____________________

Final Report
ALDOT Project 930-554
(Rigid Pavement Design Portion)

ANALYSIS OF THE 2002 DESIGN

GUIDE FOR RIGID PAVEMENT

Prepared by
Dr. David H. Timm, P.E.
William E. Barrett, III

OCTOBER 10, 2005

TABLE OF CONTENTS
Page
Chapter 1 Introduction ......................................................................................................1
Background ..............................................................................................................1
Objectives ................................................................................................................9
Scope and Overview ................................................................................................9
Chapter 2 Literature Review...........................................................................................11
Pavement Design Background...............................................................................11
Existing Rigid Pavement Design Methodologies ..................................................15
Key Parameters in Rigid Pavement Design ...............................................16
Traffic ............................................................................................16
Climate...........................................................................................18
Concrete Properties........................................................................20
Performance Indices...................................................................................21
Reliability...................................................................................................21
Current AASHTO Methodology............................................................................22
PCA Methodology .................................................................................................27
Overview of Mechanistic-Empirical Pavement Design.........................................29
M-E Pavement Design Guide Overview................................................................31
Chapter 3 Methodology ..................................................................................................34
Introduction............................................................................................................34
Phase 1 PI Regression Analysis .......................................................................34
Traffic ........................................................................................................38
Subgrade ....................................................................................................41
Load Transfer Between Slabs ....................................................................41
Design Simulations ....................................................................................42
Regression Analysis...................................................................................43
Phase 2 Sensitivity Analysis ...............................................................................43
Phase 3 Design Slab Thickness Comparison ......................................................46
Summary ................................................................................................................48
Chapter 4 Results and Discussion...................................................................................49
Introduction............................................................................................................49
Phase 1 PI Regression Analysis .......................................................................49
Phase 2 Sensitivity Analysis ...............................................................................51
Phase 3 Design Slab Thickness Comparison ......................................................54
Summary ................................................................................................................55
Chapter 5 Conclusions and Recommendations...............................................................56
References..........................................................................................................................58

LIST OF TABLES
Page
Table 3.1 Traffic Parameters.............................................................................................38
Table 3.2 J-FACTORS (After AASHTO, 1993) ..............................................................42
Table 3.3 Design Simulation Input Parameters ................................................................43
Table 3.4. Sensitivity Analysis Parameters.......................................................................45
Table 3.5 Test Matrix for Slab Thickness Investigation....................................................48
Table 4.1 PSI Regression Statistics ................................................................................50
Table 4.2 Sensitivity Analysis Results..............................................................................52

ii

LIST OF FIGURES
Page
Figure 1.1 2002 Design Process (after Design Guide, 2002) .............................................6
Figure 1.2 Current AASHTO Empirical Design Approach................................................8
Figure 2.1 Serviceability vs. Time....................................................................................12
Figure 2.2 Mechanistic-Empirical Design Approach .......................................................14
Figure 2.3 k-value Relationships with Other Soil Support Values (after 1993 AASHTO
Design Guide) ....................................................................................................................18
Figure 2.4 Maximum Stresses due to Curling and Warping..............................................19
Figure 3.1 Faulting Output From 2002 Design Guide.......................................................35
Figure 3.2 Slabs cracked Output From 2002 Design Guide .............................................36
Figure 3.3 IRI Output From 2002 Design Guide..............................................................36
Figure 3.4 1993 PSI vs. 2002 Distress Criteria Analysis Schematic..............................37
Figure 3.5 Single Axle Weight Distribution By Vehicle Class .....................................38
Figure 3.6 Tandem Axle Weight Distribution By Vehicle Class ..................................39
Figure 3.7 Tridem Axle Weight Distribution By Vehicle Class....................................40
Figure 3.8 Vehicle Type Distribution ...............................................................................40
Figure 3.9 Average Number of Axles by Vehicle Class...................................................41
Figure 4.1. 2002 PSI vs. 1993 Predicted PSI ...............................................................50
Figure 4.2 Example of Influence of Input Parameters on Predicted Distress...................53
Figure 4.3. Slab Thickness Comparison Results ..............................................................55

iii

CHAPTER 1 INTRODUCTION

BACKGROUND
Traditionally, rigid pavement design has been accomplished through empirically
based procedures. Probably the most well-known procedure for highway pavements, the
AASHTO Method, was based on the AASHO Road Test held near Ottowa, Illinois
between 1958 and 1960. The design procedure utilized empirical relationships developed
from the AASHO Road Test and is therefore limited to the conditions of that test. All
empirically-based methods share this same common disadvantage in that they are limited
to the conditions and observations of the particular road sections on which the procedure
was based. This can cause problems by forcing designers to extrapolate outside the
original bounds of the method.
Conversely, mechanistic-empirical (M-E) design is more robust since it combines
the elements of mechanical modeling and performance observations in determining the
required pavement thickness for a set of design conditions. The mechanical models are
based on physics, not empirical relationships, and determine pavement stresses and
strains due to wheel loads. The empirical part of the design uses the calculated stresses
and strains to predict the life of the pavement based on site-specific field performance.
Basically, M-E design has the capability of changing and adapting to new developments
in pavement design by relying primarily on the mechanics of materials
The 2002 Design Guide for flexible and rigid pavements is a mechanisticempirical design approach that is a product of years of research undertaken by the
AASHTO Joint Task Force, the National Cooperative Highway Research Program

