BEHAVIOUR OF UNSATURATED SOILS FOR ROAD PAVEMENT

STRUCTURE UNDER CYCLIC LOADING

by

ANEKE FRANK IKECHUKWU

A dissertation submitted to the Faculty of Engineering and information

Technology
Central University of Technology, Free State, South Africa
In fulfilment of the requirement for the degree of
Doctor of Engineering in Civil Engineering

SUPERVISOR: Prof. Mohamed M H Mostafa

CO- SUPERVISOR: Dr. Azza Moubarak

BLOEMFONTEIN
October 2018

© Central University of Technology, Free State

DECLARATION

I, Aneke Frank Ikechukwu hereby declare that the work presented in this thesis titled
‘BEHAVIOUR OF UNSATURATED SOILS FOR ROAD PAVEMENT STRUCTURE
UNDER CYCLIC LOADING’ submitted to Central University of Technology for a
Doctoral degree of Engineering (D. Eng.) in Civil Engineering, is my personal distinctive
research work carried out by me under the supervision of Prof. Mostafa and Dr.
Moubarak. I therefore confirm that no part of this thesis has been presented for any
other degree or examined in any other University.

Signature…

Date: October 2018

Bloemfontein, South Africa

ii

© Central University of Technology, Free State

ABSTRACT
The geotechnical engineering practice has not been advanced in South Africa. As no
significant integration of unsaturated soil mechanics (USM) is yet to be covered in
pavement design. Stiffness of subgrade soils that is determined by resilient modulus
(𝑀𝑟), is an important component in the design of flexible pavement structures and
railway embankments. Environmental effects, such as seasonal variations in moisture
content considerably influence the subgrade properties that are failed to be considered
in any realistic pavement and railway embankment design. The seasonal changes in
moisture condition affects the stress state of subgrade due to changes in the matric
suction and this is considered an important stress state variable in unsaturated soil
mechanics. Therefore, particular attention was directed in this research towards
evaluating the response of unsaturated pavement structure under cyclic loading.

Several factors are needed to be considered during pavement design, which are
dependent on 𝑀𝑟 in order to provide an accurate assessment of the support provide by
the subgrade. Roadbed with high plasticity index (PI) and swelling potential, are
prevalent across Free State and Northern Cape and this possess an uncommon
challenges to design Engineers. This challenge is majorly as a result of 𝑀𝑟 value of high
plasticity index subgrades depends on the moisture content. Furthermore, the pavement
structures are generally under unsaturated conditions, such that the Groundwater Table
depth is below the depth of the subgrade in consideration for pavement design. The
classic soil mechanics considers the pavement structures to be under saturated
conditions. This means, that the soil mass under consideration consists of two phases,
solids (soil particles) and water. This assumption is generally acceptable, as it makes it
possible to develop simple analytical solutions that lay the foundation for geotechnical
engineering. However, since subgrade soils for pavements exist largely in an
unsaturated state. It is reasonably realistic to employ the fundamentals of unsaturated
soil mechanics in order to explore the resilient performance and deformation
characteristics of subgrades using 𝑀𝑟- suction correlation.

The evaluation of 𝑀𝑟 properties of unsaturated subgrade soils and evaluation effects

of suction and swelling stress on 𝑀𝑟 of the studied subgrades were summerized in this
research. Furthermore, this study developed mathematical predictive models. As well,

iii

© Central University of Technology, Free State

reviewed the design of flexible pavement, as per AASHTO 1993 flexible pavement
design guide by comparing designs made with both unsaturated 𝑀𝑟 design value and
conventional laboratory 𝑀𝑟 design value. Prior to the design of flexible pavement
exercise carried out in this study, some geotechnical tests were laboratory conducted
using both the unsaturated soil mechanics and classic soil mechanics approach i.e.
unsaturated CBR, shear stress, resilient modulus tests and filter paper test to measure
suction.
The laboratory result of the tested soils reveal that the unsaturated CBR and 𝑀𝑟 values
were 1.5 to 2.5 times higher than that of the conventional CBR and 𝑀𝑟 values. Among
other results, are shear strength result that followed similar trend, on the contrary, the
shear stress parameter i.e. ∅𝑏 of the tested soils were 1 to 1.5% lower than that of the
classical soil mechanics.
The SWRC curves were evaluated through the entire range of volumetric water content
using filter paper techniques. It is evident that 𝑀𝑟 depends on matric suction, which also
varies with moisture content, thereby, a 𝑀𝑟 -matric suction relationship provides sound
theoretical framework to account for moisture variation in unsaturated subgrade soils.
Three different SWRC models (Seki, Van Genuchten, and Fredlund and Xing) were
used for curve fitting, the SWRC revealed that Seki’s SWRC model best fitted the
laboratory data with coefficient of determination, R2 values ranging from 0.95458 -
0.99986. Whereas, Van Genuchten R2 values were in the bracket of 0.85796 - 0.93317,
and Fredlund and Xing R2 values were within the range of 0.89959 – 0.96142.

The SWRC curves evidenced that the subgrade soils with fine content (50%>P 200) like
FSS 1, 2 and NCS 2 and 3 recorded higher air entry values (AEV) within the range of
152 kPa – 250 kPa. Whereas, the subgrade soils with lower fine content yielded AEV
between the range of 90kPa -120kPa. The means that soils with high fine content starts
to desaturate at a very high AEV compared to soils with lower fine content, due to
inability of the soils to maintain saturation.

The analysis of experimental data obtained from the prepared specimens at different
moisture contents were used for multi-regression analysis using "NCSS11” software
package. The predictive mathematical models were developed for unsaturated CBR.
This model performed well against Ampadu’s (2007) model for prediction of unsaturated
CBR.

iv

© Central University of Technology, Free State

In addition, all the predictive models developed in this study i.e. Models 7, 8, and
extended Yang et.al model yielded more satisfactorily results than, Yang et.al (2005)
model, Liang et. al (2008) model and when compared with the laboratory measured 𝑀𝑟.
Thus, the predicted 𝑀𝑟 values using all these models were 1.2 to 1.5 times higher than
laboratory measured 𝑀𝑟 values with R2 within the range of 0.91 – 0.96 on curve
validation. The results showed that all the tested subgrade soils are highly depended on
𝑘1 parameter. Whereas, the effects of 𝑘2 , and 𝑘3 are proportional to deviatoric and
confining stresses relatively to 𝑀𝑟 values.

Lastly, AASHTO 1993 pavement design guide was used for the design exercise, on the
samples prepared on the dry side of optimum. Subgrade 𝑀𝑟 reflect the range of stress
states, commonly developed beneath the pavements that are subjected to moving
wheel loads. According to the design exercise in this study, the predicted 𝑀𝑟 design
value for the subgrade provided sufficient thickness that can support the entire
pavement structure. Whereas, the measure resilient modulus design value, under-
designed the pavement thus, required higher asphalt thickness layer.

Keywords: Subgrade soils, unsaturated CBR, CBR, swelling stress, soil suction, soil
water retention curve (SWRC) properties, unsaturated shear strength, shear strength
unsaturated resilient modulus, resilient modulus regression analysis, pavement design.

© Central University of Technology, Free State

ACKNOWLEDGEMENTS

It has been my gratification to pursue a Doctoral degree program in Civil Engineering

with a speciality in application of Geotechnical engineering in pavement. My Doctoral
degree program offered me a true reflection of Geotechnical engineering, as a unique
and versatile discipline with the challenges Geotechnical engineers solves. Pursuing a
Doctoral degree to me, has been challenging, time consuming, somewhat requires
sacrifices through the strength of my courage though it was truly an effectual
experience.

My profound gratitude goes to my puissant main supervisor Prof. Mohamed Mostafa,

for his expert guidance, valuable assistance, constructive suggestions and positive
criticisms throughout the course of this study. I also want to say a very good thanks to
Dr. Azza Moubarak of Civil Engineering Department Suez Canal University for accepting
to serve as a co-supervisor and also for her valuable guidance and assistance towards
this dissertation.

My sincere thanks go to Central University of Technology, for providing me scholarship

and opportunity to be part of their special graduate program. Truthfully, I appreciate the
provision of workspace, through which all my laboratory programs were conducted.

Similarly, my gratitude and appreciations are due to my mom Mrs Hanna Aneke who
has been a great pillar of support ever since my father died. To my Father, late Mr
Innocent Nzekwe Aneke, this is the part you have always wanted for me, to be an
academic Doctor. Further greetings and thanks go to my brothers: Chinedu Aneke,
Emeka Aneke and Onyebuchi Aneke for their unconditional love and words of
admonition given to me throughout the period of my research study.

vi

© Central University of Technology, Free State

TABLE OF CONTENTS
Page
Declaration .................................................................................................................... ii
Abstract ..................................................................................................................... iii-v
Acknowledgements ...................................................................................................... vi
Table of Contents ..................................................................................................... vii-x
List of Abbreviations ................................................................................................ xi-xii
Notations and Symbols.......................................................................................... xiii-xvi
Greek letters ....................................................................................................... xvii-xviii
List of Tables .............................................................................................................. xix
List of Figures ...................................................................................................... xx-xxiv
List of Appendices ............................................................................................. xxv-xxvii

CHAPTER 1 : INTRODUCTION .................................................................................. 1

1.1 Overview ............................................................................................................... 1
1.2 Problem statement ................................................................................................ 2
1.3 Hypothesis............................................................................................................. 3
1.4 Justification for unsaturated soil mechanics (USM) ............................................... 3
1.5 Research objectives .............................................................................................. 3
1.6 Research scope .................................................................................................... 4
1.7 Research organisation........................................................................................... 4

CHAPTER 2 : UNSATURATED SUBGRADES ......................................................... 6

2.1 Overview .............................................................................................................. 6
2.2 Fundamentals of unsaturated geotechnics ............................................................ 6
2.2.1 Capillarity ...................................................................................................... 9
2.2.2 Soil energy potential ................................................................................... 11
2.2.3 Matric suction ............................................................................................. 13
2.2.4 Osmotic suction .......................................................................................... 13
2.3 Soil-water retention curve (SWRC) ..................................................................... 14
2.3.1 SWRC model .............................................................................................. 15
2.4 Unsaturated soil shear strength ........................................................................... 17
2.4.1 Effective stress approach ........................................................................... 18
2.4.2 Independent state stress variable ............................................................... 22

vii

© Central University of Technology, Free State

 2.4.3 Suction stress characteristic curve (SSCC) approach ............................... 26
2.5 Previous studies on resilient modulus .................................................................. 29
2.5.1 Overview ..................................................................................................... 30
2.6 Resilient modulus (Mr ) concept ............................................................................ 30
2.6.1 Mr correlation with geotechnical soil properties ........................................... 31
2.6.2 Correlation between Mr and Unconfined compressive strength .................. 32
2.6.3 Correlation between Mr and CBR empirical ................................................ 33
2.6.4 Correlation of Mr with soil physical properties ............................................. 35
2.7 Factors affecting Mr .............................................................................................. 38
2.7.1 Stress state effect ........................................................................................ 38
2.7.2 Moisture condition effect .............................................................................. 38
2.7.3 Effects of soil type ....................................................................................... 40
2.7.4 Matric suction effects .................................................................................... 40
2.8 Mr models............................................................................................................. 41
2.8.1 Universal models ......................................................................................... 41
2.8.2 Model incorporating moisture variation ........................................................ 42
2.8.3 Model incorporating matric suction ............................................................. 45

CHAPTER 3 : SOIL PROPERTIES AND TESTING PROGRAMS ............................ 54

3.1 Overview .............................................................................................................. 54
3.2 Materials ............................................................................................................... 54
3.2.1 Soil geological location ................................................................................ 54
3.3 Soil chemical and mineralogical analysis ............................................................. 56
3.3.1 Soil XRD test ............................................................................................... 56
3.3.2 Soil XRF test................................................................................................ 57
3.4 Soil classification .................................................................................................. 57
3.4.1 Particles size distribution ............................................................................. 58
3.4.2 Atterberg limit test........................................................................................ 61
3.5 Geotechnical testing programs ............................................................................. 65
3.5.1 Free swell index (FSI) test ........................................................................... 65
3.5.2 Zero swelling test (ZST) .............................................................................. 66
3.5.3 Modified proctor compaction test ................................................................. 68
3.5.4 Filter paper test............................................................................................ 69
3.5.5 Filter paper calibration exercise ................................................................... 72

viii

© Central University of Technology, Free State

 3.5.6 California bearing ration (CBR) test ............................................................. 76
3.5.7 Unconfined compressive strength (UCS) test .............................................. 77
3.5.8 Consolidated undrain test (CUT) ................................................................. 78
3.5.9 Repeated load triaxial test (RLTT) ............................................................... 80

CHAPTER 4 : GEOTECHNICAL ANALYSIS OF THE TESTED SOIL .................... ..85

4.1 Over view ............................................................................................................. 85
4.2 Density-moisture relationship ............................................................................... 85
4.2.1 Effect of moisture on compacted FSS ......................................................... 85
4.2.2 Effect of moisture on compacted NCS......................................................... 86
4.2.3 Effect of moisture on compacted GPS......................................................... 87
4.2.4 Post-compaction soil behaviour ................................................................... 88
4.3 Analysis of swell potentials ................................................................................... 89
4.3.1 Free swell index (FSI) result ........................................................................ 90
4.3.2 Zero swell stress result ................................................................................ 90
4.4 Analysis of filter paper result ................................................................................ 93
4.4.1 Total, matric and osmotic suction results ..................................................... 94
4.4.2 Soil-H2O retention curve (SWRC) of the tested soils ................................. 100
4.5 Analysis of CBR results ...................................................................................... 106
4.5.1 Behaviour of unsoaked compacted soils ................................................... 107
4.5.2 Behaviour of soaked compacted soils ....................................................... 109
4.5.3 Analysis of unsaturated CBR result ........................................................... 112
4.5.4 Unsaturated CBR effect on matric suction ................................................. 116
4.5.5 Developed predictive equation for unsaturated CBR ................................ 120
4.6 Unconfined compresive strength result .............................................................. 124

CHAPTER 5 : CONSOLIDATED UNDRAINED RESULT...... ………………………..127

5.1 Overview ............................................................................................................ 127
5.2 Saturated soil shear strength (τ) result ............................................................... 127
5.2.1 Stress-strain response of the studied soils ................................................ 127
5.2.2 Effects of molding H2O content on shear strength parameters ................. 128
5.2.3 Shear strength and swelling pressure relationship .................................... 129
5.2.4 Shear strength and suction relationship ................................................ ….131

ix

© Central University of Technology, Free State

CHAPTER 6: UNSATURATED RESPONSE OF PAVEMENT STRUCTURE ...... …138
6.1 Overview ............................................................................................................ 138
6.2 Laboratory Mr test results ................................................................................... 138
6.2.1 FSS Mr result ............................................................................................ 138
6.2.2 NCS Mr result stress and plasticity index .................................................. 139
6.2.3 GPS Mr result. ........................................................................................... 140
6.2.4 Effect of moisture content on Mr .. .............................................................. 142
6.2.5 Effect of swelling stress on Mr .. .... ............................................................ 144
6.3 Mr Estimation based on soil geotechnical properties.. ....................................... 146
6.4 Mr -matric suction relationship............................................................................. 149
6.5 Unsaturated soil Mr ........................................................................................... 152
6.5.1 Validation of the develop and existing models ........................................... 157
6.6 Design of flexible pavement structure................................................................. 161
6.6.1 Design of flexible pavement structure using laboratory Mr value .............. 162
6.6.2 Flexible pavement design using unsaturated Mr value from model 4 ........ 165

CHAPTER 7 : CONCLUSION AND RECOMMENDATIONS ...... …………………….168

7.1 Summary ........ …………………………………………………………………………168
7.2 Conclusions ........................................................................................................ 169
7.3 Recommendation for further works .................................................................... 172

REFERENCES .................................................................................................. 173-187

APPENDIXES ................................................................................................... 188-245

© Central University of Technology, Free State

LIST OF ABBREVIATIONS
AASHTO American Association of State Highway and Transportation Officials
AIC Akaike Information Criterion
AEV Air Entry Value
ASTM American Society for Testing and Material
AWI Air-Water Interface
CBR California Bearing Ratio
CBRu Unsaturated California Bearing Ratio
CH High Plastic Clay
CL Medium Plastic Clay
CPC Capillary Pressure Curve
CSIR Council of scientific and Industrial Research
CU Consolidated Undrained
CUT Consolidated Undrain Test
DOP Dry Side of the Optimum
DCP Dynamic Cone Penetration (DCP)
EICM Enhanced Integrated Climatic Model (EICM)
FSI Free Swell Index
FSS Free State Soil
FPM Filter Paper Method
GPS Gauteng Province Soil
GWT Ground-Water Table
IS Indian Standards
LFWD Light Falling Weight Deflectometer
LVDT Linear Variable Differential Transducer
Mc Moisture Content
MDD Maximum Dry Density
M-E Mechanistic-Empirical
MEPDG Mechanistic-Empirical Pavement Design Guide
M𝐸𝑅 Effective Resilient Modulus Adjusted for Seasonal Variation
ML Low Plastic Silt
-PWP Negative Pore-Water Pressure
NCS Northern Cape Soil
NCHRP National Cooperative Highway Research Program

xi

© Central University of Technology, Free State

OMC Optimum Moisture Content
OC Percentage Organic Content
Ps Swelling Stress
PSD Particles Size Distribution
PVC Poly-Vinyl Chloride
RAM Recycled Asphalt Material
RAP Recycled asphalt pavement
RLTT Repeated Load Triaxial Test
S0 Combined Standard Error of the Traffic and Performance Predictions
SAPDA South Africa Pavement Design Methods
SSR Sum of Squared Residuals
SSCC Suction Stress Characteristic Curve
SN Structural Number of the Total Pavement Thickness
SWCC Soil Water Characteristic Curve
SWRC Soil Water Retention Curve
TMH Technical Method for Highways
UCS Unconfined Compressive Strength
USCS Unified Soil Classification System
USM Unsaturated soil mechanics
XRF X-ray fluorescence
XRD X - Ray Diffraction
WOP Wet Side of the Optimum
ZR Standard normal Deviate
ZST Zero Swelling Test
ΔPSI Difference Between the 𝑃𝑆𝐼𝑜 and Terminal 𝑃𝑆𝐼𝑡 Serviceability Indices

xii

© Central University of Technology, Free State

NOTATIONS AND SYMBOLS
Roman letters
𝐴 The corresponding cross−sectional area
𝐴0 The initial cross−sectional area of the specimen
af Soil parameter related to the air entry of the soil
𝑎𝑖 Layer coefficient
Ac Activity of clay
a,n and m Fitting Parameters for SWRC
br Beam ratio
C Correction factor
C' Intercept of the Mohr-Coulomb failure envelope at specific suction
𝐶𝑐 Coefficient of curvature
𝐶𝑢 Coefficient of uniformity
𝐷𝑖 Layer thickness
e Void ratio
ec Unit electron charge
e Natural constant 2.718
f* Interaction function between the equilibrium of the soil structure and
the equilibrium of the contractile skin.
g Gravitational acceleration
Gs Specific gravity
H Relative humidity of the pore air above the meniscus
hc Capillary height
𝑘1, 𝑘2, 𝑘3 Regression constant
K Boltzmann’s constant
Ko Number of estimated parameter
Lo The initial length of the specimen
LL Liquid limit
m Number of relevant soil parameter
m1 Mass of wet filter paper + cold tare
m2 Mass of wet filter paper + hot tare mass
mf Soil parameter related to the residual water content condition
𝑚𝑖 Layer Drainage Coefficient

xiii

© Central University of Technology, Free State

M Total mass
M1 Empty mass of volumetric flask
M2 Mass of pycnometer+ oven dry soil
M3 Mass of pycnometer+ oven dry soil+ filled water
M4 Mass of pycnometer+ filled with water only
Ma Mass of air
Mw Mass of water
Ms Mass of solids
Mc Mass of the contractile skin
Mc Mass of container
Mf Mass of the dry filter paper
Mi Unit mass of surcharge
Mm Mass of the mould and base plate
𝑀𝑟 Resilient modulus
Msoil Mass of the dry soil
Mt Mass of the mould, base plate, and wet soil
Mw Mass of water to be added
Mv Mass of water in the filter paper
n Number of surcharge
nf Soil parameter related to the rate of desaturation
nw Porosity relative to the water phase
nc Porosity relative to the contractile skin
ns Porosity relative to the soil particles
P Plunger−load KN/m2 for the tested soil
p′ Pressure Deficiency
Pa Atmospheric pressure
PI Plasticity index
Ps Plunger−load in KN/m2 for the standard soil
Ps Swelling stress
PL Plastic limit
P200 Percentage of passing #200 sieve (75μm)
𝑞 Unconfined compressive strength
R Radius of the capillary tube
R2 Correlation coefficient

xiv

© Central University of Technology, Free State

R Universal gas constant
Rd Relative density of water according to temperature
Rs Sheath radius of curvature/ Radius of curvature of the meniscus
R1, R2 Radius of curvature of warped membrane
S Degree of Saturation
Sr Residual of saturation
Se Effective saturation
T Absolute Temperature
T Temperature
Tc Cold tare mass
Th Hot tare mass
Ts Surface Tension
Tk Absolute temperature
Tzy Shear stress on the z-plan in y direction
Ua Pore air pressure
Uw Pore water pressure
Uv Partial pressure of pore
Uv0 Saturation pressure of water
V Total volume
Vw0 Specific Volume of Water
Va Volume of air
Vc Volume of contractile skin
Vf Final volume of the specimen
Vi Initial volume of the specimen
Vk Volume of the soil containing Kerosene.
Vr Residual Water Volume
Vs Volume of solids in water
Vm Volume of the mould
Vw Volume of water
Wv Molecular mass of water vapour
W Moisture content
W1 Mass of container + wet soil
W2 Mass of container + wet soil
Wc Moisture content

xv

© Central University of Technology, Free State

Wf Water content of the filter paper
Wi Initial water content
Wopt Optimum moisture content
Wt Targeted moisture content
𝑊18 Predicted Number of 18−kip (80kN) Single Axle Load
Xij Independent variables
Y Dependent variables

xvi

© Central University of Technology, Free State

Greek letters
α1 Angle of contact
β Angle between the tension surface and horizontal.
εc Dielectric constant medium
εr = ε2 -ε1 Recoverable strain
ε Random error representing the discrepancies in the approximation
𝜀𝑟 Resilient strain
𝑢𝑒 Air entry value
η Electrolyte concentration
qu Unconfined compressive strength
θ =σ1 + σ2 + σ3 Bulk stress
𝜃𝑔 Gravimetric water content
𝜃𝜈 Volumetric water content
τoct Octahedral shear stress
ρw Density of water
ρs Soil particle density
ψt Total soil suction
ψm Matric suction
ua -uw Matric suction
ψo Osmotic suction
τxy Shear stress on the x-plan in y-direction

τzy Shear stress on the z-plan in y-direction

σy Total normal stress in y-direction

θs Volumetric water at saturation

θr Residual volumetric water content
θω Volumetric water content
γd Dry density
τ Shear strength of unsaturated soil
γdmax Maximum dry density
σ-ua Net normal stress

xvii

© Central University of Technology, Free State

σ3 Minor principal stress or confining stress in the triaxial
θ Bulk stress
σd = σ1 -σ3 Deviator stress
𝜎𝑐𝑦𝑐 Applied cyclical stress
σ' Effective stress
ϕ′ Effective angle of friction
ϕb Angle of friction due to suction
λo , ηo , ξo , ζo , βo , μo Intercepts
λi , ηi , ξi , ζi , βi , μi Multi-regression analysis coefficient
ϕ (x) Normalized form of the cumulative normal distribution function
ϕ Internal diameter of the consolidation ring
Θ Normalised water content between saturation and residual
water content
α,k,β Fitting constants
α Intercept of 𝑀𝑟 at given 𝜃𝜏𝑜𝑐𝑡 against suction relationship
β Slope of 𝑀𝑟 at given 𝜃𝜏𝑜𝑐𝑡 against suction relations
τsr Shear strength at residual suction
Pa Atmospheric pressure
ΔC Chemical concentration difference across a semipermeable
membrane
Δ𝜇𝑐 Change due to curvature of the air−water interface (capillarity)
Δ𝜇0 Change due to osmotic effects (dissolved solutes)
Δ𝜇𝑒 Change due to electric field Δ𝜇𝑓=change due to Van der Waals
forces
ΔU Difference in stress on a two dimension curved arc
ΔV Initial change in volume of a specimen
Δ𝑃 Pressure Difference
𝜎𝑠 Air entry pressure of water saturated soil

xviii

© Central University of Technology, Free State

LIST OF TABLES
Page

Table 2.1: Soil properties used to validate the closed equation ........ ………………29

Table 2.2: 𝑀𝑟 -CBR emperical correlations ........ ……………………………………...34

Table 3.1: Soils chemical composition ........ …………………………………………..57

Table 3.2: Grain size analysis of the studied subgrades ........ ………………………61

Table 3.3: Consistency limits values ......... …………………………………………….62

Table 3.4: USCS, AASHTO and TRH soil classification ..... ………………………..65

Table 3.5: Time duration and various methods of measuring suction……………...69

Table 4.1: Free swell index result ......………………………………………………….90

Table 4.7: Description of CBR values subgrade quality ........ ……………………...107

Table 4.8: Four days unsoaked soils for CBR ........ …………………………………108

Table 4.9: Four days soaked for CBR ........ ………………………………………….110

Table 4.10: Parameters symbols and their corresponding values ........……………121

Table 4.11: Values of unsaturated CBR........ …………………………………………..123

Table 6.3: Unsaturated reisilient modulus using model 4 ....... ……………………….153

Table 6.4: Unsaturated reisilient modulus using model 5 …………………………….154

Table 6.5: Unsaturated reisilient modulus using model 6 …………………………….155

Table 6.6: Parameters for 3-layered pavement structure design one . ……………...162

Table 6.7: Resilient modulus layers thickness and SN for design one .. ……………..163

Table 6.8: Parameters for 3-layered pavement structure design two .. ………………165

Table 6.9: Resilient modulus layers thickness and SN for design two .. ………….....165

xix

© Central University of Technology, Free State

LIST OF FIGURES
Page

Figure 1.1: Typical unsaturated pavement structure… ..………………………………………1

Figure 1.2: Extremely arid, and semiarid regions………………………… .. . ……………......2

Figure 2.1: Hydrologic system of unsaturated geotechnics.. ………………………………....7

Figure 2.2: Air water interphase .. ………………………………………………………………8

Figure 2.3: Phase diagrams for unsaturated soil system .. ……………………………………8

Figure 2.4: Capillarity model . …………………………………………………………………...10

Figure 2.5: Soil-water retention curveRelationship in Atterberg limits . …………………….14

Figure 2.6: Normal and shear stresses in unsaturated geotechnics . ………………………17

Figure 2.7: Effective stress parameters versus suction ratio. ……………………………….20

Figure 2.8: Unsaturated Mohr-coulomb failure envelope . …………………………………...23

Figure 2.9: Two dimensional projections of failure enevlopes suctions . …………………...24

Figure 2.10: Subgrade stress-strain behaviour under repeated load ………………………..31

Figure 2.11: Variation of moisture content on Mr ………………………………………………39

Figure 2.12: Typical moisture changes on Mr using EICM ……………………………………43

Figure 2.13: Collected database against EICM model …………………………………………44

Figure 2.14: Adjustment factor, Fu, for (S-𝑆𝑜𝑝𝑡 ) and 𝑤𝑃𝐼 ……………………………………....45

Figure 2.15: Predicted against measured Mr values using Yang et. al (2005).……………….46

Figure 2.16: A-6 soil predicted against measured Mr values . Liang et. al (2008)…………...50

Figure 2.17: Good line fit measure against predicted Mr values for PI=5…………………….52

Figure 3.1: GPS precision indicating visited sites ..……………………………………………55

Figure 3.2: Sample borrowed pits at different sites…………………………………………….56

Figure 3.3: FSS grian size distribution curve ........…………………………………………….59

Figure 3.4: NCS grian size distribution curve ........…………………………………………….59

Figure 3.5:GPS grain size distribution curve……………………………………………..............60

Figure 3.6: Casagrande plasticity chart for FSS . ……………………………………………...63

Figure 3.7: Casagrande plasticity chart for NCS..……………………………………………..63

xx

© Central University of Technology, Free State

Figure 3.8: Casagrande plasticity chart for GPS ………………………………………………64

Figure 3.9: Free swell index setup of the subgrade soil ………………………………………66

Figure 3.10: Fitted specimens in oedometer ring ……………………………………………….67

Figure 3.11: Zero swell test setup ………………………………………………………………...67

Figure 3.12: Sliced soil and sandwiched filter paper ……………………………………………70

Figure 3.13: Sandwiched in temperature-controlled ice chest .. ………………………………71

Figure 3.14: Weighing the filter paper inside moisture can …………………………………….71

Figure 3.15: Filter calibration curve……………………………………………………………….72

Figure 3.16: Total suction of NaCl at 2200 c ...…………………………………………………..73

Figure 3.17: Plastic support holding filter paper…………………………………………………74

Figure 3.18: Authors filter papers calibration curve……………………………………………..75

Figure 3.19: Automated CBR testing machine…………………………………………………..76

Figure 3.20: Cured soil specimens .. ……………………………………………………………..77

Figure 3.21: Complete triaxial testing arrangement .. …………………………………………..79

Figure 3.22: Laboratory specimen for Mr testing………………………………………………..80

Figure 3.23: Vibro-compactor set-up...……..………………………………………………….......81

Figure 3.24: Wrapped specimens awaiting Mr testing …………………………………………81

Figure 3.25: Setting-up Mr equipment ……………………………………………………………82

Figure 3.26: Complete repeated load triaxial device ……………………………………………83

Figure 4.1: Moisture-density relationship of FSS………………………………………………86

Figure 4.2: Moisture-density relationship of NCS……………………………………………...87

Figure 4.3: Moisture-density relationship of GPS……………………………………………...87

Figure 4.4: Post compacted soils behaviour ...…………………………………………………88

Figure 4.5: Swelling pressure with varying dey density for FSS ……………………………..91

Figure 4.6: Swelling pressure with varying dey density for NCS …………………………….92

Figure 4.7: Swelling pressure with varying dey density for GPS …………………………….93

Figure 4.8: Suction versus gravimetric water content for FSS1.. ……………………………95

Figure 4.9: Suction versus gravimetric water content for FSS 2. ……………………………95

xxi

© Central University of Technology, Free State

Figure 4.10: Suction versus gravimetric water content for FSS 3. . …………………………..96

Figure 4.11: Suction versus gravimetric water content for NCS 1.. . ………………………….97

Figure 4.12: Suction versus gravimetric water content for NCS 2. . …………………………..97

Figure 4.13: Suction versus gravimetric water content for NCS 3. . …………………………..98

Figure 4.14: Suction versus gravimetric water content for GPS 1. .. ………………………….99

Figure 4.15: Suction versus gravimetric water content for GPS 2. .. ………………………….99

Figure 4.16: Suction versus gravimetric water content for GPS 3. . …………………………100

Figure 4.17: FSS 1 soil water retention curve. . ………………………………………………..101

Figure 4.18: FSS 2 soil water retention curve. …………………………………………………102

Figure 4.19: FSS 3 soil water retention curve. …………………………………………………102

Figure 4.20: NCS 1 soil water retention curve. . ……………………………………………….103

Figure 4.21: NCS 2 soil water retention curve. . ……………………………………………….103

Figure 4.22: NCS 3 soil water retention curve. . ……………………………………………….104

Figure 4.23: GPS 1 soil water retention curve. . ……………………………………………….104

Figure 4.24: GPS 2 soil water retention curve. . ……………………………………………….105

Figure 4.25: GPS 3 soil water retention curve. . ……………………………………………….105

Figure 4.26: Soaked and unsoaked compacted specimens.. ………………………………..107

Figure 4.27: Unsoaked FSS CBR variation with water content .. …………………………….112

Figure 4.28: Unsoaked NCS CBR variation with water content. . ……………………………113

Figure 4.29: Unsoaked GPS CBR variation with water content. . ……………………………113

Figure 4.30: Soaked FSS CBR variation with water content. . ……………………………….114

Figure 4.31: Soaked NCS CBR variation with water content….………………... .. ……........115

Figure 4.32 Soaked GPS CBR variation with water content….………………... . …….........115

Figure 4.33: Variation of unsoaked FSS CBR versus matric suction…………..... ……........116

Figure 4.34: Variation of unsoaked NCS CBR versus matric suction...………..... ……........117

Figure 4.35: Variation of unsoaked GPS CBR versus matric suction……...…..... ……........117

Figure 4.36: Variation of soaked FSS CBR versus matric suction…………... . …………….119

Figure 4.37: Variation of soaked NCS CBR versus matric suction…………... .. …………...119

xxii

© Central University of Technology, Free State

Figure 4.38: Variation of soaked GPS CBR versus matric suction…………...... ..... ……….120

Figure 4.39 Measured versus predicted unsaturated CBRu model 1………... ........... …...122

Figure 4.40: Measured versus predicted unsaturated CBRu model 2………... ........... …...122

Figure 4.41: FSS seven day UCS values………………………………..………... ........... ….125

Figure 4.42: NCS seven day UCS values ………... ........... ………………………………….125

Figure 4.43: NCS seven day UCS values ………... ........... ………………………………….126

Figure 5.51: FSS shear strength-swelling pressure relationship……………………………….130

Figure 5.52: NCS shear strength-swelling pressure relationship ………………………………130

Figure 5.53: Shear strength-swelling pressure relationship …………………………………….131

Figure 5.54: Usaturated failure envelope for FSS 1…………….………………………………..133

Figure 5.55: Usaturated failure envelope for FSS 2…………….………………………………..133

Figure 5.56: Usaturated failure envelope for FSS 3…………….………………………………..134

Figure 5.57: Usaturated failure envelope for NCS 1.…………….………………………………134

Figure 5.58 Usaturated failure envelope for NCS 2…………….………………………………..135

Figure 5.59: Usaturated failure envelope for NCS 3…………….………………………………135

Figure 5.60: Usaturated failure envelope for GPS 1…………….………………………………136

Figure 5.61: Usaturated failure envelope for GPS 2…………….………………………………136

Figure 5.62: Usaturated failure envelope for NCS 2…………….………………………………137

Figure 6.1: Mr of FSS 1 at 8.55% moisture with deviatoric stress…………………………….139

Figure 6.6: Mr of NCS 1 at 8.38% moisture with deviatoric stress……………………………..140

Figure 6.10: Mr of GPS 1 at 9.28% moisture with deviatoric stress……………………………141

Figure 6.16: Mr -moisture relationship for FSS……………………………………....................142

Figure 6.17: Mr -moisture relationship for NCS………………………………….…..................143

Figure 6.18: Mr -moisture relationship for NCS………………………………….…..................144

Figure 6.19: Mr swelling pressure for FSS………………………………….….........................145

Figure 6.20: Mr swelling pressure for NCS………………………………….….........................145

Figure 6.21: Mr swelling pressure for GPS………………………………….….........................146

xxiii

© Central University of Technology, Free State

Figure 6.22: Measured versus Predicted Mr for model 1…..................................................148

Figure 6.23: Measured versus Predicted Mr for model 2…..................................................148

Figure 6.24: Measured versus Predicted Mr for model 3…..................................................149

Figure 6.25: Mr -matric suction relationship FSS…………..................................................150

Figure 6.26: Mr -matric suction relationship NCS…………..................................................151

Figure 6.27: Mr -matric suction relationship GPS…………..................................................151

Figure 6.28: Validation of model 4…………............................................................................158

Figure 6.29: Validation of model 5…………............................................................................158

Figure 6.30: Validation of model 6…………............................................................................159

Figure 6.31:Validation of model 7…………...........................................................................160

Figure 6.32: Validation of model 8…………..........................................................................160

Figure 6.33: Nomograph for flexible pavement......................................................................163

Figure 6.34: Pavement design thickness one.......................................................................164

Figure 6.35: Pavement design thickness one.......................................................................166

xxiv

© Central University of Technology, Free State

LIST OF APPENDICES

APPENDIX A: Pages

X-ray diffraction analysis for FSS 1....................................................................................... 188

X-ray diffraction analysis for FSS 2 ...................................................................................... 189
X-ray diffraction analysis for FSS 3 ...................................................................................... 190
X-ray diffraction analysis for NCS 1 ...................................................................................... 191
X-ray diffraction analysis for NCS 2 ...................................................................................... 192
X-ray diffraction analysis for NCS 3 ...................................................................................... 193
X-ray diffraction analysis for GPS 1 ...................................................................................... 195
X-ray diffraction analysis for GPS 2 ...................................................................................... 196
X-ray diffraction analysis for GPS 3 ...................................................................................... 197
X-ray fluorescence analysis for FSS 1 .................................................................................. 198
X-ray fluorescence analysis for FSS 2 .................................................................................. 199
X-ray fluorescence analysis for FSS 3 .................................................................................. 200
X-ray fluorescence analysis for NCS 1 ................................................................................. 201
X-ray fluorescence analysis for NCS 2 ................................................................................. 202
X-ray fluorescence analysis for NCS 3 ................................................................................. 203
X-ray fluorescence analysis for GPS 1 ................................................................................. 204
X-ray fluorescence analysis for GPS 2 ................................................................................. 205
X-ray fluorescence analysis for GPS 3 ................................................................................. 206

APPENDIX B:
Table 4.2: Soils moisture conditions and swelling stres characteristics ................. 207
Table 4.3 Suction values at varying moisture content ........................................... 208
Table 4.4: Fitting parameters of difference SWRC models for FSS ....................... 209
Table 4.5: Fitting parameters of difference SWRC models for NCS ....................... 211
Table 4.6: Fitting parameters of difference SWRC models for GPS ....................... 212

APPENDIX C:
Figure 5.1: Stress-displacement curve of FSS 1 at 𝜎𝑐 = 20kPa. .………………………………214
Figure 5.2: Stress-displacement curve of FSS 2 at 𝜎𝑐 = 20kPa. .………………………………214
Figure 5.3: Stress-displacement curve of FSS 3 at 𝜎𝑐 = 20kPa. .………………………………215
Figure 5.4: Stress-displacement curve of FSS 1 at 𝜎𝑐 = 50kPa. .………………………………215
Figure 5.5: Stress-displacement curve of FSS 2 at 𝜎𝑐 = 50kPa. .………………………………216

xxv

© Central University of Technology, Free State

Figure 5.6: Stress-displacement curve of FSS 3 at 𝜎𝑐 = 50kPa. .……………………………….216
Figure 5.7: Stress-displacement curve of FSS 1 at 𝜎𝑐 = 100kPa. ..…………………………….217
Figure 5.8: Stress-displacement curve of FSS 2 at 𝜎𝑐 = 100kPa. .……………………………..217
Figure 5.9 Stress-displacement curve of FSS 3 at 𝜎𝑐 = 100kPa. .………………………………218
Figure 5.10: Stress-displacement curve of NCS 1 at 𝜎𝑐 = 20kPa. .……………………………..218
Figure 5.11: Stress-displacement curve of NCS 2 at 𝜎𝑐 = 20kPa. .……………………………..219
Figure 5.12: Stress-displacement curve of NCS 3 at 𝜎𝑐 = 20kPa. .……………………………..219
Figure 5.13: Stress-displacement curve of NCS 1 at 𝜎𝑐 = 50kPa. .………………………………220
Figure 5.14: Stress-displacement curve of NCS 2 at 𝜎𝑐 = 50kPa. .……………………………..220
Figure 5.15: Stress-displacement curve of NCS 3 at 𝜎𝑐 = 50kPa. .……………………………..221
Figure 5.16: Stress-displacement curve of NCS 1 at 𝜎𝑐 = 100kPa.……………………………..221
Figure 5.17: Stress-displacement curve of NCS 2 at 𝜎𝑐 = 100kPa. .……………………………222
Figure 5.18: Stress-displacement curve of NCS 3 at 𝜎𝑐 = 100kPa. .………………..................222
Figure 5.19: Stress-displacement curve of GPS 1 at 𝜎𝑐 = 20kPa. .……………………………..223
Figure 5.20: Stress-displacement curve of GPS 2 at 𝜎𝑐 = 20kPa. .……………………………..223
Figure 5.21: Stress-displacement curve of GPS 3 at 𝜎𝑐 = 20kPa. .……………………………..224
Figure 5.22: Stress-displacement curve of GPS 1 at 𝜎𝑐 = 50kPa. .……………………………..224
Figure 5.23: Stress-displacement curve of GPS 2 at 𝜎𝑐 = 50kPa. .……………………………..225
Figure 5.24: Stress-displacement curve of GPS 3 at 𝜎𝑐 = 50kPa. .……………………………..225
Figure 5.25: Stress-displacement curve of GPS 1 at 𝜎𝑐 = 100kPa. …..………………………..226
Figure 5.26: Stress-displacement curve of GPS 2 at 𝜎𝑐 = 100kPa. …..………………………..226
Figure 5.27: Stress-displacement curve of GPS 3 at 𝜎𝑐 = 100kPa. …..………………………..227
Figure 5.28: FSS 1 failure envelope at 8.55% and 13.46% moisture content…………………227
Figure 5.29: FSS 1 failure envelope at 17.65% and 24.27% moisture content…...................228
Figure 5.30: FSS 1, 2 failure envelope at 31.75% and 9.97% moisture content……………..228
Figure 5.31: FSS 2 failure envelope at 13.67% and 18.58% moisture content…...................228
Figure 5.32: FSS 2 failure envelope at 24.49% and 32.15% moisture content…...................229
Figure 5.33: FSS 3 failure envelope at 7.49% and 14.15% moisture content….....................229
Figure 5.34: FSS 3 failure envelope at 18.54% and 22.94% moisture content…...................229
Figure 5.35: FSS 3,NCS 1 failure envelope at 28.33% and 8.38% moisture content………..230
Figure 5.36 NCS 1 failure envelope at 12.73% and 17.49% moisture content……………….230
Figure 5.37: NCS 1 failure envelope at 24.40% and 28.33% moisture content……………….230
Figure 5.38: NCS 2 failure envelope at 9.18% and 14.53% moisture content………………..231
Figure 5.39: NCS 2 failure envelope at 18.38% and 24.89% moisture content………….......231
Figure 5.40: NCS 3 failure envelope at 30.22% and 9.97% moisture content………………..231
Figure 5.41: NCS 3 failure envelope at 14.53% and 18.9% moisture content………………..232

xxvi

© Central University of Technology, Free State

Figure 5.42: NCS 3 failure envelope at 25.24% and 32.88% moisture content………………232
Figure 5.43: GPS 1 failure envelope at 9.28% and 12.89% moisture content………………..232
Figure 5.44: GPS 1 failure envelope at 16.77% and 23.65% moisture content…..................233
Figure 5.45: GPS 1, 2 failure envelope at 29.88% and 9.97% moisture content…..................233
Figure 5.46: GPS 2 failure envelope at 12.53% and 18.53% moisture content……………….233
Figure 5.47: GPS 2 failure envelope at 23.25% and 30.33% moisture content……………….234
Figure 5.48: GPS 3 failure envelope at 9.28% and 14.33% moisture content…….................234
Figure 5.49: GPS 3 failure envelope at 18.88% and 23.53% moisture content……………….234
Figure 5.50: GPS 3 failure envelope at 27.45% moisture content………………………………235
Table 5.1: Saturated and unsaturated shear strength parameters ...... …………….235

APPENDIX D:
Figure 6.2: Mr of FSS 1 at 13.67% and 17.65% moisture with deviatoric stress……………...237
Figure 6.3: Mr of FSS 2 at 9.97% and 13.67% moisture with deviatoric stress……………..237
Figure 6.4: Mr of FSS 2, 3 at 18.58% and 7.28% moisture with deviatoric stress…………..238
Figure 6.5: Mr of FSS 3 at 14.15% and 18.45% moisture with deviatoric stress…………….238
Figure 6.7: Mr of NCS 1 at 12.73% and 17.49% moisture with deviatoric stress……………239
Figure 6.8: Mr of NCS 2 at 9.18% and 14.53% moisture with deviatoric stress……………..239
Figure 6.9: Mr of NCS 2, 3 at 18.38% and 9.97% moisture with deviatoric stress................240
Figure 6.10: Mr of NCS 3 at 14.85% and 18.89% moisture with deviatoric stress................240
Figure 6.12: Mr of GPS 1 at 12.89% and 16.77% moisture with deviatoric stress................241
Figure 6.13: Mr of GPS 2 at 9.97% and 12.53% moisture with deviatoric stress..................241
Figure 6.14: Mr of GPS 2, 3 at 18.53% and 9.28%% moisture with deviatoric stress….........242
Figure 6.15: Mr of GPS 3 at 14.37% and 18.88%% moisture with deviatoric stress…………242
Table 6.1: Parameters and symbols for models 1, 2 and 3 .................................... ………243
Table 6.2: Parameters and symbols for models 7 and 8 .................................................. 244
Figure 6.29: Layer coefficient chart for asphalt concrete ................ …………………………244

xxvii

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPER 1: INTRODUCTION
1.1 Overview
The unsaturated subgrade layer plays major role in the overall performance of flexible
pavement structure. The cyclic response of subgrade layer depends greatly upon
moisture content and matric suction, but these effects have been conventionally difficult
to quantify. Following the development of Mechanistic-Empirical (M-E) design
technique. Resilient modulus (𝑀𝑟 ) is considered an essential input design parameter in
characterising cyclic behaviour of pavement structure under cyclic loading. The 𝑀𝑟
values are determined in laboratory through repeated load triaxial test (RLTT). Basically,
M-E design analysis is currently been used by American Association of State Highway
and Transportation Officials (AASHTO) and other transportation agencies around the
globe. Perhaps, pavement structures are typically constructed under unsaturated
condition within the degrees of saturation that ranges from 75 to 90%. In spite of this
fact, design engineers still uses the conventional pavement design approach that are
based on saturated soil mechanics, rather than unsaturated soil mechanics (USM)
principles. Scholars like: Fredlund and Morgenstern, (1977), Fredlund et al., (1978),
Fredlund and Rahardjo, (1993), Fredlund, (1996), and Vanapalli et al., (1996) introduced
a theoretical network, for expressing moisture response of unsaturated soils as regards
pavement design.

