Evaluation AASHTO 2002

http://irc.nrc-cnrc.gc.
ca
Evaluation of the Mechanistic Empirical

Pavement Design Guide (NCHRP 1-37A)
RR-216
Ali, O.
September 2005
Evaluation of the Mechanistic Empirical
Pavement Design Guide
(NCHRP 1- 37 A)
Author: Osman Ali
Report No: UR 3002.1

Report Date: September 2005
Contract No: Research Report 216
Program: Urban Infrastructure
i
EXECUTIVE SUMMARY
Current empirical pavement design approaches are in the process of being replaced
with mechanistic models. This study reviewed such a model entitled “AASHTO 2002
Design Guide” to examine its sensitivity to variables with known influence on the
performance of roads. The study focuses on evaluating the effectiveness of the
manner in which material characteristics were incorporated in the analytical model.
The structure module relies on linear elastic analysis ignoring visco-elastic/plastic

responses. Incorporating nonlinear analysis in the future will improve model
performance prediction capabilities.
The proposed guide is sensitive to asphalt concrete related variables such as mix
stiffness, layer thickness as influenced by traffic characteristics and elements of
the environment.
The guide is less successful in addressing variables that are related to unbound
materials. The study identified a need to introduce nonlinear analytical
capabilities to improve the model sensitivity to variation in the materials response.
Measures, introduced in the proposed guide to circumvent the need for
mechanical testing of materials in lower design levels are not effective. The study
recommends the use of actual laboratory test results or alternatively use generic
properties produced by limited testing of typical local materials.
The study identified the need to revise AASHTO soil classification to make it
more compatible with mechanistic characterization and design.
i
TABLE OF CONTENTS
EXECUTIVE SUMMARY……………………………………………………….………ii
LIST OF TABLES.....................................................................................……………..... v
LIST OF FIGURES......................................................................................………….... vii
CHAPTER 1- INTRODUCTION....................................................................…………....1
1.1 Background………………………………………………………………..1
1.2 Objective………………………………………………….………….........2
1.3 Scope..............................................................................………………......2
1.4 Organization of the Report…………………………………………………2
CHAPTER 2- STATE OF THE ART........................................................................……..4
2.1 Pavement Design and Analysis……………………………….…………...4
2.2 Overview of AASHTO ‘93…………………………………………..........5
2.3 Research Work...............................................…………………………..…9
2.4 Summary……………………………………………………………........18
CHAPTER 3- OVERVIEW of THE PROPOSED AASHTO DESIGN GUIDE………..19
3.1 Introduction………………………………………………………………19
3.2 General Features of the Design Model…………………………………..19
3.3 Reliability…………………………………………………………...........22
3.4 Traffic………………………………………………………………........24
3.5 Environmental Effects…………………………………………………...25
3.6 Performance Prediction Models………………………………………….26
3.7 Comparison Between AASHTO’93 and the Proposed 2002
Guide……………………………………………………………………..32
CHAPTER 4- EVALUATION of THE PROPOSED AASHTO 2002 GUIDE…………34
4.1 Introduction………………………………………………………………34
4.2 Asphalt Concrete Characteristics………………………………………...34
4.3 Unbound Materials……………………………………………………….44
4.4 Traffic Data……………………………………………………………....44
4.5 Climatic Data………………………………………………………….…47
4.6 Sensitivity Analysis……………………………………………………...47
4.7 Model Limitations in Predicting Fatigue Cracking……………………...48
ii
CHAPTER 5- RESULTS AND ANALYSIS ……...……………………………………50
5.1 Introduction………………………………………………….………...…50
5.2 Performance Based Design and Analysis………………………………..50
5.3 Unbound Material Characterization……………………………………..58
5.4 Asphalt Concrete Characterization….…………………………………...63
5.5 Influence of the Environment on Permanent Deformation………………64
5.6 Traffic Characteristics……………………………………………………66
5.7 Construction Related Variables….………………………………………69
5.8 General Comments Related to Model Predictions of Permanent
Deformations……………………………………………………………..72
5.9 Fatigue Cracking………………………………………………………....72
5.10 Comparison Between Design Input Levels……………………………....84
5.11 Measured vs. Predicted Dynamic Modulus…………………………..….86
5.12 Unbound Materials Inputs for Level 1 and 3………………………..…...89
CHAPTER 6- CONCLUSIONS AND RECOMMENDATIONS…………….………...93
REFERENCES..................................................................................................................97
iii
LIST OF TABLES
Table 3.1 Illustrative Levels of Reliability for New and Rehabilitation Design…...24
Table 3.2 Recommendations for selecting vehicle-operating speed……………….25
Table 4.1 Designation of Mixes Used in the Study……………………..………….35
Table 4.2 Aggregate Gradations……………………………………………............36
Table 4.3 Physical Properties of Mix HL3…………………………………………37
Table 4.4 Physical Properties of Mix HL8…………………………………………37
Table 4.5a Dynamic Modulus for HL3-58-22……………………………….………39
Table 4.5b Dynamic Modulus for mix HL3-52-34……………………………...…...39
Table 4.5c Dynamic Modulus for mix HL3-64-34…………………………………..39
Table 4.6 Aggregate Gradation Data Required For Level 2 and 3 Input……….…..41
Table 4.7 Superpave Binder Test Data……………………………………….…….41
Table 4.8 Recommended RTFO A and VTS Parameters Based on Asphalt PG
Grade……………………………………………………………….…….43
Table 4.9 Thermal Properties of Asphalt Concrete…………………………………43
Table 4.10 Properties of Different Granular Materials (A-1-a)……………………...44
Table 4.11 Traffic Data………………………………………………………………45
Table 4.12 Truck Traffic Classification (Default)…………………………………...45
Table 4.13 Traffic General Input………………………………………………….…46
Table 4.14 Number of Axle /Truck………………………………………………….46
Table 4.15 Axle Configuration………………………………………………………46
Table 4.16 Climatic Input……………………………………………………………47
Table 4.17 Drainage Input………………………………………………………...…47
Table 5.1 Predicted Permanent Deformation for Different AC Layer
Thicknesses………………………………………………………………53
Table 5.2 Predicted Permanent deformations for different Base Layer
Thicknesses...…………………………………………………………….55
Table 5.3 Performance Predicted for the Three Design Alternatives…………....…57
Table 5.4 Predicted Permanent Deformations Considering Different
Base Modulus…………………………………………………………….59
iv
Table 5.5 Predicted Permanent Deformations Considering Different
Subgrade Modulus ……………………………………………...……….61
Table 5.6 Predicted Permanent deformations for Different MAAT’s……………...64
Table 5.7 Permanent Deformations at Different Traffic Speeds…………...………66
Table 5.8 Permanent Deformations Predicted for Different Traffic Volumes……..68
Table 5.9 Permanent Deformations Predicted for Different Air Voids Contents…..69
Table 5.10 Permanent Deformations Predicted for Different Binder Contents……...71
Table 5.11 Predicted Fatigue Cracking (Bottom/Up) for Different AC
Thicknesses……………………………………………………………....73
Table 5.12 Predicted Fatigue Cracking (Bottom/Up) for Different Base
Modulus Values………………………………..………………………..76
Table 5.13 Predicted Fatigue Cracking (Bottom/Up) for Different Subgrade
Modulus…………………………………………………………….…...78
Table 5.14 Predicted Fatigue Cracking (Bottom/Up) - Different MAAT……….….79
Table 5.15 Predicted Fatigue Cracking (Bottom/Up) at Different Traffic speeds….80
Table 5.16 Predicted Fatigue Cracking (Bottom/Up) – Varying Traffic Volumes....81
Table 5.17 Effect of AC Mix Air Voids (AS Built) on Fatigue
Cracking (Bottom/Up) ……………………………………………….…82
Table 5.18 Effect of AC Mix Effective Binder Content (As Built) on Fatigue
Cracking (Bottom/Up) ……………………………………………….…83
v
LIST OF FIGURES
Figure 2.1 AASHTO 93 Structural Design Procedures………………………………6

Figure 2.2 Dynamic Shear Rheometer………………………………………………11
Figure 2.3 Master Curve Developed Using Husaroad Program …………………….14
Figure 2.4 Rheological Models Representation……………………………………..14
Figure 2.5 Huet-Sayegh Model………………………..…………………………….15
Figure 2.6 Triaxial Pressure Chamber for determination of MR…………………….17
Figure 2.7 MR Definition…………………………………………………………….17
Figure 3.1 M-E Design Process for Flexible Pavement………………………….….21
Figure 3.2 Design Reliability Concept for a Given Distress………………………...23
Figure 3.3 Enhanced Integrated Climatic Model Components……………………...26
Figure 3.4 Pavement Deformation Behaviour of Pavement Materials…………...…27
Figure 4.1 a Gradation Curve of Mix HL3……………………………………………35
Figure 4.1b Gradation Curve for Mix HL8…………………………………………...35
Figure 4.2 Viscous Response of Asphalt Concrete………………………………….38
Figure 5.1 Reference Pavement Structure………..………………………………….52
Figure 5.2 Permanent Deformation Predictions….………………………………….52
Figure 5.3 Effect of HMA Layer Thickness Changes on Predicted Rutting…….….54
Figure 5.4 Effect of Base Thicknesses on Rutting……………….………………….55
Figure 5.5 Combination 1……………………………..……………………………..56
Figure 5.8 Combination 3 that Satisfies the Design Limit………………………..…58
Figure 5.9 Pavement Structure with Different Granular Base Moduli..…………….59
Figure 5.10 Effect of Base Modulus on Permanent Deformation…………………....60
Figure 5.11 Pavement structure with the Different Subgrade Modulus Investigated...61
Figure 5.12 Effect of Subgrade Modulus on Permanent Deformation………………..62
Figure 5.13 Pavement Structure – Two AC Mixes……………..…………………….63
Figure 5.14 Permanent Deformation of Two HMA Mixes..………………………….63
vi
Figure 5.15 Permanent Deformation Evaluated for Different Mean Annual Air
Temperatures…………………………………………………………….65
Figure 5.16 Effect of Traffic Speed on Permanent Deformation……………………..67
Figure 5.17 Impact of Traffic Volume on Predicted Permanent Deformations………68
Figure 5.18 Pavement Structure for the Air Voids Study…………. .………………..69
Figure 5.19 Impact of Air Voids Content on Permanent Deformation……………….70
Figure 5.20 Effective Binder Content Impact on Permanent Deformation…...………71
Figure 5.21 Effect of AC Thickness Changes on Fatigue Cracking……………….....73
Figure 5.22 Dynamic Modulus of HL3 Prepared with Different Binders...........…......74
Figure 5.23 Fatigue Cracking Predicted for Three Mixes Prepared with Different
Binders…………………...................................................................…....75
Figure 5.24 Pavement Structure with Different Granular Base Modulus………….....76
Figure 5.25 Effect of Base Quality on Fatigue Cracking (Bottom/Up)……...........….77
Figure 5.26 Pavement Structure with Different Subgrade Modulus ……………...….77
Figure 5.27 Effect of Subgrade Modulus on Fatigue Cracking (Bottom/Up)……..….78
Figure 5.28 Fatigue Cracking (Bottom/Up) Predicted for Three Cities with
Different MAAT……………………………………………………..…..79
Figure 5.29 Fatigue Cracking Associated with Different Traffic Speeds ……..........80
Figure 5.30 Fatigue Cracking Predicted for Different Traffic Volumes …...….….....81
Figure 5.31 Effect of Air Voids (AS-Built) on Fatigue Cracking………………….....82
Figure 5.32 Fatigue Cracking for Different Binder contents……………………........83
Figure 5.33 Pavement Structure – Comparison of Design Levels…………………....84
Figure 5.34 Permanent Deformation Prediction Using Level 1 and Level 3………....85
Figure 5.35 Fatigue Cracking Prediction Using Level 1 and Level 3………………...85
Figure 5.36 Measured and Predicted Modulus for Two HL3 Mixes with
Different Binders…….....................................................................…......87
Figure 5.37 Predicted vs. Measured Dynamic Modulus…………………….......…....88
Figure 5.38 Base Rutting for Granular A……………………………………..…........90
Figure 5.39 Base Rutting for Granular B……………………………………....…......91
Figure 5.40 Fatigue Cracking for Granular A……………….……………….......…...92
Figure 5.41 Fatigue Cracking for Granular B………….……………………........…..92
vii
CHAPTER 1
INTRODUCTION
1.1 Background
The last 50 years have seen remarkable increases in the demand for moving people and
goods. The surface transportation system has been under pressure to sustain the
substantial increase in heavy truck traffic volume levels. Considerable annual
expenditures worldwide are allocated for developing the transportation infrastructure,
including urban roads and highways. In Canada, $ 8 Billion dollars is allocated annually
for the rehabilitation of urban roads [1], with overlays being one of the most commonly
used techniques. Durability of the pavement structure determines the initial cost of the
road and its future maintenance requirements. A reliable analytical model that ensures the
structural and functional integrity of the pavement during its design life is needed to
optimize the benefits from the significant funds dedicated to construction and
maintenance of the roadway network.
Currently, structural design approaches to flexible pavement are empirical in nature.
However limitations of the empirical approach are becoming increasingly obvious with
developments in the transportation system and increased knowledge in the fields of
pavement mechanics and material science. Premature failures of asphalt overlays within
few years of construction are not uncommon, and the need for a more comprehensive
mechanistic pavement design model has been recognized. Despite efforts by researchers
in the last decades to enhance the mechanistic part of the design, no fully satisfactory,
comprehensive alternative to the empirical approach has been found [2]. Research
conducted by the Directorate for Transportation of the European Commission, in the late
1990s, to evaluate existing advanced analytical pavement design models, showed that the
output varied considerably from model to model [3]. Now there is a major effort to
develop a new AASHTO Design Guide, [4] with a potential to address the specific
effects of the environment and traffic on the materials that relates its structural response
to a realistic pavement performance.
1.2 Objective
This study evaluates an emerging mechanistic-empirical analytical model the AASHTO
2002 Design Guide, which is intended to replace the current version used by almost all
provinces in Canada (AASHTO 93). The beta version of the software is being evaluated
by potential users in the United States and Canada to determine its effectiveness and the
need for calibration of some of its components. The current version of the software uses
US customary units as the measurement system.
1
1.3 Scope
This study focused on investigating the effectiveness of the material mechanistic
characterization techniques adopted in the new AASHTO 2002 to discriminate between
various material types and to produce the input parameters required by the proposed
mechanistic-empirical model.
• Asphalt concrete (AC) and unbound material samples (crushed stone and clays)
were tested in the laboratory to produce mechanistic properties using the complex
modulus and the resilient modulus test protocols respectively.
• The output of tests performed on different materials were used as input to the
AASHTO 2002 Design Guide to conduct sensitivity analysis. The results were then
used to evaluate the model’s ability to distinguish between materials with a known
performance level. The road structure, environmental conditions and traffic volumes
were kept constant in these model runs to focus analysis on the influence of the
material type on the predicted performance.
• Model runs were performed using different design levels of the AASHTO 2002
Design Guide to determine the degree of precision achieved by using the advance
design level (Level 1) compared with the routine design level (Level 3). This
component of the investigation focused on establishing the benefits gained from
using material models formulated based on results of actual mechanical tests
performed on the pavement materials (Level 1) with those estimated from
physical properties and limited mechanical tests on the binder used in the mix.
The effects of variations in significant independent material variables were quantified by

calculating the predicted distress as each significant variable was changed while all other
variables were held constant. The results were plotted to display the relative significance
of the independent variables.
1.4 Organization of the Report

Chapter 1 introduces the current state of the roadway network and the absence of a sound
analytical approach to pavement design. This chapter also includes the study objective
and scope of the investigation.
Chapter 2 discusses state-of-the art road design techniques adopted in Canada and
highlights the deficiencies in the most commonly used empirically based approach. The
chapter discusses the current material characterization techniques and laboratory testing
required in support of the pavement design task.
Chapter 3 describes the main features of the AASHTO 2002 Design Guide, including the
adopted structural response models, performance models, material characterization and
other input requirements. This chapter also includes critique of the theoretical construct
of the model with emphasis on the empirical nature of the performance models.
1
Chapter 4 discusses the AASHTO 2002 Design Guide evaluation process according to the
terms described in the scope of this study. The input parameters related to the material
characteristics used in the analysis are listed.
Chapter 5 discusses the results and output of model runs to reflect the impact of material
types on the predicted performance. Different asphalt mixes and unbound materials were
used in the analysis to examine the model sensitivity to variation in the mechanistic
response of these materials.
Chapter 6 provides conclusions and recommendation related to the objectives set for this
study.
2
CHAPTER 2
STATE OF THE ART
2.1 Pavement Design and Analysis

Pavement engineers are continually looking for an effective analytical tool to assist in
analyzing pavement structures, taking into consideration the in-service condition of the
road. Such a tool will facilitate the establishment of a performance-based design, capable
of extending the service life of roads. An ideal design tool consists of a structural model
capable of predicting the state of stresses and strains within the pavement structure under
the action of traffic and environmental loading. To carry out such analysis effectively, the
design tool should be equipped with material models capable of capturing the
mechanistic response of the various materials used to construct the road structure. Such a
model is considered a mechanistic model. Because of the knowledge gaps in mechanics
and material science in the early 1960s, the empirical modeling approach evolved as an
alternative. This study examined a typical empirical design model and evaluated a newly
proposed model that represents a step toward mechanistic design.
Empirical Design Approach
Current pavement design procedures are based mainly on empirical approaches. In these
empirical approaches, equations that estimate the response of the road are developed to
guide users to the appropriate design that limits stresses and strains below a critical level.
The design inputs include standardized traffic loading, physical properties of pavement
materials, and environment conditions to estimate the pavement performance based on
results from previous field experiments (AASHTO road test of 1960s) and in some cases,
on practical experience. A number of empirical design methods are in use today, such as
the Asphalt Institute Thickness Design [5], the National Crushed Stone Association [6],
the California Method and the most commonly used pavement design method the
AASHTO design guides [7, 8, 9, 10].
Mechanistic-Empirical Design Approach
In the mechanistic approach, the relationship between the structural response (stresses,
strains and deflection) and the physical causes (inputs) are described using a
mathematical model. The most common models are based on the multilayered elastic
theory and some finite element model applications. Since these theoretical approaches are
still not capable of fully analyzing pavements, mechanistic-empirical design approaches
evolved with empirical elements added to complement the mechanistic models. These
empirical elements are usually included, as performance prediction models since it
remains difficult to quantify permanent deformation using the linear elastic theory. The
performance prediction models use the calculated stresses, strains and deflections to
predict pavement distress using the adopted empirical formulas.
3
2.2 Overview of AASHTO ‘93
The AASHO road test conducted near Ottawa, Illinois, during the period 1958-62 was
the major highway design/traffic field experiment, which became the basis for the design
guides used in North America. The original design procedure was modified several times
over the last 40 years in an endeavour to improve the procedure and to incorporate new
research findings. For example, the reliability concept was introduced with the release of
the 1986 Guide to allow for uncertainties associated with design variables. The 1993
Guide, the current design guide used by the majority of Canadian provincial road
authorities [11], was an attempt to improve design procedures particularly for the
rehabilitation of existing pavements. Some of the new developments introduced in the
1993 Guide included the removal of the pavement support value and replacing it with the
resilient modulus MR (elastic response). Drainage coefficients were added and the
regional factor (R) abandoned.
Main features of the ASSHTO 93 structural design procedure are captured in Figure 2.1.
The basic equation of the guide for flexible pavement design is shown below as Equation
2.1.
⎛ ΔPSI ⎞
log10 ⎜ ⎟
⎝ 4.2 − 1.5 ⎠
log10 (W18 ) = Z R × S o + 9.36 × log10 (SN + 1) − 0.20 + + 2.32 × log10 (M R ) − 8.07
1094
0.40 +
(SN + 1)5.19
(2.1)
where: W18 = predicted number of 80 kN (18,000 lb.) equivalent

standard axle load (ESALs)
ZR = standard normal deviate
So = combined standard error of the traffic prediction and

performance prediction
SN = structural number (an index that is indicative of the

total pavement thickness required)
= a1D1 + a2D2m2 + a3D3m3+...
ai = ith layer coefficient
Di = ith layer thickness (inches)
mi = ith layer drainage coefficient
∆PSI = difference between the initial design serviceability

index, po, and the design terminal serviceability index,
pt
MR = subgrade resilient modulus (in psi)
1
Determine Serviceability Design Traffic
ESALs/ Lane for the
Loss ΔPSI = P0 - Pt design period
P0 = initial serviceability
Pt = terminal serviceability Materials:
Reliability: Overall • Resilient Modulus to

reliability to allow for characterize subgrade
chance variation in traffic AASHTO 93 Pavement
strength.
and performance Structural Design
• Adjust MR for seasonal
• Solve empirical equation variation
for structural number SN • Assign for each layer a
Standard Deviation as (or use charts) structural layer Coeff.
developed in Road Test • Use layer coeffs. to (The empirical
equal 0.45 generate different layer relationship between
thicknesses solutions SN and the thickness)
Adjust MR. for Climate

Effects
Economic Evaluation Drainage Coefficients
Final pavement
structural design
Figure 2.1: AASHTO 93 Structural Design Procedures
Traffic
The current AASHTO 93 is based on the 18-kip single standard axle. Different axle types
are converted to ESALs using the load equivalency factors (LEF), defined as the number
of repetitions of the 18-kip single standard axle that causes the same damage as a single
application of a particular configuration. The concept of the relative damage is not
quantifiable and is based on empirical results of the ASSHTO Road Test. One main
reason that makes the empirical procedure outmoded is the development that took place
in traffic loading conditions. Today, traffic volumes and vehicle characteristics have
changed significantly. There are a variety of multiple axle configurations; tire types and
pressures have increased substantially. The trucking industry has succeeded in increasing
2
legal truck loads. Weigh-in-motion (WIM) stations have provided detailed traffic data
including traffic counts, actual truck loadings, truck classifications and distribution
during the day and the whole year. All the above-mentioned developments have far
exceeded the data for which the empirical design was thought to cover. Although it does
influence pavement loading, the current AASHTO design practice does not account for
vehicle speed. Slower speeds and stop conditions for a longer period of time result in
greater rutting damage. Because of an inadequate mix or structural design, the rutting
problem becomes evident at bus stops and intersection approaches.
Reliability
Reliability is introduced in the guide as a means of incorporating some degree of
certainty into the design process to ensure that the various design alternatives will last for
the analysis period. Design performance reliability is controlled through the use of a
reliability factor FR that is multiplied by the design period traffic prediction (w18) to
produce design applications (W18) for the design equation. The reliability factor is a
function of the overall standard deviation (S0) that accounts for both chance variation in
the traffic prediction and normal variation in pavement performance prediction for a
given W18. Suggested levels of reliability in the Guide are associated with the functional
classification of the road and whether it is urban or rural.
Materials Characterization
With the exception of the native soil properties, the 93 Guide continued to rely on
physical properties to characterize other materials used in the pavement structure.
Roadbed Soil Characterization
Characterization of subgrade soils in the AASHTO 93 is based on the resilient modulus
MR parameter determined using AASHTO Test Method T274. This represents the first
attempt at implementing a mechanistic property in pavement design. The original
AASHTO MR test procedure has been modified (SHRP – P-46) to the current procedure
entitled “AASHTO T294-92”.
The resilient modulus is a measure of the elastic property of soil recognizing certain
nonlinear characteristics. It should be noted that the AASHTO Guide empirical model
(Equation 2.1) is sensitive to variations in the MR parameter. The AASHTO Guide
(AASHTO 93) recommends that the resilient modulus test be performed on laboratory
samples using relevant stress levels and moisture conditions simulating the primary
moisture seasons. When test facilities are unavailable for performing the test, estimation
of the resilient modulus can be made from standard CBR, R-value and soil index test
results using correlation relationships recommended by the Guide. It provides guidelines
for two procedures to be followed for determining the seasonal variation of the modulus
based either on the laboratory relationship between the resilient modulus and moisture
content, or from backcalculated moduli for different seasons. The Guide includes a
procedure for determining the effective roadbed soil resilient modulus based on an
estimate of relative damage (uf) that corresponds to the seasonally adjusted subgrade
modulus for each month of the year.
3
Pavement Layer Material Characterization
In the current empirical design procedure (AASHTO 93), pavement material
characteristics are not directly incorporated in the structural design. The layer coefficients
representing the various unbound layer materials are related to the resilient modulus
determined using AASHTO T274 test methods. Alternatively the bound or asphalt
concrete may be characterized using results of the repeated load indirect tensile test
(ASTM D 4123). The Guide also provides charts for estimating these coefficients based
on physical properties. The layer coefficient, which is a measure of the relative ability of
the material to function as a structural component of the pavement, expresses the
empirical relationship between the structural number (SN) and thickness [10].
SN = a1D1 + a2D2m2 + …… + aiDimi 2.2