(NCHRP), and a research team comprised of several internationally recognized pavement

design experts. This state-of-the-art design approach was developed under National
Cooperative Highway Research Program Project 1-37A.
The 2002 Design Guide is extensive and comprehensive. It includes procedures
for the analysis and design of new, reconstructed, and rehabilitated asphalt and concrete
pavements, procedures for evaluating existing pavements, procedures for subdrainage
design, recommendations for rehabilitation treatments and foundation improvements, and
procedures for life cycle cost analysis. The design procedure, based on a mechanisticempirical (M-E) methodology, is integrated into a software program called Design
Guide 2002. The mechanistic-empirical design procedure to be included in the 2002
Design Guide will allow the designer to evaluate the effect of variations in materials
(both inherent and due to construction procedures) on pavement performance. The 2002
Design Guide will provide a rational relationship between construction and materials
specification and the design of the pavement structure.
Since M-E procedures can account for climate, aging, modern materials, and
modern vehicle loadings, variation in performance with relation to design life should be
reduced. That feature will allow agency managers to make better decisions based on life
cycle costs and cash flow. According to a report entitled LTPP and the 2002 Design
Guide issued by the LTPP research staff in 2001, based on probabilistic life cycle cost
analysis, it is conservatively estimated that improved pavement design procedures will
reduce premature failures and result in average annual savings in pavement rehabilitation
costs of $1.14 billion per year over the next 50 years. This analysis was based on a design
life of 20 years and the assumption that the percentage of pavement failures within the

first 10 years of a pavement's life would drop from 5 percent to 0.5 percent. It was further
assumed that the range of performance lives for the remaining pavements, 10 to 30 years
for current practice, would increase to 15 to 30 years using the 2002 procedure.
During the development of previous pavement design guides, the AASHTO
design staff recognized that future design procedures would eventually need to be based
on mechanistic-empirical principles due to constantly changing load configurations,
different subgrade types, and different environmental conditions than that encountered
during the AASHO Road Test of 1958. However, for such an approach to be practical, it
would require that pavement designers have ready access to computers capable of
handling mechanistic design programs. Today, personal computers with computational
capabilities greater than many mainframes of the early 1980s are on virtually every
engineer's desk. Pentium and higher-level personal computers should be adequate for
the design procedures that will be developed. The following is a list of the benefits of the
mechanistic design procedure. This list was taken from the NCHRP 1-37A Fall 2001
Research Summary:
1. The consequences of non-traditional loading conditions can be evaluated. For
example, the damaging effects of increased loads, high tire pressures, and multiple
axles can be modeled.
2. Better use of available materials can be made. For example, the use of stabilized
materials in both rigid and flexible pavements can be simulated to predict future
performance.

3. Improved procedures to evaluate premature distress can be developed to analyze why

some pavements exceed their design expectations. In effect, better diagnostic
techniques can be developed.
4. Aging effects can be included in estimates of performance. For example, PCC
strength increases with time, which, in turn, affects slab cracking.
5. Seasonal effects such as thaw weakening can be included in estimates of
performance.
6. Consequences of subbase erosion under rigid pavements can be evaluated.
7. Methods can be developed to better evaluate the long-term benefits of providing
improved drainage in the roadway section.
Currently, the 1993 AASHTO Guide for Design of Pavement Structures is the
primary document used by State Highway Agencies, including the Alabama Department
of Transportation, to design new and rehabilitated highway pavements. According to a
brief historical summary developed by the NCHRP 1-37a research team, the Federal
Highway Administrations 1995-97 National Pavement Design Review found that some
80% of the States make use of either the 1972, 1986, or 1993 AASHTO Pavement
Design Guides. All of these versions are empirically based on performance equations
developed using 1958-1960 AASHO Road Test data. The 1986 and 1993 guides
contained some refinements in materials input parameters and empirical procedures for
rehabilitation design.
The AASHTO Joint Task Force on Pavements (JTFP) has responsibility for the
development and implementation of pavement design technologies. This charge has been
pursued by the JTFP since the AASHO Road Test and led to the development of the 1993

AASHTO Guide and prior Guides. More recently, and in recognition of the limitations
of earlier Guides, the JTFP initiated an effort to develop an improved Guide by the year
2002. As part of this effort, a workshop was convened in California during 1996 to
develop a framework for improving the Guide. The workshop attendeespavement
experts from public and private agencies, industry, and academiaaddressed the areas of
traffic loading, foundations, materials characterization, pavement performance, and
environment to help determine the technologies best suited for the 2002 Guide. At the
conclusion of that workshop, a major long-term goal identified by the JTFP was the
development of a design guide based as fully as possible on mechanistic principles. The
2002 Design Guide is the fruition of that goal and the subsequent research efforts. The
2002 Design Guide is based on a mechanistic empirical design procedure and is depicted
in Figure 1.1.

INPUTS
Structure Material Traffic Climate
Selection of Trial Design

Revise

Structural Responses
from Mechanistic Model
Damage Accumulation with Time

Calibrated Damage-Distress Models

No
Design
Requirements
Satisfied?

Reliability

PERFORMANCE
VERIFICATION
Failure Criteria

Yes
Final Design

Figure 1.1 2002 Design Process (after Design Guide, 2002).