𝐅𝐢𝐠𝐮𝐫𝐞𝟏. 𝟏: Typical unsaturated pavement structure Ampadu, (2007)

Page | 1

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

According to Figure 1.1, it becomes imperative to analyse pavement response under

unsaturated condition, by considering resilient modulus with respect to matric suction.

1.2 Problem Statement

The climatic state around the earth, ranges from very humid to dry and the climatic
stratification depend on the standard annual net moisture influx at the surface level.

𝐅𝐢𝐠𝐮𝐫𝐞𝟏. 𝟐: Extremely arid, and semiarid regions (Fredlund &Rahardjo, 1993)

Figure 1.2 shows that South Africa is cut-across within arid and semi-arid zone, and this
implies that 95% of roadbed in South African are under unsaturated condition. Despite
this factors, design engineers failed to account for negative pore-water pressure (-PWP)
or matric suction, as this might lead to an inappropriate design of pavement structure.
Resilient response of subgrade is measured using expensive laboratory tests, that is
somewhat time-consuming. Design engineers sometimes, uses overestimated
backcalculated values, that in turn lead to over-design of pavement. These challenges
encourage the need for a valuable and inexpensive geotechnical testing procedures,
that can easily and directly determine unsaturated 𝑀𝑟 of subgrades. As most of this
testing equipment are not readily available in the university laboratories and highway
engineering research centres in south Africa. Therefore, it becomes very difficult to
perform obtain unsaturated 𝑀𝑟 value.

Page | 2

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

1.3 Hypothesis
Matric suction, swelling pressure and soil type influences the shear strength, resilient
modulus of unsaturated soils.

1.4 Justification for USM

Approximately, South Africa is constituted of arid and semi-arid regions. This means
that groundwater table (GWT) is deep and soils are identified under unsaturated state.
Hence, stresses associated with pavement structure are active above GWT. These
stresses are negative with respect to the atmospheric pressure (i.e. matric suction). This
implies that rational design of pavement in arid and semi-arid regions should be carried
out based on the principles of USM.

1.5 Research Objectives

With the view to realise the objectives of this dissertation, the following masterplan were
directly pursued to investigate the geotechnical behaviour of unsaturated soils for road
pavement structure under cyclic loading:
 Firstly, the objective is to characterise the studied subgrade soils, identified
across South Africa, using particles size distribution (PSD) analyses, consistency
limit, among others are X-Ray Diffraction (XRD) and X-ray fluorescence (XRF)
as to quantify the mineralogical constituents and elemental compounds of the
this identified soils.

 To evaluate the geotechnical properties of the identified subgrades, through

laboratory testing, by conducting standard civil engineering tests in both
saturated and unsaturated conditions, ranging from compaction test, California
bearing ratio (CBR) test, free swell index (FSI) test, zero swelling stress, suction
test, consolidated undrain test (CUT), unconfined compressive strength (UCS)
test, and repeated load triaxial (RLT) test.

 To develop linear mathematical predictive model for unsaturated CBR and 𝑀𝑟

using other soil mechanical properties. As well, compare the predicted CBR and
𝑀𝑟 values with laboratory measured values.

Page | 3

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

 To review AASHTO pavement design method using saturated and unsaturated

design parameters, in order to identify the most reliable, sustainable and
conservative pavement design value.

1.6 Research Scope

The jurisdiction of this research is primarily experimental, with the objectives outlined
above, the scopes are within the confinements of the following:
o Develop linear mathematical predictive model, that can predict CBR and resilient
modulus, using other geotechnical properties.

o Analyse the experimental data for both saturated and unsaturated soil
mechanics, and further use this data to empirically design a pavement.

o The design exercise did not consider Mechanistic-Empirical (M-E) design

technique, rather AASHTO 1993 pavement design guide, that is purely empirical
method is considered.

1.7 Research Organization

This dissertation is presented in seven-chapter format, listed as follows:

Chapter 1 is the introductory chapter and it covers background to the research, problem
statement, justification of study, objectives, scopes and lastly, the organisation of the
research.

Chapter 2: presents the characteristics of unsaturated soils and provides technical

background on the basic knowledge needed for good understanding of USM. These
principles knowledge include: the three zones of unsaturated soil, stress state variables,
suction matric and the soil-water retention curve. Furthermore, this Chapters presents
previous studies on 𝑀𝑟 , factors affecting 𝑀𝑟 , 𝑀𝑟 model, 𝑀𝑟 generated from other
geotechnical properties and some correlation between 𝑀𝑟 and other geotechnical soil
properties.

Chapter 3: gives a description of identified soils and methodology adopted to achieve

the objectives of this research and this followed by standard civil engineering laboratory
Page | 4

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

testing methods, leading to the classification of this soils with their mineralogical
contents. This chapter also provided illustrations of civil engineering testing procedure
and equipment used to conduct these tests. i.e. consolidated undrained test (CUT), RLT
test and filter paper test.

Chapter 4: Following the laboratory investigations, this chapter, covers basic test
results, analyses and scientific report of all the laboratory tests conducted in this study.

Chapter 5: Provides the shear strength results of the studied soils, the effects of
moulding water content, suction matric and swelling stress on shear strength of the
roadbeds. This chapter further described the failure envelops and shear stress
parameters of the subgrades under unsaturated soil mechanics.

Chapter 6: this chapter presents 𝑀𝑟 results of the tested soils and further evaluates the
influenced of confining pressure, swelling pressure moisture content and suction on 𝑀𝑟 .
subsequently, validation of existing and developed 𝑀𝑟 predictive models were carried
out ,through comparing the models with laboratory measured 𝑀𝑟 . This chapter further
demonstrates a superlative comparison between unsaturated soil mechanics concept
of pavement design method and conventional pavement method, using unsaturated 𝑀𝑟
values and laboratory 𝑀𝑟 values.

Chapter 7: serves as the summary, conclusions and recommendations.

Page | 5

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPTER 2: UNSATURATED SUBGRADES

2.1 Overview
Pavement structure is comprised of different layers, which collectively act to provide
support for applied vehicular loads while serviceability measure is maintained.
Pavement is usually situated in an unsaturated state, i.e. located above groundwater
table (GWT) where it is considered an Active Zone. Fredlund’s effective stress theory,
basically explained shear strength of unsaturated soil, as partially saturated. Since, most
natural subgrades are located above GWT. Whereas, in state the saturated subgrade
is considered to be totally saturated. These conventional idealizations assume that the
pore water pressures (PWP) for soils beneath the water table is positive (+PWP). On
the other hand, the PWP above the GWT is certified to be negative (-PWP). Terzaghi’s
effective stress theory is widely accepted to evaluate the shear strength of saturated
soils. Realistically, soil moisture is expected to vary seasonally, because moisture
variation is known as negative pore-water pressure (-PWP). Thus, is prevalent in
pavement hence its above GWT and is referred to as soil suction.

2.2 Fundamentals Of Unsaturated Geotechnics

Classical soil mechanics advanced from empirical to a science basis, after Karl
Terzaghi, in 1936 developed the concept of effective stress. Conventionally, the
differences between saturated and unsaturated states of soil are well explained using
the phenomena of vadose zone region and water retention around soil porous system
i.e. adhesive, cohesive and capillary pressures (Figure 2.1). Currently, applied soil
mechanics is categorized within a horizontal line representing the GWT. Below GWT
(+PWP) is considered positive, hence the soil is located on generally saturated zone.
Aloft the GWT, the PWP is negative in reference to atmospheric pressure. The entire
soil region above GWT is called vadose zone, above the GWT this zone is designated
as capillary fringe, where degree of saturation (Sr), approaches 100%. This zone ranges
from > 1m to approximately 10m in thickness, depending on soil type (Fredlund,1996).
Water phase is deduced to be continuous within the capillary region, while air phase is
regarded unsteady. Above this capillary region, two-distinct region is identified in which
air and water phases are conceptualised as continuous.
Within this region, degree of saturation fluctuates ranging from 20% to 90%, depending
on soil type. Above this two-distinct region, drier soil emerged and the water phase is

Page | 6

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

considered discontinuous. Whereas, the air phase remains continuous, as both the
saturated and unsaturated zones are influenced by climatic feature i.e. evaporation,
evapotranspiration and precipitation.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏: Hydrologic system of unsaturated geotechnics (Fredlund, 1996)

The unsaturated zone is characterised by -PWP that is called suction and there are
other terms synonymous to suction such as capillary suction, matric suction, capillary
water stress, pore water tension, soil moisture deficiency, capillary potential, soil water
pressure deficiency, soil water free energy and soil moisture tension. Moving towards
zero atmospheric pressure near soil surface, PWP and increasing desiccation becomes
highly negative. Curve liquid bridges can be clearly observed linking soil particles and
this curvature at resultantly difference pressure, between air and water phase. The
characteristics of the air-water interface (AWI) possess an important bearing on the
hydro-mechanical characteristic of unsaturated soils. The saturated and the air-dried
soil is made up of two phases, i.e., fluid in the voids (e.g., water or air). While, the soil
structure includes other pursuant principles of unsaturated soil mechanics. Unsaturated
soils have more than two phases, i.e., soil structure, water, and air according to Fredlund
et al. (2012). Though, contract skin is another component phase of unsaturated soil as
the air phase is continuous, AWI interacts with soil particles and provides an influence
on the soils geotechnical behavior.

Page | 7

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟐: Air water soil interphase (Fredlund et. al 2012)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟑: Phase diagrams (a)Fourphase unsaturated soil system (b)Simplified three

phase diagrams (Fredlund and Rahardjo, 1993)

Additionally, unsaturated soil is perceived as a four-phase system due to unique role of

AWI on soil behaviour. The AWI portrayed a thin membrane interlinked through soil
voids, forming a fixed layer within the air-water phases. Stress state varies around AWI
causing changes in moisture content, volume and shear strength of unsaturated soils.
Contractile skin forms barrier around air-water interphase and determines the air to the
volume ratio in soil voids.

Page | 8

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

2.2.1 Capillarity
Basically, to understand the physical phenomenon of how negative pore-water pressure
(-PWP) influences the effective stress (𝜎 ′ ) in unsaturated soil. The principle of capillary
forces, is necessary as capillarity is consistence with matric suction. When water level
increases in the capillary tube, radius of curvature of AWI directly affects soil moisture
content and matric suction, (𝜓𝑚 ). Though rise in capillary differs from soil wetting-drying
processes due to soil void discrepancy. More commonly, water increases in the capillary
tube due to surface tension (𝑇𝑠 ) and water tendency, to wet the surface of capillary tube.
Furthermore, capillary activities could be analysed by considering 𝑇𝑠 revolving within
meniscus circumference. The 𝑇𝑠 acts at an angle 𝛼1 vertically to the capillary tube and
this angle is called contact angle. The magnitude of capillary tube is controlled by
adhesion between the molecules in the AWI. The perpendicular component of 𝑇𝑠 ,
governs holding of water column weight, which has a height, ℎ𝑐 as expressed
mathematically :

2𝜋𝑟𝑇𝑠 𝐶𝑜𝑠𝛼1 = 𝜋𝑟 2 ℎ𝑐 𝜌𝑤 𝑔 (2.1)

Where:
r = radius of the capillary tube
Ts = surface tension of water
α1 = contact angle
ℎ𝑐 = capillary heightcontact angle
g = gravitional acceleration

When rearranged, Equation 2.2 could give the maximum water elevation in the capillary
tube, ℎ𝑐 :
2𝑇𝑠
ℎ𝑐 = 𝜌 (2.2)
𝑤 𝑔𝑅𝑠

Page | 9

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟒: Capillarity model, (Fredlund and Rahardjo, 1993)

This factor contributes to soil water rise beyond water table and the pores in the soil,
portrays tortuous capillary tubes with varying tube diameters. This is termed capillarity
model. At the top of capillary bore, where the elastic film exists (contractile skin), the
pressure difference (∆P) over the film is demonstrated by the Young – Laplace equation:

∆𝑃 = (𝑢𝑎 − 𝑢𝑤 ) = 2𝑇𝑠 (1⁄𝑟) (2.3)

Where:
Ts = surface tension of water
r = radius of the capillary meniscus

Lu and Likos, (2004) presented mechanical equilibrium diagram for capillary rise in a
small diameter tube 𝑢𝑎 − 𝑢𝑤 acting within meniscus region, and the vertical projection
of 𝑇𝑠 acting over meniscus circumference lead to correlation in Equation 2.4:

𝜋
(𝑢𝑎 − 𝑢𝑤 ) 𝑑 2 = 𝑇𝑠 𝜋𝑑𝐶𝑜𝑠𝛼 (2.4)
4

It is evidenced that the term 𝑢𝑎 − 𝑢𝑤 , is equivalent to matric suction in Equation 2.4. Lu

and Likos suggested that matric suction, depends on the tube diameter, relative to soil
pore sizes.

Page | 10

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

This means that smaller the soil pores diameter, the larger the matric suction (𝜓𝑚 )
values. Thus, this concept is important when in geotechnical engineering as matric
suction is a state stress variables achieved by difference soils. The effect of -PWP and
𝑇𝑠 gives rise to suction stress in soils under unsaturated condition. Suction stress
denotes as free interparticle force gained around a matric of unsaturated soil particles
due to combined effects of -PWP and 𝑇𝑠 , which occurs at the pore water-air-soil grain
interface. As suction pressure tends to pull soil particles towards each other and this
gave rise to Kelvin equation:

∆𝑃 = −311𝑙𝑜𝑔10 𝐻 (2.5)

Where:
Ts = surface tension of water
r = radius of the capillary meniscus
H = relative humidiy of the pore air above the meniscus

Substituting the suction matric (𝑢𝑎 − 𝑢𝑤 ) for ∆𝑃 in Equation 2.4 enables it to be

demonstrated as a function of the relative humidity:

(𝑢𝑎 − 𝑢𝑤 ) = −311𝑙𝑜𝑔10 𝐻 (2.5)

2.2.2 Soil potential energy

Soil suction is described as soil potential for water attraction. This is the key variable
that regulates hydromechanical principles of unsaturated soils. Primarily, soil suction is
the fundamental factors of unsaturated soils that is categorized into two parts i.e.
osmotic suction ( 𝜓𝑜 ) and matric suction (𝜓𝑚 ). The summation of these two suction
components is known as total suction, mathematically it is expressed as:

𝜓𝑡 = 𝜓𝑚 + 𝜓𝑜 (2.6)

However, ‘Total Suction’, is considered as the potential energy of water in the soil and
it describes potential thermodynamic difference (𝑢𝑤 ) between the soil compared to free
water. The thermodynamic potential of the soil 𝑢𝑤 is decreased by capillarity effects,
short-range adsorption, and effect of dissolved salts.
Page | 11

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Thus, matric suction is triggered by short-range adsorption effects and capillarity; while
osmotic suction is caused by dissolved salts effect. Osmotic suction is only present in
marine and leached soils, because of short-range adsorption effects are only prominent
at low water contents when the adsorbed water is mainly in the form of thin films around
the soil particles. Basically, the soil potential energy, 𝑢𝑤 is the mathematical expression
of the potentials different:

∆𝜇𝑡 = ∆𝜇𝑐 + ∆𝜇0 + ∆𝜇𝑒 + ∆𝜇𝑓 (2.7)

Where:
∆𝜇𝑐 = change due to curvature of the air − water interface (capillarity)
∆𝜇0 = change due to osmotic effects (dissolved solutes)
∆𝜇𝑒 = change due to electric field
∆𝜇𝑓 = change due to Van der Waals forces

Excluding dissolved solutes effects, all the other terms added up represent matric
suction. While, all the terms represent a negative value, matric suction value is positive
because it constitutes free water state potential change. Nam et. al, (2009) reported that
matric potential contributes the highest potential of pore water during suction
measurement. Matric suction is generally defined as (𝑢𝑎 − 𝑢𝑤 ), and this stands for the
amount of -PWP in unsaturated soils.
Fredlund and Rahardjo, (1993) developed a relationship that represented total suction
as a function of partial vapour pressure (PVP) of the pore water, using the fundamentals
of total suction and thermodynamic of the soil pore water:

−𝑅𝑇 𝑢𝑣
𝜓𝑇 = 𝑙𝑛 ( ) (2.8)
𝑣𝑤0 𝑤𝑣 𝑢𝑣0

Where:
R = universal gas constant (J/mol K)
T = absolute temperature (K)
vw0 = specific volume of water (m3 /k)
wv = molecular mass of water vapour (g/mol)
uv = partial pressure of pore − water (kPa)
uv0 = saturation pressure of water (kPa)
Page | 12

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝑢
Note, the term ( 𝑣⁄𝑢𝑣0 ) stands for measured relative humidity (RH). Therefore, relative
humidity of pore water vapor, can be employed to measure suction.

2.2.3 Matric suction

Matric suction (𝑢𝑎 − 𝑢𝑤 ), is relatively determined as the different between partial
pressure of water vapour in equilibrium with the soil water. Matric suction results through
capillarity and surface absorptive forces of the soil.

𝜓𝑚 = (𝑢𝑎 − 𝑢𝑤 ) (2.9)

2.2.4 Osmotic suction

Osmotic suction (𝜓𝑜 ) is the expression of amount of dissolved salts in the pore fluid.
Therefore, matric suction is caused by capillary pressure in soil structure, thus its relates
to mechano-chemical interactions between soil minerals and 𝑢𝑤 (Wan et. al, 1995).
However, matric suction is of high interest, as many pavement problems related to
unsaturated soil due to variation in moisture content, which principally influence soil
matric suction (Nishimura et. al 2007). Osmotic changes are less significant in
geotechnical engineering, because change in total suction, (𝜓𝑇 ) is equivalent to different
in matric suction i.e.

∆𝜓 = ∆(𝑢𝑎 − 𝑢𝑤 ) (2.10)

Furthermore, mechanics of unsaturated soil is considered as a three-phase system that

is made up of pore-air, pore-water and granular solid particles. Matric suction in such
system made-up off two parts: the hydration forces and capillary forces that arises from
capillary activities attributed to interactions within air-water menisci that are originated
from soil particles and 𝑇𝑠 . The two forces are available in high clays content soil, such
as bentonite (Pusch and Yong, 2003; Arifin and Schanz, 2009). In case of sands and
low plastic clays, suction matric is associated to capillary pressure (Fredlund and
Rahardjo,1993).

Page | 13

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

2.3 Soil-Water Retention Curve (SWRC)

Thermodynamic potential decrease of pore water is associated with degree of pore
water present in the soil and a correlation exists within soil suction and moisture content
of the soil. This link is described by the Soil Water Retention Curve (SWRC). The SWRC
consists of different suction ranges, where the main moisture holding mechanism is
different and differs by soil type. Various researchers have successfully applied axis-
translation technique, to study SWRC of unsaturated soils (Ng and Pang, 2000a, b), by
considering volume change and shear strength characteristic of unsaturated soils (Ng
and Chiu, 2001; Ng and Chiu, 2003a, b; Zhan, 2003; Ng and Zhou, 2005). Therefore,
SWRC is relationship between suction and the corresponding soil wetness. Moisture
quantity, could be gravimetric water content, w, volumetric water content, θ, or degree
of saturation (Sr). SWRC could also be called capillary pressure curve (CPC). The
SWRC categorises soil behaviour into three distinct classes of desaturation (Figure 2.5).
The classes of desaturation are designated as the "boundary effect class" at a very low
suction of soil, the "transition class" at intermediate soil suction, and the "residual class"
at a great soil suction that extend to 1,000,000 kPa (Fredlund, 2006).

𝐅𝐢𝐠𝐮𝐫𝐞: 𝟐. 𝟓: Soil − water retention curve (Fredlund and Xing, 1994)

The hysteresis in the SWRC is triggered by non-uniformity of pore-size distribution in

the soil. During the wetting and drying process, the soil water-content varies at any
particular matric suction (Fredlund and Rahardjo, 1993). Figure 2.5 above, illustrated
that the end-point of adsorption curve differs from the start-point of the curve; due to

Page | 14

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

hedged air in the soil (Fredlund and Xing, 1994). The slope steepness over a range of
soil suctions is representative of the soil storage potential (Leong and Rahardjo, 1997).
Whereas, the difference between soil matric suction and moisture content is presences
soil’s adsorption potential. In addition, soil subjected to wetting may never reach full
saturation due to the entrapped air bubbles (Pham, 2005). Therefore, hysteresis effect
exists for SWRCs for two major reasons and this include:
I. Hysteresis is mainly present in the capillary regime, which explains the importance
of the pore-size distribution effect.
II. Changes in geometry of the pore-size distribution (Lu and Likos, 2004).

2.3.1 SWRC models

The most widely utilised models were formulated by (Fredlund and Xing, 1994), as
presented in Equations 2.11 and 2.12. Van Genuchten, (1980), proposed Equation 2.13,
though the models was used to describe a SWRC exist in 3 or 4 parameter forms. These
models allow analytical correlation with soil suction and volumetric water content.
Moreover, the sigmoidal shape of this model, replicates that of particle size distribution
(PSD) curve, as it was originally derived based on relationships from the pore-size
distribution of soils.
𝜃𝑠
𝜃𝑤 = (2.11)
(ln(𝑒+𝜓⁄𝑎 )𝑛 ))𝑚

𝜓
ln(1+ ) 𝜃
𝜓𝑟
𝜃𝑤 = [1 − 1,000,000 ] (ln(𝑒+𝜓⁄𝑠 𝑎)𝑛 ))𝑚 (2.12)
𝑙𝑛(1+ 𝜓 )
𝑟

𝜃𝑠 −𝜃𝑟
𝜃𝑤 = 𝜃𝑟 + (2.13)
[1+(𝜓⁄𝑎)𝑛 ]𝑚

Where:
θw = volumetric water content
θs = saturated water content
θr = residual water content
ψ = suction matric/ soil moisture deficiency
ψr = residual suction
a, n and m = fitting parameters
Page | 15

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Leong and Rahardjo, (1997) suggested that ‘a’ in the three models presented above, is
related to suction at inflection point of the curve (i.e., air-entry value), parameter ‘𝑛’
affects the gradient of the curve in the desaturation zone, and the parameter ‘𝑚’ affects
the symmetry of the slope of the curve about the inflection point. The degree of
saturation (𝑆𝑟 ) or gravimetric water content (𝑤) or volumetric water content (𝜃𝑤 ) are
used to illustrates the SWRC. Mathematically, the relations between volumetric water
content 𝜃𝑤 , gravimetric water content, 𝑤, and degree of saturation, Sr are given by the
relation below:
For saturated soil condition:
𝑉𝑤
𝜃𝑤 =
𝑉𝑤 + 𝑉𝑠

𝑀𝑤 𝑉𝑣 𝜌𝑤 𝑉𝑤 𝑒
𝑊= = = = , 𝑒 = 𝑤𝐺𝑠
𝑀𝑠 𝐺𝑠 𝜌𝑤 𝑉𝑠 𝐺𝑠 𝑉𝑠 𝐺𝑠

𝑆𝑟 𝑒 𝑊𝐺𝑠
𝜃𝑤 = = (2.15)
1+𝑒 1+𝑤𝐺𝑠

For not saturated soil condition:

wGs
e=
Sr

wGS Sr wGs
θw = wGs = (2.16)
1+ ⁄S Sr +wGS
r
Where:
e = void ratio
Gs = specific gravity
ρw = density of water
Ms = mass of soil solids
Mw = mass of water

Equations 2.15 and 2.16 are useful theoretical relationships. For practical purposes,
SWRCs is always measured in line with gravimetric water content, 𝑤 non-recoverable
energy consumed by compressed soil due to hysteresis (Blight, 2013).

Page | 16

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Conclusively, SWRC can vary according to a number of factors and these includes
hydraulic hysteresis, transient nature of the pore size distribution of a deformable soil.
The account of these factors should be considered, during modelling of SWRC.

2.4 Unsaturated Soil Shear Strength

Bishop, (1959) propounded an equation to evaluate shear strength of unsaturated soil,
while exploring effective stress concept, as stated in Equation 2.17 below:

𝜏 = 𝑐 ′ + (𝜎 − 𝑢𝑎 )𝑡𝑎𝑛𝜙 ′ + (𝑢𝑎 − 𝑢𝑤 )𝜒𝑡𝑎𝑛𝜙 ′ (2.17)

Where:
τ = shear strength of unsaturated soil,
c′ = effective cohesion,
ϕ′ = effective angle of friction
(σ − 𝑢𝑎 ) = net normal stress,
(𝑢𝑎 − 𝑢𝑎 ) = is matric suction; and χ is parameter dependent on the
degree of saturation (between 0 and 1).

Fredlund et al. (1978) extended the shear strength equation of saturated soil, by
suggesting an equation that demonstrates the shear strength of unsaturated soils.

τ = 𝑐 ′ + (𝜎𝑛 − 𝑢𝑎 )𝑡𝑎𝑛𝜙 ′ + (𝑢𝑎 − 𝑢𝑤 )𝑡𝑎𝑛𝜙 𝑏 (2.18)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟔: Normal and shear stresses in unsaturated geotechnics:

Page | 17

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

(I) independent stress state variable, (Fredlund 1993)

(𝐼𝐼) 𝑒𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑠𝑡𝑟𝑒𝑠𝑠 𝑎𝑝𝑝𝑟𝑜𝑎𝑐ℎ (𝐿𝑢 𝑎𝑛𝑑 𝐿𝑖𝑘𝑜𝑠, 2004)

Furthermore, the principal concept of unsaturated soil mechanics can be grouped into
three categories: (i) effective stress approach by (Bishop,1960), (Bishop and
Blight,1967), (ii) independent stress variable approach of (Fredlund and Morgenstern,
1977) and (Fredlund et. al, 1978) and (iii) suction stress characteristic curve (Lu and
Likos, 2006) and (Lu et. al, 2010).

2.4.1 Effective stress approach:

Shear strength of soil is the resistance a soil needed to sustain, before slide on a failure
plane (Das, 2006). Shear strength measurements are based on Mohr’s theory of
material rupture. The theory states that material failure occurs at a critical combination
of shear (𝜏) and normal (𝜎𝑛 ) stresses. Thus, shear strength at failure (𝜏𝑓 ) could be
demonstrated with Mohr-Coulomb law as:

𝜏𝑓 = 𝑐 ′ + 𝜎𝑛 𝑡𝑎𝑛𝜙 ′ (2.19)

Where:
c′ = cohesion of the soil
ϕ′ = angle of internal friction

Equation 2.19 was later modified by (Terzaghi, 1936), through the incorporation of PWP
in saturated soils. It was argued that total normal stress, 𝜎 ′ is summation of stress
supported by soil, and the stress due to PWP (𝑢𝑤 ). Thus, shear stress is defined as the
stress carried by soils, as effective stress(𝜎 ′ = 𝜎𝑛 − 𝑢𝑤 ) and Mohr-Coulomb formulation
for saturated soils is stated as:

𝜏𝑓 = 𝑐 ′ + (𝜎𝑛 − 𝑢𝑤 )𝑡𝑎𝑛𝜙 ′ (2.20)

This equation was further revised by (Bishop et. al, 1960) for unsaturated soil, as to
account for suction effects, (𝑢𝑎 − 𝑢𝑤 ) in the soil. They suggested that, since unsaturated
soil is a three-phase system (solid, pore water, and pore air) and water in voids is not
continuous.

Page | 18

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The total stress will be the sum of intergranular stress, the pore air pressure, 𝑢𝑎 and
pore water pressure, 𝑢𝑤 and they further stated that effective stress(𝜎 ′ ) in unsaturated
soils could be demonstrated as:

𝜎 ′ = (𝜎𝑛 − 𝑢𝑎 ) + 𝜒 (𝑢𝑎 − 𝑢𝑤 ) (2.21)

Where:
χ = fractional cross − sectional area of the water in voids, For dry soil, χ
= 0 and for saturated soil χ = 1.
Thus, Equation 2.22 is for Mohr-Coulomb relationship shear strength for unsaturated
soils:
𝜏𝑓 = 𝑐 ′ + [(𝜎𝑛 − 𝑢𝑎 ) + 𝜒(𝑢𝑎 − 𝑢𝑤 )] + 𝑡𝑎𝑛∅′ (2.22)

Jennings and Burland, (1962) explored the limitations in the use of effective stress
principles and discovered that it may not be adequate for description of collapse
behaviour of soils. Also, the material parameter 𝜒 includes single valued effective stress
equation, which leads to difficulties both in theory and its measurement. The material
parameter 𝜒 depends on the soil mineralogy and stress path (Khalili, N. et al, 2004;
Alsherif, N.A et. al, 2014; Baille, W. et.al 2014). Similarly, other researchers presented
the relationships in Equation 2.20 and all pointed out to the challenges of using effective
stress concept due to the error in prediction of 𝜒 value.

Bishop’s approach received wide criticism concerning the use of 𝜒 which some
researchers call an elusive parameter. Coleman, (1962) argued that 𝜒 is associated with
soil structure and no correlation can be found between 𝜒 and volumetric parameter such
as the saturation degree. Fredlund, et al. (1978) presented suction as an independent
state variable, but prediction of shear strength from the concept of effective stress is
rarely utilised. Basically, efforts were made by (Khalili and Khabbaz,1998) and (Khalili
et .al, 2004), when shear strength data from 17 studies were used (including that of
Fredlund and his associates) for the evaluation of 𝜒 value and this showed that 𝜒 value
was uniquely correlated with suction ratio and air-entry value.

Page | 19

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

(u −u ) η
χ = [(u a−u w) ] (2.23)
a w b

Where:
χ = effective stress coefficient
(ua − uw ) = matrix suction in the soil samples at failure conditions
(ua − uw )b = soil air − entry value
η = −0.55. which is the correlation coefficient.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟕: Effective stress parameter versus suction ration (from Khalili and Khabbaz, 1998)

Khalili et al. (2004) additionally reviewed previous studies from other researchers and
concluded that shear strength can be predicted using effective stress concept. These
scholars further proposed a sequential progression form of effective stress parameter
to account for suction effects. They further suggested that for suctions > air entry values,
𝜒 values should vary as demonstrated in Equation 2.23. whereas, for suction < air entry,
𝜒 value was equal to 1.0. These scholars tested their concept of sequential progression
for effective stress on shear strength data from (Cui and Delage,1996, Maatouk et. al,
1995, and Geiser, et. al 2000), and volume change data from (Fleureau et. al, 1993)
observed a good curve of fit between the measured and predicted values in all cases.
The effective stress 𝜎 ′ controls stiffness and strength of soil and its application is
important in design of pavement and other geotechnical structure.

Page | 20

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Basically, three phases presences changes in equilibrium equation, as relative air and
water pressure also contribute to the behaviour. Therefore, evaluation of strength
parameter in soils, modified effective stresses equation for unsaturated soil. These
equivalent effective stresses equation for unsaturated soils requires independent stress
variables, which might be computed combining total stress, PWP, and pore air pressure,
(PAP). The commonly used stress variables are formed by the net normal stress (𝜎 −
𝑢𝑎 ) and matric suction (𝑢𝑎 − 𝑢𝑤 ) (Fredlund and Rahardjo, 1993). Clearly, matric suction
is one of the variables, because it increases the existing forces at interparticle points
contact due to capillary pressure presence. Variable, like net normal stress, uses the air
pressure as a reference which is almost constant if taken as the atmospheric pressure.

𝜎 ′ = (𝜎⏟− 𝑢𝑎 ) + 𝜒(𝑢𝑎 − 𝑢𝑤 ) Unsaturated soils (2.24)

where 𝜒 = zero for dry condition and one for full saturation. It is necessary to state that
the impact of suction matric and net stresses induced by external loads at particle
contact points are uncoupled. Thus, the stress variables must be independent (Vinale
et.al, 2001; Cho and Santamarina, 2001). Although the Bishop’s single tensor equation
is often used and it summed menisci water pressure and total stress effects, but it has
several limitations because it mixes local and global conditions within the medium. it is
better to present stress-strain results in terms of the two state variables, to avoid
challenges: net pressure(𝜎 − 𝑢𝑎 ) and suction (𝑢𝑎 − 𝑢𝑤 ). Equation 2.25 is applicable to
pure water only as the presence of soluble will add another term to suction. Osmotic
suction 𝜋 is also important in soil systems (Tindall and Kunkel, 1999):

𝜋 = 𝐾𝑇Δ𝑐 (2.25)
Where:
K = Boltzmann’s constant
T = absolute temperature
Δ_c = chemical concentration difference across a semipermeable membrane

Aitchison, (1961) proposed the equation:

𝜎 ′ = 𝜎 + 𝜓𝑝′ (2.26)

Page | 21

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where:
p′ = pressure deficiency = (ua − uw )

The Aitchison equation is a special case of Bishop equation, when the air pressure is
zero (atmospheric). Whereas, (Jennings, 1961) proposed an equation similar to that of
Aitchison using a different symbol, β, in place of either 𝜒 or 𝜓. This equation was
formulated in 1958, though it was made public in 1960s:

2.4.2 Independent state stress variable approach

This approach was recommended by Fredlund and associates in a series of publications
(Fredlund and Morgenstern, 1977; and Fredlund et al., 1978; Fredlund and Rahardjo,
1993; Fredlund, 1996). These researchers demonstrated that stress state of
unsaturated soil can be stated using any two of the three-possible stress variables
correlation namely: total normal stress(𝜎), pore air pressure (𝑢𝑎 ), and pore water
pressure (𝑢𝑤 ). Possible combinations are: (𝜎 − 𝑢𝑎 ) and (𝑢𝑎 − 𝑢𝑤 ), (𝜎 − 𝑢𝑤 ) and (𝑢𝑎 −
𝑢𝑤 ) and (𝜎 − 𝑢𝑎 ) and (𝜎 − 𝑢𝑤 ). These researchers argued that (𝜎 − 𝑢𝑎 ) and (𝑢𝑎 −
𝑢𝑤 ), were the most benefited combination as only one stress state variable was affected
when pore water pressure changes. Using these combinations, the following correlation
was demonstrated for describing shear strength of unsaturated soils.

𝜏𝑓𝑓 = 𝐶 ′ + (𝜎𝑓 − 𝑢𝑎 )𝑓 𝑡𝑎𝑛∅′ + (𝑢𝑎 − 𝑢𝑤 )𝑓 𝑡𝑎𝑛∅𝑏 (2.27)

Where:
tan ∅𝑏 = angle indicating the rate of increase in shear strength with respect to a change
suction matric, (𝑢𝑎 − 𝑢𝑤 )𝑓 .
These researchers further observed that Mohr-Coulomb failure plot for saturated soil is
plotted in two dimensions, while the corresponding plot for unsaturated soil must be a
3-dimensional diagram. Equation 2.28 additionally expressed that since the intercept of
the failure envelop intersects the shear stress versus suction matric plane the correlation
between the shear stress versus matric suction, see Figure. 2.9.

𝑐 = 𝑐 ′ + (𝑢𝑎 − 𝑢𝑤 )𝑡𝑎𝑛∅𝑏 (2.28)

Page | 22

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where:
𝑐 = intercept of the Mohr-Coulomb failure envelope at specific suction and zero net
normal stress.

𝐅𝐢𝐠𝐮𝐫𝐞𝟐. 𝟖: Unsaturated soil Mohr − Coulomb failure envelope (Fredlund 1993)

The pair of two profitable stress state variables formed was (𝜎 − 𝑢𝑎 ) and (𝑢𝑎 − 𝑢𝑤 ) and
it was applied in formulation of constitutive models describing strength and deformation
of unsaturated soils. Muraleetharan and Wei, (2000) develop sets of governing
equations for unsaturated porous media. These equations stand on the theory of
Interfaces that explicitly considers the interfacial effects and provides a theoretical basis
for the use of two independent stress variables. Generally, stress tensors are required
to illustrate three-dimensional state of stress in soils:

𝜎𝑥 − 𝑢𝑎 𝜏𝑦𝑥 𝜏𝑧𝑥
𝜎 − 𝑢𝑎 = [ 𝜏𝑥𝑦 𝜎𝑦 − 𝑢𝑎 𝜏𝑧𝑦 ] (2.29)
𝜏𝑥𝑧 𝜏𝑦𝑧 𝜎𝑧 − 𝑢𝑎

𝑢𝑎 − 𝑢𝑤 0 0
𝑢𝑎 − 𝑢𝑤 = [ 0 𝑢𝑎 − 𝑢𝑤 0 ] (2.30)
0 0 𝑢𝑎 − 𝑢𝑤

Some boundary conditions exist, accurate components value for state stress is limited
as demonstrated by the inequality: 𝜎 ≥ 𝑢𝑎 ≥ 𝑢𝑤 .
In the condition that, air pressure must exceed total pressure, the solid particles would
not intercept the soil at the extreme, but rather deform or shear. In fact, this restriction
Page | 23

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

is often applicable in pressure plate apparatus and specimen surrounded by a flexible

membrane, could be considered and this makes air pressure to acts as the total
pressure. The water pressure can rise until it equates the air pressure. In this case, the
suction matric will be zero and the soil condition is fully saturated. Additionally, the water
pressure cannot exceed the external pressure. This case is studied in classical soil
mechanics.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟗: Two dimensional projections of failure envelopes at various suctions (Fredlund 1993)

Equation 2.28 implies that shear strength is linearly a function of suction. However,
several equations demonstrated this correlation a non-liner parameter. Furthermore,
this non-linearity is somewhat similar to the variation in 𝜒 as a function of saturation
degree in effective stress concept. Different types of revised equations have been
submitted and some of these revised equations included the same parameters to that
of degree of saturation (Vanapalli et al., 1996; Oberg and Sallfours, 1997). Whereas
other revised equation deal with inclusion of parameters related to suction (Abramento
and Carvalho, 1989; Rassamand and Cook, 2002).

Oberg and Sallfours (1997):

𝜏𝑓 = 𝑐 ′ + (𝜎𝑛 − 𝜇𝑎 )𝑡𝑎𝑛∅′ + (𝑆)(𝜇𝑎 − 𝜇𝑤 )𝑡𝑎𝑛∅𝑏 (2.31)

Fredlund et al (1996) and Vanapalli et al (1996):

𝜏𝑓 = 𝑐 ′ (𝜎𝑛 − 𝜇𝑎 )𝑡𝑎𝑛∅′ + (𝛩𝑘 )(𝜇𝑎 − 𝜇𝑤 )𝑡𝑎𝑛∅𝑏 (2.32)

Page | 24

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

(Abramento and Carvalho,1989)

𝜏𝑓 = 𝑐 ′ + (𝜎𝑛 − 𝜇𝑎 )𝑡𝑎𝑛∅′ + 𝛼(𝜇𝑎 − 𝜇𝑤 )𝛽 (2.33)

(Rassam and Cook, 2002)

𝜏𝑓 = 𝑐 ′ + (𝜎𝑛 − 𝜇𝑎 )𝑡𝑎𝑛∅′ + (𝜇𝑎 − 𝜇𝑤 )𝑡𝑎𝑛∅ − 𝜑((𝜇𝑎 − 𝜇𝑤 ) − (𝜇𝑎 − 𝜇𝑤 )𝛽 (2.34)

(μ −μ )tan∅−τsr
φ = ((μ a−μ w)−(μ β (2.35)
a w a −μw )

tan∅((μa −μw )r −(μa −μw )b )

β= (2.36)
(μa −μw )r tan∅−τsr

Where:
𝑆𝑟 = degree of saturation
Θ = normalised water content between saturation and residual water content;
α, k, β are fitting constants.
τ𝑠𝑟 = shear strength at residual suction.

Oberg and Sallfours,(1997) included saturation degree in Equations 2.31 and 2.32 while
Fredlund et. al, (1996) and Vanapalli et. al (1996) equation was on a conceptual basis
like Bishop’s formulation. These authors also argued that since water is present only in
a fraction of cross-sectional area in unsaturated soils, the effect of soil suction needs to
be reduced by saturation degree. This provides a mechanism to account for the variation
in SWRC between soil types. The fitting factor “𝑘” in Equation 2.32 was based on the
best fit of experimental data.
Nonetheless, this approach was supported by null-type triaxial tests, which showed
unsaturated soil volume specimen remains constant while decreasing and increasing
the stress state variables by an equal amount. Very small discrepancies in the specimen
volume under null-type test condition prove the applicability of stress state variables in
unsaturated soils.
Through the concept, for solving unsaturated soils mechanics problems using a realistic
approach was established. Although, effective stress principles have proved workable
in explaining stress and volume changes that occur in saturated soils due to differences
in the applied external load. Thus, extending these principles to unsaturated soils has
Page | 25

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

been problematic because of the presence of pore fluid consisting of two phases; air
and water.
Sridharan and Venkatappa Rao, (1973) asserted that osmotic suction plays part in
illustration of soil behaviour. This implies that current formulations of the effective stress
equation do not account for osmotic suction.
Allam and Sridharan, (1987) brought modifications to the effective stress and stress
state variables approach to include osmotic suction and the effect of air-water interface.
The modification, showed that high salt content contributes to a high suction which in
turn greatly influences the physical and volumetric changes of soil according to the
studies conducted by (Noorany, 1984, Feng et al., 2003).

𝜎 ′ = (𝜎 − 𝑢𝑎 ) +  (𝑢𝑎 − 𝑢𝑤 ) (2.37)
𝑤

Equation 2.37 was used by (Khalili and Khabbaz, 1998; Zienkiewicz et .al, 1999 and
Gallipoli et. al, 2003), but other researchers like (Muraleetharan and Wei, 2000)
indicated that Equation 2.38 is valid under certain conditions. This is because 
𝑤

primarily depends on degree of saturation 𝑆𝜏 , soil type, compaction procedures and

stress path. This parameter is equal to the saturation for the completely dry and
saturated cases, which are governed by the same limits as saturation (0 ≤  ≤ 1).
𝑤

The relationship between 𝑆𝜏 and  of various soils was illustrated by (Blight, 1967).
𝜔

The relation in this range, is not quite explanatory, due to lack of data at very low
saturation. Thus, the consequence is very insignificant to unsaturated granular
subgrade. However, modelling at low saturation levels are improbable to pose error
under field conditions.

