Simulation by Abaqus FWD

SIMULATION OF FLEXIBLE PAVEMENT UTILIZING
FLY ASH AS ALTERNATIVE STABILIZER

By
JACOB ADEDAYO ADEDEJI
Submitted in fulfilment of the requirements for the degree:
MAGISTER TECHNOLOGIAE (M-Tech): Civil Engineering
Department of Civil Engineering
Faculty of Engineering and Information Technology
Central University of Technology
Free State, Bloemfontein
Supervisor: Dr. Mohamed Mostafa
2015
© Central University of Technology, Free State

Declaration and Copyright
Declaration and Copyright
I hereby declare that this dissertation submitted for the degree of Magister
Technologiae: Engineering: Civil, at the Central University of Technology, is my own
original work and has not previously been submitted to any other institution of higher
education. I further declare that all sources cited or quoted are indicated and
acknowledged by means of a comprehensive list of references.
_______________________ __________________
ADEDEJI J A DATE

Dedication
Dedication
This dissertation is dedicated to the ALMIGHTY GOD, my family and loved ones.
ii

Acknowledgements
Acknowledgements
Firstly, with a grateful heart, I give thanks to ALMIGHTY GOD for HIS love, grace,
strength, wisdom, knowledge and understanding. I thank HIM for always being by my
side. Without HIM, the completion of this research study would not have been
possible.
My deepest gratitude goes to my supervisor, Dr. Mohamed Mostafa. I appreciate you

for building me up in the field of research and for contributing immensely towards the
completion of my research study. Furthermore, my gratitude also goes to Mr. Michiel
Heyns, for making all laboratory data results used for this study available.
I would like to show my appreciation to my beloved parents, Mr Isaac Adedeji and

Mrs Rachel Adedeji for their love, prayers and encouragement. Also, a big thank you
goes to my uncle, Mr Taiwo Ogundana and his wife, Fadesayo Ogundana, and my
own aunt, Mrs Folashade Adedeji for their care, support and encouragement.
Likewise, I would like to acknowledge my colleagues who stood in as family here in
South Africa; Admire Mhlaba, Chimezie Jerry and Damian Franklin Okolie: thank you
all.
With deep sincerity of heart, I would like to thank the Central University of
Technology (CUT) Research and Innovation office for their financial support and also
commend the great effort of Ms Riana Dessels and Ms Sandra Nel.
iii

Abstract
Abstract
The stabilization process in pavement construction is not a new process, but hitherto
this process has not been fully implemented in the design methods for pavement
structure. Its partial implementation in design has contributed to the failures
experienced in pavement structure, which result in such pavement needing
excessive maintenance and rehabilitation, thereby increasing the operational cost of
the roads. Additionally, the use of an empirical design method for pavement structure
has led to the over-design of pavement, resulting in wasteful design and construction
of pavement structure. Nevertheless, Mechanistic-Empirical seems to be the way
out. Consecutively, with the advent of powerful design software based on different
methods such as the Finite Element (FE), Discrete Element, Finite Difference,
Boundary Element Methods, the possibility of design and construction of quality
pavement structures are enhanced. Therefore, the main focus of this study is to
provide a modelling tool for using fly ash as alternative stabilizer for base layers of
flexible pavement. To achieve the aim of the study, various objectives were set in
place based on literature reviews which are documented in this study.
Considering the fact that FE is the method most adopted in pavement analysis and
with the ability to obtain stresses and strains at the bottom of the surface layer, and
compressive stress/strain within the base layer and at the top of sub-grade, it was
considered in this study. Validations of a 3D FE model over 2D were conducted for
fly ash stabilized base layer. Thereafter, the importance of an asphalt layer on a
stabilized base layer was checked, and the efficiency of non-linear model for material
characterization was also checked. Overall, a comparative analysis of FE modelling
and an empirical method of pavement design was conducted. The results show that
the use of 3D FE models is more efficient than 2D axisymmetric models; use of a
non-linear material characterization model is more efficient than linear material
characterization, and the use of empirical design methods results in the over-
designing of pavement structure. Thus, the overall results suggest the use of 3D FE
models, coupled with a non-linear material characterization model are suitable for
the design of flexible pavement with a stabilized base layer.
iv

Abbreviations
Abbreviations
® Registered trademark symbol

2D Two dimensional
3D Three dimensional
AASHTO American Association of State Highway and Transportation Officials
ACAA American Coal Ash Association
Al2O3 Aluminium oxide
ASTM American Society for Testing and Materials Classification
C3 Cemented natural gravel materials used as base or sub-base
according to COLTO 1998 classification
C3D8R 8-node solid continuum elements with reduction integration
C4 Cemented natural gravel materials used as base or sub-base
according to COLTO 1998 classification
Ca Calcium
CaO Calcium oxide
CAX4R 4-node bilinear axisymmetric quadrilateral elements with reduction
integration
CBR California Bearing Ratio
COLTO Committee of Land Transport Officials
CSIR Council for Scientific and Industrial Research
CSLHA Coconut shell, leaf and husk ash
D-P Drucker-Prager plasticity model
FE Finite Element
Fe2O3 Iron (III) oxide or ferric oxide
FEM Finite Element Method
FM5-410 Field Manual 5-410
G5 Unbound granular used as sub-base material according to COLTO
1998 classification
GBS Granular blastfurnace slag
K Flow-stress ratio
kN kilo Newton

Abbreviations
ksi kilo-pound per square inch

m metre
M-C Mohr-Coulomb plasticity model
ME Mechanistic-Empirical method
MEDG Mechanistic-Empirical design guide
MgO Magnesium oxide
mm millimetre
MPa Mega Pascal
MR Resilient Modulus
Psi pound per square inch
SAMDM South African Mechanistic-Empirical Design Method
SANRAL South African National Road Agency Ltd
SAPDM South African Pavement Design Method
SiO2 Silicon dioxide
SN Structural Number
UCS Unconfined Compressive Strength Test
WAC West Africa Compaction
WASHO Western Association of States Highway Officials
ψ Dilation angle
vi

Table of Contents
TABLE OF CONTENTS
Abstract………………………………………………………………………………………iv
Abbreviations ............................................................................................................. v
List of Tables .............................................................................................................. x
List of Figures .............................................................................................................xi
CHAPTER 1: INTRODUCTION .................................................................................. 1
1.1. Background ............................................................................................ 1
1.1.1. Materials .................................................................................................... 1
1.1.2. Design and Analysis .................................................................................. 3
1.2. Problem Statement ................................................................................. 4
1.3. Research Aim ......................................................................................... 4
1.4. Specific Objectives ................................................................................. 4
1.5. Delimitation ............................................................................................. 5
1.6. Significance of the Study ........................................................................ 5
1.7. Outline of Dissertation ............................................................................ 5
CHAPTER 2: LITERATURE REVIEW ........................................................................ 7
2.1. Flexible Pavement ............................................................................................... 7
2.2. Pavement Composition and Behaviour ............................................................... 8
2.3. Secondary Materials............................................................................................ 9
2.3.1. Soil Stabilization Concept ....................................................................... 10
2.3.2. Fly Ash as Stabilizer ............................................................................... 12
2.4. Pavement Design Background .......................................................................... 14
2.5. Summary ........................................................................................................... 17
CHAPTER 3: NUMERICAL SIMULATION OF FLEXIBLE PAVEMENT ................... 18
3.1. Stresses, Strains and Deflections in Flexible Pavement ................................... 18
3.2. Approach of Mechanistic Empirical (M-E) Design ............................................. 20
3.3. Layered Elastic Simulation ................................................................................ 20
vii

Table of Contents
3.4. Finite Element Modelling (FEM) ........................................................................ 23
3.4.1. Critical Factors in FEM Simulation of Pavement ......................................... 24
3.4.1.1. Geometry............................................................................................... 24
3.4.1.2. Material Characterization....................................................................... 26
3.4.1.3. Analysis Type, Boundaries and Loading Conditions ............................. 27
3.4.2. Correlation Equations in FEM Simulation .................................................... 28
3.4.3. FEM Material Characterization via Direct Testing Results .......................... 30
3.5. 2D versus 3D Analysis ...................................................................................... 34
3.5.1. Motivation for 3D FEM ............................................................................ 36
3.5.2. Comparative Study on FEM Software ..................................................... 36
3.5.3. Abaqus® as FEM Software..................................................................... 37
3.6. Failure Criteria in Numerical Simulation ............................................................ 38
3.7. Summary ........................................................................................................... 39
CHAPTER 4: DESIGN AND SIMULATION .............................................................. 41
4.1. Introduction ....................................................................................................... 41
4.2. Development of Flexible Pavement Model ........................................................ 41
4.3. Abaqus®/CAE Usage for Flexible Pavement Model ......................................... 42
4.4. 3D versus Axisymmetric .................................................................................... 43
4.4.1. Description .................................................................................................. 43
4.4.2. Model Geometry and Material Properties .................................................... 43
4.4.3. Model Mesh and Element types .................................................................. 46
4.4.4. Boundary Conditions and Loading .............................................................. 47
4.5. Paved Stabilized Base Layer ............................................................................ 48
4.5.1. Description .................................................................................................. 48
4.5.2. Model Geometry and Material Properties .................................................... 49
4.5.3. Model Mesh and Element type .................................................................... 49
4.5.4. Boundary Conditions and Loading .............................................................. 49
viii

Table of Contents
4.6. Non-Linear versus Linear Material Characterization ......................................... 49
4.7. Comparative Analysis ........................................................................................ 50
4.8. Summary ........................................................................................................... 53
CHAPTER 5: RESULTS AND DISCUSSION ........................................................... 54
5.1. 3D versus Axisymmetric Results ....................................................................... 54
5.2. Paved Stabilized Base Layer Results................................................................ 58
5.3. Non-Linear versus Linear Material Characterization Results ............................ 62
5.4. Comparative Analysis Results ........................................................................... 65
5.5. Summary ........................................................................................................... 67
CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS .................................. 68
6.1. Introduction ....................................................................................................... 68
6.2. Conclusions ....................................................................................................... 68
6.3. Recommendations ............................................................................................ 69
6.4. Further Studies .................................................................................................. 69
REFERENCES ......................................................................................................... 70
APPENDICES ............................................................................................................ A
ix

List of Tables
List of Tables
Table 3.1. Summary of correlations between the unconfined compressive strength

and modulus............................................................................................................. 30
Table 3. 2. Summary of resilient modulus constitutive models. ................................ 33
Table 4. 1. Material properties of the stabilized base layer (obtained from Heyns and
Mostafa Hassan, 2013) ............................................................................................ 48
Table 4.2. Material properties of other pavement layers .......................................... 48
Table 4.3. Layer coefficients .................................................................................... 51
Table 4. 4 Comparison between Abaqus and mePADS ........................................... 52
Table 5.1. Rutting failure analysis based on Asphalt Institute response model (1982)
for Axisymmetric and 3D model ............................................................................... 58
Table 5.2. Fatigue and rutting failure analysis based on Asphalt Institute Response
Model (1982) for 3D model....................................................................................... 61
Table 5.3. Effect of asphalt layer thickness for non-linear material model................ 64
Table 5. 4 Effect of asphalt layer thickness for Abaqus and mePADS ..................... 65
Table 5.5. Structural capacity results for 1993 AASHTO, mePADS and 3D FEM
models...................................................................................................................... 66

List of Figures
List of Figures
Figure 2.1. Typical Flexible Pavement and Load Distributions ................................... 9

Figure 2.2. Flexible pavement design background ................................................... 16
Figure 3.1. Critical Pavement Responses and Locations ......................................... 19

Figure 3.2. FEM application stages .......................................................................... 23
Figure 3. 3 Some element families in ABAQUS ....................................................... 25
Figure 4.1. General steps for the development of flexible pavement (Abaqus®/CAE
usage) ...................................................................................................................... 45
Figure 4.2. Axisymmetric model geometry of the stabilized base and sub-grade layer
with meshing, load and boundary conditions............................................................ 46
Figure 4.3. 3D model geometry of the stabilized base and sub-grade layer with
meshing, load and boundary conditions ................................................................... 47
Figure 5.1. Effect of stabilized base layer thickness on vertical compressive strains
on the top of sub-grade (Axisymmetric model) ......................................................... 55
Figure 5.2. Effect of stabilized base layer thickness on vertical compressive stress on
the top of sub-grade (Axisymmetric model) .............................................................. 56
Figure 5.3. Effect of stabilized (18 percent fly ash + 1 percent cement) base layer
thickness on vertical compressive strains on top of sub-grade (Axisymmetric and 3D
model) ...................................................................................................................... 56
Figure 5.4. Effect of stabilized (18 percent fly ash + 1 percent cement) base layer
thickness on vertical compressive stress on top of sub-grade (Axisymmetric and 3D
model) ...................................................................................................................... 57
Figure 5. 5 Contour plots showing deformation of vertical compressive strain at the
top of sub-grade (A- 3D Model and B- 2D Model) .................................................... 57
Figure 5.6. Effect of asphalt layer thickness on surface deflection over a stabilized
base layer (3D Model) .............................................................................................. 59
Figure 5.7. Effect of asphalt layer thickness on vertical compressive strains on the
top of sub-grade (3D Model) .................................................................................... 59
xi

List of Figures
Figure 5.8. Effect of asphalt layer thickness on vertical compressive stress within
stabilized base/on the top of sub-grade (3D Model) ................................................. 60
Figure 5.9. Effect of asphalt layer thickness on tensile horizontal strain at the bottom
of asphalt layer (3D Model) ...................................................................................... 60
Figure 5.10. Effect of asphalt layer on vertical compressive strain within stabilized
base layer (3D Model) .............................................................................................. 61
Figure 5.11. Displacement contour plot for 25 mm thickness asphalt layer (A- Linear
model and B- Non-linear model) .............................................................................. 63
Figure 5.12. Strain contour plot for 25 mm thickness asphalt layer (A- Linear model
and B- Non-linear model) ......................................................................................... 63
Figure 5.13. Stress contour plot for 25 mm thickness asphalt layer (A- Linear model
and B- Non-linear model) ......................................................................................... 64
xii

Chapter 1 Introduction
CHAPTER 1: INTRODUCTION
1.1. Background
Road transportation among transportation modes has expanded the most over the
past 50 years, both for passengers and freight transportation (Rodrigue, Slack and
Comtois, 2013). In South Africa, there are 750 000 kilometres of road network and
9.7 million vehicles, which make all sectors of the economy depend on roads to
transport goods. The majority of goods, estimated at 83 percent, are transported by
road, and in addition, forecasts reveal that freight transport demand will grow by 200
percent to 250 percent over the next 20 years (Ndebele, 2012). Considering its
significant role in the economic and communication activities of the modern
societies, researchers have been searching to attain the most suitable road
pavement behaviour (Shafabakhsh, Motamedi and Family, 2013a), and
consequently design and construct safe, stable, cost-effective and environment-
friendly roads. With all the attention from researchers, pavement structures
experience failure before the desirable design life resulting from the low bearing
capacity of soil (Kordi, Endut and Baharom, 2010), overloading of the pavements,
inadequacy in designs and unsuitable design methods used (Kordi et al. 2010;
Shafabakhsh et al., 2013a). Its construction becomes uneconomical most often
because of the cost incurred on materials used. With an appropriate method of soil
stabilization, the soil’s stability may be improved; resulting in stable pavements as
well as the cost of construction may be reduced. However, the challenges with
respect to the design of pavements remain. With the advent of powerful design
software based on different methods such as the Finite Element, Discrete Element,
Finite Difference, and Boundary Element Methods, the possibility of design and
construction of quality pavement structures is enhanced. Therefore, in this study an
attempt is made to simulate the behaviours of the flexible road pavements having fly
ash as an alternative soil stabilizer, by using Finite Element Method (FEM).
1.1.1. Materials
Selection of materials for road pavement design is based on a combination of

suitable materials, environmental consideration, construction methods, economics

and previous experience (Bureau for Industrial Cooperation, 2012). Previously, road
construction had depended mainly on the virgin materials from the nearest borrow
pit, but in situations where the available soil lacks some geotechnical properties such
soil needs to be stabilized. Soil stabilization, refers to the method aimed at
increasing or maintaining the stability of soil mass and the chemical alteration of soils
to enhance their engineering properties via different techniques, such as mechanical
compaction, dewatering and addition of materials which are more advantageous
(Gyanen, Savitha and Gudi, 2013; Yadu and Tripathi, 2013). According to Aminaton,
Nima and Houman (2013), stabilizing soil using lime, cement, chemicals, plastics,
rice husk ash, millet husk ash, corn cob ash, coconut shell ash, foundry sand,
cement kiln dust, granular blastfurnace slag (GBS), or fly ash increases the soil’s
resistance, strength and permeability. Furthermore, results and experience show that
lime as a stabilizer yields better results than others, but its use will make pavement
structure uneconomical, which in turn makes fly ash an alternative stabilizer.
Fly ash, a finely divided residue that results from the combustion of pulverized coal,
an amorphous ferro-alumino silicate with a matrix very similar to soil and its
elemental composition varies with types and source of coal (Comberato, Vance and
Someshwar, 1997). These ash particles are transported from the combustion
chamber by exhaust gases as a result of their light weight and collected in control
devices such as filter bags and electrostatic precipitators. They are spherical in
shape and range in size from 0.5 micron to 100 micron (Heyns and Mostafa Hassan,
2013). From the point of view of the American Coal Ash Association (ACAA) (1995),
fly ash particles are composed of glass with crystalline matter, carbon, and varying
quantities of lime. Its chemical and physical properties depend greatly on several
factors such as production type, raw feed and the handling method. This in turn
gives the two classes of fly ash based on the chemical composition. Class C ashes
are from sub-bituminous and lignite coals and may contain more than 20 percent
CaO with 1 percent to 3 percent free lime, while Class F ashes are generally
obtained from bituminous and anthracite coal and contain less than 20 percent CaO
with no free lime ASTM C618 (ASTM-C618 2011).This industrial by-product is
considered in this research because it is readily available and various measures of
success have been achieved when used as stabilizer in pavement structures.

1.1.2. Design and Analysis
Pavement structural design is a daunting task with the basic geometry being quite
simple, while everything else is not. Its traffic loading is a heterogeneous mix of
vehicles, axle types, and axle loads with distributions that vary with time throughout
the day, from season to season, and over the pavement design life (Schwartz and
Carvalho, 2007). Also, pavement material characteristics such as viscoelasticity,
non-linearity and linearity, respond to these loads in complex ways coupled with
stress state and magnitude, temperature, moisture, time, loading rate, and other
factors. Previously, design was done by the empirical method, then by layered
elastic method, but as a result of the assumptions of the aforementioned, design
sometimes results in errors (Huang, 2004). Thus, to model pavements correctly, it is
necessary to use numerical methods, such as the finite difference method, the
boundary element method and FEM (Áurea, Evandro and Lucas, 2006). However,
FEM is the most adopted in pavement analysis and will be considered.
FEM is a numerical technique for finding approximate solution to boundary value

problems for differential equations, also with the ability of handling changes of
material properties such as Resilient Modulus and Poisson’s Ratio in both vertical
and horizontal directions and having successfully been used not only for designing
pavement structures, but also for optimizing the design by stimulation (Brooks,
Hutapea, Obeid, Bai, and Takkalapelli, 2008; Shafabakhsh et al., 2013a).
Additionally, it is suitable for eliminating tensile stresses in granular layers by stress
transfer method and also enables pavement designers to predict with some amount
of certainty the life of the pavement (Brooks et al. 2008). FEM includes two-
dimensional (2D) and three-dimensional (3D) methods, both of which can be
employed to capture the structural response of flexible pavements.
Lastly, the failures experienced in pavement structures has been a long-standing

global challenge to roads engineers, but with the success recorded through soil
stabilization (fly ash as stabilizer), it would be ideal to feature this in the FEM to cut
costs and time spent on laboratory work. In addition, the structural characteristics
and mechanical behaviours of stabilized soil bases have not been investigated so
extensively (Peng and He, 2009). Hence, this brought about this research work.