There are many new inputs and parameters that are incorporated into the 2002
Design Guide as well as new methods that greatly differ from the 1993 Design Guide in
order to more effectively determine the proper pavement, base, and subbase types and
thicknesses for a given traffic flow and climate. This poses a special challenge to
agencies currently using the 1993 Design guide since use of the new guide may require
additional training and equipment to fully implement the 2002 Design Guide.
Generally speaking, the 2002 Design Guide is an iterative process and includes
the following basic steps for rigid pavement design:

1. Inputs are gathered for the trial design. These include the structure of the pavement
(i.e.base type, subbase type, type of subgrade, slab specifications, etc.), the material
strengths used, the expected traffic loading on the pavement structure, and the
environmental conditions.
2. The designer inputs a trial design. This includes slab lengths, slab thickness, base
thickness, load transfer mechanisms (i.e. dowels or no dowels).
3. Once the trial design is set by the designer, the 2002 Design Guide program is
executed. The software first calculates structural responses due to traffic loading and
environmental conditions.
4. The program then predicts the damage and key distresses over the design life. These
key distresses include mean joint faulting, percent of slabs cracked, and a smoothness
prediction, International Roughness Index (IRI).
5. The trial design is checked against the preset distress criteria through output files
displayed in a spreadsheet format, and the design may be modified as needed to meet
performance and reliability requirements.
This procedure is expected to be a vast improvement over the current empirical
AASHTO methodology pictured schematically Figure 1.2.

AASHO
ROAD TEST

ITERATION

OBSERVATIONS

REGRESSION
EQUATIONS

DESIGN
AND
EVALUATION

Figure 1.2 Current AASHTO Empirical Design Approach.

As shown in Figure 1.2, the empirical approach is based on the AASHO Road
Test. In essence, it is based, one type of traffic loading, one climate, one subgrade soil,
and one base type. The observations made from this road test were transformed into
regression equations for the design of rigid and asphalt pavements. The limited design
variables used to generate the data at the AASHO Road Test cause much extrapolation
from the original observations through the regression equations, and thus shows the
necessity for a more mechanistic-based approach to pavement design.
While there is a national movement toward M-E based pavement design, and the
reasons for the shift are well documented, there is a serious concern about adopting this
method at the state agency level. The current design methods have been in place for
some time and there is extensive experience regarding their use in practice. Therefore,
prior to adoption of a new approach, there must be an extensive investigation of the

approach to warrant its use in practice. Specifically, in regard to rigid pavement design,
there are three main questions that must be answered:
1. When designing for the same design conditions, what slab thickness will the new
approach yield relative to the old approach?
2. How sensitive is the new approach to the various design inputs?
3. Is there a definable relationship between the old and new methods of rigid pavement
design?

OBJECTIVES
The objectives of this research serve to answer the questions above and include:
1. Establish a link between the existing methodology (1993 Design Guide) and the new
methodology (2002 Design Guide).
2. Identify inputs that are not significant in the final design and analysis so that greater
resources may be devoted to those that are significant.
3. Compare design thicknesses generated by the 1993 Design Guide with those
obtained through the 2002 Design Guide.

SCOPE AND OVERVIEW

A literature review was conducted to explore current mechanistic-empirical rigid
pavement design methods as well as the current accepted pavement design methodologies
for rigid pavement. Information was obtained regarding analysis techniques within each
approach.

An analysis was performed to compare the output of the 2002 Design Guide with
the output of the 1993 Design Guide. Several hundred simulations were run comparing
the 1993 and the 2002 Design Guide for rigid pavements and the output received was
analyzed through regression analyses. The 2002 Design Guide was further analyzed
through a slab thickness comparison between the 2002 Design Guide and the current
1993 AASHTO Design Guide and also through a sensitivity analysis based on the new
inputs for rigid pavement design in the 2002 Design Guide.

10

CHAPTER 2 LITERATURE REVIEW

PAVEMENT DESIGN BACKGROUND
Although pavement design has gradually evolved from art to science, empiricism
still plays an important role. Prior to the 1920s, the thicknesses of pavements were
based purely on experience. The same thickness was used for a section of highway even
though widely different soils were encountered. As experience was gained throughout
the years, various methods were developed by different agencies for determining the
thickness of pavement required. It is neither feasible nor desirable to document all the
methods that have been used so far; however the majority of pavement design has been
done empirically.
Empirical design is based on a pavements ability to withstand traffic over a given
time frame known as a design period (Huang, 1993). The ability to withstand traffic is
measured by an overall ride quality factor called the Present Serviceability Index (PSI), a
concept introduced at the AASHO Road Test. The PSI is a rating of the performance of
the pavement based on smoothness of travel over the pavement. PSI ranges from 5
(optimum performance) to 0 (worst performance) (Yoder and Witczak, 1975). Test data
were obtained from the American Association of State Highway Officials (AASHO)
Road Test, and for each section a PSI vs. Time plot was formed from the start of traffic
operations until the PSI dropped to a minimum tolerance, typically 1.5. This is shown in
Figure 2.1.

11

PSI
Initial Serviceability

Terminal Serviceability

DESIGN PERIOD

Time

Figure 2.1 Serviceability vs. Time.