2.4.3 Suction stress characteristic curve (SSCC) approach

The third concept explains stress state of unsaturated soil is called the suction stress
characteristic curve approach. Lu & Likos, (2006) proposed a form of suction stress that
is similar to Terzaghi’s effective stress for saturated soils (Terzaghi, 1943) and Bishop’s
effective stress for unsaturated soil (Bishop, 1954).
This concept aimed at proposing a single stress variable that can model hydro-
mechanical response of soil. Forces such as the van der Waals, double layer forces,
surface tension and adhesive forces are interactive within soil solid surface. These
Page | 26

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

forces generate energy that triggers suction stress variables. Consequently, suction
stress characteristic curve approach involves thermodynamic method and this approach
is better than the both effect stress and independent state variable approach, due to
these reasons:
(i) Suction stress is solely a function of soil suction and therefore does not
require that the effective stress coefficient 𝜒 be used to define effective stress.
(ii) The SSCC is similar to the soil water characteristic curve, so a single valued
function is not required.
(iii) Hysteresis could also be conveniently handled in the SSCC

The effective stress equation by the SSCC is expressed as:

𝜎 ′ = ( 𝜎 − 𝑢𝑎 ) − 𝜎 𝑠 (2.38)
Where:
𝑢𝑎 = pore air pressure
σ = total stress,
σ′ = effective stress
𝜎 𝑠 = the SSCC of soil

Where:
𝜎 𝑠 = −(σ − 𝑢𝑎 )S and S = saturation proportion
(σ − 𝑢𝑎 ) = is the matric suction

Using functions of thermodynamic justifications (Lu et. al, 2010) also evaluated tensile
stress using virtual work by increasing the volume of the soil system with bound residual
water. They arrived at an expression for the SSCC as:

σs = −(ua − uw )Se for Vw > Vr (2.39)

Where:
Se = effective saturation
(ua − uw ) = suction matric
Vw = total water volume
Vr = residual water volume

Page | 27

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

From equation 2.38, (Lu et. al, 2010) suggested an effective stress equation as an
extension of Bishop’s equation and an expansion of Terzaghi’s equation for all
saturations by modifying the contribution to effective stress as:
σ′ = −(σ − uw ) − [−Se (ua − uw )] (2.40)
S−S
= (σ − ua ) − 1−Sr (ua − uw ) = (σ − ua ) − σs (2.41)
r

Where:

Sr = residual saturation

The above equation is different from Bishop’s equation with respect to the degree of
saturation, but can become Terzaghi’s effective stress equation, 𝜎 ′ = 𝑢𝑎 − 𝑢𝑤 , when it
is saturated. An additional extension could be carried out by applying the correlation,
linking degree of saturation and suction matric. Using (Van Genuchten, 1980) soil water
content curve model, the normalised saturation degree is expressed as:

1 1−1⁄𝑛
𝑆𝑒 = { } (2.42)
1+[𝛼(𝑢𝑎 −𝑢𝑤 )]𝑛

where 𝑛 and α are empirical fitting parameters of unsaturated soil properties, 𝑛 being
the pore size distribution parameter and α the inverse of the air entry pressure of water
saturated soil.
1
1 𝑛
𝑠𝑒
𝜎𝑠 = (𝑠𝑒 1−𝑛
− 1) 0 ≤ 𝑆𝑒 ≤ 1 (2.43)
𝛼

Similar suction stress closed-form equation for full range of matric suction is obtained
when equation (2.43) is substituted into equation (2.44) and eliminating the degree of
saturation giving equation 2.45 un saturated soils:

𝜎 𝑠 = −(𝑢𝑎 − 𝑢𝑤 ) 𝑢𝑎 − 𝑢𝑤 ≤ 0 (2.44)

1 1−1⁄𝑛
𝑠
𝜎 = {1+[𝛼(𝑢 𝑛 } 𝑢𝑎 − 𝑢𝑤 ≥ 0 (2.45)
𝑎 −𝑢𝑤 )]

Substituting equations (2.44) and (2.46) into (2.45) yields equation 2.47, for unsaturated
soils:

Page | 28

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝜎 ′ = 𝜎 − 𝑢𝑎 + ( 𝑢 𝑎 − 𝑢𝑤 ) 𝑢𝑎 − 𝑢𝑤 ≤ 0 (2.46)

(𝑢𝑎 −𝑢𝑤 ) 1−1⁄𝑛

𝜎 ′ = 𝜎 − 𝑢𝑎 + { )]𝑛
} 𝑢𝑎 − 𝑢𝑤 ≥ 0 (2.47)
1+[𝛼(𝑢𝑎 −𝑢𝑤

Because of scarcity of data for SWCC and SSCC for same soils Lu et al, 2010 used
existing data to validate this model. Soils, which have both the SWCC and SSCC their
data was validated by comparing them with equations 2.42 for SWCC and equations
2.45 for SSCC. Table 2.1 gives some characteristics of the used soils.

𝐓𝐚𝐛𝐥𝐞 𝟐. 𝟏: Soil properties used to validate the closed equation (Lu et. al. 2010)

In conclusion, it is evidenced that three schools of thought have been established with
regard to unsaturated soil behaviour. The revised effective stress approach was
attributed to Bishop, the two-independent stress approach was propounded by
(Fredlund and Morgenstern, 1977) and the SSCC, was suggested by (Lu and Likos,
2006) and (Lu et. al, 2010).

2.5 Previous Studies on Resilient Modulus

2.5.1 Overview
One of the functions of pavement structure, is to provide support at different layers. The
key factor in determining the thicknesses of other layers is the support specified through
the subgrade resilient modulus (𝑀𝑟 ). Generally, subgrade is elucidate as having an
infinite thickness and the support given by subgrade is commonly compute from
modulus of subgrade reaction (k). This reaction is `expressed as the sustained pressure
of the soil under a rigid plate at a specified settlement. However, pavement design has

Page | 29

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

evolved over time, with regards to quantifying brace provided by pavement layers, under
repeated loading rather than static loading condition. Formerly, plate load tests under
static loading is use to evaluate modulus of subgrade reaction and as well use to
quantify subgrade provided support. Although, pavements experience series of loading
as a result of vehicular traffic load, perhaps loading due to vehicular traffic can be fittingly
evaluated in the laboratory using repeated load triaxial test (RLTT) through
mathematical use of cyclic stress equations.

2.6 Resilient Modulus (𝑴𝒓 ) Concept

The key pavement design parameter, is 𝑀𝑟 and it is considered an important input
design property particularly during application of mechanistic-empirical pavement
design guide (MEPDG). Recently, National Cooperative Highway Research Program
(NCHRP) for flexible pavement design guide was developed in the united states and the
current South Africa pavement design methods (SAPDM) requires repeated load triaxial
testing to determine the 𝑀𝑟 for characterising subgrade soil (NCHRP, 2004, SANRAL,
2008). 𝑀𝑟 is illustrated as the ratio of maximum cyclical stress to elastic strain under
repeated loading.

𝜎𝑑
𝑀𝑟 = (2.48)
𝜀𝑟

Where:
Mr = resilient modulus (Mpa)
σd = σ1 − σ3 = deviator stress (kPa)
εr = ε2 − ε1 = recoverable strain (m)

Since pavement materials are repeatedly loaded and unloaded when subjected to cyclic
loading. Most pavements materials, like soil, exhibits both elastic and plastic behaviour
when subjected to loading and unloading. Plastic strain is permanent while elastic strain
is recoverable. This means that 𝑀𝑟 is analogous to the stiffness of subgrade under
repeated loading thus 𝑀𝑟 is determined with regards to recoverable strain (i.e., elastic
strain). Figure 2.11 shows the strain behaviour of a specimen subjected to repeated
axial loading. Initially, there is considerable plastic strain. However, the plastic strain
caused by each load cycle decreases as the number of cycles increases.

Page | 30

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟎: Subgrade resilient modulus behaviour (Hopkins et al. , 2004)

Notably, as the number of loading cycles increases, an accumulation of plastic strain

occurs (i.e., non-recoverable deformation). While 𝑀𝑟 is based just on the elastic strain,
pavements experiences both plastic and elastic strain under repeated loading. Plastic
deformation manifests itself in a pavement as rutting (permanent deformation) and is an
undesirable property that could trigger failure in pavement.
Generally, the larger the 𝑀𝑟 value the better the subgrade soil would be considered. A
large 𝑀𝑟 value could indicate that the subgrade can handle certain repeated loading
with little deformation (i.e. subgrade is stiff). In reality 𝑀𝑟 continually changes due to the
effects of moisture flux, therefore it becomes imperative to appropriately evaluate 𝑀𝑟
pavement with respect to matric suction. Darter et al., (1991) proved that a decrease in
𝑀𝑟 of pavement during its service life, results in increased deflection of the pavement
structure, which in turn causes fatigue and cracks.

2.6.1 Resilient Modulus Correlation With Soil Geotechnical Properties

Although, 𝑀𝑟 laboratory testing procedures is overly tricky, time-dependent and
somewhat costly. On that regard, attempts have been made to correlate 𝑀𝑟 with other
geotechnical properties, such as CBR and unconfined compressive strength (UCS) etc.
However, it is acknowledged that these reported correlations are empirical, whereas 𝑀𝑟
values are mechanistic parameter and it depends on soil stress despite the correlated
values been reasonably close to the measured values in most cases.

Page | 31

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Thus, this gave reasons for many correlations been proposed to evaluate 𝑀𝑟 based on
in-situ and laboratory testing results such as: Geogauge, and Light Falling Weight
Deflectometer (LFWD), Dynamic Cone Penetration (DCP) test. Among other, is the
laboratory civil engineering test method i.e. unconfined compressive strength (UCS) and
California Bearing Ratio (CBR), this method is widely utilized in correlating 𝑀𝑟 to obtain
a rational design values. These methods are acceptable and useful in obtaining
𝑀𝑟 values for subgrades in design practice. Whereas in research, direct laboratory
testing method is widely-accepted for measuring 𝑀𝑟 because of its accuracy and ability
to control multiple factors that directly affect 𝑀𝑟 . Laboratory evaluation of 𝑀𝑟 involves
conducting RLTT and this test is generally conducted in a triaxial cell environment on a
cylindrical either disturbed or undisturbed soil specimens. The main advantages of
laboratory RLTT is the capability to apply multiple stress states to soil specimen by
utilizing a combination of confining, deviatoric stresses among and recording the
number of cyclic sequences required for a soil to fail.
Ng et al. (2013) performed RLLT, to investigate 𝑀𝑟 of subgrade soil under various
stress and matric suction conditions. The results revealed that matric suction influenced
𝑀𝑟 of subgrade soil, as 𝑀𝑟 depend mainly on the soil’s stress states.
Ekblad and Isacsson (2008) presented experimental results from cyclic triaxial testing
at various moisture content, by keeping confining pressure constant on two different
graded granular materials. The result revealed increased moisture content caused a
reduction in 𝑀𝑟 and an increased in strain ratio. Therefore, it is important to evaluate
effects of gravimetric moisture content on 𝑀𝑟 .

2.6.2 Correlation between 𝑴𝒓 and unconfined compressive strength

The estimation of 𝑀𝑟 reported by Mallela et al. (2004) on lime-stabilized subgrade using
unconfined compressive strength (𝑞𝑢 ) tests and this is recommendation for 𝑀𝑟 highway
design. Mechanistic-Empirical Pavement Design Guide, Interim Edition: A Manual of
Practice (MEPDG) (AASHTO 2008) is in conformity with this Mallela et al. (2004)
approximation.

𝑀𝑟 (𝑀𝑃𝑎) = 0.124𝑞𝑢 (𝑘𝑃𝑎) + 68.8, (𝑅 2 = 0.46) (2.49)

Where:
q u = unconfined compressive strength (MPa)
Page | 32

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

This equation was developed through comparing shear strength (kPa) and a secant
modulus of elasticity 𝐸 (MPa) obtained from static unconsolidated-undrained (UU)
triaxial compression tests (Thompson 1966). Specifically, Equation 2.49 is based on 𝑞𝑢
values from specimens tested at zero confining stress and 𝐸 values from soil tested at
100 kPa confining stress. Thompson (1998) further performed three 𝑀𝑟 and 𝑞𝑢 tests
(per AASHTO 1994) on one A-7-6 soil mixed with 6% quicklime. The results were
identified as “duplicate” sets of three specimens in their referenced report. This test
results generally agreed with Thompson’s correlation for 𝑞𝑢 values obtained between
1,000 and 1,400 kPa. In order to employ the above equation for 𝑀𝑟 prediction,
unconfined compressive test should be conducted in an in-situ conditions. Furthermore,
the low value for coefficient of determination (COD), 𝑅 2 was as result of wide gap
between strength (kPa) and secant modulus of elasticity 𝐸 (MPa).
Lee et al (1997) suggested an alternative relationship to estimate 𝑀𝑟 with small strain
level and they proposed Equation 2.50 which has a strong COD and can be used for
laboratory compacted samples.

𝑀𝑟 (MPa) = 4.795 (𝑆𝑢1.0% ) -0.041(𝑆𝑢1.0% )2 (𝑅 2 = 0.97) (2.50)

Where:
S(u1.0%) = stress at 1.0% strain in the unconfined compression test
(strain rate is 1%/minute).
This concept of unconfined compression test was suggested as an option testing
method to the laboratory RLTT for 𝑀𝑟 values at the small strain levels Drumm et al. 1990
and Lee et al. 1997.

2.6.3 Correlation between 𝑴𝒓 and CBR empirical

The correlation of CBR to estimate in-situ 𝑀𝑟 of subgrade and granular materials, is
extensively used empirical equation by pavement engineers. Moreover, National
Cooperative Highway Program (NCHRP), AASHTO Guide (2008) and Austroads
pavement design guide (2012) equally recommended this correlation. Some of the
summarised correlations between 𝑀𝑟 and CBR are shown below.

Page | 33

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Table 2.2: 𝑀𝑟 -CBR empirical correlations

References Equation Comments
Power et .al (1984) Mr(psi)=2555(CBR)0.64 None

Hopkins (1994a) Mr (MPa)=17.914(CBR)0.874 None

Webb and Mr(psi)= 3116 (CBR)0.478 Mr showed no significant

Campbell (1986) change
within ± 1.5% of optimum
AASHTO moisture
Pavement Design Mr (MPa)= 17.6 (CBR)0.64 content.
Guide (2002) pavement sections (Lister
and
Powell 1987).

Valid from 5 x CBR to 20 x

CBR
(Sparks and Potter 1982).
Austroads Mr (MPa) = 10 (CBR) (For
Pavement Austroads 2012)
Design Guide
(2012)

Though, CBR is technically a strength index parameter and could not necessarily be
correlated with the modulus of reaction, k or stiffness, but correlation of CBR with 𝑀𝑟
has comprehensively been applied by some pavement engineer and highway agencies
to evaluate 𝑀𝑟 with outstanding results. Furthermore, CBR is stress independent while
𝑀𝑟 is basically a mechanistic parameter and it is represented as a repeated stress.
Notably, the correlation between 𝑀𝑟 and CBR is sensitive to soil mechanical properties,
i.e. percentage of fines contents; environmental factors, and stress level. Logically, there
should not good correlations between 𝑀𝑟 and CBR, despite the wide use of this
correlation among highway engineers.

Thompson and Robnett (1976) stated that developing a suitable correlation that can
exactly predict 𝑀𝑟 from CBR testing under certain experimental setup, could be
challenging. Thus, some experimental data has shown that 𝑀𝑟 values may differ over
a wide range of CBR value, which can lead to a rough calculation of 𝑀𝑟 value when it
is empirically correlated for design purpose.

Page | 34

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

2.6.4 Correlation of 𝑴𝒓 With Soil Physical Properties

Researchers like Thompson and Robnett (1979) proved that soil geotechnical properties
affect 𝑀𝑟 . Thus, the correlation of low plasticity index, liquid limit and specific gravity
resulted in lower 𝑀𝑟 value. In addition, Lekarp et al. (2000) recorded that 𝑀𝑟 decreases
with an increase in fines quantity. Janoo and Bayer II (2001) observed that an increased
particle size lead to an increase in 𝑀𝑟 value. Another observation by (Pezo and Hudson
1994), stated that the older the specimen, the lesser the resilient strain becomes. This
implies that 𝑀𝑟 of subgrade pavement will decrease with time. Majority of 𝑀𝑟 values
used for pavement design are not measured directly from laboratory test, but estimated
either from experience based on previous projects or from physical properties George
(2004). Nevertheless, AASHTO (2008) recommended estimation of 𝑀𝑟 values from soil
mechanical properties where direct measurement is not available. The good benefit of
this method is that, it provides a means of taking seasonal variations into account for
𝑀𝑟 calculation and these are reflected in the physical properties.
Jones and Witczak (1977) proposed two correlation equations for subgrade soils that
combines moisture content and degree of saturation. RLTT were performed at deviator
stresses of 6, 12, and 18 psi and confining stresses of 2, 4, 6, 8, and 12 psi for disturbed
and undisturbed samples. The correlation portrayed some shocking results; thus, it was
revealed that disturbed sample yielded reasonable 𝑀𝑟 value compared to undisturbed
sample.

For disturbed sample (𝑅 2 = 0.94)

Log(Mr ) = −0.1328(Mc) + 0.0134(S) + 2.319 (2.51)

For undisturbed sample (𝑅 2 = 0.45)

𝐿𝑜𝑔(𝑀𝑟 ) = −0.1111(𝑀𝑐) + 0.0217(𝑆) + 1.179 (2.52)

Where:
Mr = resilient modulus (ksi)
Mc = moisture (%)
S = degree of saturation (%)

Page | 35

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Carmichael and Stuart (1986) explored highway research information service database
and proposed a correlation that compute 𝑀𝑟 from mechanical properties, stress level
and soil classification according to the Unified Soil Classification system (USCS) as
follows. The correlated 𝑀𝑟 outputted fair values as compared to directly measured value.

Mr = 37.431 − 0.457(PI) − 0.618(Mc) − 0.1424(P200 ) + 0.179(σd ) + 36.722(CH)

+ 17.097(MH)
(2.53)
Where:
Mr = resilient modulus (psi)
PI = plasticity index (%)
P200 = percentage of passing #200 sieve (75m)
σd = deviator stress (psi)
CH = 1 for (CH)soil
= 0 for MH, ML and (CL)soil
MH = 1 for(MH)soil
= 0 for CH, ML and (CL)soil

An investigation was conducted on thirteen Wyoming fine-grain soils and correlation

was proposed to estimate 𝑀𝑟 from index properties and stress level Farrar and Turner
(1991). It was observed that 𝑀𝑟 was negatively correlated to degree of saturation, (S)
and positively correlated with 𝑃𝐼 and 𝑃200 as stated in Equation 2.54. This correlation
was vigorously interrogated, but they were able to justify the equation, by stating that
higher S will reduce the value of 𝑀𝑟 , thereby argued that degree of saturation should be
negatively correlated.

Mr = 34280 − 359(S) − 325(σd ) + 236(σ3 ) + 86(PI) + 107(P200 ) (2.54)

Where:
Mr = resilient modulus (psi)
S = degree of saturation (%)
σd = deviator stress (psi)
Page | 36

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝜎3 = confining stress (psi)

PI = plasticity index (%)
𝑃200 = percentage of passing #200 sieve (75 micron) (%)

Hall and Thompson (1994) suggested an equation that corelated 𝑀𝑟 with clay
percentage, PI and organic percentage content. The value of 𝑀𝑟 predicted from this
correlation outputted significant value closely compared to 𝑀𝑟 value measured directly.
Even though percentage of organic content was negatively correlated, but the limitation
of this equation is that it cannot be reasonably apply to inorganic soils.

Mr = 6.90 + 0.0064(C) + 0.216(PI) − 1.970(OC)(R2 = 0.76) (2.55)

Where:
𝑀𝑟 = resilient modulus at the AASHTO T − 99 optimum
Moisture content and 95% compaction (ksi)
C = percentage of clay (< 2m)
OC = percentage organic content (%)
An equation was presented by (Rahim and George, 2004), on an investigation carried
out on soil properties index for predicting 𝑀𝑟 of 12 fine-grained soils in Mississippi. The
values of predicted 𝑀𝑟 values from Equation 2.56, showed some impressive correlated
values when compared with experimentally measured 𝑀𝑟 .

𝐿𝐿 2.06 𝑃200 −0.59

𝑀𝑟 = 16.75 {1 ( ) +( ) } (2.56)
𝑀𝑐 ×𝛾𝑑𝑟 100

Where:
𝑀𝑟 = resilient modulus (psi)
LL = liquid limit (%)
𝑀𝑐 = moisture content (%)
𝛾𝑑𝑟 = confining stress (psi)
𝑃200 = percentage of passing #200 sieve (75 micron) (%)

However, several made attempts were recorded by (Lee at al. 1997; Burczyk et al. 1994;
Santha 1994; Lee at al. 1995; Drumm et al. 1997; Von Quintus and Killingsworth 1998;
Mohammad et al. 1999; Dai and Zollars 2002) to develop predictive equations for 𝑀𝑟
using soil physical properties. The equations by these listed scholars, closely predicted
Page | 37

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝑀𝑟 of subgrade with an error of 0.05%. Furthermore, from equations 2.56 it is observed

that moisture content, degree of saturation, material passing #200 sieve (75 micron), PI
and dry density are the parameters used to predict 𝑀𝑟 . Contrary to other equations, it is
only Equation 2.54 that considered applied stresses such as 𝜎𝑑 and 𝜎3 . Notably, the
equations are dependent of experimental results from soils used in each corresponding
study and can be used strictly for basic prediction of 𝑀𝑟 values for pavement design.

2.7 Factors Affecting 𝑴𝒓

Illustration of non-linear behaviour of pavement materials under dynamic traffic loading
is challenging, as 𝑀𝑟 is a complicated parameter, due rigorous procedures involved.
Although, many studies have been conducted in order to critically evaluate the factors
that can possibly affect 𝑀𝑟 of subgrade. Basically, it was discovered that applied
stresses, moisture content, swelling pressure and density are the three factors that
influences 𝑀𝑟 .

2.7.1 Stress state effect

Stress state experienced by highway pavement layers, is affected by subgrade 𝑀𝑟 . The
stress state of a subgrade is considered by the deviatoric stress and confining pressure
experienced on the subgrade during cyclic loading. Most importantly, pavement
structures are under unsaturated conditions, therefore matric suction is an important
factor that defines stress state of subgrade (Yang, 2008). Increase in deviatoric stress,
causes decrease in 𝑀𝑟 as the shear stresses experienced by the soil specimen
increases. Whereas, increase in confining pressure serves to increase the 𝑀𝑟 value of
subgrade soil, as it increases the bulk stress resistance of the soil, therefore providing
a stiffening effect on the specimen. Furthermore, effects of confining pressure are
evidenced on granular soils, while the effect of deviatoric stress is observed on cohesive
soils.

2.7.2 Moisture condition effects

Subgrades are compacted at a point close to optimum moisture content (OMC) with
corresponding maximum dry density (MDD). Most times, it becomes difficult to achieve
desired compaction moisture. Thus, leading to pavements been constructed on under
in-situ moisture conditions.

Page | 38

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

However, during pavement service, subgrade moisture content comes to equilibrium

with its surrounding moisture conditions and vary thereafter as a result of seasonal
variation (Yang, 2005; Uzan, 1998). Although, subgrade is prepared at a particular
moisture content and density, the moisture content of a subgrade will never be constant.
Therefore, it is necessary to evaluate the impact of moisture content on the 𝑀𝑟 value.
Though, the post-construction moisture content (Mc) varies significantly and as such
affects subgrade 𝑀𝑟 . Researchers like: (Pezo et al. 1992, Fredlund et al. 1977;
Mohammad et al. 1999; Huang 2001; Butalia et al. 2003; Heydinger 2003; Ooi et al.
2004; Wolfe and Butalia 2004; Hopkin et al. 2004; Richter 2006; Zaman and Khoury
2007; Cary and Zapata 2010) proved in their respective studies that 𝑀𝑟 decreases
drastically on the average between 42% and 60% with an increase in moisture content.
Their studies further revealed that Mc of subgrade soils increases, while the stiffness
tends to decrease, therefore contributed to reduced 𝑀𝑟 values. Figure 2.12
demonstrates behaviour of subgrade 𝑀𝑟 with four different Mc levels, including the
(OMC), dry of OMC (OMC - 2.5%), wet of OMC (OMC + 2.5%) and soaked condition.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟏: Variation of Mc on Mr (Nguyen and Mohajerani 2015)

Pavement under seasonal moisture variation, eventually absorbs moisture and increase
the moisture content to the wet side. During the first five years of pavement service life,
Uzan (1998) observed that the subgrades recorded an increase in Mc of about 30%
higher than soil’s initial plastic limit. Apparently, Von Quintus and Killingsworth (1998)
argued that Mc of a pavement subgrade will continuously be on the wet-side of
Page | 39

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Optimum, this result was published as 59 subgrade soils were tested for Mc after five
years of construction. This report was an evidence, that subgrade’s Mc increases
subsequently after construction.

2.7.3 Effects of soil type

In the last section, it is well understood that increase in Mc results to a decrease in 𝑀𝑟 .
Reasonably, the influence of Mc on 𝑀𝑟 differs with soil types. Drumm (1997) observed
that A-7 soils tended to have higher 𝑀𝑟 values at optimum conditions, compared to A-4
and A-6 soils, they also exhibited a larger decrease in 𝑀𝑟 once the Mc increased to
values greater than that of the optimum.
Furthermore, scholars like: Hicks and Monismith (1971), Thom and Brown (1987),
Barksdale and Itani (1989) and Kamal et al. (1993) claimed that fine content has great
influence on 𝑀𝑟 . Hicks and Monismith (1971) reported on a 𝑀𝑟 of partially crushed
aggregates. The study showed that decrease in 𝑀𝑟 , was triggered with an increase in
fines content. Among other studies on fine content, was recorded by (Barksdale and
Itani, 1989) and the study proved that 𝑀𝑟 decreases with 60% fines content increased
from 0% to 10% when fine content was reduced. Generally, it was observed that 𝑀𝑟
value decreases as the fine contents increases Kancherla (2004).

2.7.4 Matric suction effect

Matric suction effects on 𝑀𝑟 is necessary, because suction is fundamental to
unsaturated soils that impart on stress state variables. Sauer and Monosmith (1968),
conducted an in-situ study. The result showed that suction and pavement deflection are
interrelated, such that suction values increases with decrease in deflection.
Furthermore, suction increase will cause subgrade stiffness increase and thus increase
𝑀𝑟 of not saturated subgrade soils. Because subgrade suction is composed of two
components, matric suction and osmotic suction.

However, Khoury et al. (2004) demonstrated that changes in 𝑀𝑟 for unsaturated soils
are triggered by changings in matric suction. Thus, justifying matric suction definition as
different between pore-air pressure and pore-water pressure (𝑢𝑎 − 𝑢𝑤 ). The intensity of
suction present in subgrade is related to Mc, which changes in subgrade over time.

Page | 40

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Therefore, evaluation of suction influence, that will account for seasonal variation on 𝑀𝑟
for unsaturated subgrades, become valuable. Yang et al. (2005) replicated field
conditions, with laboratory compacted A-7 specimens at OMC and were then subjected
to wetting. This was justified because subgrades are basically compacted at OMC and
allowed to come to equilibrium with the surrounding soils. According to their study,
suction was measured using the Filter Paper Method, followed by 𝑀𝑟 testing. The soil
samples were wetted at two different stages, equilibrium moisture content (EMC) and
Mc between OMC and EMC. There result revealed a drastic decrease in 𝑀𝑟 when
moisture content increased from moisture content between OMC and EMC, therefor
resulting in a low matric suction.

2.8 Resilient Modulus Models

2.8.1 Universal model
Various stress-dependent models are available to establish stress sensitive relationship
between 𝑀𝑟 and various stress state, for material parameter input in AASHTO 2002
design guide. This relationship, describes stress dependency of both fine-grained soils
and unbound materials for mechanistic analysis (NCHRP 1-37A, 2004). In the same
line, Witczak and Uzan (1998) proposed a model that is widely known as universal
model. This model is adopted to evaluate 𝑀𝑟 behaviour with respect to stress state. The
generalized model adopted by MEPDG is presented in Equation 2.57

𝜃 𝑘2 𝜏 𝑘3
𝑀𝑟 = 𝑘1 𝑃𝑎 (𝑃 ) ( 𝑃𝑜𝑐𝑡 + 1) (2.57)
𝑎 𝑎

Where:
θ = bulk stress = σ1 + σ2 + σ3
σ1 = major principal stress
σ2 = σ3 = for triaxial test on cylindrical specimen
σ3 = minor principal stress or confining stress in the triaxial
1
3
τoct = octahedra: shear stress = √(σ1 − σ2 )2 + (σ1 − σ3 )2 + (σ2 − σ3 )2
√2
= (σ − σ3 ) for cylindrical specimen in triaxial test
3 1
Pa = normalising stress atmospheric pressure
k1 , k 2 , k 3 = model parameter obtained from regression analysis
Page | 41

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Equation 2.57 is useful and widely accepted, it only accounts for stress state effect on
𝑀𝑟 but does not consider the effect of moisture disparity on 𝑀𝑟 due to seasonal variation.
Research have geared up towards solving this challenge by relating the regression
constants to soil physical properties. Thus, incorporating seasonal variation effects on
𝑀𝑟 predictions (Nazzal and Mohammad, 2010; Yau and Von Quintus 2002). Nazzal and
Mohammad (2010) introduced physical meanings for the regression constants by
evaluating them across different moisture conditions as to determine moisture change,
can affect regression constants.
It was discovered that 𝑘1 is related to the stiffness of the material, which increases with
increasing effective stress. 𝑘2 describes the stiffening increase effect of bulk stress on
soil. Whereas, 𝑘2 decreases with increase Mc; 𝑘3 describes the softening of the material
with increasing shear stress, such that 𝑘3 decreases and becomes more negative as
Moisture content increases and this implies that model parameters for regression
constant greatly influenced by Moisture content.

2.8.2 Model incorporating moisture variation

Due to variation in Moisture content, several models were developed to accomplish the
task of predicting the ability of changes in 𝑀𝑟 values. However, the environmental effects
particularly the seasonal variation MEPDG introduced Enhanced Integrated Climatic
Model (EICM) to predict changes in properties of pavement structures. EICM evaluates
changes in Mc for flexible asphalt pavements and it requires the user to input 𝑀𝑟 at a
specified moisture condition for subsequent evaluation of the effect of the seasonal
changes in pavement structure. Most commonly, EICM is considered a powerful tool,
due to its capability to creates a set of adjustment factors that account for moisture
changes, freezing, thawing, and post thawing effects. MEPDG then combines the
adjustment factors obtained from EICM with the effects of cyclical loading from the traffic
and applies the total effect to the material properties. Once this is accomplished,
MEPDG makes use of transfer functions to predict pavement performance considering
the effect of EICM adjustment factor and the external loading on material properties.
The link EICM adjustment factor is presented in Equation 2.58.

Mr b−a
𝑙og M =a+ −b (2.58)
ropt 1×Exp(ln +km ∗(S−Sopt ))
a

Page | 42

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where:
Mr
= resilient modulus ratio
Mropt
Mr
a = minimum of log
Mropt
Mr
b = maxmum of log
Mropt
k m = regression parameter
(S − Sopt ) = changes in degree of saturation

The right part of Equation 2.58 represents adjustment factor, this can be solved by
applying the anti-logarithm to obtain the adjusted 𝑀𝑟 by multiplying the adjustment
factor with the 𝑀𝑟 value at optimum moisture condition. The MEPDG recommends value
of -0.5934 for 𝑎, 0.4 for 𝑏, and 6.1324 for 𝑘𝑚 , for fine-grained soils. A graphical
presentation of Equation 2.58 is given in Figure 2.13 for fine-grained soils. However,
MEPDG provides a different set of values for 𝑎, 𝑏, and 𝑘𝑚 for coarse-grained soils.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟐: Typical moisture changes on Mr using EICM (NCHRP, 2004)

Cary and Zapata (2010), evaluated the rationality of Equation 2.58 for a broader range
of moisture conditions and it was discovered that the EICM models tends to wrongly
predict 𝑀𝑟 in dry conditions, particularly for high PI soils. However, inadequate data was
available to investigate the validity of the model for more wet conditions. Figure 2.14

Page | 43

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

illustrates how EICM model fits the data collected by Cary and Zapata (2010). There is
remarkable data scatter when the S is below the optimum condition.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟑: Collected database against EICM model (Cary and Zapata, 2010)

Cary and Zapata (2010) specified that the soil type effect is analytical, considering the
rise in 𝑀𝑟 value due to decrease in Mc, particularly for soils compacted on the dry side
of optimum. Hence, soils with high plasticity index (PI) values tend to attain much higher
suction values at lower degree of saturation, compared to soils with lower PI values, this
postulation is considered based on Figure 2.15. Furthermore, they proposed a model to
accommodate the effect of soil type on 𝑀𝑟 , by including the term 𝑤𝑃𝐼, which is the
product of PI and soil fine (No.200 sieve size). The model is presented in Equation 2.59

(δ+γ∗wPI.5 )−(α+β∗e−wPI )−1

LogFu = mx [(α + β ∗ e−wPI )−1 + −(δ+γ∗wPI.5 S−Sopt ] (2.59)
1+e(ln( )+(ρ+w∗e−wPI ).5 ( )
α+β∗e−wPI )−1 100

Where:
𝑎 = α + β ∗ e−wPI
𝑏 = δ + γ ∗ wPI .5
𝑘𝑚 = (ρ + w ∗ e−wPI ).5
𝑚 = correction factor = 1.002
𝛼 = −0.600, 𝛽 = −1.87194, 𝛿 = 0.800, 𝛾 = 0.080, 𝜌 = 11.96518, 𝑎𝑛𝑑 𝜔 =
−10.19111

Page | 44

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Similar to the EICM model presented in Figure 2.12, Equation 2.59 was used to create
the model in Figure 2.14. As this model allows for more appropriate predictions in the
dry range by considering additional stiffness gain by higher PI soils in the lower
saturation range.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟒: Adjustment factor, Fu, for (S − Sopt) and wPl (Cary and Zapata, 2010)

2.8.3 Models incorporating matric suction

The need to develop a model that incorporates soil suction in predicting 𝑀𝑟 was also
recognised by Yang et al. (2005), since suction has a direct effect on the stiffness of
unsaturated soils. This model is a variation of the deviator stress model initially
introduced by AASHTO T 292-91 and the original deviator stress model for 𝑀𝑟 is
presented in Equation 2.60.
𝑀𝑟 = 𝑘1 (𝜎𝑑 )𝑘2 (2.60)
Where:
𝜎𝑑 = deviatoric stress
𝑘1 , 𝑘2 = regression constants
Yang et al. (2005) proposed a new correlation based on Equation 2.60 that considered
soil suction as stated in Equation 2.61 by utilizing unsaturated soils effective stress
concept in Equation 2.32 in chapter 2, this gives Equation 2.61.

Page | 45

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝑀𝑟 = 𝑘1 (𝜎𝑑 + 𝑥𝜓𝑚 )𝑘2 (2.61)

Where:
x = parameter representing contribution of suction to effective stress
(Note: 0 for completely dry soil and 1 for saturated soils)
ψm = suction matric
k1 , k 2 = regression constants

Conclusively, Equation 2.61 accurately adopted the effect of suction, especially at

decreased moisture contents because its effect is very remarkable and the effect of
deviator stress which is significant at higher moisture content. Thus, changes in Mc
affect suction, the effect of seasonal variation on 𝑀𝑟 , is absolutely included in Equation
2.61. This equation provides a good fit between the measured and predicted 𝑀𝑟 data.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟓: Predicted against measured Mr values using equation 2.75 (Yang et. al. , 2005)

Based on three stress variables the bulk stress, matric suction (𝑢𝑎 − 𝑢𝑤 ) and deviator
stress, Gupta et al. (2007) developed a model to predict 𝑀𝑟 for unsaturated soils. This
model was based on the model proposed by Vanapalli et al. (1996), this describes the
shear strength of unsaturated soils across the entire SWCC range.

𝜏𝑢𝑠 = 𝐶 ′ + (𝜎𝑛 − 𝑢𝑎 )𝑡𝑎𝑛∅′ + (𝑢𝑎 − 𝑢𝑤 )(𝛩𝑘 𝑡𝑎𝑛∅′ ) (2.62)

Page | 46

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where:
C′ = 𝑒𝑓fective cohesion of a saturated soil
∅′ = effective friction angle of saturated soil
(σn − ua ) = net normal stress
Θ
Θ = normalized volumetric water =
θs
k = fitting parameters

The first part of Equation 2.62 represents shear strength of soil under saturated
condition. Whereas, the second part represents the contribution to shear strength due
to matric suction. The volumetric water content, 𝛩 was introduced into the model to
reflect the quantities of water in the soil and it varies from large to a very small value at
residual conditions when the soil is saturated. This model comprised normalizing Mc to
actually evaluate the contribution of suction, since the area of contact between the soil
particles, which is wetted, reduces with an increase in suction and when suction is
decreased. The link between the increase and decrease of contact wetted area between
soil particles is connected with the rate at which changes occurs in shear strength under
unsaturated conditions. Therefore, there is a significant connection between the
strength of unsaturated soil and the SWCC that explains the relationship between water
content and suction (Vanapalli et al, 1996).

Gupta et al. (2007) explicitly expressed that the inclusion of one of the parameters that
describes SWCC into Equation 2.62, generated a power correlation between soil suction
and shear strength similar to the model presented in Equation 2.63.

𝑘 𝑡𝑎𝑛∅′ )𝛽
𝜏𝑢𝑠 = (𝜎𝑛 − 𝑢𝑎 )𝑡𝑎𝑛∅′ + 𝐶 ′(𝛩 (2.63)

The advantage of Equation 2.56 over Equation 2.64 is that there is no need to evaluate
normalized Mc and suction. Using the relationship presented in Equation 2.62 with the
Universal 𝑀𝑟 model presented by NCHRP 2003, Equation 2.63 can be formatted by
considering suction in 𝑀𝑟 equation.

Page | 47

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

θ k2 τoct k3
Mr = (k1 Pa ( ) ( + 1) + α(ua − uw )β (2.64)
Pa Pa

Where:
α = intercept of 𝑀𝑟 at given 𝜃𝜏𝑜𝑐𝑡 against suction relationship
β = slope of 𝑀𝑟 at given 𝜃𝜏𝑜𝑐𝑡 against suction relationship

Liang et al. (2008) aimed at improving the model presented by Yang et al. (2005). As
the model is devoid of regression constants calibration at each moisture content, for the
same soil type, to be effective. Furthermore, Liang et al. (2008) also intended to propose
a model which can accommodate seasonal variation effects in predicting 𝑀𝑟 . This model
is based upon the Universal Model utilized by MEPDG (NCHRP 2004).

𝜃 𝑘2 𝜏 𝑘3
𝑀𝑟 = (𝑘1 𝑃𝑎 (𝑃 ) ( 𝑃𝑜𝑐𝑡 + 1) (2.65)
𝑎 𝑎

By incorporating the effective stress equation for unsaturated soils, Liang et al. (2008)
w able to propose a new model to include suction for evaluating 𝑀𝑟 as follows:

𝜃+𝑥𝜓𝑚 𝑘2 𝜏 𝑘3
𝑀𝑟 = (𝑘1 𝑃𝑎 ( ) ( 𝑃𝑜𝑐𝑡 + 1) (2.66)
𝑃𝑎 𝑎

Where:
Pa = atmospheric pressure (100kPa)
x = Bishop′ s effective stress parameter
ψm = matric suction
τoct = octahedral shear stress
θ = bulk stress
k1 , k 2 , k 3 = regression constants

Liang et al. (2008) suggested a model introduced by Khalili and Khabbaz (1998) for
evaluation of 𝜒 parameters, which is presented in Equation 2.68. While, in Liang et al.
(2008), 𝜒 was only computed at suction values greater than the air-entry value (AEV)
since soil would be saturated before AEV and 𝜒 = will be equal to 1.

Page | 48

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

(ua −uw )b 0.55

xw = ( ) (2.68)
ua −uw

Where:
(ua − uw )b = air − entry pressure
ua − uw = suction matric

Liang et al. (2008) further carried out repeated load triaxial test to validate the model for
𝑀𝑟 values and filter paper method to obtain suction values. They conducted regression
analysis at OMC during 𝑀𝑟 test, as to obtain regression constants. Thus, regression
constants along with the model in Equation 2.68, were introduced to specimens at
different Mc to predict 𝑀𝑟 values. Liang et al. (2008) also differentiated the total stress
approach, by neglecting suction against the effective stress approach, but included
suction, to predict 𝑀𝑟 values. The 𝑀𝑟 values predicted were significantly better
compared to when suction was incorporated. A comparison between the total stress
approach and effective stress approach for A-6 soil is displayed in Figure 2.17. It was
observed that including matric suction helps in improved version of the predictive model
for 𝑀𝑟 values.
Cary and Zapata (2011) also expressed a model that included the effect of suction in
evaluating 𝑀𝑟 of unsaturated soils. Contrary to other models, this model included the
effects of pore-water pressure (PWP) build up during cyclic loading. Dissipation
happens in the delay time between applied loads, whereas excessive PWP in soil is
usually generated under moving vehicle loads. When the delay in time is long, then the
possibility of PWP accumulation between load cycles and PWP will not occur.

Page | 49

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟔: A − 6 soil predicted against measured Mr values (Liang et al. , 2008)

Cary and Zapata, 2011 further stated that, if the delay time is short (i.e., fast moving
traffic) there will be remarkable accumulation of excess PWP as the number of applied
loads increases. The dissipation of PWP is a function of the coefficient of permeability, 𝑘
and time delay between load repetitions. At high coefficient of permeability 𝑘 or if there
is large time delay between load repetitions, this condition can be simulated in the
laboratory through performing a drained 𝑀𝑟 test. Most commonly, at a low coefficient of
permeability 𝑘, an undrained 𝑀𝑟 test is recommended, as to accurately simulate field
conditions.
θnet −3∗∆uw−sat k2 τ k3 (ψm −∆ψm ) k4
Mr = k1 Pa ( ) ( Poct + 1) ( + 1) (2.69)
Pa a Pa

Where:
Pa = atmospheric pressure
θnet = θ − 3ua = the net bulk stress (θ=bulk stress =σ1 +σ2 +σ3 and σa = Pore − air
pressure)
∆uw−sat =pore − water pressure build − up under saturated condition (ψm =0)
1⁄
3
τoct = Octahedral shear stress= √(σ1 − σ2 )2 + (σ1 − σ3 )2 + (σ2 − σ3 )2
ψmo =initial matric soil suction and
∆ψm = relative change in soil matric suction with respect to ψm o due to pore − water
pressure build − up under unsaturated condition (∆uw−sat =0)
k1 ≥ 0, k 2 =≥ 0, k 3 ≤ 0 and k 4 ≥ 0 are regression constants
Page | 50

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The model presented in Equation 2.69 was formulated using 𝑀𝑟 testing conducted with
the concept of an unsaturated soil triaxial cell. That allows the use for axis-translation
technique as to obtain suction and PWP during 𝑀𝑟 testing. Thus, using 𝜃𝑛𝑒𝑡 instead of
θ to represent the bulk stress (𝜃𝑛𝑒𝑡 = 𝜃 − 𝑢𝑎 ) as the soil tends to reach saturation, 𝑢𝑎
approaches towards 0 and 𝜃𝑛𝑒𝑡 becomes θ. Cary and Zapata (2011) validated Equation
2.69, by performing several different comparisons and there was no much divergence
with these results.
Witicizak et al. (2000) suggested a model that incorporates environmental adjustment
factor along with 𝑀𝑟 at an applied effective stress to predict changes in 𝑀𝑟 as a function
of changes in degree of saturation.

b−a θ k2 τ k3
(a+ −b )∗k1 Pa ( ) ( oct +1)
1+EXP(ln km∗(S−S Pa Pa
a
Mr = 10 0pt) (2.70)

Cara and Zapata (2011) used 𝑀𝑟 test data to obtain regression constants, 𝑘1 to 𝑘4 , in
Equation 2.69. The predicted 𝑀𝑟 results from Equation 2.69 were compared to those
obtained using Equation 2.70. The comparison, is presented in Figure 2.18, shows that
Equation 2.69 tends to give a better prediction of measured 𝑀𝑟 values.

a.) Using Equation 2.70

Page | 51

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

b.) Using Equation 2.69

𝐅𝐢𝐠𝐮𝐫𝐞 𝟐. 𝟏𝟕: good line fit measured against predicted Mr values for soil PI
= 5 (Cary and Zapata, 2011).