1.2. Problem Statement

The poor performance of flexible pavements results from the use of poor-quality
materials, inappropriate stabilization (Paige-Green, 2008) and/or inadequacy in
designs (Kordi et al., 2010; Shafabakhsh et al., 2013a). These later result in higher
expenditure on maintenance and rehabilitation. Consequently, this influences the
national annual budget, which is observed to be increasing at a higher rate every
year (Ndebele, 2012). Furthermore, it is observed that flexible pavements in most of
the South African roads experience permanent deformation (usually referred to as
rutting), cracking of surface course and creation of potholes (Council of Scientific and
industrial Research (CSIR) 2010), which are generally caused by various factors,
such as soil expansion (Kordi et al., 2010), inadequate soil stabilization,
inappropriate use of materials in the base courses and provision of inadequate
thickness of pavement layers. As a result, the roads need regular maintenance and
rehabilitation, which increases the operational cost of the roads. However, with the
help of FEM, the behaviour of the flexible pavements can be simulated and the
adequacy of pavement design can be examined by considering soil stabilizers as
one of the major influential parameters for flexible road pavements. Based on the
developed simulated scenarios, appropriate design and construction interventions
can be taken to design and construct the pavements adequately using fly ash as a
base course stabilizer, consequent upon which the cost of the maintenance and
rehabilitation of flexible pavement roads will be reduced. Hence, this study pertains
to the use of FEM to simulate the behaviour of flexible pavements of South African
roads in which fly ash will be considered as a soil stabilizer in the base course.
1.3. Research Aim
This research aims at providing a modelling tool for the use of fly ash as alternative
stabilizer for base layers of road. Moreover, this tool can be extended to other non-
traditional materials as well.
1.4. Specific Objectives
To achieve the aim of this research, the following specific objectives need to be
considered:

1. To evaluate the efficiency of using 3D FE model for design of flexible

pavement.
2. To determine the structural response of stabilized base layers in flexible
pavement system due to traffic loads using the 3D FE model.
3. To validate the use of fly ash as stabilizer through a 3D FE model.
4. To compare laboratory test empirical results already available against the
3D FEM results.
1.5. Delimitation
In this dissertation the development/formulation of a new mathematical model for the

characterization of the material (fly ash stabilized base) is not considered. However,
appropriate selection is made from the existing material characterization model.
Lastly, this selection is based on the proper findings from literature reviews and the
ability of the model to represent the behaviour of the material under loading.
1.6. Significance of the Study
This research is worth doing because of the important information it renders to future
road engineers and researchers. Overall, the design of new road projects located in
areas short of high-quality materials would result in very long material hauls. Thus, it
may require pavement structure alternatives other than the conventional granular
base. Understanding such material behaviour under loading is of great importance
for effective pavement design. Recent studies undertaken on the use of waste and
by-product materials as soil stabilizers have left a gap, between fly ash as stabilizer
(empirical design approach) and its computer-aided design for pavement structures.
Simulation of pavement structures has been carried out for different purposes, but
not in the use of fly ash as alternative stabilizing material. As a result, this study will
save time, human error and cost of laboratory experiments in carrying out projects
and address the problems relating to road construction industry.
1.7. Outline of Dissertation
The remainder of this dissertation is organized as follows:

Chapter 2: This chapter presents the literature review on flexible pavement,

pavement composition and behaviour, secondary materials in road construction and
a brief background to pavement design.
Chapter 3: This chapter continues with the literature review on numerical simulation
of flexible pavement with reviews on stresses, strains and deflections in flexible
pavement, approach of mechanistic empirical design, layered elastic and finite
element simulation of flexible pavement.
Chapter 4: This chapter presents the detail simulation design used for fly ash
stabilized base layer flexible pavement via Abaqus® 3D FEM analysis; four models’
analysis were used in line with the set objective and comparative analysis of
laboratory test empirical results, and 3D FEM was carried out.
Chapter 5: Presentations and discussions of results obtained from the models in

Chapter 4 were undertaken. These presentations were presented graphically and
through contour plots.
Chapter 6: This chapter gives the conclusions, general recommendations and

further studies for this dissertation.

Chapter 2 Literature Review
CHAPTER 2: LITERATURE REVIEW
2.1. Flexible Pavement
Before any design of pavement structure, an appropriate pavement type selection is

of importance as it is usually based on some critical factors such as soil composition,
climate, traffic volume, life cycle, constructability and cost. In addition, there are
secondary factors that need to be also considered, including: tire-pavement noise
generation, surface smoothness and environmental sustainability. Flexible
pavements have suitably met all the requirements, which made it to be used most
frequently (Asphalt Pavement Alliance, 2010).
Flexible pavements with asphalt on the surface are used all around the world. The
various layers of this pavement structure have different strength and deformation
characteristics which make the layered system difficult to analyse in pavement
engineering. At the surface there is a viscous material with its behaviour depending
on time and temperature, and pavement foundation geomaterials; coarse-grained
unbound granular materials in base/sub-base course; and fine-grained soils in the
sub-grade, exhibiting stress-dependent non-linear behaviour (Kim, 2007).
Furthermore, with the introduction of soil stabilization which brought about the use of
new materials with different characteristics such as cementitious and polymeric, the
design of flexible pavement has become more complex. However, the analysis of
pavement via empirical methods, as previously mentioned, sometimes result in
errors, but if material characterization is properly understood, finite element analysis
can be successfully used in the design of flexible pavement, which in turn makes
design adequate.
This chapter covers the literature review on pavement composition and behaviour,
secondary materials, soil stabilization concept, fly ash as stabilizing agent, and lastly,
details on pavement design background.

2.2. Pavement Composition and Behaviour
Pavement structure a composite system, consisting of superimposed layers of

processed materials above the natural soil sub-grade, with the primary function of
distributing the applied vehicle loads to the sub-grade. This structure’s ultimate aim
is to ensure that the transmitted stresses due to the loading are sufficiently reduced,
so that they will not exceed sub-grade bearing capacity (Adu-Osei, 2001). In other
words, the tensile and compressive stresses induced on the pavement by heavy
wheel loads decreases with increasing depth (Figure 2.1). In order to take maximum
advantage, pavement layers are usually arranged in order of descending load
bearing capacity, with the highest load-bearing capacity material on the top and the
lowest load-bearing capacity material at the bottom, as seen in flexible pavement
(Figure 2.1). However, in flexible pavements the unbound granular layers serve as a
major structural component of the structure (Adu-Osei, 2001). Further, in developing
countries like South Africa, the main structural element is formed by the unbounded
granular layer as thick base and sub-base layers placed over the sub-grade; and for
economic reasons, the asphalt layer is very thin, with a limited structural function,
which mainly provides protection against water ingress (Araya, 2011).
Overall, the material properties and their influence on pavement behaviour must be
thoroughly understood. According to The South African National Road Agency Ltd.
(SANRAL) (2013a), there are a number of fundamental properties that influence the
behaviour of a material regardless of its situation. These are: inter-particle friction,
particle distribution, cohesion, elasticity, particle hardness, durability and porosity. In
addition to fundamental properties, there are the situational properties that influence
the behaviour, such as density, moisture content and temperature. The majority of
these properties are considered in the design of pavement, but the most essential of
these are the engineering properties which are actually the basic results in the
design. Some of the engineering properties are: ultimate strength, elastic modulus,
resistance to deformation and crack propagation and fatigue, all obtained from
various laboratory tests.
Furthermore, various factors that have significant effects on the soil behaviour can
be loading condition, stress state, soil composition, compaction and soil physical
states (Kim, 2007). As a result of these factors the material characteristics of the

entire pavement change continuously over time with environmental changes which
later result in pavement failure. To avert pavement failure and reduce the cost of
hauling natural materials, researchers introduced the use of secondary materials.
Figure 2.1. Typical Flexible Pavement and Load Distributions (Steve Muench,
2003)
2.3. Secondary Materials
Using by-products, recycled and waste materials as alternatives to naturally

occurring aggregates in the construction of roads helps to conserve the supplies of
good-quality aggregates, leads to less energy and environmental cost associated
with the extraction and transportation of conventional aggregates, and assists in
problems arising from the disposal of unwanted materials (Sherwood, 1974). Such
materials are referred to as secondary materials or aggregates. This practice results
from the current and projected high demand for conventional aggregates and the
increasing difficulty of obtaining planning consent for their extraction. Moreover, this
is combined with a greater awareness of the considerable quantities of ‘waste
aggregates’ that are stockpiled and currently arising from the mineral extraction
industries, the construction/demolition industry and industrial processes. All these
have stimulated greater interest in the use of secondary materials in road
construction (Nunes, Bridges and Dawson, 1996; Brennan and O’Flaherty, 2002).
Some of the secondary materials considered for road works are blast furnace and
steel slag, spent oil shale, china clay waste, slate waste, rice husk ash, millet husk

ash, corn cob ash, coconut shell ash, waste foundry sand, cement kiln dust, fly ash,
bottom ash and demolition and construction waste (Sherwood, 1974; Mostafa
Hassan and Khalid, 2010; Amin, 2012; Bindu and Vysakh, 2012; Yadu and Tripathi,
2013). These materials are subjected to various laboratory tests before considering
their use for road construction work. Such laboratory tests may include grain size
analysis, specific gravity, compaction, Triaxial and leaching tests, etc., depending on
the material type. Overall, the use of any of these materials depends on its
availability at a particular location.
All in all, secondary materials are inferior to the natural materials used in
construction, but the lower cost of these inferior materials makes it an alternative if
adequate performance can be achieved (Heyns and Mostafa Hassan, 2013). In the
quest to certify the use of secondary materials, researchers discovered that the one
or a mixture of these materials with unstable natural materials yields an increase in
its engineering properties. Hence, this relates to the process called soil stabilization.
2.3.1. Soil Stabilization Concept
In South Africa, the bearing capacity of the pavement is provided by the unbound
base and sub-base or by the unbound base and stabilized sub-base (Araya, 2011).
The asphalt layer provides a smooth riding surface and provides skid resistance.
These structures have been successfully used in South Africa for moderately and
heavily loaded roads. However, the minimum California Bearing Ratio (CBR)
required for the sub-grade is 15 percent; when this is not reached, improvement of
the sub-grade should take place (Molenaar, 2009).
Yet the concept of soil stabilization is not new, as it can be dated back to 5000 years
ago. McDowell (1959) mentioned that stabilized earth roads were used in ancient
Mesopotamia and Egypt, and that the Greeks and Romans once used soil-lime
mixtures. Over the years, research has focused on improving the durability, safety
and efficiency of pavement materials and structures within both economic and
environmental constraints. This brought about the various means of stabilizing soil
which are practical and economical.
10

Soil stabilization mainly aims at improving soil strength and increasing resistance to
softening by water through bonding the soil particles together, waterproofing the
particles or a combination of the two (Sherwood, 1993). It is used to treat a wide
range of materials including expansive clays to granular materials (Openshaw,
1992). The stabilization process can be accomplished by several methods. All these
methods fall into two broad categories (FM5-410 2012), namely:
 Mechanical stabilization
Stabilization is achieved via a physical process by altering the physical nature of
natural soil particles by either induced vibration or compaction and also by
introducing coarse or fine materials and geosynthetic materials. Recently,
mechanical stabilization has been used for pavement structure through
geotextiles materials (Hejazi, Sheikhzadeh, Abtahi, and Zadhoush, 2012) which
yielded a great increase in the property strength of the structures. Further, using
a geogrid, Al-Azzawi, (2012) noted that placing this reinforcement at the base-
asphalt interface leads to the highest reduction of the fatigue strain.
 Chemical stabilization
Stabilization depends mainly on chemical reactions between stabilizer
(cementitious material) and soil minerals (pozzolanic materials) to achieve the
desired effect, including lime, cement, secondary materials and chemicals.
Recently, with the increase in the problem posed by secondary materials and its
availability locally, researchers considered their use as an alternative stabilizer.
Some of these are: Mgangira (2006) Waste Foundry Sand on clayey soils,
Bindu and Vysakh (2012) Coconut Shell, Leaf and Husk ash (CSLHA) on
lateritic soils, Yadu and Tripathi (2013) GBS and Fly ash on soft soils, and Amin
(2012) reviewed on soil stabilization using low-cost methods. Based on these
studies, GBS, foundry sand, CSLHA, fly ash and scrap tyres are low-cost and
effective as stabilizer. Further review will be done on fly ash as stabilizer as it is
a centre to this dissertation.
Nevertheless, among these stabilization methods, results have shown that chemical
stabilization is more advantageous (Makusa, 2012; Gyanen et al., 2013; Yadu and
Tripathi, 2013). Overall, researchers noted that the presence of organic matters,
11

sulphate, sulphide and carbon dioxide in the stabilized soils may inhibit the
stabilization process (Makusa 2012). Likewise, compaction, moisture content,
temperature and freeze-thaw further contribute (Sherwood, 1993; Makusa, 2012).
Additionally, Paige-Green (2008) noted that failure in stabilization process may
further result from lack of suitable skill and experience, inadequate specification,
change in construction equipment and construction techniques.
2.3.2. Fly Ash as Stabilizer
South Africa being the fourth largest producer of fly ash at 30 mega ton per year
after China, USA and India, results from the fact that coal plays an important role in
its economy and is the primary energy source for electricity generation (Furter,
2011). Fly ash is a heterogeneous material with SiO2, Al2O3, Fe2O3 and occasionally
CaO as its main chemical components. This ash also contains Ca-bearing minerals
such as Anorthite, Gehlenite, Akermanite and various Calcium Silicates and Calcium
Aluminates identical to those found in Portland Cement (Snellings, Mertens and
Elsen, 2012). Considering its production per year in South Africa, the government is
at the stage where it is strategically finding ways to reduce fly ash through treatment,
re-use and beneficiation (Heyns and Mostafa Hassan, 2013).
All over the world, fly ash is being used for various purposes such as cement
production, concrete production (Torii, Hashimoto, Kubo and Sannoh, 2013), soil
stabilization, asphalt (Lin Li, benson and Edil, 2007), embankment, flow-able fill and
waste stabilization owing to its cement-like property, yet in South Africa only 6
percent of the annual production is utilized. Further, in pavement structure, fly ash
has a wide application which is incorporated in sub-grade, granular base/sub-base,
asphalt base/surface and structural fill (United States Environmental Protection
Agency, 2009). Also, it has been combined with other products or by-products to
improve pavement materials and its light weight and ability to be handled easily on
construction site with little safety precaution (Kim, Prezzi and Salgado, 2005; Mathur,
2011; Heyns and Mostafa Hassan, 2013), contributes to it’s usage.
Various studies have been conducted on its utilization as stabilizer and as an

alternative to the use of virgin materials. Senol, Bin-Shafique, Edil, and Benson,
(2002) carried out a study on the use of self-cementing class C fly ash for the
12

stabilization of soft sub-grade. In this study, the optimum mix design and stabilized
layer thickness were estimated by strength and modulus-based approaches. The
results obtained showed that the engineering properties such as unconfined
compressive strength (UCS), CBR and resilient modulus increase substantially after
fly ash utilization. Also, in 2002, Pandian and Krishna conducted laboratory CBR
tests on the stabilized fly ash-soil mixtures and observed that fly ash is an effective
admixture for improving the soil quality. In addition, Brooks (2009) reported the soil
stabilization with rice husk ash and fly ash mixed together with natural soil, the study
showed improvement in CBR values and UCS. Also, researchers have proven that
mixtures of fly ash with inert materials reach 50 percent to 70 percent of the strength
of the corresponding cement-inactive materials (Eskioglou and Oikonomou, 2008).
Further, an innovative research was undertaken by Heyns and Mostafa Hassan

(2013), utilizing three different types of fly ash (Kendal Dump Ash, Durapozz and
Pozzfill) at 16, 18, 20 and 22 percent enhanced with cement on G5 sub-base
material classified according to the Committee of Land Transport (COLTO) (1998)
officials, and results show that G5 sub-base material is stabilized to meet up with the
C3 and C4 stabilized standard according to COLTO. Fly ash controls the shrink-swell
by cementing the soil grains together and also have the tendency to increase the
maximum dry density (Ban and Park, 2014) and improve the CBR of soil by 100
percent (Umar, Alhassan, Abdulfatah, and Idris, 2013). Consequently, any fly ash
that has at least some self-cementitious properties can be engineered to perform in
transportation projects.
Furthermore, studies have been concluded that, if fly ash is used properly, it is not
hazardous to the environment when used for soil stabilization. This was done with a
combination of batch-leaching tests to determine potential impact on the
environment of fly ash as trace element mobility in soil stabilization (Heebink and
Hasselt, 2001). Similarly, Tanosaki, Yu and Nagasaki (2011) studied an image of fly
ash being an ‘environmentally friendly’ product. The measurements were carried out
only on Hunter brightness or reflectivity. Three hundred lots of coal ash samples
were analysed, whereby it was determined that coal ash possesses a wide range of
colour hues. Due to strong correlations between hue and spherical rate, Chroma(C)
and CaO+MgO content of coal ash, it could be used as a base for quality control
13

standards. Overall, the use of fly ash is accepted worldwide due to saving in cement,
consuming industrial waste and making durable materials, especially due to
improvement in the quality when used as stabilizer (Heyns and Mostafa Hassan,
2013).
Generally, the compressive strength of fly ash-stabilized soils is dependent on in-

place soil properties, delay time, moisture content at time of compaction and fly ash
addition ratio as discussed by ACAA (1995). Summarily, the performance of
pavement structure depends on the satisfactory performance of each material used,
thus proper evaluation is required in respect to the properties of each material
separately. Overall, the proper understanding of the behaviour of natural materials,
secondary materials and the soil stabilization process are important to the successful
implementation of any design method.
2.4. Pavement Design Background
Pavement design is the process of developing the most economical combination of

pavement layers (in relation to both thickness and materials type) to suit the soil
foundation and the traffic to be carried during design life. From the SANRAL (2013b)
point of view, pavement design is to ensure that materials within the pavement layers
are not overstressed at any time during the course of these changes in the
pavement’s life. Over the years, in the pursuit of accurate simulation of pavement
structure behaviour under loading, various design methods have been developed.
These design methods, on individual capacity, have been used to simulate
pavement behaviour based on some assumptions. Figure 2.2 gives a background on
the design of flexible pavement.
At the outset, pavement designs were based on empirical methods which are back-
dated to the development of the Public Roads soil classification system in the 1920s
(Huang, 2004; Schwartz and Carvalho, 2007). Empirical methods are derived from
experience in terms of field observation performance of in-service pavement or
laboratory test sections. The purpose of laboratory methods is to subject a
representative pavement material sample to an environment (consisting of simulated
traffic loading and environmental conditioning) that closely simulates field conditions
(Adu-Osei, 2001). These methods also define the interaction between pavement
14

performance, traffic loads and pavement thickness for a given set of paving
materials, soil, location and environmental conditions (Schwartz and Carvalho,
2007). Although the design of flexible pavements is still largely empirically based,
these methods remain accurate only for the exact conditions for which they were
developed, and perhaps invalid outside the range of variables used in its
development.
This brought about the various examples of empirical design methods developed
with different location such as the American Association of State Highway and
Transportation Officials (AASHTO) in the USA (1993), Road Note in the UK (Road
Research Laboratory, 1970), Western Association of States Highway Officials
(WASHO) in Malad and West Africa Compaction (WAC) in West Africa Countries
(Fall, Ba, S., Sarr, Ba, M. and Ndiaye, 2011), to mention but a few.
Even though empirical methods tend to be simple and easy to use, these methods
are associated with various limitations such as one climate condition, limited traffic,
material type, and new construction only (i.e. cannot be used for rehabilitation). If
these conditions change, the design is no longer valid (Wang, 2001; Huang, 2004).
To further buttress this point, Huber, Andrewski, and Gallivan (2009) found that the
AASHTO 1993 pavement design guide typically over-designed pavements in Indiana
by 1.5 to 4.5 inches, amounting to approximately 600 to 800 tons of materials per
lane-mile beyond what is needed.
Before the final introduction of mechanistic-empirical design guide (MEDG) in the