These historical data were used to develop equations for flexible and rigid
pavements (McAuliffe et al., 1994). The main drawback to empirical pavement
modeling is at the foundation of the AASHO Road Test. At the AASHO road test only
certain subgrade types and strengths and only certain slab properties as well as a single
environment type were utilized. Any differing design conditions from the original
properties at the AASHO Road Test are to be considered an extrapolation. In most cases,
design inputs vary geographically and are rarely unique or constant, thus available
empirical procedures are, in a sense, inadequate.
Traffic weights and volume are the most variable and uncertain design inputs due
to regional differences in economies. Subgrade strength is also a highly variable input to
the design process. Within a particular state, there may be a wide variety of soil types
that require different design considerations. Pavement layer strength is highly variable
due to material type, layer location, and construction control exercised to obtain

12

uniformity in the material. Some empirical or deterministic design methods apply a

general variability term for all inputs; however this is inadequate as all inputs vary at
different levels (Basma et.al., 1989).
Selecting the most effective and economical design for a given project is
imperative to the pavement engineer and is at the core of all engineering practice. This
demand for better design processes in pavement technology arises from limited pavement
funds and materials, public need for better performance, and less traffic delay due to
maintenance (Basma et al., 1989).
From this need for better design methods, Mechanistic-Empirical (M-E) design
evolved. M-E design evolved in the 1970s on the basis of known relationships between
material behavior under stressed conditions and corresponding empirical performance of
pavement structure for all important combinations of loading and environmental
conditions. The M-E approach uses the principles of engineering mechanics to calculate
pavement responses and relate them to rates of deterioration as shown in Figure 2.2.

13

Inputs:
Materials
Characterization

Traffic

Climate

Pavement Responses
,,

DESIGN ITERATION

Paving Materials
Subgrade Soils

Structural Model

Transfer Functions

Pavement Distress / Performance

Design Reliability
Final Design

Figure 2.2 Mechanistic-Empirical Design Approach.

Miners hypothesis is typically employed in the M-E approach to sum the damage
caused by traffic and environmental loading. Miners hypothesis is defined as the sum of
applied loads on the pavement structure over the total allowable loads on the pavement
structure:
k

D=
i =1

ni
Ni

(2.1)

where:
D = damage factor
If D>1 then the pavement distress level has exceeded the limit

14

If D<1 then the pavement distress level has not exceeded the limit
ni = expected number of load repetitions on the pavement
Ni = allowable number of load repetitions calculated by the transfer function
i = loading or seasonal condition
M-E design is more robust than simple empirical design since it combines the
elements of mechanical modeling with empirical performance observations in
determining required pavement thickness for a set of design conditions. The mechanical
load-displacement model is based upon physics, not empirical relationships, and
determines pavement stresses and strains due to wheel loads. The empirical part of the
design uses calculated stresses and strains to predict the life of the pavement based upon
site-specific field performance. In short, M-E design has the capability of changing and
adapting to new developments in pavement design by relying primarily on the mechanics
of materials (McAuliffe et al., 1994).

EXISTING RIGID PAVEMENT DESIGN METHODOLOGIES

In rigid pavement design, the most widely used procedures for design is the Guide
for Design of Pavement Structures published in 1993 by the American Association of
State Highway and Transportation Officials (AASHTO). A few states use the 1972
AASHTO Interim Guide procedure, the Portland Cement Association (PCA) procedure,
their own empirical or mechanistic-empirical procedure, or a design catalog (Hall, 2000).
Regardless of the method, the key input parameters in any concrete pavement design
procedure are traffic over a design period, subgrade, environment, concrete material

15

properties, base properties, performance criteria, and design reliability. These key input
parameters are outlined in the following subsections.

Key Parameters in Rigid Pavement Design

Traffic
The number of heavy truck axle loads anticipated over the design life must be
estimated from current truck traffic weights and volumes and growth projections. In the
AASHTO methodology, the anticipated spectrum of truck loads over the design period is
expressed in terms of an equivalent number of 18-kip single axle loads, computed using
load equivalency factors that relate the damage done by a given axle type and weight to
the damage done by this standard axle. The 18-kip equivalency for single axle loads, also
known as an ESAL, is a widely accepted standard axle in the U.S. and around the world.
A standard legal axle load limit has generally been imposed for highway travel, hence
maximum gear loads in highway design have not appreciably increased with time. For
this reason, the effects of other vehicles have normally been accounted for in the design
phase by the use of 18-kip single axle loads (Yoder et al., 1975).
It is important to note that the 2002 Design Guide will no longer utilize ESALs
and rely instead upon load spectra. Load spectra are simply the weight distributions of
various axle configurations. In fact, load spectra are used in the 1993 AASHTO method
to computer ESALs.

16

Subgrade Support
The modulus of subgrade reaction of the foundation can be measured by plate
bearing tests. Load is applied at a predetermined rate until a pressure of 10 psi is
reached. The pressure is held constant until the deflection increases not more than 0.001
in. per minute for three consecutive minutes. The average of the three dial readings is
averaged to determine the deflection. A more detailed description of the test is provided
in (Huang, 1993). The modulus of subgrade reaction (k) is important in rigid pavement
design because it defines both the subgrade and subbase support. The modulus of the
subgrade reaction is given by the following equation.

k=

p
w

(2.2)

where,
p = pressure on the plate (psi)
w = deflection of the plate (in)
Since the plate loading test is time-consuming and expensive, the k-value is
usually estimated from correlations to simpler tests such as California Bearing Ratio
(CBR) and R-value tests. Figure 2.3 shows the approximate relationship between the kvalue and other soil properties.