Where:
SN = Structural number of the pavement structure
ai = Layer coefficient i.
Di = Thickness of layer i
m = Drainage factor for layer I
It was found that many factors could influence the layer coefficients such as the thickness
of the layer and the underlying support and position within the pavement structure. The
Guide noted that the laboratory resilient modulus values were significantly different from
what may exist for an in situ condition. In practice, experience play a major role in
determining the layer coefficients, and long-term pavement response data are considered
the most reliable method for determining the structural layer coefficients.
Critique
The empirical equations developed for the specific conditions of the AASHTO Road Test
hindered advancement of the earlier developed design approach. Limitations of the
approach were recognized, and researchers noted that it is not realistic to use empirically
derived relationships to describe phenomenon that occur outside the range of the original
data used to develop the empirical relationships. Accordingly, applications of the
AASHTO 93 Guide [10] involved significant limitations including the following.
• Since the adopted empirical equations were developed based on only one type of
pavement material and roadbed soil at the AASHTO Road Test, extending
analysis beyond those materials may result in design errors.
• Limitations associated with the environment at which the AASHTO Road Test
was conducted render the results of the design procedure in uniquely different
environmental conditions less accurate.
• There are limitations associated with the age at which loading tests were
performed. An accelerated two-year testing period was used to develop the test
results. The Guide did not consider aging effects, and it is unrealistic to
extrapolate the results for a typical 20-year pavement life.
• The maximum load used to develop the equations was approximately one million.
The vehicles used in the test had identical axle loads and configurations, as
opposed to mixed traffic.
4
Pavement performance is categorized according to the structural performance and the
functional performance. The structural performance relates to the physical conditions of
the pavement, such as the occurrence of cracking or any other conditions that might
affect the load-carrying capacity of the road. On the other hand,0 the functional
performance describes how well the pavement serves the user. It is concerned with the
riding comfort and is quantified using the serviceability-performance concept.
In the current AASHTO 93 system, the loss of serviceability, the difference between the
initial serviceability and terminal serviceability is the only performance indicator. One
major deficiency of the empirical approach is that it proved to be inadequate in evaluating
the performance of the pavements. There is no provision to predict other modes of
failure, such as fatigue, rutting and thermal cracking. In fact, these types of failures, in
many cases dictate the scheduling of maintenance and planning of rehabilitation.
The empirical nature of the design approach does not account for the mechanistic
response of asphalt concrete. The mix design procedure relies on physical properties,
which cannot be linked to the pavement structural response. Road agencies select a mix
type based solely on experience without direct reference to the mix’s ability to withstand
relevant vehicle load characteristics. The long-term effects of climate on material
properties were not considered. In the end, the design process yields a set of layer
combinations that satisfy the structural number with the selection of the best solution
based mainly on economical analysis.
2.3 Research Work

The limitations of AASHTO road design guides, evolving from the early road test,
motivated researchers to pursue improvements in the design process. These
improvements benefited from advancements in mechanics and material science leaning
toward the development of a mechanistic based design approach.
Materials Characterization
Mechanistic pavement design procedures are complex and involve the interaction of the
pavement structure model that estimates the state of stresses and strains with the imposed
traffic loads and environmental conditions. The state of stresses and strains changes in
response to traffic and seasonal variations in the environment. The state of the different
materials forming the road layers and, consequently, the response of the structure,
changes with variation in temperature and moisture condition. An effective analytical
model should account for all of these factors in analysis leading to a performance-based
design. Material characterization models interact with the environmental models. The
structural response model should then use material properties to estimate the stresses and
strains within the pavement structure. The shift to a mechanistic design approach
necessitates the development of procedures to define the material characteristics
necessary for the implementation of the proposed analytical model and to establish
performance-based specifications for road design and construction.
The following is a review of recent developments in material characterization techniques.
5
Binder Characterization:
The Strategic Highway Research Program (SHRP), completed in the late 1990’s, was one
of the largest research projects conducted after the AASHTO Road Tests. One of its
major deliverables was the SuperPave system, which included new binder specifications
that replaced the penetration grade and a mix design procedure to replace the Marshall
approach.
In the past, binders were characterized according to their physical properties, such as
penetration, ductility and viscosity. The tests associated with these specifications are
empirical and do not provide a comprehensive interpretation of the test results. However
the conventional tests do not allow for cold temperature effects. SuperPave uses a binder
selection process based on binder performance grading (PG) system and its associated
specifications. New binder tests and classification system are designated for expected
high and low in-service temperatures. The high pavement temperature is based on the
seven-day average high air temperature of the surrounding area while the low pavement
temperature is based on the one-day low air temperature of the surrounding area. Using a
reliability concept the appropriate PG asphalt binder is determined. SuperPave allows for
high pavement temperature adjustment for slow transient loads and stationary loads.
Additionally, if the anticipated 20-year loading is in excess of 30 million ESALs, the high
temperature grade is increased by one grade.
SuperPave has adopted a new set of tests that can be related to pavement performance.
Rolling Thin-Film Oven (RTFO) Test
The test simulates short-term aging by heating a moving film of asphalt binder in an oven
for 85 minutes at 163° C. The quantity of volatiles lost from the binder is an indication of
the aging that may occur in the asphalt during mixing and construction time. The
AASHTO T240 test, “Effects of Heat and Air on a Moving Film of Asphalt”, and the
ASTM D 2872 test, “Rolling Thin-Film Oven Test”, were adopted to account for the
impact of short-term binder aging on the response of asphalt concrete layers.
Pressure Aging Vessel (PAV)
The SuperPave pressure aging vessel test simulates the effects of long-term asphalt
binder aging that occur during the in-service time of pavement (AASHTO PP1: Practice
for Accelerated Aging of Asphalt Binder Using a Pressurized Aging Vessel).
Dynamic Shear Rheometer (DSR)
This test is used in the SuperPave system for estimating medium to high temperature
viscosities of binders. The test is performed between 46°C (115° F) and 82° C (180° F).
The test simulates the shearing action of traffic at a certain speed and determines two
important parameters used to predict the performance of the hot mix asphalt concrete
(HMA). The first is the complex shear modulus (G*) and the second is the phase angle
(δ), which represent the time lag between the maximum applied shear stress and the
maximum resulting shear strain. By obtaining the values of G* and δ, it is possible to
determine the elastic and viscous components of the total complex shear modulus. These
binder properties help in predicting the performance of the HMA and specify the
minimum value for the elastic component if rutting is of primary concern. On the other
hand, it is possible to specify the maximum value of the viscous component when fatigue
1
is under consideration (see AASHTO TP5 test “Determining the Rheological Properties
of Asphalt Binder Using a Dynamic Shear Rheometer”).
Binder Oscillating Plate

Sample
Fixed
Base Plate
Figure 2.2 Dynamic Shear Rheometer
Bending Beam Rheometer (BBR)

This test measures binder properties at low temperatures (AASHTO TP1: Method for
Determining the Flexural Creep Stiffness of Asphalt Binder Using the Bending Beam
Rheometer). A simple asphalt beam is subjected to a 980 μN load over 240 seconds to
calculate the beam creep stiffness S(t) and the rate of change of that stiffness known
as the m-value. The creep stiffness is related to thermal stresses while the m-value
indicates how the binder relaxes under load. Cold region specifications may call for a
maximum limit on creep stiffness and a minimum limit on the m-value to minimize
the potential for thermal cracking.
Direct Tension Tester (DDT)
The DTT test measures the amount of binder strain before failure at low temperatures.
The test is implemented for testing asphalt binders with high BBR creep stiffness in
excess of 300 MPa (see AASHTO TP 3: Method for Determining the Fracture
Properties of Asphalt Binder in Direct Tension). SuperPave specifications call for a
minimum failure strain for asphalt binders that have a high BBR creep stiffness to
ensure that they will stretch rather than break.
Although the SuperPave goal was to use the above mentioned tests to quantify asphalt
performance during mixing, construction and in-service, these tests received little
attention by the industry. However, the performance-based binder specifications (PG)
was the most popular product of SuperPave and gained wide acceptance and
application.
2
SuperPave Mix Design Method
The superPave mix design method targets the establishment of sustainable aggregate
gradation and performance-based binder specifications [12] to produce a durable HMA
product. The method was designed to replace the conventional Marshal and Haveem
methods. The Marshal method developed in the late 1930s depends on a set of physical
characteristics of the mix to arrive at the intended material quality during construction
[13]. However, there is a limitation associated with the Marshal tests (Stability and
Flow), because of their empirical nature.
SuperPave specifies aggregate by placing restrictions on aggregate gradation using broad
control points. In addition, it applies consensus requirements on coarse and fine aggregate
angularity, flat and elongated particles, and clay content.
The SuperPave approach relies on using the gyratory compactor for preparing laboratory
samples. It was claimed that the gyratory compactor simulates closely actual field
compaction.
Density and voids analysis are similar to the conventional mix design methods. The
optimum asphalt binder content is selected as the asphalt binder content that results in
four percent air voids at the design number of gyrations required to produce a sample
with the same density as that expected in the field after the indicated amount of traffic.
SuperPave mix design procedure adopted the modified Lottman test, “AASHTO T 283:
Resistance of Compacted Bituminous Mixture to Moisture Induced Damage” for
moisture susceptibility testing.
Characterization of Asphalt Concrete Mix

Linear viscoelasticity
One principal feature of elastic behaviour is the capacity for materials to store mechanical
energy when deformed by loading, and to release this energy totally on removal of loads.
Conversely, in viscous flow, mechanical energy is continuously dissipated. A number of
engineering materials, such as HMA, simultaneously store and dissipate mechanical
energy in varying degrees during a loading and unloading cycle. This behaviour is
referred to as viscoelastic [14]. Two main ways to characterize this property in
bituminous mixes are the complex modulus and the creep compliance.
The Dynamic Modulus

The bituminous mixture behaviour, similar to its binder, depends on temperature and load
frequency. Viscoelastic properties of asphalt materials are generally determined within
the linear viscoelastic domain where the material is subjected to sinusoidal loading at
different frequencies at small values of strain to keep it within the linear zone.
Experimentally, loading can be applied under either a stress or strain controlled mode.
The complex modulus (E*) defines the relationship between the stress and strain for a
linear viscoelastic material subjected to sinusoidal loading. The complex modulus
contains both real and imaginary components as shown in equation 2.3. The absolute
value of the complex modulus |E*| is defined as the dynamic modulus.
E* = |E*|cosφ + |E*|sinφ (2.3)
3
The phase angle φ is an indication of the elastic-viscous properties of the mix or binder
material. For a perfect elastic material φ is equal to zero and for a perfect viscous material
φ is equal to 90º.
The complex modulus test is conducted at different temperatures (usually –10, 0, 10, 20,
30 and 40 °C) over a range of frequencies say (0.1, 0.3, 1, 5, 10, 20 Hz). Isothermal
curves can be displayed by plotting the dynamic modulus |E*| as a function of reduced
frequency for each temperature tested in a bi-logarithmic scale.
Master curves may be constructed based on the time-temperature superposition properties
of asphalt mixes. Different combinations of temperature and frequency can produce the
same dynamic modulus. The principle of time-temperature superposition makes it
possible to merge the isothermal curves into a single curve at a reference temperature by
shifting those curves horizontally relative to that reference temperature.
The resulting master curve of the modulus, as a function of time, describes the time
dependency of the material. The amount of shift at each temperature required to form the
master curve describes the temperature dependency of the material. In general, the
dynamic modulus master curve can be mathematically modeled by a sigmoidal function
described by Equation 2.2.
α
log(E*) = δ + (2.4)
1 + e β + γ (log(tr)
Where:
tr = time of loading at the reference temperature.
δ, α = fitting parameters; for a given set of data, δ represents the minimum value
of E* and δ+α represents the maximum value of E* and δ+α represents
the maximum value of E*.
β, γ = parameters describing the shape of the sigmoidal function.
The fitting parameters δ and α depend on the physical properties of the asphalt mixes [4].
The model developed by Witczak for predicting the dynamic modulus of HMA has a
sigmoidal form similar to the general relation suggested for the master curve in Equation
(2.4) [15]. The predictive model is intended to provide an approximation for the dynamic
modulus in the absence of conducting the test protocol.
4
IE*I [MPa]
40000
-10 0
20 30
30000 40 Master Curve
20000
10000
0
0.0001 0.01 1 100 10000 1000000
Reduced freq. [Hz]
Figure 2.3 Master Curve Developed Using Husaroad Program [16]
Rheological Models
Rheological models are used to capture the material’s response to an imposed load. This
response is generally used as an input to performance models that predict various types of
distress. Many rheological models were developed to describe the mechanistic response
of bituminous materials such as the Generalized, Burgers, and Huet-Sayegh (see Figure
2.4). The rheological models are composed of various combinations of springs and
dashpots to describe the viscoelastic behaviour of materials.
Figure 2.4 Rheological Models Representation
5
The Huet-Sayegh model [17, 18] has two branches as shown in Figure 2.5. One branch
consists of two parabolic dashpots, k and h, associated in series with a spring E∞ - E0. The
other branch consists of a single spring E0. The model can be described mathematically
using the formula in Equation 2.5.
E∞ – E0
E0
k
h
Figure 2.5 Huet-Sayegh Model
E*(iω) = E0 + (E∞ - E0 ) /1+ δ (iωτ)-k + (iωτ)-h (2.5)
E0 is the high temperature stiffness while E∞ represents the purely elastic modulus; δ, k
and h are parameters of the biparabolic elements of the model. The parameter τ is a
temperature dependent variable and is referred to as the characteristics time. The dynamic
modulus at any given temperature and frequency can be obtained by fitting the
rheological model into the experimental data.
Creep Compliance
The creep compliance is another concept used to characterize the viscoelastic behaviour
of bituminous mix. If a constant stress is applied, the strain will increase with time. When
this strain is normalized and formulated as a function of time, it is called creep
compliance as defined by Equation (2.6)
D (t) = ε(t) / σ (2.6)
Where
6
σ = constant applied stress.
ε(t) = increasing strain with time
The creep compliance must be defined for a number of temperatures. In the laboratory,
tensile creep is determined by applying a static load of fixed magnitude along the
diametral axis of a specimen. The horizontal and vertical deformations measured near
the center of the specimen are used to calculate tensile creep compliance as a function of
time (see AASHTO TP9-96 Method for Determining the Creep Compliance and Strength
of Hot Mix Asphalt (HMA) Using the Indirect Tensile Test Device). Loads are selected
to keep horizontal strains in the linear viscoelastic range. The Poisson ratio can be
determined from the test. The tensile strength is determined immediately after
determining the tensile creep or separately by applying a constant rate of vertical
deformation to failure.
Tensile creep and tensile strength test data are required to determine the master relaxation
modulus curve and fracture parameters. This information is used to calculate fatigue
performance and thermal cracking of the HMA. The master relaxation modulus curve
controls thermal crack development while the fracture parameter defines a mixture’s
resistance to fracture [19].
Characterization of Unbound Materials

The mechanical response of naturally occurring soils, such as sand, gravel and clays, can
be influenced by a variety of factors. These include:
the shape, size and mechanical properties of the individual soil particles;
the configuration of the soil structure;
the intergranular stresses and stress history; and
the presence of soil moisture, the degree of saturation and the soil
permeability.
These factors generally contribute to stress-strain phenomena, which display markedly
non-linear, irreversible and time-dependent characteristics, and to soil masses, which
exhibit anisotropic and non-homogeneous material properties [20]. The non-linear
behaviour of a soil is important for the analytical design and evaluation of pavements.
The resilient modulus, MR, is commonly used in pavement engineering as an appropriate
measure of stiffness for the unbound layers.
The selection of an appropriate design MR value for base, subbase and/or subgrade
materials has long been complicated by various test and analysis problems [21]. Since the
MR was introduced in the AASHTO 1986 edition, a set of test protocols was developed
and proposed. They include AASHTO T 292-91, AASHTO T 294-92, AASHTO T P46-
94, NCHRP 1-28 Draft-96, and the Harmonized Resilient Modulus Test Method for
Unbound Pavement Materials NCHRP 1-28A. The main differences between these
protocols are in the material type definition, specimen size, compaction method, loading
time, stress sequence and the model predictive equation.
7
Hydraulic actuator
Chamber top
plate
Load cell
Cap
LVDT assembly
Corundum stone
Tie rod
External wall of tri-axial
-
chamber Tests specimen
Membrane
To pore—water High air entry disc
Pressure control (1/4 in. thick, 5 bar)
And volume
Chang indicator Bottom Plate
Pore- air pressure
-
Pore-water Chamber base plate Sensors lead wires
Confining Pressure
Pressure line
transducer
To flush
Figure 2.6 Triaxial Pressure Chamber for Determination of MR
Figure 2.7 MR Definition.
Many models were suggested to establish the stress-strain relationship. Course-grained

soils exhibit an increase in stiffness with increasing confining stress and are usually
represented with a power-law model of the form:
8
MR = k1 pa (θ/Pa)k2 (2.7)
Where:
θ = Bulk stress (σ1 + σ2+ σ3),
k1, k2 = material parameters, with k1> 0 and k2 ≥ 0
Pa = atmospheric pressure.
For fine-grained soils, an alternative equation representing the decrease in stiffness with
the increase of deviator stress (σd) has the form:
MR = k1 pa (σd/Pa)k3 (2.8)
Here, the parameters k1> 0 and k3 ≤ 0.
Witzak and Uzan proposed a universal model [22] that combined both the effects into a
single equation. The model later modified and adopted in the NCHRP Project 1-37A and
has the following form:
k2 k3
⎛ θ ⎞ ⎛τ ⎞
M r = k1 p a ⎜⎜ ⎟⎟ ⎜⎜ oct + 1⎟⎟
⎝ pa ⎠ ⎝ pa ⎠
(2.9)
Where:
τoct is the octahedral shear stress.
2.4 Summary
Review of current pavement design practices in this study revealed that the AASHTO 93
Guide is relied upon across Canada. Deficiencies of the design process were identified,
which were mainly related to limitations in the AASHO Road test performed in the
1960s. The limitations included dependency on information collected for a single native
soil and road construction material within unique environmental and traffic conditions.
Advancements made in engineering mechanics and analytical modeling encouraged
researchers to explore the potential for mechanistic design methods. Material
characterization techniques were investigated to produce mechanistic properties suitable
for the new trend in modeling. Encouraged by these developments, the road community
is now prepared for the mechanistic design approach and the AASHTO 2002 Guide
represents the first attempt toward that goal. The next chapter discusses the AASHTO
2002 Guide to reflect improvements achieved with the new approach to pavement design.
9
CHAPTER 3
OVERVIEW OF THE PROPOSED AASHTO DESIGN GUIDE
3.1 Introduction
The new guide was developed through a National Cooperative Highway Research
Program initiative (NCHRP Project 1-37A). The objective of the new Design Guide was
to provide the highway community with a state-of-the practice tool for the design of new
and rehabilitated pavement structures, based on mechanistic-empirical principles.
This chapter includes a brief overview of the theoretical construct of the mechanistic
empirical model and its applications. General features of the design procedures and the
interaction between materials and of environmental elements are discussed. The chapter
also covers the two main features of the analytical model; the structural response model
and the distress response models.
3.2 General Features of the Design Model

The design process adopted in the new guide is an iterative process using computational
software and a set of documentation. Three design levels were incorporated in the model
for users to select according to their needs. Main features of the design process are
illustrated in Figure 3.1.
Design Input Levels

The guide provides three levels of input depending on the criticality of the project, the
sensitivity of the pavement performance to a given input, the resources available to the
designer, and the availability of input information at the time of the design. These three
levels are categorized as follows.
Level 1
Site and/or material-specific inputs for the project are to be obtained through direct
testing or measurements. This level of input uses the state of the art techniques for
characterization of the materials, such as the dynamic modulus of HMA, as well as
characterization of traffic through collection of data from (WIM) stations.
Level 2
This level uses correlations to determine the required inputs. For example, the dynamic
modulus could be estimated based on results of tests performed on binders, aggregate
gradation and mix properties. The level of accuracy for this category is considered as
intermediate.
Level 3
This level produces the lowest accuracy. Inputs are typically user selected from national
or regional default values, such as characterizing the HMA using its physical properties
(gradation) and type of binder used.
10
Processing of Input Over Design Analysis Period
The design period is divided into incremental analysis periods typically by the month or
for two weeks for frost conditions. The design inputs are processed to obtain seasonal
values of the traffic, material and climatic inputs needed for each analysis increment. The
traffic level and the layer moduli are computed for each increment period. Temperature
and moisture profiles in pavement layers are obtained using the Enhanced Integrated
Climatic Model (EICM) built in the software. The asphalt global aging model modifies
the hot-mix asphalt properties for long-term aging so the dynamic modulus of the asphalt
mix is increasing over time.
Pavement Structural Response Models

Two flexible pavement analysis methods have been incorporated in the Design Guide.
1. Multilayer linear elastic analysis method