Cary and Zapata (2011) did a comparison of the model presented in Equation 2.58 with
that of Liang et al. (2008) suction dependent of 𝑀𝑟 model. When compared with Figure
2.18 (part b), Cary and Zapata (2011) model provides a better prediction of 𝑀𝑟 for this
soil type.
Nokkaew et al. (2014) carried out a study, to investigate effects of matric suction on 𝑀𝑟
of Recycled asphalt pavement (RAP) and Recycled Asphalt Material (RAM) in a post
compaction state. These authors investigated the relationship further for RAP and RAM
since this materials are repelling materials. In order, to determine the correlation
between specimens prepared at OMC and 95% of Maximum Dry Density (MMD).
Materials were subsequently saturated, and then dried to a target suction value before
𝑀𝑟 testing commenced. In an attempt to dissect the data obtained from 𝑀𝑟 testing,
Nokkaew et al. (2014) employed the model proposed by Liang et al. (2008) to predict
𝑀𝑟 values, however slight alteration was made, using the definition of 𝜒 presented in
Equation 2.71.

𝜃−𝜃 𝑘
𝑥 = 𝛩𝑘 = (𝜃 −𝜃𝑟 ) (2.71)
𝑠 𝑟

Page | 52

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where:
Θ = volumetric water content
θr = residual water content
θs = saturated water content
k = fitting parameter to fit measured values to predicted values of x

This resulted in Equation 2.72 being utilized for the prediction of 𝑀𝑟 .

k k2 k3
θ+Θ ψ τ
Mr = k1 Pa ( ) ( Poct + 1) (2.72)
Pa a

Clearly, it was observed by Nokkaew et al (2014) that similar curve fitting of the
measured data was observed when compared with Liang et .al (2008) equation.
Nonetheless, Nokkaew et al. (2014) argued that Liang et al. (2008) model cannot fittingly
predict 𝑀𝑟 near saturation and at residual condition. As a result of the definition of 𝜒
used by Liang et al. (2008), that linearly assumes a relationship between 𝜒 and soil
suction in a logarithmic scale when the suction value is greater than the air-entry
pressure.

Page | 53

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPTER 3 : SOIL PROPERTIES AND TESTING PROGRAMS

3.1 Overview
This chapter illustrates the selection of natural subgrades for the experimental
programme, together with the basic geotechnical properties of the soils. Standardised
civil engineering tests procedures were followed, for all the experiments conducted in
this study. This chapter further presents series of laboratory tests carried out on the nine
selected subgrade soils as to accomplish the aim of this study:

• Modified compaction test

• Free swell index test (FSI)
• Zero swelling test (ZST)
• Filter paper test.
• California Bearing Ratio (CBR) test and unsaturated CBR test
• Triaxial tests (consolidated undrained test, CUT),
• Repeated load triaxial test (RLTT),

3.2 Materials
3.2.1 Soil geological location
The natural subgrade soils for this investigation were selected across three provinces
in South Africa i.e. Free state, Northern cape and Gauteng province. These soils are
metamorphosed by weathering of basic crystalline rock, to an expansive black clays
known as cotton soils or turf. Lager area of South Africa is susceptible to subgrade
problems triggered due to expansive activities. In addition, collapsible soil is among the
problematic soil considered for this study. This soil is found largely in Gauteng province
between Johannesburg and Pretoria area of South Africa. This soil stands higher risk of
collapse and subsidence, as it is formed by weathering of dolomitic and limestone rocks.
This type of soil undergoes precipitous decrease in volume due to wetting load and
easily erodible by acidic water. As this causes sudden collapse or sinkholes on
pavement structure. The studied soils, were sampled from three different locations at
each province and labelled according to the province from which they were collected
such as: Gauteng province soil (GPS 1 GPS 2 and GPS 3), Free State soil (FSS1, FSS
2 and FSS 3) and Northern Cape soil (NCS 1, NCS 2 and NCS 3), making it a total of
nine different soil samples.

Page | 54

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

3.3.2 Soils XRF test

Chemical composition of the nine subgrade soils was determined using thermo scientific
RIGAKU 3000 energy-dispersive x-ray fluorescence spectrometry analyser. Having an
excitation source as miniaturised 30 kV X-ray tube. Dried pulverised soil passing
through sieve size of 250 µm aperture were used. RIGAKU 300 XRF analyser was
automatically adjusting for matrix effects, as to determine the various chemicals
contented in the soils for an hour, without requiring any empirical input from the
instrument operator. The XRF results of the subgrade soils, determined the existence
of the following major and minor oxide: Al2O3, SiO2, k2O, CaO, MnO, Fe2O3 (major) and
Na2O, MgO, TiO2 BaO (minor). The average concentrations of heavy metals CuO,
Rb2O, SrO, ZrO2, Cr2O3, NiO, Rb2O and Nb2O5 were detected to the lowest minimum,
for the oxides such as AgO2, CdO, HgO, Sb2O3, Se2O3 and SnO2 were observed to
have low concentrations and were below trace level. The chemical compositions of the
nine studied soils are summarised in Tables 4.1 while the detailed results are presented
in Appendix A.

Table 3.1: Soils Chemical Composition

FSS 1 FSS 2 FSS 3 NCS 1 NCS 2 NCS 3 GPS 1 GPS 2 GPS 3
Chemical Mass Mass Mass Mass Mass Mass Mass Mass Mass
oxides (%) (%) (%) (%) (%) (%) (%) (%) (%)
SiO2 58.39 55.34 49.15 55.14 53.43 56.17 66.59 61.12 64.34
Al2O3 21.41 25.16 27.31 19.80 22.71 20.13 17.50 19.65 16.83
Fe2O3 12.09 12.09 13.93 18.99 11.56 12.56 8.29 11.34 10.29
TiO2 1.13 0.25 2.12 0.76 0.88 1.57 0.78 0.92 0.72
MgO 1.96 4.37 1.81 0.64 3.23 2.35 0.72 0.56 0.63
CaO 1.75 2.25 0.23 0.22 0.56 0.54 1.27 1.74 1.46
Na2O 0.53 2.13 0.89 0.052 0.11 0.13 0.37 0.52 0.61
K2O 1.89 6.38 2.25 3.77 4.13 2.98 2.98 2.56 3.11
Others 0.85 3.12 2.31 0.628 4.87 3.57 1.50 1.68 2.01
LOI (%) 0.23 0.84 0.67 0.27 0.45 0.17 0.32 0.27 0.13

3.4 Soil Classification

The particle size distribution (PSD) analysis and Atterberg limits test were carried out
for the classification of the soils. However, under unsaturated soils mechanics, soil
classification takes an additional significance meaning. Though the PSD curve provides
information on the percentage distribution of the soil particle sizes.

Page | 57

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The distribution of the soil particles bears a correlation with pore sizes distribution.
Information from pore-size distribution can be used to estimate water content–soil
suction relationship for the soil (i.e., SWCC). Consequently, the PSD becomes of
increased value for understanding the mechanics of unsaturated soil.
Furthermore, various standards are used for soil classification, though it depends on the
geotechnical engineer’s choice of standard. As long as the engineer provide justification
that is in line with the local standard, besides most of these standards are written under
the same principle. In this research, American Society for Testing and Materials (ASTM)
and American Association of State Highway and Transportation Officials (AASHTO)
which are in conformity with the local South Africa standard i.e. Technical Methods for
Highways (TMH) were used.

3.4.1 Soils particles size distribution (PSD)

The objective of this research was to investigate the geotechnical response of natural
subgrades that are prone to swelling, fatigue and cracks for pavement, when subjugated
to cyclic moisture stresses and repeated loading. These problematic soils were carefully
identified and collected from three provinces across South Africa. Figures 3.3 through
3.5 shows the PSD analysis of the respective studied soils. Whereas, summary of the
average particle size (D60), effective size (D10), uniformity coefficient (Cu), coefficient of
curvature (Cc) and the GPS coordinate location across nine sites from where the soils
were sampled are summarized in Table 3.2. The result of PSD in Figures 3.3 through
3.5 reveal that fine content of the soils varies. Apparently, the curves show that FSS
contained the highest number of fines among other studied soils, followed by NCS and
GPS when traced from the curve. The soils particle content, ranges from well graded
soils to poorly soils with no possibility of gap as the samples had very high percentages
of fines with over 50% passed through No. 200 sieves.

Page | 58

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

100
FSS 1
90
FSS 2
80 FSS 3
Percentage finer (%) 70

60

50

40

30

20

10

0
0.0001 0.001 0.01 0.1 1 10 100
Sieve Sizes (mm)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟑: FSS grain size distribution curve

100

90 NCS 1
NCS 2
80
NCS 3
70
Percentage finer (%)

60

50

40

30

20

10

0
0.0001 0.001 0.01 0.1 1 10 100
Sieve sizes (mm)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟒: NCS grain size distribution curve

Page | 59

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

100

90
GPS 1
80 GPS 2
Percentage finer (%)
70 GPS 3

60

50

40

30

20

10

0
0.0001 0.001 0.01 0.1 1 10 100

Sieve sizes (mm)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟓: GPS grain size distribution curve

The coefficient of uniformity, (Cu) values for FSS 1, 2 and 3 are; Cu=19, 33 and 11, this
depicted well-graded soils. While NCS 1, 2 and 3 have Cu values of 11, 5.2 and 52.3.7
which qualifies the soils as poorly-graded. Furthermore, Cu for GPS 1 and 3 are 15.3,
45 and 9, this show that GPS 1 and 3 are uniformly-graded, whereas GPS 2 is well-
graded.
The Coefficient of curvature, (Cc) determines the shape of particle sizes. The values
obtained for the studied soils are summerized in Table 3.2. The values of Cc for FSS 1,
2 and 3 are 1.89, 0.08 and 1. FSS 2 portrayed an irregular shape, because its value for
Cc is not stipulated within the standard range i.e. 1 – 3 The NCS 1, 2 and 3 portrayed
irregular shapes because they have Cc values to be15.8, 0.77 and 5.63 which are not
within the range, due to high percentage of fines content. In addition, GPS 2 is within
the stipulated range because it has Cc value 2.2 and GPS 1 and 3 is found to be outside
the range. Therefore, GPS 2 is well-graded whereas GPS 1 and 3 are poorly graded.
This result is in consistency with Cu results.

𝐷60
𝐶𝑢 = Coefficient of uniformity ( ⁄𝐷 ) (3.1)
10

(𝐷30 )2⁄
𝐶𝑐 = Coefficient of curvature ( (𝐷10 ) × (𝐷60 )) (3.2)

Page | 60

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Table 3.2: Grain size analysis of the studied subgrades

Gradation (%) Modulus of gradations Coordinates

Soils Fines Latitude

Sand Silt Clay D10 D30 D60 Cu Cc longitude
290 11’ 22.77” S;
FSS 1 8.62 34.44 56.94 0.001 0.006 0.019 19 1.89 260 12’ 58.18” E
-280 06’ 6.0” S; 260
FSS 2 3.61 37.19 59.20 0.0003 0.003 0.01 33 0.08 53’ 59.99” E
-270 58’ 6.0” S; 260
FSS 3 24.62 31.21 44.17 0.0045 0.015 0.05 11 1.00 44’ 6.22” E

-280 44’ 4.19” S,

NCS 1 11.82 29.44 57.24 0.001 0.008 0.011 11 15.8 240 46’ 6.59” E
-280 33’ 60” S, 240
NCS 2 7.16 30.02 59.80 0.0025 0.005 0.013 5.2 0.77 17’ 60” E
290 2’ 19.2” S, 240
NCS 3 4.11 39.14 55.75 0.00012 0.0025 0.008 53.3 5.63 36’ 39.6” E

250 51’ 0.67” S,

GPS 1 27.62 28.40 43.98 0.0022 0.008 0.035 15.9 0.83 280 11’ 16.54” E
-25°59'53.99" S
GPS 2 17.38 45.08 37.54 0.0004 0.004 0.018 45 2.2 28°06'25.99" E
250 51’ 6.84” S,
GPS3 45.34 28.57 26.09 0.01 0.028 0.09 9 0.87 280 11’ 22.56” E

3.4.2 Atterberg limit test

Atterberg limit results are presented in Table 3.3. The soils were classified as fine-
grained soils, because they have over 50% fine by dry weight of the soils that passes
through sieves #200 (75-μm). The analysed test data and corresponding plasticity chart
for the nine soil samples are illustrated in Figures 3.6 through 3.8. The plasticity curves
demonstrate plasticity characteristics of the soils. Each symbol is explained using
Universal Soil Classification System (USCS) from which the characteristic description
of each soil is obtained. FSS and NCS, are of high Plasticity with Liquid limit greater
than 35%, and GPS has a low plasticity. The data were recorded from three sets of
performed tests, in order to obtain a reliable and most consistent results.

Page | 61

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Table 3.3: Consistency limits values

Natural
Soil Liquid Plastic Plasticity Degree of Specific Moisture
Samples Limit Limit (%) Index (%) expansion Gravity Gs Content,
(%) 𝒘𝒏 (%) AASHTO

FSS 1 62.10 21.32 40.78 High 2.71 26.00 A-7-6

FSS 2 68.03 23.82 44.21 High 2.73 24.23 A-7-6
FSS 3 60.28 23.56 36.72 High 2.69 25.64 A-7-6

NCS 1 54.91 20.93 33.98 Marginal 2.69 22.14 A-7-5

NCS 2 61.32 21.13 40.19 Marginal 2.70 23.11 A-7-5
NCS 3 66.88 21.21 45.67 Marginal 2.72 22.58 A-7-5

GPS 1 45.33 22.13 23.20 Low 2.68 17.58 A-6

GPS 2 52.28 24.34 27.94 Low 2.72 17.78 A-6
GPS 3 40.12 22.81 17.31 Low 2.65 16.90 A-6

Furthermore, FSS plotted data are positioned above A-line and their plasticity index is
greater than 4% (PI>4%), NCS is classified as high plastic soils because the plotted
data is traced above A-line. However, FSS 1, 2 and 3 has a liquid limit of 60.10%, 68.03
and 60.28% with plasticity index of 40.78%; 44.21% and 36.72% respectively. This area
is identified as CH (high plastic clay). NCS 1, 2 and 3 are classified as CH as well,
because the liquid limits obtained for these soils are: 54.91%, 66.88% and 66.88%.
Whereas, the plasticity indexes of soils recorded are 33.98%, 40.91%, and 45.67%
respectively. This area is identified as CH (high plastic clay), because it is positioned
above A-line. The GPS 1 and 2 are classified as CL because they recorded liquid limits
of 45.33% and 52.28% with corresponding plasticity indexes of 23.20% and 27.94% this
area is identified as CL (low plastic clay) and it plotted above A-line but on the left side
of the curve (Figure 3.8). GPS 3 is classified as low plastic silt (ML), because its plotted
data is located below A-line on the left side of the chat with a liquid limit and a
corresponding plasticity index of 40.12% and 17.31% respectively.

Page | 62

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Table 3.4: USCS, AASHTO and TRH Soil Classification

Soil Sample Group Index USCS AASHTO TRH
FSS 1 38.92 CH A-7-6 G10
FSS 2 48.72 CH A-7-6 G10
FSS 3 30.12 CH A-7-6 G10

NCS 1 39.31 CH A-7-6 G10

NCS 2 34.53 CH A-7-6 G10
NCS 3 49.11 CH A-7-6 G10

GPS 1 12.84 CL A-7-5 G10

GPS 2 23.39 CL A-7-5 G10
GPS 3 8.006 ML A-7-5 G10

3.5 Geotechnical Testing Programs

With the view of realizing the objective of this study, some laboratory testing methods
were adopted. However, both local and international testing specification were carefully
crafted before laboratory testing commenced. These testing standards are: TMH, ASTM
and AASHTO.

3.5.1 Free swell index (FSI) test

Free swell index test was conducted on the studied soils according to Indian standard
(Is: 2720 (1977, Part 40) test method as to evaluate the swelling potential of the soils.
These soils were passed through 425 μm sieve size (#40) and the soils were oven-dried
at a temperature of 115°C. Ten grams of each soil samples were poured into two
separate graduated cylindrical glass jars of 100 ml capacity, with one containing distilled
water plus the 10g of the soil and the other 10g in a jar contained kerosene. The jars
were allowed to stand for 48 hours and the final volumes of the soil were recorded and
FSI of the soils were determined and recorded. Figure 3.5 shows the test setup.

Page | 65

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟗: Free Swell Index setup of the subgrade soils

𝐹𝑆𝐼 = [(𝑉𝑆 − 𝑉𝐾 )/𝑉𝐾 ] × 100% (3.4)

Where:
FSI= free Swell Index
Vk = volume of soil in kerosene,
Vs = volume of soil in water

3.5.2 Zero swelling test (ZST)

The method adapted for this research is zero swell test (ZST). The pressure which must
be applied to the soil as to overcome the soil’s swelling pressure from any further
swelling through wetting process is called swelling stress. This procedure is also
designated as Zero Swell Test (ZST) (Basma et al, 1995; Fattom and Barakat, 2000).
The ZST was conducted in line with ASTM D4546 (2013). The soil samples passing
through ASTM sieve size of 4.75mm (#4) were used. The specimens were prepared at
varying moisture content equivalent to the one gotten from compaction test. Hence,
total of 5kg by weight of the soils were Oven-dried at a constant temperature of 1050C
for 48 hours. Followed by compaction in a modified mould at different dry unit weight, 𝛾𝑑 .
The specimens were demoulded and cut into slices of 25mm height and 75 mm in
diameter bearing the shape of the consolidation ring and oedometer ring was fitted into
the sliced specimens.
Page | 66

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟎: Fitted Specimens in oedometer ring

The inner part of the ring, was greased to reduce resistance friction between the inner
surface of the ring and the soil sample during testing. The specimen was held inside the
metal ring and placed between two porous stones. The upper porous stone, which can
easily move inside the ring with a small spacing, is positioned below a metal loading cap
through which pressure can be applied to the specimen and a collar was attached. The
sample was levelled and covered with filter in order to preserve the two exposed sides.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟏: Zore swell test setup

The soils were then loaded under sustained stress condition equal to the overburden on
the consolidation set-up. The representative soils were loaded and set at surcharge
force. Upon wetting, the height of the specimen begins to increase and additional loads
were applied each time change in height was recorded by the dial gauge, as to
overcome swelling at that particular height. The same procedure continued, until no

Page | 67

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

additional swelling was recorded on the dial gauge. The applied loads that was on the
prepared soils to counter the active swelling stress were recorded and the swelling
stress of the soils were calculated using Equation 3.5.

(∑𝑛
𝑖 =1𝑀𝑖 )×𝑔×𝑏𝑟
𝑃𝑠 (𝑘𝑃𝑎) = ( 𝜋(∅2 )
)
⁄
4
1000 (3.5)

Where:
𝑃𝑠 = swelling pressure in kPa
(∑𝑛𝑖 = 1𝑀𝑖 ) = total number of surcharges
𝑔 = acceleration due to gravity 9.81m/𝑠 2
𝑏𝑟 = beam ratio of the oedometer arm
𝜋(∅2 )
= internal area of the ring
4

3.5.3 Modified proctor compaction

The Proctor compaction test is a geotechnical laboratory testing procedure used to
evaluate moisture-density relationship of soils. Specifically designed to determine the
optimum moisture content of soils with their corresponding maximum dry density. The
test procedure was carried out by adding water to each soil sample to bring it to the
desired moisture content. Five layers of the soil were then compacted by evenly
distributing 55 blows in a mould, having 127 ± 1mm height and 152.4 ± 1mm diameter
using a modified Proctor hammer of weight 4.536kg, striking at a distance of 457.2 ±
1mm in accordance with TMH1 Method 7 (1986). After which, some small quantities of
the soils were taken from top, bottom and middle to determine the moisture content.

𝑀1 −𝑀2
𝑤𝑐 = × 100% (3.8)
𝑀2 −𝑀𝑐

where
M1 = mass of compacted sample + Can
M2 = oven dry mass of compacted sample + Can
Mc = mass of Can
wc = Moisture content

Page | 68

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

3.5.4 Filter paper test (Whatman No. 42)

During sample preparation for each test i.e. CBR, ZST, consolidated undrain test, and
resilient modulus test. The identical sets of specimens were prepared at different
moisture content. One set of the identical specimens were used to determine the soils
suction, while remaining specimens were used to measure the above-mentioned tests.
Average of two results for each tested specimen were taken as the most consistence
results determined in the laboratory. Filter paper test is considered in this study, due the
ability of this methods to covers full range of suction. Soil suction tests were performed
using filter paper technique according to ASTM D 5298-10 test method. The filter paper
moisture contents were converted to matric suction using the calibration curves in ASTM
5298-10 and the soil water retention curve was established.
Though, several methods for measuring suction and the range of suction values vary
from one standard to another. Measuring the negative pore water pressure directly can
be limiting, because devices typically used for this measurement only measures very
low suctions. Table 3.5 (Ridley 1993) summarized different devices used to measure
suction and range of suction that can be measured.

Table 3.5: Time duration and various methods of measuring suction (Ridley 1993).

The filter paper method for both the contact and non-contact filter papers to measure
the matric and total suctions was selected for this research, because it can cover a wide
range of suction measurement. The working principle behind the filter paper technique,
is that the filter paper moisture content will come to equilibrium with the soil’s moisture
Page | 69

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

content either through vapour flow or liquid flow. As the filter paper absorbs water
through vapour flow (no contact between the filter paper and soil), then only total suction
is measured. Whereas, absorption of moisture through fluid flow (contact between the
filter paper and soil), allows for matric suction is measurement.
Following the ASTM D 5298 – 10 Standard Test Method for measurement of Soil
Potential (Suction) Using Filter Paper. Dry filter papers were used, as obtained directly
from their boxes. Throughout testing, the filter papers with a pair of clean tweezers with
gloved hands were used to handle the filter papers, as to avoid contamination. A number
of soil specimens were compacted and were carefully cut into sizes. Approximately,
these cut specimens occupied 75% volume of the air-tight containers having 85 mm and
70 mm of diameter and height respectively. After which, the cylindrical specimens were
equally sliced at the middle and three contact filter paper was sandwiched between the
two sliced soils. Thus, the joint was sealed off using electrical tape to ensure no loss of
moisture from the filter paper.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟐: Sliced soil and sandwiched filter paper

The wrapped specimen is then placed into the air-tight container and A poly-vinyl
chloride (PVC) ring (diameter 40mm and thickness 5mm) was placed on top of each soil
specimens onto which the non-contact filter paper was placed.
The test specimens were then sealed in the air-tight container and transferred to an ice-
chest box, for 14 days as to allow for moisture equilibrium between the filter papers and
soil, under controlled temperature of 250C.

Page | 70

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟑: Sandwiched in temperature − controlled ice chest

After 14 days, the contact and no-contact filter papers were retrieved, weighed and their
water contents were determined following the procedures suggested by Bulut et al.
(2001).

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟒: Weighing the filter paper inside moisture can

Then, the filter paper calibration curve for water content versus suction, with the
corresponding suction values were calculated from the curve. Therefore, a calibration
curve was adopted (i.e., the one curve presented for two different filter papers in ASTM
D 5298 – 10 Standard Test Method for Measurement of Soil Suction, Using Filter Paper).

Page | 71

© Central University of Technology, Free State

© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

a) NaCl solutions were prepared from 0 to 2.7 molality. The molality is the number
of moles of NaCl in 1000ml of distilled water. For example, one mole of NaCl is
58.4428g. Whereas, 2 molality of NaCl means 2 times 58.4428g which is
equivalent to 116.8856g NaCl in 1000ml distilled water. Figure 3.16 gave the
NaCl weight at various suction values.

16000
*pF= 1+ Log kPa ψ (kPa)
14000

12000
R² = 0.998
10000
Suction (kPa)

8000

6000

4000

2000

0
0 0.5 1 1.5 2 2.5 3

Molality (g)

𝐅𝐢𝐠𝐮𝐫𝐞. 𝟑. 𝟏𝟔: Total suction of NaCl at 20°C (adopted from Lang, 1967)

b) A 300 ml glass jar was filled with 200 ml of a known NaCl molality and the glass
jar was labelled according to the solution molality the jar contained.
c) Then, a plastic support is put into the glass jar, Figure: 3.17 shows the setup of
the glass jar, the plastic support and filter papers placed a few millimetres above
the solution level.
d) Two filter papers were placed on top of the plastic support in order to minimise
error scale readings and in case if one filter paper is mistakenly dropped during
the exercise, the other filter will be used. The lid of the glass jar was air-tightened,
and a plastic tape was used to seal off the glass jar, as shown in Figure 4b.

Page | 73

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞: 𝟑. 𝟏𝟕: Plastic support hold filter papers; (b) glass jar tightly closed

a) Step b and d were repeatedly performed for each different NaCl concentrations.
Then, glass jars were kept in a controlled temperature chamber, 2 weeks
equilibrium period was adopted for the calibration exercise.

After equilibrium was attained, the filter paper moisture content was determined
following the below listed procedures
a) Prior to filter papers moisture content determination, all the related items for
calibration procedure were kept cleaned and the gloves were used throughout
the entire exercise. Furthermore, each glass jars containing the filter paper, the
NaCl solution and plastic support were retrieved from the controlled temperature
chamber one at a time. The moisture tins were weighed to the nearest 0.0001g
accuracy, and the filter paper moisture content of each NaCl concentration were
recorded on the laboratory data sheet.
b) Throughout the entire exercise, all the measurements were performed buy two
persons. i.e. one person was opening the sealed glass jar, while the other person
was inserting the filter paper into the moisture tin swiftly in less than 5 seconds
using the tweezers.
c) Following procedure “b” the mass of each moisture tin with the wet filter papers
were recorded with the moisture tin labels for the bottom and the top filter paper.
d) Afterwards, all the moisture tins were placed in the oven at a temperature of 105
± 5 °C for 24 hours with the lids half-closed as to allow for moisture evaporation.
e) The moisture tins were then fully closed with theirs lids, prior to retrieving from
the oven and were allowed to equilibrate for 5 minutes in the oven, as the mass
Page | 74

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

of the dried filter papers and moisture tins were being taken. Immediately, each
moisture tins were removed from the oven, they were placed on a metal block for
20 seconds, in order to step down the tins hotness. Then, each moisture tin
containing the filter paper were weighed again. Subsequently, each dry filter
paper was retrieved from the moisture tin and the weight of each cold tin were
taken in few seconds. Thus, recording each value of the dry filter paper from each
concentration on the laboratory data sheet.
f) Lastly, the values of each filter paper moisture content were used to calculate the
suction values of each Nacl concentration. The curve of suction and each
corresponding moisture content was plotted. Thus, the curve was designated as
Authors calibration curve (Figure: 4.10).
The calibration curve of moisture content against the corresponding suction values of
the filter paper was obtained from calibration process. The calibration curve of the filter
paper is obtained when the suction value in pF or Log (kPa) units are represented with
the corresponding moisture content. The curves were plotted using Whatman No. 42
type filter papers and Schleicher & Schuell No589 White Ribbon filter paper as
previously given by ASTM D 5298 (1994) in Figure: 3.15. Therefore, the Author’s
calibration curve for Whatman No. 42 type filter papers was used for this study, for the
calculation of suction.

6 Linear (Whatman
Log(kPa) = -0.0769wf + 5.2266 No.42)
R² = 0.9962 Linear (Whatman
5 No.42)
Suction, Log(kPa)

Linear (Schleicher &

Schuell No.589)
4 Linear (Schleicher &
Schuell No.589)

2 Log(kPa) = -0.0126wf + 2.3361

R² = 0.8647
Log(kPa) = -0.0674w + 4.9771
1 R² = 0.9973

Log(kPa) = -0.0131w + 2.0618

R² = 0.9599
0
0 20 40 60 80 100
Filter paper water content, 𝜔 (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟖: Authors filter papers calibration curve

Page | 75

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

3.5.6 California bearing ratio (CBR) test

CBR test is experimental approach for evaluating the bearing capacity of pavement
subgrade. This test was performed by applying load on the penetration piston over
compacted cylindrical specimens with the size of 152mm and height of 127mm inside
a rigid mould at the rate of penetration which is approximately 1.27 mm/min. The load
readings at penetrations was recorded as follows: 0.64mm, 1.27 mm, 1.91 mm, 2.54
mm, 3.18 mm, 3.81 mm, 4.45 mm, 5.08 mm, 5.72 mm, 6.35 mm, 7.62 mm, 8.26 mm,
8.90 mm and 9.53 mm as described in TMH1 Method 8. Essentially two types of CBR
test associated with the testing of the soils was conducted i.e. soaked and unsoaked.
The unsoaked CBR test was performed to simulate natural conditions. Whereas, the
soaked test was performed to replicate severe and extreme conditions in the field, such
as a high-water table rise in the pavement due to moisture variation.
Furthermore, specimens at different moisture content were prepared equivalent to the
moisture content obtained from moisture-density test, as to determine moisture
influence on the studied soils. Uniform compactive efforts were maintained to produce
specimens with uniformed unit weight. The specimens were subjected to soaked and
unsoaked curing conditions for four days. Subsequently, after four days soaking each
specimen were weighed again and subjected to penetration by a cylindrical rod using
automated CBR machine as shown below.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟏𝟗: Automated CBR testing machine

The CBR values of the tested soils were determined by plotting CBR curve with the load
on the vertical axis and penetration depth on the vertical axis. In addition, equation 3.9
Page | 76

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

was employed to calculate CBR values at three different strain loading level i.e. 2.54mm,
5.08mm and 7.62mm. The California standard values for these depths are 13,344,
20,016 and 25,354 kN respectively.

P
CBR = P × 100% (3.9)
s

Where:
𝐶𝐵𝑅 = 𝐶𝑎𝑙𝑖𝑓𝑜𝑟𝑛𝑖𝑎 𝑏𝑒𝑎𝑟𝑖𝑛𝑔𝑠 𝑟𝑎𝑡𝑖𝑜
P = is the plunger − load KN/m2 for the tested soil.
𝑃𝑠 = is the plunger − load in KN/m2 for the standard soil.
Moisture content was determined using Equation 4.8 above

3.5.7 Unconfined compressive strength (UCS) Test

Unconfined compressive strength is the load per unit area at which an unconfined
cylindrical specimen of soil will fail under axial compression. However, UCS of a soil is
determined by measuring the load required to allow a standard compressive machine
to crush a compacted soil according to (TMH1 METHOD 14). The UCS specimens were
prepared at different moisture content, equivalent to the moisture content gotten from
compaction test. Cylindrical specimens with diameter of 152mm and height of 127mm
were stored in the curing chamber at a temperature of 230 and 98% relative humidity.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟎: Cured soil specimens

Afterward, the compacted specimens were tested for 7 days curing period at the rate of
1.2 mm/min and UCS the compacted soils were determined using Equation 3.10.
Page | 77

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

P kN
𝑞𝑢 = A = πr2 (3.10)

Where:
𝑞 = Unconfined compressive strength (kPa),
𝑘𝑁 = Load required to crush specimen (kN)

𝑟 = radius of specimen face (0.025m),

Stress − strain values were calculated using Equation 3.11

∆L
The axial strain, ε = L (3.11)
o

Where ∆L = the change in the specimen length

Lo = the initial length of the specimen

A0
A = (1−ε) (3.12)

𝐴0 = the initial cross − sectional area of the specimen (mm2 )

𝐴 = the corresponding cross − sectional area (mm2 )

3.5.8 Consolidated undrain triaxial (CUT) test

Shear strength test such as CUT test procedures are described in this section. Samples,
obtained from soil passing through No. 4 sieve, were used to prepare the specimens at
different moisture content targeting the moisture content obtained from compaction test.
The method used in this research to measure shear strength of the prepared specimens
is consolidated undrained (CU) triaxial and the total stress was measured through
consolidation phase. The shear strength test was conducted on the selected soil
samples in accordance with ASTM D 4767 – 99. The shear strength parameters (𝐶’, ∅0 )
of the soils were determined. The prepared specimens were cylindrical, having 150mm
diameter and 300mm height corresponding with height-to-average diameter ratio of 2:1
with largest particle size smaller than 0.167mm of the specimen diameter. Specimens
were allowed to drainage under initially applied normal stress only and full consolidation
was allowed to take place. The curves of shear stress at failure against normal stress
with the shear strength parameters at each moisture content were automatically plotted

Page | 78

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

by the Instron software. Thus, the shear strength of the studied soil was calculated at
each moisture content using Equation 2.19.This is usually isotropic in most standard
testing and it is denoted as consolidated isotropic undrained triaxial test (CIU) with or
without pore pressure measurement. Different combinations of axial stress and cell
pressure are possible only on the triaxial test. The triaxial equipment was preferable for
this because both the saturated and unsaturated soil tests could be performed on them
with very minimal adaptations.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟏: Complete triaxial testing arrangement

Extension of the Mohr-Coulomb shear strength criterion to unsaturated soils requires

determination of the shear strength parameters 𝑐’ and 𝜙 ′ . conventional triaxial and direct
shear apparatus are used to determine these parameters for saturated soils. However,
for unsaturated soils this apparatuses needed some modification to make room for pore
air and pore water pressure measurements simultaneously. Blight (2013) discovered
several challenges for the experimentation of shear strength for unsaturated soils as:
• The need to make a large number of tests to establish the variation of shear
strength with matric suction.
• The long time required to achieve equilibrium in soil samples before testing.
• Specialised equipment for unsaturated soil testing is complicated and costly.
As a result of these, filter paper method was adopted in this study to measure pore air
and pore water pressure which equivalent to suction for the subgrade soils.

Page | 79

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

3.5.9 Repeated load triaxial test (RLTT)

This test was conducted as to evaluate the resilient modulus (𝑀𝑟 ) of the studied soils.
The soils were compacted at a different moisture content as obtained from the moisture-
density relationship. To explore subsequent impact of seasonal moisture variation in
pavement. Matric suction and 𝑀𝑟 relationship of each soil was developed.

Samples, obtained from material passing through No. 4 sieve were oven-dried, as to
eliminate possible hygroscopic moisture content. Once the samples had achieved
constant weight in the oven, they were removed and allowed to cool. The 𝑀𝑟 specimens
were compacted and tested at different moisture contents equivalent to the moisture
content gotten from compaction test. Standard RLTT was carried out to evaluate 𝑀𝑟 of
the investigated soils, following CSIR Protocols which is in line with AASHTO T 307, but
differs only on specimen size: Standard method of test resilient modulus of soils and
aggregate materials. The prepared specimens are 300 mm in height and 150 mm in
diameter (Figure 3.22). The machine includes loading frame and software that controls
the materials dynamic stress level.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟐: Laboratory specimen for Mr testing

Prior to specimen preparation, good quantity of the soil samples were oven-dried at a
temperature of 80° C. After which, the soil samples were passed through Sieve 4.75mm
size. The maximum particles size material passing sieve 4.75mm was utilized for
preparation of the 𝑀𝑟 specimens. Once the samples had been processed through the
sieve 4.75mm, and appropriate amount of demineralised water was added to achieve

Page | 80

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

the target moisture content. Following the addition of water, the samples were
thoroughly mixed, covered with plastic bags and left overnight to attain mellowing and
moisture equilibrium.
The specimens were compacted in the laboratory mould using a vibro-compaction
machine, in five layers of equal weight, with the quantities of moisture content obtained
from moisture-density relationship.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟑: Vibro − compactor setup

The weight of each specimens was recorded after compaction using scale balance, thus
height and diameter were of each specimen were measured using Vernier caliper to
ensure that the specimens maintained 2:1 ratio of height to diameter. The specimens
were wrap with plastic bags, labeled and kept in the chamber for three days to avoid
moisture lost (Figure 3.24).

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟒: Wrapped specimens awaiting resilient modulus testing

Page | 81

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Then the prepared specimens were tested one at a time by removing the plastic bag
and placing latex membrane around the specimen to protect the inside part of the triaxial
cell. A porous stone and filter paper were placed on the bottom base plate. The
specimen is then placed on top of the filter paper and porous stone, and another filter
paper and porous stone are placed on top of the specimen as well. Immediately, the
specimen, porous stones, and filter papers are in place on the base plate, the prepared
specimen was loaded to the triaxial frame.

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟓: Setting − up of resilient modulus equipment

The applied load is measured by RLT device through a load cell, which is positioned
around the specimen. This system helps minimise errors associated with the measured
loads. Three linear variable differential transducers (LVDTs) were placed between the
top plate, base plate and one on the load frame to record the axial displacements. The
LVDTs on the body of specimen is installed to decrease the amount of error in the
measured axial deformation. The filter paper and porous stone were allowed for free
drainage of water from the specimen and the drainage valves were kept close during
𝑀𝑟 testing.

Page | 82

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟑. 𝟐𝟔: Complete Repeated load triaxial device

Prior, to 𝑀𝑟 testing, shear strength (Mohr circles) of the subgrade soils using static tests
at a range of confining pressures was done. This is use as a basis of defining the stress
regimes to be used for repeated load testing. Harmonized test methods for laboratory
determination of resilient modulus for flexible pavement design,” specifies 0.2 -second
haversine load pulse and 0.8-second rest period. The total resilient axial deformation
response and applied deviator stress were measured. The loading system are function
of three different confining pressures, with five different cyclic deviatoric stresses applied
at each confining pressure, however, seven different cyclic deviatoric stresses were
used in this research. Moreover, each specimen was subjected to 15 different stress
states during the course of the testing. A total of 1000 cycles were applied during the
conditioning stage to remove imperfections on the top and bottom surface that might
occur during compaction. The cyclical load was applied in the form of a sine-wave
movement shaped load pulse and this loading sequence best represent the loading
conditions experienced by a pavement structure under vehicular or rail track loading.
During vehicular loading, some zone in the pavement experiences minimal deviatoric
stresses when the wheel load is at considerable distance away from that point. The
zone experiences the maximum deviatoric stress when the wheel load is directly on top.

The acquisition system records the data from the last seven load cycles at every stress
state, the data obtained from the last five cycles, at each stress, is averaged to provide
𝑀𝑟 value. The 𝑀𝑟 values were automatically obtained from the software using Equation

Page | 83

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

3.13, with each test providing 21 𝑀𝑟 values, at different deviatoric stress each
specimen.

𝜎𝑐𝑦𝑐
𝑀𝑟 = (3.13)
𝜀𝑟

Where:
𝜎𝑐𝑦𝑐 = Applied cyclical stress
𝜀𝑟 = Resilient strain

Following the completion of RLT testing, the specimens were carefully removed from
the triaxial cell and as well from the latex membrane, and moisture content of the
specimens were measured after 𝑀𝑟 testing was completed. The test was considered
acceptable as the moisture content of each tested specimens was within 0.5% of the
target moisture content and the dry density was within 1.3% of the target dry density.

Page | 84

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPTER 4: GEOTCHNICAL TEST RESULTS OF THE SOILS

4.1 Overview
The following results and analysis are based on the data gathered from experimental
tests conducted on the tested soils. The presented laboratory test results covered the
following geotechnical tests: free swell index, swelling stress, modified compaction,
suction, CBR, unsaturated CBR and UCS test. The laboratory testings analysis
illustrated the geotechnical properties of these subgrades and the results were
scientifically interpreted using an average of two specimens of each tested soils.

4.2 Density–Moisture Relationship

Soil compaction is one of the most critical components in the pavement construction,
airfields, rail tracks, embankments, earth dam and foundations. Stability of structure
depends mostly on the achievement of proper soil compaction. Failure to achieve proper
ground compaction, is somewhat causes failure of pavement, rail track and airfield.
However, higher compaction degree aggrandizes the geotechnical specifications of the
subgrade. This aids to achieve the desired degree of relative compaction, required to
ascertain specified properties, fittingly for shear strength resistance (Jaquin et al., 2009).
Compaction process has proven the link between the density-moisture relationships
with shear strength and helped in clarifying the aspects of cementation. Tarantino and
De Cole, (2008) reported on the response of clay materials under compaction as the
most important factor influencing compaction process (i.e. moisture status of soil).

4.2.1 Effect of moisture content on compacted FSS

Moisture-density relationships for FSS was established according to TMH1 method A7,
using modified proctor compaction technique. Basically, it was observed that the
density of the FSS decreases with increase in moisture content (Mc) and this lead to the
soil displaying flocculated structure at moisture content less than the OPT. Whereas, at
optimum moisture content (OPT) the soil portrayed a more aggrandize structure.
Though, moisture content on the dry side is low due to more predominant attractive
force compared to repulsive forces. This resulted into a flocculated structure as the Mc
increased beyond the optimum, as a result of increased repulsive forces that caused
the soil particles to orient into a more dispersed structure Figure 4.1.

Page | 85

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

20
FSS 1
18 FSS 2
16 FSS 3

14

12
𝛾𝑑 kN/m3

10

0
0 5 10 15 20 25 30 35
Moisture content (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏: Moisture − Density Relationship of FSS

4.2.2 Effect of moisture content on compacted NCS

Moisture content strongly influenced NCS (Figure 4.2), due to high percentage of fine
content in the soils > 50% with specific gravity of 2.69, 2.72 and 2.70 for NCS 1, 2 and
3 respectively. The curve showed that NCS portrayed bell-shaped compaction curves
which confirmed the soils, typical clayey soil. Furthermore, the dry unity weight of the
soil starts to drop as the moisture content begins to increase. This is caused by the
capillary force effects, as the pore space of the soils begins to inhibits moisture. The soil
particles tends to move around and portrayed less densely compacted soil matrix, due
to more absorbed moisture on the wet side of the curves. However, the soil possesses
a flocculent structure on the dry side of the optimum, as a result of the low moisture
content. While on the wet-side, a diffused behaviour of the soil was observed causing
the layers of ions surrounding the clay particles to be scattered. Hence, the interparticle
repulsion was reduced resulting in a more random particle orientation thereby caused
the soils to record higher MDD compared to FSS.

Page | 86

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

20
NCS 1
19
NCS 2
18 NCS 3
17

𝛾𝑑 kN/m3
16

15

14

13

12

11

10
0 5 10 15 20 25 30 35
Moisture Content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟐: Moisture − Density Relationship of NCS

4.2.3 Effect of moisture content on compacted GPS

Generally, the comprehensive results of compaction experiment displayed in the form
of moisture content-unit weight showed in Figures 4. Considerably varied, but showed
reasonable well-defined maximum dry unit weight at given moisture content. As the Mc
of the compacted soil increases, the diffuse moisture around the soil particles expanded.
This increased the repulsion forces within the clay particles of the soils and resulted in
higher degree of flocculation that caused higher dry unit weight on the dry side.

24
GPS 1
GPS 2
22
GPS 3

20
𝛾𝑑 kN/m3

18

16

14

12

10
0 5 10 15 20 25 30 35
Moisture content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑: Moisture − Density Relationship of GPS
Page | 87

© Central University of Technology, Free State

© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

show any form of cracking. Nonetheless, at a point before the optimum, no cracks were
observed. The post compaction behaviour of the soils is in line with the XDR result in
Appendix A. The result classified FSS as highly expansive and NCS as moderate
expansive. This is considered as one of the most serious challenges geotechnical
engineer faces, due to potential danger of unpredictable upward movements of the
subgrade. This swelling stress causes fatigue and cracks to the pavement structures
identified across Free state and Northern cape. Hence, swelling behaviour of the
compacted soils is importance, for pavements within this province. The high swelling
potential of this studied soils could be attributed to the soil capillarity, as the voids within
the soil structure are filled with moisture. Basically, during capillary process, moisture is
absorbed by the interlayer particles of the clay minerals causing an increase in volume.
After the moisture capillarity, voids are filled by the swollen expansive clay minerals like
montmorillonite, kaolinite, bentonite etc. As such, the montmorillonite within the soil
swells and the swelling increases as the amount of clay minerals increases until the
swelling pressure equilibrates with the amount of moulding moisture content. However,
after 3 days of compaction the total volume of the soil is restricted due to the compaction
efforts. Furthermore, at a period beyond 7 days the soils swell and occupied all the
volume available, thereby initiating cracking on the compacted soils.
The pressure caused by this process is known as the swelling pressure. Rationally, it
will be ideal to compact expansive subgrade at a point below the optimum moisture
content and provide high thickness of pavement wearing surface, in order to overcome
swelling pressure during pavement design.