17

100

60

20

50
10

20

40

30

Modulus 1,000psi k = Modulus / 19.4

70

Texas Triaxial Value

CBR

30

80

R - Value

70
50
40

90

Figure 2.3 k-value Relationships with Other Soil Support Values (after 1993
AASHTO Design Guide).
The subgrade and subbase support strengths varies over the course of a year. The
k-values are low during spring thaws and high during the longer freezing periods.
However, as evident at the AASHO Road Test, k-values do not have a great effect on
required thickness of concrete pavements. Other methods, such as the PCA method,
avoid the tedious method of considering seasonal variations in k-values by using normal
summer or fall k-values for design purposes (PCA, 1984).

Climate
The climatic variables, particularly temperature variation, can greatly influence
concrete behavior The following list, taken from a Fall 2001 NCHRP 1-37A progress
update on climatic data research for the 2002 Design Guide displays how daily and

18

seasonal variations in temperature and moisture influence the behavior of concrete

pavements in many ways, including:
1. Opening and closing of transverse joints in response to daily and seasonal variation in
slab temperature, resulting in fluctuations in joint load transfer capability, which is
the ability of each slab group to transfer wheel loads from one slab to the next.
2. Upward and downward curling of the slab caused by daily cycling of the temperature
gradient through the slab thickness as pictured in Figure 2.4.

= 0

Neutral Axis
Max. Compression

T = + upward
Daytime
Conditions

Max. Tension

T = + down

Max. Tension

Nightime
Conditions

Max. Compression
Figure 2.4 Maximum Stresses due to Curling and Warping.
3. Permanent upward curling of the slab, which in some circumstances may occur
during construction, as a result of the dissipation of a large temperature gradient that
existed in the concrete while it cured.
4. Upward warping of the slab caused by seasonal variation in the moisture gradient
through the slab thickness.

19

5. Erosion of base and foundation materials caused by inadequate drainage of excess

water in the pavement structure, primarily from precipitation.
6. Freeze-thaw weakening of the subgrade soil.
7. Freeze-thaw damage to certain types of coarse aggregates in the concrete mix.
8. Corrosion of dowel bars, steel reinforcement, or both, especially in coastal
environments and in areas where deicing salts are used in winter.
Although the effects of the climate on concrete pavement behavior and
performance have been recognized since the time of the earliest concrete pavement
design experiments, concrete pavement thickness design practice traditionally has not
explicitly considered most of these climatic effects. Several recent field and analytical
studies have contributed greatly to better understanding and quantifying these effects so
that they may be more adequately considered in thickness design (Hall, 2000). Climatic
effects will be more adequately considered in thickness design in the 2002 Design Guide
through a program that compiles much of the recent field and analytical studies into
weather stations that allow the designer to triangulate the design location or to input its
exact latitude and longitude from weather stations in major cities. The program is called
the Enhanced Integrated Climate Model (EICM).

Concrete Properties
For the purpose of pavement thickness design, concrete is characterized by its
flexural strength (Sc) as well as its modulus of elasticity (E). Concrete flexural strength
is usually characterized by the 28-day modulus of rupture from third point loading tests
of beams or it may be estimated from compressive strength as shown in Equation 2.3.

20

The flexural strength of concrete is a measure of the quality and durability of the
concrete. A higher flexural strength of the concrete will most likely result in a lower
concrete slab thickness. The modulus of elasticity can also be predicted from
compressive strength as indicated in Equation 2.4 for normal weight concrete.

S c = 8 f 'c

to 10 f 'c

Ec = 57,000 f 'c

(Huang, 1993)

(2.3)

(ACI Code 318)

(2.4)

where:
Sc = concrete modulus of rupture, psi
Ec = concrete elastic modulus, psi
fc = concrete compressive strength, psi

Performance Indices
All pavement thickness design procedures incorporate performance criteria that
define the end of the performance life of the pavement. In the current AASHTO
methodology, the performance criterion is the loss of serviceability, which occurs as a
result of accumulated damage caused by traffic load applications. The Portland Cement
Association (PCA) procedure uses both fatigue cracking and erosion criteria.

Reliability
The reliability level for which a pavement is designed reflects the degree of risk
of premature failure that the agency is willing to accept. Facilities of higher functional
classes and higher traffic volumes warrant higher safety factors in design. In the
AASHTO methodology, this margin of safety is provided by applying a reliability

21

adjustment to the traffic ESAL input. The magnitude of the adjustment is a function of
the overall standard deviation associated with the AASHTO model, which reflects error
associated with the estimation of traffic and strength inputs and error associated with the
quality of fit of the model to the data on which it is based (AASHTO, 1993). When
reliability adjustments are made to the traffic input in this manner, AASHTO
recommends that average values should be used for the material inputs.
Before the introduction of reliability concepts in pavement thickness procedures,
the traditional approach to introducing a margin of safety into concrete pavement
thickness design was to apply a safety factor to the concrete modulus of rupture. This
approach is still used in the PCA procedure. The PCA procedure also accounts for
uncertainty by the reduction of the concrete modulus of rupture by one coefficient of
variation and increasing the traffic weights by a percentage depending upon the type of
roadway.

CURRENT AASHTO METHODOLOGY

The AASHO Road Test data and the research on the rigid pavement inputs were
compiled to develop a general regression equation for rigid pavement design for
AASHTO. The empirical model for the performance of the jointed plain concrete
pavement (JPCP) and jointed reinforced concrete pavement (JRCP) sections in the main
loops of the AASHO Road Test predicts the log of number of axle load applications as a
function of the slab thickness, axle type and weight, and terminal serviceability. The
empirical model used in the 1993 AASHTO Design Guide is shown below:.