The assumptions associated with this method are that each layer is homogeneous, has
finite thickness except for the subgrade, is isotropic, full friction is developed
between layers at each interface, and there are no surface shearing forces. The elastic
modulus and Poisson’s ratio of each layer are essential for the stress solution of the
problem.
2. Finite element method
The finite element module is initiated when the unbound material’s non-linear
behaviour is considered and the coefficient and exponents of the resilient modulus
prediction model are entered for level 1 input (of unbound layers).
Critical stress and/or strain values computed by the structural model include:
• tensile horizontal strain at the bottom/top of the HMA layer to predict fatigue
cracking;
• compressive vertical stresses/strains within the HMA layer for prediction of HMA
rutting;
• compressive vertical stresses/strains within the base/subbase layers for prediction of
rutting of unbound layers; and
• compressive vertical stresses/strains at the top of the subgrade for prediction of
subgrade rutting.
11
• Establish pavement Selection of Trial Design
performance criteria
e.g. rutting, thermal Structure
cracking, smoothness. Traffic Climate Materials Properties
• Determine a level of & Drainage
design reliability for
each applicable
performance
indicator. • Design period is divided into incremental analysis
periods typically two weeks or a month).
• Process input to obtain values of traffic and seasonal
variation of materials and climatic inputs for each of the
specified incremental periods.
Compute structural responses using multilayer elastic theory

or finite element for each load and for each damage-
calculation increment.
Calculate accumulated distress and or damage at the end of

each analysis period for the entire design period.
Use calibrated M-E performance models

to predict key distress and smoothness.
Performance Verification
Failure Criteria
Strategy Selection
• Study viable alternatives Does design
• Constructability issues Yes meet No
• Life cycle cost analysis criteria?
• Select strategy
Figure 3.1 M-E Design Process for Flexible Pavement
12
Distress Prediction
The critical stress and/or strain values obtained from the structural response model are
converted to incremental distresses, either in absolute terms, such as in rut depth
calculation or in terms of a damage index in fatigue cracking. Distress models, sometimes
called transfer functions, relate the calculated damage to field observed distresses. The
cumulative damage is converted to physical cracking using calibrated distress prediction
models and the output at the end of each analysis period is tabulated by the Design Guide
software and plotted for each distress type.
The structural distresses considered in flexible pavement design and analysis include:
• bottom-up fatigue cracking (alligator);
• surface-down fatigue cracking (longitudinal);
• fatigue in chemically stabilized layers (in semi-rigid pavements);
• thermal cracking; and
• permanent deformation (rutting).
Smoothness
All types of structural distresses mentioned above contribute to the loss of pavement
smoothness. Users usually rate the road performance by its roughness. The smoothness is
defined as the variation of surface elevation that induces vibrations in traversing vehicles.
The new guide adopted the international roughness index (IRI) as a measure for
smoothness. In addition to the structural distresses, the performance criteria for
smoothness is achieved through specifying a terminal IRI at a defined level of design
reliability. The smoothness model adds changes to the initial smoothness (IRI) over the
design period. These changes are due to the increase in individual distress, site conditions
and maintenance activities.
3.3 Reliability
The design of flexible pavements is associated with many factors that introduce a
substantial measure of variability. These factors include traffic levels, material properties
and construction quality, in addition to model prediction errors and calibration
measurement errors. There are two methods of pavement design to cater to these
uncertainties: deterministic and probabilistic. In the deterministic method, each design
factor has a fixed value based on the factor of safety assigned by the designer. In the
probabilistic method, each design factor is assigned a mean and a variance.
13
Figure 3.2 Design Reliability Concept for a Given Distress
(Source: Design Guide Documentation- Part1. Introduction)
Reliability is defined as the probability that each of the key distress types and smoothness
levels will be less than a selected critical level over the design period. For each trial
design, the software provides a prediction based on mean or average values for all inputs
that correspond to 50% reliability. The designer usually specifies a higher probability that
the design will meet the performance criteria over the design life. The distribution of the
error term for a given distress about the mean expected prediction is a function of the
many sources of variation and uncertainty mentioned earlier. Distress and IRI are
approximately normally distributed. The standard deviation for each distress type was
determined from the model prediction error from calibration results. The standard
deviation of IRI was determined using a closed form variance model estimation approach
while the standard deviation of the distribution of distress was determined as a function
of the predicted cracking. The reliability of the design can be calculated using the mean
and standard deviation of a normal distribution in two steps
Predict the cracking level over the design period using mean inputs to the
model.
Estimate cracking at the desired reliability level using the following
relationship:
Crack _ P = Crack_ mean + STDmeas * Zp (3.1)
Where:
Crack_ P = cracking level corresponding to the reliability level p
Crack_ mean = cracking predicted using the deterministic model with mean inputs
14
STDmeas = standard deviation of cracking corresponding to cracking predicted
using the deterministic model with mean inputs
Zp = standard normal deviate corresponding to reliability level p
Recommended levels of reliability for each distress are based on the functional class of
the roadway as shown in Table 3.1.
Table 3.1 Illustrative Levels of Reliability for New and Rehabilitation Design
Functional Recommended Level of Reliability

Classification
Urban Rural
Interstate/Freeway 85 – 97 80 – 95
Principal Arterials 80 – 95 75 – 90
Collectors 75 – 85 70 – 80
Local 50 – 75 50 – 75
Source: Design Guide Documentation- Part1. Introduction.
3.4 Traffic
The traffic characterization method proposed in the new guide is more precise than the
conventional ESALs technique adopted in previous guides. The load spectra for single,
tandem, tridem and quad axles are introduced. Traffic data, including truck count by
class, by direction and lane, are required for traffic characterization. Axle load spectra
distributions are developed for each vehicle class from axle weight data. Traffic volumes
by vehicle class are forecasted for the design analysis period. The traffic module
determines the total number of axle applications for each axle type and load group over
the design period. The number of applications for each axle type and load increment is
then used in the computation of pavement responses, damage and distress prediction.
Additional data related to the axle configuration, such as average axle width, dual tire
spacing, tire pressure and axle spacing are used in the pavement response module.
Traffic wander influences the number of load applications over a point. This parameter
affects prediction of fatigue and permanent deformation. Wander is assumed normally
distributed and the standard deviation is used as a measure for it.
An important feature introduced in the guide is the provision for the special axle
configuration. This option allows the designer to analyze pavement performance due to
special, heavy, non-conventional off-road vehicle systems. Another important input for
the flexible pavement design is the vehicle operational speed. It directly influences the
stiffness response of the visco-elastic asphalt concrete layers. The magnitude and
15
duration of stress pulses caused by the moving traffic depend on the vehicle speed, type
and geometry of pavement structure, and the location of the element under consideration.
The frequencies corresponding to various speeds are used to calculate the dynamic
modulus of the asphalt concrete layer. Table 3.2 shows recommendations for selecting
vehicle operating speed and the estimated frequency at mid-depth of the asphalt concrete
layer.
Table 3.2 Recommendations for Selecting Vehicle-Operating Speed
Estimated Frequency at Layer Mid-Depth (Hz)

Operating
Type of Road Representative Thin HMA Thick HMA
Speed
Facility HMA layer layer wearing layers
(mph)
(4-12 in) surface (1-3in) binder/base
(3-12in)
Interstate 60 15-40 45-95 10-25
State primary 45 10-30 35-70 15-20
Urban street 15 5-10 10-25 5-10
Intersection 0.5 0.1-0.5 0.5-1.0 0.1-0.25
3.5 Environmental Effects

Environmental variations can have a significant impact on pavement performance.
Moisture and temperature changes can significantly affect the pavement material’s
properties and, hence, its strength, durability and load carrying capacity. The asphalt
bound materials modulus value can rise during the cold winter months by 20 times its
value during the hot summer months. Excessive moisture can drastically lead to the
stripping of asphalt mixture. Similarly, the resilient modulus of unbound materials at
freezing temperatures exhibits high values compared to the thawing months. The
moisture content affects the state of stress of unbound materials and it breaks up the
cementation between soil particles. Increased moisture contents lower the modulus of
unbound materials.
The Design Guide incorporated the Enhanced Integrated Climatic Model (EICM) to
simulate changes in the behaviour and characteristics of pavement and subgrade materials
that concur with climatic conditions over the design period. The model computes and
predicts the modulus adjustment factors, pore water pressure, water content, frost and
thaw depths, frost heave and drainage performance. The EICM consists of three
components as shown in Figure 3.3. The model role is to record the user-supplied
resilient modulus MR of all unbound layer materials at an initial or reference condition,
generally at near optimum moisture content and maximum dry density. The model then
16
evaluates expected changes in moisture content and the effect on the user-entered
resilient modulus.
The model also evaluates the effect of freezing on the layer MR and the effect of thawing
and recovery from the frozen MR condition. The model provides varying MR values in the
computation of critical pavement response parameters and damage at various points
within the pavement system. For the asphalt bound layer, the model evaluates the
changes in temperature as a function of time to allow for the calculation of the dynamic
modulus and thermal cracking.
Enhanced Integrated Climatic Model
CMS (Model) CRREL ID (Model)

Climatic Material (Model) The Infiltration
Structural Model Frost Heave and and Drainage
Thaw Settlement Model
Model
Figure 3.3 Enhanced Integrated Climatic Model Components
The climate module in the Design Guide software requires the user to specify a climate
file for the project location stored in the database. The database includes over 800
weather stations around the United States. The user has an opportunity to generate a
weather file by interpolating climatic data from selected locations inversely weighted by
the distance from the required location.
3.6 Performance Prediction Models

Permanent Deformation
The Design Guide predicts the rutting of each layer in the pavement structure as a
function of time and traffic. The Guide models only the initial and secondary permanent
deformation stages shown in Figure 3.4. The primary stage is modeled using an
extrapolation of the secondary stage trend. The tertiary stage is not considered in the
model.
17
Primary Secondary Tertiary
High initial small rate of rutting at constant rate of
rutting change of rutting- associated with
with volumetric volumetric change
εp
change
Permanent Strain
Flow Point
High level of rutting

associated with plastic
deformation
No volume change
Load Repetitions
Figure 3.4 Pavement Deformation Behaviour of Pavement Materials
The structural response model calculates the vertical strain in a layered pavement cross-
section at any given depth using the elastic properties of the material. Equation 3.2 gives
the vertical strain
εrz = 1/E* (σz – μσx – μσy ) (3.2)
From the knowledge of the resilient strain εrz, the plastic strain is predicted using an
empirical relationship. The selected laboratory model has the following form:
εp / εr = a1 Ta2 Na3 (3. 3)
Where:
εp = accumulated plastic strain at N repetitions of load (in/in)
εr = resilient strain of the asphalt material as a function of mix properties,
temperature and time rate of loading (in/in)
N = number of load repetitions
T = temperature (deg F)
a = non-linear regression coefficients

i
18
The total permanent deformation needs to be accumulated for different conditions over
the design period. To account for the variation in temperature, resilient modulus, and
moisture over the incremental periods, the Guide use a special approach called the strain
hardening approach to incorporate these variable parameters in a cumulative deformation
subsystem. Details of the procedure for calculating the permanent deformation of each
layer are given in Part 3-Chapter 3 of the guide documentation.
Fatigue Cracking
Repeated traffic loads cause tensile and shear stresses to develop in the asphalt layer.
Fatigue cracking initiates at locations of critical strain and stress. These locations mainly
depend on the stiffness of the asphalt layer and the load configuration. Two types of
fatigue cracks initiate in the asphalt layer. The first is the commonly known fatigue
cracking that initiates due to bending action, which results in flexural stresses at the
bottom of the asphalt layer. The second type, which propagates from the surface to the
bottom, is believed to be due to critical tensile and or shear stresses developed at the
surface due to high contact pressures at the tire edges-pavement interface. Highly aged
thin asphalt layers facilitate the initiation of fatigue type two cracks.
Modelling of both types of fatigue cracking is based on one approach. The estimation of
the fatigue damage is based on Miner’s law [23] given by Equation 3.4.
T
D = ∑ ( ni / Ni ) (3.4)
i=1
Where:
D = damage
T = total number of periods
n = actual traffic for period i

i
N = traffic allowed under conditions prevailing in i

i
The model used for the characterization of the fatigue cracking is as follows:
N = Ck1(1/ ε )k2(1/E)k3 (3.5)

f t
Where:
N = number of repetitions to fatigue cracking

f
ε = tensile strain at the critical location

t
19
E = stiffness of the material
k , k , k = laboratory regression coefficients

1 2 3
C = laboratory to field adjustment factor
The transfer function for bottom-up cracking and top-down cracking is calculated from
the fatigue damage model and calibrated separately.
Thermal Cracking Model

Thermal fracture analysis in the new Design Guide is based on the visco-elastic
properties of the asphalt mixture. The thermal cracking model, TCMODEL, is an
improved version of a model developed earlier by (SHRP). The procedure requires the
characterization of the HMA mix in an indirect tensile mode to measure the creep
compliance at one or three temperatures depending on the level of analysis. The master
creep compliance curve is then developed and fit to a power model defined by:
D (ξ) = D0 + D1ξm (3.6)
Where
D (ξ) = creep compliance at reduced time ξ
D0, D1 = Prony series parameters
m = power slope parameter
The next step is to relate the compliance D(t) to the relaxation modulus Er of the asphalt
mix, using viscoelastic transformation theory. The relaxation modulus is represented by a
generalized Maxwell model expressed by a Prony series relationship:
N+1
E(ξ) = ∑ Ei e-ξ /λi (3.7)
i=1
Where:
E(ξ) = relaxation modulus at reduced time ξ
Ei, λi = Prony series parameters for master relaxation modulus curve (spring
constants or moduli and relaxation times for the Maxwell elements)
20
Once the relaxation modulus is known, the thermal stresses in the pavement can be
computed using the following constitutive equation:
ξ dε
σ(ξ) = ∫ E(ξ – ξ’) ─ dξ’ (3.8)
0 dξ’
Where:
σ(ξ) = stress at reduced time ξ.
E(ξ-ξ′) = relaxation modulus at reduced time ξ-ξ′
ε = strain at reduced time ξ = (α (T(ξ′ ) - T0))
α = linear coefficient of thermal contraction
T(ξ′) = pavement temperature at reduced time ξ′
T0 = pavement temperature when σ = 0
ξ′ = variable of integration
The amount of crack propagation induced by a given thermal cooling cycle is given is
predicted using the Paris law of crack propagation
∆C = A∆Kn (3.9)
Where:
∆C = change in the crack depth due to a cooling cycle
∆K = change in the stress intensity factor due to a cooling cycle
A,n = fracture parameters for the asphalt mixture
The stress intensity K is estimated from the following formula, which was developed
based on an analytical finite element method analysis.
K = σ (0.45 + 1.99 Co 0.56) (3.10)
21
Where:
K = stress intensity factor
σ = far-field stress from pavement response model at depth of crack tip
Co = current crack length, feet
The value of the parameter (n) in Equation 3.9 is given by Equation 3.11:
n = 0.8(1+ 1/m) (3.11)
The value of m is obtained from the master creep compliance curve power function (3.6).
The fracture parameter (A) is given by Equation 3.12:
A = 10 (β* (4.389-2.52 log (E σm n)) (3.12)
Where
E = mixture stiffness, psi
σm = undamaged mixture tensile strength, psi
β = calibration parameter
Finally, the length of the thermal cracks is predicted from an assumed relationship
between the probability distribution of the log of the crack depth to HMA layer thickness
ratio and the percent of cracking. The relation between the computed crack depth and the
crack frequency is given by the following expression:
Cf = β * N (log C/hac/σ) (3.13)
Where:
Cf = observed amount of thermal cracking
β1 = regression coefficient determined through field calibration
22
N (z) = standard normal distribution evaluated at (z)
σ = standard deviation of the log of the depth of cracks in the pavement
C = crack depth
hac = thickness of asphalt layer
The thermal cracking model assumes a maximum crack length of 400 ft in every 500ft,
which is equivalent to a crack across a lane width of 12 feet spaced at 15 feet along the
pavement length. The model can only predict 50% of this maximum amount. The model
assumes failure occurs when the average crack depth reaches the thickness of the asphalt
layer.
Smoothness Models (IRI)
Distresses predicted by the mechanistic-empirical models, such as fatigue cracking,

permanent deformation and thermal cracking are correlated to smoothness. In addition
the smoothness model optionally considers other distresses, such as potholes, longitudinal
cracking outside the wheel path, and block cracking if there is potential of occurrence.
An initial IRI value, typically between 50 to 100 in/mile is required to estimate the
terminal distress. There are three models for predicting IRI depending on whether the
base type is an unbound aggregate base and subbase, asphalt treated base, or chemically
stabilized base.
3.7 Comparison between AASHTO 93 and the Proposed 2002 Guides
Review of the current and proposed AASHTO guides, conducted in this study, revealed
significant improvements in the design process achieved in the new Guide.
The new Guide is capable of addressing a wide range of pavement structures covering
new construction and rehabilitation of existing pavements. Analysis of composite
structures is now possible, which address the need to overlay AC pavement with PCC
or the opposite.
• Significant changes were incorporated in the new guide to improve treatment of

traffic related variables. The current guide relies on an equivalent standard axles load
(ESAL), which obscures the impact of the various characteristics of vehicles using
the roadway network on the type and extent of damage. The new guide expands
characterization of road vehicles to account for tire pressure, axle load and
distribution.
• The impact of the climate on the material response is adequately captured in the new
guide where climate conditions are linked to the material library to facilitate
23
adjustment of AC dynamic modulus and unbound resilient modulus according to
seasonal changes in moisture and temperature condition.
• With the introduction of mechanistic material characterization, the model is now
capable of assessing the performance of newly developed materials, such as
engineered binders, unconventional gradations and recycled materials. Reliance in the
past on field trials proved to be expensive and takes a long time to rate these newly
developed materials in comparison with conventional materials.
• Foundation material (native soil) and existing pavement are analyzed adequately in
the new guide making it possible for the model to assess a number of design options
before implementation.
• The introduction of a performance model (distress prediction) promotes the
implementation of performance-based design instead of stress and strain based
designs, which proved unreliable in the past. Accumulation of damage with time is an
effective tool for designers to plan rehabilitation activities anticipated in the future.
Providing designers with the ability to incorporate local preference using acceptance
criteria based on a number of distress types and smoothness of the road supports this
process.
Capabilities of the proposed guide are expected to encourage users of the current guide to
switch to the new application. However, the ability of the new guide to address design
requirements should be examined before moving to implementation. Chapter 5 of this
study includes the results of a sensitivity analysis performed to test the ability of the new
guide in performing a number of design functions. Parameters of the sensitivity analysis
and input data used in the model runs are described in Chapter 4.
24
CHAPTER 4
EVALUATION OF THE PROPOSED AASHTO 2002 GUIDE
4.1 Introduction
The evaluation of an analytical approach to design and analysis usually includes a
validation component in the form of field investigations where controlled road sections
are built and performance monitored. Comparisons between predicted and actual
performance provides a direct indication of the ability of the model to perform the
evaluated functions. The Long-Term Pavement Performance (LTPP) program, an
element of the Strategic Highway Research Program (SHRP), was one source used to
evaluate the AASHTO 2002 Guide. However, since some of the LTPP data has been
originally used to generate some of the empirical performance models of the proposed
guide, the evaluation process is not adequate for concluding that the model will
effectively address combinations of parameters that were not covered in the LTPP
experiments such as material characteristics, layer configurations, construction quality
and environmental conditions. Accordingly, the AASHTO project committee launched a
second series of validation projects (NCHRP 1-30A, NCHRP 1-40A) to further test the
model. These new tests will take two to three years and will still fall short of validating
the model since they will not address long-term performance.
This study focuses on sensitivity analysis to examine the ability of the model to account
for the impact of variables with a known influence on performance, especially material
characteristics. This chapter discusses input parameters used in the sensitivity analysis;
the results are discussed in Chapter 5.
4.2 Asphalt Concrete Characteristics
Following on the recommendations of the Design Guide, a laboratory investigation was

performed at the National Research Council Canada (NRCC) to produce necessary AC
input data. Standards of the AASHTO and ASTM were used to perform physical testing
on the HMA fractions and to design the different mixes used in the sensitivity analysis. A
complete list of the AC and binder tests required by the software for all input levels is
shown in Appendix A. For this study the Marshal approach was followed in designing the
AC mixes shown in Table 4.1. The two mixes were selected to represent commonly used
asphalt mixes in Ontario Canada. The Marshal mixes were designed according to
AASHTO standard designated T 245 [24] with different binder types leading to slightly
different binder contents.
Aggregate Proportioning
Sieve analysis was performed according to Ontario Ministry of Transportation (MTO)

specifications [25]. The MTO specifications were followed to determine the
combinations of aggregate fractions that achieve the job mix formulas. The gradation
curves for the two mixes are shown in Figures 4.1a and 4.1b. The designed aggregate
gradation is shown in Table 4.2.
34
Table 4.1 Designation of Mixes Used in the Study.
MTO Mix Designation Function in Road Nominal Maximum

Structure Aggregate Size (mm)
HL3 Surface layer 13.2
HL8 Stabilized base course 19
HL 3 Mix Design
120
100
80
% Passing
Mix Design
60 Coarse Limit
Fine Limit
40
20
0
0.10 1.00 10.00 100.00
Sieve Size (mm)
Figure 4.1a Gradation Curve of Mix HL3
HL8 Mix Design
120
100
80
% Passing
Mix Design
60 Fine Limit
Coarse Limit
40
20
0
0.10 1.00 10.00 100.00
Sieve Size (mm)
Figure 4.1b Gradation Curve for Mix HL8
35
Table 4.2 Aggregate Gradations
MTO HL3 HL8
Sieve Mix Design Mix Design

Designation
(mm)
% Passing % Passing
26.5 100.0 100.0
19 100.0 96.5
16 100.0 88.5
13.2 97.5 72.2
9.5 79.3 53.8
4.8 59.0 48.6
2.4 52.1 39.1
1.2 46.3 31.9
0.600 22.5 17.3
0.300 7.9 7.8
0.150 3.0 3.6
0.075 0.5 0.5
Binder Characteristics
Asphalt cements were classified according to SuperPave performance grades (PG)
specifications. Three binder types were used to prepare the AC specimens. Two binders,
PG 58-22 and PG 52-34, were described as neat binders. The third binder (PG 64-34) is
an engineered binder and was used in the HL3 mix only, because it is relatively
expensive to use in stabilized base courses. Engineered binders are generally used to
improve the performance of AC by reducing the potential for rutting or cracking under
severe traffic and temperature conditions.
Mix Design Procedure
The Marshal Hammer was used to compact specimens prepared using the Marshal mix
design procedure. The optimum binder content was selected following local
specifications based originally on the Marshal design approach. The selected binder
contents were then used to prepare the specimens for the complex modulus test.
A mechanical mixer was used to prepare the aggregate–binder mixture. It was aged in an
oven for two hours at the compaction temperature of the binders in accordance with the
short-term aging procedure specified by AASHTO Designation PP2 [26]. The binder
manufacturer specifies compaction temperatures. Samples were then compacted using the
gyratory compactor. The gyratory compactor was used to produce samples at a specified
height and air voids content. The number of gyrations was selected to produce samples
with the desired size at the specified air voids percentage. The size of the samples
36
prepared for the complex modulus tests was determined according to ASTM specification
D3497 [27], which is 100 mm for height and diameter.
Standard Physical Tests
Physical properties of the compacted mix specimens were determined to ensure the mix
compliance with local criteria. Two different measures of densities were typically taken
according to AASHTO specifications T166 [28] and 269-97 [29]: bulk specific gravity
(Gmb) and theoretical maximum specific gravity ( Gmm).
These densities were then used to calculate the volumetric parameters of the HMA: air
voids (Va), voids in the mineral aggregate (VMA) and voids filled with asphalt (VFA)
The Marshall stability and flow test was conducted according to AASHTO standard
designated T 245 [24]. Physical properties of the selected mix designs that satisfied the
mix design criteria are summarized in Tables 4.3 and Table 4.4.
Table 4.3 Physical Properties of Mix HL3
Mix Type
Physical Property
HL3- PG 58-22 HL3- PG 64-34 HL3 – PG 52-34
Pb (%) 5 5 4.9
Gmm 2.53 2.477 2.50
Gmb 2.396 2.337 2.375
Gsb 2.72 2.72 2.72
VMA 16.3 18.4 17.0
Va 5.3 5.7 5.1
Vbeff 11.0 12.7 11.9
Table 4.4 Physical Properties of Mix HL8
Mix Type
Physical Property
HL3- PG 58-22 HL3 – PG 52-34
Pb (%) 4.5 4.5
Gmm 2.54 2.56
Gmb 2.412 2.429
Gsb 2.74 2.74
VMA 15.9 15.3
Va 5.0 5.3
Vbeff 10.9 10.0
37
Mechanical Testing
The primary stiffness property required for the AC analysis is the dynamic modulus (E*).
Mechanical testing was performed on HMA samples prepared for the selected mixes and
compacted using a gyratory compacter. The NRC test protocol was followed in
conducting the complex modulus test. Tests were performed to cover a wide range of
pavement temperatures (400C to –100C) and a frequency sweep from 0.1 Hz to 20 Hz.
Typical test output produced from the stress and strain signals generated under a specific
temperature and loading frequency is shown in Figure 4.2. Test results for the HL3 mixes
made with the three binders for this study are listed in Tables 4.5a, 4.5b and 4.5c in the
format required for running the AASHTO 2002 model.
F
ΔΗ
Strain (MPa)
Stress (MPa) Phase Lag
Strain Amplitude εο
Stress Amplitude σο
Time
Stress Amplitude σο Strain Amplitude εο
Stress Signal Strain Signal
Stress Cycle Duration
Strain Cycle Duration
Figure 4.2 Viscous Response of Asphalt Concrete
38
Tables 4.5 Dynamic Modulus Data Prepared in the Format Required for Running
AASHTO 2002 Model
Table 4.5a Dynamic Modulus for Mix HL3- 58-22