21st century, other design methods aside, empirical methods were developed
between 1940 and the 1960s which are: limiting shear failure method, limiting
deflection method, and regression method, all based on pavement performance
and/or road test. However, these methods have various limitations and likewise do
not satisfy all necessary requirements for an ideal design which makes them
obsolete (Huang, 2004). Ultimately, pavement design methods differ from one to
another yet, they are affected by the same factors which are: traffic and loading,
structural models, materials, environment, and failure criteria. Nevertheless, a better
approach to the design of perpetual pavements is the mechanistic-empirical method.
15

Figure 2.2. Flexible pavement design background
16

2.5. Summary
Chapter Two of this dissertation dealt with the literature review on flexible pavement,
pavement composition and behaviour, secondary materials, the soil stabilization
concept, and fly ash as stabilizer and pavement design background. Research has
shown that the use of fly ash and other industrial by-products as stabilizer in
pavement structure has recorded great success and also, fly ash is proven to be
environment-friendly if proper precaution is taken into consideration when used.
Further, on the aspect of design, the inadequacy of empirical, limiting shear failure,
limiting deflection and regression based method for pavement design as contributed
to pavement failures, but the FEM is seen as a way out. With the foundation made in
this chapter, the next chapter will be building upon it by reviewing in detail on
numerical simulation of flexible pavement.
17

Chapter 3 Numerical Simulation of Flexible Pavement
CHAPTER 3: NUMERICAL SIMULATION OF FLEXIBLE PAVEMENT
3.1. Stresses, Strains and Deflections in Flexible Pavement
Pavement analysis has been transitioning from empirical methods to numerical

approaches (Kim, 2007). However, beforehand - due to the limitations of
computational capabilities - pavement designs were dominated first by empirical
methods which are limited to a certain set of environmental and material conditions,
but with the advent of powerful computer storage capabilities, this design can now be
done on personal computers (Adu-Osei, 2001; Huang, 2004; Shafabakhsh et al.,
2013a). In numerical approach (also known as mechanistic), the pavement is treated
as a layered structure with the proper understanding of its components in respect to
the constituent materials (Kim, 2007).
Mechanistic Analysis exploits mathematical capability to calculate the stress, strain,

or deflection in a multi-layered system such as pavement, when subjected to
external loads (Hafeez, 2010). The stresses, strains and deflections generated in
flexible pavements result from the material properties and thickness of each layer
and loading condition (Al-Khateeb, Saoud and Al-Msouti, 2011). Further, with the
use of computer programs one can evaluate the theoretical stresses, strains, and
deformations anywhere in the structure. However, there are a few critical locations
which are of interest and are often used in pavement analysis (NCHRP, 2004;
Pavement Interactive, 2008; Darwish, 2012; SANRAL, 2013b) (Figure 3.1) such as;
 Surface deflection
 Tensile horizontal strain at the bottom of the surface course (for surface course
fatigue cracking)
18

 Compressive vertical stresses/strains within the base/sub-base layers (for

rutting of unbound layers)
 Compressive vertical stresses/strains at the top of the sub-grade (for sub-grade
rutting)
Figure 3.1. Critical Pavement Responses and Locations (Pavement Interactive,

2008)
Furthermore, in mechanistic analysis, the material’s resistant behaviour is

characterized using mathematical models. Thus, this method translates the
analytical calculations of pavement response to performance. Nevertheless, the
design of pavement structure is not totally mechanistic, as dependence on observed
performance is necessary because theory alone has not proved sufficient to realistic
pavement design (Huang, 2004), and also laboratory testing is often required to
provide a relationship between loadings and failure which enhance the development
of a proper mathematical model. Hence, this brought about the concept of the
mechanistic-empirical (M-E) method of pavement design.
This chapter gives detailed background on pavement design through layered elastic
simulation and finite element simulation with more attention on critical factors, such
as the geometry selection, material characterization, and boundary and loading
condition. Further, in review of FEM types, the software ABAQUS® will be
introduced and thereafter the concept of failure analysis was discussed.
19

3.2. Approach of Mechanistic Empirical (M-E) Design
An M-E design approach uses empirical relationships between cumulative damage

and pavement distress to determine the adequacy of a pavement structure to carry
the expected traffic load (Nicholas and Lester, 2010). This approach combines
theory and physical testing with the observed performance in pavement design. This
design is also historically aimed at developing more accurate pavement models with
a lot of emphasis on developing the mechanistic parts of the model (Theyse and
Muthen, 2000). As a result, this gives M-E analysis advantage over empirical
methods. Some of such advantages are: accommodation of new materials and
changing load types, better utilization of available materials, capability of being used
for the design of both existing pavement rehabilitation, and new pavement design in
which empirical methods are limited.
Despite the advantages of M-E analysis, many developing countries still rely on
empirical methods, realizing that more sophisticated mechanistic design procedures
often require too many assumptions regarding material behaviour and too
complicated material testing techniques to be of direct practical use (Araya, 2011).
Nevertheless, the end results outweigh its complexity. Via M-E analysis, two major
approaches are employed to compute the stresses and strains in pavement
structures which include layered elastic theory and FEM, which are further
discussed. In addition, the effectiveness of any M-E method relies on the accuracy of
the predicted stresses and strains. Hence, this gives FEM an edge over the layered
elastic theory. Further, the success of this method hinges on some critical variables,
which are material properties, traffic, environmental conditions, and pavement
geometry. Nonetheless, for accuracy in pavement response prediction through M-E
methods, more focus should be placed on constituent materials’ behaviour and their
accurate characterization (Johnson, Sukumaran, Mehta, and Willis, 2007; Araya,
2011).
3.3. Layered Elastic Simulation
Layered elastic simulation is the most common and easily understood procedure of
the M-E design methods. In this simulation the pavement structure is divided into an
arbitrary number of horizontal layers with the thickness of each individual layer and
20

materials assumed to be homogeneous and linearly elastic (Wang, 2001). Firstly,

Burmister (1943) obtained the primary equations for a two-layer, three-layer and later
multilayer system. These equations were derived from the original elastic theory by
Boussinesq (1885). The original elastic theory was used to compute stress and
deflection in a half-space soil composed of homogeneous, isotropic and linearly
elastic material which is still widely used in soil mechanics and foundation design
(Wang, 2001; Huang, 2004). Yet, Burmister’s equations led to the significant
development in pavement analysis using mechanistic method; these equations are
included in the earliest software CHEV 1963. In addition, other software was
developed based on these equations but with different modifications. A number of
these are: BISAR 1973, developed to incorporate rate independence; VESYS 1974,
to incorporate the serviceability and reliability concept; ELSYM5 1986, to incorporate
multilayers; DAMA 1979, to incorporate nonlinear elastic granular materials; and also
KENLAYER for nonlinearity in granular materials, which is still commonly used
(Wang, 2001; Huang, 2004; NCHRP, 2004).
In South Africa, a great contribution has been made through the development of the
South African Mechanistic-Empirical Design Method (SAMDM), which is now known
as the South African Pavement Design Method (SAPDM) (Van Vuuren, Otte and
Paterson, 1974; Theyse, de Beer, Maina, and Kannemeyer, 1996; SANRAL, 2013b).
The SAMDM analysis for flexible pavement is based on linear elastic multilayer
theory and here the structural pavement layers are assumed to be isotropic (Steyn,
Maina and Repsold, 2013). Although SAMDM is sound in principle and has been
applied successfully to the design of pavement, this method is faced with the intense
challenge of its inability to cater for the cross-anisotropic behaviour of materials
(Steyn et al., 2013) and its over-sensitivity to the changes in the input variables,
which lead to inadmissible and counter-intuitive results and provide unrealistic
pavement design (Theyse et al., 2011). These in turn contribute to the increases in
its scrutiny and criticism in the recent past (Jooste, 2004). However, for SAMDM to
achieve more realistic values of predicted life for pavement section, it must include
cross-anisotropic analysis (Steyn et al., 2013); as a result, SAMDM is being revised
(SANRAL, 2013b).
21

Overall, considerable efforts have been reported regarding linear elastic simulation
of pavement structures, yet the assumptions, on which this approach works, make it
inappropriate for the real pavement properties and actual scenario on-site. Such
assumptions are (Tutumluer and Thompson, 1997; Wang, 2001; Huang, 2004);
 Each layer is homogeneous, isotropic and linearly elastic with a finite
thickness.
 Material is weightless.
 Circular uniform pressure is applied on the surface.
 Continuity and frictionless interface condition.
However, Mansurkhaki, Hesami, Khajehhassani, and Khordehbinan (2014) maintain

that there are no significant differences between the mean values of the parameters
obtained from layered elastic analysis and FEM, which is similar to the opinion of
Ameri, Salehabadi, Nejad, and Rostami, (2012), but on the other hand Ameri et al.
state that results from FEM are most appropriate compared with that of multi-layer
system. Also, Gupta and Kumar (2014) reported discrepancies in results from
KENLAYER; compared with those of FEM it shows that maximum vertical deflections
are lower in KENLAYER. In addition, various studies have shown that using linear
elastic simulation for pavement vertical stress and strain prediction results in error,
especially in low-thickness layers of asphalt pavement (Theyse et al., 1996; Abed
and Al-Azzawi, 2012; Al-Azzawi, 2012; Shafabakhsh et al., 2013a). Yet the simplicity
and speed of multi-layer analysis has been used as justification for relative results
obtained (Zaghloul, 1993). Overall, since stress, strain and relative conditions of
different layers in pavement structure are used in predicting pavement failures, the
need for considering materials’ behaviour in nonlinear form increased significantly.
This substantiates the fact that many researchers have found that the nonlinear
elastic behaviour of base and sub-grade materials is important in accurately
estimating stresses and strains in pavements (De Beer, Fisher, and Jooste, 1997;
Mun, 2003; Tiliouine and Sandjak, 2014). In view of the aforementioned limitations,
FEM is more preferred because it provides a more realistic analysis for predicting
pavement response (Zaghloul, 1993) and its capability to accommodate nonlinearity
of pavement materials (Tiliouine and Sandjak, 2014).
22

3.4. Finite Element Modelling (FEM)
FEM has wide application in lamella mechanics, hydrodynamics, soil mechanics, and
structural mechanics because of its great capability for finding approximate solution
to boundary value problems (Peng and He, 2009). In FEM, the whole problem is
divided into small and simpler parts through mesh generation which are called finite
elements and solved by calculus of variation in order to minimize associated error
function (Reddy, 2005; Dixit, 2007; Yagawa, 2011). Over the years, FEM has been
applied extensively in road engineering (Peng and He, 2009) and so far, it is the
most versatile of all analysis techniques, with capabilities for 2D and 3D geometric
modelling, able to analyse stable (static), time-dependent problems, non-linear
material characterization, large strains/deformations, dynamics analysis and other
sophisticated features (NCHRP, 2004). Furthermore, FEM can deal with complicated
loading (static, dynamic and spatially distributed form) conditions and more accurate
than the multilayer elastic method. The application of FEM to solve any problem
consists of three separate stages, as shown in Figure 3.2.
Pre- processing Processing Post Processing

(Modelling) (Evaluation & Simulation) (Visualization)
Figure 3.2. FEM application stages (Abaqus, 2013)
 Pre-processing (Modelling)
This is the first stage in any FEM analysis, and here can be referred to as the
input files stage, which is the most critical for the accurate prediction of the
result in terms of stress, strain and deflection. At this stage the following
selection/input are made: the geometry of pavement (in terms of dimensions),
material characterization, relationship between parts (assembling and
interactions), loading and boundary conditions, and analysis type. Further
discussion will be introduced on the input files in this thesis.
23

 Processing (Evaluation and Simulation)

In this stage, the job step is the main step and the input files are processed to
produce the results (output file). Basically, at this stage the analysis process is
only monitored in case an error is detected.
 Post-processing (Visualization)
This stage is a graphic rendering phase of the output file from the processing
stage. Results are well represented in the realistic format and the maximum
and critical area of interest can easily be accessed. Further, results in graph
format can be obtained as well.
3.4.1. Critical Factors in FEM Simulation of Pavement
Generally, creating a FE model for flexible pavement analysis involves the

consideration of all the steps in the pre-processing, with a critical look at some
factors. Any FEM generated must capture important features of the physical situation
without irrelevant details (Abaqus Inc., 2003). Overall, the success of any FEM
simulation depends greatly on these factors as it can lead to error in the design of
pavement if not properly put into consideration. Some of these factors are discussed
below, with their effects on pavement design.
3.4.1.1. Geometry
In pavement simulation via FEM, geometry in terms of dimension and

sharpness is of importance as it affects the overall analysis, time efficiency
and accuracy of the results. Regarding geometry, there are some major areas
of concern such as: dimension size, element type and mesh. Generally, the
larger the dimension size and complexity of model, the more analysis time is
required, as a result, Duncan, Monismith, and Wilson (1968) reported that a
reasonable pavement response can be obtained by using 50 times and 12
times the circular loading area in vertical and horizontal direction respectively.
FEM achieves its aim by dividing the problem domain into a number of
simpler subdomains - the finite elements. Various element types exist in the
use of FEM for pavement simulation, such as: linear or first-order, quadratic or
24

second-order, modified second-order, continuum and infinite element, to

mention but a few (Figure 3.3). These element types further utilize the
reduction integration techniques for reducing analysis running time (Abaqus,
2013; Zaghloul, 1993). Thus, careful consideration must be given to the
element aspect ratios; however, the use of infinite elements was discouraged
as it is not necessary in achieving accurate result (Sukumaran, 2004). In
addition to 3D FEM analysis, Zaghloul (1993), Rahman, Mahmud and Ahsan,
(2011) and Ibrahim, Gandhi and Zaman (2014) recommend the use of the
solid continuum element with reduction integration as it has the capability of
representing large-scale deformation and material nonlinearity.
Figure 3. 3 Some element families in Abaqus (Psarras et al. 2002)
On mesh, FEM employs mesh generation technique for dividing a complex

problem into small elements (Lo, 2002; Yagawa, 2011). It is known that the
finer the mesh, the more accurate the results obtained (Al-Jhayyish, 2014)
and the more analysis time required. On this note, Hjelmstad, Kim and Zuo
(1996) investigated issues on mesh construction aspects of modelling
pavement structures with 3D finite element analyses such as mesh
refinement, domain extent, computational memory, and element size
transitions, result shows that there is a great relationship between mesh
refinement and accuracy of results obtained and good aspect ratio resulted in
25

accurate results and reduce computation time. More so, on the aspect of
mesh construction and refinement reports as shown that the more fine the
meshing is, the more the accuracy in the result generated, consequently,
researchers concluded that, the mesh should be fine near loading area and
coarse at distances away from applied load for efficient model (Hjelmstad,
Kim and Zuo, 1996; Sukumaran, 2004; Peng and He, 2009; Tiliouine and
Sandjak, 2014).
3.4.1.2. Material Characterization
Proper material characterization is another major aspect of FEM-based

design of pavement for accurate response prediction as the reliability of
pavement design depends on it (Çöleri, 2007). However, an accurate material
characterization is the selection or formulation of proper constitutive equations
to represent the behaviour of the materials under loading (Kim, 2007).
Qualitative choice is needed in material characterization and it is important
that the model captures the major features of material behaviour while minor
features may be ignored in the model (Abaqus Inc., 2003). Furthermore,
resilient modulus (MR) is one of the important inputs alongside with Poisson’s
Ratio and it is a primary material property for characterizing all unbounded
layers and soils in any FEM model for flexible pavement design (Kim and
Siddiki, 2006; Harold and Von Quintus, 2007; Ji, Siddiki, Nantung, and Kim,
2014). MR values may be estimated directly from laboratory testing such as:
Triaxial, Oedometer and Shear test (level 1 input), indirectly through
correlation with other laboratory/field tests (CBR, Isotropic compression test,
Uniaxial strain test, Indirect tensile strength and UCS) (level 2 input) or back-
calculated from deflection measurements (level 3 input) (Mallela, Harold, Von
Quintus, Smith, and Consultants, 2004; Harold and Von Quintus, 2007;
Eluozo, 2013; Ji et al., 2014). Yet, correlation (level 2) is commonly used,
based on the fact that level 1 depends on difficult laboratory testing. However,
further discussions are presented in section 3.4.2 and 3.4.3 for level 2 and
level 1 respectively, based on the available results for this study.
Primarily, two material constitutive models are used in pavement structures,

which are Elasticity; Elastic and Viscoelastic, and Plasticity; Viscoelastic,
26

Drucker-Prager (D-P), Mohr-Coulomb (M-C), Modified Cam-Clay model, and

Modified Cap model to mention but a few (Abaqus Inc., 2003; Ti, Huat,
Noorzaei, Jaafar, and Sew, 2009; Desai, 2012). Ultimately, a realistic
constitutive model should better understand the mechanical behaviour of the
represented material and must be capable of representing material behaviour
in any relevant spatial situation (i.e. 1-dimensional, 2D and full 3D analysis). If
the models are not properly selected it may lead to under- or over-design of
the pavement structure.
3.4.1.3. Analysis Type, Boundaries and Loading Conditions
In FEM analysis, there are two major types of analysis procedure (also called
STEP in Abaqus®) depending on the modelling nature; these analyses are:
general and linear perturbation. In these procedures, there are forms such as;
Geostatic, Mass diffusion, Heat transfer, Static, dynamic analysis, etc.
However, these two analyses’ procedure can be used in pavement analysis.
The linear perturbation analysis procedure is usually employed for linear
analysis work, while the general analysis procedure goes with the non-linear
analysis work (Abaqus, 2013). As a result, the use of linear perturbation for
non-linear analysis will only consider the linear effects, thus resulting in error.
Furthermore, boundaries conditioning is of importance and has a significant

influence on the predicted response (Zaghloul, 1993). Boundary conditions
are the degrees of freedom at each node in an element. A model can either
be restrained in vertical, horizontal direction or set of nodes; on the other
hand, if boundary conditions are not properly selected it may lead to
generation of excess stresses and strains, both in vertical and horizontal
direction. In view of these, researchers have suggested the use of fixed
constraints at the bottom of the element (sub-grade) and roller constraints on
the vertical boundaries (Peng and He, 2009; Al-Khateeb et al., 2011;
Rahman, Mahud and Ahsan, 2011; Abed and Al-Azzawi, 2012; Sinha,
Chandra and Kumar, 2014). Furthermore, various forms of contact interaction
(mechanical and thermal) occur between the pavement layers. This
interaction usage was encouraged (Peng and He, 2009) as it improves the
results. Additionally, most researchers (Peng and He, 2009; Shafabakhsh et
27

al., 2013a; Shafabakhsh et al., 2013b) prefer the use of perfect bond between
the layer so allow uninterrupted distribution of stresses, strains and
deflections through the layers, yet this is not the real scenario in reality as full
bounding is not always achieved (Sutanto, 2009).
Regarding loading, tyre load representation is another critical factor to be

considered in pavement simulation. Representing the tyre contact wrongly will
affect the overall results. Over the years, various methods have been
suggested in representing the loading in pavement design. Initially, a circular
representation is used (Al-Khateeb et al., 2011; Sinha et al., 2014; Tiliouine
and Sandjak, 2014), but at present, various representations have also been
made in different studies (Peng and He, 2009; Rahman et al., 2011) with
positive results. Furthermore, in reality, pavement is subjected to a moving
loading, yet several researchers (Rahman et al., 2011; Shafabakhsh,
Talebsafa, Motamedi, and Badroodi, 2013b; Sinha et al., 2014) have used
static load for analysis rather than dynamic load because of the theoretical
and practical difficulties involved in the analysis when using a dynamic load
(Kim, 2002).
In a nutshell, with the great aptitude of FEM to analyse stable problems, time-
dependent problems and those problems with non-linear properties of materials
(Salehabadi, 2012), a careful balance is required in all the above-mentioned factors
to meet the demand for solution and memory without sacrificing accuracy
(Sukumaran, 2004). FEM has been successfully used in the analysis of the major
forms of failure in pavement structure such as rutting and fatigue cracking at different
layers (Walubita and van de Ven, 2000; Al-Khateeb et al., 2011; Abed and Al-
Azzawi, 2012), and also used to determine the accurate positioning of geogrid
materials (Al-Azzawi, 2012), thickness of each layer (Shafabakhsh et al., 2013a;
Sinha et al., 2014) and the interaction between pavement and its instrumentation
(Zafar, Nassar and Elbella, 2005; Yin, 2013).
3.4.2. Correlation Equations in FEM Simulation
MR is the measure of material stiffness (i.e. stress divided by strain for rapidly applied
loads). This can be mathematically expressed as the ratio of applied deviator stress
28

to recoverable strain (George, 2004; Pavement Interactive, 2007; Ji et al., 20014).