4.3 Analysis of Swell Potentials

Determination of swelling stress in expansive subgrade soil is important for numerous
engineering applications. Damages caused per year by expansive soils has been
reported to be more than all other natural hazards combined, including earthquakes,
floods, and tornadoes (Chen, 1988). Expansion susceptible stress occurs as the Mc of
the subgrade soils increases. Upon expansion, the soil exerts an upward stress on
pavement and when this stress exceeds the limit state of stress in pavement, fatigue
and cracks manifest. Then uplift or differential uplift occurs causing pavement to cracks.
Therefore, assessing swelling pressure becomes important step in designing of
pavement on expansive soil. Two swelling tests were used in this study to measure the
swelling potential and swelling stress i.e. Free swell index test (FSI) and Zero swell test
Page | 89

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

(ZST) of the studied soils confirmed the tested soils to possess some expansive
characteristic.

4.3.1 Free swell index (FSI) results

FSI results of the studied soils is consistent with Atterberg limit results, which identified
FSS as high expansive soil, while NCS and GPS were quantified to be moderate with
low degree of expansiveness.

Table 4.1: Free Swell Index Results

Soil Samples Vk (ML) Vs (ML) FSI (%) Degree of
expansiveness
FSS 1 10 15 70 35-50 very high
FSS 2 10 15 80 Very high
FSS 3 10 15 65 Very high

NCS 1 10 15 70 35-50 high

NCS 2 10 15 80 Very high
NCS 3 10 15 80 Very high

GPS 1 10 15 30 20-35 moderate

GPS 2 10 15 50 50 high
GPS 3 10 15 20 <35 low

4.3.2 Zero swell stress result (𝑷𝒔 )

Figure 4.5 through 4.7 illustrates pressure stress against five sets of dry densities at
varying Mc for the studied soils. This result shows that variables such as dry density,
moisture content and percentage of clay fractions greatly affected swelling pressure of
the subgrade soils. Furthermore, FSS 1, 2 and 3 in Figure 4.5 showed that higher Ps is
obtained at lower moisture content and lower Ps is recorded at higher Mc. The soils
behaviour is in consistence with the report published by (Rasheed, 1985). Values of Ps
for different Mc are summarised in Table 4.2 Appendix B.

Page | 90

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

4.4.1 Total, matric and osmotic suction results

Filter paper method was adopted to measure suction of the studied soils. This technique
is on the basis that water potential of the soil reaches equilibrium with the filter paper
water potential. Whatman No.42 filter paper calibration curve developed for this study
was used to estimate the soil suction at equilibrium moisture content. Total and matric
suctions were determined on the samples compacted at varying different moisture
conditions. Whereas, osmotic suction was calculated from the difference between total
and matric suction.

Various properties of unsaturated soil, such as the swelling stress, the volume variation,
and hydraulic conductivity, is relative to moisture quantity within the soil’s voids for a
given soil potential. Thus, the relation between the water content (gravimetric water
content, volumetric water content) and the soil potential is an essential feature under
unsaturated soil mechanics. Suction curves against gravimetric moisture content for the
soil sampled FSS 1, 2 and 3 are presented in (Figures 4.8 through 4.10). Generally, it
was observed that total and matric suction recorded high suction values, due to low Mc.
Whereas, FSS 2 yielded the highest matric suction values among FSS 2 and 3. This
implied that high capillary stresses in the soils causes movement of water in an
unsaturated state. This replicates real filed condition for typical unsaturated pavement
structure. Thus, this result is consistence with the reported by Yang et al. (2005).
Furthermore, on the dry side of the optimum FSS 2 recorded the highest matric suction
value of 6517 kPa, while FSS 1 and 3 yielded 5941 kPa and 5598 kPa respectively. The
curves demonstrated that matric suction dominates 82% component of the soils total
suction (Leong et. al, 2003). This was as a result of soils capillarity, pore size distribution
and the physiochemical interactions, which are highly dependent on the soil mineralogy
that governs suction. The data for the curves is presented in Table 4.3, Appendix B.

Page | 94

© Central University of Technology, Free State

© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

10000
FSS 3 as compacted
Total suction
9000
Matric suction
8000
Osmotic suction
7000

6000
𝜓 (kPa)

5000

4000

3000

2000

1000

0
0 5 10 15 20 25 30

Gravimetric water content, 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟎: Suction, ψ versus gravimetric water content curve

Figure 4.11 through 4.14 depicts the suction matric behaviour of NCS. These curves
demonstrated that the suction response of the soils is associated with capillary action.
The interface tension between water and air within soils voids created a curve interface
boundary in a narrow opening, leading to a high suction pressure. The capillary rise that
occur within these soils was affected by its particle size and grading. It was observed
that NCS, sustained large pressure difference between pore water and air, allowing
large capillary rises, because of the qualities of soils fine-grained. The soils would have
behaved differently, if it contained coarse grained with larger voids.

Page | 96

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

4000
NCS 1 as compacted Total suction
3500 Matric suction

Osmotic suction
3000

2500
𝜓 (kPa)

2000

1500

1000

500

0
0 5 10 15 20 25 30
Gravimetric water content, 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟏: Suction, ψ versus gravimetric water content curve

5000
NCS 2 as compacted Total suction
4500
Matric suction
4000

3500

3000
𝜓 (kPa)

2500

2000

1500

1000

500

0
0 10 20 30 40

Gravimetric water content, 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟐: Suction, ψ versus gravimetric water content curve

Page | 97

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

3000
NCS 3 as compacted Total suction

Matric suction
2500
Osmotic suction

𝜓 (kPa) 2000

1500

1000

500

0
0 10 20 30 40

Gravimetric water content, 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟑: Suction, ψ versus gravimetric water content curve

The relation between the soil suctions total, matric, and osmotic suction with the
corresponding gravimetric water content was investigated for each tested soil and are
depicted graphically. The compacted soils suction i.e. GPS 1, 2 and 3 obtained by filter
paper method are presented in Figures 4.14 through 4.16. The curves of the tested soils
showed that total and matric suctions insidiously decreased with an increase in the initial
water content. The measured total and suction matric response of the tested GPS were
found to be similar to FSS and NCS. The difference between total and matric suctions
at low water content was as a result of proper contact between the filter paper and the
soil specimens. The measured total suctions were greater than the measured matric
suctions as was anticipated.

Page | 98

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

7000
GPS 1 as compacted
6000
Total suction
Matric suction
5000
Osmotic suction

4000
𝜓 kPa

3000

2000

1000

0
0 5 10 15 20 25 30 35

Gravimetric water content 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟒: Suction, ψ versus gravimetric water content curve

8000
GPS 2 as compacted
7000 Total suction

Matric suction
6000
Osmotic
suction
5000
𝜓 kPa

4000

3000

2000

1000

0
0 5 10 15 20 25 30 35

Gravimetric water content 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟓: Suction (ψ) versus gravimetric water content curve

Page | 99

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

8000
GPS 3 as compacted
7000
Total suction

6000 Matric suction

Osmotic suction
𝜓 kPa 5000

4000

3000

2000

1000

0
0 5 10 15 20 25 30

Gravimetric water content 𝜃𝑔 (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟏𝟔: Suction (ψ) versus gravimetric water content curve

Conclusively, the difference between total and matric suction values for FSS 1, 2 and 3
varied from 1742 kPa to 404 kPa, 1641 kPa to 721 kPa and 2056 kPa to 578 kPa
respectively from the dry side of the optimum (DOP) to the wet side of the optimum
(WOP). While, the differences between total and matric suctions for NCS 1, 2 and 3 on
the varies between 1064 kPa to 208 kPa, 1254 kPa to 490 kPa and 971 kPa to 208 kPa
respectively. The differences between the total and matric suction values for GPS 1, 2
and 3 changes between 1483 kPa to 344 kPa, 1808 kPa to 289 kPa and 1949 kPa to
235 kPa. This result show that fine content and moulding water content of the soils have
measurable effect on suction values of all the tested soils. In addition, the relationship
between the total, matric suction and the filter paper water content. are approximately
linear. This indicated decrease of suction with increase of the filter paper water content.

4.4.2 Soil-water retention curve (SWRC) of the tested soils

SWRC is the principle property of unsaturated soils (Fredlund et al., 2012). The SWRC
for the studied soils are plotted using suction matric and volumetric water content ((𝜃𝜈 )
as recommended by Fredlund et al. (2001) for geotechnical engineering Figures 4.17
through 4.25.

Page | 100

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Based on the SWRC in Figures 4.17 through 4.25 it is clear that the matric suction-𝜃𝜈
relationship is unique for each soil type due to the dependence of the soils pore size
distribution and the physiochemical interactions, which are highly dependent on the clay
minerals. Quick comparison of suction results with the soil plasticity test, revealed that
high PI soils recorded higher range of suction during desaturation as compared to the
soils with lower PI values that undergo desaturation over a small range of suction values.
This was due to capillary forces, which is the main water holding mechanism in low PI
soil and this dominates lower suction range of the soil. Whereas, surface adsorptive
forces play a large role in holding water in high PI soils. As such, the AEV of the tested
soils depends on the pore size distribution, such that soils with smaller pores have
higher AEV than soils with larger pores. This is evidenced in particles sizes distribution
analysis. Figures 3.3 through 3.5 show GPS with a significant silt content, yielded the
lowest AEV value amongst soils evaluated.

4.5 Analysis of CBR Results

CBR values are used as an indication of index bearing capacity of soil strength. These
values are predominantly used to execute the pavement design. Though CBR does not
represent fundamental soil property. Thus, CBR is relevant on the basis of point load
bearing capacity design method, rather than repetitive traffic load of pavement design.
Nicholson et al. (1994) reported that CBR test is a common index test used for
evaluation of strength of sub-base layers. In the same accord, CBR-test was conducted
to characterise strength and bearing capacity of the studied subgrades. The CBR is
influenced by different factors such as: moisture content, dry density, texture among
other factors is clay minerals and soil’s pore size distributions. The CBR test is
conducted in the laboratory, under soaked and unsoaked conditions on soil samples
prepared at the dry density and Mc likely to simulate field condition. The CBR test was
carried out in this research, in order to use it as one of the independent variables for the
prediction of resilient modulus.

Page | 106

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟐𝟔: Soaked and unsoaked compacted specimens

Table 4.7 below showed, the description of CBR values with respect to subgrades
qualities, this table serves as a measure to determine the quality of subgrades as
regards to the type of pavement required.

Table 4.7: Description of CBR-value subgrade quality

CBR-values Subgrade quality Subgrade status
0-3 very poor subgrade
3-7 poor to fair subgrade
7-20 fair subgrade
20-50 good subgrade
>50 excellent subgrade

4.5.1 Behaviour of unsoaked compacted soils

The unsoaked CBR test was performed as stipulated in South African standard
(Technical Methods for Highway (TMH) method 8, 1986). The load (kN) and penetration
(mm) reading from a circular piston of 1935mm 2 were observed at a rate of 1,27 mm
per minute. The loads at 2.54 mm, 5.08 mm and 7.62 mm were expressed as a
percentage of standard load value for respective deformation level. CBR test were
conducted at different dry densities and moisture content, as to evaluate the response
of the subgrades over range of moisture content. The test results of unsoaked and
soaked CBR are summarised in Tables in 4.8 and 4.9.

Page | 107

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The calculation depicted the soils as fair subgrades from the dry side of the optimum.
Whereas on the wet side, the subgrades are designated as poor to fair. The soils with
clay fraction greater than 70%, recorded relatively smaller swelling values. Therefore,
potential swelling for soil containing fine grain smaller than 70% is relatively small. At
optimum moisture content (OPT), the CBR values of FSS 1, 2 and 3 are 12.59%,11.79%
and 13.52% respectively. Whilst on the dry side of the optimum (DOP) the soils recorded
the highest CBR values of 15.65%, 13.71% and 17.46%. Furthermore, wet side of the
optimum (WOP) FSS 1, 2 and 3 obtained CBR values of 8.43%, 6.52% and 10.83%
respectively. However, similar trend was observed in all the studied subgrades. In
addition, significant decrease was observed in CBR values of the subgrades as the Mc
increases. This could be ascribed to the arrangements of the soil particles, mineralogy
and desperation of the soil particles because on increased moisture content. However,
tabulated results of both the soaked and unsoaked subgrade soils are presented in
Tables 4.8 and 4.9.

Table 4.8: Four days Unsoaked Soils for California Bearing Ration (CBR)
BEARING RATIO AT= SWELL
𝑻𝒆𝒔𝒕 𝑳𝒐𝒂𝒅 (%) MOD. AASHTO COMPACTION DATA
× 𝟏𝟎𝟎
𝑺𝒕𝒂𝒏𝒅𝒂𝒓𝒅 𝒍𝒐𝒂𝒅 DATA

Soil CBR CBR CBR MDD OPTIMUM DRY COMPACTION

moisture (%) at (%) at (%) at 3
(Kg/m ) MOISTURE DENSITY S (%) MOISTURE
conditions: 2.54mm 5.08mm 7.62mm (%) (Kg/m3) QUANTITY
(g)
FSS 1
8.55 14.65 13.23 11.75 0.58 1766 17.65 1269 95 767
13.46 13.11 12.52 12.21 0.52 1766 17.65 1695 95 798
17.65 12.59 11.73 10.47 0.41 1766 17.65 1766 95 815
24.27 10.23 10.16 9.35 0.37 1766 17.65 1621 95 833
31.75 8.43 6.45 4.11 0.33 1766 17.65 1323 95 859

FSS 2
9.97 12.11 11.22 10.09 0.65 1726 18.78 1271 95 799
13.67 11.11 11.54 10.42 0.58 1726 18.78 1592 95 812
18.78 10.19 9.76 8.96 0.52 1726 18.78 1726 95 832
24.09 8.14 6.83 4.28 0.46 1726 18.78 1443 95 856
30.16 6.52 4.56 3.49 0.44 1726 18.78 1236 95 894

FSS 3
7.28 17.46 15.33 13.79 0.43 1895 18.45 1371 95 729
14.15 15.39 13.70 12.33 0.40 1895 18.45 1795 95 754
18.45 13.52 11.86 11.02 0.36 1895 18.45 1895 95 787
22.94 11.95 11.09 10.67 0.33 1895 18.45 1726 95 808
28.33 10.33 8.36 6.87 0.28 1895 18.45 1373 95 834
Page | 108

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

NCS 1
8.38 17.52 13.56 11.34 0.78 1972 17.49 1638 95 727
12.73 15.49 15.76 14.19 0.67 1972 17.49 1929 95 815
17.49 13.66 11.87 10.67 0.52 1972 17.49 1972 95 833
24.49 12.83 10.99 8.73 0.48 1972 17.49 1785 95 856
28.33 10.75 7.78 5.88 0.40 1972 17.49 1588 95 892

NCS 2
9.18 14.14 13.34 12.83 0.58 1888 17.38 1575 95 712
13.13 12.44 11.58 10.48 0.55 1888 17.38 1786 95 789
17.38 11.45 11.16 10.43 0.49 1888 17.38 1888 95 816
24.09 10.54 9.88 8.49 0.40 1888 17.38 1722 95 834
30.22 9.34 7.66 6.23 0.33 1888 17.38 1588 95 866

NCS 3
9.97 12.34 11.59 10.88 0.53 1716 18.98 1445 95 718
14.15 10.89 11.45 10.32 0.47 1716 18.98 1617 95 756
18.98 9.74 8.73 7.89 0.43 1716 18.98 1716 95 788
24.64 8.83 6.45 5.56 0.40 1716 18.98 1626 95 809
32.88 7.67 6.84 4.56 0.36 1716 18.98 1327 95 818

GPS 1
9.28 15.21 13.45 11.41 0.47 2225 18.76 1750 95 728
12.53 13.54 12.56 10.39 0.43 2225 18.76 1984 95 743
18.76 12.18 11.92 9.49 0.23 2225 18.76 2225 95 779
23.94 11.48 10.17 8.66 0.20 2225 18.76 2046 95 794
29.33 10.33 10.13 8.45 0.12 2225 18.76 1647 95 807

GPS 2
9.89 12.16 11.56 8.76 0.52 2130 18.88 1575 95 743
14.15 11.88 9.62 6.34 0.48 2130 18.88 1960 95 756
18.88 10.84 8.88 6.63 0.43 2130 18.88 2130 95 776
24.09 10.31 8.48 6.31 0.40 2130 18.88 1933 95 818
30.33 9.21 6.76 4.33 0.38 2130 18.88 1627 95 883

GPS 3
8.67 18.45 16.32 13.83 0.33 2328 16.77 1909 95 703
11.89 16.43 11.55 10.45 0.28 2328 16.77 2052 95 728
16.77 14.78 14.52 11.93 0.40 2328 16.77 2328 95 755
22.65 12.84 11.67 10.42 0.38 2328 16.77 2213 95 768
28.13 11.94 10.89 9.56 0.32 2328 16.77 1920 95 793

4.5.2 Behaviour of soaked compacted soils

The CBR values of soaked compacted soils are commonly used parameter for design
of highway pavement, in the sense that it simulates worst field condition. The soaked
CBR values of the subgrades dropped tremendously compared to unsoaked specimens.
This was as result of absorbed water, that eventually led to higher voids ratio which
caused the soils to yield lower CBR values.

Page | 109

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Under soaked condition, the soils are considered as poor to fair subgrades from dry side
to the optimum and very poor from optimum to wet side. Soaked CBR values of FSS 1,
2 and 3 decrease with respect to unsoaked CBR are 4.05%, 3.96% and 4.02% on the
dry side of the optimum. This, indicated that on the dry side, the soil voids were less and
the soil particles were more flocculated, thus causing the soils to be less affected by
water during wetting period. In addition, the same trend of decrease in soaked CBR
values with respect to unsoaked CBR was observed for the rest of the studied
subgrades. The CBR value continues to decrease from dry side to the OMC up to the
wet side, thus decrease in CBR values in soaked condition on dry side of the optimum
for NCS 1, 2 and 3 were evaluated to be 4.95%, 4% and 1.04. These values showed
that the difference in CBR values for soaked specimens continues to increase as the
testing shifts from dry side to wet side.

This CBR-value of the soaked compacted soils was reported in the context of the
general relationship between the CBR values and geotechnical quality of the soils used
in pavement applications (Bowles, 1992). These values showed that unsoaked soils at
wet side are classified as good subgrades, whilst under soaked condition, the soils are
categorised as poor to fair. However, this CBR values cannot be recommended for the
design pavement. Hence, it does not meet the minimum requirement for CBR design
values. Thus, significant decrease in CBR value of the soaked subgrade soils were high
compared to the unsoaked soils. This can be ascribed to the deformation of capillary
forces under soaked conditions. Generally, the CBR values decreases significantly with
number of soaking periods.

Table 4.9: Four days Soaked Soils for California Bearing Ration (CBR)
BEARING RATIO AT=
𝑻𝒆𝒔𝒕 𝑳𝒐𝒂𝒅 SWELL MOD. AASHTO COMPACTION DATA
× 𝟏𝟎𝟎 (%) DATA
𝑺𝒕𝒂𝒏𝒅𝒂𝒓𝒅 𝒍𝒐𝒂𝒅
Soil CBR CBR CBR MDD OPTIMUM DRY COMPACTION
moisture (%) at (%) at (%) at (Kg/m3) MOISTURE DENSITY S (%) MOISTURE
conditions: 2.54mm 5.08mm 7.62mm (%) (Kg/m3) QUANTITY
(g)
FSS 1
8.88 11.60 10.14 9.83 0.62 1772 17.93 1293 95 770
13.18 10.11 8.45 6.65 0.58 1772 17.93 1680 95 794
17.93 9.59 7.73 5.47 0.50 1772 17.93 1772 95 818
24.87 8.28 6.77 4.36 0.45 1772 17.93 1636 95 830
31.40 4.53 3.56 2.88 0.40 1772 17.93 1330 95 862

FSS 2
9.28 9.73 7.28 5.15 0.68 1729 18.29 1274 95 801
Page | 110

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

13.82 7.80 5.58 3.42 0.60 1729 18.29 1589 95 809

18.29 5.79 4.76 3.96 0.55 1726 18.29 1729 95 835
24.33 5.14 4.54 3.23 0.50 1726 18.29 1440 95 859
29.71 4.52 3.28 2.60 0.48 1726 18.29 1233 95 891

FSS 3
7.49 13.44 11.55 9.88 0.46 1891 18.63 1388 95 733
14.09 11.53 9.65 7.28 0.43 1891 18.63 1782 95 759
18.63 10.47 8.33 6.82 0.37 1891 18.63 1891 95 790
22.68 9.68 7.87 5.65 0.35 1891 18.63 1737 95 812
28.18 7.66 4.88 3.86 0.30 1891 18.63 1392 95 837

NCS 1
8.47 13.31 11.59 10.44 0.75 1975 17.45 1641 95 731
12.69 12.94 10.66 9.45 0.70 1975 17.45 1926 95 812
17.45 10.66 9.89 8.75 0.55 1975 17.45 1975 95 836
24.71 8.45 6.99 4.16 0.51 1975 17.45 1788 95 853
28.25 7.47 5.78 3.88 0.45 1975 17.45 1585 95 895

NCS 2
9.36 11.35 10.34 8.83 0.60 1885 17.55 1578 95 714
12.92 10.55 9.60 7.46 0.55 1885 17.55 1790 95 791
17.55 9.68 7.59 5.48 0.50 1885 17.55 1885 95 813
23.87 7.43 6.88 4.49 0.44 1885 17.55 1725 95 837
30.10 6.98 4.66 3.53 0.38 1885 17.55 1585 95 863

NCS 3
9.89 9.84 8.29 6.45 0.56 1716 18.66 1445 95 718
14.38 9.17 8.33 7.22 0.50 1716 18.66 1617 95 756
18.66 7.48 5.66 3.45 0.47 1716 18.66 1716 95 788
24.48 5.87 4.32 3.76 0.45 1716 18.66 1626 95 809
32.67 4.94 2.84 1.59 0.40 1716 18.66 1327 95 818

GPS 1
9.32 13.67 11.98 10.41 0.52 2225 18.56 1750 95 730
12.34 12.54 10.86 8.57 0.49 2225 18.56 1984 95 740
18.56 10.66 8.78 7.88 0.35 2225 18.56 2225 95 783
23.85 9.98 7.17 5.69 0.30 2225 18.56 2046 95 790
29.49 7.47 6.13 5.15 0.20 2225 18.56 1647 95 815

GPS 2
9.84 11.23 10.76 9.89 0.55 2132 18.91 1571 95 746
14.18 10.13 8.86 5.89 0.50 2132 18.91 1963 95 73
18.91 9.45 7.75 5.68 0.48 2132 18.91 2132 95 779
24.03 7.31 5.41 4.54 0.44 2132 18.91 1928 95 813
30. 21 5.89 6.99 3.98 0.40 2132 18.91 1623 95 880

GPS 3
8.46 15.55 13.44 11.86 0.36 2331 16.85 1912 95 708
11.65 13.96 10.63 8.45 0.30 2331 16.85 2049 95 725
16.85 12.73 10.52 7.93 0.45 2331 16.85 2331 95 758
22.63 11.84 9.67 7.49 0.38 2331 16.85 2209 95 763
28.28 10.94 8.89 6.56 0.32 2331 16.85 1923 95 795

Page | 111

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

16
NCS 1
OMC
14 NCS 2

NCS 3
12

Soaked CBR (%) Dry side Wet side

10

4
OMC

0
0 10 20 30 40

Moisture content, (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟏: Soaked NCS CBR Variation with water content

18
GPS 1
OMC GPS 2
16
GPS 3
14 Dry side
Soaked CBR (%)

Wet side
12

10

6
OMC
4

0
0 5 10 15 20 25 30 35
Moisture content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟐: Soaked GPS CBR Variation with water content

Consequently, the influence of matric suction on CBR of compacted soils were

investigated as part of the unsaturated CBR test. Sets of identical specimens were
prepared for both soaked and unsoaked CBR test. One set of the prepared samples

Page | 115

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

were used for CBR testing, while the other set was used for suction measurement (filter
paper method Whitman No. 42).

4.5.4 Unsaturated CBR effect on matric suction

The result portrayed that decrease in moisture content, led to the increase in matric
suction and subsequently caused increase in CBR value. Figures 4.33 through 4.35
explained moisture content decrease due to drying, and this caused the rise in capillary
pressure within the soil’s particles thus led to increase in matric suction. Probably, a bi-
linear relationship between CBR and matric suction was observed from the curve.

20
FSS 1
18 FSS 2
FSS 3
16

14
Unsoaked CBR (%)

12

10

0
100 1000 10000

𝜓𝑚 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟑: Variation of unsoaked FSS CBR versus matric suction

Page | 116

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

20
NCS 1
18 NCS 2
NCS 3
16

14
Unsoaked CBR (%)
12

10

0
100 1000 10000

𝜓𝑚 (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟒: Variation of unsoaked NCS CBR versus matric suction

20
GPS 1
18
GPS 2
16
GPS 3
Unsoaked CBR (%)

14

12

10

0
100 1000 10000
𝜓𝑚 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟓: Variation of unsoaked GPS CBR versus matricsuction

Page | 117

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Furthermore, the experimental results in Figures 5.34 through 4.35 showed increment
of matric suction due to the drying process that increased CBR values of unsoaked
specimens. However, the CBR values increases as matric suction increased along the
drying and wetting paths. This trend agrees with the observations of Paraire (1987) and
Ampadu (2007) for the soil samples examined along the wetting paths. Based on this
extrapolation of the dry densities used in this study and the corresponding matric suction
values, (𝑢𝑎 − 𝑢𝑤 ), all test points were estimated using Equation 4.1. The suction matric
values are plotted against the unsaturated CBR, (CBRu) in Figures 4.36 and 4.38. The
results portrayed an equation that is expressed as:

𝑢𝑎 −𝑢𝑤 𝑛
𝐶𝐵𝑅𝑢 = 𝐶𝐵𝑅𝑠 × ( ) (4.1)
𝑢𝑒

where 𝐶𝐵𝑅𝑠 is the soaked CBR, 𝑢𝑒 is the air-entry value and 𝑛 is the regression
parameters for the model presented in Equation 4.1. A regression model was developed
for parameter 𝑛 and as well used to optimise the parameter 𝑛 in Equation 4.1. The
values for parameter 𝑛, in this study ranges from 0.21 to 0.382. This values differs from
Ampadu’s values for parameter 𝑛, that ranges from the order of 1.4 and about 0.5, for
lower and for higher dry densities respectively, and constant for suction values up to
about 15,000 kPa. The parameter 𝑛 obtained from the regression model differs from
that proposed by Ampadu (2007). The differences in values could be attributed to soil
type, variation in moisture content and dry densities. However, the results of CBRu were
discovered to be 1.5 to 2 times higher than that of the conventional CBR values.
Whereas, the CBRu values obtained using the developed models were found to be close
to the predicted unsaturated CBR values obtained from Ampadu’s models.

Page | 118

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

16
FSS 1
14 FSS 2
FSS 3
12

Soaked CBR (%)

10

0
100 1000 10000

𝜓𝑚 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟔: Variation of soaked FSS CBR versus matric suction

16
NCS 1
14 NCS 2

12 NCS 3
Soaked CBR (%)

10

0
100 1000 10000

𝜓𝑚 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟕: Variation of soaked NCS CBR versus matric suction

Page | 119

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

18
GPS 1
16
GPS 2
14 GPS 3

Soaked CBR (%)

12

10

0
100 1000 10000
𝜓𝑚 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟖: Variation of soaked GPS CBR versus matric suction

4.5.5 Developed predictive equation for unsaturated CBR

NCSS 11 mathematical software was used in this study and the software package has
excellent predictability analytical capacity and graphical features. The interrelationship
of variables is formulated by an equation that linked the dependent variable and one or
multiples predictor variables. The response variable is designated as 𝑍 and the set of
the predictor variables are denoted as 𝑌1 , 𝑌2 , … … . 𝑌𝑛 , where n depicts the number of
predictor variable. Accurate relationship between 𝑍 and 𝑌1 , 𝑌2 , … … . 𝑌𝑛 can be
approximated by the regression model.

𝑍 = 𝑓(𝑌1 , 𝑌2 , … . . 𝑌𝑛 ) + 𝜀, (4.2)

Where ε is assumed to be the random error representing approximation divergence and

it accounts for the failure of the model that failed to fittingly optimize the data. The
function 𝑓(𝑌1 , 𝑌2 , … … . 𝑌𝑛 ) describes the relationship between 𝑍 and 𝑌1 , 𝑌2 , … . . 𝑌𝑛 . An
example is the linear regression model.

𝑍 = 𝛼0 + 𝛼1 𝑌1 + 𝛼2 𝑌2 + ⋯ … . . +𝛼𝑛 𝑌𝑛 + 𝜀, (4.3)

Page | 120

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where 𝛼0 , 𝛼1 , … … . . , 𝛼𝑛 , are known as the regression parameters and are usually

determined from the data. The predictor variables are also known as independent
variables, regressors, factors and carriers. Most commonly, the term independent
variable is used during regression exercise.
The NCSS software was used to develop a mathematical predictive models for CBRu,
using suction matric, dry density, AEV and soaked CBR (CBRs) as independent
variables. While, CBRu is used as dependent variable. The regression analysis results
demonstrate that AEV has the highest effect on CBR u, followed by matric suction, dry
density and lastly CBRs. The R2 for each independent variable are presented in Table
4.10. The developed models are presented as follow:

CBR u (%) = −β0 + β1 (CBR s ) − β2 (γd ) − β3 log(ψm ) + β4 (Se ) … … … … … … … … … … (4.4)

CBR u (%) = −ξ0 + ξ1 log(ψm ) − ξ2 (𝑃𝑠 ) + ξ3 (𝐶𝐵𝑅𝑠 ) + ξ4 (𝑆𝑒 ) + ξ5 (𝛾𝑑 ) … … … … … … . (4.5)

Where: CBRu is the unsaturated CBR and the values of each symbol are summarized
in the table below:

Table 4.10: Parameters symbols and their corresponding values

Denotations Meaning Values R2
Equation 4.4
−𝜷𝟎 Intercept 16.9485 -
𝛃𝟏 Coff. for CBRs 4.6696 -
𝐂𝐁𝐑 𝐬 CBRs - 0.8511
𝛃𝟐 Coff. for dry density 0.9766 -
𝛄𝐝 Dry density 0.9182
𝛃𝟑 Coff. for suction matric 0.0001 -
𝛙𝐦 Suction matric - 0.9234
𝛃𝟒 Coff. for Air-entry values (AEV) 0.1085 -
𝑺𝒆 Air-entry values - 0.9332

Equation 4.5
−𝛏𝟎 Intercept -14.9392 -
𝛏𝟏 Coff. for matric suction 6.237E-05 -
𝛙𝐦 Matric suction - 0.9415
𝛏𝟐 Coff. for swelling pressure 0.0016 -
𝑷𝒔 Swelling pressure - 0.7865
𝛏𝟑 Coff. for CBRs 4.4837 -
𝐂𝐁𝐑 𝐬 CBRs - 0.9177
𝛏𝟒 Coff. for Air-entry values (AEV) 0.0983 -
𝑺𝒆 Air-entry values - 0.9195
𝛏𝟓 Coff. for Dry density 0.9173 -
𝛄𝐝 Dry density - 0.9501

Page | 121

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The curve of measured CBRu against predicted CBRu values showed that Models 1 and
2 closely predicted the measured values because of the convergence data with high R2
values of 0.9496 (Figure 4.39).

55
y = 0.9824x + 0.1161
50 R² = 0.9496

45
Measured CBRu (%)

40

35

30

25

20

15

10

5
5 15 25 35 45 55
Predicted CBRu (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟑𝟗: Measured versus predicted unsaturated CBRu model 1

55
y = 1.0071x - 0.18
50
R² = 0.9499
45
Measured CBRu (%)

40

35

30

25

20

15

10

5
5 15 25 35 45 55

Predicted CBRu (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟒𝟎: Measured versus predicted unsaturated CBRu model 2

Page | 122

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Table 4.11: Values of unsaturated CBR

Soaked Dry Measured Model 1 Model 2

samples density 𝐂𝐁𝐑 𝐬 𝛙𝐦 𝐧 𝑺𝐞 𝐂𝐁𝐑 𝐮 (%) (%) (%)
(%) (%) (kPa) (kPa)
Authors Model
FSS 1
1293 11.60 6541 0.30 132 37.41 38.63 37.91
1680 10.11 5976 0.28 132 29.40 27.66 27.52
1772 9.59 4689 0.26 132 24.26 24.61 24.71
1636 8.28 2793 0.25 132 17.76 20.19 20.12
1330 4.53 921 0.33 132 8.60 5.85 6.10

FSS 2
1274 9.73 8517 0.25 158 26.36 32.48 31.86
1589 7.80 6843 0.23 158 18.56 20.58 20.45
1729 5.79 4989 0.21 158 11.95 10.95 10.32
1440 5.14 2895 0.26 158 10.95 10.05 10.11
1233 4.52 1913 0.30 158 9.55 9.24 9.37

FSS 3 1388 13.44 6398 0.30 113 45.11 44.20 43.68

1782 11.53 5345 0.28 113 33.95 31.34 31.32
1891 10.47 3450 0.26 113 25.47 25.65 25.64
1737 9.68 1998 0.24 113 19.29 23.75 23.63
1392 7.66 1240 0.29 113 15.34 17.77 17.23

NCS 1
1641 13.31 2534 0.35 101 41.11 40.18 39.04
1926 12.94 2098 0.33 101 35.21 35.72 34.82
1975 10.66 1643 0.30 101 24.61 24.70 24.30
1788 8.45 1134 0.28 101 16.63 16.24 16.16
1585 7.47 778 0.35 101 15.26 13.00 13.13

NCS 2
1578 11.35 3278 0.38 128 38.92 34.48 33.55
1790 10.55 2465 0.36 128 30.60 28.81 28.22
1885 9.68 2087 0.34 128 26.44 23.82 23.46
1725 7.43 1598 0.32 128 16.67 16.83 16.71
1585 6.98 1091 0.35 128 14.78 14.20 14.15

NCS 3
1445 9.84 1895 0.36 135 25.47 29.60 28.23
1617 9.17 1525 0.34 135 20.91 24.86 23.82
1716 7.48 1272 0.32 135 15.33 16.05 15.46
1626 5.87 1010 0.30 135 10.74 9.42 9.24
1327 4.94 681 0.35 135 8.70 7.98 7.93

GPS 1
1750 13.67 5071 0.31 124 43.19 42.99 42.32
1984 12.54 3884 0.28 124 32.90 35.61 35.12
2225 10.66 2959 0.27 124 25.10 24.63 24.60
2046 9.98 2183 0.25 124 20.44 23.26 23.16
1647 7.47 1128 0.34 124 15.83 15.48 15.58

GPS 2
1571 11.23 5263 0.33 145 36.74 35.53 34.77
Page | 123

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

1963 10.13 3989 0.31 145 28.30 26.85 26.37

2132 9.45 2914 0.29 145 22.56 22.18 21.86
1928 7.31 2029 0.27 145 14.91 14.17 14.10
1623 5.89 667 0.32 145 9.60 10.63 10.56

GPS 3
1912 15.55 5331 0.30 102 50.96 47.85 47.12
2049 13.96 3708 0.28 102 38.18 39.22 38.76
2331 12.73 2707 0.26 102 29.86 30.95 30.74
2209 11.84 1715 0.24 102 23.31 28.00 27.88
1923 10.94 763 0.31 102 20.42 26.73 26.59

In summary, pavement design procedures which are commonly based on CBR test, can
be extended to account for suction under unsaturated condition. Moreover, resilient
modulus (𝑀𝑟) test is widely accepted in the design of pavements in various
transportation agencies around the world. However, determination of 𝑀𝑟 value is costly,
time-consuming, and somewhat complicated. Based on this reason, Unsaturated CBR
test can be used as an alternative, as demonstrated in is study. Recommendation for
the interpretation of CBR results considering the influence of suction, gave an insightful
knowledge about suction correlation with CBR. These results are encouraging as they
not only provide a valid frame work to understand the influence of soil suction on the
engineering behaviour of highway pavements, but proved to be reliable for unsaturated
soil mechanics.

4.6 Unconfined Compressive Strength (UCS) Results

Compressive strength of soil is a significant factor, used in evaluating subgrade strength
of pavement. UCS test was prepared according to TMH1 METHOD 14. The studied
subgrade soils were demoulded after compaction and kept in a curing chamber for
7days, as to evaluate the effects of moulding Mc on soil’s strength. The result revealed
that the compressive strength of the subgrades decreases as water contented
increases. Furthermore, the strength gained by these soils were as result of compaction
effort and the soils are categorized as stiff and very stiff subgrade according to (Das,
1994) classification.
The results demonstrated that the densest specimens absorbed less water and were
found to have higher compressive strength. Similarly, the least dense material (soils
with higher fine content) tends to have more water intake and this led to lower
compressive strength values.

Page | 124

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The curves in Figures 4.41 through 4.43 indicated that density and strength are highly
sensitive to moulding water content. Furthermore, the curves demonstrated that the
effect of moulding water content is prominent for specimens with higher percentage of
fine. This showed that specimens compacted on the dry side portrayed higher strength
than those specimens compacted on the wet-side.

400
FSS 1
350 FSS 2
FSS 3
300

250
𝑞𝑢 (kPa)

200

150

100

50

0
0 10 20 30 40
Moisture content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟒𝟏: FSS seven days UCS values

400
NCS 1
350 NCS 2
NCS 3
300

250
𝑞𝑢 (kPa)

200

150

100

50

0
0 10 20 30 40
Moisture content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟒𝟐: NCS seven days UCS values
Page | 125

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

450
GPSS 1
400 GPS 2
GPS 3
350

300

250
𝑞𝑢 (kPa)

200

150

100

50

0
0 10 20 30 40

Moisture content, (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟒. 𝟒𝟑: GPS seven days UCS values

The low strength values with the increased moulding water content (wet-side) can be
attributed to relatively high-water content. Thus, on the wet-side soil particles of the
specimen dispersed, thereby becomes too difficult to be compacted especially with the
soils that contained higher amount of expansive clay minerals. Whilst on the dry side,
the soil particles fussed together resulting in closed packed pore water within the soil,
this caused the soil particles to be densest and thus led to more compacted specimens
that yielded relatively high strength.

Page | 126

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPTER 5: CONSOLIDATED UNDRAINED TEST RESULT

5.1 Overview
This part of the experimental work, gives emphases on the changes in shear strength
parameters with respect to moulding water content and matric suction. Further
description of Mohr-Coulomb failure criteria with shear strength parameters i.e. friction
angles, (𝜙 0 ) and cohesion, C’ under unsaturated soil mechanics for studied soils were
also illustrated.

5.2 Saturated Soil Shear Strength (𝝉) Result

Shear stresses are induced as the soil is loaded, thus when shear stresses reach a
limiting value, shear deformation sets in, leading to soil failure. Therefore, shear strength
is the resistance of soil to deform by continuous shear displacement of the soil particles
upon the action of stress. The failure conditions for soil is expressed as ultimate shear
stress, known as shear strength. Nonetheless, series of consolidated undrained shear
test on the prepared specimens was performed to obtain effective 𝜏 parameters (c’ and
𝜙 0 ) for all the tested soils at different Mc. Three different confining stresses i.e. 20kPa,
50kPa, and 100kPa were considered, to significantly evaluate the behaviour of the
studied soils at different confining stresses.

5.2.1 Stress- strain response of the studied soils

During consolidated undrained test, the stress-displacement responses of the nine
tested subgrades were assessed. The soils were tested at different Mc, ranging from
the dry to wet side of the optimum. The influence of Mc on stress-strain behaviour of the
tested soils were evaluated. Figures 5.1 through 5.27 in Appendix C shows stress-strain
as a function of vertical displacement for FSS, NCS and GPS at normal stresses (𝜎𝑛 ) of
20, 50 and 100kPa. The strain at failure was found to be within 0.030 to 0.035%, at 𝜎𝑛
of 20kPa for FSS1 on the dry side. Furthermore, the strain at failure was obtained with
the range of 0.06% to 0.07% at 𝜎𝑛 of 50k and 0.12% to 0.14% at 𝜎𝑛 of 100 kPa for FSS
2 and 3 respectively. The shear stress increases with respect to strain up until the
maximum value and at the point beyond failure strain. The stress start to decrease at
faster rate (residual stress), hence increasing the elastic modulus of the soils. The
behaviour was as a result of the flocculated structure formed on the dry side, as the soil

Page | 127

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

particles oriented randomly, with an increase in stress and suddenly failed at a higher
stress value.
All tested soils followed similar trend, due to gradually increase of stress with respect
to increase in strain to the ultimate stress capacity of each soil. Furthermore, the soil
structures were dispersed, as the soil particles were more in a parallel arrangement
perpendicular to the direction of applied stress for the specimens prepared with higher
moisture content. The stress-strain curves of the soils portrayed brittle stress-strain
behaviour on dry side of the optimum. Whereas, ductile responses were observed on
the wet-side of the optimum, this result agrees with the report published by (Fang, 2006).

5.2.2 Effects of moulding water content on shear strength parameters

Changes in shear strength parameters with varying moisture content were assessed.
The results demonstrated the behaviour of the studied soils at different Mc, ranging from
dry side of the optimum (DOP) to the wet side (WOP) with respect to shear strength
parameters i.e. c’ and 𝜙 ′ . The parameter values at 8.55% water content for specimen
FSS1, gave . c’ and 𝜙 ′ 58.10 kPa and 49.290 values. At 13.46% Mc, significant change
was observed, thereby causing the parameter values to decrease to the values of 53.52
kPa and 43.380 c’ and 𝜙 0 . More changes were recorded in shear strength parameters
at the optimum moisture content (OPT). When compared with strength parameters
values at 8.55% Mc as the soil yielded 51.2 kPa and 41.40 for both c’ and 𝜙 0 . Whereas,
at 24.27% Mc, the soil shear strength parameter values were measured to be 46.59
kPa and 37.670. Furthermore, at 31.75% Mc, . c’ and 𝜙 ′ values decrease of 8.1 kPa and
6.620 were recorded. Therefore causing the soil to yield shear strength parameter
values of 43.11 kPa and 34.780 respectively for c’ and 𝜙 0 . Furthermore, the FSS
recorded lower shear strength parameters, compared to NCS and GPS, as FSS
contained higher clay content than rest of the soils. Similar trends were observed on the
rest of the tested soils, as the soils exhibits bilinear relationship between the shear stress
and bulk stress for Mohr-Coulomb failure envelop as the Mc increases. The failure
envelops of the studied soils are presented in Appendix C, ranging from Figures 5.28
through 5.50.

Page | 128

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Generally, the shear strength response of the studied soils is attributed to the amount
of clay fraction and clay minerals within the subgrades, as the soils with higher content
of expandable clay i.e. montmorillonite reduces the shear strength. Thus, shear strength
is basically low for fully expanded clays (Morrow et al., 1984).
The shear strength parameters of the studied soils were obtained automatically at each
Mc from the intron’s software and failure envelopes were produced by plotting shear
stress and bulk stress with a tangent line drawn to the failure stress circles. The shear
strength values for each tested soil were calculated using equation 2.19. The shear
strength parameter of the saturated and unsaturated tested soils is summarised in Table
5.1 in Appendix C.

5.2.3 Shear strength and swelling pressure relationship

Shear strength is important engineering properties of soil and it is the property, that
materially influences the bearing capacity of pavement. Figures 5.73 through 5.75
showed that an exponential relationship exists between shear strength and swelling
pressure. The results proved that swelling pressure reduces shear strength of
compacted soil. This implies that increased swelling pressure in subgrade, causes
decrease in shear resistance of the subgrade and this will in turn trigger fatigues that
eventually will lead to cracking. Considering the swelling potential of the compacted
studied soils, the shape of the exponential shear strength versus swell pressure curve
demonstrated a smooth decrease in shear strength values with increase in moisture
content for all the studied soils. Alternatively, exponential relation between shear
strength and swelling pressure for sample FSS, which recorded the highest swelling
potential values among the other studied soils, showed a sudden decrease in shear
strength of 16% and 28 % when compared to NCS and GPS respectively. This response
is associated with the quantities of expansive clay mineral that are present in FSS. Thus,
FSS has the highest content of expandable clay montmorillonite according to XRD
results in Appendix A. This result is consistent with the report published by (Morrow et
al., 1984) which stated that the expandable clay montmorillonite is by far the weakest of
the clay minerals. Therefore, the soil-moisture reaction of the is clay minerals, will
reduce shear strength of soils due to swelling potentials.

