strength is about 70%

to 80% of the peak strength for confining stress beyond 100 kPa Figure 2.9 (Cement Deep Mixing Association of Japan, 1994) shows the residual and peak strengths under different confining pressures: in CIU the residual strength is 30% to 60% lower than the peak strength; while in CID the residual is 60% to 80%

Estimation of shear strength of cement-treated soil from direct shear tests is not common So far the available published data was by Saitoh et al (1980) as shown in Figure 2.10 for low strength, the shear strength can be roughly represented by half of the unconfined compressive strength It is worth noting that in these tests, the normal stress during shearing was zero and

is therefore more appropriate in representing soil-cement at shallow depth

The increase in strength of cemented soil is attributed to hydration and pozzolanic reactions Hydration reaction dominates the early stage of curing while pozzolanic reaction is more significant at prolonged curing period Hence, effective friction angle for the treated soil increases more predominantly at early curing period during the formation of particle interlocking in the soil-cement skeleton Increased component of pozzolanic reaction products contributes to the increase in strength parameter c’ Figure 2.11 shows the effect of cement content and curing time on effective strength parameters of treated clay by Kamruzzaman (2002) Effective stress path of the cement-treated soil in triaxial test (Figure 2.12) revealed

the existence of deviatoric stress (q) beyond the original critical state of untreated soil as evidence of cementation bond takes place during the shearing state

It is well known that addition of cement helps to improve the compressibility and strength of in-situ soil Conversely, it results in a brittle material that is weak in tension As a result, designer often ignores the tensile strength in design; consequently the tensile behavior of cement-treated soil is always neglected in the study related to cemented soil Unixial tension test also known as direct tension test is the direct method of determining the tensile strength but in most cases it was determined by split tension test Very often, the tensile strength is correlated to the key parameter qu as illustrated in Table 2.2 Nevertheless, these empirical correlations should be applied with care, as they differed from soil to soil

Table 2.2: Summary of some t – qu relationships of cemented soil

in strength after peak strength is observed Addition of cement in the untreated soil has changed the material from ductile to brittle manner Once the treated material experiences the

maximum load, the material cannot maintain the maximum load under further straining This post-peak strength reduction in further straining is known as softening behavior To the best knowledge of this author, so far the tension-softening behavior of cement-treated soil was only investigated by Namikawa and Koseki (2006) The study was conducted for Toyoura sand with three-point bending notched beam test (BNT) From the test, the tension-softening relation of the cemented Toyoura sand (Figure 2.14) was interpreted using the energy balance approach

c) Poisson’s ratio, ν

When a material is compressed in one direction, it usually tends to expand in the other two directions perpendicular to the direction of compression The ratio between these two quantities is named as Poisson’s ratio CDIT (2002) suggested a range of 0.25 to 0.45 irrespective of the unconfined compressive strength, qu of the cement-treated soil The measured Poisson’s ratio for cement-treated clay by Fang and Yu (1998) varied from 0.13 to 0.24 A value of 0.167 was evaluated from the triaxial drained compression test on cemented Toyoura sand by Namikawa (2006) Larsson et al (2012) proposed a value of 0.15 in their study

2.3 Existing design approaches

The early application of cement-treated soil columns is mainly used in foundation of structures to increase bearing capacity and reduce settlement On top of these main criteria of improvement, the stability of whole system is indeed affected by the layout of improvement pattern Therefore, the design considerations based on limit equilibrium approach so far have been focused on the stability assessment of the improved ground There are several improvement geometrical patterns: a) block-type, b) wall-type, c) lattice-type, d) group column-type, and e) columns-in-contact type as illustrated in Figure 2.15 Since there is no