Determining MR is of vital importance for any mechanistically based design/analysis
procedure for pavements because it represents the structural strength of pavement
layer on through which the thickness design is based (Eluozo, 2013; Ji et al., 2014).
However, AASHTO recommends that MR be obtained from repeated Triaxial testing,
but due to the complexity of the test and time required, its results are not readily
available. In view of this, researchers improvise through the use of correlation
equations for readily available test results, for example CBR (Heukelom and Klomp,
1962; Sas, Głuchowski, and Szymański, 2012) and UCS (Little, Snead, Godiwalla,
Oshiro, and Tang, 2002; Kim and Siddiki, 2006; Rao, Titus-Glover, Bhattacharya,
and Darter, 2012; Al-Jhayyish, 2014). This material input method is referred to as
level 2 inputs. Correlation equations help to convert readily available results to
corresponding MR values. However, results from UCS testing is more common and
popularly used as its data are readily available (Rao et al., 2012) and a better
property to predict design MR (George, 2004). As a result, it is a necessity to
evaluate design MR of stabilized base layer based on the available UCS data. Table
3.1 suggests few of several equations to estimate MR with results from UCS test.
Of all the promising equations suggested in the above table, according to Little et al.
(2002) and Al-Jhayyish (2014), the correlation equations proposed by Barenberg
(1977) for cement-stabilized soils are in good agreement with the laboratory results.
However, a cement-fly ash-base layer is considered in the research, yet there is no
direct correlation equation for it. Considering and validating the two equations by
Barenberg, it was found the equation for ‘cement-stabilized coarse-grained sandy
soils’ gives a closer result when compared with the recommended M R for cemented
materials used in SAMDM 1996 (SANRAL, 2013b). Since the material used in the
previous research is a G5 material (usually gravel with coarse-grained properties)
which is stabilized to C3 and C4 by using cement-fly ash as stabilizer (Heyns and
Mostafa Hassan, 2013). Thus, the correlation equation by Barenberg (1977) for
‘cement-stabilized coarse-grained sandy soils’ is suitable and will be used to
estimate the design MR which will serve as the input for material property in the
software. Largely, careful consideration should be given to the unit of parameters in
the equation and their conversion to avoid error.
29

Table 3.1. Summary of correlations between the unconfined compressive

strength and modulus
Correlation Source of the Application Area

Correlation
American Coal Ash Lime-cement-fly ash
MR (ksi) = 500 + UCS (psi)
Pavement Manual (1990) stabilized soils
Cement-stabilized
Barenberg (1977)
MR (psi) = 1200 UCS (psi) coarse-grained sandy
soils
MR (psi) = 440 UCS (psi) + Cement-stabilized
Barenberg (1977)
0.28 UCS2 (psi) fine-grained soils
MR (ksi) = 0.124 UCS (psi)
Thompson (1966) Lime-stabilized soils
+ 9.98
McClelland Engineers Lime-cement-fly ash
MR (psi) = 0.25 UCS2 (psi)
(unpublished) mixtures
MR (MPa) = 2240 UCS0.88 Australian Road Research Cemented natural
(MPa) + 110 Laboratory (1998) gravel
3.4.3. FEM Material Characterization via Direct Testing Results
Although, the use of Triaxial, Oedometer and Shear test results as material
characterization are level 1 input, considered more accurate (Mallela et al., 2004),
but as a result of these tests’ unavailability, it is considered second in the research
studies. The use of direct testing results (level 1 inputs) in material characterization
gives a more realistic constitutive model, which consequently gives a better
understanding of the mechanical behaviour of the material (in terms of material non-
linearity) (Abaqus Inc., 2003; Mallela et al., 2004). While the level 2 (correlation input
methods) only gives room for obtaining limited parameters (such as MR, Poisson
ratio) which therefore, results in the use of linear material characterization and are
basically considered for preliminary study. Using any of the direct test results
requires at least one to two laboratory tests for calibration in the FE model.
Additionally, these test results are used in obtaining the MR and further inputted into
30

various constitutive models in the FE model for the characterization of the material in
question.
Over the years, various models have been developed for obtaining MR through
Triaxial laboratory results. Table 3.2 suggests a few of the several models available.
Overall, amongst the listed models in Table 3.2, the LTTP model, – a modification of
the Universal model – is adopted in the NCHRP 1-37A Design Guide (United States
Department of Transportation – Federal Highway Administration (USDT-FHA) 2014),
thus will be considered in this study based on its general acceptance. Further study
can be found on these various models for MR calculation in a report by George
(2004). However, these models are affected by important parameters such as
Atterberg limits, grain size distribution, moisture content and density, which are used
in the calculation of coefficients (k) to form regression analysis (George, 2004;
Dione, Fall, Berthaud, and Makhaly, 2013; Ji et al., 2014). Also, the result in terms
of MR obtained is inputted in constitutive material models in the FE Model.
As mentioned earlier (section 3.4.1.2), the two constitutive material models are
Elasticity and Plasticity, but the Plasticity model has got various models which can
be used as a close representative of non-linearity of geotechnical materials such as
gravel and soil (Abaqus Inc., 2003; Shafabakhsh et al., 2013a). However, out of the
various Plasticity models (Viscoelastic, D-P, M-C, Modified Cam-Clay, Modified Cap
model, etc.), the D-P and M-C Plasticity model had been considered to be a better
representation for base, sub-base and sub-grade layer materials in pavement. Yet,
more consideration has been given to D-P because of its capability to model material
behaviour in high stresses, volumetric shear and strain (Peng and He, 2009; Ti et al.,
2009; Al-Azzawi, 2012; Shafabakhsh et al., 2013a; Maharaj and Gill, 2014) and
simplicity (Al-Khateeb et al., 2012), therefore, it is considered in this study.
D-P model is a Plasticity model and a modified version of Mises criteria which is
approximate to M-C criterion for simulating frictional materials (Abaqus Inc., 2003;
Peng and He, 2009). In this model, there is a period of purely elastic response, after
which some material deformation is not recoverable (plastic), thus it should be used
along with Elasticity models, which makes this model elasto-plastic in nature
(Abaqus Inc., 2003; Abaqus, 2013; Shafabakhsh et al., 2013a). The D-P model has
a choice of three different yield criteria, such as: linear, hyperbolic and a general
31

exponent form (Abaqus Inc., 2003; Abaqus, 2013). Nevertheless, the most common
of the three yield criteria is the exponent form, which provides the most flexibility in
matching Triaxial test data, such that Abaqus® determines the material parameters
required for this model directly from the Triaxial test data, thus minimizing relative
error (Abaqus, 2013). However, D-P is not non-linear, yet according to Rodriguez-
Roa (2003), there is no much difference between non-linear elastic and elasto-plastic
behaviour, thus, the elasto-plastic model such as D-P can be used as a close
representation of non-linearity in pavement materials.
Furthermore, the yield criteria for the general exponent form provide the most
general yield criteria available which is expressed in equation 3.1. Overall, other
parameters used in the D-P model – such as: Dilation angle (ψ), Flow-stress ratio (K)
– can be determined by the M-C model.
𝐹 = 𝑎𝑞 𝑏 − 𝑝 − 𝑝𝑡 = 0…………………………Equation 3.1
Where: 𝐹 = Yield surface

𝑎𝑏 = Constant with respect to stress
𝑝 = Mean normal stress
𝑝𝑡 = Hardening parameter that represents the hydrostatic tension
strength of material
𝑞 = Mises equivalent stress
Essentially, various research studies have been done on the layers in flexible
pavement via FEM. Yet granular materials do not feature strongly, as more focus is
given to designing the asphalt layer and sub-grade condition (Araya, 2011).
Similarly, only limited work has been done on stabilized base and sub-base layers
(Peng and He, 2009). Hence, stabilized granular material as a base layer will be
considered. As earlier mentioned in introduction, FEM can be applied in two ways:
2D and 3D. However, the use of 3D appears to be the best approach (Wang, 2001;
Sukumaran, 2004; Rahman et al., 2011; Shafabakhsh et al., 2013a). Nevertheless,
there are various sources of error in pavement performance predictions and most
are more difficult to control than the response model (NCHRP 2004). Therefore, a
reality check through validation of results with field testing or available results is of
importance.
32

Table 3. 2. Summary of resilient modulus constitutive models (George, 2004).
S/N Model Source of Model Comments

This model does not incorporate
𝜃 𝑘2 Seed et al. (1967) the realistic responses of confining
1 𝑀𝑅 = 𝑘1 ( )
𝑃𝑎 (k–θ model) and deviator stress in MR
properties.
𝜎3 𝑘2 It does not consider the effect of
2 𝑀𝑅 = 𝑘1 ( ) Dunlap (1963)
𝑃𝑎 deviator stress on the MR.
Adequate for cohesive soils but
𝜎𝑑 𝑘2 Moossazadeh and Witczah (1981) does not consider the effect of
3 𝑀𝑅 = 𝑘1 ( )
𝑃𝑎 (k-σd model) confining stress on MR for clay
soils.
It describe the nonlinear behaviour
𝜃 𝑘2 𝜎𝑑 𝑘3 in Triaxial test by considering the
4 𝑀𝑅 = 𝑘1 𝑃𝑎 ( ) ( ) May and Witczah (1981) effect of shear stress, confining
𝑃𝑎 𝑃𝑎
stress and deviator stress in terms
of bulk and deviator stress.
Universal cause of the ability to
𝜃 𝑘2 𝜏𝑜𝑐𝑡 𝑘3 Uzan (1992) conceptually represent all types of
5 𝑀𝑅 = 𝑘1 𝑃𝑎 ( ) ( )
𝑃𝑎 𝑃𝑎 (Universal model) soil from pure cohesive to non-
cohesive.
𝜃 𝑘2 𝜏𝑜𝑐𝑡 𝑘3 Yau and Quintus (2002) It combines both the stiffening
6 𝑀𝑅 = 𝑘1 𝑃𝑎 ( ) [( ) +1] effect of the confinement or bulk
𝑃𝑎 𝑃𝑎 (LTTP model)
stress.
Where; MR = Resilient modulus; θ = Bulk stress (σ1 +σ2 +σ3); Pa = Atmospheric Pressure; σd = Deviator stress; σ3 = Confining stress;
2
𝜏𝑜𝑐𝑡 = Octahedral stress (√3 σd ); k i = Regression coefficient
33

3.5. 2D versus 3D Analysis
The 2D FEM analysis generally assumes plan strain or axis-symmetric condition in

the development of the model (pavement structure) utilizing the horizontal and
vertical (X and Y) dimensions. Comparing this against the layered elastic method, it
is more practical as it considers material anisotropy, non-linearity and variety of
boundary conditions, yet the method is challenged by some limitations as it cannot
accurately capture non-uniform tyre contact and multiple wheel loads (Stoffels,
Solaimanian, Morian, and Soltani, 2006; Rahman et al., 2010). In contrast, 3D FEM
analysis is more of the real-life representation of the pavement structure utilizing the
horizontal, vertical and depth (X, Y and Z) dimensions – not only that, it has the
ability of accounting for multiple wheel loads as well as moving wheels (Wang,
2001).
Effectively, through the use of 2D FEM analysis programs such as DSC2D, JULEA,
MICHPAVE, ILLIPAVE and ABAQUS® (NCHRP 2004), investigation of flexible
pavements’ responses has been done and also utilized for single-wheel load
analysis (Harichandran, Yeh and Baladi, 1989; Sukumaran, 2004). Al-Khateeb et al.
(2011) predicted rutting in flexible pavement using 2D FEM: results show that the
use of linear-elastic models to predict stresses and strains in pavement structures
can lead to significant errors and rut depth increases with decreasing sub-grade
strength. Further, Tiliouine and Sandjak (2014) used 2D axis-symmetric in simulation
of granular materials behaviour on the basis that it can adequately represent the
granular material non-linearity under various stress conditions. On a comparison
note, Cho, McCullough, and Weissmann, (1996) present 2D axis-symmetric as a
good alternative over 2D plan strain and 3D, when traffic load is away from the
edges, considering the failure of 2D plan strain in calculating the appropriate stress
distribution and high computation resources of 3D, but it must be noted that axis-
symmetric cannot model moving traffic load only static loading. Likewise, Hua
(2000), using both 2D and 3D model to predict surface profile under 5000 wheel
passes, shows that there is no significant difference (< 2 percent) between the two
models.
Although, the 2D FEM analysis has been adequate for the study of nonlinear
analysis, the 3D FEM is believed to be used for more accurate pavement responses
34

(Kim, 2007; Al-Khateeb et al., 2011). Considering this fact, 3D via Abaqus® has
been used in the study of flexible pavement under spatially varying tyre/contact
pressure by which 2D is limited (Wang, 2001; Rahman et al., 2011). The 3D-based
EverStressFE1.0 Software for analysis of flexible pavement was developed by
Davids (2009), addressing the shortcomings of traditional analysis software
packages such as EverStress, mePADS, BISAR, or KENLAYER. Further, through
Abaqus® 3D, Zaman, Pirabarooban, and Tarefder (2003) developed a FEM to
simulate the laboratory testing of asphalt mixes in asphalt pavement analysis for
rutting; results show that the speed of moving load has a significant effect on
predicted rutting. Shafabakhsh et al. (2013a) via Abaqus® presented the influence of
asphalt thickness on settlement of flexible pavement: an increase in the pavement
thickness and a decrease displacement value. Also, Shafabakhsh et al., (2013b)
reported the consistency of results for a 3D model with moving load impact on the
tensile strain at the bottom of the asphalt layer when compared with that of field
measured pavement responds, and similarly, Abaza (2007) discovered that cyclic
and non-linear materials give results close to field measurement results. In 2011,
Rahman et al., using Abaqus® 3D FEM, study a preliminary research of traffic-
related factors in the design of flexible pavement under specific material properties,
model geometries, etc.
Conclusively on 3D, Peng and He (2009) simulated the design and construction
process of flexible pavement with cement-stabilized base layer using ADINA FEM
software. However, the construction process has little effect on the outcome of
results, yet the use of 3D is encouraged, based on its ability for layer-contact
modelling. As mentioned earlier, the use of 3D is a drawback because it’s difficult,
high demand in data preparation and computation time (Wang, 2001; Zafar et al.,
2005; Al-Khateeb et al., 2011). In view of this, through Abaqus®, Sukumaran (2004)
tried to discover a less computationally intensive 3D model that would still maintain
accuracy; as a result, the use of 3D symmetric model was presented as a suggestion
on mesh construction, mesh refinement and element aspect ratio. Besides, the
newer versions of the 3D software have been improved by making it user-friendly
and interactive and overall increased speed in the analysis time.
35

3.5.1. Motivation for 3D FEM
Among the three models of representation in FEM analysis (2D plan strain, 2D axis-
symmetric and 3D), the 3D FEM model is used to achieve the aim of this research
work. According to the above review, the following are the tangible reasons why 3D
FEM is used:
 Its ability to capture the effect of non-linear materials or the effect of
combination of loads (Abaza, 2007; Shafabakhsh et al., 2013b).
 Its capability to account for multiple wheel load as well as moving wheel load
(Wang, 2001; Zaman et al., 2003).
3.5.2. Comparative Study on FEM Software
Quite a number of FEM software programs are available and many have been found
useful for pavement design purposes. Basically, in pavement design, there are two
major categories of FEM software, the general purpose and the specific purpose
software (NCHRP 2004). The general purposes are those with a wide range of
applications aside from pavement design, in areas such as medicine, lamella
mechanics, hydrodynamics, soil mechanics, structural mechanics; examples are
Abaqus®, ADINA, ANSYS and DYNA3D, while the specific purposes are developed
particularly for analysis of pavement design. Examples are EverStressFE, ILLI-
SLAB, ILLI-PAVE and MICH-PAVE. Various successes have been recorded through
the use of the aforementioned software, yet the general purpose is more powerful
and capable of conducting 3D non-linear dynamic analysis perfectly. Additionally, it
provides optimum flexibility to manipulate a variety of FE models with sophisticated
geometry and boundary conditions (Wu, Chan and Young, 2011). Further, NCHRP
(2004) did a comparative study on the software regarding issues of efficiency issues
and operational issues; in that study, Abaqus® is considered as a potential
candidate based on its technical capabilities and its extensive past usage in research
oriented pavement analysis, but it was disregarded because of its high licensing
costs and restrictive licensing terms. Yet, Abaqus® has wide applications in the
aspect of pavement design; this software is introduced and used in this study.
36