Page | 129

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

700
FSS 1
y = 1117.5e-0 003x
600 R² = 0.9895 FSS 2

FSS 3

Shear Strength (kPa)

500
y= 719.79e-0 002x
R² = 0.9844
400
y = 1885.6e-0.004x
R² = 0.9482
300

200

100

0
100 300 500 700 900 1100
Swelling Pressure (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟏: FSS Shear strength − swelling pressure relationship

1000
NCS 1
900 NCS 2
y = 2137e-0 003x
R² = 0.9368 NCS 3
800
Shear Strength (kPa)

700 y = 841.19e-0.002x
R² = 0.9678
600

500
y = 1449.4e-0.006x
R² = 0.9756
400

300

200

100

0
0 200 400 600 800 1000
Swelling Pressure (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟓𝟐: NCS Shear strength − swelling pressure relationship

Page | 130

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

900
y = 1320.1e-0.003x
GPS 1
R² = 0.9803
800 GPS 2
GPS 3
700

Shear Strength (kPa)

600

500
y = 1096.9e-0.003x
R² = 0.9924
400

300

200 y = 1776.4e-0.004x
R² = 0.9542
100

0
0 200 400 600 800
Swelling Pressure (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟑: GPS Shear strength − swelling pressure relationship

The mathematical expression of shear strength relationship with swell pressure for the
studied soils are given in Figures 5.51 through 5.53 by adding trendline in Excel
software. These relationships were analysed with several functions such as linear,
exponential and polynomial functions. Thus, exponential function was suitable for all the
studied soils. The high values of coefficient of determination, R2 was the decisive factor
on the part of regression analysis fittingly for the tested soils.

5.2.4 Shear strength and suction relationship

The relation between the shear strength and suction, depicts the unsaturated failure
criteria of the studied soils. The friction angle (∅𝑏 ) due to shear is associated with matric
suction. This gives resistance of the particles to being pulled apart due to surface tension
of the thin layer of water surrounding each particle. Figures 6.76 through 6.84 illustrates
the variation of shear strength and matric suction values of the studied soil. The curves
proved that non-linear relationship between shear strength and matric suction exists.
Thus, the soil shear strength values increase with respect to an increase in matric
suction which is defined by the angle (∅𝑏 ). This result agrees with research published
by Escario and Juca (1989) which demonstrated an experimental study performed over
a large range of suction values shown that the variation of shear strength with respect
to suction is non-linear.

Page | 131

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Furthermorehe, slope of the curve denoted by ∅𝑏 , implied that the rate of shear strength
increase due to the increase in matric suction. According to the results gotten from this
study, the friction angles of the unsaturated soil was evaluated to be less compared to
friction angles of the saturated soil. According to Fredlund and Rahardjo (1993), the
magnitude of ∅𝑏 , was normally equal or smaller than the effective friction angle ∅’.
Owning to the fact that ∅𝑏 recorded in this study is low compared to the saturated 𝜙 0 .
The laboratory result in this research for unsaturated angle of internal friction, is in
consistence with the report published by (Donald, 1957, Likos, et al. 2010, Nam et al.,
2011) were ∅𝑏 is less than ∅’ (∅𝑏 > ∅’). The curves showed that ∅𝑏 is the angle indicating
the rate of variation in shearing strength due to the contribution of matric suction. The
result portrayed that shear strength increased with increasing matric suction, this implied
that increase in shear strength contribution for matric suction is characterised by ∅𝑏 .
This response of unsaturated shear strength of the tested soils, are simply governed by
soil type, high suction values, dilation and different mechanisms of particle interaction
due to initial void ratios as a result of the complex response of unsaturated clay,
significantly contributed to the low ∅𝑏 values.

The stress at failure and the corresponding matric suction of each tested specimens
from the consolidated undrained shear strength tests were plotted as a function of
unsaturated soil version of failure envelopes. The saturated shear strength of the
studied soils was calculated using equation 2.19, while the unsaturated shear strength
of the studied soils was calculated using equation 2.27. Whereas, the unsaturated
friction angle of soils was determined from the slope between the shear strength and
matric suction curves. The shear strength and friction angle for saturated and
unsaturated soil mechanics are summarized in Appendix C Table 5.1. Hence, it
evidenced from the curves that the shear strength values of the unsaturated soil
increase with increase in negative pore water pressure (matric suction), though the
increase is non-linear.

Page | 132

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

600
FSS 1

500

y = 0.0555x + 84.956
Shear strength (kPa) 400 R² = 0.8941

300

∅𝑏=3.660
200

100 y = 133.98e0.0002x
R² = 0.97

0
0 2000 4000 6000 8000

Matric suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟒: Unsaturated failure envelope for FSS1

450
FSS 2
400

350
y = 0.0387x + 32.817
Shear strength (kPa)

300 R² = 0.9799

250

200
∅𝑏=1.660
150

100

y = 84.528e0.0002x
50
R² = 0.9775

0
0 2000 4000 6000 8000 10000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟓: Unsaturated failure envelope for FSS2

Page | 133

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

700

FSS 3
600

500
Shear strength (kPa)

y = 0.1256x - 1.2949
R² = 0.9402
400

300

∅𝑏=7.410
200

y = 126.5e0 0003x
100
R² = 0.9756

0
0 1000 2000 3000 4000 5000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟔: Unsaturated failure envelope for FSS3

700
NCS 1
600

500 y = 0.2124x + 35.624

Shear strength (kPa)

R² = 0.999

400

300

∅𝑏=15.750
200
y = 134.84e0.0006x
R² = 0.9708
100

0
0 500 1000 1500 2000 2500 3000

Matric suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟕: Unsaturated failure envelope for NCS 1

Page | 134

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

500
NCS 2
450
y = 0.1425x - 62.466
R² = 0.9909
400

Shear Strength (kPa) 350

300

250

200
∅𝑏=8.530
150

100
y = 57.336e0 0006x
50
R² = 0.9815
0
0 1000 2000 3000 4000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟖: Unsaturated failure envelope for NCS2

450
NCS 3
400

350
Shear Strength (kPa)

300

250 y = 0.2534x - 86.075

R² = 0.9761

200 ∅𝑏=15.640

150

100

y = 50.789e0 0011x
50 R² = 0.9913

0
0 500 1000 1500 2000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟗: Unsaturated failure envelope for NCS3

Page | 135

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

800
GPS 1
700

600 y = 0.0884x + 201.76

Shear Strength (kPa)

R² = 0.989

500

400

300 ∅𝑏=5.710

y = 254.53e0 0002x
200 R² = 0.9854

100

0
0 1000 2000 3000 4000 5000 6000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟔𝟎: Unsaturated failure envelope for GPS 1

600
GPS 2

500
Shear Strength (kPa)

400
y = 0.0712x + 75.893
R² = 0.9807
300

∅𝑏=3.660
200

100
y = 106.84e0 0003x
R² = 0.8822
0
0 1000 2000 3000 4000 5000 6000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟔𝟏: Unsaturated failure envelope for GPS 2

Page | 136

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

900
GPS 3
800

700

Shear Strength (kPa) y = 0.1052x + 198.77

600 R² = 0.9823

500

400

300
∅𝑏=6.420
200
y = 250.46e0.0002x
100 R² = 0.9565

0
0 1000 2000 3000 4000 5000 6000

Matric Suction (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟔𝟐: Unsaturated failure envelope for GPS 3

Conclusively, in the view of the experimental data, the connection between shear
strength and matric suction under consistent net normal stresses is assessed to be non-
straight as appeared in Figures 5.54 through 5.62. The collective comparison of the
tested soils results and different conditions of saturated and unsaturated shear strength
and shear strength parameters are presented in Appendix C Table 5.1. The shear
strength of the studied soils i.e. FSS, NCS and GPS, under unsaturated condition
increases with matric suction. Perhaps, the unsaturated shear strength values of the
tested soils are 1.5 to 2.5 higher than the saturated shear strength values. The failure
envelope moves up with the expanding net normal stress, subsequently causing the
stress at failure to be high and thusly resulted in higher shear strength values under
unsaturated condition.

Page | 137

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPTER 6: UNSATURATED RESPONSE OF PAVEMENT STRUCTURE

UNDER CYCLIC LOADING
6.1 Overview

This chapter analyse the laboratory results of resilient modulus (𝑀𝑟 ) test and as well
demonstrated the effects of confining stress, deviatoric stress, moisture content,
swelling stress and suction on 𝑀𝑟 . The development of 𝑀𝑟 predictive models using the
test results obtained from various laboratory tests conducted in this research were also
presented. A critical evaluation and validation of the developed 𝑀𝑟 models and existing
𝑀𝑟 models from literature review were assessed. In addition, comparison of pavement
designs, using unsaturated soil 𝑀𝑟 value and laboratory 𝑀𝑟 value were also presented.

6.2 Laboratory Resilient Modulus Test Results

6.2.1 FSS resilient modulus result
The resilient modulus values for the subgrade soils were evaluated using Repeated
Load Triaxial Test (RLTT). The axial load and displacement data recorded by 𝑀𝑟 data
collection was stored in 15 separate data files following the laboratory testing. Each of
these data files consisted of the load stroke, and three LVDT displacement values
recorded during the test. A software program installed into computer system that are
connected to RLTT instron was use to convert these data files into resilient modulus
values. The program searched for local maxima in the load and three displacement data
sets; these peak values correspond to the peak load and displacement pulses observed
during the haversine load pulse. Therefore, 100 load and displacement values were
identified in each column. The average 𝑀𝑟 values of last five sequences for the soils
were recorded. Three identical specimens were tested at the same moisture content to
illustrate the variation of 𝑀𝑟 with deviatoric stress (𝜎𝑑 ) at different confining pressures
(𝜎𝑐 ) of 19.96, 49.61, and 100.93 kPa as obtained from shear strength test. The results
showed that the differences in resilient modulus values due to change in confining
pressure are small. Figures 6.1 through 6.5 showed that the average difference in
resilient modulus values relative to deviator stress is < 5% in FSS. Figure 6.1 is
presented below while the rest of the Figures are presented in Appendix D.

Page | 138

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

6.2.4 Effect of moisture on resilient modulus

The specimens were prepared at varying moisture content, as was obtained from
moisture-density relationship in chapter 4 from section 4.2.1 through 4.2.3. The result
demonstrated that moisture content has significant impact on the 𝑀𝑟 of the studied
subgrade soils. Subgrade 𝑀𝑟 is critical in characterizing the support provided to the
pavement structure by the underlying subgrade. Considering that the moisture content
of a subgrade varies cyclically due to seasonal moisture changes. Therefore, it is
imperative to evaluate the effects of moisture changes on 𝑀𝑟 . Figure 6.16 illustrate
decrease in 𝑀𝑟 with increasing moisture content for FSS 1, 2 and 3. The soils prepared
beyond the optimum moisture content failed at conditioning stage. Therefore, no 𝑀𝑟
values were obtained at moisture content beyond the optimum. FSS 1 and 2 were more
weakens by the increase in moisture content, the soils were more susceptible to failure
at the optimum moisture content. Thus, failed completely at a point +2.5% moisture
beyond the optimum moisture content. This behaviour could be attributed to an
increased swelling pressure of the soils with higher PI and fine contents, because of
expansive nature of the tested soils.

160
FSS 1
140 FSS 2
Resilient Modulus (MPa)

FSS 3
120

100

80

60

40

20

0
0 5 10 15 20

Gravimetric moisture content (%)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟏𝟔: Mr − moisture content relationship for FSS

The pavement structures are usually compacted at a point below optimum water
content. On the calculation that the subgrade moisture content will come to equilibrium
with the surrounding subgrade soils due to seasonal moisture variation. However, the
Page | 142

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

degree of saturation of subgrade layer of pavement structures varies due to seasonal

moisture changes. Therefore, it is important to examine the effects of moisture on the
resilient deformation characteristics for the subgrade soils i.e. NCS 1, 2 and 3. To
evaluate the resilient deformation characteristics of the subgrade soils, the resilient
modulus tests for NCS 1, 2 and 3 under different moisture content were tested with the
combinations of three confining 20, 50 and 100 kPa at five deviatoric stress. The effect
of moisture content caused appreciable decrease in 𝑀𝑟 values, but the decrease was
more pronounced in NCS 3 as the soil recorded the lowest 𝑀𝑟 values among the studied
soils as shown in Figure 6.17. The results followed the same trends with the studied
published by (Hossain, 2010), which shows decrease in 𝑀𝑟 with increasing moisture
content.
100
NCS 1
NCS 2
80 NCS 3
Resilient Modulus (MPa)

60

40

20

0
0 5 10 15 20
Gravimetric moisture content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟏𝟕: Mr − moisture content relationship for NCS

The variation in 𝑀𝑟 values for specimens compacted at varying moisture content are
represented in Figure 6.18. Comparatively, at a given moisture content, GPS 3 recorded
the highest 𝑀𝑟 value of 158 MPa at a moisture content of 8.28%. Whereas, at the same
moisture content the corresponding 𝑀𝑟 values of GPS 1 and 2, where obtained to be
120MPa and 90MPa respectively. This implied that moisture content strongly affected
the 𝑀𝑟 values of GPS 2, compared to GPS 1 and GPS 3. In addition, the 𝑀𝑟 of the tested
soils generally decreased with increase in moisture content, irrespective of soil type
and deviatoric stress. Therefore, it was observed from the results that moisture content
is among the factors that governs 𝑀𝑟 of subgrade soils.

Page | 143

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

180
GPS 1
160 GPS 2

Resilient Modulus (MPa)

140 GPS 3

120

100

80

60

40

20

0
0 5 10 15 20
Gravimetric moisture content (%)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟏𝟖: Mr − moisture content relationship for GPS

In addition, the highest 𝑀𝑟 value was obtained on the dry of the OMC, even though the
dry density of the sample at the dry side of OMC is not the maximum dry density. This
can be explained due to the capillary suction and lack of lubrication at the dry side of
the optimum. At the effect of capillary suction, soil particles moved close to each other,
and this increases the shear stress with the soil’s particles, therefore resulting to
increase in 𝑀𝑟 values. The specimens prepared on the wet side of the OMC failed during
pre-conditioning stage. Thus, the smallest 𝑀𝑟 values were recorded at the OMC as the
specimens prepared beyond the OMC failed. Though, GPS 1, 2 and 3 survived the
conditioning stage at a moisture content +2% beyond the OMC, but recorded 𝑀𝑟 values
of 20 MPa, 19MPa and 30MPa for GPS 1, 2 and 3 respectively.

6.2.5 Effect of swelling stress on 𝑴𝒓

Appropriate pavement design requires knowledge of stress-strain response of subgrade

layer under cyclic loading. However, to accurately design a pavement, it is significant to
consider swelling stress, hence the soil is expansive dealing with nature. Resilient
modulus is an important property that characterizes subgrade behaviour through
repeated load triaxial test (RLTT), with relatively high reliable testing procedures. Series
of repeated load tests were conducted on FSS, NCS and GPS samples at different

Page | 144

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

moisture content, as to evaluate the stress relationship between of 𝑀𝑟 and swelling

stress. Figures 6.19 through 6.21 of the tested soils revealed and exponential
decreasing in 𝑀𝑟 values with increasing swelling pressure. This result, is similar to the
result obtained from shear strength modulus variation with swelling stress presented in
section 5.2.3, page 125.

500

450 y = 12.442e0.0051x FSS 1

R² = 0.9909
400 FSS 2
Resilient modulus (MPa)

y = 20.991e0.0023x
350 R² = 0.9533 FSS 3

300
y = 18.659e0.0049x
250 R² = 0.9942
200

150

100

50

0
0 200 400 600 800

Swelling stress (kPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟏𝟗: 𝑀𝑟 − swelling pressure relationship for FSS

200
y = 11.302e0 0039x
180 R² = 0.9979 NCS 1
160 NCS 2
y = 13.59e0.0025x
R² = 0.9844 NCS 3
140
Resilient modulus (MPa

120
y = 6.5634e0.0039x
100 R² = 0.941

80

60

40

20

0
0 100 200 300 400 500 600 700

Swelling pressure (KPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟎: 𝑀𝑟 − swelling pressure relationship NCS

Page | 145

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

500
GPS 1
y = 9.3848e0.0067x
450 R² = 0.9947 GPS 2
400 GPS 3
y = 18.942e0.0027x

Mesilient modulus (MPa)

350 R² = 0.9479

300 y = 19.659e0.0062x
R² = 0.859
250

200

150

100

50

0
0 200 400 600
Swelling stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟏: 𝑀𝑟 − swelling pressure relationship for GPS

The tested soils were allowed to swell by absorbing water before 𝑀𝑟 test was conducted
on the specimens. The results showed reduction range of 2.18%-3.17% in resilient
modulus values of the compacted soils, compared to the 𝑀𝑟 values of the specimens
that were not subjected swelling. This implies that increased swelling pressure in the
studied subgrade, caused some decrease in shear resistance of the subgrades. As this
is one the factors that triggers fatigues that leads to cracking on pavement structures
constructed on expansive soil. Considering the swelling potential of the compacted
studied soils, the shape of the exponential 𝑀𝑟 versus swell pressure curve demonstrated
a smooth decrease in 𝑀𝑟 values with increase in swelling stress of all the studied soils.

6.3 Resilient modulus Estimation Based on Soil Geotechnical Properties

This section centres on build-up and development of mathematical predictive models,

for estimating 𝑀𝑟 using other geotechnical soil properties. Multiple regression analysis
was used to develop mathematical model to predict 𝑀𝑟 based on selected conducted
geotechnical tests with respect to saturated soil mechanics. Forward and backward
stepwise regressions were performed using NCSS11, with 𝑀𝑟 as the dependent variable
and independent variables are: 𝛾𝑑 , 𝐶𝐵𝑅, 𝑀𝑐 , 𝑃𝑠 , 𝑈𝐶𝑆, 𝜏𝑆 , 𝑃200 and 𝐿𝐿. Forward stepwise
regression starts without independent variables and sequentially adds variables to the

Page | 146

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

model in order of their significance in predicting the dependent variable. A backward

stepwise regression begins with all independent variables and sequentially removes
variables from a model that are least significant in predicting the dependent variable
were both explored. Independent variables were added or removed until only those
variables that are mathematically significant as measured by a T-test are included in the
model. A significance level of 0.05 accuracy was obtained as a result of mathematical
significance, of each independent variable and developed mathematical predictive. The
accuracy level showed how reliable and accurate the developed models are and the
models are presented as follows:

𝑀𝑟 (MPa) = ζ0 + ζ1 (𝛾𝑑𝑟𝑦 ) − ζ2 (𝑃#200 ) − ζ3 (𝑀𝑐 ) − ζ4 (LL) − ζ5 (𝑃𝑠 ) … … … … … (Model 1)

𝑀𝑟 (MPa) = λ0 + λ1 (𝛾𝑑𝑟𝑦 ) − λ2 (𝐶𝐵𝑅) − λ3 (𝑀𝑐 ) − λ4 (𝑃𝑠 ) + λ5 (𝑞𝑢 ) + λ6 (𝜏𝑠 ) … … (Mode 2)

Mr (MPa) = η0 + η1 (γdry ) − η2 (CBR) − η3 (P#200 ) − η4 (Mc ) + η5 (LL) − η6 (Ps ) + η7 (q u )

+ η8 (τs ) … … … … … … … … … … … … … … … … … … … … … … … … … . (Model 3)

The values of each symbol and the intercept are summarized in the Table 6.1 Appendix
D. Three predictive models were developed using sets of laboratory data. The validation
of the models was satisfactory. Comparing the 𝑀𝑟 values obtained from the laboratory
exercise and 𝑀𝑟 values predicted by the developed models (Figures 6.22 through 6.24).
The model has an adjusted coefficient of determinations (R2) of 0.866, 0.936 and 0.939
for model 1, 2 and 3 respectively. The independent variables had p-values < 0.0001
(i.e., << 0.05). The negative coefficient of swelling pressure (𝑃𝑠 ) on the three models
infers that 𝑀𝑟 decreases with increasing swelling pressure.

Page | 147

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

160
y = 0.9422x + 1.4909
R² = 0.8658
140

120

Measured Mr (MPa)
100

80

60

40

20

0
0 20 40 60 80 100 120 140 160

Predicted Mr (MPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟐: Measured versus Predicted Mr for Model 1

160
y = x - 5E-13
R² = 0.9355
140

120
Measured Mr (MPa)

100

80

60

40

20

0
0 20 40 60 80 100 120 140 160
Predicted Mr (MPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟑: Measured versus Predicted Mr for Model 2

Page | 148

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

160
y = 0.9697x + 0.9924
R² = 0.9385
140

120
Measured Mr (MPa)
100

80

60

40

20

0
0 20 40 60 80 100 120 140 160

Predicted Mr (MPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟒: Measured versus Predicted Mr for Model 3

The 𝑀𝑟 values obtained from the developed models were compared with the measured
values. The data trend in Figure 6.22 through 6.24 indicates that there is a good
agreement between the measured and predicted 𝑀𝑟 values and this result proved the
validity of the developed models.

6.4 Resilient Modulus- Matric Suction Relationship

Resilient modulus-matric suction relationship in Figures 6.25 through 6.27 shows
variation of 𝑀𝑟 with different suction values obtained at various moisture contents. This
curves analysed the effects of suction on 𝑀𝑟. This was achieved by correlation of
gravimetric moisture content of 𝑀𝑟 specimens to the corresponding volumetric moisture
content on the SMR, for the matric suction values for each 𝑀𝑟 test soils. The results
demonstrated that matric suction changes with respect to change in soil moisture
conditions. The curves were plotted to understand the overall correlations between 𝑀𝑟
and matric suction of the soils. The scattered plot in Figures 6.25 through 6.27 showed
that compacting soils in unsaturated state induced suction increase as the 𝑀𝑟 values
increases. Therefore, a bi-linear relationship exists among 𝑀𝑟 and matric suction.

Thus, increase in 𝑀𝑟 values is pronounced due to the fact that dry state of soil initiated
high suction. This is due to much lower capillary suction and these behaviours are typical
Page | 149

© Central University of Technology, Free State

© Central University of Technology, Free State
 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

120
y = 0.0277x - 8.3359
100 R² = 0.6171

Resilient modulus (MPa) 80

60

40

Mr-Matric suction
20

0
0 1000 2000 3000 4000

𝑢𝑎−𝑢𝑤 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟔: Mr − matric suction relationship for NCS

180
y = 0.0231x + 0.3484
160 R² = 0.8146
Resilient modulus (MPa)

140

120

100

80

60

40
Mr-Matric suction
20

0
0 1000 2000 3000 4000 5000 6000 7000

𝑢𝑎−𝑢𝑤 (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟕: Mr − matric suction relationship for GPS

Generally, the scattered plots in Figures 6.25 and 6.27 provided a better theoretical framework
for unsaturated soils with respect to suction. This relationship can adequately take into account
the effect of moisture variation on changes in 𝑀𝑟 .

Page | 151

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

6.5 Unsaturated Soil 𝑴𝒓

In the field, subgrade soils are normally subjected to continual seasonal moisture
variations and this behaviour of unsaturated subgrade soils is governed by two stress
state variables which are net normal stress and matric suction. Therefore, it is imperative
to evaluate the unsaturated resilient modulus of a soil applying suction. The unsaturated
𝑀𝑟 values of the specimens were determined by preparing two sets of identical
specimens as mentioned earlier on section 3.5.4 page 68, at different moisture content
as obtained from compaction exercise. One set of the identical specimens were used to
measure the soils suction, while remaining set of the specimens were used to measure
𝑀𝑟. Furthermore, the unsaturated 𝑀𝑟 of the studied soils were computed using Equation
2 2.61 and 2.65 by Yang et. al (2005) and Lian et.al (2008) in the literature review on
pages 42 and 46 respectively.

This models considered the influence of deviator, matric suction and the lower impact
of confining stress in terms of bulk stress. The test data were fitted into this model and
the regression parameters K1 , K 2 and K 3 for each model at different moisture content
and dry density, were obtained, using multiple linear regression statistical software
package named NCSS11. Each model parameters K1 , K 2 and K 3 are presented in
Tables 6.3 through 6.5. Furthermore, two more models were developed using matric
suction, air-entry values, unsaturated CBR, dry density and swelling stress, to evaluate
the unsaturated 𝑀𝑟 of the studied soils. The model coefficients of each developed
models are presented in Table 2 Appendix D. Amongst other developed model is the
extension of Yang’s model in which swelling stress is negatively integrated, as the
swelling pressure negatively affects 𝑀𝑟 by reducing the soils 𝑀𝑟 values. The five
models that are used to evaluate 𝑀𝑟 values of the tested soils are listed as follows:

Mr (MPa) = k1 (σd + 𝑥ψm )k2 Yang et. al 2005 … … … … … … … … … … … … … … … … … … . . (Model 4)

Mr (MPa) = k1 (σd + 𝑥ψm )k2 − Psn Extension of Yang et. al 2005 model … … … … … … . . (Model 5 )
θ+𝑥ψm k2 τoct k3
Mr (MPa) = (k1 Pa ( Pa
) (P + 1) Liang et. al 2008 … … … … … … … … … … … … . . (Model 6)
a

𝑀𝑟 (MPa) = ξ0 + ξ1 (ψm ) − ξ2 (𝑆𝑒 ) + ξ3 (𝐶𝐵𝑅𝑢 ) − ξ4 (𝑃𝑆𝑛 ) Author′s … … … … … … … … … . (Model 7)

Mr (MPa) = α0 + α1 (ψm ) − α2 (Se ) − α3 (PSn ) + α4 (CBR u ) − α5 (γdry ) Author′s … … … … . (Model 8)

Page | 152

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The 𝑀𝑟 values of the studied soils were calculated at a constant deviatoric of σ1 - σ3 and
confining stress of, 100 kPa were considered. Thus, the studied soils recorded the
highest 𝑀𝑟 values, at this stresses. The of value 𝑥 = 1 is considered for the calculation
of the 𝑀𝑟 , as 𝑥 is the parameter representing contribution of suction on soil resilient
stress. The equation coefficients for models 7 and 8 are available in Appendix D.

Table 6.3: Unsaturated resilient modulus using model 4

𝝎% 𝛄𝐝 𝑲𝟏 𝑲𝟐 𝑹𝟐 𝛙𝐦 Measured Predicted
Soils (kN/m3) (MPa) 𝐌𝐫 (MPa) 𝑴𝒓 (MPa)
FSS 1
8.35 12.45 375 -0.363 0.459 6.541 130 190.6
13.56 16.63 289 -0.395 0.472 5.676 100 145.4
17.48 17.32 198 -0.438 0.493 4.989 60 97.83
24.27 15.90 - - - 2.793 Failed Failed
31.75 12.88 - - - 0.921 Failed Failed

FSS 2
10.13 12.47 415 -0.542 0.456 8.517 90 130
13.61 15.62 376 -0.687 0.428 6.343 70 106
18.54 16.93 312 -0.753 0.416 5.189 50 90
24.09 14.15 - - - 2.895 Failed Failed
30.16 12.12 - - - 1.913 Failed Failed

FSS 3
7.33 11.45 525 -0.546 0.432 6.398 140 190.4
14.19 17.61 473 -0.687 0.387 5.345 100 149.3
18.48 16.59 418 -0.999 0.342 3.450 70 121
22.91 16.93 - - - 1.998 Failed Failed
28.30 13.47 - - - 1.240 Failed Failed

NCS 1
8.40 16.07 410 -1.256 0.348 2.534 90 127
12.70 18.92 270 -1.289 0.377 2.098 70 103
17.51 19.35 150 -1.297 0.421 1.643 50 78
24.52 17.52 - - - 1.134 Failed Failed
28.35 16.18 - - - 0.778 Failed Failed

NCS 2
9.20 15.45 377 -0.980 0.543 3.878 70 99.57
14.18 17.52 200 -0.985 0.457 2.465 50 80
18.40 18.52 100 -0.997 0.632 2.087 20 48
24.91 18.89 - - - 1.598 Failed Failed
30.23 15.58 - - - 1.091 Failed Failed

NCS 3
9.94 14.17 150 -1.181 0.567 1.595 60 86
14.55 15.86 100 -1.198 0.584 1.545 40 59
18.87 16.83 60 -1.299 0.632 1.172 20 42.49
25.26 15.95 - - - 0.990 Failed Failed
32.92 13.02 - - - 0.681 Failed Failed

GPS 1
Page | 153

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

8.69 16.47 415 -0.588 0.437 5.071 120 160

13.48 19.46 387 -0.793 0.421 3.884 100 132
17.30 21.83 200 -0.757 0.875 2.959 70 88
24.31 20.07 295 -0.976 0.654 2.183 30 70
32.72 16.16 - - - 1.128 Failed Failed

GPS 2
9.95 15.45 377 -0.689 0.432 5.263 90 120
13.69 19.23 318 -0.792 0.451 3.989 70 106
18.56 20.89 200 -0.884 0.865 2.914 50 77.41
24.11 18.96 249 -0.979 0.435 2.029 20 60
30.14 15.96 - - - 0.667 Failed Failed

GPS 3
9.30 18.13 525 -0.589 0.412 5.331 150 196
14.18 21.63 474 -0.876 0.418 3.708 120 150
18.85 22.84 250 -0.949 0.389 2.707 90 96.74
22.91 21.71 130 -0.984 0.421 1.715 40 76
27.30 18.83 - - - 0.763 Failed Failed

*𝛄𝐝 = 𝐝𝐫𝐲 𝐮𝐧𝐢𝐭 𝐰𝐞𝐢𝐠𝐡𝐭 *𝛙𝐦 = 𝐦𝐚𝐭𝐫𝐢𝐜 𝐬𝐮𝐜𝐭𝐢𝐨𝐧 * 𝐌𝐫 = 𝐫𝐞𝐬𝐢𝐥𝐢𝐞𝐧𝐭 𝐦𝐨𝐝𝐮𝐥𝐮𝐬 * 𝝎 = 𝐦𝐨𝐢𝐬𝐭𝐮𝐫𝐞 𝐜𝐨𝐧𝐭𝐞𝐧𝐭

Table 6.4: Unsaturated resilient modulus using model 5

Soils 𝝎% 𝛄𝐝 𝛙𝐦 𝐏𝐬𝐧 Measured Predicted
(kN/m3) 𝑲𝟏 𝑲𝟐 𝑹𝟐 (kPa) (kPa) 𝑴𝒓 (MPa) 𝑴𝒓 (MPa)

FSS 1
8.35 12.45 375 -0.363 0.459 6.541 0.699 130 189.90
13.56 16.63 289 -0.395 0.472 5.676 0.600 100 144.8
17.48 17.32 198 -0.438 0.493 4.989 0.450 60 97.38
24.27 15.90 - - - 2.793 0.390 Failed Failed
31.75 12.88 - - - 0.921 0.330 Failed Failed

FSS 2
10.13 12.47 415 -0.542 0.456 8517 0.899 90 129.101
13.61 15.62 376 -0.687 0.428 6343 0.749 70 105.251
18.54 16.93 312 -0.753 0.416 4989 0.630 50 89.37
24.09 14.15 - - - 2895 0.450 Failed Failed
30.16 12.12 - - - 1913 0.300 Failed Failed

FSS 3
7.33 11.45 525 -0.546 0.432 6398 0.499 140 189.901
14.19 17.61 473 -0.687 0.387 5345 0.450 100 148.85
18.48 16.59 418 -0.999 0.342 3450 0.390 70 120.67
22.91 16.93 - - - 1998 0.330 Failed Failed
28.30 13.47 - - - 1240 0.250 Failed Failed

NCS 1
8.40 16.07 410 -1.256 0.348 2.534 0.650 90 126.35
12.70 18.92 270 -1.289 0.377 2.098 0.600 70 102.4
17.51 19.35 150 -1.297 0.421 1.643 0.520 50 77.48
24.52 17.52 - - - 1.134 0.450 Failed Failed
28.35 16.18 - - - 0.778 0.350 Failed Failed

Page | 154

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

NCS 2
9.20 15.45 377 -0.980 0.543 3878 0.630 70 98.94
14.18 17.52 200 -0.985 0.457 2465 0.500 50 79.5
18.40 18.52 100 -0.997 0.632 2087 0.450 20 47.55
24.91 18.89 - - - 1598 0.400 Failed Failed
30.23 15.58 - - - 1091 0.350 Failed Failed

NCS 3
9.94 14.17 150 -1.181 0.567 1595 0.830 60 85.17
14.55 15.86 100 -1.198 0.584 1545 0.725 40 58.28
18.87 16.83 60 -1.299 0.632 1172 0.650 20 41.84
25.26 15.95 - - - 990 0.520 Failed Failed
32.92 13.02 - - - 681 0.400 Failed Failed

GPS 1
8.69 16.47 415 -0.588 0.437 5071 0.490 120 159.51
13.48 19.46 387 -0.793 0.421 3884 0.450 100 131.55
17.30 21.83 200 -0.757 0.875 2959 0.375 70 87.63
24.31 20.07 295 -0.976 0.654 2183 0.350 50 69.65
32.72 16.16 - - - 1128 0.250 Failed Failed

GPS 2
9.95 15.45 377 -0.689 0.432 5263 0.650 90 119.35
13.69 19.23 318 -0.792 0.451 3989 0.600 70 105.4
18.56 20.89 200 -0.884 0.865 2914 0.520 50 76.89
24.11 18.96 249 -0.979 0.435 2029 0.450 30 59.55
30.14 15.96 - - - 667 0.350 Failed Failed

GPS 3
9.30 18.13 525 -0.589 0.412 5331 0.520 150 195.48
14.18 21.63 474 -0.876 0.418 3708 0.425 120 149.58
18.85 22.84 250 -0.949 0.389 2707 0.400 70 96.34
22.91 21.71 130 -0.984 0.421 1715 0.300 40 75.7
27.30 18.83 - - - 763 0.200 Failed Failed

*𝛄𝐝 = 𝐝𝐫𝐲 𝐮𝐧𝐢𝐭 𝐰𝐞𝐢𝐠𝐡𝐭 *𝛙𝐦 = 𝐦𝐚𝐭𝐫𝐢𝐜 𝐬𝐮𝐜𝐭𝐢𝐨𝐧 * 𝐌𝐫 = 𝐫𝐞𝐬𝐢𝐥𝐢𝐞𝐧𝐭 𝐦𝐨𝐝𝐮𝐥𝐮𝐬 * 𝝎 = 𝐦𝐨𝐢𝐬𝐭𝐮𝐫𝐞 𝐜𝐨𝐧𝐭𝐞𝐧𝐭
* 𝐏𝐬𝐧 = 𝐬𝐰𝐞𝐥𝐥𝐢𝐧𝐠 𝐩𝐫𝐞𝐬𝐬𝐮𝐫𝐞

Table 6.5: Unsaturated resilient modulus using model 6

Soils 𝝎% 𝛄𝐝 𝑲𝟏 𝑲𝟐 𝑲𝟑 𝑹𝟐 𝛙𝐦 Measured Predicted
(kN/m3) (kPa) 𝑴𝒓 (MPa) 𝑴𝒓 (MPa)
FSS 1
8.35 12.45 750 0.119 0.339 0.824 6.541 130 138.17
13.56 16.63 500 0.148 0.453 0.943 5.676 100 105.44
17.48 17.32 200 0.258 0.673 0.952 4.989 60 68
24.27 15.90 - - - - 2.793 Failed Failed
31.75 12.88 - - - - 0.921 Failed Failed

FSS 2
10.13 12.47 650 0.054 0.332 0.975 8.517 90 92.17
13.61 15.62 500 0.067 0.432 0.934 6.343 70 76
18.54 16.93 350 0.075 0.464 0.927 4.989 50 54.44
24.09 14.15 - - - - 2.895 Failed Failed
Page | 155

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

30.16 12.12 - - - - 1.913 Failed Failed

FSS 3
7.33 11.45 900 0.088 0.348 0.953 6.398 140 146
14.19 17.61 700 0.093 0.359 0.843 5.345 100 114
18.48 16.59 500 0.098 0.427 0.884 3.450 70 81.47
22.91 16.93 - - - - 1.998 Failed Failed
28.30 13.47 - - - - 1.240 Failed Failed

NCS 1
8.40 16.07 650 0.075 0.364 0.829 2.534 90 98
12.70 18.92 500 0.084 0.395 0.942 2.098 70 73.43
17.51 19.35 400 0.097 0.439 0.854 1.643 50 61
24.52 17.52 - - - - 1.134 Failed Failed
28.35 16.18 - - - - 0.778 Failed Failed

NCS 2
9.20 15.45 500 0.079 0.456 0.873 3.878 70 77.47
14.18 17.52 400 0.085 0.487 0.842 2.465 50 62
18.40 18.52 200 0.095 0.493 0.838 2.087 20 31.45
24.91 18.89 - - - - 1.598 Failed Failed
30.23 15.58 - - - - 1.091 Failed Failed

NCS 3
9.94 14.17 530 0.066 0.532 0.852 1.595 60 76
14.55 15.86 410 0.069 0.583 0.689 1.545 40 60
18.87 16.83 280 0.072 0.632 0.732 1.172 20 38
25.26 15.95 - - - - 0.990 Failed Failed
32.92 13.02 - - - - 0.681 Failed Failed

GPS 1
8.69 16.47 800 0.068 0.579 0.965 5.071 120 126
13.48 19.46 650 0.073 0.591 0.987 3.884 100 103
17.30 21.83 510 0.079 0.637 0.852 2.959 70 81.18
24.31 20.07 380 0.085 0.689 0.842 2.183 50 62
32.72 16.16 - - - - 1.128 Failed Failed

GPS 2
9.95 15.45 650 0.076 0.648 0.821 5.263 90 108
13.69 19.23 470 0.079 0.681 0.873 3.989 70 78.07
18.56 20.89 370 0.087 0.686 0.834 2.914 50 61.79
24.11 18.96 260 0.094 0.734 0.794 2.029 30 43.70
30.14 15.96 - - - - 0.667 Failed Failed

GPS 3
9.30 18.13 1185 0.046 0.437 0.843 5.331 150 160
14.18 21.63 900 0.055 0.495 0.892 3.708 120 129
18.85 22.84 570 0.063 0.561 0.836 2.707 70 84.10
22.91 21.71 340 0.071 0.583 0.821 1.715 40 50.33
27.30 18.83 - - - - 0.763 Failed Failed

*𝛄𝐝 = 𝐝𝐫𝐲 𝐮𝐧𝐢𝐭 𝐰𝐞𝐢𝐠𝐡𝐭 *𝛙𝐦 = 𝐦𝐚𝐭𝐫𝐢𝐜 𝐬𝐮𝐜𝐭𝐢𝐨𝐧 * 𝐌𝐫 = 𝐫𝐞𝐬𝐢𝐥𝐢𝐞𝐧𝐭 𝐦𝐨𝐝𝐮𝐥𝐮𝐬 * 𝝎 = 𝐦𝐨𝐢𝐬𝐭𝐮𝐫𝐞 𝐜𝐨𝐧𝐭𝐞𝐧𝐭

Page | 156

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Conclusively, Tables 6.3 through 6.5 presents the results obtained from the regression
analysis performed on the measured 𝑀𝑟 values by utilizing models 4 through 6. The
values of the regression constants, along with coefficient of determination (R2), are
shown for each soil and moisture content tested. Some general trends due to changes
in moisture conditions was observed amongst the regression constants. Generally, it
was observed that 𝑘1 coefficient attained maximum value on the dry side and this values
decreases with increasing moisture content. This regression trend of the studied soils
is in agreement with result published by Nazzal and Mohammad (2010). The regression
result of the soils showed that 𝑘1 is proportional to the stiffness of the material which is
dependent on the effective stress of the soil. Under unsaturated soils, effective stress is
dependent on matric suction, and matric suction increases with decreasing water
content, therefore the increase in 𝑘1 can be attributed to an increase in matric suction.
The coefficient for 𝑘2 is related to deviator stress for model 4. Thus, has negative values
as it tend to decrease with increasing moisture content. While coefficient for 𝑘2 in model
5 is positive, as it is related to bulk stress.
Generally, the value of 𝑘3 coefficient is positive, as the regression analysis identified
the best subset of independent variables that results in accurate correlation between
resilient modulus model parameters 𝑘𝑖 and basic soil properties. However, the 𝑘3
coefficient describes the softening of the soils with increasing octahedral shear stress
and values reduces with increasing moisture content. This implied that the soils at a
higher moisture content are more susceptible to softening due to increase in shear
stress.

6.5.1 Validation of the developed and existing models

In order to validate the models used in this study, five models with varying degrees of
predictability were plotted against the laboratory measure Mr data. The 𝑀𝑟 calculated
using model 4, 5 and 6 best fitted the laboratory data, because this models coefficient
of determination (R2) were obtained to 0.9438, 0.9607 and 0.9533 respectively. In
addition, the 𝑀𝑟 of unsaturated subgrade soil mainly depend on suction, deviatoric
stress level and confining pressure, which are the parameter coefficient in this models.

Page | 157

© Central University of Technology, Free State

Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Yang et.al 2005, Model 4

160
y = 0.7827x - 7.1228
R² = 0.9438
140

Measured Mr (MPa) 120

100

80

60

40

20

0
0 20 40 60 80 100 120 140 160 180 200 220
Esitimated unsaturated Mr (MPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟖: Validation of model 4

Extension of Yang et.al 2005 model, Model 5

160
y = 0.775x - 9.7707
140 R² = 0.9607

120
Measured Mr (MPa)

100

80

60

40

20

0
0 50 100 150 200 250

Esitimated Unsaturated Mr (MPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟗: Validation of model 5

Page | 158

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Liang et.al 2008, Model 6

160
y = 1.0422x - 15.609
R² = 0.9533
140

Measured Mr (Mpa) 120

100

80

60

40

20

0
0 20 40 60 80 100 120 140 160 180

Estimated unsaturated Mr (MPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟑𝟎: Validation of model 6

The results from the 𝑀𝑟 tests were correlated with other unsaturated soil properties i.e.
Air entry value, suction, unsaturated CBRu and dry densities. The developed models 7
and 8 for estimating 𝑀𝑟 of studied soils showed strong correlation with R2 of 0.9134
and 0.9139 for models 7 and 8 respectively. Though, models 4, 5 and 6 were found to
have higher R2 compared to models 7 and 8. The low value of R2 for models 7 and 8
were expected as this models is given as function of suction, air-entry values
unsaturated CBR and dry density.
Generally, this models were acceptable though deviatoric and confining stresses were
not considered as independent variables during regression exercise. Moreover, there is
a strong correlation between the predicted and measured Mr as shown in Figures 6.31
and 6.32. This implies that models 7 and 8 are suitable for predicting unsaturated 𝑀𝑟 of
expansive subgrade with high degree of accuracy.

Page | 159

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Author's Model 7
162
y = 1.018x - 1.0381
R² = 0.9134
142

122
Measured Mr (Mpa)

102

82

62

42

22

2
2 22 42 62 82 102 122 142 162

Estimated unsaturtaed Mr (MPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟑𝟏: Validation of model 7

Author's Mode 8
162
y = 1.0179x - 1.027
R² = 0.9139
142

122
Measured Mr (MPa)

102

82

62

42

22

2
2 22 42 62 82 102 122 142 162

Estimated unsaturated Mr (MPa)

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟑𝟐: Validation of model 8

Figures 6.27 through 6.31 shows validation of different 𝑀𝑟 models. The models were
demonstrated as a function to simulate the behaviours of the 𝑀𝑟 with respect to soil
Page | 160

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

suction using curve-fitting procedures and laboratory test data. It was observed that the
predicted 𝑀𝑟 default the measured 𝑀𝑟 values by ± 0.048 on the average. However, the
increase in 𝑀𝑟 values were recognized at the dry side of the optimum. Whilst low 𝑀𝑟
values were recorded as the moisture content of soil increases. Thus, the higher 𝑀𝑟
values on the dry side is attributed to the integrity of soil structure and rigidity of soil
skeleton, as higher suction stiffens the structure of the soil specimen, higher 𝑀𝑟 is
obtained.