22


PSI
log

4.5 1.5

log W18 = Z R So + 7.35 log(D + 1) 0.06 +

1 + 1.624 107

8.46
(D + 1)

0.75
ScCd D 1.132
+ (4.22 0.32 pt ) log

0.75

D 18.42

215.63

Ec 0.25

(2.5)

where:
ZR = z-statistic corresponding to level of design reliability
So = overall standard deviation of design (0.35 often assumed for rigid pavement design)
W18= Number of ESALs before the terminal level of serviceability is reached
D = slab thickness (in)

PSI = Change in serviceability index = po pt

pt= terminal serviceability
Sc = Concrete modulus of rupture (psi)
Cd = Drainage coefficient
Ec = Concrete modulus of elasticity (psi)
k = modulus of subgrade reaction (pci)
This original model applies only to the designs, traffic conditions, climate,
subgrade, and materials of the AASHO Road Test. It has been modified and extended to
make possible the estimation of allowable axle load applications to a given terminal
serviceability level for conditions of concrete strength, modulus of subgrade reaction, and

23

concrete elastic modulus different than those of the AASHO Road Test. The AASHTO
design methodology has also been extended to accommodate the conversion of mixed
axle loads to ESALs through the use of load equivalency factors (AASHTO, 1993).
A limitation of the extended AASHO model is that the loss of serviceability that
corresponds to a predicted number of axle load applications does not include any
contribution of faulting to pavement roughness because, although the doweled pavements
in the AASHO Road Test experienced substantial loss of support, they did not fault. The
design loss of serviceability is presumed to be entirely due to slab cracking (Hall, 2000).
Furthermore, it is an extrapolation of the AASHTO model to apply it to the
prediction of performance of undoweled jointed pavements, jointed pavements with
stabilized bases, jointed pavements with spacings other than those used in the AASHO
Road Test, continuously reinforced concrete pavement (CRCP), or concrete pavements of
any type in climates that may produce significantly greater curling and warping stresses
than those experienced by the AASHO Road Test sections. These limitations provide
further justification for moving toward an M-E based methodology.
The 1986-1993 AASHTO Guide incorporates many modifications to the
procedures for concrete pavements, although the basic design models for both remained
the same as in previous versions. The principal modifications to the AASHTO concrete
pavement design methodology in the 1986-1993 procedure are the following (Hall,
2000):
1. Addition of drainage adjustment factor, Cd; a multiplier of the slab thickness that
presumably is less than 1.0 for drainage conditions worse than those in the AASHO
Road Test and greater than 1.0 for better drainage conditions. The quality of drainage

24

is measured by the length of time for water to be removed from the bases and
subbases and depends primarily on their permeability. The percentage of time during
which the pavement structure is exposed to moisture levels approaching saturation
depends on the average yearly rainfall and the prevailing drainage conditions.
2. Incorporation of the design subgrade reaction modulus, or k-value, as a function of
the subgrade resilient modulus, depth to rigid layer, base thickness and elastic
modulus, erodability of the base material, and seasonal variation in soil support.
3. Incorporation of a load transfer adjustment (J-factor) as a function of pavement type,
load transfer, and shoulder type
4. Addition of a reliability adjustment applied to the design ESAL input instead of using
a factor of safety on the modulus of rupture.
The revised AASHTO design model for concrete pavements presented in the
1998 Supplement to the AASHTO Guide was developed under NCHRP Project 1-30
(Hall, 2000). The purpose of the NCHRP Project 1-30 study was to evaluate and
improve the AASHTO Guides characterization of subgrade and base support. The
original AASHO empirical model was calibrated to the springtime k-value measured in
plate load tests on the granular base, whereas the 1986 Guides method for determining
the design k-value was based on seasonally adjusted annual average k-value for the
composite k-value. A key recommendation of the 1-30 study was that, for purposes of
concrete pavement design in the existing AASHTO methodology, both the AASHO Road
Test subgrade and the subgrade of the project under design should by characterized by
the seasonally adjusted annual average static elastic k-value. The 1998 AASHTO
Supplement presents guidelines for determination of an appropriate design k-value on the

25

basis of the plate bearing tests, correlations with soil types and properties, CBR, or
deflections measured on in-service pavements.
Using the same process by which the original AASHO Road Test empirical model
was extended in 1961, a new AASHTO design model was derived to be consistent with
the recommended characterization of the design k-value and consider the effects on the
stress in the slab of base modulus, base thickness, slab and base friction, joint spacing,
edge support, temperature and moisture gradients, and traffic loading. The stress
analyses were conducted using a three-dimensional finite element model, which was
validated by comparison with stresses in the AASHO Road Test, and measured slab
deflections. Regression equations were then developed to relate the computed stresses to
the design factors. The three-dimensional finite element model was also used to develop
a design check for corner loading for undoweled jointed pavements.
As in the earlier versions of the AASHTO rigid pavement design procedure, the
computed slab thickness is that which is required to support the anticipated ESALs to a
selected terminal serviceability level, assuming that the serviceability level loss is due
only to slab cracking. If faulting were to develop on a pavement to such a degree that it
contributed significantly to loss of serviceability, the pavement would have been under
designed; that is, it would have reached terminal serviceability sooner than predicted.
The appropriate way to prevent this is not to increase the slab thickness, but rather to
design the joint load transfer system so that faulting will not develop to the degree that it
contributes significantly to loss of serviceability.
One major output of the AASHO Road Test was the load equivalency factor
(LEF). LEFs were used to quantify the damage different axle loads and configurations