3.5.3. Abaqus® as FEM Software
Abaqus®, a general purpose and commercial FEM modelling software, has widely
been applied for pavement analysis. As mentioned above, in Finite element
simulation (section 3.4), it contains three major process stages: pre-processing,
processing and post-processing. In 1990, Chen Marshek, and Saraf,
comprehensively studied various pavement analysis programs and showed that the
results from the Abaqus® program were comparable to those from other programs.
Also, from the above review (section 3.4.3 and 3.4.5) on FEM it can be seen that
Abaqus® has been preferred above others.
Further, Abaqus®/CAE is a complete Abaqus® environment that provides a simple,

consistent interface for creating, submitting, monitoring and evaluating results from
Abaqus®/Standard, Abaqus®/Explicit simulation and others (Abaqus, 2013). It is
divided into modules and in each module logical representation of the modelling
process can be defined such as defining the geometry, generating a mesh and
assigning material properties, etc. The model built in Abaqus®/CAE generates an
input file to submit to the Abaqus®/Standard or Abaqus®/Explicit analysis product.
The analysis product performs the analysis, sends information to Abaqus®/CAE to
allow the progress of the job to be monitored, and generates an output database.
Finally, the visualization module of Abaqus®/CAE is used to read the output
database and view the results of the analysis (Abaqus, 2013).
Furthermore, Abaqus® is a modular code consisting of a library of over 300 different

element types and a comprehensive material model library with materials ranging
from linear to nonlinear and isotropic to anisotropic behaviour, which are useful for
pavement analysis and also allows for the introduction of new materials through its
user-defined sub-route (Rahman et al., 2011; Abaqus, 2013). Also, it is capable of
analysing a variety of problems (linear, nonlinear, static, dynamic, structural and
thermal) (Britto, 2010). Overall, ABAQUS® usage is enhanced by its friendly and
interactive user-guide which is available in PDF and HTML version.
Motivation for Abaqus®

ABAQUS® will be used in this research because of its capabilities in solving
pavement engineering problems:
37

 Material modelling as linear and nonlinear elastic, viscoelastic, and elasto-

plastic in 2D and 3D analysis
 Static, harmonic dynamic and transient dynamic loading simulation
 Contact/Interface modelling with friction
 Cracking propagation modelling (Abaqus, 2013)
 Analysis which involve temperature gradient (Sukumaran et al., 2004; Britto,
2010; Shafabakhsh et al., 2013a).
3.6. Failure Criteria in Numerical Simulation
Failure criteria based on fracture mechanics have been developed for pavement
layers, with the aim of enhancing design to provide sufficient resistance to pavement
failure (Mamlouk and Mobasher, 2004). This analysis requires models which relate
the output from FEM or elastic-layered analysis (stress, strain, or deflection) to
pavement behaviour in terms of performance, cracking, rutting, roughness and life
span (Ekwulo and Eme, 2009). It is one of the empirical portions of M-E design and
also known as damage models (SANRAL, 2013b). Equations used for these models
are derived from observation and performance of pavement with relation to observed
failure and initial strain under various loads, thereby computing the number of
loading cycles to failure (Pavement Interaction, 2008). Various types of failure criteria
exist depending on the type of pavement layer in question, such as: Asphalt surface
– Fatigue cracking; Unbound granular base and sub-base layer − Permanent
deformation; Cemented base and sub-base layers − Crushing failure, Effective
fatigue and Permanent deformation; Sub-grade – Permanent deformation or rutting.
Nonetheless, two are widely recognized: fatigue cracking in asphalt and deformation
in the sub-grade (Pavement Interaction, 2008; Ekwulo and Eme, 2009; SANRAL,
2013b).
In South Africa, failure analysis has been checked through damage models
suggested by SAMDM (1996), but according to SANRAL (2013b), (1996) SAMDM
fatigue transfer functions for asphalt are not that reliable and permanent deformation
transfer functions for granular materials are on the conservative side. As a result, the
SAMDM damage model is out-dated (SANRAL, 2013b), therefore, it is appropriate to
consider other damage models, such as: Shell (Huang, 2004), Transport and Road
38

Research Laboratory, Asphalt Institute (Asphalt Institute, 1982), etc. However, the
fatigue criterion in the M-E approach is centred on limiting the horizontal tensile
strain at the bottom of the asphalt layer due to repetitive loads on the pavement
surface. If this strain is excessive, it will result in cracking (fatigue) of the layer
(Ekwulo and Eme, 2009), and the relationship is given in the equation 3.2 by Asphalt
Institute (Asphalt Institute, 1982), which is commonly accepted. Permanent
deformation can initiate in any layer of the structure, making it more difficult to predict
than fatigue cracking (Pavement Interaction, 2008). However, critical rutting can be
attributed mostly to a weak pavement layer (sub-grade). This is typically expressed
in terms of the vertical compressive strain at the top of the sub-grade layer and is
given by Asphalt Institute by equation 3.3.
Nf = 0.0796(εt )−3.291 (E)−0.854 ……………………… Equation 3.2

Where: Nf = Number of repetitions for fatigue cracking
εt = Tensile strain at the bottom of the asphalt surface in microstrain
E = resilient modulus of asphalt in psi
Nr = 1.365x10−9 (Ec )−4.477 ……..…………………….. Equation 3.3
Where: Nr = Number of repetitions for sub-grade rutting failure

Ec = Compressive strain on top of the sub-grade.
Overall, the failure analysis models are used to define the point at which failure
occurs in a pavement by determining the incremental damage.
3.7. Summary
Chapter Three of this dissertation presented reviews on numerical simulation of

flexible pavement. Away from the empirical method of design, numerical simulation
uses the level of stress, strain and deflection in the design of pavement structure.
However, there are two major approaches in numerical design of pavement, but the
effectiveness of FEM in predicting the stress and strain gives it an edge over the
layered elastic method. 3D FEM has its own challenges of input parameters and
computational time, but reviews have shown that it is the more preferred of the two
FEMs because of its ability to design more complex problems relating to the actual
conditions of pavement structures.
39

Furthermore, various material characterization inputs (level1 and 2 inputs) were

explored and thereafter, failure criteria in numerical simulation were considered and
the Asphalt Institute model for both fatigue and rutting in terms of number of
repetitions before failure would be used in this study. Abaqus® provides user
flexibility and as a result; it has wide attention in pavement design. Therefore, based
on the useful information from this chapter, the following study methods will be
presented in the next chapter:
 Development of the geometry model for representation/description of a fly-

ash- stabilized base layer in a typical South African road through Abaqus®;
 Appropriate selection of a material model for the accurate characterization of
the stabilized base layer; and
 Appropriate selection of boundary conditions and loading analysis that
suitably represent the already available laboratory results.
40

Chapter 4 Design and Simulation
CHAPTER 4: DESIGN AND SIMULATION
4.1. Introduction
This chapter establishes detail on the steps for development of FEM and its analysis.
Likewise, it presents a description of two FE models in line with the set objectives
and a comparative analysis of the laboratory and FEM results. The first model was
used to validate the efficiency of using 3D FEM over axisymmetric in the design of
pavement structure; it was also used to examine the structural response of a
cement-fly ash-stabilized base layer in terms of the stresses and strains on the top of
the sub-grade, while the second model was developed to evaluate the protective
importance of surface layer over stabilized base layer by estimating the tensile strain
at the bottom of the surface layer and the surface of the sub-grade. Thirdly, a
comparative analysis of the non-linear and linear material characterization will be
undertaken. Lastly, the results obtained from FEM analysis will be compared with the
already available laboratory empirical results for validation. Overall, all the models
were developed using Abaqus® 6.13 software.
4.2. Development of Flexible Pavement Model
Abaqus® analysis modules starts with a batch program, with the objective of
assembling an input file which describes a problem so that Abaqus® can provide an
analysis (Liang, 2000). The input file for Abaqus® contains model data and history
data. Model data defines a FEM in terms of geometry, element properties, material
definitions and any data that specifies the model itself (Liang, 2000; Britto, 2010;
Abaqus, 2013). Further, the history data define what happens to the model and the
sequence of loading for which the model’s response is sought, including the
procedure type, control parameters for time integration or non-linear solution
procedures, loading and output request (Liang, 2000; Britto, 2010; Abaqus, 2013).
Data can be defined by the user with relevant option blocks provided in the modules
(Abaqus, 2013).
Applying the file, Abaqus® automatically controls the time step and increments of the
load and records the message and data in all the analysis procedures according to
41

data defined in the file; afterwards the results are obtained by using Abaqus®/post.
Overall, there are two basic methods of inputting data into Abaqus® software
(Abaqus, 2013) which are
 Input file usage
 Abaqus®/CAE usage
However, of these two input methods, Abaqus®/CAE usage is preferred because of

simplicity in terms of not including code writing, so it is used in this study. The
general steps for the development of flexible pavement through Abaqus®/CAE
usage input method are summarized in Figure 4.1 and briefly explained thereafter.
4.3. Abaqus®/CAE Usage for Flexible Pavement Model
In Abaqus®/CAE usage, there is no particular order for modelling of a member in

Abaqus®, but in any FEM analysis the input (in terms of geometry) are basically
considered first. Since a conventional flexible pavement which contains surface-,
base- and sub-grade layer is used in this study; the use of this 3-layer pavement
system is to properly understand the behaviour of the stabilized base layer without
the interference of other layers such as a sub-base. The part module is used to input
the pavement layers’ geometries by creating a 2D sketch which is extruded in 3D,
with other features such as partitioning, generating meshes for parts, creating sets
and assigning of names to all members. On the material properties module, the
characterization of each part, such as surface, base and sub-grade layer, are
inputted; thereafter, via the assembling and steps module, the parts are put together
to form a composite conventional pavement structure. The step module is used to
capture changes in the loading and boundary conditions with respect to the parts’
interaction with each other.
Various forms of contact interaction (mechanical and thermal) occur in pavement,

thus the interactions and load module are used to define the interface in line with the
already created steps in the step module; the load defines the transferred load
(traffic load) and boundary condition in an attempt to represent the on-site condition
of the pavement. Lastly, job module in Abaqus is used to submit, analyse and
42

monitor the created pavement model and afterwards the results are viewed in the
visualization module. Conclusively, in using Abaqus® software, careful attention
should be given to all the modules to avoid warnings and prevent errors.
4.4. 3D versus Axisymmetric
4.4.1. Description
This FEM is a scenario of unpaved pavement structures which are developed for a
two-layered system (base and sub-grade layer) with the aim of achieving the set
objectives which are;
1. To evaluate the efficiency of using 3D FE model for design of flexible
pavement.
2. To determine the structural response of stabilized base layers in flexible
pavement system due to traffic loads using 3D FE model.
The scenario consists of 16, 18, 20 and 22 percent fly ash with 1 percent cement
stabilized base layer over a sub-grade and would be modelled in axisymmetric and
3D FEM. In this model, the thickness of the sub-grade layer is kept constant at a
specific depth (2000 mm), while the base layer thickness changes over a range (100
mm – 500 mm). Comparative analyses of the results obtained from axisymmetric
and 3D FEM would be undertaken for the structural response of the base and sub-
grade layer in terms of:
1. Compressive vertical stresses/strains within the base layer

2. Compressive vertical stresses/strains at the top of the sub-grade layer.
4.4.2. Model Geometry and Material Properties
The axisymmetric model is basically 3000 mm radius with a total depth that varies
based on the thickness of the base layer, which changes over a range of 100 mm –
500 mm (Figure 4.2). This geometry, particularly the radius (breadth), is similar to
that used by Al-Jhayyish (2014). However, sub-grade depth is infinite, but for the
purpose of boundary conditioning it is assumed to 2000 mm (Rahman, 2011), since
there is no deformation after a certain depth. On material properties, level 2 input
43

methods (section 3.4.1.2) would be used and material properties of the stabilized
base layers were obtained from laboratory testing (UCS) conducted by Heyns and
Mostafa Hassan (2013).
The UCS results used were those of 16, 18, 20, 22 percent fly ash with 1 percent
cement, where AFRISAM was the cement and Pozzfill as the fly ash (Table 4. 1)
(Heyns and Mostafa Hassan, 2013). These material properties are obtained using
correlation formula by Barenberg (1999) (section 3.4.2) and other material properties
are selected from SANRAL (2013b), as presented in Table 4. 1 and 4.2. All material
properties are assumed to be linearly elastic for simplicity as non-linear properties
require many input parameters which are not readily available (Al-Jhayyish, 2014).
The 3D FE model utilizes 3000 mm length by 3000 mm breadth with the total depth
varying based on the thickness of the base layer as in the axisymmetric model
(Figure 4.3). This geometry is similar to that used by Ahmed (2006), with the aim of
avoiding edge error when loaded. Materials properties are all assumed to be linearly
elastic, thus a static linear perturbation analysis procedure type will be used. These
material properties are presented in Table 4. 1 and Table 4.2; these data were
utilized to define the material properties of the model layers in ABAQUS®.
44

Figure 4.1. General steps for the development of flexible pavement

(Abaqus®/CAE usage)
45

Figure 4.2. Axisymmetric model geometry of the stabilized base and sub-grade
layer with meshing, load and boundary conditions
4.4.3. Model Mesh and Element types
In the axisymmetric model, 4-node bilinear axisymmetric quadrilateral elements

(CAX4R) with reduction integration were used. The stabilized base and sub-grade
layer were seeded at 0.025 m at the loading area because displacement gradients
are higher in this region, while other areas were seeded at 0.1 m; as a result,
meshes are fine in/near loading area and coarse at distances away from applied
load for efficient modelling, as suggested by Peng and He (2009) and Tiliouine and
Sandjak (2014) (Figure 4.2). The total number of elements range from 888 – 1480.
For the 3D FEM model, 8-node solid continuum elements (C3D8R) with reduction
integration were used; similarly, the stabilized base and surface layer were seeded
as in axisymmetric model, overall the total number of elements range from 19443 –
36075 (Figure 4.3).
46

Figure 4.3. 3D model geometry of the stabilized base and sub-grade layer with
meshing, load and boundary conditions
4.4.4. Boundary Conditions and Loading
The pavement layers were assumed to bond together perfectly; although in reality
full bond is not always achieved (Sutanto, 2009) but proper distribution of stresses,
strains and deflections between the layers, it is assumed to be perfectly bonded.
Also, the models are fixed at the bottom of the sub-grade and roller constraints on
the vertical boundaries (i.e. the model can move only in y-direction) (Figure 4.2 and
4.3). On loading (section 3.4.1.3), a static standard equivalent single-axle load with
dual tyres was used in these models, since Wu et al. (2011) specified that the
maximum stress at a specific point in the pavement occurs when the wheel load is
directly above it, while the stress can be assumed at zero when the load is quite far
from that point. In an axisymmetric model, the breadth of tyre load (224 mm)
proposed by Huang (2004) (Figure 4.2) was used while, in the 3D model contact
area of 72557 mm2 (Figure 4.3) with a rectangular area of contact was placed above
the stabilized layer (Huang, 2004; Al-Jhayyish, 2014). These loads were standard
equivalent single-axle load (80 kN) with dual tyres and applied uniformly with a
47

pressure of 0.65 MPa in accordance with South African standard (TRH 4, 1996;
Theyse et al., 2011). Conclusively, this analysis will be run as a static linear
perturbation analysis procedure type.
Table 4. 1. Material properties of the stabilized base layer (obtained from

Heyns and Mostafa Hassan, 2013)
Stabilized Base Material code USC Modulus of Poisson’s

(percent Flyash+1 (Colto, 2008) (Kpa) Elasticity Ratio
percent Cement) (MPa)
16 C3 3310 3972 0.35
18 C3 2133 2560 0.35
20 C3 3830 4596 0.35
22 C3 2298 2758 0.35
Table 4.2. Material properties of other pavement layers

Layer Material code Modulus of Poisson’s Ratio
(Colto, 2008) Elasticity (MPa)
Surface AG 3000 0.44
Granular Base G5 200 0.35
Sub-grade G10 45 0.35
4.5. Paved Stabilized Base Layer
4.5.1. Description
A scenario of paved flexible pavement is developed for a three-layered system of the

pavement structure, which are: asphalt surface, 18 percent fly ash with 1 percent
cement stabilized base, and sub-grade layer. The 3D FEM was used in the
development of these models. The thicknesses of the stabilized base and sub-grade
layer were kept constant at a specific depth in accordance with results from 3D vs
axisymmetric case (300 mm and 2000 mm respectively). Here, the protective
importance of surface layer over stabilized base layer will be evaluated in terms of:
1. Surface layer deflection;

2. Tensile strain at the bottom of the surface layer;
48

3. Compressive vertical stresses/strains within the base layer; and

4. Compressive vertical stresses/strains at the top of the sub-grade layer.
4.5.2. Model Geometry and Material Properties
A 3D model with 3000 mm length by 3000 mm breadth and the total depth varying
based on the thickness of the surface layer over a range of 25 mm–100 mm was
developed. This geometry is also similar to that used by Ahmed (2006), with the aim
of avoiding edge error when loaded. The material properties and analysis procedure
type are similar to the above, with properties for 18 percent fly ash with 1 percent
cement stabilized as the material for the base layer (Table 4. 1).
4.5.3. Model Mesh and Element type
In order to keep the size of the problem manageable in terms of analysis time and
storage capacity (Saad, Mitri and Poorooshasb, 2006), the meshing is fine in/near
loading area and coarse at distances away from applied load; this is similar to those
in the 3D vs axisymmetric case. Additionally, 8-node solid continuum elements
(C3D8R) with reduction integration were used; as they have the capability of
representing large deformation and material nonlinearity.
4.5.4. Boundary Conditions and Loading
Similarly, pavement layers were also assumed to bond together perfectly and the
models are fixed at the bottom of the sub-grade and roller constraints on the vertical
boundaries (Figure 4.3) (section 4.4.4). Here, a rectangular contact area of 72 557
mm2 was placed on the asphalt surface layer and was applied uniformly with a
pressure of 0.65 MPa (Theyse et al., 2011).
4.6. Non-Linear versus Linear Material Characterization
A comparative analysis of the two material characterizations input in the FE model is

undertaken here. Although, the linear material characterization method has been
used in the first two analyses, which is justifiable by the fact that results are easily
available. In this analysis, a scenario (Non-Linear Material model) of paved flexible
49

pavement is developed for a three-layered system of the pavement structure: asphalt

surface, 18 percent fly ash with 1 percent cement stabilized base, and sub-grade
layer. The model geometry, model mesh and element type, and boundary conditions
and loading are the same as in the paved stabilized base layer (section 4.5), with the
introduction of non-linear material characterization for the stabilized base layer in a
static-general analysis procedure, so as to consider the non-linear effect.
As mentioned in the review (section 3.4.3), LTTP model (Yau and Quintus, 2002)
which was adopted in the NCHRP 1-37A Design Guide (USDT-FHA 2014) was used
in obtaining the MR (1301 MPa) for 18 percent fly ash with 1 percent cement-
stabilized base layer, parameters such as (bulk stress = 1854kPa) are obtained from
Heyns and Mostafa Hassan (2013) and regression coefficients (k1 = 3000psi and k2 =
0.5) suggested by AASHTO (as cited in USDT-FHA 2014). Coupled with the elastic
model, D-P plasticity model in Abaqus was used for the material characterization to
be non-linear. In the D-P model, the shear criterion is selected to be ‘exponent form’
so as to allow for the use of sub-option (Triaxial test data) (Appendices A1 – C1) and
the dilation angle is assumed to be 15o. Furthermore, to validate the results obtained
from D-P model, a quick M-C model will be run in the model as well. Thereafter, the
results obtained will be compared with those obtained for linear material
characterization.
4.7. Comparative Analysis
The fourth objective of this dissertation is to compare laboratory-test empirical results

already available against that of 3D FEM. A comparative analysis of the empirical,
multilayer linear elastic software (mePADS) (Theyse and Muthen, 2000) and 3D
FEM (Non-Linear and Linear Material) results for a paved three-layered system with
18 percent fly ash plus 1 percent cement stabilized base layer is carried out.
However, estimating the structural capacity of flexible pavement through empirical
methods can be undertaken by the following: Pavement Structural Number, dynamic
cone penetrometer, 1993 AASHTO Structural Number (SN), etc. (SANRAL 2013b).
Nevertheless, the use of 1993 AASHTO SN is widely accepted, yet it has its own
disadvantages based on its assumptions (Pavement Interactive, 2008; SANRAL,
2013b). This method is based on the results of the AASHTO road test executed in
50