6.6 Design Of Flexible Pavement Structure

Appropriate pavement design requires knowledge of stress-strain response of subgrade

layer under cyclic loading. Pavement structure is generally under unsaturated
conditions. Thus, to appropriately design a pavement, it is significant to consider
unsaturated 𝑀𝑟 values, as 𝑀𝑟 is an important property that characterizes subgrade
performance and pavement strength under unsaturated soil mechanics.
However, AASHTO Guideline for the Design of Pavements (1993) was followed in this
research. This guideline gives a full description of the functional and structural
performance of pavement. The method can be used for new and rehabilitation pavement
design. The design equations for asphalt pavements structure include is stated below:
∆𝑃𝑆𝐼
log ( )
log(𝑊18 ) = 𝑍𝑅 × 𝑆𝑜 + 9.36 log(𝑆𝑁 + 1) − 0.20 + 4.2 − 1.5 + 2.32 log(𝑀𝑅 )
1094
0.40 +
(𝑆𝑁 + 1)5.19
− 8.07
(6.1)
Where:
𝑊18 = predicted number of 18 − kip (80kN) single axle load applications
ZR = standard normal deviate
S0 = combined standard error of the traffic and performance predictions
SN = structural number of the total pavement thickness
∆PSI = difference between the 𝑃𝑆𝐼𝑜 and terminal 𝑃𝑆𝐼𝑡 serviceability indices
M𝐸𝑅 = effective resilient modulus adjusted for seasonal variation (MPa)
However, the structural number is given by the equation 6.2 below:

𝑆𝑁 = 𝑎1 𝐷1 + 𝑎2 𝐷2 𝑚2 + 𝑎3 𝐷3 𝑚3 (6.2)

Page | 161

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

Where:
SN = structural number of the total pavement thicknes
𝑎𝑖 = layer coefficient
𝐷𝑖 = layer thickness
𝑚𝑖 = layer drainage coefficient
The design equations are provided to design pavement structure to the predetermined
degree of assurance in order to guarantee performance of service life period which it
was designed.

6.6.1 Design of flexible pavement structure using laboratory 𝑴𝒓 value

This design approach followed layer thickness determination using layered analysis
technique. Moreover, salient features were considered during pavement design in this
research, such as:
• Traffic consideration in terms of equivalent standard axle load repetitions (18-
kip).
• Equivalent axle load factors.
• Layer coefficients for different types of materials
• Single index structural number was used to represent the pavement structure at
each layer.

𝑊18 = f(Mr , SN, ZR , So, ∆PSI)

Table 6.6: Parameters for design of a 3-layered pavement structure

Selected design input Design values
Expected ESAL, 𝐒𝐂𝐖 𝟏𝟖 10,000,000, 18 − kip standard axles
𝐑𝐞𝐚𝐥𝐢𝐚𝐛𝐢𝐥𝐢𝐭𝐲, 𝐑 95%
𝐙𝐑 −1.645
𝑺𝑶 0.45
𝐌𝐫 surface asphalt (assumed) 2758 MPa
𝐌𝐫 base (assumed) 172.37MPa
𝐌𝐫 subbase (assumed) 89.632
FSS 1 𝐌𝐫 subgrade obtained from Lab 60MPa at Optimum moisture content
𝑷𝒊 4.5
𝑷𝒕 2.1
∆𝑷𝑺𝑰 2.4

The nomograph for flexible pavement, as presented in Figure 6.33 below, was used to
trace the required structural number for each layer.

Page | 162

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟑𝟑: Nomograph for flexible pavement

Tests to determine parameters associated with pavement design need to be effective.

Routine testing of soil subgrades to determine various parameters is an important
feature in pavement design and construction. The resilient modulus is one important
parameter in the determination of the structural capacity of pavements. However, the
Mr and structural number of each layer are summerized in the Table 6.7 below:

Table 6.7: Resilient modulus layers thickness and structural number (SN) design one
Structural number Next lowest layer Next lowest SNi
layers Mr
Surface Asphalt SN1 Base 2758 MPa 3.20
Base SN2 Subbase 172.37MPa 4.00
Subbase SN3 Subgrade 60MPa 4.55

𝐋𝐚𝐲𝐞𝐫 𝟏, 𝐒𝐍 𝟏 = 𝒂𝟏 𝒅𝟏, 𝐒𝐍 𝟏
From coefficient for asphalt chart, a1 = 0.44, in Appendix D
SN 1 = 3.20 = 4.1(𝑑1 )
3.22
𝑑1 = = 7.3 𝑖𝑛𝑐ℎ𝑒𝑠 = 190𝑚𝑚, ∴ 𝑑𝑒𝑝𝑡ℎ 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑, 𝑑1∗ = 7.5𝑖𝑛𝑐ℎ𝑒𝑠, 200𝑚𝑚
0.44
Min. thickness = 3.5 inches = 90mm ∴ 𝑑1 > 𝑑1∗ , 𝑝𝑟𝑜𝑣𝑖𝑑𝑒 𝑑1 = 200𝑚𝑚 𝑑𝑒𝑝𝑡ℎ 𝑜𝑘𝑎𝑦

𝐋𝐚𝐲𝐞𝐫 𝟐, 𝐒𝐍 𝟐 = 𝒂𝟏 𝒅∗𝟏 + 𝒂𝟐 𝒅𝟐 𝒎𝟐 ,
Where:
Page | 163

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝑑2 = 𝑖𝑠 𝑑𝑎𝑟𝑖𝑛𝑎𝑔𝑒 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 = 1, 𝑓𝑎𝑖𝑟

𝑎2 = 0.249(𝑙𝑜𝑔Mr ) − 0.977 = 0.249(𝑙𝑜𝑔13000 𝑝𝑠𝑖) − 0.977, 𝑎2 = 0.047
∴ SN 2 = 4.00, 𝑎2 = 0.047
m2 = is darinage coefficient = 1, fair
SN 2 = 𝑎1 𝑑1∗ + 𝑎2 𝑑2 𝑚2 = 4.00 = 0.44(7.85) + 0.047𝑑2 (1)

0.546
∴ 𝑑2 = = 11.62 𝑖𝑛𝑐ℎ𝑒𝑠 = 295𝑚𝑚, ∴ 𝑑𝑒𝑝𝑡ℎ 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑, 𝑑2 = 300𝑚𝑚
0.047

SN 2 − 𝑆𝑁1∗ 4.00 − 3.3

𝑑2∗ > > > 14.89 𝑖𝑛𝑐ℎ𝑒𝑠 𝑝𝑟𝑜𝑣𝑖𝑑𝑒, 𝑑2 = 380𝑚𝑚
𝑎 2 𝑚2 0.047 × 1

𝐋𝐚𝐲𝐞𝐫 𝟑, 𝐒𝐍 𝟑 = 𝐚𝟏 𝐝∗𝟏 + 𝐚𝟐 𝐝∗𝟐 𝐦𝟐 + 𝐚𝟑 𝐝𝟑 𝐦𝟑

m3 = is darinage coefficient = 1, fair
𝑎3 = 0.227(𝑙𝑜𝑔Mr ) − 0.839 = 0.227(𝑙𝑜𝑔8702.26 𝑝𝑠𝑖) − 0.839, 𝑎3 = 0.055
∴ SN 3 = 4.55, 𝑎3 = 0.055
SN 3 = 𝑎1 𝑑1∗ + 𝑎2 𝑑2∗ 𝑚2 + 𝑎3 𝑑3 𝑚3 = 4.55 = 0.44(7.3) + 0.047(14.89) + 0.055(1) 𝑑3
4.55 − 3.990
∴ 𝑑3 = = 10.18 𝑖𝑛𝑐ℎ𝑒𝑠 = 254𝑚𝑚, ∴ 𝑑𝑒𝑝𝑡ℎ 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑, 𝑑2∗ = 460𝑚𝑚
0.055

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟑𝟒: Pavement design thickness one

The designed depth for subgrade is not sufficient to provide support for the pavement
structure, therefore the roadbed/subgrade required stabilization or reinforcement. The
design showed that structural performance of the subgrade required higher Mr value in
order to provide the required support for pavement structure.

Page | 164

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

6.6.2 Flexible pavement design using unsaturated 𝑴𝒓 value from model 4

The second pavement design was done using unsaturated Mr value evaluated using
model 4. The subgrade thickness of pavement layers are optimized to fulfil both
structural and economical requirements based on the resilient strength of the roadbed.
The selected design parameters structural numbers are summarized in Tables below:

Table 6.8: Parameters for 3-layered pavement structure for design two
Selected design input Design values
Expected ESAL, 𝐒𝐂𝐖 𝟏𝟖 10,000,000, 18 − kip standard axles
𝐑𝐞𝐚𝐥𝐢𝐚𝐛𝐢𝐥𝐢𝐭𝐲, 𝐑 95%
𝐙𝐑 −1.645
𝑺𝑶 0.45
𝐌𝐫 surface asphalt (assumed) 2758 MPa
𝐌𝐫 base (assumed) 172.37MPa
𝐌𝐫 subbase (assumed) 89.63 MPa
FSS 1 𝐌𝐫 subgrade obtained from Lab 100MPa at Optimum moisture content
𝑷𝒊 4.5
𝑷𝒕 2.1
∆𝑷𝑺𝑰 2.4

According to the design presented below, the wearing course portrayed response of a
dense graded hot mix asphalt (HMA) that has high resistance again deformation due to
its value for 𝑀𝑟 . Stone mastic asphalt (SMA) could be recommended for a high-volume
roads as to increase the resistance against deformation.

Table 6.9: Resilient modulus layers thickness and structural number design two
Structural number Next lowest layer Next lowest SNi
layers Mr
Surface Asphalt SN1 Base 2758 MPa 3.22
Base SN2 Subbase 172.37MPa 4.00
Subbase SN3 Subgrade 100MPa 6.50

𝐋𝐚𝐲𝐞𝐫 𝟏, 𝐒𝐍 𝟏 = 𝒂𝟏 𝒅𝟏, 𝐒𝐍 𝟏 ,
From coefficient for asphalt chart, a1 = 0.44, in Appendix D
SN 1 = 3.20 = 4.1(𝑑1 )
3.22
𝑑1 = = 7.3 𝑖𝑛𝑐ℎ𝑒𝑠 = 190𝑚𝑚, ∴ 𝑑𝑒𝑝𝑡ℎ 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑, 𝑑1∗ = 7.5𝑖𝑛𝑐ℎ𝑒𝑠, 200𝑚𝑚
0.44
Min. thickness = 3.5 inches = 90mm ∴ 𝑑1 > 𝑑1∗ , 𝑝𝑟𝑜𝑣𝑖𝑑𝑒 𝑑1 = 200𝑚𝑚 𝑑𝑒𝑝𝑡ℎ 𝑜𝑘𝑎𝑦

Page | 165

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

𝐋𝐚𝐲𝐞𝐫 𝟐, 𝐒𝐍 𝟐 = 𝒂𝟏 𝒅∗𝟏 + 𝒂𝟐 𝒅𝟐 𝒎𝟐 ,
Where:
𝑑2 = 𝑖𝑠 𝑑𝑎𝑟𝑖𝑛𝑎𝑔𝑒 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 = 1, 𝑓𝑎𝑖𝑟
𝑎2 = 0.249(𝑙𝑜𝑔Mr ) − 0.977 = 0.249(𝑙𝑜𝑔13000 𝑝𝑠𝑖) − 0.977, 𝑎2 = 0.047
∴ SN 2 = 4.00, 𝑎2 = 0.047

SN 2 = 𝑎1 𝑑1∗ + 𝑎2 𝑑2 𝑚2 = 4.00 = 0.44(7.85) + 0.047𝑑2 (1)

0.546
∴ 𝑑2 = = 11.62 𝑖𝑛𝑐ℎ𝑒𝑠 = 295𝑚𝑚, ∴ 𝑑𝑒𝑝𝑡ℎ 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑, 𝑑2 = 300𝑚𝑚
0.047

SN 2 − 𝑆𝑁1∗ 4.00 − 3.3

𝑑2∗ > > > 14.89 𝑖𝑛𝑐ℎ𝑒𝑠 𝑝𝑟𝑜𝑣𝑖𝑑𝑒, 𝑑2 = 380𝑚𝑚
𝑎 2 𝑚2 0.047 × 1

𝐋𝐚𝐲𝐞𝐫 𝟑, 𝐒𝐍 𝟑 = 𝐚𝟏 𝐝∗𝟏 + 𝐚𝟐 𝐝∗𝟐 𝐦𝟐 + 𝐚𝟑 𝐝𝟑 𝐦𝟑

d3 = is darinage coefficient = 1, fair
𝑎3 = 0.227(𝑙𝑜𝑔Mr ) − 0.839 = 0.227(𝑙𝑜𝑔8702.26 𝑝𝑠𝑖) − 0.839, 𝑎3 = 0.055
∴ SN 3 = 4.55, 𝑎3 = 0.055
m3 = is darinage coefficient = 1, fair
𝑎3 = 0.227(𝑙𝑜𝑔Mr ) − 0.839 = 0.227(𝑙𝑜𝑔14504 𝑝𝑠𝑖) − 0.839, 𝑎3 = 0.106
∴ SN 3 = 6.5, 𝑎3 = 0.106
SN 3 = 𝑎1 𝑑1∗ + 𝑎2 𝑑2∗ 𝑚2 + 𝑎3 𝑑3 𝑚3 = 3.86 = 0.41(7.85) + 0.047(16.62) + 0.106(1) 𝑑3
6.50 − 3.997
∴ 𝑑3 = = 24 𝑖𝑛𝑐ℎ𝑒𝑠 = 609.6𝑚𝑚, ∴ 𝑑𝑒𝑝𝑡ℎ 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑, 𝑑3∗ = 620𝑚𝑚
0.106

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟑𝟓: Pavement design thickness design two

Page | 166

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

The designed okay, design depth > d1 and d2 .there for the subgrade is sufficient enough
to provide support for the pavement structure. In addition, the empirical equation in the
AASHTO method shows the different factors that are considered for design, for instance
traffic, pavement performance (serviceability), road bed soil (subgrade), and drainage
coefficients. The subgrade layer is characterized by its resilient modulus.

Page | 167

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

CHAPTER 7: SUMMARY, CONCLUSION AND RECOMMENDATIONS

7.1 SUMMARY
This research was carried out to study the effects of the geotechnical response of
unsaturated subgrades for pavement structure under cyclic loading. Following the
research objective, characterization of the identified subgrades soils using particles size
distribution analysis (PSD) and consistency limit tests was accomplished. Base on the
results, the subgrades were designated as high plastic clay (CH) for FSS and NCS
respectively, and lean clay (CL) for GPS 1 and 2. Thus, GPS 3 was classified as elastic
slit. Nonetheless, these soils were subjected to free swell index and swelling stress
tests, to ascertain the their degree of expansiveness. In addition, XRD and XRF tests
were also carried on the soil with the objective to quantify the clay minerals responsible
for swelling behaviour of the soils and as a confirmation for the classification tests
mentioned above.

However, an extensive laboratory testing exercise was carried out to assess the impact
of suction, swelling stress and moisture content on 𝑀𝑟 values of unsaturated subgrade
soils of different plasticity indices. Repeated Load Triaxial (RLT) tests were conducted
to evaluate the 𝑀𝑟 response of nine identified subgrade soils, representing the range of
problematic subgrade soils found across three provinces in South Africa. The influence
of suction, swelling stress, deviator stress and confining stress variation on 𝑀𝑟 were
evaluated. As these stress state variable are mainly the parameters that governs the
dynamics of unsaturated 𝑀𝑟. Additionally, Soil Water Retention Curves (SWRC) were
established to evaluate the correlation between moisture content and matric suction for
the tested subgrade soils.

After these laboratory tests were completed and results were obtained. Development of
mathematical predictive models for both CBR and 𝑀𝑟 under classical soil and
unsaturated soil mechanics were achieved. The development of the mathematical
models exercise, were followed by regression analysis as to develop and compute the
regression constants (𝑘1 , 𝑘2 , 𝑘3 ,) for Yang et.al (2005) and Hang et. al constitutive model
using NCSS 11 mathematical software Thus, the developed models were tested using
laboratory data and results revealed that the values obtained under unsaturated soil

Page | 168

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

mechanics are 1.5 to 2 times higher compared to the results gotten under classical soil
mechanics.
7.2 CONCLUSIONS
The aim of this research was directly pursued, as to characterize the geotechnical
response of unsaturated soils for road pavement structure under cyclic loading. Based
on the results of the laboratory testing program and the subsequent analyses, the
following conclusions were reached:

• The particles size distribution analyse of the soils, showed that the soils fine
content ranges from 75% to 95%. Whereas, the consistency limit test results
revealed that the soils have liquid limit range of 40 to 70% and Plasticity index
(PI) range of 17% to 45%. Furthermore, the soils free swelling index and swelling
stress results were obtained to be within the bracket of 30% to 80% and 200 kPa
to 900 kPa respectively. Based on these tests results, the soils were accurately
classified as high plastic clay (CH) for FSS, low plastic clay (CL) for NCS, while,
GPS 1 and 2 were classified as lean clay (CL) and GPS 3 as elastic slit (ML).

• The suction filter paper results for the soils, confirmed lower total and matric
suction values on the wet side. Thus, this implies high capillary stresses in the
soils and caused higher absorption of water within soils voids. Furthermore, the
soils with higher PI on the dry side of the optimum recorded the highest matric
suction values ranging from 5000 kPa to 10000kPa, due to higher percentage of
fine content. However, matric suction of the soils were found to be 83.23% higher
than that osmotic suction. This proved that matric suction is a dominant
parameter of total suction. This is ascribed to the influence of soils capillarity,
void size and the physiochemical interactions, that depends on the soil
mineralogy and morphology. Thus, the result is in line with the investigation
published by (Yang et al. 2005) which proved that hydromechanical response of
soils are governed by matric suction.

• An increase in matric suction results stiffens the unsaturated soil specimens, as

evidenced by the effective stress for unsaturated soils. Whereas, effective stress
increases with increasing matric suction. The incorporation of matric suction in

Page | 169

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

predicting Mr values, provides the best theoretical approach since matric suction
is a key parameter in describing the stress state variables of unsaturated soils.

• Based on the filter paper for soil-water retentivity, FSS, NCS and GPS displayed
adsorptive unimodal shapes. Thus experimental test data were fitted with three
different SWRC models developed by Van Genuchten 1980; Fredlund and Xing
1994; and Seki 2007 with a correction factor using a least squares regression. An
Optimizer was used to optimize the parametric models to the measured data. This
was followed by an iterative exercise, for the best sum of squared residuals (SSR)
differences between the predicted and measured SWRC values. The sum of the
SSR and Akaike information criterion (AIC) is an indication of how well the models
fits with the measured data. Generally, it was observed that Seki’s model best fitted
the experimental data of the studied soils with R2 ranging from 0.95458 - 0.99986,
compared to the other models.

• SWRC curves establishes the relationship between the volumetric water content
and matric suction. This demonstrated that matric suction increases with
decreasing volumetric water content. However, the soils with higher PI values
recorded higher air-entry values compared to the soils with lower PI values. This
was expected due to unimodal shape of the SWRC. Hence higher PI soils have
a higher water holding capacity due to surface charges and short-range
adsorption.

• The resilient modulus values for subgrade soils were observed to be reliant on
the stress state condition of the soil. The 𝑀𝑟 results of the studied soils exhibited
decrease in 𝑀𝑟 with increasing deviatoric stress and as well an increase with
confining pressure.

• The moisture content has a significant impact on the 𝑀𝑟 of soils, such that 𝑀𝑟
values decreased with increase in moisture content. CH and CL soils were the
most susceptible to decreases in 𝑀𝑟 when moisture content increased to the wet
side of the optimum. The subgrades resilient response, displayed significant
strain-softening behaviour on specimens prepared at the optimum moisture

Page | 170

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

contents. Whereas, the specimens prepared at the point beyond the optimum
moisture content failed at the conditioning stage.

• The equations (models) developed in this research were based on statistical

analysis of laboratory test results that were limited to the soil geotechnical
properties specified. Estimation of resilient modulus of subgrade regression
parameters (𝑘1 , 𝑘2 , 𝑘3 ) varies with soils moisture content, dry density and
percentage fines.

• Existing constitutive models that incorporate matric suction for predicting

𝑀𝑟 values for unsaturated soils i.e. Yang et al. (2005) and Liang et al. (2008)
were validated. Through comparing the predicted 𝑀𝑟 values with measured 𝑀𝑟
values from this study. The results showed that Yang et al. (2005) and Liang et
al. (2008) models had the ability to capture the effect of moisture variation on 𝑀𝑟
through incorporating matric suction. The Yang et al. (2005) and Liang et al.
(2008) models generally provided a good fit to the measured 𝑀𝑟 data.

• Based on the experimental data of soils, three models were proposed in this
study. These model includes the effect of dry densities, swelling pressure and
air-entry values (AEV) on the contribution of 𝑀𝑟 and the models also establishes
an explicit correlation to matric suction by incorporating swelling stress and AEV
into the model. Therefore, strong relationship between the matric suction and 𝑀𝑟
is highlighted. Thus, the proposed models in this research can be a useful tool in
the rational prediction of the 𝑀𝑟 for routine engineering practice for unsaturated
soil mechanics in pavement design practice.

• The swelling pressure decreased the resilient modulus and shear strength 1.45
to 3,2% of the initial 𝑀𝑟. Therefore, the effects of swelling should be considered
in pavement design exercise and be neglected if the subgrade soil is not an
expansive one.

• The unsaturated values of all the geotechnical tests results ranging from CBR
shear strength and resilient modulus were 1.5 to 2 times higher than that of the
classical laboratory soil mechanics values. This simply implied that designing
Page | 171

© Central University of Technology, Free State

 Behaviour of Unsaturated soils for Road Pavement Structure Under Cyclic Loading

pavement structure using unsaturated soil mechanics value is very conservative,

contrary to the design of pavement using classical soil mechanics values, that
resulted in under-design of pavement structure, as demonstrated in this study.

7.3 RECOMMENDATIONS
In light of this research, the following recommendations are made to incorporate the effects
of suction variation in assessing design value of 𝑀𝑟 for unsaturated subgrades:

• A simpler procedures are needed to be established by adapting the conventional

triaxial resilient modulus instrument in various research centres and universities
in South Africa , in order to directly measure suction as at the same time of
measuring resilient modulus and shear stress values. As this development will
aid to lessen the time required to separately measure suction and resilient
modulus. Therefore, reduce the time require to prepare many specimens and as
well lessen the time specimens require to attain equilibration period.

• The proposed constitutive 𝑀𝑟-matric suction relationship was validated based on

the results from the nine soil types utilised in this research. Wider range of soil
types are required to validate the applicability of the developed model for a variety
of soil types with different PI and moisture contents.

• The undergraduate geotechnical engineering courses has not been significantly

modified in decades in South African universities. There, is no significant
coverage of unsaturated soil mechanics. The student’s depth of understanding
of soil engineering would be greatly enhanced via introduction of geotechnical
principles for unsaturated soil mechanics with the saturated soil case being
presented as a subset of the broader theory. This will aid engineers in pavement,
foundation, earth dam and rail track for better design.

Page | 172

© Central University of Technology, Free State

 References

REFERENCES:
AASHTO. 1994. Standard method of test for resilient modulus of unbound granular
base/subbase materials and subgrade soils. AASHTO T294, Washington, DC.

AASHTO. 2002. Guide for the design of new and rehabilitated pavement structures.
National Cooperative Highway Research Program, AASHTO NCHRP Project 1-
37A. American Association of State Highway and Transportation Officials
(AASHTO), Washington, D.C.

AASHTO. 2003. Standard method of test for determining the resilient modulus of soil
and aggregate materials.”

AASHTO T307. 2005. Standard methods of test for determining the resilient modulus of
soils and aggregate materials.

AASHTO T 307-99. 2007. Standard method of test for determining the resilient modulus
of soils and aggregate materials, American Association of State Highway and
Transportation Officials, Washington, D.C.

AASHTO. 2008. Mechanistic-empirical pavement design guide, interim edition: A

manual of practice. Washington, DC.

Abramento, M, and Carvalho, S. 1989. Goetechnical parameters for study of slope

insatbilization at Serra do Mar-Brazilian Southeast. Proceedings of the 12th
International Conference on Soil Mechanics and Foundation Engineering, Rio de
Janeiro, 3: 1599-1602.

Agus, SS and Schanz, T, 2006a. Drying, wetting, and suction characteristic curve of
bentonite-sand mixture. In: Proc 4th Int. Conf. on unsaturated soil, Arizona, USA,
pp1405-1414.
Aitchison, GD. 1961. Relationship of moisture and effective stress functions in
unsaturated soils Pore pressure and suction in soils, Butterworth, London.

Alam, TB Abdelrahman, M and Scharm, SA. 2009. Laboratory characteristics of

Recycled Asphalt Pavement as a base layer. International Journal of Pavement
Engineering, Vol. 11, No. 2, 123-131.

Allam, MM, and Sridharan, A. 1987. Stresses present in unsaturated soils. Journal of
Geotechnical Engineering. ASCE. Vol. 113, 11, 1395-1399.

Alsherif, NA. and McCartney, JS. 2014. Effective stress in unsaturated silt at low
degrees of saturation. Vadose Zone J. 13 doi:10.2136/ vzj2013.06.0109 (this
issue).
Page | 173

© Central University of Technology, Free State

 References

Ampadu, S. (2007). A laboratory investigation into the effect of water content on the
CBR of a subgrade soil. In T. Schanz (Ed.), Experimental Unsaturated Soil
Mechanics (pp. 137–144). Berlin: Springer.

Arifin, YF and Schanz, T. 2009. Osmotic suction of highly plastic clays, Acta
Geotechnical, 4 (3), pp. 177-191.

ASTM D4767 .1999. Standard test method for consolidated-undrained triaxial

compression test on cohesive soils, 1999 Annual Book of ASTM, Vol..04 .08,
Philadelphia P.A, pp.860-867.

ASTM D 5298-10. 2010. Standard test method for measurement of soil potential
(suction) using filter paper.

ASTM D4546. 2013. Standard test method for expansion index of soils. Annual Book of
ASTM Standards, PA 4546.

Austroads. 2012. A guide to the structural design of road pavements. Austroads.

Basma, AA., Al-Homoud, AS and Malkawi, AH. 1995. Laboratory assessment of

swelling pressure of expansive soils. Applied Clay Science, 9, 355-368. 1995.

Baille, W, Tripathy, S and Schanz, T .2014. Effective stress in clays of various

mineralogy. Vadose Zone J. 13 doi:10.2136/vzj2013.06.0112 (this issue).

Barksdale, RD and Itani, SY. 1989. Influence of aggregate shape on base behaviour.
Transportation Research Record, No. 1227, Transportation Research Board,
Washington, D. C., 173–182.

Bishop, AW. 1954. The use of pore pressure coefficients in practice. Geotechnique, 4,
No 4,148-152.

Bishop, AW. 1959. The principle off effective stress. Norwegian Geotechnical Institute,
vol. 32, pp 1–5.

Bishop, AW, Alan, I, Blight, GE and Donald, IB. 1960. Factors controlling the strength
of partially saturated cohesive soils. Proc. Res. Conf. Shear Strength Cohes.
Soils. ASCE, pp 503 532.

Bishop, AW, and Blight, GE. 1967. Some aspects of effective stress in saturated and
unsaturated soils.” Géotechnique, 13(3), 177–197.

Page | 174

© Central University of Technology, Free State

 References

Blight, GE. 1967. Effective stress evaluation for unsaturated soils. ASCE journal of the
soil mechanics and foundation division. 125-149.

Blight, GE. 2013. Unsaturated Soil Mechanics in Geotechnical Practice. Leiden The
Netherlands, CRC Press/Balkema.

Bowles, J. 1992. Engineering properties of soil and their measurements. McGraw-Hill

Boston, 4thEdition.
Bulut, R, Lytton, RL and Wray, WK. 2001. Soil suction measurements by filter paper.
Geotechnical special publication number 115, Proceedings of Geo-Institute
Shallow Foundation and Soil Properties Committee Sessions at the ASCE 2001
Civil Engineering Conference, pp. 243-261.

Butalia, TS, Huang, J, Kim, DG and Croft, F. 2003. Effect of moisture content and pore
water pressure builds on resilient modulus of cohesive soils. Resilient Modulus
Testing for Pavement Components ASTM STP 1437.

Burczyk, James, M, Ksaibati, Khaled., Anderson-Sprecher, Richard. 1994. Factors

influencing determination of a subgrade resilient modulus value. in
Transportation Research Record 1462, TRB, National Research Council,
Washington, D.C. 1994, pp. 72-79.

Carmichael III RF. and Stuart, E.1978. Predicting Resilient Modulus: A study to
determine the mechanical properties of subgrade soils. Transportation Research
Record 1043, TRB, National Research Council, Washington, D.C., 1978, pp 20-
28.

Cary, CE and Zapata, CE. 2010. Enhanced model for resilient response of soils resulting
from seasonal changes as implemented in mechanistic-empirical pavement
design guide.

Cary, CE, Zapata, CE, Witcizak, MW, Kaloush, K E. 2011. Pore- water pressure
response of a soil subjected to traffic loading under saturated and unsaturated
conditions. In ASU Electronic Dissertations and Theses. Arizona State
University, 2011.

Cekerevac, C, Baltzer, S, Charlier, R, Chazallon, C, Erlingsson, S, Gajewska, B and

Pavšic, P. 2009. ˇ Water influence on mechanical behaviour of pavements:
Experimental investigation. In A. Dawson (Ed.), Water in road structures (pp.
217–242). Springer.
Chen, F.H. 1988. Foundations on expansive soils. Elsevier scientific publishing
company, Amsterdam.

Page | 175

© Central University of Technology, Free State

 References

Cho, GC and Santamarina, JC. 2001. Unsaturated Particulate Materials - Particle Level
Studies. ASCE Geotechnical Journal. Vol. 127. No. 1. pp. 84-96.

Coleman, JD. 1962. Stress strain relations for partly saturated soils. Geotechnique, 12,
348-350.

Cui, YJ and Delage, P. 1996. Yielding and plastic behaviour of an unsaturated silt,
Géotechnique 46, No. 2, pp. 291-311.

Dai, S and Zollars, J. 2002. Resilient modulus of Minnesota road research project
subgrade soil. Transportation Research Record No. 1786, Transportation
Research Board, pp. 20–28.

Darter, MI, Elliot, RP and Hall, KT. 1991. Revision of AASHTO pavement overlay design
procedures, Appendix: Documentation of Design Procedures. National
Cooperative Highway Research Program Study 20-7, TRB, National Research
Council, Washington, D.C., April 1991.

Das, BM. 1984. Principles of foundation engineering. USA: Brooks/Cole Eng. Division.
Elias, V. (1994). Strategies for managing unknown bridge foundations. USA:
federal highway administration.

Das, BM. 2006. Principles of geotechnical engineering. Stamford, CT: Learning College.

Donald, IB. 1957. Effective stresses in unsaturated non-cohesive soils with controlled
negative pore pressure. M.Eng. Sc. Thesis, University of Melbourne, Melbourne,
Australia.

Drumm EC, Boateng-Poku, Y and Johnson Pierce, T. 1990 Estimation of Subgrade

Resilient Modulus from Standard Tests. Journal of Geotechnical Engineering,
Vol. 116, No.5, May 1900. pp 774-789.

Drumm, EC, Reeves, JS, Madgett, MR, and Trolinger, WD.1997. Subgrade resilient
modulus correction for saturation effects. Journal of Geotechnical and
Geoenvironmental Engineering, 123(7): p. 663–670.

Ekblad, J. 2008. Water in coarse granular materials: Resilient and retentive properties.
pp.117–123.

Erzin, Y and Erol, 0. 2004. Correlations for quick prediction of swell pressures. The
Electronic Journal of Geotechnical Engineering [serial online], 9(F): Paper No.
0476.

Page | 176

© Central University of Technology, Free State

 References

Escario, V and Juca, JFT. 1989. Shear strength and deformation of partly saturated
soils. Proceedings of the 12th International Conference on Soil Mechanics and
Foundation Engineering, Rio de Janerio, 2: 43-46.

Fang, Z and Yin JH. 2006. Physical modelling of consolidation of Hong Kong marine
clay with prefabricated vertical drains. Canadian Geotechnical Journal 43(6):
638-652.

Farrar, MJ and Turner, JP. 1991. Resilient Modulus of Wyoming Subgrade Soils.
Mountain Plains Consortium Report No 91-1, The University of Wyoming,
Luramie, Wyoming.

Fattom, M and Barakat, S. 2000. Investigation of Three methods for Evaluating Swelling
pressure of Soils. Environmental & Engineering Geoscience, Vol. VI, No. 3, pp.
293-299. 2000.

Feng, G, Meiri, A and Letey, J. 2003. Evaluation of a model for irrigation management
under saline conditions. Effects on plant growth. Soil Science Society of America
Journal, 67, 71-76.

Fleureau, JM, Kheirbek-Saoud, S, Soemitro, R and Taibi, S. 1993. Behaviour of clayey

soils on drying-wetting paths. Can. Geotech. J. 30: 287-296.

Fredlund, DG, Bergan, AT and Wong, PK. 1977. Relation between resilient modulus
and stress research conditions for cohesive subgrade soils. Transportation
Research Record No. 642, Transportation Research Board, Washington, D.C.,
pp. 73–81.

Fredlund, DG and Morgenstern. NR. 1977. Stress state variables for unsaturated soils.
Journal of Geotechnical and Geoenvironmental Engineering, 103, 447-466.

Fredlund, DG, Morgenstern, NR and Widger, RA. 1978. Shear strength of unsaturated
soils. Canadian Geotechnical Journal, 15, 313-321.

Fredlund, DG and Rahardjo, H. 1993. Soil mechanics for unsaturated soils. New York.
John Wiley and sons, Inc.

Fredlund, DG and Xing, A. 1994. Equations for the Soil-Water Characteristic Curve.
Canadian Geotechnical Journal 31, no. 4 (August 1, 1994): 521–32.

Fredlund, DG. 1996. The scope of unsaturated soil mechanics. An overview. In Alonso
and Delage (eds.). Proceedings 1st International Conference on Unsaturated
Soils, Paris, France. Balkema Press.

Page | 177

© Central University of Technology, Free State

 References

Fredlund, DG. 1996. The emergence of unsaturated soil mechanics. The fourth Spencer
J. Buchanan Lecture, College Station, Texas, A&M University Press, p. 39.

Fredlund, DG, Xing, A, Fredlund, MD and Barbour, S. L. 1996. The relationship of the
unsaturated soil shear strength to the soil-water characteristic curve. Canadian
Geotechnical Journal, Vol. 33, No. 3, pp. 440-448.

Fredlund, DG., Rahardjo, H, Leong, EC and Ng, CWW. 2001. Suggestions and
recommendations for the interpretation of soil-water characteristic curves. In:
Southeast Asian Geotechnical Conference, Hong Kong, (Vol. 1,), pp. 503–508.

Fredlund, DG. 2006. Unsaturated soil mechanics in engineering practice, Journal of

Geotechnical and Geoenvironmental Engineering, Vol. 132, No. 3, pp. 286-321.

Fredlund, DG, Rahardjo, H and Fredlund, MD. 2012. Unsaturated soil mechanics in
engineering practice., A Wiley-Interscience Publication, John Wiley & Sons, Inc,
pp. 535.

Gallipoli, D, Gens, A., Sharma, R and Vaunat, J. 2003. An elasto-plastic model for
unsaturated soil incorporating the effects of suction and degree of saturation on
mechanical behaviour. Géotechnique,53, 123-136.

Geiser, F, Laloui, L and Vulliet, L. 2000. On the volume measurement in unsaturated

triaxial test. Proc., The Asian Conference on Unsaturated Soil, pp. 669-679.

George, KP. 2004. Prediction of resilient modulus from soil index properties. Final
report. University of Mississippi, Department of Civil Engineering, Oxford.

Goudie, A.S. 1973. Duricrusts in tropical and subtropical landscapes. Oxford.

Gupta, S, Ranaivoson, A, Edil, T, Benson, C and Sawangsuriya, A. 2007. Pavement

design using unsaturated soil technology.

Hall, DK and Thompson, MR. 1994. Soil-property-based subgrade resilient modulus

estimation for flexible pavement design. Transportation Research Record, 1449,
TRB, National Research Council, Washington, D.C., pp. 30-38.

Heydinger, AG, Xie, QL, Randolph, BW and Gupta, JD. 1996. Analysis of resilient
modulus of dense and open-graded aggregates. Transportation Research
Record, No. 1547, Transportation Research Board, Washington, D.C., 1–6.

Page | 178

© Central University of Technology, Free State

 References

Heydinger, AG., 2003. Evaluation of seasonal effects on subgrade soils. Transportation

Research Record No. 1821, Transportation Research Board, pp.47–55.

Hicks, RG. 1970. Factors influencing the resilient properties of granular materials. Ph.D.
thesis, University of California, Berkeley, Berkeley, CA.

Hicks, RG and Monismith, CL. 1971. Factors influencing the resilient properties of
granular materials. Highway Research Record. 345.

Hopkins, T, Beckham, TL., Sun, C and Ni, B. 2004. Resilient modulus of Kentucky Soils.
Final research report for Kentucky transportation cabinet, Kentucky
Transportation Centre, University of Kentucky.

Hossain, Z, Zaman, M, Doiron, C and Solanki, P. 2010. Characterization of Subgrade

Resilient Modulus for Pavement Design. In Geo-Frontiers 2011. American
Society of Civil Engineers, Reston, VA, 2011, pp. 4823-4832.

Huang, J., 2001. Degradation of resilient modulus of saturated clay due to pore water
pressure build-up under cyclic loading. Thesis (Master), Department of Civil
Engineering and Environmental Engineering, Ohio State University.

India Standard (IS). 2020. Method of test for soils part xl determination of free swell
index of soils, 24 Bureau of Indian Standards, 2720, Part XL.

Jaquin, PA, Augarde, CE, Gallipoli D. and Toll DG. 2009. The strength of rammed earth
materials, Geotechnique. 59(5): 487–490.

Kamal, MA, Dawson, AR, Garouki, T, Hughes, DAB and Shaa’at, AA. 1993. Field and
laboratory evaluation of the mechanical behaviour of unbound granular materials
in pavements, Transportation Research Record No. 1406. 1993. pp 88-97.

Kancherla, A. 2004. Resilient modulus and permanent deformation testing of unbound

granular materials., Master’s Thesis, Texas A&M University.

Kang DH, Gupta, Ranaivoson, AZ, Roberson R and Siekmeier. 2011. Recycled
materials a substitute for virgin aggregates in road construction II, Inorganic
contaminant leaching.” science society of America journal, Vol.75. No. 4 1276-
1284.

Khoury, NN., and Zaman, MM. 2004. Correlation between resilient modulus, moisture
variation, and soil suction for subgrade soils. Transportation Research Record:
Journal of the Transportation Research Board 1874, no. 1 (2004): 99–107.

Page | 179

© Central University of Technology, Free State

 References

Khalili, N and Khabbaz, M. 1998. A unique relationship of chi for the determination of
the shear strength of unsaturated soils. Géotechnique, 48.

Khalili, N, Geiser, F and Blight, GE. 2004. Effective stress in unsaturated soils: Review
with new evidence. Int. J. Geomech., 4(2), 115–126.

Leong, EC., He, L and Rahardjo, H. 2002, Factors affecting the filter paper method for
total and matric suction measurements. Geotech. Test. J., Vol. 25, No. 3, pp.
322–333.

Likos, WJ, Wayllace A, Godt, J, Lu, N. 2010. Modified direct shear apparatus for
unsaturated sand at low suction and stress, Geotechnical Testing Journal,
Vol.33, No. 4: p. 1-13.
.
Janoo, VC and Bayer II JJ. 2001. The effect of aggregate angularity on base course
performance. U.S. Army Corps of Engineers, ERDC/CRREL TR-01-14.
Jennings, J. 1961. A revised effective stress law for use in the prediction of the
behaviour of unsaturated soils. Pore Pressure and Suction in Soils, 26-30.

Jennings, JE and Burland, JB. 1962. Limitations to the use of effective stresses in partly
saturated soils. Géotechnique, 12(2), 125–144.

Jones, MP and Witczak, M.W. 1977. Subgrade modulus on the San Diego Test Road.
Transportation Research Record, TRR 641, Transportation Research Board,
National Research Council, Washington, DC, pp. 1 6.

Lee, WJ., Bohra, NC., Altschaeffl, AG., and White, TD., 1995. Resilient modulus of
cohesive soils and the effect of freeze-thaw. Canadian Geotechnical Journal, Vol.
32, 1995, pp. 559-568.

Lee, W, Bohra, NC, Altschaeffl, AG and White, TD. 1997. Resilient modulus of cohesive
soils. Journal of Geotechnical and Geoenvironmental Engineering, Vol. 123, No.
2, 1997, pp. 131-135.

Leite, FC, Motta, RS, Vasconcelos, KL. 2011. Laboratory evaluation of recycled
construction and demolition waste for pavements Construction and Building
Materials, 25 (6) (2011), pp. 2972–2979.

Lekarp, F, Isacsson, U and Dawson, A. 2000. State of Art I: Resilient response of

unbound aggregates. Journal of Transportation Engineering, ASCE, Vol (1)
Jan/Feb. 2000.

Page | 180

© Central University of Technology, Free State

 References

Leong, EC, Tripathy, S and Rahardjo, R. 2003. Total suction measurement of

unsaturated soils with a device using the chilled-mirror dew-point technique.
Géotechnique, 53(2):173-182.

Liang, RY, Rabab'ah, S and Khasawneh, M. 2008. Predicting moisture-dependent

resilient modulus of cohesive soils using soil suction concept. Journal of
Transportation Engineering, 134(1): 34– 40.

Lu, N and Likos, WJ. 2004. Unsaturated soil mechanics. John Wiley & Sons, Inc.,
Hoboken, New Jersey.

Lu, N and Likos, WJ. 2006. Suction stress characteristic curve for unsaturated soil.
Journal of Geotechnical and Geoenvironmental Engineering, 132, 131.

Lu, N, Godt, JW and Wu, DT. 2010. A closed-form equation for effective stress in
unsaturated soil. Water Resources Research, 46, W05515.

Maatouk, A, Leroueil, S and La Rochelle, P. 1995. Yielding and critical state of a

collapsible unsaturated silty soil. Geotechnique, 45(3):465-477.

Mallela, J, VonQuintus, H and Smith, KL. 2004. Consideration of lime-stabilized layers

in mechanistic-empirical pavement design. Rep. prepared for The National Lime
Association, National Lime Association, Arlington VA.

Mohammed, LN, Huang, B., Puppala, AJ., and Allen, A. 1999. Regression model for
resilient modulus of subgrade soil. Transportation Research Record No. 1687,
Transportation Research Board, Washington, D.C.

Morrow, C, Shi, LQ and Byerlee. J. 1984. Permeability of fault gouge under confining
pressure and shear stress, J. Geophys. Res., 89, 3193–3200,
doi:10.1029/JB089iB05p03193.

Muraleetharan, KK and Wei, C. 2000. A fully coupled analysis procedure for dynamic
behaviour of unsaturated soils”, in Advances in Unsaturated Soils, Geotechnical
Special Publication No. 99 (ed. C. Shackleford, S.L. Houston and N-Y. Chang),
Reston: American Society of Civil Engineers, pp. 165-179.

Nam, S, Gutierrez, M, Diplas, P, Petrie, J. 2011. Determination of the shear strength of

unsaturated soils using the multistage direct shear test. Eng. Geol. 122 (3-4),
272–280.
National Cooperative Highway Research Program (NCHRP). 2002. Development of the
2002 guide for the design of new and rehabilitated pavement structures. Project
No. 1-37A, NCHRP Transportation Research Board, Washington, D.C.
Page | 181

© Central University of Technology, Free State

 References

National Cooperative Highway Research program (NCHRP). 2004. Development of the

2002 Guide for Design of New and Rehabilitated Pavement structures”. Phase2.

National Cooperative Highway Research program (NCHRP). 2003. Harmonized test

method for laboratory determination of resilient modulus for flexible pavement
design. Final Report 1-28A, National Cooperative Highway Research Program
(NCHRPA), Transportation Research Board, National Research Council,
Washington, D.C.

National Cooperative Highway Research program NCHRP. 2004. Guide for Mechanistic
Empirical Design of New and Rehabilitated Pavement Structure: Part 2 Design
Inputs: Chapter 3 Environmental Effects. Transportation Research Board,
National Research Council.

Nazzal MD, Mohammad LN. 2010. Estimation of resilient modulus of subgrade soils for
design of pavement structures. Journal of Materials in Civil Engineering 22(7):
726-734.