Number of temperatures: 5
Number of frequencies: 6
Temperature Mixture E* (psi)

o
F 0.1 0.3 1 5 10 20
14 2612482 2859216 3120464 3367199 3381713 3497823
32 1582002 1886792 2206096 2656023 2830189 3062409
68 671988 985486 1310595 1785196 1959361 2104499
86 49637 76197 126851 249637 339623 461538
104 20882 26560 42671 87083 119739 172714
Table 4.5b Dynamic Modulus for Mix HL3- 52-34


o
F 0.1 0.3 1 5 10 20
14 785195 955007 1169811 1494920 1683599 1901306
32 381712 519593 693759 978229 1120464 1342525
50 99129 142670 211901 349782 431059 551523
86 10885 16981 27866 56313 77503 111901
104 3628 8461 8461 17851 25253 39477
Table 4.5c Dynamic Modulus for Mix HL3- 64-34


o
F 0.1 0.3 1 5 10 20
14 917271 1174165 1509433 1770682 1857764 2089985
32 349782 468795 622641 880986 1010159 1178519
68 41074 58055 89550 159651 201741 1178519
86 15674 21770 32801 61683 79390 106966
104 8461 10885 15820 29027 37155 50798
39
Mix Gradation Used in Predicting the Dynamic Modulus in Level 3
For Levels 2 and 3 analyses, the proposed guide employs a master curve that is developed
directly from dynamic modulus values estimated using the predictive equation shown
below (Equation 4.1). It predicts the dynamic modulus of asphalt mixtures over a range
of temperatures, rates of loading, and aging conditions using data from the volumetric
design of the mixture and binder properties. The required mix gradation information for
Levels 2 and 3 are shown in Table 4.6.
log E * = −1.249937 + 0.029232 P200 − 0.001767( P200 ) 2 + 0.002841P4 − 0.058097Va
Vbeff [3.871977 − 0.0021P4 + 0.00395P38 − 0.000017( P38 )2 + 0.00547 P34 ]
−0.802208 +
(Vbeff + Va ) 1 + e( −0.603313−0.313351log f −0.393532 logη )
(4.1)
Where:
E * = asphalt mix complex modulus, in 105 psi
η = bitumen viscosity, in 106 poise

f = loading frequency, in Hz
Va= percent air voids in the mix, by volume
Vbeff = percent effective bitumen content, by volume
P34 = percent retained on ¾-inch sieve, by total aggregate weight (cumulative)
P38 = percent retained on 3/8-inch sieve, by total aggregate weight (cumulative)
P4 = percent retained on No. 4 sieve, by total aggregate weight (cumulative)
P200 = percent passing No. 200 sieve, by total aggregate weight.
Witczack’s model (Equation 4.1) is one of numerous methods used for predicting the
modulus of AC mixtures. However, the equation needs the binder viscosity as an input.
The model uses default viscosity data and empirical approximations to obtain viscosity
values at intermediate and lower temperatures. The actual binder characteristics in all
cases would not match the empirically approximated values. In Chapter 5, the difference
in performance predictions between Levels 1 and 3 is discussed while looking into the
difference between the dynamic modulus measured in the laboratory and that estimated
by the prediction equation used in Level 3.
40
Table 4.6 Aggregate Gradation Data Required For Level 2 and 3 Input:
Mix Type
HL3 HL8
Cumulative % Retained on 3/4 in. sieve:
0.0 3.5
Cumulative % Retained on 3/8 in. sieve:
20.7 46.2
Cumulative % Retained on #4 sieve
41.0 59.0
% Passing #200 sieve:
0.5 0.5
Binder Characterization
The binder complex modulus and phase angle data are needed over a range of
temperatures for a loading rate of 1.59 Hz (10 rad/sec) as input for Level 1 design.
Typical values of dynamic shear rheometer measurements for PG 58-22 are shown in
Table 4.7.
Table 4.7 SuperPave Binder Test Data
Temperature (oF) Angular Frequencies = 10 rad/sec

G* (psi) Delta (o)
59 2605000 50
77 572800 56
95 120800 63
113 23440 69
140 3267 78
The design guide software converts the binder stiffness data at each temperature to
viscosity using Equation 4.2.
η = G*[1/sin δ ]4.8628 (4.2)

10
Where:
G* = binder complex shear modulus, Pa
δ = binder phase angle.
41
The parameters of the A and VTS of the ASTM Equation below [30] are then found by
the linear regression of Equation 4.3.
Log log η = A + VTS Log TR. (4.3)
Where:
η = viscosity, cP
T = temperature, Rankine
R
A = regression intercept
VTS = regression slope of viscosity temperature susceptibility.
Master Curve for Level 1

After establishing the viscosity–temperature relationship, the laboratory dynamic
modulus test data are shifted to construct the master curve represented by Equation 4.4.
α
log(E*) = δ + (4.4)
β + γ [log(t)-c(log(η)-log(ηTr ))]
1 +e
Where:
E* = dynamic modulus, psi
t = time of loading, sec
η = viscosity at temperature of interest, CPoise
η = viscosity at reference temperature, CPoise
Tr
α,β,δ,γ,c = mixture specific fitting parameters.
The shift factors are calculated using Equation 4.5. The viscosity at the age and
temperature of interest are obtained from the global aging model incorporated in the
software. The Global Aging System comprises a number of models for adjusting the
viscosity of the original binder for short-term aging that occurs during mixing and lay-
down operations and for in-service long-term aging.
log(t ) = log(t) – c( log(η) - log(η )) (4.5)

r Tr
Where:
t = reduced time, sec
r
t = loading time, sec
η = viscosity at the age and temperature of interest, CPoise
η = viscosity at reference temperature and RTFO aging, CPoise
Tr
42
Master Curve for Level 3
The input at Level 3 includes the mix data, as mentioned before, in addition to the
performance grade of the asphalt binder. As no testing is required for the binder, the
viscosity is estimated from default temperature-viscosity A and VTS values. Typical
values built in the software are shown in Table 4.8.
Table 4.8 Recommended RTFO A and VTS Parameters Based on Asphalt PG Grade
Binder Grade A VTS

PG 58-22 11.787 -3.981
PG 52-34 10.707 -3.602
PG 64-34 9.461 -3.134
Poisson’s Ratio for Bituminous Materials
Poisson’s ratio for asphalt concrete usually ranges between 0.15 and 0.5 and is a function
of the asphalt mix and temperature. This study used a typical value of 0.35.
Other HMA Material Properties

Tensile Strength
Tensile strength values required as input for Level 3, calculated internally by the
software, were used in this study. These values were based on a regression equation
developed under NCHRP 1-37A. The indirect tensile strength at 14 ºF is calculated as a
function of as-constructed air voids, as constructed voids filled with asphalt, the binder
penetration at 77 ºF and the viscosity-temperature susceptibility intercept A.
Creep Compliance
Values for creep compliance used in this study as input for Level 3 design were estimated
using a regression equation developed under NCHRP 1-37A, and were calculated by the
software using the as built air voids (Va), as built voids filled with asphalt (VFA) in
addition to the binder penetration at 77 ºF.
Table 4.9 Thermal Properties of Asphalt Concrete
Thermal conductivity BTU/hr-ft-oF 0.67

Specific heat (BTU/lb-oF) 0.23
Reference temperature 70 oF
43
4.3 Unbound Materials
Attempts were made to adopt non-linear resilient modulus models in the design and
analysis of unbound pavement layers including the granular base course and subgrade,
which is usually a native soil consisting of a cohesive material. However, the finite
element model incorporated in the AASHTO model for non-linear analysis is not
calibrated and is being reviewed. Instead, linear elastic analysis is available which is
performed with Level 3 input data covering a number of parameters that include the
AASHTO soil classification designated M 145[31] and other physical properties. The
data for two soils used in this study were produced in NRC laboratories based on testing
of different soil categories as shown in Table 4.8. Both Granular materials A and B were
classified as A-1-a according to AASHTO specifications. The properties of these granular
materials are used in Chapter 5 of this study to assess the implications of using Level 3 in
estimating the resilient modulus based on AASHTO classifications.
The sensitivity analysis conducted in this study covered other parameters that directly or
indirectly influence the material response. These parameters included traffic and seasonal
changes in moisture and temperature conditions as discussed below.
Table 4.10 Properties of Different Granular Materials (A-1-a)
Measured Property Guide Proposal
Property
Granular A Granular B Lower Limit Upper Limit
Plasticity index 0 0 0 6
Passing #200 (%) 6 7 0 15
Passing #4 (%) 19 21 15 30
D60 (mm) 11 7 2 25
MDD (pcf) 147.7 150 120 150
Specific gravity 2.72 2.73 2 4
OMC (%) 5.0 5.4 2 40
Resilient modulus (psi) 26124 51700 38500 42000
4.4 Traffic Data
Data input for the sensitivity analysis were formulated to account for the wide range of
traffic variables and requirements of the different levels of design included in the
proposed AASHTO Design Guide. The variables cover the wide axle load spectra, speed,
gear/axle configuration/tire spacing, tire pressure, traffic wander and the factors that
capture monthly and daily traffic distribution. Traffic data for the sensitivity analysis
were selected to establish categories that represent low, medium, medium to heavy and
heavy volumes as shown in Table 4.11. The model uses these data to calculate pavement
responses and the number of axle load applications over a point for predicting distress
and performance. (See the results in Chapter 5) The AADTT in Table 4.11 is the average
daily number of trucks traffic (FHWA vehicle classes 4-13) expected over the base year.
The traffic growth factor was considered as 3% compound growth. The default traffic
volume adjustment factors were selected since there were no WIM station data and were
found reasonable; for example the monthly adjustment factor was set to one to indicate
44
that the volume of trucks is evenly distributed over the months of the year. The default
truck traffic classification is shown in table 4.12. The truck traffic classification
represents the percentage of each type of vehicles in the traffic stream. The axle load
distribution was taken as the default. General traffic input values are shown in Tables
4.13 to 4.15.
Table 4.11 Traffic Data
Initial Two-Way 200 Low,

AADTT 1000 Medium,
2000 Medium to High
4000 High
Number of lanes in
2
design direction
Percent of trucks in
50
design direction
Percent of trucks in
95
design lane
Operational speed 5 Low,
(mph) 25 Medium,
45 Medium to High
60 High
Table 4.12 Truck Traffic Classification (Default)
Truck Traffic Classification %

Class 4 1.3
Class 5 8.5
Class 6 2.8
Class 7 0.3
Class 8 7.6
Class 9 74.0
Class 10 1.2
Class 11 3.4
Class 12 0.6
Class 13 0.3
Total percentage 100
45
Table 4.13 Traffic General Input
Mean wheel location (inch from the lane marking) 18

Traffic wander standard deviation (in.) 10
Design lane width (ft.) 12
The average number of axles for each truck class (classes 4-13) for each axle type (single,
tandem, tridem and quad) is calculated from WIM data measured over time by dividing
the total number of a specific axle type measured for a truck class by the total number of
trucks in that class. The traffic module assumes that the number of axles for each axle
type is constant with time. The proposed Design Guide software contains a default set of
values estimated using LTPP data.
Table 4.14 Number of Axle/Truck
Vehicle Single Tandem Tridem Quad

Class Axle Axle Axle Axle
4 1.62 0.39 0.00 0.00
5 2.00 0.00 0.00 0.00
6 1.02 0.99 0.00 0.00
7 1.00 0.26 0.83 0.00
8 2.38 0.67 0.00 0.00
9 1.13 1.93 0.00 0.00
10 1.19 1.09 0.89 0.00
11 4.29 0.26 0.06 0.00
12 3.52 1.14 0.06 0.00
13 2.15 2.13 0.35 0.00
Table 4.15 Axle Configuration
Average axle width edge to edge outside dim. (ft.) 8.5

Dual tire spacing (in.) 12
Tire pressure – Single tire (psi) 120
Tire pressure – Dual tire (psi) 120
Average axle spacing – Tandem axle (in.) 51.6
Average axle spacing – Tridem axle (in.) 49.2
Average axle spacing – Quad axle (in.) 49.2
46
4.5 Climatic Data
The database incorporated in the current version of the proposed guide does not include
climatic information for Canadian regions. However, the Transportation Association of
Canada (TAC) is considering such a project to compile a climatic database using
Environment Canada data. The input parameters include air temperature, precipitation,
water table depth, wind speed and sun radiation. The model uses EICM input to adjust
unbound material properties, mainly the resilient modulus. An hourly temperature profile
was also used to adjust the dynamic modulus and all other parameters built in the distress
module to estimate accurately the potential for rutting. In this study, USA geographical
sites with climatic conditions available in the database were selected to represent warm
and cold regions to test the model’s ability in accounting for the impact of seasonal
variation in moisture and temperature conditions on predicted performance.
Table 4.16 Climatic Input
Weather Station Description Mean Annual Air Temperature

Minnesota Relatively cold region 461°F
Chicago Intermediate 52 °F
Dallas Relatively warm region 67 °F
Table 4.17 Drainage Input
Cross slope 2%
Drainage and
Drainage path from the center line to the edge
surface 12
drain adjacent to the lane shoulder (ft.)
properties
Surface shortwave absorptivity 0.85
4.6 Sensitivity Analysis
The collected input data discussed above was designed to serve the sensitivity analysis
performed in this study covering four major items while focusing on material
characteristics to evaluate the impact on performance associated with the type of
pavement structure, seasonal variations in climatic conditions and traffic volumes.
Results produced by running the proposed AASHTO design software associated with
variations in those design parameters are discussed in Chapter 5.
47
Structural considerations
The performance of roads constructed over subgrades with different levels of support was
examined. Native soils with a different resilient behaviour were used to perform the
analytical runs. The predicted performance was used to evaluate the adequacy of the
structure based on comparisons with a preset tolerance level for specific performance
criteria (rutting, fatigue cracking). Additional runs were then performed after introducing
of some design solutions, such as increasing the thickness of the AC layer or the granular
base.
Material Considerations
The model application covered a number of material variables to examine the sensitivity
of the design guide to material characteristics. Mix designs involving different aggregate
gradations and binder types were used in the analysis. The AC layers involving mixes
that include binders with different performance grades (based on SHRP specification)
were simulated to determine the effectiveness of typical design solutions based on
manipulation of binder characteristics (engineered binders).
Design Levels
The above model applications were performed using Level 1 AC input and Level 3 for
unbound materials. However, separate runs were performed to assess performance
predictions based on applications of Level 3 and the output was compared with that
produced earlier using Level 1 AC input.
4.7 Model Limitations in Predicting Fatigue Cracking

Sensitivity analysis outlined in Section 4.6 produced results that facilitated assessment of
the model ability to evaluate the resistance of AC mixes to permanent deformation and
fatigue cracking. Fatigue-related damage is well defined in the literature, and most
models use an approach similar to that adopted in the proposed Design Guide. Most of
the adopted transfer functions relate the AC tensile strains to the allowable number of
load repetitions. The asphalt institute and shell models relate the allowable number of
load repetitions that cause fatigue cracking to the tensile strain at the bottom of the AC
layer and the AC modulus (E). However, it is essential to carefully calibrate the transfer
function by applying appropriate calibration parameters so that the predicted distress can
match field observations. Such a process will be needed in the calibration of the proposed
guide before it can be used with confidence.
Longitudinal Fatigue Cracking (Surface/Down) Model
The longitudinal fatigue cracking model in the 2002 Design Guide was based on the same
concept on which the conventional fatigue model was constructed [32]. This type of
fatigue is not as well defined in the proposed Design Guide from a mechanistic viewpoint
as the more classical bottom/up fatigue. The Design Guide suggested that the cause of
48
this type of fatigue is due to critical tensile and/or shear stresses developed at the surface
and caused by extremely large contact pressures at the tire edges pavement interface,
coupled with a highly aged stiff thin surface layer that has become oxidized.
The results of model runs conducted in this study produced fluctuating and unstable
trends when Level 1 AC input was used. On the other hand, input Level 3 functioned
better producing stable results that were similar to the conventional fatigue model.
However, it is not yet shown which of the two approaches quantifies better the damage
associated with fatigue.
Currently NCHRP Project 1-42 is investigating the top-down fatigue cracking of hot
asphalt layers. The conducted literature search suggests that top-down cracking in
flexible pavements is primarily caused by traffic-associated fatigue and/or thermal
stresses. Top-down cracking is significantly affected by the interaction between many
factors, including poor compaction and segregation during construction, pavement
structure, modulus gradients within the pavement, age hardening of the pavement surface,
moisture damage and mixture fracture toughness [33].
Thermal Cracking Model

Thermal stresses are predicted by the TCMODEL based on viscoelastic characteristics of
the AC mix at low temperatures. The model was last calibrated in 2003 to be consistent
with current SuperPave IDT protocols. The original 1000-second creep data files were
changed to 100-second creep data. The calibration report [34] included in the design
documentation has highlighted future research areas to improve the model prediction
capabilities. After the study, the effect of mix aging in the field on thermal cracking
development will be incorporated in the model. Another topic relates to fracture and the
utilization of tensile strength data. The current model uses mixture tensile strength at a
single test temperature of -10 ºC, and then develops fracture parameters by taking this
value as an estimate of the undamaged tensile strength of the mixture. In the future, the
model can be improved once a better understanding of mixture fracture behaviour is
obtained. A third study area relates to the recent finding that traffic loads applied during
critical cooling events can increase tensile stresses by more than 50% [35].
The most recent calibration process is considered insufficient, because of the shortage of
field data and site-specific information. The 14 CSHRP sites could not be used, because
the Design Guide does not include climatic data for Canadian weather stations.
Additionally, not all GPS (SHRP general pavement sections) sites could be interpolated
from suitable weather stations at currently available sites.
Two main points affect the accuracy of the thermal cracking prediction process: the
reliability of the climatic model that will be used to determine the pavement temperature
and eventually the stresses as discussed in the following subsection and the definition and
identification of the thermal cracking patterns in the field. There is a misconception that
all transverse cracks are due to thermal cracking. Poor compaction methods using
conventional rollers would result in hairline cracks [36, 37, 38, 39, 40], which widen and
propagate with time in the presence of specific environmental factors. These cracks, left
behind by conventional compactors on a freshly compacted asphalt layer, are oriented in
a transverse direction. Another point is associated with the fact that temperature is not
unidirectional. The thermal crack may follow any direction similar to the thermal cracks
that occur on plain concrete ground slabs, a pattern similar to block cracking.
49
The Climatic Model
The climatic model has an essential and direct influence on the pavement performance
prediction process. It is essential to develop unique environmental regions and construct
the climatic database files. The validation of the EICM using regional data is important.
A recent study [41] to validate the Model, pointed out that the variation in climatic
patterns from region to region coupled with the variation in site-specific conditions across
North America make it difficult to develop standard models to account for seasonal
variation in material properties for all regions. The results of the validation study did not
indicate a high correlation between EICM-predicted temperature and moisture profiles
for the various pavement layers as compared to measured values. The findings of this
study should receive high consideration, because of the direct impact of the climatic
model on the whole analysis process
50
CHAPTER 5
RESULTS AND ANALYSIS
5.1 Introduction:
Results of the sensitivity analysis exercises discussed in Chapter 4, designed to evaluate
capabilities of the Design Guide are discussed in this chapter. Analysis addresses the
impact of changes in each individual design feature on performance as reflected on the
extent of damage accumulating under the action of traffic loading. It should be noted that
the stress/strain profiles are not presented as an output of the model application (see
typical software run in Appendix B). Accordingly, it is not possible to evaluate stresses
and strains predicted by the proposed AASHTO model using another analytical model or
field data. Instead, this study reviews model output to examine its conformance with
basic mechanics rules and make sure performance predictions agree with trends known
for the investigated materials. Analysis also covers the performance of the modules
incorporated in the guide to identify areas of future research that may be pursued and the
outcome used to improve the accuracy of the Design Guide.
This study covers the primary distress types: permanent deformation and fatigue
cracking. It also discusses fatigue longitudinal cracking and thermal cracking models, and
evaluated the impact of using hierarchical input levels on the performance prediction
process.
5.2 Performance Based Design and Analysis

The guide predicts permanent deformation in each individual layer and finally sums up
these values to arrive at the total permanent deformation in the entire pavement structure.
The plastic strain is estimated empirically under specific pavement conditions for a total
number of traffic repetitions, and the strain hardening approach is then used to
accumulate the deformations from sub-seasons to obtain the total plastic deformation.
The output of several runs conducted to investigate the impact of changes in the
magnitude of various input variables on the predicted permanent deformation is discussed
in the following section. The runs were conducted for a design life of 20 years and a
reliability level of 90%.
Road Built over a Weak Subgrade
A number of runs were performed to examine the ability of the Design Guide in
evaluating the impact of a problem, such as a weak subgrade, and the information it
provides to assist the designer in arriving at a sound design alternative. A trial design was
modified several times to improve the performance of the road structure. The trial design
assumed an urban road with medium traffic (AADTT = 1000) with an operating speed of
25 mph, located in a relatively warm region and having a mean annual air temperature of
67° F. The adopted road structure is shown in Figure 5.1
51
AC: 4 in. HL3 – PG 58-22
Base: 10 in. GB type A-3 (MR = 24000 psi)
Subgrade: A-7-6 (MR = 5000 psi)
Figure 5.1 Reference Pavement Structure