Ottawa, Illinois during the late 1950s to early 1960s and can be used for new and
rehabilitation pavement design. AASHTO SN empirical method is presented by
equation 4.1:
∆PSI
log10 ( )
log10 (SC) = ZR X So + 9.36 X log10 (SN + 1) − 0.2 + 4.2−1.5
1094 +
0.40+
(SN+1)5.19
2.32 X log10 (MR ) − 8.07 ………………………….………………. Equation 4.1
Where: SC = Predicted number of 80 kN ESALs
ZR = Standard normal deviate

So = Combined standard error of the traffic and performance predictions
SN = Structural number of the total pavement thickness
∆PSI= Difference between the initial (PSI0) and terminal (PSIt) serviceability
indices
MR = Sub-grade resilient modulus (in psi)
SN = a1 D1 + a2 D2 m2 + a3 D3 m3 …. ………………………..…..Equation 4.2
Where: SN = Structural number of the total pavement thickness
a = ith layer coefficient (per inch) (Table 4.3)

D = ith layer thickness (inches)
m = ith layer drainage coefficient; assumed = 1.0
Table 4.3. Layer coefficients (SANRAL, 2013b)
Materials Ranges for South African Materials

Asphalt concrete 0.20 – 0.44
Crushed stone 0.06 – 0.14
Cemented-treated material 0.10 – 0.28
Bituminous-treated material 0.10 – 0.30
Using the above equation, the structural capacity (in terms of ESALs) of flexible
pavement is calculated. To use this equation, the following input assumptions were
extracted from AASHTO design procedure (1993), Pavement Interactive (2008), and
51

SANRAL (2013b); the pavement was assumed to be a category B with the following
characteristic: reliability = 90 percent (ZR = 1.282), So = 0.45, total equivalent traffic
loading = 0.3 – 10 X 106, PSI0 = 4.5 and PSIt = 2.0 and MR = 45 MPa (Table 4.2).
Careful consideration should be given when using this equation as it is in imperial
units. Results obtained for empirical method (AASHTO SN) were compared with
those obtained for the 3D model in paved stabilized base layer. Furthermore, the use
of mePADS (see Appendices A2 – F2 for inputs data) serves as a check for the
performance of the 3D models. mePADS is mechanistic pavement design software,
which combines a stress-strain computational engine with pavement material models
and it’s capable of analysing pavement for bearing capacity. mePADS generates
outputs inform of pavement layer lives and contour plots of stresses and strains
(CSIR Built Environment, 2009). Although, there are various multilayer linear elastic
software but mePADS was selected based on its availability and suitability for South
Africa pavement design. Further, table 4.4 presents brief comparison between
Abaqus® and mePADS. Yet, recent report states that mePADS is currently been
updated since it works with the SAPDM principle which is currently under review
(CSIR Built Environment, 2009; SANRAL, 2013b) (see Section 3.3 and 3.6).
Table 4. 4 Comparison between Abaqus® and mePADS
Comparison Criterial Abaqus mePADS

Analysis Method 3D and 2D 2D-Axisymmetric
Developer Abaqus Inc. CSIR
Development Status Actively Developed Under Review
Loading Type Capacity Static and Dynamic Static
Operation Techniques FEM Multi-layer Elastic Method
Pavement Layer Non-limited 5 layers maximum
Problem Type Capacity Linear, Non-linear, etc. Linear only
Required Disk Space Very Small Required A lot of Space Required
Seconds to Hours
Time of Analysis Seconds
(analysis dependent)
Year of Originally Released 1978 2000
52

4.8. Summary
In this chapter, four different basic scenarios were developed to achieve the set
objectives of this study. These scenarios are: axisymmetric versus 3D, paved
stabilized base layer, non-linear versus linear material characterization, and a
comparative analysis of empirical and 3D FEM results, and a check by mePADS
software. Essentially with these scenarios the following will be achieved:
1. The efficiency of using 3D FEM for design of flexible pavement over

axisymmetric and the structural response of stabilized base layer in flexible
pavement;
2. The structural response effect of asphalt layer over stabilized base layer;
3. Efficiency of non-linear material characterization over linear; and
4. Benefits of 3D FEM design for flexible pavement over empirical methods.
53

Chapter 5 Results and Discussion
CHAPTER 5: RESULTS AND DISCUSSION
5.1. 3D versus Axisymmetric Results
This model was employed to study the efficiency of using the 3D model for design of
flexible pavement and the effect of unpaved stabilized base layer thickness on the
vertical compressive strain and stress at the top of the sub-grade. Firstly, from Figure
5.1 and 5.2, it was observed that the addition of stabilizer to natural G5 material
decreases the vertical compressive strain and stress at the top of the sub-grade
layer. On the other hand, the vertical compressive strain and stress at the top of the
sub-grade layer decrease with increase in the thickness of the stabilized layer for
axisymmetric model (Figure 5.1 and 5.2). It was observed that the increase in the fly
ash percentage was added, which resulted in an increase of the modulus of elasticity
of the base contributing to the reduction of vertical strain and stress at the top of the
sub-grade layer. Thus, increase in the modulus of elasticity of a layer reduces the
vertical strain and stress in the underneath layer. Similarly, the vertical compressive
strain and stress at the top of sub-grade layer decrease with the stabilized layer
thickness increase in 3D model. However, the results obtained from 20 percent fly
ash-stabilized base layer shows better results when compared with others; this
results from the high modulus of elasticity of the 20 percent stabilization. However,
considering the economical aspect and the fact that beyond 20 percent fly ash
strength starts to decrease, thus, 18 percent fly ash is considered best and
economical (Figure 5.1 and 5.2). However, it is recommended that a lower
stabilization percentage (10 percent–15 percent) should be experimented. Secondly,
comparing the results obtained from axisymmetric and 3D model for 18 percent fly
ash-stabilized base (Figure 5.3 and 5.4); results show that the 3D model is more
efficient, as vertical strain is centralized in the model against that of the axisymmetric
which tends to diverge toward a side of the model (Figure 5.5), which is far from
reality and overall vertical strains at the sub-grade are smaller. This implies that
numbers of load repetitions will be very small for axisymmetric (4.92 x 103) when
compared with that of 3D (1.30 x 106) and ASSHTO SN (11.54 x 106) results, taking
300mm stabilized base layer as an example. Thus, the axisymmetric model tends to
under-design, which is not economically wise. In both models the stabilized base of
54

100 mm generated excess strain and stress; this shows that the use of thinning
stabilized base layer would quickly result in pavement failure.
Furthermore, sub-grade rutting failure criteria analysis (section 3.5) using Asphalt
Institute model equation 2 (section 3.5) (Asphalt Institute 1982) for 18 percent fly
ash-stabilized base layer are presented in Table 5.1. The 3D model shows better
results, as it is obvious that using a 100-mm stabilized base layer would initiate
permanent deformation in the sub-grade layer under some loadings, such as 2.57
load repetitions against the result from axisymmetric model (62.6 number of load
repetitions). Thus using thinning stabilized base layer should not be encouraged and
overall, proper curing of stabilized base is necessary. Additionally, the 3D model
shows an increase in the number of load repetitions to failure for 200 mm–500 mm
thickness of base layer over that of axisymmetric model, implying that axisymmetric
tends to under-design for deep thickness and over-design for thin thickness.
However, using 300-mm stabilized base layer thickness against 100-mm increases
the capacity of the structure by approximately 100 percent.
800 0%FLYASH+0%CEMENT
VERTICAL COMPRESSIVE STRAIN AT THE
16%FLYASH+1%CEMENT
700
TOP OF SUBGRADE , (10-5)
18%FLYASH+1%CEMENT
600
20%FLYASH+1%CEMENT
500
22%FLYASH+1%CEMENT
400
300
200
100
0
100 200 300 400 500
STABILIZED BASE LAYER THICKNESS, MM
Figure 5.1. Effect of stabilized base layer thickness on vertical compressive

strains on the top of sub-grade (Axisymmetric model)
55

0.70
VERTICAL COMPRESSIVE STRESS AT THE
0.60
TOP OF SUBGRADE, MPA
0%FLYASH+0%CEMENT
0.50
16%FLYASH+1%CEMENT
0.40 18%FLYASH+1%CEMENT
20%FLYASH+1%CEMENT
0.30
22%FLYASH+1%CEMENT
0.20
0.10
0.00
100 200 300 400 500
Figure 5.2. Effect of stabilized base layer thickness on vertical compressive

stress on the top of sub-grade (Axisymmetric model)
900
3D FEM
800
TOP OF SUBGRADE, (10-5)
700 AXISYMMETRIC
600
500
400
300
200
100
0
100 200 300 400 500
Figure 5.3. Effect of stabilized (18 percent fly ash + 1 percent cement) base
layer thickness on vertical compressive strains on top of sub-grade
(Axisymmetric and 3D model)
56

VERTICAL COMPRESSIVE STRESS AT THE 0.40
0.35
3D FEM
0.30
AXISYMMETRIC
0.25
0.20
0.15
0.10
0.05
0.00
100 200 300 400 500
Figure 5.4. Effect of stabilized (18 percent fly ash + 1 percent cement) base
layer thickness on vertical compressive stress on top of sub-grade
(Axisymmetric and 3D model)
Figure 5. 5 Contour plots showing deformation of vertical compressive strain

at the top of sub-grade (A- 3D Model and B- 2D Model)
57

Table 5.1. Rutting failure analysis based on Asphalt Institute response model
(1982) for Axisymmetric and 3D model
Rutting Criterion
Base Layer Vertical Strain No. of Vertical Strain No. of
Thickness Ԑc (10-6) in Repetitions to Ԑc (10-6) in Sub- Repetitions to
(mm) Sub-grade Failure Nr (3D) grade (2D) Failure Nr (2D)
(3D)
100 8481 2.57 4155 62.6
200 999.9 36.8 x 103 2276 0.9 x 103
300 451.0 13.0 x 105 1568 4.9 x 103
400 272.7 12.4 x 106 1228 14.7 x 103
500 187.9 65.6 x 106 1039 31.0 x 103
5.2. Paved Stabilized Base Layer Results
The protective importance of surface layer over stabilized base layer is evaluated by
the paved stabilized base layer model. On surface deflection, it can be seen from
Figure 5.6; that asphalt layer-deflection decreases with an increase in thickness,
which is related to the conclusion in a study by Shafabakhsh et al. (2013a). The
conclusion was that increase in asphalt layer thickness reduces the surface
deflection and the other layer. Similarly, vertical compressive stress/strain at the top
of the sub-grade layer compared with the unpaved also decreases in the same
manner (Figure 5.7 and 5.8). However, the use of 50 mm thickness of asphalt is
recommended in developing countries for economic reasons (Araya, 2011).
Similarly, the tensile horizontal strain at the bottom of the surface layer (Figure 5.9)
shows a decrease: quite the reverse for 100 mm thickness, as there was an increase
in strain; this implies high probabilities for bottom-up fatigue cracking to occur with
increase of asphalt surface thickness over a stabilized base layer. Additionally, the
stabilized base layer-vertical strain increases initially and tends to decrease on 100
mm thickness of asphalt (Figure 5.10); this implies that an increase in asphalt layer
has a significant effect on stresses and strains generated in all layers of flexible
pavement.
58

0.40
0.35
SURFACE DEFLECTION, mm (10-5)
0.30
0.25
0.20
0.15
0.10
0.05
0.00
25 50 75 100
ASPHALT SURFACE LAYER THICKNESS, mm
Figure 5.6. Effect of asphalt layer thickness on surface deflection over a

stabilized base layer (3D Model)
500
450
400
TOP OF SUBGRADE, (10-6)
350
300
250
200
150
100
50
0
0 25 50 75 100
Figure 5.7. Effect of asphalt layer thickness on vertical compressive strains on

the top of sub-grade (3D Model)
59

VERTICAL COMPRESSIVE STRESS AT THE 0.70

STABILIZED BASE
0.60
SUBGRADE
0.50
0.40
0.30
0.20
0.10
0.00
0 25 50 75 100
ASPHALT SURFACE LAYER THICKNESS, MM
Figure 5.8. Effect of asphalt layer thickness on vertical compressive stress

within stabilized base/on the top of sub-grade (3D Model)
40
TENSILE HORIZONTAL STRAIN AT THE
BOTTOM OF SURFACE LAYER, (10-6)
35
30
25
20
15
10
0
25 50 75 100
Figure 5.9. Effect of asphalt layer thickness on tensile horizontal strain at the
bottom of asphalt layer (3D Model)
60

140
VERTICAL COMPRESSIVE STRAIN WITHIN
135
STABILIZED BASE LAYER, (10-6)
130
125
120
115
110
25 50 75 100
Figure 5.10. Effect of asphalt layer on vertical compressive strain within

stabilized base layer (3D Model)
Table 5.2. Fatigue and rutting failure analysis based on Asphalt Institute
Response Model (1982) for 3D model
Fatigue Criterion Rutting Criterion
Asphalt Layer Tensile Strain No. of Load Vertical No. of Load
Thickness Ԑt (10-6) Repetitions Strain Repetitions
(mm) bottom of to Ԑc (10-6) in to
Asphalt Layer Failure Nf Sub-grade Failure Nr
25 31.94 7.6 x 108 376.60 2.92 x106
50 27.53 12.39 x 108 325.50 5.60 x 106
75 23.08 22.14 x 108 286.70 9.89 x 106
100 35.60 5.32 x108 253.90 17.04 x106
According to the model suggested by Asphalt Institute, the capacity in terms of

number of load repetition before failure is calculated for the paved stabilized base
layer pavement. From Table 5.2, it was observed for both asphalt fatigue and sub-
grade rutting; that increase in thickness of asphalt surface layer increase the number
of load repetition before failure. Although, for 100 mm thickness there was an
61

increase load repetition (17.04 x106) before rutting failure, yet the fatigue failure in
terms of load repetition (5.32 x108) decreases at this thickness. Thus, increase in the
asphalt surface layer does not necessary increase the bearing capacity of the
pavement structure as other pavement layers are contributing factors to flexible
pavement bearing capacity.
5.3. Non-Linear versus Linear Material Characterization Results
According to Abaza (2007), non-linear material characterization over linear gives a

close field measurement, thus here a comparative analysis of non-linear and linear
material characterization was undertaken. Figures 5.11 – 5.13 show the contour
plots for displacements, strains and stresses in 25 mm asphalt thickness layer
respectively. From Figure 5.11, it was observed that the maximum magnitude of
deflection (rutting - 4.544 x 10-4 m) was higher in Figure 5.11B, which is for non-
linear model, implying that material acts like an elasto-plastic thus did not totally
return to the original state. Similarly, from Figure 5.12, the maximum strain (1.838 x
10-4 m) was higher in the non-linear model but also worth to noting that the minimum
strain (-5.076 x 10-6 m) was higher in the linear model, thus implying that strain in the
linear model extended to the lower part of the sub-grade which will overall result in
failure.
In Figure 5.12, the maximum stress transfer (tyre load) through the linear model was
high, thus implying that more stress is transferred to the rest of the layers. Overall,
there are not many differences in the results obtained, despite the M R (1301 MPa)
used in non-linear model is smaller when compared with that of linear model (2560
MPa).
62

Figure 5.11. Displacement contour plot for 25 mm thickness asphalt layer (A-
Linear model and B- Non-linear model)
Figure 5.12. Strain contour plot for 25 mm thickness asphalt layer (A- Linear
model and B- Non-linear model)
63

Figure 5.13. Stress contour plot for 25 mm thickness asphalt layer (A- Linear
model and B- Non-linear model)
Table 5.3. Effect of asphalt layer thickness for non-linear material model
Vertical Strain
Asphalt Layer Thickness -6
Tensile Strain Ԑt (10-6)
Ԑc (10 ) in Stabilized
(mm) bottom of Asphalt Layer
base Layer
25 259.1 38.57
50 285.7 30.92
75 273.9 41.46
100 247.5 61.55
Furthermore, from Table 5.3 above it is of a great interest to note that it is against the
trend in the linear model for increase in thickness of asphalt layer which was
reported in Figures 5.9 and 5.10 for vertical compressive strain in stabilized base
and tensile horizontal strain in asphalt layer. The results for the non-linear model
experienced an increase in the compressive strain for stabilized base in 50 mm
thickness asphalt layer and thereafter a decrease. Conversely, the horizontal
decreases in the 50 mm thickness and thereafter increases for subsequent
thickness, thus, implying that the thickness of asphalt layer beyond 50 mm may
result in bottom-up fatigue cracking. On a comparative note, the results obtained
64

from D-P model compared with those of the M-C model; at first (i.e. in 25 mm
Asphalt thickness layer) experienced a difference of about 0.2 percent in the results
obtained for displacements, strains and stresses. However, the subsequent results
were comparable. This implies that the D-P model or M-C model is a good non-linear
material representation for stabilized base layers in pavement design. Overall, it is
worth noting that the use of 50 mm thickness of asphalt layer over the stabilized
base layer by developing countries, is not only justifiable by economic reasons, but
also on its effectiveness to prevent failure such as bottom-up fatigue cracking which
can be experienced in thicknesses beyond 50 mm.
5.4. Comparative Analysis Results
Comparing the results obtained from Abaqus (Linear model) and that of mePADS in
terms of horizontal strain at the bottom of asphalt layer and the vertical strain in the
subgrade (Table 5.4). Results show that the strains generated in the mePADS are
generally low when compared to that of Abaqus. On like in Abaqus, results at 100
mm asphalt layer thickness did not follow the regular pattern but that of mePADS
was consistent. Thus, the results from Abaqus can be said to be dynamic in nature.
Table 5. 4 Effect of asphalt layer thickness for Abaqus and mePADS

Abaqus 3D (Linear Model) mePADS
Asphalt Layer Tensile Strain Vertical Tensile Strain Vertical
-6 -6
Thickness Ԑt (10 ) Strain Ԑt (10 ) Strain
(mm) bottom of Ԑc (10-6) in bottom of Ԑc (10-6) in
Asphalt Layer Sub-grade Asphalt Layer Sub-grade
25 31.94 376.60 14.13 195
50 27.53 325.50 26.26 169
75 23.08 286.70 40.58 149
100 35.60 253.90 41.81 134
Furthermore, Table 5.5 presents the pavement structural capacity results obtained
from the use of 1993 AASHTO SN empirical method, mePADS and those obtained
using 3D FEM (Non-Linear and Linear Material (Table 5.2)) with the Asphalt Institute
model. Results from the mePADS (see Appendices A2 – F2); which serves a check
65

for the performance of 3D FEM models, although within a close range yet, tends to
be higher than those of AASHTO SN and those of 3D FEM models. This is so
because the SAPDM damage model used in software in question is outdated and
currently under review (SANRAL, 2013b).
However, there are not many differences in the results obtained from AASHTO SN
and those of 3D FEM-linear materials, yet those of 1993 AAHSTO were higher. In
the report by Huber, Andrewski and Gallivan (2009), the AASHTO 1993 pavement
design guide was found to have typically over-designed pavements in Indiana by 1.5
to 4.5 inches beyond what was needed. Thus, it can be concluded that the 1993
AASHTO SN tends to over-design, which makes its use uneconomical. Additionally,
from Table 5.5, results from linear models are higher than those of non-linear, which
also show that the linear model tends to over-design as a result of the MR of the
stabilized base layer used. This MR is obtained using level 2 inputs (lower reliability
when compared with level 1), thus it can be concluded that M R has a significant
effect on the design of pavement through FEM. Overall, the 3D FE non-linear model
tends not to be partial in its design as there are few assumptions to be made in using
it for the design of pavement structure and the fact that M R was obtained through
Triaxial testing, which gives the true strength of materials used in pavement
structure.
Table 5.5. Structural capacity results for 1993 AASHTO, mePADS and 3D FEM
models
Asphalt Predicted No. Sub-grade No. of Load No. of Load
Layer of 80 kN Bearing Capacity Repetitions Repetitions to
Thickness ESALs (1993 (mePADS to Failure Nr Failure Nr
(mm) ASSHTO SN) Results) (Linear (Non-Linear
Model) Model)
7 12 6
25 10.59 x 10 30.70 x 10 2.92 x10 5.41 x 105
50 31.00 x 107 12.70 x 1014 5.60 x 106 1.13 x 106
75 79.00 x 107 43.11 x 1014 9.89 x 106 2.13 x 106
100 185.20 x 107 10.00 x 1015 17.04 x106 3.91 x 106
66

5.5. Summary
Results of this study were presented in this chapter. As expected, based on the
literature reviews, the following results were observed:
1. 3D FE model results for design of flexible pavement were more efficient when
compared with those of axisymmetric;
2. The structural response of stabilized base and asphalt layer were discovered
and are of great importance in flexible pavement;
3. The current update for mePADS software is quite necessary, especially in
terms of the damage models;
4. Non-linear material characterization model is efficient over linear model; and
5. Overall 3D FEM design for flexible pavement is efficient over empirical
methods.
67

Chapter 6 Conclusions and Recommendations
CHAPTER 6: CONCLUSIONS AND RECOMMENDATIONS
6.1. Introduction
The stabilization process in pavement construction is not a new process, but hitherto
this process has not been fully implemented in the design methods for pavement
structure. Although, in recent decades, researchers have tried to implement it in the
existing empirical methods (Al-Jhayyish, 2014), but these methods are already
inaccurate in their design and are limited in their capacity (Huang, 2004), thus,
bringing about the use of FEMs. Considering the success recorded using FEMs, it is
a necessity to incorporate the stabilization process such as fly ash-stabilized base
layer into it, which was the essence of this study. As a result, an attempt was made
to simulate the behaviour of the flexible road pavements having fly ash as an
alternative soil stabilizer using FEM. This simulation study was undertaken by
creating FEMs using Abaqus® to study the structural responses of the stabilized
base layer and the responses of flexible pavement when constructed with fly ash-
stabilized base layer. Therefore, in this final chapter the main conclusions of this
thesis are summarized and some recommendations are given.
6.2. Conclusions
As a result of the modelling and analysis which were performed in this study the
following conclusions were obtained;
 3D FE models are more efficient than 2D axisymmetric models.