26

caused to the pavement relative to an 18-kip single axle load. One shortcoming of LEFs
is that they are based on the AASHO Road Test concrete pavements, which failed due to
slab cracking only. There are several types of failure modes prevalent in rigid pavement
structures (Hall, 2000) not represented by the AASHO LEF equations. Most rigid
pavement structures fail due to faulting and fatigue cracking.
Some further limitations of the AASHTO Design Guide are that the effects of
widened lanes and tied concrete shoulders cannot be analyzed in detail. The AASHTO
Design Guide also does not directly consider joint spacing and curling stresses in rigid
pavements (Roesler et al., 2000). The 1993 AASHTO Design Guide also is limited as to
incorporating climate, predicting subgrade, base, and pavement layer strengths, and also
accounting for different types of pavement loading due to traffic, particularly heavy
vehicle types and different axle configurations. These shortcomings are due primarily to
the limited pavement design inputs available at the AASHO Road Test.

PCA METHODOLOGY
The PCA rigid pavement design procedure evaluates a candidate pavement design
with respect to two potential failure modes: fatigue and erosion. This M-E procedure
was developed using the results of a finite element analyses of stresses induced in
concrete pavements by joint, edge, and corner loading. The analyses took into
consideration the degree of load transfer provided by dowels or aggregate interlock and
the degree of edge support provided by a concrete shoulder (PCA, 1984). The PCA
procedure, like the 1986-1993 AASHTO procedure, employs the composite k concept in
which the design k is a function of the subgrade soil k, base thickness, and base type.

27

The fatigue analysis incorporates the assumption that approximately 6% of all

truck loads will pass sufficiently close to the slab edge to produce a significant tensile
stress. The erosion analysis quantifies the power with which a slab corner is deflected by
a wheel load as a function of the slab thickness, foundation k-value, and estimated
pressure at the slab-foundation interface (Hall, 2000).
For each load level considered, the expected number of load repetitions over the
design life is expressed as a percentage of allowable load repetitions of that load level
with respect to both fatigue and erosion. An adequate thickness is one for which the sum
of the contributions of all axle load levels to fatigue and erosion levels is less than 100
percent. This evaluation method is consistent with Miners Hypothesis described
previously.
Other concrete pavement design methods range from empirical adaptations of the
AASHTO method to calibration and mechanistic-empirical extension of the AASHTO
method and methods that combine mechanistic stress calculation with an empirical
fatigue cracking model.
The latest versions of the PCA thickness design for concrete highway and street
pavements have more mechanistic features than the empirically based AASHTO guide.
The PCA uses the load spectra analysis to calculate the bending stress in the concrete due
to various axle loads and configurations. Load spectra analysis is more theoretically
sound than ESAL analysis because fundamental stresses and strains are calculated and
related to the performance of laboratory concrete fatigue beam tests. Load spectra
analysis also allows for calculation of pavement stresses due to axle loads and
configurations not originally considered in the AASHO Road Test. The limitations of the

28

PCA guide include no ability to analyze widened lanes or different joint spacings and no
consideration of load transfer across the shoulder-lane joint (Packard, 1984). This is
significant because widened lanes are an often used type of edge support, and also
because differences in length between joints in concrete slabs are often encountered in a
rigid pavement design.

OVERVIEW OF MECHANISTIC-EMPIRICAL PAVEMENT DESIGN

The development of mechanistic-empirical (M-E) design procedures was needed
to account for situations where existing empirical studies could not be extrapolated to
find a reasonable thickness design solution. Mechanistic-based design addresses the
theoretical stresses, strains, and deflections in the pavement structure due to the
environment, pavement materials, and traffic. These stresses, strains, and deflections are
then related to the field performance of in-service rigid pavements through transfer
functions (Roesler et al., 2000).
In an M-E design procedure, new, old, and current pavement features may be
analyzed to determine their effect on pavement performance. The pavement engineer can
make changes to these design features to accommodate the specific location and
constraints of the proposed pavement structure (Roesler et al, 2000). For example, the
behavior of a pavement in a high desert environment should not be expected to be the
same as a pavement in a coastal environment, and some pavement design features may
need to be adjusted to account for the different environments.
In contrast, with an empirical design guide such as the AASHTO Design Guide,
changes can be made only to the pavement features that are included in the original road

29

test. Extrapolation of designs not included in the original road test could result in
unrealistic designs. In M-E procedures, analysis can be used to describe the failure of
field tests in terms of stresses, strains, and deflections. Future designs can be outside the
scope of any field testing because the mechanisms of pavement failure are quantified
with theoretical analysis (Barenberg, 1992).
A mechanistic-based model is verified through calibration with field test results.
A purely mechanistic model would not have to be calibrated with field data, but an M-E
design approach still needs calibration to account for unknown slab behaviors. These
unknown behaviors are also addressed in applying reliability to the design. Applying
design reliability gives a factor of safety against premature failure.
For rigid pavement M-E design, typically two models are used to estimate
pavement life; a model for predicting cracking in the slabs and a model for predicting
faulting in the joints between each slab. For the fatigue cracking model, typical M-E
procedures use critical stresses in the slab to estimate total wheel load applications before
cracks begin, propagate, and ultimately fracture the pavement. For the faulting model,
the example below is of a mechanistic-empirical equation where performance evaluation
relates faulting rates to the maximum concrete bearing stress (COPES, 1993):
F = (N18 )