A one-inch tolerance level (design limit) for total rutting and 0.33 in. for AC rutting were
selected based on 90% reliability. Analysis performed on the output of the initial design
trial revealed there would be excessive rutting in the road exceeding the one-inch design
limit as shown in Figure 5.2. The AC layer and subgrade showed the greatest permanent
deformation levels. Rutting in the AC layer exceeded the design limit of 0.33 in.
Although exposed to low stresses, rutting in the subgrade exceeded that predicted in the
granular base. Accordingly, the initial design should be adjusted to improve the road
performance as discussed in the following subsection.
1.80 SubTotal AC Total Rutting at 90% Reliability

SubTotal Base
SubTotal SG
Total Rutting (50% Reliability)
Total Rutting (90% Reliability)
Design Limit for Rutting
1.40
Total Rutting
(50% Reliability)
Rutting Depth (in)
Design Limit for Total Rutting

1.00
AC Rutting
0.60 Subgrade
0.20 Base
0 24 48 72 96 120 144 168 192 216 240 264
Pavement Age (month)
Figure 5.2 Permanent Deformation Predictions
52
Effect of HMA Thickness on Permanent Deformation
The solution to the weak subgrade problem lies in stiffening the pavement layers above
the subgrade. Because of the absence of good quality base course material (a scenario
selected here because of the lack of the current model sensitivity to variation in unbound
material parameters), adjustment of the AC layer thickness was investigated. Five AC
layer thicknesses were used as input to investigate the model’s ability to capture the
reduction in traffic-induced stresses and strains transmitted to the weak subgrade and as a
consequence, limiting build up of rutting in the pavement.
The AC thickness was increased from 4 to 6, 8 and 10 inches, while keeping the base
course thickness constant at 10 inches. The effects of AC thickness increase on the total
rutting as well as the contribution of individual pavement layers were examined. The
results are shown in Table 5.1 and plotted in Figure 5.3.
Increasing the AC layer thickness reduces the total permanent deformation in the
pavement structure as shown in Figure 5.3. The result agrees well with what have been
established theoretically and in field performance records, where increasing the HMA
layer thickness increases the stiffness of the asphalt layer and the overall structure and
hence decreases stresses and strains in layers below. Rutting, in the subgrade, diminished
by 34% when the AC thickness was increased from 4 to 10 inches. Similarly, AC rutting
dropped by 26% resulting in an overall drop in rutting of 43%. No visible change was
observed in base course rutting as a result of AC thickness changes. None of the
evaluated design options, based on changing the AC layer thickness, satisfied the design
limit involving a maximum tolerance of 1.0 in rutting.
Table 5.1 Predicted Permanent Deformation for Different AC Layer Thicknesses

(Base course thickness 10 in., MR = 24000 psi - weak subgrade, MR = 5000 psi)
Decrease
AC
Total Rutting AC Rutting Base Rutting SG Rutting in Total
Thickness
(in.) (in.) (in.) (in.) Rutting
(in)
%
4 1.4 0.83 0.13 0.44 Reference
6 1.27 0.79 0.1 0.38 9
8 1.12 0.71 0.08 0.32 20
10 0.97 0.61 0.067 0.29 43
53
1.6 AC: PG58-22
Base: 10 (in)
1.4 MR = 24000 psi
Subgrade:
1.2 MR = 5000 psi
1 AC Rutting
Rutting (in)
Total Rutting
0.8
Base Rutting
0.6 Subgrade Rutting
0.4
0.2
0
0 2 4 6 8 10 12
AC Thickness (in)
Figure 5.3 Effect of HMA Layer Thickness Changes on Predicted Rutting.
Effect of Base Thickness on Permanent Deformation

The effect of increasing the base course thickness on predicted total permanent
deformation of the pavement structure as well as on individual layers of the pavement
was also investigated. The base course thickness was increased from 10 to 12, 14,16, and
18 inches while keeping the AC layer thickness constant at 6 inches. Results are shown in
Table 5.2 and were plotted in Figure 5.4. Increasing the base thickness has no effect on
the rutting of the AC layer. However, the increase in the base resulted in a slight decrease
in permanent deformation in the subgrade, probably as a result of a drop in the vertical
compressive strain. None of the evaluated design options, based on changing the granular
base layer thickness, satisfied the design limit involving a maximum tolerance of 1.0 in.
rutting.
54
Table 5.2 Predicted Permanent Deformations for Different Base Layer Thicknesses
(AC thickness 6 in., binder PG 58-22 , weak Subgrade MR =5000 psi)
Change in
Base Total
AC Rutting Base Rutting Subgrade Total Rutting
Thickness Rutting
(in.) (in.) Rutting (in.) %
(in.) (in.)
8 1.28 0.78 0.0887 0.4115 + 0.3
10 1.275 0.79 0.101 0.3802 Reference
12 1.294 0.805 0.1228 0.3656 -1.5
14 1.286 0.814 0.1311 0.3412 -0.86
16 1.295 0.818 0.1474 0.3296 -1.56
18 1.28 0.83 0.147 0.303 -0.4
AC: 6(in)-HL3
1.4 PG58-22
Base: var.
1.2 MR = 24000 psi
Subgrade:
1 MR = 5000 psi
Rutting (in)
0.8 Total Rut

AC Rut
0.6 Base Rut
SubG Rut
0.4
0.2
0
0 4 8 12 16 20
Base Thk (in)
Figure 5.4 Effect of Base Thicknesses on Rutting.
55
In summary, model predictions indicate that none of the attempted design alternatives
(increasing the thickness of the AC or base course layers) will satisfy the set design
criteria (maximum of 1.0 in. total rutting depth). Accordingly, a number of other options
involving a combination of design solutions were attempted to overcome the problem; the
results are shown in Table 5.3.
Combination 1: Increase AC layer thickness to10 in. base course to 18 in.(see Figure 5.5)
AC: 10” HL3 – PG 58-22
Base: 18” GB type A-3 (MR = 24000 psi)
SG: A-7-6 (MR = 5000 psi)
Figure 5.5 Combination 1
Analysis of the results produced by the model application considering the configuration
of the road structure and materials included in Combination 1 reflected a drop in total
permanent deformation (at 50% reliability) to 0.98 in, which is below the tolerance level
of 1.0 in.. However, predicted damage at 90% reliability level is still above the design
limit (predicted rutting at 90% reliability = 1.18 in.) rendering this solution unsuitable for
the investigated subgrade condition.
Combination 2: This design trial involves a combined change in layer thicknesses

(compared with the reference structure shown in Figure 5.1). Application of a stiffer
HMA was achieved by using a binder grade with more stability at relatively high
temperatures (see Figure 5.6). A binder designated PG 70-22 was incorporated in the mix
to replace the PG 58-22 binder. Total permanent deformation of 0.66 in. was predicted by
the model, which satisfies the design limit specified under a reliability level 90%.
AC: 10 in. HL3 – PG 70-22
Base: 18 in. GB type A-3 (MR = 24 000 psi)

(Note: the only base material available
SG: A-7-6 (MR = 5000 psi)

Combination 3: This combination was selected in conformance with practice in many
geographical zones, where an HMA mix with lower binder content (4%) and larger
maximum nominal aggregate size is used beneath the surface course and is referred to as
either the binder course or a stabilized base. A typical mix in MTO specifications,
designated as HL8, was investigated in this study. The depth of the surface course was
limited to 6 in. with a 6 in. HL8 layer, prepared using a PG 58-22 binder, introduced
56
beneath the HL3 layer. The aggregate base course was 14 in thick resting directly on the
native subgrade as shown in Figure 5.7. Results of the model run performed using details
of Combination 3 as inputs are shown in Table 5.3 and plotted in Figure 5.8. The results
indicated that this combination is superior to all others evaluated in this exercise where
predicted permanent deformations in all layers are lower.
AC: 6 in. HL3 – PG 70-22

AC: 6 in. HL8– PG 58-22
Base: 14 in. GB type A-3 (MR = 24 000 psi)
SG: A-7-6 (MR = 5000 psi)
According to analysis of model outputs, combinations 2 and 3 seem to meet the design
limits, and the user may make a final choice between the two options based on the cost
effectiveness of the design solution.
The exercise performed to investigate the three combinations demonstrates the
opportunity offered by the application of the proposed AASHTO Design Guide where an
evaluation of design alternatives is possible. No such capabilities are offered by any of
the current design approaches.
Table 5.3 Performance Predicted for the Three Design Alternatives
Combination AC Base Course Total Total

Thickness Thickness Rutting at Rutting at
(in.) (in.) 50% 90%
Reliability Reliability
1 10 18 0.98 1.18
2 10 18 0.66 0.83
3 12 18 0.57 0.73
57
1.20
Design Limit Control

Line
1.00
0.80
Rutting Depth (in)
Total Rutting at specified Reliability Level
0.60
Total Rutting
0.40
Subgrade Rutting
0.20
AC Rutting Base Rutting
0.00
0 24 48 72 96 120 144 168 192 216 240 264
Figure 5.8 Combination 3 that Satisfies the Design Limit
5.3 Unbound Material Characterization

Unbound material characterization based on the resilient modulus approach,
implemented in the proposed guide was examined in this study. In the absence of a
working module that captures the nonlinear behaviour of the material (currently being
redeveloped), only input Levels 2 and 3 were used in this study.
A conventional application of the model was used to gauge the sensitivity of the model to
changes in unbound material types with a different resilient modulus. The results are
discussed below.
Granular Aggregate Used as a Base Course
Properties of the granular material types, classified by AASHTO M145 (The
Classification of Soils and Soil-Aggregate Mixtures for Highway Construction Purposes)
as A-2-7, A-2-4, A-1-b, A1-a, were used as input to the model considering the road cross
section and materials shown in Figure 5.9. Other input parameters covering load
characteristics were also specified (medium traffic with an AADTT of 1000, operating
speed of 25 mph, and an MAAT = 67°F. Permanent deformations that accumulated in the
58
road as predicted by the model for the different material types attempted as a base course
are shown Table 5.4. Permanent deformation results that reflect the impact of changing
the resilient modulus of the granular base material are shown in Figure 5.10.
AC: 10 in. HL3 – PG 58-22
Base: 18 in. GB-MR Varies 21 500, 28 000, 36 000, 44 000

(psi)
SG: A-6 (MR = 15000 psi)
Figure 5.9 Pavement Structure with Different Granular Base Moduli
Table 5.4 Predicted Permanent Deformations Considering Different Base Modulus

(AC 10 in. thick, binder PG 58-22, Medium subgrade strength, MR 15 000 psi)
Base Modulus Total Rutting AC Rutting Base Rutting Subgrade
(psi) (in.) (in.) (in.) Rutting (in.)
21 500 0.82 0.60 0.09 0.13
28 000 0.82 0.61 0.08 0.13
36 000 0.81 0.62 0.07 0.12
42 000 0.81 0.63 0.06 0.12
59
AC: 10(in)-PG58-22
0.9 Base: 18 (in)
0.8 MR = Var.
Subgrade:
Permanent Deformation (in)
0.7 MR = 15000 psi

0.6
AC Rut
0.5 Total Rut
0.4 Base Rut
SubG Rut
0.3
0.2
0.1
0
0 10000 20000 30000 40000 50000
Base Modulus (psi)
Figure 5.10 Effect of Base Modulus on Permanent Deformation
The effect of changing the quality of the granular base material on the magnitude of
predicted permanent deformation is insignificant. A slight decrease in permanent
deformation is predicted in the base course with the increase in the base modulus.
However there seems to be no effect on the subgrade layer as the base layer modulus
changed. The result does not agree with basic rules of pavement mechanics where
changes in the base course modulus from 21 500 to 42 000 psi are expected to have a
considerable effect on the stresses transmitted to the subgrade.
Cohesive Native Soils Representing Road Subgrade Layers

Similar to the base course investigation discussed above, inputs pertaining to a number of
native soils that may exist in the road as a subgrade, were used to perform model runs
analyzing the structure depicted in Figure 5.11. The resilient modulus of these soils
ranged from 10 000 to 35 000 psi. The AASHTO classification for these materials
includes an A-7-5 (MR = 10 000 psi), an A-2-7 (MR = 21 500 psi), A-2-4 (MR = 28 000
psi) and A-2-4 (MR = 35 000 psi). These different materials and the permanent
60
deformations predicted for the structure shown in Figure 5.11 are listed in Table 5.5. The
results are plotted in Figure 5.12.
AC: 4 in. HL3 – PG 58-22
Base: 10 in. GB-MR = 38 500 psi
SG: var MR = 10 000, 21 500, 28 000, 35 000 (psi)
Figure 5.11 Pavement Structure with the Different Subgrade Moduli Investigated
Table 5.5 Predicted Permanent Deformations Considering Different Subgrade Modulus

(AC 4 (in.) thick –PG 58-22, base 10 (in.) MR = 38 500 (psi))
Subgrade
Total Rutting AC Rutting Base Rutting Subgrade
Modulus
(in.) (in.) (in.) Rutting (in.)
(psi)
10 000 1.19 0.84 0.10 0.25
21 500 1.15 0.85 0.10 0.21
28 000 1.11 0.84 0.09 0.17
35 000 1.08 0.84 0.09 0.14
61
1.4 AC: 4in-PG58-22
Base: 10 (in)
1.2 MR = 38500(psi).
Subgrade:
1 MR = Var (psi)
Total Rutting
0.8
AC Rutting
Subgrade Rutting
0.6
Base Rutting
0.4
0.2
0
0 10000 20000 30000 40000
Subgrade Modulus (psi)
Figure 5.12 Effect of Subgrade Modulus on Permanent Deformation
Results of model runs related to various soils evaluated as subgrades suggest the increase
in the modulus has a slight effect on the total permanent deformation predicted for the
pavement structure (10%). This effect is directly related to the permanent deformation
acquired in the subgrade. On the other hand, Figure 5.12 shows that the influence of the
subgrade modulus on the AC and granular base layers is insignificant. This result has
been expected, since the structural model employs linear elastic analysis, with permanent
deformation prediction depending on the empirical model, which may require further
calibration.
In summary, the model reflected insufficient sensitivity to unbound materials. Changes in
one unbound material do not reflect on the other unbound material in the structure (from
a granular base to a subgrade and vice versa). Accordingly, further investigation of
unbound material characterization application in the Design Guide was performed; the
results are offered in Subsection 5.12 comparing output of the different input levels
(comparison between input Levels 1 and 3).
62
5.4 Asphalt Concrete Characterization
This section investigates the AC characterization focusing on the HMA dynamic modulus
as implemented in the Design Guide. Permanent deformation in the pavement structure
associated with two mix types prepared using different binders was predicted to evaluate
the model sensitivity to variables associated with the AC mix. Initially, an HL3 mix
prepared with a PG 58-22 binder and an HL3 mix prepared with a PG 52-34 binder were
evaluated considering the road structure shown in Figure 5.13. The magnitude of
permanent deformations shown in Figure 5.14 reflects sensitivity of the model to
variations in the dynamic modulus of the mix.
AC: thickness 8 in., Mix 1: HL3-binder PG 58-22 / Mix 2: HL3-52-34
Base: Thickness 10 in. GB-A-1-a, MR = 38 500 psi
SG: MR = A-2-7, MR = 21 500 (psi)
Figure 5.13 Pavement Structure – Two AC Mixes
3.00
PG52-34
2.50
2.00
Rutting Depth (in)
1.50
Total Rutting Design Limit

1.00
0.50 PG58-22
0.00
0 24 48 72 96 120 144 168 192 216 240 264
Pavem ent Age (m onth)
Figure 5.14 Permanent Deformation of Two HMA Mixes
63
According to the results shown in Figure 5.14, permanent deformation of the HL3 mix
prepared with the PG 58-22 binder was substantially lower than that predicted for the
HL3 mix prepared with the PG 58-22 binder. Permanent deformation in the mix prepared
with the PG 58-22 binder evaluated at an average annual air temperature of 67° F,
exceeded the design limit set for this study (1.0 in.). Permanent deformation at the end of
the second month in the less stiff mix (prepared with PG 52-34 binder) is more than twice
that in the relatively stiffer mix prepared with the PG 58-22 binder. The results agree with
the design objectives intended by the producer of the binder where high stability at
relatively high temperatures offers adequate resistance to permanent deformations. The
mix with the PG 58-22 binder decreased substantially the accumulation of permanent
deformation in the course of the 20-year design life (70% less rutting compared with the
HL3 mix prepared with the PG 52-34 binder).
The Design Guide showed sufficient sensitivity to mix stiffness while predicting
permanent deformation potential. Sensitivity of the model for other HMA characteristics
is discussed below based on the results of evaluations performed using the appropriate
performance criteria and under relevant environmental and traffic loading conditions.
5.5 Influence of the Environment on Permanent Deformation

This section discusses sensitivity of the model to the effect temperature has on permanent
deformation. The climatic files for three cities with uniquely different mean annual air
temperatures (MAAT) were implemented in the model runs. The temperature and load
repetitions are the only implicit principal parameters in the empirical model used to
predict permanent deformation.
Results of permanent deformations predicted by the model for the three MAAT are listed
in Table 5.6 and plotted in Figure 5.15. Rutting that accumulated throughout the design
life of 20 years for the warm city (Dallas) was 80% more compared with that in the
relatively cold region (Minnesota). Clearly, the designer in the warm region will opt to
use a different mix design in order to reduce rutting below the design limit of 1.0 in. The
model proved capable of capturing the effect of the temperature on stiffness of the AC
layer.
Table 5.6 Predicted Permanent Deformations for Different MAATs

(AC 4 (in.) thick, PG 58-22, base 10 (in.) MR = 38 500 (psi))
Mean Annual Total Rutting AC Rutting Base Rutting Subgrade
Air Temp. (in.) (in.) (in.) Rutting (in.)
46 ºF 0.64 0.34 0.08 0.21
53 ºF 0.68 0.38 0.09 0.21
67 ºF 1.15 0.85 0.09 0.21
64
AC: 4in-PG58-22
1.4 Base: 10 (in)
MR = 38500(psi).
1.2 Subgrade:
MR = 21500 (psi)
1
AC Rutting
0.8 Total Rutting

Base Rutting
0.4
0.2
0
0 20 40 60 80
Mean Annual Air Temperature (F)
Figure 5.15 Permanent Deformation Evaluated for Different Mean Annual Air
Temperatures
65
5.6 Traffic Characteristics
Traffic speed
Sensitivity of the AC response to the loading rate makes traffic speed an important
parameter to incorporate in road structure analysis and design tools. The magnitude and
duration of stress pulses resulting from moving vehicles depend on the vehicle speed.
This effect diminishes toward deeper sub-layers. Lower frequencies from slower traffic
are associated with a lower AC modulus and, accordingly, higher permanent
deformations. To evaluate the sensitivity of the proposed guide to this phenomenon,
permanent deformation predictions were made using the input data shown in Table 5.7.
Model predictions are also shown in Table 5.7 and plotted in Figure 5.16 reflecting a
drop of 42% in total rutting associated with the increase in speed from 10 to 60 mph. The
results also indicate that the difference is mainly due to rutting in the AC layer with no
change in rutting magnitude in other layers associated with a change in speed. The model
is clearly sensitive to the dependency of the HMA response to the loading rate as
reflected in the increase in permanent deformation associated with the drop in vehicle
speed.
Table 5.7 Permanent Deformations Predicted at Different Traffic Speeds

(AC 4 in. thick, PG 58-22, granular base 10 in., MR =38 500 psi, subgrade MR =21 500)
(psi))
Traffic Speed Total Rutting AC Rutting Base Rutting Subgrade
(mph) (in.) (in.) (in.) Rutting (in.)
10 1.55 1.24 0.10 0.21
25 1.15 0.85 0.09 0.21
45 0.97 0.67 0.09 0.21
60 0.90 0.60 0.09 0.21
66
AC:4in-HL3-PG58-22
1.8 Base: 10 (in)
MR = 38500(psi).
1.6
Subgrade:
1.4 MR = 21500 (psi)
1.2
0.8 Total Rutting

AC Rutting
0.6
Base Rutting
0.2
0
0 20 40 60 80
Speed (mph)
Figure 5.16 Effect of Traffic Speed on Permanent Deformation
Traffic Volume
Traffic volume, quantified as the number of load repetitions, is known to influence the
magnitude of permanent deformations in AC roads. Different average annual daily truck traffic
(AADTT) values and other inputs shown in Table 5.8 were used to investigate the model
sensitivity to the number of load repetitions that reflect traffic volume differences. Predicted
permanent deformations are shown in Table 5.8 and plotted in Figure 5.17. Permanent
deformation increased with higher traffic volume. The AC layer absorbed the majority of the
impact of the increase in traffic volumes. The proposed Design Guide is noticeably sensitive to
the number of load repetitions, which is incorporated in the model as a principal input parameter.
67
Table 5.8 Permanent Deformations Predicted for different Traffic Volumes
(AC 6 in. thick –PG 58-22, base 10 in., MR =38 000 psi - subgrade MR =15 000 psi)
Traffic Volume Total Rutting AC Rutting Base Rutting Subgrade
AADTT (in.) (in.) (in.) Rutting (in.)
200 0.56 0.37 0.06 0.13
1000 1.05 0.80 0.069 0.18
2000 1.39 1.11 0.07 0.20
4000 1.86 1.55 0.08 0.23
2
1.8
1.6
1.4
AC Rutting
1.2
Total Rutting
1
Base
0.8
Subgrade
0.6
0.4
0.2
0
0 1000 2000 3000 4000 5000
Average Annual Daily Truck Traffic (AADTT)
Figure 5.17 Impact of Traffic Volume on Predicted Permanent Deformations
68
5.7 Construction Related Variables
The proposed Design Guide requires input for three as built volumetric properties related to AC;
the effective binder content percentage, air voids percentage and total unit weight. The effects of
these parameters are discussed below.
5.7.1 Influence of HMA Air Voids on Predicted Permanent Deformations
The impact of air voids content has been incorporated in the empirical predictive equation. They
are mainly used to construct the master curve for levels 2 and 3 inputs. The global aging system
model included in the Design Guide also uses the air voids to adjust the viscosity of the original
binder for in-service short and long-term aging.
The use of Level 1 AC input for a complex modulus yielded no changes in the output and the
investigation switched to the use of level 3. As constructed air voids of 4%, 6%, 8% and 10%
were used as input to examine the model sensitivity to variations in this important construction
variable (see Figure 5.18 for structural details and air voids investigated). Results of the runs are
shown below.
The model captures the effect of increasing the air voids by showing an increase in the
permanent deformation. The increase in air voids will result in lower mix stiffness. The voids
filled with binder have a direct impact on the mix tendency for permanent deformation.
AC: 6 in. HL3 – PG 58-22, air voids as built 4%, 6%, 8%, 10%
Figure 5.18 Pavement Structure for the Air Voids Study
Base: 10 in. GB-A-1-a, MR = 38 000 psi
SG: MR = A-7-5, MR = 15 000(psi)