 Increase in the MR of any material in pavement structure, increases the overall
pavement resistivity to failure.
 Increase in the thickness of fly ash-stabilized base layer increases the
resistance of pavement to failure in terms of surface deflection, vertical
compressive stresses/strains on top of the sub-grade layer; however, increase
beyond 300 mm results in strength decrease.
 In the same manner, increase in the thickness of asphalt layer increases
pavement resistivity to failure; however, increase in thickness beyond certain
68

Chapter 6 Conclusions and Recommendations
thickness, especially over a stabilized base may result in bottom-up fatigue

cracking.
 The use of non-linear material characterization model is more efficient than
linear material characterization. However, as a result the unavailability of
Triaxial test results, the linear material characterization model can be used as
a preliminary study.
 The results obtained from D-P and M-C models are comparable, thus either
can be used in a material characterization model in pavement design.
Overall, the uses of empirical deign methods result in over-designing of pavement

structure, consequently resulting in uneconomical pavement design and
construction. However, the use of 3D FE models and most especially, the non-linear
material characterization model provides better results and gives some amount of
certainty on the design life of the pavement.
6.3. Recommendations
Since the structural element in the pavement is formed by the thickness and strength
base and sub-base layers placed over the sub-grade, there is a need for further
study on the materials used in these layers. Furthermore, it is recommended that the
fly ash as a stabilizer should be experimented with a lower percentage (10 percent –
15 percent), as percentages beyond 20 percent result in strength reduction and
economical unwise.
6.4. Further Studies
Firstly, it was discovered that the fly ash stabilizer for pavement materials lacks
correlation equations for deriving MR using UCS test data; secondly, there is a need
to develop resilient modulus constitutive material models for South Africa granular
material, especially for stabilized materials as it is commonly used as a base and
sub-base layer in flexible pavement. Lastly, for further study there is a need to put
into consideration the effect of climate conditions in terms of temperature, rainfall,
etc., on the material characterization model in FEM design of pavement structures.
69

References
REFERENCES
Abaqus Inc. (2003). Analysis of geotechnical problems with ABAQUS. 2003.

ABAQUS, Inc.
Abaqus, (2013). Analysis User's Guide (6.13). [Online] Available at:

http://129.97.46.200:2080/v6.13/books/usb/default.htm [Accessed: 2 Apr.
2014].
Abaza, O. (2007). Simplified ME Approach for the Design of Flexible Pavement

Structures. An-Najah University Journal for Research, 21, pp.129-150.
Abed, A.H. and Al-Azzawi, A.A (2012), Evaluation of Rutting Depth in Flexible
Pavements by Using Finite Element Analysis and Local Empirical Model.
American Journal of Engineering and Applied Sciences, 5(2), pp.163-169.
Adu-Osei, A. (2001). Characterization of Unbound Granular Layers in Flexible

Pavements. Technical Reported, Texas A&M University Texas Transportation
Institute College Station, Texas.
Al-Azzawi, A. A. (2012). Finite element analysis of flexible pavements strengthened

with geogrid. ARPN Journal of Engineering and Applied Sciences, 7(10),
pp.1295-1299.
Al-Jhayyish, A. K. (2014). Incorporating Chemical Stabilization of the Subgrade in

Pavement Design and Construction Practices. PhD Thesis, Ohio University.
Al-Khateeb, L.A., Saoud, A. and Al-Msouti, M.F. (2011). Rutting prediction of flexible
pavements using finite element modelling. Jordan Journal of Civil
Engineering, 5(2), pp.173-189.
Ameri, M., Salehabadi, E.G., Nejad, F.M. and Rostami, T. (2012). Assessment of
Analytical Techniques of Flexible Pavements by Final Element Method and
Theory of Multi-Layer System. Journal Basic Applied Science Research,
2(11), pp.11743-11748.
70

References
American Association of State Highway and Transportation Officials (AASHTO).

(1993), AASHTO Guide for design of pavement structures. Washington, DC.
American Coal Ash Association. (1990). Flexible Pavement manual. Alexandria,

Virginia.
American Coal Ash Association. (1995). Fly Ash Facts for Highway Engineers.
American Coal Ash Association, Inc., Alexandria, Virginia FHWASA-94-081.
Amin, E. R. (2012). A Review on the Soil Stabilization Using Low-Cost Methods.

Journal of Applied Sciences Research, 8 (4), pp.2193-2196.
Aminaton, M., Nima, L. and Houman, S. (2013). Stabilization of Laterite Soil using
GKS Soil Stabilizer. Electronic Journal of Geotechnical Engineering, 18,
pp.521-532.
Araya A. A. (2011). Characterization of Unbound Granular Materials for Pavement.,

Master’s Thesis, Road and Railway Engineering Section, Faculty of Civil
Engineering and Geosciences, Delft University of Technology.
Asphalt Institute. (1982). Research and Development of Asphalt Institute’s Thickness

Design Manual. 9th Ed., Research Report 82-2, the Asphalt Institute.
Asphalt Pavement Alliance (APA). (2010). Perpetual asphalt pavements: A

synthesis. APA 101, Lanham, Maryland.
ASTM C618. (2011). Standard Specification for Fly Ash and Raw or Claimed Natural
Pozzolans for Use as a Mineral Admixture in Portland Cement Concrete.
American Society for Testing and Materials, Annual Book of ASTM Standards,
Volume 04.02, West Conshohocken, Pennsylvania.
Áurea, S. H., Evandro, P.J. and Lucas, T. B. (2006). Finite Element Modelling of
Flexible Pavement.
Australian Road Research Laboratory (1998). Correlations between compressive

strength and modulus. Sydney, New South Wales, Australia.
71

References
Ban, H. and Park, S. (2014). Characteristics of Modified Soil-Aggregate System and

Their Application in Pavements. KSCE Journal of Civil Engineering, (0), pp.1-
7.
Barenberg, E. J. (1977). Evaluating stabilized materials. National cooperative

highway research program, contract no. HR 63-4-1.
Bindu J. and Vysakh P. (2012). Stabilisation of Lateritic soil using coconut shell, leaf
and husk ash. Institution of Electrical and Electronics Engineers, 978-1-4673-
2636-0/12, pp.247-279.
Boussinesq, J. (1885). Application of the potential for the study of equilibrium and
motion of elastic solids. Paris. Lille. pp. 57 – 75.
Brennan, M. J. and O’Flaherty, C. A. (2002). Materials used in road pavements, In

O’Flaherty, C.A., Highways: the location, design, construction & maintenance
of road pavement. Butterworth-Heinemann, Oxford, pp.118-162.
Britto A. Running ABAQUS 6.9 [Online]. University of Cambridge. 2010. Available at:
http://www-h.eng.cam.ac.uk/help/programs/fe/abaqus/doc/aba69fdouble.pdf.
[Accessed: 6 December 2014].
Brooks, R. M. (2009). Soil stabilization with fly ash and rice husk Ash. International
Journal of Research and Reviews in Applied Sciences, 1(3), pp.209-217.
Brooks, R. M., Hutapea, P., Obeid, I., Bai, L. and Takkalapelli, K.V. (2008). Finite
Element Method - a Tool for Learning Runway Design. ASEE Mid-Atlantic
(Zone I) Conference, West Point, NY.
Bureau for Industrial Cooperation. (2012). FINAL REPORT - Preparation of a

Transport Facilitation Strategy for the East African Community’, [Online]
Available at: www.eac.int/infrastructure/index.php. [Accessed: 20 September
2014].
72

References
Burmister, D. M. (1943). The Theory of Stresses and Displacements in Layered

Systems and Application to the Design of Airport Runways. Proceedings,
Highway Research Board, 23.
Chen, H. H., Marshek, K. M. and Saraf, C. L. (1990). Effects of Truck Tire Contact
Pressure Distribution on the Design of Flexible Pavements: A Three
Dimensional Finite Element Approach. Transportation Research Record, No.
1095, pp.7278.
Cho, Y. H., McCullough, B. F., Weissmann, J. (1996). Considerations on Finite-

element method application in pavement structural analysis. Transportation
Research Board, 1539, pp. 96-101.
Çöleri, E. (2007). Relationship between Resilient modulus and Soil index properties
of unbound materials. Master’s Thesis, Middle East Technical University.
Comberato, J. J., Vance, E. D. and Someshwar, A. V. (1997). Composition and land

application of paper manufacturing residuals. In: Rechcigl, J., Mackinnon, H.
(Eds.), Agricultural Uses of By-products and Wastes. ACS, Washington, DC,
185–203.
Committee of Land Transport Officials (COLTO), 1998 Edition.
Council of Scientific and Industrial Research (CSIR). (2010). Potholes: Technical

guide to their causes, Identification and Repair. [Online] Available at:
www.csir.co.za.pothole_guides. [Accessed: 21 September 2014].
CSIR Built Environment (2009). Pavement Analysis and Design Software. [Online]
Available at: http://asphalt.csir.co.za/samdm/. [Accessed: 29 July 2015].
Darwish S. G., (2012). Stress and Strain in Flexible Pavement. [Online] Available at:
http://www.unimasr.net/ [Accessed: 27 March 2014].
Davids, W. G. (2009). EverStressFE1.0 Software for 3D Finite-Element Analysis of

Flexible Pavement Structures. The Washington State Department of
Transportation.
73

References
De Beer, M., Fisher, C., and Jooste, F. J. (1997). Determination of pneumatic

tire/pavement interface contact stresses under moving loads and some effects
on pavements with thin asphalt surfacing layers. In: Proceedings 8th
International Conference on Asphalt Pavements, University of Washington,
Seattle, pp.179–227.
Desai, C. S. (2012). Application of Finite Element and Constitutive Models.
Dione, A., Fall, M., Berthaud, Y., and Makhaly, B. (2013). Estimation of Resilient
Modulus of Unbound Granular Materials from Senegal (West Africa).
Geomaterials, 2013(3), pp.172-178.
Dixit, U. (2007). Finite Element methods in Engineering.
Duncan, J. M., Monismith, C. L. and Wilson, E. L. (1968). Finite Element Analysis of

Pavements. HRR-228, Highway Research Board, National Research Council,
Washington, D.C.
Dunlap, W. S. (1963). A report on a mathematical model describing the deformation

characteristics of granular materials. Technical Report 1, Project 2-8-62-27,
TTI, Texas A & M University.
Ekwulo, E. O., and Eme, D. B. (2009). Fatigue and rutting strain analysis of flexible
pavements designed using CBR methods. African Journal of Environmental
Science and Technology, 3(12), pp.412-421.
Eluozo, S. N. (2013). Predictive Model to Determine Resilient Modulus on

Bituminous Mixtures in Load Repetition on Pavement Failure. International
Journal of Materials, Methods and Technologies, 1(5), pp.82-89.
Eskioglou, P. and Oikonomou, N. (2008). Protection of Environment by the use of Fly

Ash in Road Construction.
Fall, M., Ba, S., Sarr, D., Ba, M. and Ndiaye M. (2011). An Alternative Method to the
West African Compaction (WAC) Test Procedure. Geomaterials, 1(2), pp.25-
27.
74

References
FM5-410, (2012). Soil Stabilization for Road and Airfield. [Online] Available at:
www.itc.nl/~rossiter/Docs/FM5-410. [Accessed: 25 March 2014].
Furter, E. (2011). Coal Fly ash holds African opportunities. [Online] Available at:
www.sheqafrica.com/fly-ash-conference/ [Accessed: 25 March 2014].
George, K. P. (2004). Prediction of Resilient Modulus from Soil Index Properties –

Final Report. The University of Mississippi.
Gupta, A. and Kumar, A. (2014). Comparative Structural Analysis of Flexible

Pavements Using Finite Element Method. The International Journal of
Pavement Engineering and Asphalt Technology (PEAT), 15(1), pp.11-19.
Gyanen, T., Savitha, A.L. and Gudi, K. (2013). Laboratory study on soil stabilization
using fly ash mixtures. International Journal of Engineering Science and
Innovative Technology, 2(1), pp.477-482.
Hafeez, I. (2010). Empirical and mechanistic flexible pavement designed. [Online]

Available at: http://pec.org.pk/sCourse-files/Lec1.ppt. [Accessed: 25 March
2014].
Harichandran, R. S., Yeh, M. S., and Baladi, G. Y. (1989). MICH-PAVE: A Nonlinear

Finite Element Program for Analysis of Pavements. Transportation Research
Record 1286, TRB, National Research Council, Washington, D.C., pp.123-
131.
Harold, L. and Von Quintus, P. (2007). Evaluation of Procedure to Estimate

Subgrade Resilient Modulus for use in Pavement Structural Design. Kansas:
Kansas Department of Transportation Bureau of Materials and Research.
Heebink, L. V. and Hassett, D. J. (2001). Coal Fly Ash Trace Element Mobility in Soil
Stabilization. www.flyash.info, International Ash Utilization Symposium,
Centre for Applied Energy Research, University of Kentucky, Paper #64.
75

References
Hejazi, S. M., Sheikhzadeh, M., Abtahi, S.M. and Zadhoush, A. (2012). A simple
review of soil reinforcement by using natural and synthetic fibres. Construction
and Building Materials, 30, pp.100–116.
Heukelom, W., and Klomp, A. G. (1962). Dynamic Testing as a Means of Controlling

Pavements during and after Construction. In: Proceedings of the International
Conference on the Structural Design of Asphalt Pavement, Ann Arbor,
Michigan, U.S.A.
Heyns, M. W. and Mostafa Hassan, M. (2013). South Africa Class F Fly Ash for
roads: physical and chemical analysis. Interim: Interdisciplinary Journal, 12(3),
pp.28-41.
Hjelmstad, K. D., Kim, J. and Zuo, Q. H. (1996). Finite Element Procedures for Three
Dimensional Pavement Analysis. Transportation Research Record 1539,
TRB, National Research Council, Washington, D.C, pp.66-71.
Hua, J. (2000). Finite Element Modelling and Analysis of Accelerated Pavement

Testing Devices and Rutting Phenomenon', Ph.D. Thesis, Purdue University,
West Lafayette. Indiana, USA.
Huang, Y. H. (2004). Pavement Analysis and Design. 2nd Edition, Pearson Education
Inc, NJ.
Huber, G., Andrewski, D. and Gallivan, V. (2009). Design and Construction of

Highways for Very Heavy Trucks. In: Proceedings International Conference of
Perpetual Pavements. Ohio University. Columbus. CD-ROM.
Ibrahim, S. F., Gandhi, G. S. and Zaman, T.T. (2014). An Approach in Evaluating of

Flexible Pavement in Permanent Deformation OF Paved and Unpaved Roads
over Sand Dunes Subgrade under Repeated Loads. Journal of Environment
and Earth Science, 4(14), pp.78-89.
Ji, R., Siddiki, N., Nantung, T. and Kim, D. (2014). Evaluation of Resilient Modulus of
Subgrade and Base Materials in Indiana and Its Implementation in MEPDG.
The Scientific World Journal, pp.1-14.
76

References
Johnson, D., Sukumaran, B., Mehta, Y. and Willis, M. (2007). Three Dimensional
Finite Element Analysis of Flexible Pavements to Assess the Effects of
Wander and Wheel Configuration. Presented for the 2007 FAA worldwide
airport technology transfer conference, Atlantic City, New Jersey, USA.
Jooste F. (2004). Re-Evaluation of Some Aspects of the Mechanistic-Empirical

Design Approach. In: Proceedings of the 8th Conference on Asphalt
Pavements for Southern Africa. Sun City, South Africa. [CD].
Kim B., Prezzi M. and Salgado R. (2005). Geotechnical properties of fly and bottom
ash mixtures for use in highway embankments. Journal of Geotechnical and
Geoenvironmental Engineering, 131(7), pp.914–924.
Kim, D. (2002). Effects of Super-Single Tire Loading on Subgrade. Ph.D.

Dissertation, Purdue University, West Lafayette, Indiana.
Kim, D. and Siddiki N. Z. (2006). Simplification of Resilient Modulus Testing for

Subgrades. Joint Transportation Research Program, Indiana Department of
Transportation and Purdue University, West Lafayette, Indiana.
Kim, M. (2007). Three-Dimensional Finite Element Analysis of Flexible Pavements

Considering Nonlinear Pavement Foundation Behaviour. Department of Civil
Engineering, Graduate College of the University of Illinois at Urbana-
Champaign.
Kordi, N. E., Endut, I. R. and Baharom, B. (2010). Types of Damages on Flexible

Pavement for Malaysian Federal Road. In: Proceeding of Malaysian
Universities Transportation Research Forum and Conferences.
Liang, A. (2000). Finite Element Analysis of Distortional bucking of Cold-formed

Stainless Steel Columns. Master’s Thesis, University of Johannesburg.
Lin Li, C. H., Benson, T. B. and Edil, B. H. (2007). Sustainable Construction Case
History: Fly Ash Stabilization of Recycled Asphalt Pavement Material.
Transportation Research Board, 86th Annual Meeting.
77

References
Little, D., Snead, B., Godiwalla, A., Oshiro, P., and Tang, P. (2002). Characterization
of Design Properties (Compressive Strength and Resilient Modulus) Of Lime,
Cement, Fly Ash Stabilized Structural Recycled Concrete Base As A Function
Of Curing Time. In Presentada en The 2002 Federal Aviation Administration
Airport Technology Transfer Conference, pp. 5-17.
Lo, S. H. (2002). Finite element mesh generation and adaptive meshing. Progress in
Structural Engineering and Materials, 4(4), pp.381-399.
Maharaj, D. K. and Gill, S. (2014). The behaviour of flexible pavement by non-linear

finite element method. International journal of latest research in science and
technology, 3(1), pp.53-57.
Makusa, G. P. (2012). Soil Stabilization Methods and Materials. PhD Thesis,

Department of Civil, Environmental and Natural resources, Luleå University of
Technology, Luleå, Sweden.
Mallela, J., Harold Von Quintus, P. E., Smith, K. L. and Consultants, E. R. E. S.