Ng, CWW and Pang, YW. 2000a. Experimental investigation of soil–water

characteristics of a volcanic soil. Can. Geotech. J., 37(6), 1252–1264.

Ng, CWW and Pang, YW. 2000b. Influence of stress state on soil–water characteristics
and slope stability. J. Geotech. and Geoenviron. Eng., ASCE. 26(2), ASCE, 157–
166.

Ng, CWW and Chiu, CF. 2001. Behaviour of a loosely compacted unsaturated volcanic
soil.” J. of Geotech. and Geoenviron. Engrg., ASCE. 127(12), 1027–1036.
Ng, CWW and Chiu, CF. 2003a. Laboratory study of loose saturated and unsaturated
decomposed granitic soil. J. of Geotech. and Geoenviron. Engrg., ASCE. 129(6),
550–559.

Ng, CWW. and Chiu, CF. 2003b. Invited keynote paper. Constitutive modelling of
unsaturated loose soil slopes subjected to rainfall infiltration. Proc. 9th China
National Conf. on Soil Mech. and Geotech. Engrg., Beijing. 1, pp. 187–200.

Ng, CWW. and Zhou, RZB. 2005. Effects of soil suction on dilatancy of an unsaturated
soil. Proc. 16th Int. Conf. Soil Mech. & Geotech. Engrg. Osaka, Japan, 2, 559–
562.

Ng, CWW, Zhou, C, Yuan, Q and Xu. 2013. Resilient modulus of unsaturated subgrade
soil. Experimental and technical investigations Canadian Geotech. 50(2), 223-
232.

Page | 182

© Central University of Technology, Free State

 References

Nicholson, P, Kashyap, V and Fuji, C. 1994. Lime and fly ash admixture improvement
of tropical Hawaiian soils. Transportation Research Record 1440, National
Research Council, Washington, D.C., 71–78.

Nishimura, T, Toyota, H, Vanapalli, SK and Oh, WT. 2007. Evaluation of critical state
parameters for an unsaturated soil. Proc., 60th Canadian Geotechnical
Conference, Ottawa (Canada), 1029-1036.

Noorany, I. 1984. Phase relations in marine soils. Journal of Geotechnical Engineering,

ASCE, 110, 539-543.

Nokkaew, K, Tinjum, J. William L and Edil, T. 2014. Effect of matric suction on resilient
modulus for compacted recycled asphalt base course in post compaction state.
transportation research record (2014): i-iv.

Ooi, PSK, Archilla, AR., and Sandefur, KG. 2004. Resilient modulus for compacted
cohesive soils. Transportation Research Record, No. 1874, pp. 115–124.

Oberg, A and Sallfors, G. 1997. Determination of Shear Strength Parameters of

Unsaturated Silts and Sands Based on the Water Retention Curve. Geotechnical
Testing Journal, Vol. 20, No. 1, pp. 40-48.

Paraire, J. 1987. Suction tests on CBR-diameter specimens. The bearing capacity

suction relation. Crowthorne, GB: Transport and Road Research Lab (TRRL).

Pezo, R, Claros, G, Hudson, WR, and Stoke, KH. 1992. Development of a Reliable
Resilient Modulus Test for Subgrade and Non-Granular Subbase Materials for
Use in A Routine Pavement Design. Research Report 1177-4F., University of
Texas-Austin.

Pham, HQ, Fredlund, DG and Lee Barbour, S. 2005. A study of hysteresis models for
soil-water characteristic curves. Canadian Geotechnical Journal 42, no. 6
(December 2005): 1548–68.

Purwana, Y, Nikraz, H and Jitsangiam, P. 2012. Experimental study of suction-

monitored CBR test on sand–kaolin clay mixture. International Journal of
Geomaterials, 3(2), 419–422.

Pusch, R. and Yong, R. 2003. Water saturation and retention of hydrophilic clay buffer
microstructural aspects, Applied Clay Science, Elsevier, 23, pp. 61-68.

Page | 183

© Central University of Technology, Free State

 References

Rahim, AM and George, KP. 2004. Subgrade soil index properties to estimate resilient
modulus. Transportation Research Board, CD-ROM of Transportation Research
Board Annual Meeting.

Rashed K. A. 1985. Swelling characteristics of compacted soil in Mosul Area. MSc.

Thesis Mosul University.

Rassam, DW, and Cook, FJ. 2002. Predicting the shear strength envelope of
unsaturated soils. Geotechnical Testing J. 28: 215-220.

Richter, C. 2006. Seasonal Variations in the Moduli of Unbound Pavement Layers.

Publication No. FHWA-HRT-04-079, Turner–Fairbanks Highway Research
Centre, McLean, Va.

Ridley, A. M. 1993. The measurement of soil moisture suction. PhD dissertation.

Imperial College, University of London.

Saklecha, PP, Kedar, RS. and Saklecha, PK. 2013. Correlation of CBR with mechanical
properties of foundation soil, Proc. of Indian Geotechnical Conference December
22-24, Roorkee, pp. 1-8, 2013.

Sánchez-Leal, F. 2002. Interpretation of CBR test results under the shear strength
concept of unsaturated soil mechanics. 3rd international conference on
unsaturated soils (pp. 663–668). Recife: CRC Press.
Santha, BL. 1994. Resilient modulus of subgrade soils: comparison of two constitutive
equations. Transportation Research Record No. 1462, Transportation Research
Board, Washington, D.C.

Sauer, EK. and Monismith, CL. 1968. Influence of Soil Suction on Behaviour of a Glacial
till Subjected to Repeated Loading. Highway Research Record, No. 215,
Washington, D.C., pp. 18-23.

Seki, K. 2007. SWRC fit – A nonlinear fitting program with a water retention curve for
soils 10 having unimodal and bimodal pore structure. Hydrol. Earth System Sci.
Discus, 4, 407- 11 437.

Sibley, JW and Williams, DJ. 1990. A new filter material for measuring soil suction.
Geotech. Test. J., Vol. 13, No. 4, pp. 381–384.

Sivakumar, V and Wheeler, S. 2000. Influence of compaction procedure on the

mechanical behaviour of an unsaturated compacted clay. Part 1: Wetting and
isotropic compression. Geotechnique 2000; 50(4): 359-368.

Page | 184

© Central University of Technology, Free State

 References

South African national road agency limited (SANRAL). 2008. Framework for a highway
system: integration of the pavement design method and a highway planning
system.” Report No. PB/2007/HPS.

Sreedeep, S. and Singh, D. 2006. Methodology for determination of osmotic suction of

soils. Geotechnical and Geological Engineering, 24, 1469-1479.

Sridharan, A and Venkatappa Rao, G. 1973. Mechanisms controlling volume change of

saturated clays and the role of the effective stress concept. Géotechnique 23(3),
359–382.

Tarantino, A. and De Col, E. 2008. Compaction Behavior of Clay. Geotechnique 58 (3):

199–213.

Terzaghi, K. 1936. The shearing resistance of saturated soils and the angle between
the planes of shear. 1st International Conference on Soil Mechanics and
Foundation Engineering.

Terzaghi, K. 1943. Theoretical Soil Mechanics.” New York, USA, Wiley.

Theyse, HL. 2000. The development of mechanistic-empirical permanent deformation

design models for unbound pavement materials from laboratory and accelerated
pavement test data. Proc., Int. Symp. on Un-Bound Aggregates in Road
Construction, A. A. Balkema, Rotterdam, The Netherlands, 285–293.

Tindall, JA. and Kundell, JR. 1999. Unsaturated zone hydrology for scientists and
engineers. Prentice-Hall. 624 pages.

TMH1 (METHOD 7). 1986. Determination of the maximum dry density and optimum
moisture content of gravel, soil and sand. Technical Methods for Highway.

TMH1 (METHOD 8). 1986. The determination of the determination of the California
bearing ratio of untreated soils and gravels. Technical Methods for Highway.

TMH1 (METHOD 14). 1986. The determination of the dry bulk density, apparent relative
density and water absorption of aggregate retained on a 4,75 Mm Sieve.
Technical Methods for Highway.

Traylor. KL. 1971. Finite element analysis of layered road pavements. PhD Thesis,
University of Nottingham.

Page | 185

© Central University of Technology, Free State

 References

Thompson, MR and Figueroa, JL. 1989. Mechanistic thickness design procedure for
soil-lime layers.Transp. Res. Rec., 754, 32–36.

Thomson, MR and Robnett, QL. 1976. Resilient properties of subgrade soils. Final
Report-Data Summary, Transportation Engineering Series 14, Illinois
Cooperative Highway Research and Transportation Program Series, University
of Illinois, Urbana.

Thomson MR, Robnett QL. 1979. Resilient properties of subgrade soils. Transportation
Engineering J. ASCE 105.

Thom, NH and Brown, SF. 1987. The effect of moisture on the structural performance
of a crushed limestone road base. Annual Transportation Research Board
Meeting, Washington, D. C. 1987.

Thom, NH., and Brown, SF. 1989. The mechanical properties of unbound aggregates
from various sources.’’ Unbound aggregates in roads, edited by RH. Jones and
AR. Dawson.

Vanapalli, SK, Fredlund, DG, Pufahl, DE and Clifton, AW. 1996. Model for the prediction
of shear strength with respect to soil suction. Canadian Geotechnical Journal,
Vol. 33, No. 3, pp. 379-392.

Van Genuchten, MT. 1980. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Science Society of America Journal, 44,
892-898.

Vinale, F, d’Onofrio, A., Mancuso, C, Santucci de Magistris, F and Tatsuoka, F. 2001.

The pre-failure behaviour of soils as construction materials. Pre-Failure
Deformation Characteristics of Geomaterials. Edited by M. Jamiolkowski, R.
Lancellotta and D. LoPreti. Swets & Zeitlinger, Lisse. pp. 955-1007.

Von Quintus, H and Killingsworth, B. 1998. Analyses relating to pavement material

characterizations and their effects on pavement performance. FHWA-RD-97-
085. Federal Highway Administration, McLean, VA.

Wan, AWL, Gray, MN and Graham, J. 1995. On the relations of suction moisture content
and soil structure in compacted clays. Proceedings of the 1st International
Conference on Unsaturated Soils, Paris, France, pp. 215-222.

Wang, B, 2000. Stress effects of soil water characteristics curve on slope stability in
expansive soils. MPhil Thesis, the Hong Kong University of Science and
Technology, Hong Kong, China.

Page | 186

© Central University of Technology, Free State

 References

Wilson, BE, Sargand, SM, Hazen, GA and Green, R. (1990). “Multiaxial Testing of
Subgrade”, Transportation Research Record, 1278, TRB, National Research
Council, Washington, D.C., pp. 91-95.

Witczak, M and Uzan, J. 1998. The universal airport design system, Report I of IV.
Granular material characterization. Rep. to Department of Civil Engineering,
Univ. of Maryland, College Park, Md.

Wolfe, WE and Butalia, TS. 2004. Seasonal instrumentation of SHRP pavements. Final
Report for Ohio Department of Transportation, Ohio State University, Columbus.

Yau, A and Von Quintus, H. 2002. Study of laboratory resilient modulus test data and
response characteristics.” Report. No. FHWA-RD-02-051. FHWA, U.S.
Department of Transportation, Washington, D.C.

Yang, SR, Huang, WH and Tai, YT. 2005. Variation of resilient modulus with soil suction
for compacted subgrade soils. Transportation Research Record,1913: 99–106.
doi:10.3141/1913-10.

Yang, SR, Lin, HD, Kung, JHS and Huang, WH. 2008. Suction controlled laboratory test
on resilient modulus of unsaturated compacted subgrade soils. Journal of
Geotechnical and Geo-Environmental Engineering, Vol. 134, No. 9, pp. 1375-
1384.
Zaman, M and Khoury, N. 2007. Effect of soil suction and moisture on resilient modulus
of subgrade soils in Oklahoma. Final Report for Oklahoma Department of
Transportation. Norman. Oklahoma.

Zhan, LT. 2003. Field and laboratory study of an unsaturated expansive soil associated
with rain-induced slope instability. PhD Thesis. Hong Kong University of Science
and Technology.

Zienkiewic, ZOC, Chan, AHC, Pastor, M, Schrefler, BA and Shiomi, T. 1999.

Computational geotechnics with special reference to earthquake engineering
Wiley London.

Page | 187

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Appendix A

Qualitative analysis results

Phase name Formula Figure of merit Phase reg. detail DB card number
quartz low, alpha-Si
Si O2 0.887 ICDD (PDF2010) 01-086-1629
O2
Iron Oxide Fe2 O3 2.669 ICDD (PDF2010) 00-056-1302

K Al (Mg0.2 Al0.8)
Muscovite-1M,
(Al0.42 Si3.58) O10 1.33 ICDD (PDF2010) 01-070-1868
magnesian, syn
(O H )2

Kaolinite-1A Al2 Si2 O5 (O H )4 1.435 ICDD (PDF2010) 01-072-2300

(Ca0.06 Na0.21
K0.27) (Al1.64
Bentonite Fe0.06 Mg0.31) 6.014 ICDD (PDF2010) 01-076-7629
(Al0.29 Si3.71 O10
(O H )2)

Goethite, cadmian, (Fe0.99 Cd0.01) O

1.722 ICDD (PDF2010) 01-072-8205
syn (O H)

(Fe +2, Fe +3, Al )3

Berthierine-1H 1.530 ICDD (PDF2010) 00-031-0618
(Si, Al )2 O5 (OH )4

Illite- K Al4 (Si, Al )8 O10

0.514 ICDD (PDF2010) 00-035-0652
montmorillonite (O H )4 ·4 H2 O
(NR)

Phase name Formula Space group Phase reg. detail DB card number
quartz low, alpha-Si Si O2 152: P3121 ICDD (PDF2010) 01-086-1629
Iron Oxide Fe2 O3 63: Cmcm ICDD (PDF2010) 00-056-1302
K Al (Mg0.2 Al0.8)
Muscovite-1M, 5: C121, unique-b,
(Al0.42 Si3.58) O10 ICDD (PDF2010) 01-070-1868
magnesian, syn cell-1
(O H )2
Kaolinite-1A Al2 Si2 O5 (O H )4 1: C1 ICDD (PDF2010) 01-072-2300
(Ca0.06 Na0.21
K0.27) (Al1.64 Fe0.06 12: C12/m1, unique-
Bentonite ICDD (PDF2010) 01-076-7629
Mg0.31) (Al0.29 b, cell-1
Si3.71 O10 (O H )2)

Goethite, cadmian, (Fe0.99 Cd0.01) O (O 62: Pbnm ICDD (PDF2010) 01-072-8205

syn H)

(Fe +2, Fe +3, Al )3

Berthierine-1H 147: P-3 ICDD (PDF2010) 00-031-0618
(Si, Al )2 O5 (O H )4

Illite-montmorillonite K Al4 (Si, Al )8 O10 (O

- ICDD (PDF2010) 00-035-0652
(NR) H )4 ·4 H2 O

Page | 194

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Appendix B

APPENDIX B
Table 4.2: Soils Moisture Conditions and Swelling Stress Characteristics
Soils 𝝎% 𝛄𝐝 Surcharge force (kN) 𝑷𝒔𝒏 (kPa) Ring area (m2)
(kN/m3)
FSS 1
8.35 12.45 1.37 699 0.01963
13.56 16.63 1.18 600 0.01963
17.48 17.32 0.88 450 0.01963
24.32 15.90 0.46 390 0.01963
31.64 12.88 0.59 300 0.01963

FSS 2
10.13 12.47 1.77 899 0.01963
13.61 15.62 1.47 749 0.01963
18.54 16.93 1.23 630 0.01963
24.12 14.15 0.88 450 0.01963
30.19 12.12 0.59 300 0.01963

FSS 3
7.33 11.45 0.98 499 0.01963
14.19 17.61 0.88 450 0.01963
18.48 16.59 0.77 390 0.01963
22.91 16.93 0.65 330 0.01963
28.30 13.47 0.49 250 0.01963

NCS 1
8.40 16.07 1.28 650 0.01963
12.70 18.92 1.18 600 0.01963
17.51 19.35 1.02 520 0.01963
24.52 17.52 0.88 450 0.01963
28.35 16.18 0.69 350 0.01963

NCS 2
9.20 15.45 1.28 630 0.01963
14.18 17.52 0.98 500 0.01963
18.40 18.52 0.88 450 0.01963
24.91 18.89 0.78 400 0.01963
30.23 15.58 0.69 350 0.01963

NCS 3
9.94 14.17 1.67 830 0.01963
14.55 15.86 1.42 725 0.01963
18.87 16.83 1.28 650 0.01963
25.26 15.95 1.02 520 0.01963
32.92 13.02 0.88 400 0.01963

GPS 1
8.69 16.47 0.96 490 0.01963
13.48 19.46 0.88 450 0.01963
17.30 21.83 0.74 375 0.01963
24.31 20.07 0.69 350 0.01963
32.72 16.16 0.49 250 0.01963

GPS 2
9.95 15.45 1.28 650 0.01963
13.69 19.23 1.18 600 0.01963
Page | 207

© Central University of Technology, Free State

 Appendix B

18.56 20.89 1.02 520 0.01963

24.11 18.96 0.88 450 0.01963
30.14 15.96 0.69 350 0.01963

GPS 3
9.30 18.13 1.02 520 0.01963
14.18 21.63 0.83 425 0.01963
18.85 22.84 0.78 400 0.01963
22.91 21.71 0.59 300 0.01963
27.30 18.83 0.39 200 0.01963
*γd = dry unit weight

* 𝑷𝒔𝒏 = swelling pressure

* 𝜔 = moisture water content

Table 4.3: Suction Values at Varying Moisture Content

Total suction
Soils GWC (%) FPWC (%) 𝛙𝐓 (kPa) 𝛙𝐌 (kPa) (𝛙𝐎 ) (kPa)
FSS 1
8.55 14.63 8283 6541 1742
13.46 18.78 7311 5676 1335
17.65 24.15 5593 4989 904
24.27 28.32 3993 2793 600
31.75 33.50 1327 921 406

FSS 2
9.97 14.93 10158 8517 1641
13.53 18.48 8164 6343 1321
18.58 23.93 6052 4989 1063
24.09 28.11 3797 2895 902
30.16 32.78 2634 1913 721

FSS 3
7.28 14.22 8454 6398 2056
14.15 19.43 6768 5345 1423
18.45 24.21 4689 3450 1239
22.94 27.43 2931 1998 933
28.33 31.12 1818 1240 578

NCS 1
8.38 18.97 3598 2534 1064
12.73 25.91 2781 2098 683
17.49 29.11 2198 1643 555
24.49 32.31 1495 1134 361
28.33 35.78 1093 778 315

NCS 2
9.18 19.23 4532 3878 1254
14.15 26.52 3495 2465 1030
18.38 30.34 3035 2087 948
24.89 32.82 2474 1598 876
30.22 34.34 1581 1091 490

Page | 208

© Central University of Technology, Free State

 Appendix B

NCS 3
9.97 19.45 2166 1595 971
14.53 25.76 1998 1545 453
18.98 29.65 1592 1172 310
25.24 31.54 1193 990 303
32.88 34.35 889 681 208

GPS 1
8.67 15.78 6554 5071 1483
13.46 22.91 5192 3884 1308
17.33 26.62 4253 2959 1294
24.27 30.14 2895 2183 712
32.75 33.32 1472 1128 344

GPS 2
9.97 14.96 6771 5263 1808
13.67 19.65 5699 3989 1210
18.58 25.98 4081 2914 1067
24.09 29.23 2758 2029 529
30.16 31.33 956 667 289

GPS 3
9.28 13.35 7280 5331 1949
14.15 19.46 4974 3708 1266
18.88 21.84 3642 2707 935
22.94 30.29 2297 1715 582
27.33 34.39 998 763 235
*GWC = Gravimetric water content
*FPWC = Filter paper water content
*ψT , ψ𝑀 and ψ𝑂 = Total, Matric and Osmotic suction

Table 4.4: Fitting Parameters of Difference SWRC Models for FSS

FSS 1
Model Equation Parameters R2 AIC

𝜃𝑠 =0.54642
𝑚
Van Genuchten Se = [
1
] (m= 1-1/n) 𝜃𝑟 =3.97841e-05 0.93016 -75.337
1+(𝛼ℎ)𝑛 𝛼 = 0.0025426
n= 1.3251
𝑚
1 𝜃𝑠 =0.59715
Fredlund and Se =𝑄 [ 𝑛 ] 𝜃𝑟 = 4.4119e-06
𝑙𝑛[𝑒+(ℎ⁄𝑎) ]
Xing 𝑎= 2255.8 0.95550 -79.198
𝑚=2.5655
𝑛 = 0.59392

Page | 209

© Central University of Technology, Free State

 Appendix B

𝜃𝑠 =0.60402
ln(ℎ⁄ℎ 1) 𝜃𝑟 = 5.9737e-08
𝑚
Seki Se =𝜔1 𝑄 [ ] + 𝜔1 =0.18115
𝜎1
ℎ𝑚1 =16.297 0.97754 -84.087
ln (ℎ⁄ℎ 2)
(1 − 𝜔1 )𝑄 [ 𝑚
] 𝜎1 = 1.7275
𝜎2 ℎ𝑚2 = 3767.5
𝜎2 = 1.6344

FSS 2
Van Genuchten 𝜃𝑠 =0.53788
𝑚
Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑟 =2.7835e-05
1+(𝛼ℎ)𝑛 𝛼 = 0.0017918 0.93317 -78.051
n= 1.3507

𝜃𝑠 = 0.55708
𝑚
1
𝜃𝑟 = 3.4180e-06
Fredlund and Se = [ ℎ 𝑛
] 𝑎=4384.7
𝑙𝑛[𝑒 + (𝑒 ) ]
Xing 𝑎 𝑚= 2.8559 0.96142 -83.195
𝑛=0.73219

𝜃𝑠 =0.56810
ln(ℎ⁄ℎ 1) 𝜃𝑟 = 2.2280e-06
𝑚
Seki Se =𝜔1 𝑄 [ ] + 𝜔1 =0.18665 0.99813 -118.58
𝜎1

ln(ℎ⁄ℎ 2) ℎ𝑚1 = 54.936

𝑚
(1− 𝜔1 )𝑄 [ ] 𝜎1 = 0.93116
𝜎2
ℎ𝑚2 = 5053.5
𝜎2 = 1.0062

FSS 3
𝑚
Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.50863
Van Genuchten 1+(𝛼ℎ)𝑛 𝜃𝑟 =1.9479e-07
𝛼 = 0.0032176 0.91957 -76.980
n= 1.3068

𝜃𝑠 = 0.52975
𝑚
1 𝜃𝑟 = 1.6845e-05
Fredlund and Se = [ 𝑛 ] 𝑎=2672.7
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
Xing 𝑚= 2.6044 0.94750 -80.527
𝑛=0.64936

𝜃𝑠 =0.53867
𝜃𝑟 = 1.4097e-06
Seki ln(ℎ⁄ℎ 1) 𝜔1 =0.20857 0.9898513 -97.887
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 =40.659
𝜎1

ln(ℎ⁄ℎ 2) 𝜎1 = 1.1023
𝑚
(1−𝜔1 )𝑄 [ ] ℎ𝑚2 = 3791.6
𝜎2
𝜎2 = 1.0595

Page | 210

© Central University of Technology, Free State

 Appendix B

Table 4.5: Fitting Parameters of Difference SWRC Models for NCS

NCS 1
Model Equation Parameters R2 AIC
𝑚
Van Genuchten Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.49334
1+(𝛼ℎ)𝑛 𝜃𝑟 =3.2549e-05
𝛼 = 0.0019231 0.91356 -76.028
𝑛 = 1.5498

𝜃𝑠 = 0.51977
𝑚
1 𝜃𝑟 = 3.6868e-06
Fredlund and Xing Se = [ 𝑛 ] 𝑎=1568.6
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
𝑚= 2.7477 0.93197 -77.141
𝑛=0.84386

𝜃𝑠 =0.53365
𝜃𝑟 = 2.2035e-06
Seki ln(ℎ⁄ℎ 1) 𝜔1 =0.20987 0.99876 -125.25
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 =29.232
𝜎1
𝜎1 = 1.2094
ln (ℎ⁄ℎ 2)
(1 − 𝜔1 )𝑄 [ 𝑚
] ℎ𝑚2 = 1754.1
𝜎2 𝜎2 = 0.66531
NCS 2
𝑚
Van Genuchten Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.52915
1+(𝛼ℎ)𝑛 𝜃𝑟 =3.25871e-05
𝛼 = 0.0014801 0.93152 -77.458
𝑛 = 1.6255

𝜃𝑠 = 0.57650
𝑚
1 𝜃𝑟 = 1.3018e-05
Fredlund and Xing Se = [ 𝑛 ] 𝑎=1151.1
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
𝑚= 2.1934 0.89959 -70.481
𝑛=0.60755

𝜃𝑠 =0.57028
𝜃𝑟 = 0.024685
Seki ln(ℎ⁄ℎ 1) 𝜔1 =0.22872 0.99986 -152.10
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 =43.896
𝜎1
𝜎1 = 1.1845
ln (ℎ⁄ℎ 2)
(1 − 𝜔1 )𝑄 [ 𝑚
] ℎ𝑚2 = 2109.3
𝜎2 𝜎2 = 0.57823
NCS 3
𝑚
Van Genuchten Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.58368
1+(𝛼ℎ)𝑛 𝜃𝑟 =4.4307e-06
𝛼 = 0.005422 0.87339 -66.375
𝑛 = 1.3433

𝜃𝑠 = 0.60885
𝑚
1 𝜃𝑟 = 2.9804e-05
Fredlund and Xing Se = [ 𝑛 ] 𝑎=1064.0
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
𝑚= 2.5958 0.92085 -70.482
𝑛=0.79245

Seki 𝜃𝑠 =0.62055 0.99970 -138.97

𝜃𝑟 = 0.043033
Page | 211

© Central University of Technology, Free State

 Appendix B

ln(ℎ⁄ℎ 1) 𝜔1 =0.23381
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 = 26.594
𝜎1

ln(ℎ⁄ℎ 2) 𝜎1 = 1.1599
𝑚
(1−𝜔1 )𝑄 [ ] ℎ𝑚2 = 1196.9
𝜎2
𝜎2 = 0.52036

Table 4.6: Fitting Parameters of Difference SWRC Models for GPS

GPS 1
Model Equation Parameters R2 AIC
𝑚
Van Genuchten Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.59867
1+(𝛼ℎ)𝑛 𝜃𝑟 =2.6553e-05
𝛼 = 0.004833 0.91231 -70.455
𝑛 = 1.3266

𝜃𝑠 = 0.61057
𝑚
1 𝜃𝑟 = 8.8741e-08
Fredlund and Se = [ 𝑛 ] 𝑎= 2102.1
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
Xing 𝑚= 2.6498 0.94392 -74.265
𝑛=0.69976

𝜃𝑠 =0.61742
𝜃𝑟 = 5.4004e-06
Seki ln(ℎ⁄ℎ 1) 𝜔1 =0.20791 0.99942 -129.62
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 = 41.449
𝜎1

ln(ℎ⁄ℎ 2) 𝜎1 = 1.0729
𝑚
(1−𝜔1 )𝑄 [ ] ℎ𝑚2 = 2847.3
𝜎2
𝜎2 = 0.79985

GPS 2
𝑚
Van Genuchten Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.51900
1+(𝛼ℎ)𝑛 𝜃𝑟 = 5.3120e -05
𝛼 = 0.0051261 0.85796 -68.486
𝑛 = 1.2128

𝜃𝑠 = 0.56468
𝑚
1 𝜃𝑟 = 1.4060e-05
Fredlund and Se = [ 𝑛 ] 𝑎= 1900.6
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
Xing 𝑚= 2.6818 0.91031 -81.309
𝑛=0.55083

𝜃𝑠 =0.55899
𝜃𝑟 = 0.0040401
Seki ln(ℎ⁄ℎ 1) 𝜔1 =0.99989 0.95458 -68.463
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 = 2649.7
𝜎1
𝜎1 = 3.1068
ln (ℎ⁄ℎ 2)
(1 − 𝜔1 )𝑄 [ 𝑚
] ℎ𝑚2 =
𝜎2 5.19783+06
𝜎2 = 0.21145
GPS 3
𝑚
Van Genuchten Se = [
1
] (𝑚 = 1 − 1/𝑛) 𝜃𝑠 =0.51361
1+(𝛼ℎ)𝑛 𝜃𝑟 =2.3320e-07
Page | 212

© Central University of Technology, Free State

 Appendix B

𝛼 = 0.0046693 0.87927 -74.565

𝑛 = 1.1396

𝜃𝑠 = 0.53769
𝑚
1 𝜃𝑟 = 6.3827e-06
Fredlund and Se = [ 𝑛 ] 𝑎= 811.36
𝑙𝑛[𝑒 + (ℎ⁄𝑎) ]
Xing 𝑚= 1.9089 0.93348 -80.304
𝑛=0.46674

𝜃𝑠 =0.53473
𝜃𝑟 = 0.012003
Seki ln(ℎ⁄ℎ 1) 𝜔1 =0.47621 0.99937 -136.95
𝑚
Se =𝜔1 𝑄 [ ] + ℎ𝑚1 = 134.36
𝜎1

ln(ℎ⁄ℎ 2) 𝜎1 = 2.9783
𝑚
(1−𝜔1 )𝑄 [ ] ℎ𝑚2 = 3819.8
𝜎2
𝜎2 = 0.62959

Page | 213

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Appendix C

500 500
𝑐 ′ = 51.21 kPa 𝑐 ′ = 46.59 kPa
𝜙 = 41.420 𝜙 = 37.670
400 400

Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟐𝟗: FSS 1 failure envelope at 17.65% and 24.27% moisture content

500 500
𝑐 ′ = 43.11 kPa 𝑐 ′ = 53.32 kPa
𝜙 = 34.780 𝜙 = 42.410
400 400
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0
0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟎: FSS 1, FSS 2 failure envelope at 31.75% and 9.97% moisture content

500 500
𝑐 ′ = 45.52 kPa
𝑐 ′ = 42.32 kPa
𝜙 = 36.910
𝜙 = 32.140
400 400
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟏: FSS 2 failure envelope at 13.67% and 18.58% moisture content

Page | 228

© Central University of Technology, Free State

 Appendix C

500 500
𝑐 ′ =38.43 kPa 𝑐 ′ =35.18 kPa
𝜙 = 28.510 𝜙 = 24.630
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟐: FSS 2 failure envelope criterion at 24.49% and 32.15% moisture content

500 500
𝑐 ′ =62.30 kPa 𝑐 ′ =58.42 kPa
𝜙 = 51.210 𝜙 = 46.510
400 400
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟑: FSS 3 failure envelope at 7.28% and 14.15% moisture content

500 500
𝑐 ′ =55.64
kPa 𝑐 ′ =51.23 kPa
𝜙 = 42.480 𝜙 = 37.140
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟒: FSS 3 failure envelope at 18.54% and 22.94% moisture content
Page | 229

© Central University of Technology, Free State

 Appendix C

500 500
𝑐 ′ =47.34 kPa 𝑐 ′ =53.41 kPa
𝜙 = 34.160 𝜙 = 46.220
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟓: FSS 3, NCS 1 failure envelope at 28.33% and 8.38% moisture content

500 500
𝑐 ′ =48.91 kPa 𝑐 ′ =44.31 kPa
𝜙 = 43.130 𝜙 = 39.420
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟔: NCS 1 failure envelope at 12.73% and 17.49% moisture content

500 500
𝑐 ′ =41.37 kPa 𝑐 ′ =34.57 kPa
𝜙 = 33.650 𝜙 = 27.210
400 400
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟕: NCS 1 failure envelope at 24.40% and 28.33% moisture content

Page | 230

© Central University of Technology, Free State

 Appendix C

500 500
𝑐 ′ =47.32 kPa 𝑐 ′ =41.45 kPa
𝜙 = 42.510 𝜙 = 37.610
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟖: NCS 2 failure envelope at 9.18% and 14.53% moisture content

500 500
𝑐 ′ =37.81 kPa 𝑐 ′ =33.62 kPa
𝜙 = 33.530 𝜙 = 27.110
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100
100

0
0 500 1000 0
0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟑𝟗: NCS 2 failure envelope at 18.38% and 24.89% moisture content

500
500
𝑐 ′ =26.13 kPa 𝑐 ′ = 42.22 kPa
𝜙 = 23.840
400 𝜙 = 37.340
400
Shear Stress (kPa)

Shear Stress (kPa)

300
300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟎: NCS 2, NCS 3 failure envelope at 30.22% and 9.97% moisture content
Page | 231

© Central University of Technology, Free State

 Appendix C

500 500
𝑐 ′ = 37.12 kPa 𝑐 ′ = 32.61 kPa
𝜙 = 33.430 𝜙 = 28.840
400 400

Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000
0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟏: NCS 3 failure envelope at 14.53% and 18.9% moisture content

500 500
𝑐 ′ = 27.31 kPa 𝑐 ′ = 22.37 kPa
𝜙 = 25.470 𝜙 = 21.640
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟐: NCS 3 failure envelope at 25.24% and 32.88% moisture content

500 500
𝑐 ′ =25.36 kPa
𝑐 ′ =20.12 kPa
𝜙 = 48.410
400 400
𝜙 = 44.120
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0
0
0 500 1000
0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟑: GPS 1 failure envelope at 9.28% and 12.89%moisture content
Page | 232

© Central University of Technology, Free State

 Appendix C

500 500
𝑐 ′ =17.14 kPa 𝑐 ′ =15.22 kPa
𝜙 = 42.910 𝜙 = 41.380
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0
0
0 500 1000
0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟒: GPS 1 failure envelope at 16.77% and 23.65%moisture content

500
500
𝑐 ′ =13.52 kPa 𝑐 ′ =22.71 kPa
𝜙 = 38.930 𝜙 = 45.230
400 400
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟓: GPS 1, GPS 2 failure envelope at 29.88% and 9.97%% moisture content

500 500
𝑐 ′ =16.14 kPa
𝜙 = 37.820
400 400
𝑐 ′ =19.43 kPa
Shear Stress (kPa)
Shear Stress (kPa)

𝜙 = 40.740
300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟔: GPS 2 failure envelope at 12.53% and 18.53% moisture content
Page | 233

© Central University of Technology, Free State

 Appendix C

500 500
𝑐 ′ =14.52 kPa 𝑐 ′ =11.55 kPa
𝜙 = 35.360 𝜙 = 32.210
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟕: GPS 2 failure envelope at 23.25% and 30.33% moisture content

500 500
𝑐 ′ =33.71 kPa 𝑐 ′ =27.45 kPa
𝜙 = 54.340 𝜙 = 52.220
400 400
Shear Stress (kPa)
Shear Stress (kPa)

300 300

200 200

100 100

0 0
0 500 1000 0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟖: GPS 3 failure envelope at 9.28% and 14.37% moisture content
500 500
𝑐 ′ =23.81 kPa 𝑐 ′ =18.61 kPa
𝜙 = 46.520 𝜙 = 40.340
400 400
Shear Stress (kPa)

Shear Stress (kPa)

300 300

200 200

100
100

0
0 500 1000 0
0 500 1000
Principal Stress (kPa) Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟒𝟗: GPS 3 failure envelope at 18.88% and 23.53% moisture content

Page | 234

© Central University of Technology, Free State

 Appendix C

500
𝑐 ′ =14.52 kPa
𝜙 = 36.220
400

Shear Stress (kPa)

300

200

100

0
0 500 1000
Principal Stress (kPa)
𝐅𝐢𝐠𝐮𝐫𝐞 𝟓. 𝟓𝟎: GPS 3 failure criterion at 27.45% moisture content

Table 5.1: Saturated and unsaturated shear strength parameters

Soil 𝑪′ 𝝈𝒏𝒇 𝒕𝒂𝒏𝝓′ 𝝍𝒎 𝒕𝒂𝒏𝝓𝒃 𝝉𝒔 𝝉𝒖 (𝝉𝒖 - 𝝉𝒔 )

moisture (kPa) (kPa) (degree) (kPa) (degree) (kPa) (kPa) (kPa)
content
(%)
FSS 1
8.55 58.10 388 49.29 6541 509 927 418
13.46 53.64 356 43.38 5976 3.66 390 772 382
17.65 51.21 283 41.42 4689 301 601 300
24.27 46.59 222 37.67 2793 218 397 179
31.75 43.11 178 34.78 921 168 226 58

FSS 2
9.97 53.32 358 42.41 8517 380 627 247
13.67 45.52 306 36.91 6843 276 473 197
18.58 42.32 281 32.14 4989 1.66 219 364 145
24.49 38.43 212 28.51 2895 154 237 83
32.15 35.18 158 24.63 1913 108 163 55

FSS 3
7.28 62.30 432 51.21 4498 600 1185 585
14.15 58.42 424 46.51 4045 505 1031 526
18.54 55.64 338 42.48 3250 7.41 365 788 423
22.94 51.23 264 37.14 2298 251 550 299
28.33 47.34 259 34.10 1440 223 410 187

NCS 1
8.38 53.41 493 46.22 2534 568 982 414
12.73 48.91 468 43.13 2098 487 959 472
17.49 44.31 416 39.42 1643 12.68 387 726 339
24.40 41.37 355 33.65 1134 278 533 255
28.33 34.57 315 27.21 778 197 372 175

NCS 2
9.18 47.32 398 42.51 3278 399 904 505
Page | 235

© Central University of Technology, Free State

 Appendix C

14.53 41.45 318 37.61 2465 287 657 370

18.38 37.81 288 33.53 2087 8.53 229 542 313
24.89 33.62 229 27.11 1598 151 397 246
30.22 26.13 189 23.34 1091 108 272 164

NCS 3
9.97 42.22 356 37.34 1895 314 803 489
14.53 37.12 312 33.43 1525 15.64 243 670 427
18.98 32.61 267 28.84 1272 180 503 323
25.24 27.31 241 25.47 1010 142 398 256
32.88 25.37 227 21.64 681 116 307 191

GPS 1
9.28 25.36 567 48.41 5021 664 1166 502
12.89 20.12 522 44.12 3884 526 914 388
16.77 17.14 470 42.91 2959 5.71 454 733 279
23.65 15.22 445 41.38 2183 407 610 203
29.88 13.52 392 38.93 1128 330 443 113

GPS 2
9.97 22.71 418 45.23 5263 444 781 337
12.53 19.43 388 40.74 3989 3.66 354 589 235
18.53 16.14 358 37.82 2914 290 464 174
23.25 14.52 328 35.36 2029 247 377 130
30.33 11.55 303 32.21 667 203 245 42.4

GPS 3
9.28 33.71 529 54.34 5331 737 1371 634
14.37 27.45 487 52.22 3708 628 1045 417
18.88 23.81 458 46.52 2907 6.42 508 834 326
23.53 18.61 422 40.34 1715 358 570 212
27.45 14.52 382 36.22 763 280 380 100

Page | 236

© Central University of Technology, Free State

© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
© Central University of Technology, Free State
 Appendix D

Table 6.1: Parameters symbols and their corresponding values

Denotations Meaning Values R2
Model 1
𝛇𝟎 Intercept 369.6004 -
𝛇𝟏 Coff. for 𝛾𝑑𝑟𝑦 4.8039 -
𝜸𝒅𝒓𝒚 Dry density - 0.7271
𝛇𝟐 Coff. for Percentage passing #200 1.1861 -
𝑷#𝟐𝟎𝟎 Percentage passing #200 - 0.7555
𝛇𝟑 Coff. for Moisture content 5.3152 -
𝑴𝒄 Moisture content - 0.8640
𝛇𝟒 Coff. for Liquid limit 0.0659 -
𝐋𝐋 Liquid limit - 0.8640
𝛇𝟓 Coff. for Swelling pressure 0.0551 -
𝑷𝒔 Swelling pressure - 0.8740

Model 2
𝛌𝟎 Intercept 135.6336 -
𝛌𝟏 Coff. for 𝛾𝑑𝑟𝑦 2.1865 -
𝛄𝐝𝐫𝐲 Dry density - 0.7264
𝛌𝟐 Coff. for CBR 6.2707 -
𝐂𝐁𝐑 CBR 0.7464
𝛌𝟑 Coff. for Moisture content 3.0062 -
𝑴𝒄 Moisture content - 0.8068
𝛌𝟒 Coff. for Swelling pressure 0.0341 -
𝑷𝒔 Swelling pressure - 0.8358
𝛌𝟓 Coff. for Compression strength 0.0941 -
𝒒𝒖 Compression strength - 0.8445
𝛌𝟔 Coff. for Shear stress 0.1981 -
𝝉𝒔 Shear stress - 0.9355

Model 3
𝛈𝟎 Intercept 188.2355 -
𝛈𝟏 Coff. for 𝛾𝑑𝑟𝑦 2.1662 -
𝜸𝒅𝒓𝒚 Dry density - 0.7436
𝛈𝟐 Coff. for CBR 8.9390 -
CBR CBR - 0.7825
𝛈𝟑 Coff. for Percentage passing #200 0.7787 -
𝑷#𝟐𝟎𝟎 Percentage passing #200 - 0.7855
𝛈𝟒 Coff. for Moisture content 3.3032 -
𝑴𝒄 Moisture content - 0.8643
𝛈𝟓 Coff. for Liquid limit 0.5347 -
𝐋𝐋 Liquid limit - 0.8643
𝛈𝟔 Coff. for Swelling pressure 0.0155 -
𝑷𝒔 Swelling pressure - 0.8751
𝛈𝟕 Coff. for Compression strength 0.0940 -
𝒒𝒖 Compression strength - 0.8804
𝛈𝟕 Coff. for Shear stress 0.2106 -
𝝉𝒔 Shear stress - 0.9340

Page | 243

© Central University of Technology, Free State

 Appendix D

Table 6.2: Parameters symbols and their corresponding values

Denotations Meaning Values R2
Model 7
𝛏𝟎 Intercept 12.396 -
𝛏𝟏 Matric suction
Coff. for 0.0095 -
𝛙𝐦 Matric suction - 0.6117
𝛏𝟐 Coff. for Air entry value (AEV) 1.1861 -
𝑺𝒆 Air entry value - 0.8106
𝛏𝟑 Coff. for Unsaturated CBR u 1.9574 -
𝐂𝐁𝐑 𝐮 Unsaturated CBR u - 0.9097
𝛏𝟒 Coff. for Swelling pressure 0.0089 -
𝑷𝑺𝒏 Swelling pressure - 0.9104

Model 8
𝛂𝟎 Intercept 19.493 -
𝛂𝟏 Matric suction
Coff. for 0.0094 -
𝛙𝐦 Matric suction - 0.6117
𝛂𝟐 Coff. for Air entry value (AEV) 0.2183 -
𝑺𝒆 Air entry value (AEV) - 0.8106
𝛂𝟑 Coff. for Swelling pressure 3.0062 -
𝑷𝑺𝒏 Swelling pressure - 0.8169
𝛂𝟒 Coff. for Unsaturated CBR u 1.9782 -
𝐂𝐁𝐑 𝐮 Unsaturated CBR u - 0.9104
𝛂𝟓 Coff. for Dry density 0.3311 -
𝛄𝐝𝐫𝐲 Dry density - 0.9109

𝐅𝐢𝐠𝐮𝐫𝐞 𝟔. 𝟐𝟗: Layer coefficient for asphalt concrete

Page | 244

© Central University of Technology, Free State

 Appendix J

APPENDIX J
Table 6.23: Measurement of soil suction using filter paper- Data sheet
ASTM D 5298 (1994)
Geotechnical Laboratory, Department of Civil Engineering , CUT
Soil designation …………………
Date tested: ………………….
Tested by: …………………..
Sample 1-1 1-2 1-3 1-4 1-5
Gravimetric water content of soil sample,W, ( %)
Tin No
Top filter paper / Bottom filter paper Top Bot Top Bot Top Bot Top Bot Top Bot
Cold Tare Mass, g Tc

Mass of wet Filter paper + Cold Tare Mass, g m1

Mass of Dry Filter paper + Hot Tare Mass, g m2

Hot Tare Mass, g Th

Mass of water in Filter Paper, g M2-Th Mf

Mass of water in Filter Paper, g M1-M2-Tc+Th Mw

Water content of filter Paper,g (Mw/Mf) % Wf

Suction , kPa Ψ
Suction , logkPa Ψ
Suction, PF = logkPa+1 Ψ

Page | 245

© Central University of Technology, Free State