Figure 5.18 Pavement Structure for the Air Voids Study
Table 5.9 Permanent Deformations Predicted for Different Air Void Contents
(AC 6 in. thick, PG 58-22, base 10 in., MR =38 000 psi, subgrade MR =15 000 psi)
AC Mix Subgade
Total Rutting AC Rutting Base Rutting
Air Voids Rutting
(in.) (in.) (in.)
(As Built)% (in.)
4 0.88 0.63 0.07 0.18
5.9 0.95 0.7 0.07 0.19
8 1.08 0.82 0.07 0.19
10 1.25 0.98 0.08 0.19
69
1.4
1.2
1
Total Rutting
0.8 AC Rutting
Base Rutting
0.6
Subgrade Rutting
0.4
0.2
0
0 2 4 6 8 10 12
Air Voids % (As Built)
Figure 5.19 Impact of Air Voids Content on Permanent Deformation
Influence of Effective Binder Content on Predicted Permanent Deformations

The predictive equation, used in levels 2 and 3 includes effective binder content as an input
parameter. The sensitivity of these two design levels to variations in effective binder content was
evaluated using three uniquely different mixes as shown in Table 5.10. Permanent deformation
predictions associated with these different effective binder contents are shown in Figure 5.20.
Permanent deformations increased with the increase in effective binder content. The entire
change is related to the AC layer as indicated by the results plotted in Figure 5.20.
70
Table 5.10 Permanent Deformations Predicted for Different Binder Contents
(AC 4 in. thick, PG 58-22, base 10 in., MR =38 000 psi, subgrade MR =15 000 psi)
AC Mix
Subgade
Total Rutting Binder AC Rutting Base Rutting
Rutting
(in.) Contents (in.) (in.)
(in.)
(As Built)%
0.86 8 0.61 0.07 0.18
0.95 10.5 0.7 0.07 0.19
1.05 14 0.79 0.07 0.19
1.2
1
0.8
Total Rutting
Ac Rutting
0.6
Base Rutting
Subgrade Rutting
0.4
0.2
0
0 5 10 15
Effective Binder Content % (As Built)
Figure 5.20 Effective Binder Content Impact on Permanent Deformation
71
5.8 General Comments Related to Model Predictions of Permanent Deformations
Model applications for predicting permanent deformation as the performance criteria to be used
in selecting a design proved effective for AC layers. The model showed sensitivity to the
thickness of the AC layer and to the mix stiffness associated with the binder used in the
manufacturing process. It has also responded clearly to changes in traffic volume and operating
speed. Construction related parameters such as mix air voids and effective binder contents have
shown to be effective in input level 3 rather than level 1. The predictive equation used in levels 2
and 3 includes these two parameters as a direct input that estimates the dynamic modulus.
Accordingly, Level 1 applications should be complemented by laboratory tests performed on
samples prepared at the as built air voids content and effective binder content.
Apparently, the model is less successful in addressing analysis requirements related to unbound
materials. The model lacks sensitivity toward variations made in this study to parameters known
to influence performance, such as the base course depth and the resilient modulus, where model
predictions of permanent deformation show little or no effect. These findings are discussed
further using different model application runs in the following sections. However, model
calibration reports suggest the performed calibration exercise for unbound materials did not
progress as anticipated. The calibration of the permanent deformation characteristics within each
major pavement layer could not be achieved due to the fact that LTPP data only provides the
total (surface) rutting present in the test section. The unbound base, subbase and subgrade rut
depth predictions were calibrated to a special study involving the use of numerous test sections
designed by the 1993 AASHTO Design Guide [42]. Recommendations were issued suggesting
that problems and limitations of the performed calibrations may be avoided benefiting from the
results of a local calibration stage to achieve a more reasonable degree of confidence.
The performance of the model is expected to improve when the nonlinear behaviour of unbound
materials is included in the analysis.
The NRC is performing an R&D program involving a modification to the current resilient
modulus test geared toward developing a process for modeling permanent deformation of
unbound materials [43]. Future research is expected to incorporate AC viscoelastic behaviour
and permanent deformation in the structural model and reduce dependency on the empirical
models included in the proposed guide.
5.9 Fatigue Cracking

Fatigue cracking is a main pavement distress affecting the performance of roads. Repeated traffic
loads induce tensile and shear stresses at critical locations causing cracks to initiate and then
propagate through the asphalt layer. Environmental conditions aggravate this damage type and
cause the pavement structure to lose its structural integrity. Results of a sensitivity analysis are
discussed in the following section to address the impact of the factors that affect fatigue cracking
in flexible pavements, such as layer thickness, AC dynamic modulus, binder grade, air voids in
the asphalt layer, effective binder content, base thickness, the subgrade modulus, traffic and
environmental conditions.
The pavement structure details shown in Table 5.1.1, subjected to a medium traffic load of 2000
AADTT and a medium traffic speed of 25 mph in a relatively warm climate (MAAT = 67°F)
was analyzed using the proposed Design Guide. The results of the model runs are discussed
below
72
Effect of AC Thickness on Bottom/Up Cracking
The magnitude of fatigue cracking associated with different AC layer thicknesses was
investigated. The results of varying the AC thickness from 2 inches to 12 inches are shown in
Table 5.11 and plotted in Figure 5.21. Fatigue cracking increased between 2 inches and 4 inches,
and then decreased with the increase in the AC layer thickness. The decrease in fatigue cracking
with the increase in layer thickness, above 4 inches, is a direct result of the decrease in tensile
strain at the bottom of the AC layer and, consequently, the amount of fatigue cracking. It is
interesting to note that fatigue cracking peaks at an AC layer thickness of 4 inches, which is
supported with observations documented in the literature [44]. The decrease in fatigue cracking
for thicknesses below 4 inches may be justified by the fact that at HMA thickness below the
maximum level between 3 and 5 inches, tensile strains actually decrease and may become
compressive in nature. Similar findings were reported [45] using the Kenlayer computer
program. They concluded that for thin asphalt surfaces, say less than 2 inches thick, the use of a
single tire results in a smaller tensile strain and is unsafe for the prediction of fatigue cracking.
Table 5.11 Predicted Fatigue Cracking (Bottom/Up) for Different AC Thicknesses

(AC var. thick –HL3-PG 58-22, base 10 in., MR =38 000 psi, subgrade MR =15 000 psi)
AC thickness in.
2 4 6 8 10
Fatigue cracking
6.8 13.2 3.4 1 0.3 0.1
%
% Decrease in
48 Reference 74 92 98 99
fatigue cracking
14
AC: HL3-PG58-22
12 Base: 10 (in)
MR = 38000 (psi)
Fatigue Cracking %
10 Subgrade:
MR = 15000 (psi)
8
0
0 2 4 6 8 10 12 14
AC Thickness (in)
Figure 5.21 Effect of AC Thickness Changes on Fatigue Cracking
73
Effect of Binder Properties on Fatigue Cracking (Bottom/Up)
Three HL3 mixes prepared with different performance grade binders (PG 58-22, PG 64-34, and
PG 52-34) were used to investigate the effect of the binder properties on the amount of fatigue
cracking. The dynamic modulus of the three mixes is shown in Figure 5.22. Predicted fatigue
cracking is shown in Figure 5.23. As expected, fatigue cracking decreased for mixes with high
binder stiffness. Using the softer binder PG 52-34 resulted in an increase of 40% in fatigue
cracking compared with PG 58-22. However, the use of the mix with the PG 64-34 binder
resulted in 45% higher fatigue cracking compared with the mix prepared with the PG 58-22
binder. Although the PG 64-34 was supposed to be stiffer, predicted performance is in agreement
with the laboratory determined dynamic modulus, which indicated that the PG 58-22 produced a
stiffer mix than that produced using the PG 64-34 binder.
30000
25000
20000
Modulus MPa
HL3-58-22
15000
HL3-52-34
10000
5000
HL3-64-34
0
-20 -10 0 10 20 30 40 50
Temp
Figure 5.22 Dynamic Modulus of HL3 Prepared with Different Binders
74
10
Alligator Cracking (%)
PG64-34 PG52-34
PG58-22
0
0 24 48 72 96 120 144 168 192 216 240 264
Figure 5.23 Fatigue Cracking Predicted for Three Mixes Prepared with Different Binders
Effect of Unbound Materials Characteristics Used as a BaseCourse

Properties of the different pavement layers influence its response (stresses and strains). Thus, the
quality of the granular base course and that of the subgrade acting as a supporting foundation are
important factors in evaluating road performance. This subsection evaluates the sensitivity of the
model to changes in the resilient modulus of the base course.
The AASHTO soil classification system classifies soils based on particle-size distribution and
Atterberg limits. Properties of a number of granular material types, classified by AASHTO
(M145) as A-2-7, A-2-4, A-1-b and A-1-a were used as input to the model considering the road
cross section and material properties shown in Figure 5.24. Other input parameters covering
load characteristic were also specified (medium traffic with an AADTT of 1000, operating speed
of 25 mph, and an MAAT = 67 °F). The percentage of fatigue cracking predicted by the model
for the different material types simulating the road base course are shown in Table 5.12. These
75
percentages reflect the impact of changing the resilient modulus of the granular base material as
shown in Figure 5.25.
The results indicate that fatigue cracking predicted by the model is very low, but still seem
sensitive to changes in the resilient modulus of unbound material below the AC layer. Percentage
reduction in the magnitude does not improve the overall performance of the road related to
fatigue cracking.
AC: 10 in HL3–PG 58-22
Base:18 in.GB-MR varies 21 500, 28 000, 36 000 and 42 000 psi
SG: A-6 -MR = 15 000 psi
Figure 5.24 Pavement Structure with Different Granular Base Modulus
Table 5.12 Predicted Fatigue Cracking (Bottom/Up) for Different Base Modulus Values
(AC 10 in. thick –HL3-PG 58-22 - Base 18 in. MR = Var. - Subgrade MR =15000 psi)
Base modulus psi

21 500 28 000 36 000 42 000
Fatigue cracking % 0.9 0.5 0.3 0.2

% Decrease in
Reference 44 67 78
fatigue cracking
76
1
0.9 AC:10 (in)-PG58-22
Base: 18 (in)
0.8 MR = Var.
0.7 Subgrade:
Fatigue cracking %
MR = 15000 (psi)
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10000 20000 30000 40000 50000
Base modulus MR (psi)
Figure 5.25 Effect of Base Quality on Fatigue Cracking (Bottom/Up)
Effect of Effect of Unbound Materials Characteristics Used as a Subgrade

Similar to the above base course investigation, inputs pertaining to a number of native soils that
may exist in the road as a subgrade were used to perform model runs to analyze the structure
depicted in Figure 5.26. The resilient modulus of these soils ranged from 5000 to 30 000 psi.
AASHTO classification for these materials includes an A-7-6 (MR = 5000 psi), A-7-5 (MR =
10000 psi), A-6 (MR = 15000 psi) and A-2-4 (MR = 30000 psi). These different materials and
the percentage of fatigue cracking predicted for the structure shown in Figure 5.26 are listed in
Table 5.13. The results are plotted in Figure 5.27.
AC: 6 in. HL3 – PG 58-22
Base: 10 in. GB-MR = 38 500 psi
SG: var MR = 5000, 10 000, 15 000, 30 000 psi
Figure 5.26 Pavement Structure with Different Subgrade Modulus
77
Table 5.13 Predicted Fatigue Cracking (Bottom/Up) for Different Subgrade Modulus
(AC 6 in. thick –HL3-PG 58-22, base 10 in., MR = 38000 psi)
Subgrade modulus
5000 10 000 15 000 30 000
psi
Fatigue cracking % 4.5 3.8 3.4 1.5
% Decrease in
Reference 15 24 67
fatigue cracking
5
AC:6 (in)HL3-PG58-22
Base: 10 (in)
Fatigue Cracking(bottom/Up) %
4 MR = 38000 (psi)
Subgrade:
MR Varies
3
0
0 5000 10000 15000 20000 25000 30000 35000
Subgrade Modulus (psi)
Figure 5.27 Effect of Subgrade Modulus on Fatigue Cracking (Bottom/Up)
Similar to the base course situation, the subgrade modulus has little effect on fatigue cracking.
Results of model runs related to various soils evaluated as subgrades suggest the increase in the
modulus has affected fatigue cracking as predicted by the model. An increase in the modulus
from 5000 psi to 30 000 psi resulted in a decrease of 67% in the predicted fatigue cracking.
Influence of the Environment on Bottom/Up Fatigue Cracking

The influence of the environment on fatigue cracking, focusing on temperature effect, was
investigated in this study using the database of the model that houses climatic files of different
cities. Climatic files of three cities with uniquely different mean annual air temperatures
78
(MAAT) were used to run the model. Results of fatigue cracking predicted by the model for the
three MAAT are listed in Table 5.14 and plotted in Figure 5.28.
Fatigue cracking in the relatively warm city is 40% more than that predicted for the relatively
cold city. The results are in agreement with the impact of temperature on the AC layer stiffness
where an increase in the air temperature makes the pavement more vulnerable to fatigue
cracking. However, the percentage of fatigue cracking estimated for the warmest city after the
design life is lower than what is considered a critical level (25%).
Table 5.14 Predicted Fatigue Cracking (Bottom/Up) – Different MAAT.

(AC 4 in. thick –HL3-PG 58-22, base 10 in., MR = 38 500 psi,
subgrade MR = 21 500 psi)
MAAT (oF) 46 53 67
Fatigue cracking % 10.1 12.6 14.2

% Increase in
Reference 25 40
fatigue cracking
16
14
AC: 4in-PG58-22
12 Base: 10 (in)
Fatigue Cracking %
MR = 38500(psi).
10
Subgrade:
8 MR = 21500 (psi)
0
35 40 45 50 55 60 65 70
o
Mean Air Annual Temp. F
Figure 5.28 Fatigue Cracking (Bottom/Up) Predicted for Three Cities with Different MAAT
79
Effect of Traffic Characteristics on Fatigue Cracking
The influence of traffic speed and volume on fatigue cracking was investigated in this study.
Traffic speed was used as input to perform runs at a medium traffic volume (2000 AADTT).
Traffic speeds of 10, 25, 45 and 60 mph were used interchangeably in the model. The structure
details are shown in Table 5.15. The climate selected is relatively warm (MAAT = 67 °F).
Results of the runs are shown in Table 5.15 and plotted in Figure 5.29. The results show a
decrease in the fatigue cracking associated with the increase in traffic speed. The decrease in the
magnitude of fatigue cracking reflects the concept that higher speeds result in higher loading
frequencies and, consequently, the dynamic modulus of the asphalt layer increases and the
induced tensile strains decrease.
Table 5.15 Predicted Fatigue Cracking (Bottom/Up) at Different Traffic Speeds

(AC 4 in. thick –PG 58-22, base 10 in., MR =38 500 psi, subgrade MR =21 500 psi)
Traffic Speed
10 25 45 60
Mph
Fatigue cracking
15.1 14.2 13.5 13.2
(bottom/up)%
% Decrease in fatigue
Reference 6 12 13
cracking
15.5
15
Fatigue Cracking %
14.5
14
13.5
13
0 10 20 30 40 50 60 70
Traffic Speed Mph.
Figure 5.29 Fatigue Cracking (Bottom/Up) Associated with Different Traffic Speeds
Traffic volume is another critical traffic characteristic that influences pavement damage.
Different truck volumes were used to investigate the model sensitivity to the number of load
repetitions that simulate different traffic volumes (see input in Table 5.16). The predicted
percentages of fatigue cracking are shown in Table 5.16 and plotted in Figure 5.30. The model
80
predictions are in agreement with the fundamentally established relationship between the number
of load repetitions and fatigue cracking.
Table 5.16 Predicted Fatigue Cracking (Bottom/Up) – Varying Traffic Volumes

(AC 6 in. thick, PG 58-22, base 10 in., MR =38 000 (psi), subgrade MR =15 000 psi)
ADDT
200 1000 2000 4000
Fatigue cracking
0.6 5.5 7 14
(bottom/up)%
% Change in fatigue
-89 Reference +23 +155
cracking
16
14
12
Fatigue cracking %
10
6
4
0
0 1000 2000 3000 4000 5000
Average Annual Daily Truck Traffic (AADTT)
Figure 5.30 Fatigue Cracking (Bottom/Up) Predicted for Different Traffic Volumes
Impact of Construction Related Variables on Fatigue Cracking

This subsection investigates the impact of air voids on fatigue cracking. The Percentage of air
voids, together with the percentage of effective binder content, are used in the Asphalt Institute
model [46] as an input for obtaining “the laboratory to field adjustment factor (C)” shown earlier
in Equation 3.5. As-constructed air voids of 4%, 6%, 8% and 10% were used in this study as
input to examine the model sensitivity to variations in this construction related variable. Results
of the performed runs are shown in Table 5.17 and Figure 5.31. As expected, high air voids
resulted in high fatigue cracking.
81
Table 5.17 Effect of AC Mix Air Voids (As Built) on Fatigue Cracking (Bottom/up)
(AC 6 in thick, PG 58-22, base 10 in., MR =38 000 psi, subgrade MR =15 000 psi)
Air Voids % 4 6 8 10
Fatigue Cracking
1.6 4.8 11.8 22.2
(Bottom/Up)%
% Increase in
Reference 67 637 1288
Fatigue Cracking
25
20
Fatigue cracking %
15
10
0
0 2 4 6 8 10 12
Air Voids % (As Built)
Figure 5.31 Effect of Air Voids (As Built) on Fatigue Cracking
To examine sensitivity of the model to changes in the effective binder content, as constructed
effective binder contents of 8%, 10.5% and 14% were used as input to the model. Results of the
runs are listed in Table 5.18 and plotted in Figure 5.32. The results suggest the effective binder
content has a slight impact on fatigue cracking. Increasing the binder content within a commonly
adopted range decreased the percentage of fatigue cracking in the road surface as shown in
Figure 5.32
82
Table 5.18 Effect of AC Mix Effective Binder Content (As Built) on Fatigue Cracking
(AC 6 in thick, PG 58-22, base 10 in., MR =38 000 psi, subgrade MR =15 000 psi)
Effective binder
8 10.5 14
content %
Fatigue cracking
9.4 4.8 2.4
(bottom/up)%
% Decrease in
Reference 49 74
fatigue cracking
10
9
8
Fatigue cracking %
7
6
5
4
3
2
1
0
0 4 8 12 16
Binder Content % (by volume)
Figure 5.32 Fatigue Cracking for Different Binder Contents
83
5.10 Comparison between Design Input Levels
The main objective of the hierarchical approach of the design input levels is to provide users
with a flexible design system that allows them to use the software of the proposed Guide
depending on availability of data and local resources, taking into consideration the complexity of
the project. Concerns may arise as to whether pavement engineers can start applying Level 3 and
later progress to analysis with the ideal approach (Level 1) when testing resources in Canada
match model requirements. The impact of using Level 3 on the accuracy of the analysis is not
well known. This section discusses results of analysis using data produced with different design
input levels.
Impact of Using AC Design Inputs Level 1 and Level 3 on Performance Prediction
To investigate the impact of using level 1 and level 3 on pavement performance prediction, six
model runs were conducted. The first three runs were performed with Level 1 input data using
the complex modulus of HL3 mix determined in the laboratory. The three performance graded
binders, shown earlier in Table 4.5 were used to prepare the mix design. The other three runs
were performed using Level 3 for the same mixes. The pavement structure used to perform the
analysis is shown in Figure 5.33 together with other input parameters shown in Chapter 4.
Analysis was performed for a design life of 20 years in a relatively warm climate. The predicted
total deformations using the two levels of input are shown in Figure 5.34. Permanent
deformation varied considerably depending on the binder type used. In general, Level 3 analysis
tends to under predict permanent deformation as shown in Figure 5.34.
AC: 6 in. HL3 – PG 58-22, PG 52-34, PG 64-34
Base: 10 in. GB type A-1-b (MR = 38 000 psi)
SG: A-7-5 (MR = 15 000 psi)
Figure 5.33 Pavement Structure – Comparison of Design Levels
The calculated difference based on level 1 ranged from 10% for the mix made with the PG 58-22
binder to 60 % for the mix made with the PG 52-34 binder. Level 1 seems to be able to
discriminate between the evaluated binder types more than Level 3.
On the other hand, using Level 3 resulted in 40% overestimation of fatigue cracking for mix
made with the PG 58-22 binder and by 10% for the mix made with PG 52-34 binder as shown in
Figure 5.35.
84
3.5
3
2.5
1.5
0.5
1 3
0
HL3-PG58-22 1
HL3-PG52-34 HL3-PG64-34
Figure 5.34 Permanent Deformation Prediction Using Levels 1 and Level 3
5
Fatigue Cracking %
2
1 3 1 3 1 3
1
HL3-PG58-22 HL3-PG52-34 HL3-PG64-34
Figure 5.35 Fatigue Cracking Prediction Using Level 1 and Level 3
85
The difference in pavement performance predictions is apparently due to the accuracy of
material properties used in each modeling approaches associated with the two design levels.
With level 1 design, the model uses complex modulus values established in the laboratory,
determined at relevant temperatures and loading frequencies for the specific mix. Level 3 uses
the empirical model incorporated in the Design Guide to calculate the dynamic modulus of AC
mix using physical properties as outlined in Equation 4.1[47].
5.11 Measured vs. Predicted Dynamic Modulus

The substantial difference in performance predictions between Level 1 and 3 was discussed in
the previous subsection looking into the difference between dynamic modulus measured in the
laboratory and that estimated by the prediction equation used in Level 3. The dynamic modulus
values were determined in the laboratory simulating cold, moderate and warm operating
temperatures. The selected loading frequencies (0.1, 1 and 20 Hz) represent slow, medium and
high speeds. The ability of the dynamic modulus to discriminate between two mixes prepared
with uniquely different binder types (PG 64-34 and PG 58-22) is demonstrated in Figure 5.36.
The response of the engineered binder (PG 64-34), compared with the neat binder, suggests the
objectives set for the binder design, aimed at reducing its brittleness at low temperatures to
minimize the potential for cracking, are fulfilled. The measured dynamic modulus of the HL3
mix prepared with a PG 64-34 binder is half that prepared with a PG 58-22 binder (see Figure
5.36). However, both binders produced similar dynamic modulus values at relatively high
temperatures, reinforcing the widely accepted role played by the aggregate skeleton at such a
temperature condition.
The level 3 inputs (mix gradation and binder type) pertaining to the same mixes were entered in
the model to estimate the dynamic modulus using the empirical predictive equation incorporated
in the model. Dynamic modulus values predicted for the different temperatures for the two
binders are shown in Figure 5.36. The predictive equation managed to correctly rate the mix
response at low temperatures as influenced by the properties of the two binders. However, the
predictive equation was less successful in quantifying the difference between the two binders as
measured in the laboratory using the complex modulus test. The predicted responses of the two
binders suggest a difference of less than 10% whereas the measured response reflected a
difference of 100%.
86
25000
HMA 58-22 - Measured