(2004). Consideration of lime-stabilized layers in mechanistic-empirical
pavement design. The National Lime Association.
Mamlouk, M. and Mobasher, B. (2004). Cracking Resistance of Asphalt Rubber Mix

versus Hot-Mix Asphalt. International Journal of Road Materials and
Pavement Design, 5(4), pp.435-452.
Mansurkhaki, A., Hesami, S., Khajehhassani, I. and Khordehbinan, M.W. (2014).

Evaluation of Role of ABAQUS and KENLAYER Software in Linear Elastic
Analysis of Flexible Pavement. Nationalpark-Forschung in Der Schweiz
(Switzerland Research Park Journal), 103(1), pp.546-562.
Mathur, S.S. (2011). Utilization of Fly Ash in Roads and Embankment Works.
May, R. W. and Witczak, M. W (1981). Effective Granular Modulus to Model

Pavement Response. Transportation Research Record 810, TRB, National
Research Council, Washington, D.C.
78

References
McClelland Engineering (1986), unpublished. Correlations between compressive

strength and modulus for LCF material on Runway 9-27. Intercontinental
Airport, Houston, Texas.
McDowell, C. (1959). Stabilization of soils with lime, lime-fly ash and other lime
reactive Materials. Highway Research Board Bulletin, 231, Washington, DC,
pp.60-66.
Mgangira, M. B. (2006). Laboratory assessment of the influence of the proportion of

waste foundry sand on the geotechnical engineering properties of clayey
soils. Journal of the South African Institution of Civil Engineering, 48(1), pp.2–
7.
Molenaar, A. A. A, (2009). Part III: Design of flexible pavements. [Online] Available

at: http://www.citg.tudelft.nl/. [Accessed: 28 March 2014].
Moossazadeh, J. and Witczak, M. W. (1981). Prediction of Subgrade Moduli for Soil

that Exhibits Nonlinear Behaviour. Transportation Research Record 810,
TRB, National Research Council, Washington, D.C., pp.9-17.
Mostafa Hassan, M., & Khalid, H. (2010). Mechanical and environmental

characteristics of bituminous mixtures with incinerator bottom ash aggregates.
International Journal of Pavement Engineering, 11(2), pp.83-94.
Mun, S. H. (2003). Nonlinear Finite Element Analysis of Pavements and Its

Application to Performance Evaluation. Dissertation, Department of Civil and
Environmental Engineering, North Carolina State University.
NCHRP (2004). Guide for Mechanistic-Empirical Design of New and Rehabilitated

Pavement Structures: Appendix RR-Finite Element Procedures for Flexible
Pavement Analysis, Transportation Research Board, National Research
Council, Washington D.C.
Ndebele, S. J. (2012). Keynote address by the minister of transport. Presented at the

occasion of department’s budget vote, 25 April.
79

References
Nicholas J Garber and Lester A. H. (2010). Traffic and Highway Engineering, SI

version. 4th edition, Cengage Learning, Stamford, USA.
Nunes, N. C., Bridges, M. G. and Dawson, A.R. (1996). Assessment of secondary

materials for pavement construction: Technical and Environment aspects.
Waste management, 16(1-3), pp.87-96.
Openshaw, S. C. (1992). Utilization of Coal Fly Ash. Master’s thesis, Department of

Environmental and Civil Engineering, University of Florida.
Paige-Green P. (2008), ‘A reassessment of some road material stabilization

problems’, Proceedings of the 27th Southern African Transport conference,
July 7-11, pp. 125-134.
Pandian, N. S. and Krishna, K. C. (2002). The Pozzolanic Effect of Fly Ash on the
CBR Behaviour of Black Cotton Soil. Journal of Testing and Evaluation,
ASTM, 31(6), pp.479-485.
Pavement Interactive. (2008). Flexible pavement mechanistic models. [Online]

Available at: www.pavement-interactive.org/article/flexible-pavement-
mechanistic-models/ [Accessed: 27 March 2014].
Peng, Y. and He, Y. (2009). Structural characteristic of cement-stabilized soil bases

with 3D finite element method, Front Architect Civil Engineering China, 3(4),
pp.428-434.
Psarras, S., Falzon, B. G. and Pinho, S. T. (2012). Damage-Tolerant Design of

Stiffener Run-Outs: A Finite Element Approach. INTECH Open Access
Publisher.
Rahman, M.T., Mahmud, K. and Ahsan, S. (2011). Stress-Strain characteristics of

flexible pavement using Finite Element Analysis. International Journal of Civil
and Structural Engineering, 2(1), pp.233-240.
Rao, C., L. Titus-Glover, B. Bhattacharya, and M. I. Darter. (2012). User’s Guide:

Estimation of Key PCC, Base, Subbase, and Pavement Engineering
80

References
Properties from Routine Tests and Physical Characteristics (Technical Report;

FHWA-HRT-12-031). Federal Highway Administration, Washington, DC.
Reddy, J. N. (1993). An introduction to the finite element method (Vol. 2, No. 2.2).
New York: McGraw-Hill.
Road Research Laboratory. (1970). A Guide to structural design of pavements for

New roads, Road Note 29, Department of the Environment, H.M.S.O.,
London, 3rd edition.
Rodrigue J., Slack B. and Comtois C. (2013). Transportation modes: An overview. In

Rodrigue J., The geography of transportation systems, 3rd ed., New York:
Routledge.
Rodriguez-Roa, F. (2003). Observed And Predicted Behaviour of Maipo River Sand.

Soils and Foundations, 43(5), pp.1-11.
Saad, B., Mitri, H., and Poorooshasb, H. (2006). 3D FE analysis of flexible pavement
with geosynthetic reinforcement. Journal of transportation Engineering,
132(5), pp.402-415.
Salehabadi, E. G. (2012). The Linear Elastic Analysis of Flexible Pavement by the

Finite Element Method and Theory of Multiple-Layers System. National park-
Forschung in Der Schweiz, 101(9), pp.363-37.
Sas, W., Głuchowski, A., and Szymański, A. (2012). Determination of the resilient
modulus MR for the lime stabilized clay obtained from the repeated loading
CBR tests. Annals of Warsaw University of Life Sciences-SGGW. Land
Reclamation, 44(2), 143-153.
Schwartz, C. W. and Carvalho, R.L. (2007). Implementation of the NCHRP 1-37A

Design Guide Final Report, Volume 2: Evaluation of Mechanistic-Empirical
Design Procedure. MDSHA Project No. SP0077B41
81

References
Seed, H. B., Mitry, F. G., Monismith, C. L. and Chan, C. K. (1967). Prediction of

pavement deflection from laboratory repeated load tests, NCHRP Report 35,
USA.
Senol, A., Bin-Shafique, M., Edil, T. B. and Benson, C. H. (2002). Use of Class C fly
ash for the stabilization of soft soil as sub-base. ARI Bull Istanbul Tech
University, 53, pp.89–95.
Shafabakhsh, G. A., Motamedi, M. and Family, A. (2013a). Influence of Asphalt

Concrete Thickness on Settlement of Flexible Pavements. Electronic Journal
of Geotechnical Engineering, 18, pp.473-483.
Shafabakhsh, G. A., Talebsafa M., Motamedi, M. and Badroodi, S. K. (2013b).

Validation of Load Movement on Flexible Pavement Using Pennsylvania Field
Test. Technical Journal of Engineering and Applied Sciences, 3(22), pp.3125-
3130.
Sherwood, P. (1974). The use of waste and low-grade materials in road construction:
guide to materials available. TRB publication.
Sherwood, P. (1993). Soil stabilization with cement and lime. State of the Art
Review. London: Transport Research Laboratory, HMSO.
Sinha, A. K., Chandra, S. and Kumar, P. (2014). Finite Element Analysis of Flexible
Pavement with Different Subbase Materials. Technical Paper, Indian
Highway. 42(2), pp.53-63.
Snellings, R., Mertens G. and Elsen J. (2012). Supplementary cementitious

materials. Reviews in Mineralogy and Geochemistry, 74, pp.211–278.
South African National Road Agency Ltd (SANRAL). (2013a). Pavement

Composition and Behaviour: South African Pavement Engineering Manual. An
initiative of the South African national roads agency ltd, South Africa.
82

References
South African National Road Agency Ltd (SANRAL). (2013b). Pavement Design:
South African Pavement Engineering Manual. An initiative of the South
African national roads agency ltd, South Africa
Steyn, W., Maina, J. and Repsold, A. (2013). Comparison of isotropic and cross-
anisotropic analysis of pavement structures. Transportation Research Board
92nd Annual Meeting, 25(3), 26.
Stoffels, S., Solaimanian, M., Morian, D., and Soltani, A. (2006). Field
Instrumentation and Testing Data from Pennsylvania's Superpave In-Situ
Stress/Strain Investigation. Airfield and Highway Pavement, 191(10), pp. 107-
118.
Sukumaran, B. (2004). Three Dimensional Finite Element Modeling of Flexible

Pavement. Presented for the 2004 FAA worldwide airport technology transfer
conference, Atlantic City, New Jersey, USA.
Sutanto, M. H. (2009). Assessment of Bond between Asphalt Layers. Ph.D.

Dissertation, Department of Civil Engineering, Nottingham Transportation
Engineering Centre.
Tanosaki, T., Yu, Q. and Nagataki, S. (2011). Are Fly Ashes Green? 2011 World of
Coal Ash (WOCA) Conference, May 9-12, 2011, [Online] Available:
www.flyash.info. Accessed: 20 March 2014.
Theyse H. L., de Beer M. and Rust F. C. (1996). Overview of the South African
Mechanistic Pavement Design Method. Transportation Research Record
1539. Transportation Research Board, Washington D.C.
Theyse, H. L., and Muthen, M. (2000, July). Pavement analysis and design software
(PADS) based on the South African mechanistic-empirical design method. In:
South African Transport Conference (SATC). Action in Transport for the new
Millennium, pp.17-20.
Theyse, H.L., de Beer, M., Maina, J.W. and Kannemeyer, L. (2011). Interim Revision
of the South African Mechanistic-Empirical Pavement Design Method for
83

References
Flexible Pavements. Presented at 10th Conference on Asphalt Pavements for

Southern Africa.
Thompson, M. R. (1966). Shear strength and elastic properties of lime soil mixtures.
Highway Research Board, University of Illinois, Champaign, IL.
Ti, K. S., Huat, B. B., Noorzaei, J., Jaafar, M. S., and Sew, G. S. (2009). A review of
basic soil constitutive models for geotechnical application. Electronic Journal
of Geotechnical Engineering, 16, pp.1-18.
Tiliouine, B. and Sandjak, K. (2014). Numerical simulation of granular materials

behaviour for unbound base layers used in Algerian pavement structures.
International Journal of Civil and Structural Engineering, 4(3), pp.419-429.
Torii, K., Hashimoto, T., Kubo, T. and Sannoh, C. (2013). The effective utilization of
classified fine fly ashes for production of highly-durable concrete mixtures. In:
3rd Int. Conf. on Sustainable Construction Materials & Technologies, SCMT3,
pp.109-118.
TRH 4. (1996). Structural Design of flexible pavements for interurban and rural
roads. Pretoria: Committee of Land Transport Officials (COLTO), Department
of Transport (DoT). (DoT Technical Recommendations for Highways; Draft
TRH 4 (1996)).
Tutumluer, E. and Thompson, M. R. (1997). Anisotropic modelling of Granular Bases

in Flexible Pavements. Transportation Research Record, No. 1577.
Transportation Research Board, National Research Council, Washington, DC.
Umar, M., Alhassan, H., M., Abdulfatah, A.Y. and Idris, A. (2013). Beneficial use of
Class-C Fly Ash in Improving Marginal Lateritic Soils for Road Construction.
Electronic Journal of Geotechnical Engineering, 18, pp.2815-2822.
United State Environmental Protection Agency. Using recycled industrial materials in

roadways [Online]. 2009 Available at:
http://www.epa.gov/epawaste/conserve/imr/pdfs/roadways.pdf [Accessed 5
December 2014].
84

References
United States Department of Transportation - Federal Highway Administration

(USDT-FHA). Geotechnical Aspects of Pavements Reference Manual
[Online]. 2014 Available at:
http://www.fhwa.dot.gov/engineering/geotech/pubs/05037/05b.cfm. [Accessed
3 January 2015].
Uzan, J. (1992). Characterization of Granular Material. Transportation Research

Record 1022, TRB, National Research Council, Washington, D.C., pp.52-59.
Van Vuuren D. J., Otte E. and Paterson W. D. O. (1974). The Structural Design of
Flexible Pavements in South Africa. In: Proceedings of the 2nd Conference on
Asphalt Pavements in South Africa. Durban, South Africa.
Walubita, L.F. & van de Ven, M.F. (2000). Stresses and Strains in Asphalt-Surfacing
Pavements. South African Transport Conference Organised by: Conference
Planners, ‘Action in Transport for the New Millennium’.
Wang, J. (2001). Three-Dimensional Finite Element Analysis of Flexible Pavements.

Department of Civil Engineering, University of Maine.
Wu, Z., Chen, X., and Yang, X. (2011). Finite Element Simulation of Structural
Performance on Flexible Pavements with Stabilized Base/Treated Subbase
Materials under Accelerated Loading (No. FHWA/LA. 10/452). Louisiana
Transportation Research Centre.
Yadu, L.K. and Tripathi, R.K. (2013). Stabilization of Soft Soil with Granulated Blast
Furnace Slag and Fly Ash. International Journal of Research in Engineering
and Technology, 2(2), pp.115-119.
Yagawa, G. (2011). Free Mesh Method: fundamental conception, algorithms and

accuracy study. In: Proceedings of the Japan Academy. Series B, Physical
and biological sciences, 87(4), pp.115.
Yau, A. and Quintus, H. L. V. (2002). Study of LTTP laboratory resilient modulus test
data and response characteristics. Final Report (Publication No. FHWA-RD-
02-051), Maclean VA: Federal Highway Administration.
85

References
Yin, H. (2013). The Impact of Strain Gage Instrumentation on Localized Strain

Responses in Asphalt Concrete Pavements. International Journal of
Pavement Research and Technology, 6(3), pp.225-234.
Zafar R., Nassar, W. and Elbella, A. (2005). Interaction between Pavement

Instrumentation and Hot-Mix-Asphalt in Flexible Pavements. Emirates Journal
for Engineering Research, 10(1), pp.49-55.
Zaghloul, S.M. (1993). Non-linear dynamic analysis of flexible and rigid pavements.
Ph.D. Thesis, Purdue University, West Lafayette. Indiana, USA.
Zaman, M., Pirabarooban, S. and Tarefder, R.A. (2003). Evaluation of Rutting

Potential in Asphalt Mixes Using Finite Element Modelling. Annual
Conference of the Transportation Association of Canada, St. John’s,
Newfoundland and Labrador.
86

Appendices
APPENDICES

Appendices
Appendix A1, B1 and C1 are extract form the Triaxial test results (Heyns and
Mostafa Hassan, 2013).

Appendices

Appendices

Appendices
Appendix A2 – F2 are extract from the mePADS software results showing the
various steps and results.

Simulation by Abaqus FWD

Uploaded by

Copyright:

Available Formats

Simulation by Abaqus FWD

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Simulation by Abaqus FWD

Uploaded by

Copyright:

Available Formats

SIMULATION OF FLEXIBLE PAVEMENT UTILIZING

FLY ASH AS ALTERNATIVE STABILIZER

JACOB ADEDAYO ADEDEJI

Submitted in fulfilment of the requirements for the degree:

MAGISTER TECHNOLOGIAE (M-Tech): Civil Engineering

Department of Civil Engineering

Faculty of Engineering and Information Technology

Central University of Technology

Free State, Bloemfontein

Supervisor: Dr. Mohamed Mostafa

© Central University of Technology, Free State

Declaration and Copyright

© Central University of Technology, Free State

© Central University of Technology, Free State

My deepest gratitude goes to my supervisor, Dr. Mohamed Mostafa. I appreciate you

I would like to show my appreciation to my beloved parents, Mr Isaac Adedeji and

© Central University of Technology, Free State

© Central University of Technology, Free State

® Registered trademark symbol

© Central University of Technology, Free State

ksi kilo-pound per square inch

© Central University of Technology, Free State

List of Tables .............................................................................................................. x

List of Figures .............................................................................................................xi

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

1.1. Background ............................................................................................ 1

1.1.1. Materials .................................................................................................... 1

1.1.2. Design and Analysis .................................................................................. 3

1.2. Problem Statement ................................................................................. 4

1.3. Research Aim ......................................................................................... 4

1.4. Specific Objectives ................................................................................. 4

1.5. Delimitation ............................................................................................. 5

1.6. Significance of the Study ........................................................................ 5

1.7. Outline of Dissertation ............................................................................ 5

CHAPTER 2: LITERATURE REVIEW ........................................................................ 7

2.1. Flexible Pavement ............................................................................................... 7

2.2. Pavement Composition and Behaviour ............................................................... 8

2.3. Secondary Materials............................................................................................ 9

2.3.1. Soil Stabilization Concept ....................................................................... 10

2.3.2. Fly Ash as Stabilizer ............................................................................... 12

2.4. Pavement Design Background .......................................................................... 14

2.5. Summary ........................................................................................................... 17

CHAPTER 3: NUMERICAL SIMULATION OF FLEXIBLE PAVEMENT ................... 18

3.1. Stresses, Strains and Deflections in Flexible Pavement ................................... 18

3.2. Approach of Mechanistic Empirical (M-E) Design ............................................. 20

3.3. Layered Elastic Simulation ................................................................................ 20

© Central University of Technology, Free State

3.4. Finite Element Modelling (FEM) ........................................................................ 23

3.4.1. Critical Factors in FEM Simulation of Pavement ......................................... 24

3.4.1.2. Material Characterization....................................................................... 26

3.4.1.3. Analysis Type, Boundaries and Loading Conditions ............................. 27

3.4.2. Correlation Equations in FEM Simulation .................................................... 28

3.4.3. FEM Material Characterization via Direct Testing Results .......................... 30

3.5. 2D versus 3D Analysis ...................................................................................... 34

3.5.1. Motivation for 3D FEM ............................................................................ 36

3.5.2. Comparative Study on FEM Software ..................................................... 36

3.5.3. Abaqus® as FEM Software..................................................................... 37

3.6. Failure Criteria in Numerical Simulation ............................................................ 38

3.7. Summary ........................................................................................................... 39

CHAPTER 4: DESIGN AND SIMULATION .............................................................. 41