HMA 64-34 - Measured
20000 HMA 58-22 - Predicted
HMA 64-34 - Predicted
Dynamic modulus (MPa)
15000
10000
5000
0
-20 -10 0 10 20 30 40 50
Temperature (C)
Figure 5.36 Measured and Predicted Modulus for Two HL3 Mixes with Different Binders
Data points representing all mixes examined in this study produced using the predictive equation
and laboratory measurements were plotted in Figure 5.37. A comparison between the two
approaches represents an excellent opportunity to evaluate the ability of the predictive equation
to estimate the dynamic modulus, at least for the specific mixes used in the study. Comparison
between estimated and measured values was conducted by examining the distribution of data
points along the equality line drawn at 45o. Points located above this line indicate the equation
over-predicted the dynamic modulus value and those below suggest under-prediction. Actual co-
ordinates of data points (measured and predicted) were used in this study to quantify the
deviation of predicted values from those measured in the laboratory at specific conditions
associated with speed and temperature. Lines that represent different percentages of deviations
from the equality line were used to highlight deviation observed under the different conditions.
The results plotted in Figure 5.37 show that the predictive equation over-estimated the value of
the dynamic modulus under conditions of relatively high temperature and low loading frequency
(low speed). These conditions are marked in Figure 5.37 at measured dynamic modulus values
lower than 5000 MPa. The deviation reached as high as 300% under these conditions. Data for
measurements made above the 5000 MPa modulus value are scattered above and below the
equality line with prevalence of over-estimation, which reached as high as 100%. The average
percent error calculated considering all data points produced by the predictive equation was 77%.
87
Predictive equation
45 degree line
30000 40% underprediction
Predicted dynamic modulsu (MPa)

100% over-prediction line
300% over prediction line
20000
10000
0
0 10000 20000 30000
Measured dynamic modulus (MPa)
Figure 5.37 Predicted vs. Measured Dynamic Modulus
88
5.12 Unbound Materials Inputs for Level 1 and 3
The need for considering non-linear material behaviour in unbound materials is as
important as the case in AC materials. The results of pavement analysis using linear
elastic approximations of unbound material behaviour are no longer acceptable in
pavement analysis. Errors from such approximations have been noted and documented
[48]. Dependency of the response on the state of stress in soil-based materials must be
considered for effective analysis of road pavements. Repeated load resilient modulus test
has been recommended as the primary characterization method for unbound materials in
the proposed AASHTO Design Guide.
Different levels of input are incorporated in the proposed Guide as a mean for obtaining
the resilient modulus of unbound layers. Level 1 resilient modulus is determined from
laboratory testing of representative material samples. Two standard test methods are
recommended in the Guide (NCHRP 1-28A and AASHTO 307). Results of the
laboratory test are used to determine the non-linear elastic coefficients and exponents of
the modified universal model [21] described by Equation 2.9.
Currently, no access has been offered in the model to Level 1 input and, hence, no
evaluation has been performed in this study to examine the proposed characterization of
unbound materials at this level. At Level 2 inputs, the resilient modulus may be estimated
using certain correlations established in the model between the resilient modulus and
physical material properties, such as the CBR, R-value. At Level 3 inputs, typical
resilient modulus values and ranges are recommended based on AASHTO and unified
soil classifications.
In the absence of laboratory resources for conducting the resilient modulus, Canadian
users of the proposed Guide are expected to rely on Levels 2 and 3. The effectiveness of
the approach base on using physical properties as indicators of the expected mechanical
response of unbound materials was evaluated in this study. The study examined two
granular materials (A and B), which were tested in the laboratory using standard
AASHTO techniques for evaluation of physical and mechanical properties. Relevant
properties for the two materials, considered suitable for the construction of a base course,
are shown in Table 4.10. Based on the determined physical properties, these materials are
classified as A-1-a according to AASHTO specifications. Using input level 3 and
entering the material class, one resilient modulus range will be proposed. Commonly,
users are expected to select the average of the two limits. However, laboratory
experiments conducted in this study revealed that the measured resilient modulus of the
two materials differed substantially from the recommended average (see Table 4.10).
Actually, the measured modulus of the two materials was found to be outside the
proposed range, one above and one below the two limits.
To highlight the impact of the difference between the measured and recommended
modulus, the software of the proposed AASHTO Design Guide was used to predict
performance levels associated with the four different moduli. Base rutting and bottom-up
fatigue cracking were selected as the performance criteria for the software applications.
Base rutting predictions are shown in figures 5.38 and 5.39. Base rutting predictions were
89
different for the different modulus values used as input. The application of the two
resilient moduli limits proposed in the guide resulted in the prediction of relatively low
rutting, which is similar for both materials. However, the application of the modulus
measured in the laboratory for Granular A resulted in a base rutting magnitude that is
25% higher than that produced based on applying modulus values estimated by the guide
(see Figure 5.38).
4.5
3.5
3
Rutting depth (mm)
2.5
2
Granular A- measured
1.5
Estimated Lower Limit
Estimated Upper Limit
1
0.5
0
0 50 100 150 200 250
Pavement age (month)
Figure 5.38 Base Rutting for Granular A
Similarly, the application of the modulus measured in the laboratory for Granular B
resulted in a granular base course rutting that is 20% lower than that produced based on
applying modulus values estimated by the guide (see Figure 5.39). Accordingly, it
possible to conclude that reliance on the AASHTO classification to estimate the resilient
modulus, as suggested in the proposed Guide, may under- or over-estimate permanent
deformations.
90
3
2.5
2
Rutting depth (mm)
1.5
Granular B- measured
1
Estimated Upper Limit
0.5
0
0 50 100 150 200 250
Figure 5.39 Base Rutting for Granular B
Similarly, analysis based on another performance criteria, namely fatigue cracking,

produced high error rates when resilient modulus values estimated by the proposed
Design Guide were used instead of actual values measured in the laboratory (see figures
5.40 and 5.41). For granular A, an under-estimation of 45% was predicted (see Figure
5.40) and for granular B and over-estimation of 77% was noted (see Figure 5.41).
In summary, the application of Level 3 to predict performance may produce erroneous

results and lead to under or over design of unbound pavement layers. A research project,
currently underway at the NRC, is pursuing an alternative approach based on using
generic material properties and applying level 1 instead of Level 3 which relies on the
predictive equation. These generic values may be produced using a limited number of
tests performed on typical materials used locally.
91
18
16
Granular A-measured
14 Estimated Lower Limit
Estimated Higher Limit
12
Fatigue Cracking %
10
0
0 50 100 150 200 250
Figure 5.40 Fatigue Cracking for Granular A
4.5
4 Granular B- measured
3.5 Estimated Higher Limit
3
Fatigue cracking %
2.5
1.5
0.5
0
0 50 100 150 200 250
Figure 5.41 Fatigue Cracking for Granular B
92
CHAPTER 6
Conclusions and Recommendations
The proposed AASHTO Design Guide represents a radical change in the way pavements
are analyzed and designed. The implementation of mechanistic principles in parts of the
analytical model promotes design based on a relatively accurate assessment of the state of
stresses and strains, compared with earlier versions of AASHTO guides. However, this
first attempt at producing a mechanistic approach will require further improvements to
produce reliable design solutions and to address user expectations related to a variety of
issues encountered by road engineers.
(a) The structure module still relies on linear elastic analysis in contrast to typical field
observations suggesting that the response is viscoelastic/plastic in nature. With the
implementation of linear elasticity, it will not be possible to predict mechanistically
components of the strain that translate into permanent deformations (rutting and
cracking). The proposed model performs structural analysis to determine the elastic
response and then proceeds to estimate the magnitude of various distress types, using
empirical performance prediction models. Verification of the accuracy of the different
modules and their calibration will take considerable time. However, the design of the
software, which involves separate modules for the different functions, will make it
easy to replace those proven inaccuracies with improved versions.
(b) Evaluation of the structural model was not possible in this study, because the
calculated stresses and strains were not displayed in the current beta version of the
proposed AASHTO 2002 model.
(c) Although the design software refers to a finite element model for capturing non-linear
response of unbound pavement layers, the current version is blocked, because of
some adjustment proposed by the reviewers of the beta version. Accordingly, Level 1
inputs were used for AC and only Level 3 inputs for unbound materials were used in
this study. Results of model runs performed in this study to evaluate the sensitivity of
the proposed Design Guide to variations in parameters with a known impact on
performance, reflect the following.
Model applications for predicting permanent deformation, as the performance

criteria to be used in selecting a design proved to be effective for AC layers and
showed sensitivity to the thickness of the AC layer and to the mix stiffness
associated with the binder used in manufacturing the AC mix. They also
responded clearly to changes in traffic volume and operating speed. Construction
related parameters, such as the percent of air voids in the mix and effective binder
contents are effective in input Level 3 rather than Level 1. The predictive equation
used in Levels 2 and 3 included the percent of air voids and the mix effective
binder contents parameters as direct inputs that contribute to the determination of
the dynamic modulus. Accordingly, Level 1 applications should be complemented
93
by laboratory tests performed on samples prepared at the as built air voids content
and effective binder content.
Apparently, the model is less successful in addressing analysis requirements
related to unbound materials. The model lacks sensitivity toward variations made
in this study to parameters known to influence performance, such as the base
course depth and the resilient modulus, where model predictions of permanent
deformation showed little or no effect.
The use of nonlinear analysis is recommended in mechanistic-based models to

capture the real behaviour of unbound materials. The performance of the model is
expected to improve when the nonlinear behaviour of unbound materials is
included in the analysis and the model recalibrated to local conditions taking into
consideration the contribution of field individual layers to the total permanent
deformation.
The NRC is performing R&D program involving modifications to the current

AASHTO resilient modulus test protocol geared toward developing a process for
modeling permanent deformation of unbound materials. Future research effort is
expected to incorporate AC viscoelastic behaviour and permanent deformation in
the structural model and reduce dependency on the empirical models included in
the proposed Design Guide.
Sensitivity to climatic conditions has been demonstrated in the higher rutting

accumulated in warm regions compared with relatively cold regions.
The model proved sensitive to variables with known influence on the resistance of
AC mixes to fatigue cracking. This type of fatigue related damage is well defined in
the literature and most models use an approach similar to that adopted in the
proposed Guide.
However, it is essential to calibrate carefully the transfer function by applying
appropriate calibration parameters so the predicted distress can match field
observations. Such a process will be needed in the calibration of the proposed
Guide.
Additional research is needed to address issues related to fatigue top-down cracking

and for the development of mechanistic based models to replace the one currently
incorporated in the Design Guide. Construction related variables should be studied
to quantify their contribution to the deterioration of pavement structures. It is
necessary to identify the effects of relevant construction practices, in particular
compaction methods, as potential factors that influence the evolution of pavement
cracking.
Research is needed to address issues related to reflection cracking and to develop
mechanistic-based models for use in the Design Guide. Such an approach will
improve analysis and design of HMA overlays.
Considering the fact that current testing facilities are not capable of satisfying
requirements of the resilient modulus and dynamic modulus tests, users of the
model will rely mainly on level 3 inputs for both the AC and unbound materials.
94
Because of the approximate nature of the mechanistic response used in Level 3,
which predicts the modulus using physical properties, the results of the evaluation
highlight the importance of using a well-executed material characterization process
(Level 1) to the success of the proposed AASHTO Design Guide. The following
conclusions are based on the results of the evaluation performed in this study.
The model reflected sensitivity to variation in AC mix types with unique physical
and mechanical properties. Performance predictions produced using the proposed
AASHTO Design Guide while using laboratory-measured dynamic modulus
values as input Level 1 are in agreement with performance patterns established by
the current practice and those reported in the literature.
However, AC dynamic modulus estimates made using the predictive equation
incorporated in the Guide proved to be substantially different from the measured
values. The average percent error calculated considering data produced by the
predictive equation vs. measured dynamic modulus amounted to 77%. The error
in estimating the modulus (input Level 3) led to underestimation of accumulated
damage related to permanent deformation by 10% to 60% depending on the
binder used in the mix, which consequently results in undersigning the road
structure.
Similarly, input Level 3 for unbound materials, mainly based on the correlation
between physical properties (including the AASHTO classification) and the
resilient modulus, produced unreliable values when compared with actual
measurements made in the laboratory. Applications based on a modulus estimate
obtained using the Guide proposed values to run the software resulted in
substantially different performance predictions compared with those produced
using measured modulus values. This study has shown a difference that ranges
from -20% to +25% in the prediction of unbound layer rutting.
It is highly recommended to use a laboratory-measured dynamic modulus as the first

choice. However, in the absence of adequate facilities and enough expertise for
conducting the tests necessary to measure the required mechanical properties, reliance
will be on performing design exercises using input Levels 2 and 3. To produce more
reliable pavement design solutions using these two levels, this study proposes a
scheme for building a database (Material Library) for potential Canadian users of the
proposed Guide. The Library may be populated with generic dynamic modulus values
of local AC mixes and resilient moduli of commonly used unbound materials.
Implementation of such generic properties in running the proposed design software
will result in lower design errors compared with those produced using moduli
estimated from correlations established for the proposed Guide. These minor errors
associated with the use of generic properties may be minimized by introducing
additional results to the database from tests performed on possible local alternatives
that reflect variations in the real practice.
95
(d) The accuracy of distress prediction models depends on effective calibration and
validation using reliable sources of data. The review of the calibration studies
associated with the models incorporated in the Guide has been impeded by the lack of
required information in the LTPP as well as the CSHRP sections. It is imperative to
plan and identify data needs to calibrate and validate mechanistic-based models and
utilize advanced instrumentation to monitor the performance of individual pavement
layers.
96
REFERENCES:
1. Statistic Canada Agency, “Local Government Financial Management

Statistics-Capital Expenditures”, Index 1733, 2003.
2. David Croney, Paul Croney. Design and Performance of Road Pavements,
Mc-Graw Hill.
3. Belgian Road research Center, “Advanced Models for Analytical Design of
European Pavement Structure”
4. Guide for Mechanistic-Empirical Design of New and Rehabilitated Pavement
Structures – AASHTO 2002. American Association of State Highway and
Transportation Officials, Washington, D.C.
5. The Asphalt Institute, “Thickness Design, Asphalt Pavements for Highways
and Streets” Manual Series No 1, The Asphalt Institute, Lexington, KY,1991.
6. National Crushed Stone Association, “Flexible Pavement Design Guide for
Highways”, NCSA Publication, Washington D.C., 1972.
7. Department of Transportation U.S.,“ AASHO Interim Guide for the Design of
Rigid and Flexible Pavements” published by the American Association of
State Highway officials, Washington D.C. 1961
8. Department of Transportation, U.S.,“ AASHTO Interim Guide for the Design
of Pavement Structures” published by the American Association of State
Highway and Transportation officials, Washington D.C. 1972
9. Department of Transportation, U.S.,“ AASHTO Guide for the Design of
Pavement Structures” published by the American Association of State
Highway and Transportation officials, Washington D.C. 1986
10. Department of Transportation, U.S., “AASHTO Guide for Design of
Pavement Structures (1993)”. American Association of State Highway and
Transportation Officials, Washington, D.C. 1993
11. Transportation Association of Canada (TAC), Canadian Strategic Highway
Research Program (C-SHRP), “Pavement Structural Design Practices Across
Canada” , C-SHRP Technical Brief No. 23, Ottawa, Ontario, April2002.
12. Asphalt Institute, “SuperPave Mix Design Manual”, SuperPave Series No. 2
(SP-2), 3rd Edition, 2001.
13. The Asphalt Institute, “Mix Design Methods for Asphalt Concrete and Other
Hot-Mix Types”, Manual series No. 2 (MS-2) 6th edition 1997.
14. Mase, George Thomas. “Continuum Mechanics for Engineers”, CRC Press,
2nd edition
15. Charles W. Schwartz (2005). Evaluation of the Witczak Dynamic Modulus
Prediction Model. TRB Annual meeting 2005.
16. Hopman, P.C, “VEROAD: A Viscoelastic Multilayer Program”, 75th TRB
meeting, Transportation Research Record 1539, Washington D.C. USA, 1996,
pp. 72-80.
17. Sayegh, G. 1967. Viscoelastic Properties of Bituminous Mixtures.
Proceedings of the 2nd International conference on structural design of
asphalt pavement, pp. 743-755. Held at Rackham Lecture Hall, University of
Michigan, Ann Arbor, USA
97
18. Hopman, P.C, “VEROAD: A viscoelastic multilayer program”, 75th TRB
meeting, Transportation Research Record 1539, Washington D.C. USA, 1996,
pp.72-80.
19. Design Guide Documentation. Appendix HH Field Calibration of the Thermal

Cracking Models.
20. A.P. S. Selvadurai. Elastic Analysis of Soil Foundation Interaction”. Elsevier

Scientific Publishing Company, New York, 1979- ISBN 0-444-41663-3
(vol.17).
21. Dragos Andrei, Mathew W. Witzack, Charles W.Schwartz, and Jacob Uzan.
Harmonized Resilient Modulus Test Method for Unbound Pavement
Materials. Transportation Research record: Journal of the Transportation
Research Board, No. 1874, TRB, National Research Council, Washington,
D.C., 2004, pp.29-37
22. Witzack, M. W., and J. Uzan. The universal Airport Design System, Report I
of IV: Granular Material Characterization. Department of Civil Engineering,
University of Maryland, College Park, 1988.
23. Yang H. H. Pavement Analysis and Design. Prentice Hall, Englewood Cliffs,
New Jersey 1993.
24. American Association State Highway and Transportation Officials
(AASHTO), “Resistance to Plastic Flow of Bituminous Mixtures Using
Marshall Apparatus”, AASHTO Designation: T245.
25. Ontario Ministry of Transportation (MTO), Ontario Provincial Standard
Specification, OPSS 1149-1152, Ottawa, Ontario, July 1990
(AASHTO), “Short and Long Term Aging of Hot Mix Asphalt”, AASHTO
Designation: PP2
27. American Society for Testing and Materials (ASTM), “Standard Test Method
for Dynamic Modulus of Asphalt Mixtures”, ASTM D3497-79 (1995)
(AASHTO), “Standard Method of Test for Bulk specific gravity of
Compacted Bituminous Mixtures Using Saturated Surface-Dry Specimens”,
AASHTO Designation: T166-94
(AASHTO), “Standard Method of Test for Theoretical Maximum specific
gravity and Density of Bituminous Paving Mixtures”, AASHTO Designation:
269-97
30. American Society for Testing and Materials. “ASTM D 2493 Viscosity-
Temperature Chart for Asphalts,” 1998 Annual Book of ASTM Standards, Vol.
0.403, pp. 230-234.

(AASHTO), “Classification of Soils and Soil-Aggregate Mixtures for
Highway Construction Purposes”, AASHTO Designation: M 145
98
32. Design Guide Documentation. Appendix II Calibration of Fatigue cracking
models for flexible pavements.
33. Interim Report: Chapters 1 and 2, NCHRP 1-42: Top-Down Fatigue Cracking
of Hot –Mix Asphalt Layers. Advanced Asphalt Technologies, LLC. May
27,2004
34. Design Guide Documentation. Appendix HH Field Calibration of the Thermal
Cracking Models.
35. Waldhoff, A.S., Buttlar, W.G., and J. Kim "Evaluation of Thermal Cracking
at Mn/ROAD Using the Superpave IDT, Proceeding of the Canadian
Technical Asphalt Association, 45th Annual Conference, Winnipeg, Manitoba,
Polyscience Publications, Inc., Laval, Quebec, Canada, pp. 228-259, 2000.
36. Abd El Halim, A. O. 1986. Experimental and Field Investigation of the
Influence of Relative Rigidity on the Problem of Reflection Cracking. For
presentation and publication at the 1986 TRB Annual Conference,
Washington, D.C. U.S.A.
37. Abd El Halim, A. O. (1985). “Influence of relative rigidity on the problem of
reflection cracking.” Transportation Research Record 1007, Transportation
Research Board, Washington, D.C., 53- 58
38. Abd El Halim, A. O., Phang, W., and Hass, R.C. (1987). “Realizing structural
design objectives through minimizing of construction induced cracking.”
Proc., 6th Int. Conference on Structural Design of Asphalt Pavements, Univ.
of Michigan, Ann Arbor, Mich., Vol 1, 965-970.
39. Abd El Halim, A. O., Ralf Haas “Process and case illustration of construction
Innovation” Journal of construction Engineering and management
ASCE/July/August 2004
40. Abd El Halim, A. O., Ralf Haas “Process and case illustration of construction
Innovation” Journal of construction Engineering and management
ASCE/July/August 2004
41. Zubair Ahmed, Ivana Marukic, Sameh Zaghloul, Nick Vitillo, “ Validation of
Enhanced Integrated Climatic Model Predictions Using New Jersey Seasonal
Monitoring Data” TRB 84th annual Meeting- January,2005.
42. Design Guide Documentation. Appendix GG- Calibration of Permanent
Models for Flexible Pavements.
43. Walaa E.I. Khogali and ElHussein H. Mohamed. “Novel Approach for
Characterization of Unbound Materials”. Transportation Research record:
Journal of the Transportation Research Board, No. 1874, TRB, National
Research Council, Washington, D.C., 2004, pp.38-46
44. Design Guide Documentation. Appendix II Calibration of Fatigue cracking
models for flexible pavements.
45. Yang H. Huang (1993). Pavement Analysis and Design. Prentice-Hall, Inc,
New Jersey
46. Asphalt Institute. 1991. Thickness Design – Asphalt Pavements for Highways
and Streets. Manual Series No.1 (MS-1), Lexington, KY.
99
47. O.A. Fonseca and M.W. Witczak, “A Prediction Methodology for the
Dynamic Modulus of In Place Aged Asphalt Mixture”, Journal of the
Association of Asphalt Paving Technologist, vol. 65,1996, pp. 532-565.
48. “Assessing Pavement Layer Condition Using Deflection Data”, (NCHRP 10-
48), Transportation Research Board, Lockbox 289, Washington, DC 20055.
100
APPENDIX A
Asphalt Concrete and Unbound Materials

Inputs and Related Tests Specifications
101
Asphalt Mix
Data Input Level

Property cedure
1 2 3
ret ¾”; ret ¾”;
3 3 AASHTO
/8”; #4; /8”; #4;
Gradation T30
pass #200 pass #200
AASHTO
Stiffness Master E*
Curve TP62
Asphalt Binder
Data Input Level

Property Procedure
1 2 3
AASHTO
Layer 1 – Asphalt Concrete

Performance Binder Test Result Test Result
T315
Conventional Binder
AASHTO
Structure
Viscosity Test Result Test Result T201, T202,
T316
AASHTO
Penetration Test Result Test Result
T49
Specific Gravity Test Result Test Result ASTM D 70
Softening Point Test Result Test Result ASTM D 36
Performance Grade A & VTS Select Grade
Viscosity Grade A & VTS Select Grade
Penetration Grade A & VTS Select Grade
Asphalt General
Data Input Level

Property Procedure
1 2 3
M-E Design
Poisson’s Ratio Calculated Calculated Calculated
Guide
AASHTO R
Vbe Calculated Calculated Calculated
35
AASHTO
Va Test Result Test Result Test Result
T269
AASHTO
Unit Weight Test Result Test Result Test Result
T166
102
ASTM E
Thermal Conductivity Test Result
1952
ASTM D
Heat Capacity Test Result
2766
Thermal Cracking
Property Data Input Level Procedure
AASHTO
Indirect Tension @ 3 temps @ 1 temp Defaults
T322
Mix Coefficient of
Thermal Contraction
AASHTO
VMA Test Result Test Result Test Result
R35
ASTM
Aggregate CTC Test Result Test Result Test Result
D4535
Level 1
Typical
Property Test Procedure Range of
Drainage Values
& Surface Weathered Asphalt
Property Surface Pavement 0.80 – 0.90
Shortwave n/a New Asphalt
Absorptivity Pavement 0.90 – 0.98
Asphalt Mix
Property Test Procedure Typical
Range of
Values
Structure
Master Stiffness Dynamic Modulus AASHTO TP 62-03 Varies

Curve (E*)
Asphalt Concrete Layer
Asphalt Binder
Range of
Values
Performance G* & δ, after AASHTO M
AASHTO T 315
Binder RTFO 320
AASHTO T 201, T
202, AASHTO M
Viscosities
T 316 or ASTM 226
Conventional D4402
Binder AASHTO M
Penetration AASHTO T 49
20
Specific Gravity ASTM D70 Varies
Softening Point ASTM D36 Varies
103
Asphalt General

Range of
Values
Poisson’s Ratio Calculation M-E Software 0.15 - .50
Vbe Calculation AASHTO R 35 Varies
Va Air Voids in AASHTO T 269 Varies
Compacted HMA
Unit Weight (γ) Gmb * 62.4 AASHTO T 166 Varies
Thermal HMA Thermal ASTM E 1952 0.67
Conductivity Conductivity BTU/hr-ft-
ºF
Heat Capacity HMA Heat ASTM D2766 0.23
Capacity BTU/lb-ºF
Property Test Procedure
Indirect Tension Creep & Strength AASHTO T 322 @ 3 Varies
Thermal Tests temps
Cracking Mix Coefficient of VMA AASHTO R 35 Varies
Thermal Aggregate CTC ASTM D 4535 21 to 37
Contraction /oC
Level 2
Typical
Values
Drainage
&
Surface
Weathered Asphalt 0.80 –
Property
Surface Pavement 0.90
Absorptivity Pavement 0.90 –
0.98
Asphalt Mix
Structure
Typical
Values
Aggregate ret 3/4”; ret 3/8”;
AASHTO T 30 Varies
Gradation ret #4; pass #200
Layer
Asphalt Binder
Typical
Values
Performance AASHTO
G* & δ, after RTFO AASHTO T 315
Binder M 320
AASHTO T 201, T
Conventional 202, AASHTO
Viscosities
Binder T 316 or ASTM M 226
D4402
104
AASHTO
Penetration AASHTO T 49
M 20
Specific Gravity ASTM D70 Varies
Softening Point ASTM D36 Varies
Asphalt General
Typical
Values
Air Voids in AASHTO T 269
Va Varies
Compacted HMA
HMA Thermal ASTM E 1952 0.67
Thermal
Conductivity BTU/hr-ft-
Conductivity
ºF
HMA Heat Capacity ASTM D2766 0.23
Heat Capacity
BTU/lb-ºF
Typical
Values
Thermal Indirect Tension AASHTO T 322 @ 1
Cracking
Creep & Strength Varies
Tests temp
Mix Coefficient of VMA AASHTO R 35 Varies
Contraction /oC
Level 3
Typical
Values
Drainage
&
Surface
Weathered Asphalt 0.80 –
Property
Surface Pavement 0.90
Absorptivity Pavement 0.90 –
0.98
Structure
Asphalt Mix
Typical
Values
ret 3/4”; ret 3/8”;
Layer
Aggregate
ret #4; pass #200 AASHTO T30 Varies
Gradation
sieves
Asphalt Binder
Typical
Values
105
Performance Table
A & VTS Select Grade
Grade 2.2.10
Table
Viscosity Grade A & VTS Select Grade
2.2.11
Penetration Table
A & VTS Select Grade
Grade 2.2.12
Asphalt General
Typical
Values
Air Voids in AASHTO T 269
Va Varies
Compacted HMA
HMA Thermal ASTM E 1952 0.67
Thermal
Conductivity BTU/hr-ft-
Conductivity
ºF
HMA Heat Capacity ASTM D2766 0.23
Heat Capacity
BTU/lb-ºF
Typical
Values
Thermal Indirect Tension Default Table
Cracking
Creep & Strength Varies
Tests 2.2.20
Mix Coefficient of VMA AASHTO R 35 Varies
Contraction /oC
106
The Indirect Tensile Creep (ITC) test (AASHTO T 315-02, AASHTO T 322-03)
UNBOUND MATERIALS
AASHTO T 307 NCRHP 1-28A

Electrohydraulic Electohydraulic
Loading Device Electro pneumatic
Top-Loading, closed loop Top-Loading, closed loop
Electronic load cell located Electronic load cell located
between the actuator and the inside the triaxial cell
Load Measurement
chamber piston rod inside the
triaxial cell
Load shape Haversine-shaped load pulse Haversine-shaped load pulse
Measured externally Measured internally
Two spring loaded Linear Optical extensometer
Strain Measurement
Variable Differential Transducers Non-contact sensors
(LVDT’s) Clamps attached to the specimen
Pneumatic Base/subbbase materials
Load Duration - 0.1 sec. Load Duration – 0.1 sec.
Rest period – 0.9 to 3.0 sec Rest period – 0.9 sec.
Cycle Time Hydraulic Subgrade material
Load Duration - 0.1 sec.
Rest period – 0.9 to 3.0 sec Load Duration – 0.2 sec
Rest period – 0.8 sec
107
Type 1 Materials Type 1 Materials
Untreated granular base/subbase Unbound granular
All untreated subgrade soils base/subbase
< 70 % passing No. 10 sieve Untreated subgrade
< 20 % passing No. 200 sieve Maximum particle size > 3/8”
Type 2 Materials Type 2 Materials

All materials not meeting Type 1 Unbound granular base/subbase
Thin wall tube samples Untreated subgrade
Types of Materials Maximum particle size > 3/8”
< 10 % passing No. 200 sieve
N/A Type 3 Materials
Untreated subgrade
Maximum particle size <3/8”
> 10 % passing No. 200 sieve
N/A Type 4 Materials

Undisturbed samples
Vibratory Type 1 Materials
Type 1 Soils Impact
Type 2 Soils Vibratory
Static Loading Type 2 Materials
Specimen Compaction Type 2 Soils Vibratory
Kneading Type 3 Materials
Uniform Density Gradient Impact
Kneading
Type 4 Materials (Undisturbed)
Soils and Unbound Material Inputs
Strength Inputs
Material Property Level cedure

1* 2 3
AASHTO T307 or
X X X
Resilient Modulus, Mr NCHRP 1-28A
CBR** X AASHTO T193
R-value** X
AASHTO T190
AASHTO Guide for the
Layer Coefficient, ai** X Design of Pavement
Structures
108
AASHTO T27 and AASHTO
X
PI and Gradation** T90
Dynamic Cone
X ASTM D 6951
Penetrometer**
AASHTO M 145
X
Soil Classification Or ASTM D2487
*
Level 1 Resilient Modulus Testing is only applicable for new asphalt pavement design.
**
Only one of the five properties (CBR, Layer Coefficient, etc) is needed for Level 2.
Level 2 Correlation Equations:
Strength/Index
Model Comments
Property
CBR Mr = 2555(CBR)0.64 CBR = California Bearing Ratio, percent
R-value Mr = 1155 + 555R R = R-value
AASHTO layer ⎛ a ⎞
M r = 30000⎜ i ⎟ ai = AASHTO layer coefficient
coefficient ⎝ 0.14 ⎠
75 wPI = P200*PI
PI and gradation* CBR = P200= percent passing No. 200 sieve size
1 + 0.728( wPI)
PI = plasticity index, percent
CBR =
292 CBR = California Bearing Ratio, percent
DCP*
DCP1.12 DCP =DCP index, mm/blow
109
Material
Type Input
Group Description*
Design Level
Category
1-2 • None of these levels are considered applicable for bedrock co
• User selects typical design values:
Solid, Massive Bedrock
New Typical ranges: 750-2000 ksi
3 Default Value: 1000 ksi
Highly Fractured, Weathered
Typical ranges: 250-1000 ksi
Bedrock Default Value: 500 ksi
1-2 • None of these levels are considered applicable for bedrock co
• User selects typical design values:
Solid, Massive Bedrock
Rehab
3 Default Value: 1000 ksi
Highly Fractured, Weathered
Default Value: 500 ksi
Soils and Unbound Material Inputs
Environmental Adjustment Inputs
Material Property Level

1 2 3
Plasticity Index AASHTO T90 N/A N/A
Gradation AASHTO T27 N/A N/A
Specific Gravity AASHTO T99 Calculated from PI

AASHTO T180
N/A
and Gradation
Maximum Dry Unit AASHTO T99 Calculated from PI
Assumed Value
Weight AASHTO T180 and Gradation
Optimum Moisture AASHTO T99 Calculated from PI
Content AASHTO T180 and Gradation N/A
Calculated from PI
Saturated Hydraulic AASHT0 T215 N/A
Conductivity and Gradation
Calculated from wopt,
Degree of Saturation N/A N/A
dmax, Gs
110
Appendix B
Typical Analysis Run Using Proposed AASHTO 2002 Design Guide
111
Project: level1.dgp
General Information Description:

Design Life 20 years
Base/Subgrade construction: September, 2004
Pavement construction: September, 2004
Traffic open: October, 2004
Type of design Flexible
Analysis Parameters
Analysis type Probabilistic
Performance Criteria Limit Reliability

Initial IRI (in/mi) 63
Terminal IRI (in/mi) 172 90
AC Surface Down Cracking (Long. Cracking) (ft/500): 1000 90
AC Bottom Up Cracking (Alligator Cracking) (%): 25 90
AC Thermal Fracture (Transverse Cracking) (ft/mi): 1000 90
Permanent Deformation (AC Only) (in): 0.25 90
Permanent Deformation (Total Pavement) (in): 0.75 90
Location:
Project ID:
Section ID:
Principal Arterials - Others
Date: 13/02/2004
Station/milepost format: Miles: 0.000

Station/milepost begin: 0
Station/milepost end: 10
Traffic direction: East bound
Default Input Level

Default input level Level 3, Default and historical agency values.
Traffic
Initial two-way aadtt: 1000
Number of lanes in design direction: 2
Percent of trucks in design direction (%): 50
Percent of trucks in design lane (%): 95
Operational speed (mph): 25
Monthly Adjustment Factors (Level 3, Default MAF)
Vehicle Class
Class Class Class
Month Class 4 Class 5 Class 6 Class 7 Class 8 Class 9 Class 10 11 12 13
January 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
February 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
March 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
April 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
112
May 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
June 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
July 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
August 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
September 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
October 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
November 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
December 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
Hourly truck traffic
Vehicle Class Distribution distribution
(Level 3, Default Distribution) by period beginning:
AADTT distribution by vehicle class Midnight 2.3% Noon 5.9%
1:00
Class 4 1.3% 1:00 am 2.3% pm 5.9%
2:00
Class 5 8.5% 2:00 am 2.3% pm 5.9%
3:00
Class 6 2.8% 3:00 am 2.3% pm 5.9%
4:00
Class 7 0.3% 4:00 am 2.3% pm 4.6%
5:00
Class 8 7.6% 5:00 am 2.3% pm 4.6%
6:00
Class 9 74.0% 6:00 am 5.0% pm 4.6%
7:00
Class 10 1.2% 7:00 am 5.0% pm 4.6%
8:00
Class 11 3.4% 8:00 am 5.0% pm 3.1%
9:00
Class 12 0.6% 9:00 am 5.0% pm 3.1%
10:00 10:00
Class 13 0.3% am 5.9% pm 3.1%
11:00 11:00
am 5.9% pm 3.1%
Traffic Growth Factor
Vehicle Growth Growth

Class Rate Function
Class 4 3.0% Compound
Class
10 3.0% Compound
Class
11 3.0% Compound
Class
12 3.0% Compound
Class
13 3.0% Compound
113
Traffic -- Axle Load Distribution Factors
Level 3: Default
Traffic -- General Traffic Inputs

Mean wheel location (inches from the 18
lane marking):
Traffic wander standard deviation (in): 10
Design lane width (ft): 12
Number of Axles per Truck
Vehicle Single Tandem Tridem Quad

Class Axle Axle Axle Axle
Class 4 1.62 0.39 0.00 0.00
Class 5 2.00 0.00 0.00 0.00
Class 6 1.02 0.99 0.00 0.00
Class 7 1.00 0.26 0.83 0.00
Class 8 2.38 0.67 0.00 0.00
Class 9 1.13 1.93 0.00 0.00
Class
10 1.19 1.09 0.89 0.00
Class
11 4.29 0.26 0.06 0.00
Class
12 3.52 1.14 0.06 0.00
Class
13 2.15 2.13 0.35 0.00
Axle Configuration
Average axle width (edge-to-edge) 8.5
outside dimensions, ft):
Dual tire spacing (in): 12
Axle Configuration
Single Tire (psi): 120
Dual Tire (psi): 120
Average Axle Spacing

Tandem axle(psi): 51.6
Tridem axle(psi): 49.2
Quad axle(psi): 49.2
Climate
icm file:
C:\DG2002\Projects\dallas.icm
Latitude (degrees.minutes) 32.54
Longitude (degrees.minutes) -97.02
Elevation (ft) 559
Depth of water table (ft) 30
114
Structure--Design Features
Structure--Layers
Layer 1 -- Asphalt concrete
Material type: Asphalt concrete
Layer thickness (in): 10
General Properties
General
Reference temperature (F°): 70
Volumetric Properties as Built

Effective binder content (%): 10.5
Air voids (%): 5.9
Total unit weight (pcf): 151
Poisson's ratio: 0.35 (user entered)
Thermal Properties
Thermal conductivity asphalt (BTU/hr-ft-F°): 0.67
Heat capacity asphalt (BTU/lb-F°): 0.23
Asphalt Mix

°F 0.1 0.5 1 5 10 25
15.8 2E+06 2E+06 2E+06 2E+06 3E+06 3E+06
40 752930 1E+06 1E+06 2E+06 2E+06 2E+06
70 164415 259400 319020 512560 645725 846635
100 44810 63095 76300 120730 149020 202140
130 30525 37830 42245 56325 68710 87100
Asphalt Binder
Option: Superpave binder test data
Temperature Angular frequency = 10 rad/sec

°F G*, psi Delta (°)
59 2605000 49.75
77 572800 55.84
95 120800 62.58
113 23440 69.3
140 3267 77.96
Layer 2 -- A-3
Unbound Material: A-3
115
Thickness(in): 18
Strength Properties
Input Level: Level 3
Analysis Type: ICM inputs (ICM Calculated Modulus)
Poisson's ratio: 0.35
Coefficient of lateral pressure,Ko: 0.5
Modulus (input) (psi): 29000
ICM Inputs
Gradation and Plasticity Index
Plasticity Index, PI: 0
Passing #200 sieve (%): 10
D60 (mm): 0.2
Calculated/Derived Parameters
Maximum dry unit weight (pcf): 126 (derived)
Specific gravity of solids, Gs: 2.65 (derived)
Saturated hydraulic conductivity (ft/hr): 0.0223 (derived)
Optimum gravimetric water content (%): 9.2 (derived)
Calculated degree of saturation (%): 78 (calculated)
Soil water characteristic curve parameters: Default values
Parameters Value
a 2.89
b 7.5
c 0.488
Hr. 7.65
Layer 3 -- A-7-6
Unbound Material: A-7-6
Thickness(in): Semi-infinite
Strength Properties
Input Level: Level 3
Analysis Type: ICM inputs (ICM Calculated Modulus)
Poisson's ratio: 0.35
Coefficient of lateral pressure,Ko: 0.5
Modulus (input) (psi): 8000
ICM Inputs
Gradation and Plasticity Index
Plasticity Index, PI: 40
D60 (mm): 0.01
116
Calculated/Derived Parameters
Maximum dry unit weight (pcf): 82.2 (derived)
Specific gravity of solids, Gs: 2.77 (derived)
Saturated hydraulic conductivity (ft/hr): 3.25e-005 (derived)
Optimum gravimetric water content (%): 28.8 (derived)
Calculated degree of saturation (%): 89.4 (calculated)
Soil water characteristic curve parameters: Default values
Parameters Value
a 750
b 0.911
c 0.772
Hr. 47500
Distress Model Calibration Settings - Flexible

AC Fatigue Level 3 (Nationally calibrated values)
k1 0.0043
k2 3.9492
k3 1.281
AC Rutting Level 3 (Nationally calibrated values)

k1 -3.449
k2 1.5606
k3 0.4791
Standard Deviation Total 0.1587*POWER(RUT,0.4579)+0.001

Rutting (RUT):
Thermal Fracture Level 3 (Nationally calibrated values)

k1 5
Std. Dev. (THERMAL): 0.2474 * THERMAL + 10.619
CSM Fatigue Level 3 (Nationally calibrated values)

k1 1
k2 1
Subgrade Rutting Level 3 (Nationally calibrated values)

Granular:
k1 1.673
Fine-grain:
k1 1.35
AC Cracking
AC Top Down Cracking
C1 (top) 7
117
C2 (top) 3.5
C3 (top) 0
C4 (top) 1000
Standard Deviation (TOP) 200 + 2300/(1+exp(1.072-2.1654*log(TOP+0.0001)))
AC Bottom Up Cracking
C1 (bottom) 1
C2 (bottom) 1
C3 (bottom) 0
C4 (bottom) 6000
Standard Deviation (TOP) 32.7 + 995.1 /(1+exp(2-2*log(BOTTOM+0.0001)))
CSM Cracking
C1 (CSM) 1
C2 (CSM) 1
C3 (CSM) 0
C4 (CSM) 1000
Standard Deviation CTB*1

(CSM)
IRI
IRI Flexible Pavements with GB
C1 (GB) 0.0463
C2 (GB) 0.0012
C3 (GB) 0.1834
C4 (GB) 0.0038
C5 (GB) 0.0074
C6 (GB) 0.0012
Std. Dev (GB) 0.387
IRI Flexible Pavements with ATB

C1 (ATB) 0.01
C2 (ATB) 0.0005
C3 (ATB) 0.0024
C4 (ATB) 18.36
C5 (ATB) 0.9694
Std. Dev (ATB) 0.292
IRI Flexible Pavements with CSM

C1 (CSM) 0.0073
C2 (CSM) 0.0765
C3 (CSM) 0.0001
C4 (CSM) 0.0084
C5 (CSM) 0.0002
Std. Dev (CSM) 0.229
118
Project: level1.dgp
Reliability Summary
Distress Reliablity Distress Reliability

Performance Criteria Target Target Predicted Predicted Acceptable
Terminal IRI (in/mi) 172 90 148.7 82.52 Fail

AC Surface Down Cracking (Long.
Cracking) (ft/500): 1000 90 2.3 98.02 Pass
AC Bottom Up Cracking (Alligator
Cracking) (%): 25 90 0.4 99.999 Pass
AC Thermal Fracture (Transverse
Cracking) (ft/mi): 1000 90 1 99.999 Pass
Permanent Deformation (AC Only) (in): 0.25 90 0.52 1.05 Fail
Permanent Deformation (Total Pavement)
(in): 0.75 90 0.79 39.65 Fail
119
Asphalt Sub-Layers Modulus Vs Time
3,000,000
2,500,000
AC1(1) h=0.5
2,000,000 AC1(2) h=0.5

Modulus (psi)
AC1(3) h=1.0
1,500,000 AC1(4) h=1.0

AC1(5) h=1.0
1,000,000 AC1(6) h=4.0
AC1(7) h=2.0
500,000
0
0 24 48 72 96 120 144 168 192 216 240
120
Surface Down Cracking - Longitudinal
10
8
Maximum Damage(%)
6
Surface
5
Depth = 0.5"
4
0
0 24 48 72 96 120 144 168 192 216 240 264
121
Bottom Up Cracking - Alligator
30
25
Alligator Cracking (%)
Maximum Cracking
20
Bottom Up Reliability
15
Maximum Cracking
Limit
10
0
0 24 48 72 96 120 144 168 192 216 240 264
122
Permanant Deformation: Rutting
1.20
AC Rutting Design Value = 0.25
Total Rutting Design Limit = 0.75
1.00
Rutting Depth (in)
SubTotalAC
0.80
SubTotalBase
SubTotalSG
0.60
Total Rutting
TotalRutReliability
0.40
Total Rutting Design Limit
0.20
0.00
0 24 48 72 96 120 144 168 192 216 240 264
123
IRI
200
180
160
140
120
IRI (in/mi)
100
80
IRI
60
IRI at Reliability
40 Design Limit
20
0
0 24 48 72 96 120 144 168 192 216 240 264
124

Evaluation AASHTO 2002

Uploaded by

Copyright:

Available Formats

Evaluation AASHTO 2002

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

Evaluation AASHTO 2002

Uploaded by

Copyright:

Available Formats

http://irc.nrc-cnrc.gc.

Evaluation of the Mechanistic Empirical

Author: Osman Ali

Report No: UR 3002.1

The structure module relies on linear elastic analysis ignoring visco-elastic/plastic

Figure 2.1 AASHTO 93 Structural Design Procedures………………………………6

The effects of variations in significant independent material variables were quantified by

1.4 Organization of the Report

STATE OF THE ART

2.1 Pavement Design and Analysis

where: W18 = predicted number of 80 kN (18,000 lb.) equivalent

ZR = standard normal deviate

So = combined standard error of the traffic prediction and

SN = structural number (an index that is indicative of the

= a1D1 + a2D2m2 + a3D3m3+...

ai = ith layer coefficient

Di = ith layer thickness (inches)

mi = ith layer drainage coefficient

∆PSI = difference between the initial design serviceability

MR = subgrade resilient modulus (in psi)

Reliability: Overall • Resilient Modulus to

Adjust MR. for Climate

Economic Evaluation Drainage Coefficients

Figure 2.1: AASHTO 93 Structural Design Procedures

SN = a1D1 + a2D2m2 + …… + aiDimi 2.2

2.3 Research Work

Binder Oscillating Plate

Figure 2.2 Dynamic Shear Rheometer

Bending Beam Rheometer (BBR)

Characterization of Asphalt Concrete Mix

The Dynamic Modulus

Reduced freq. [Hz]

Figure 2.3 Master Curve Developed Using Husaroad Program [16]

Figure 2.4 Rheological Models Representation

Figure 2.5 Huet-Sayegh Model

E*(iω) = E0 + (E∞ - E0 ) /1+ δ (iωτ)-k + (iωτ)-h (2.5)

D (t) = ε(t) / σ (2.6)

ε(t) = increasing strain with time

Characterization of Unbound Materials

Pore- air pressure

Figure 2.6 Triaxial Pressure Chamber for Determination of MR

Figure 2.7 MR Definition.

Many models were suggested to establish the stress-strain relationship. Course-grained

θ = Bulk stress (σ1 + σ2+ σ3),

k1, k2 = material parameters, with k1> 0 and k2 ≥ 0

Here, the parameters k1> 0 and k3 ≤ 0.

τoct is the octahedral shear stress.

OVERVIEW OF THE PROPOSED AASHTO DESIGN GUIDE

3.2 General Features of the Design Model

Design Input Levels

Pavement Structural Response Models

1. Multilayer linear elastic analysis method

Compute structural responses using multilayer elastic theory

Calculate accumulated distress and or damage at the end of

Use calibrated M-E performance models

Figure 3.1 M-E Design Process for Flexible Pavement

Crack_ P = cracking level corresponding to the reliability level p

using the deterministic model with mean inputs

Zp = standard normal deviate corresponding to reliability level p

Functional Recommended Level of Reliability