0.5377

[2.2073 + 0.002171(S )

0.4918

+ 0.003292(JS )

1.0793

where:
F = pavement faulting (in.)
N18 = number of ESALs in millions
S = maximum bearing stress (psi)
JS = transverse joint spacing (ft)

30

2.1397(k )

0.01305

(2.6)

k = estimated modulus of subgrade reaction on the top of the subbase (pci)

For the slab cracking model, an example of a mechanistic-empirical equation is
shown in the following equation where performance evaluation relates cracking rates to
the concrete modulus of rupture and the tensile stress (Kessler et al., 1953).
Log Nf = 17.61-17.61(/Sc)

(2.7)

where:
Nf = Number of ESALs to failure

= Tensile Stress (psi)

Sc = Concrete Modulus of Rupture (pci)

Fatigue life is the number of stress repetitions, at a magnitude less than its strength,
required for structural failure of pavement. Faulting and cracking are both considered
modes of structural failure (McAuliffe et al., 1994).

M-E PAVEMENT DESIGN GUIDE OVERVIEW

NCHRP 1-37A was started in 2000 in an attempt to create a user-friendly,
mechanistic-empirical design program that would encompass all types of pavement
design. The overall objective of this project was to develop and deliver the 2002 Guide
for Design of New and Rehabilitated Pavement Structures, based on mechanisticempirical principles, accompanied by the necessary computational software, for adoption
and distribution by AASHTO. As it is now called, the Mechanistic Empirical Pavement
Design Guide (MEPDG) was to use existing structural models and be a fully M-E design
approach. No new models or databases were created for this design guide. The MEPDG

31

was to provide a uniform basis for the design of flexible, rigid, and composite pavements
and employ common design parameters for traffic subgrade, environment, and reliability.
The focus of this research is the MEPDG module for rigid pavement design. The
rigid pavement design portion includes jointed concrete pavement, jointed reinforced
concrete pavement, and continuously reinforced concrete pavement design. The MEPDG
procedure verifies trial design against user input performance criteria. The distress types
considered in the MEPDG for rigid pavements are joint faulting and transverse cracking
in jointed plain concrete pavements and punchouts in continuously reinforced concrete
pavements. The MEPDG takes directly into account the effects of
temperature/environmental conditions for the rigid pavement section trail designs by use
of the Enhanced Integrated Climate Model (EICM). The EICM algorithms are linked to
the design guide software as an independent module through interfaces and design
inputs.
This research focuses on the 2002 Design Guide for Rigid Pavements and how it
compares with the 1993 Design Guide for Rigid Pavements. Since the 1993 AASHTO
Design Guide is the most widely used empirical pavement guide, it was compared with
the new mechanistic-empirical 2002 Design Guide for Rigid Pavements. The main
difference in the 1993 Design Guide and the 2002 Design Guide is in its prediction of
pavement thickness and distress. The 1993 Design Guide predicts pavement thickness
based a pavements terminal serviceability index, anticipated traffic, and strength
parameters of the pavement. The 2002 Design Guide takes into account strength
parameters, anticipated traffic, and user input thicknesses of slab, base, and subgrade and

32

predicts distress. If the predicted distresses are less than the user input thresholds for all
distress criteria, then the pavement can be considered successful.
In the next chapter, Methodology, the process of comparing the existing empirical
1993 AASHTO Design equation for rigid pavements to the 2002 Design Guide for rigid
pavements is presented.

33

CHAPTER 3 - METHODOLOGY
INTRODUCTION
The task of comparing two different approaches for pavement design through
design simulations and statistical analysis was divided into several phases. The first
phase consisted of comparing the change in present serviceability (PSI) from the 1993
Design Guide to a derived PSI from the pavement distresses taken from the 2002
Design Guide. The second phase consisted of a sensitivity analysis that analyzed new
inputs encountered in the 2002 Design Guide with respect to their bearing on the
pavement distresses in the 2002 Design Guide. The last phase was a comparison of slab
design thicknesses determined from the 1993 Design Guide and the 2002 Design Guide.

PHASE 1 PSI REGRESSION ANALYSIS

The 2002 Design Guide and 1993 Design Guide offer two very distinct methods
of pavement design. In the 1993 Design Guide, the input data are entered into a
regression-based design equation which directly outputs the required slab thickness. The
primary performance criterion in the 1993 AASHTO Design Guide is the change in
serviceability over the design life (PSI).
In the 2002 Design Guide, the pavement cross section, along with the proper
traffic, structural, and material properties are input and predicted distresses are output,
and the user determines the efficiency of the design based on the user input distress
criteria. The rigid pavement distress criteria used in the 2002 Design Guide is IRI, %
Slabs Cracked and Mean Joint Faulting. Mean joint faulting is defined as the average
slab elevation differential at each transverse joint over the entire rigid pavement system.

34

Percent slabs cracked is the predicted percentage of slabs in the rigid pavement system
with at least one crack, and IRI is a measure of overall roughness of the pavement
system. Examples of each of these 2002 Design Guide outputs are shown in the
following figures. These figures were created by the 2002 Design Guide as output files
and are representative of predicted distresses over the design life of the pavement.
Predicted faulting
0.14

Faulting Limit
0.12

Faulting, in

0.10
0.08

Faulting
Faulting Limit

0.06
0.04
0.02
0.00
0

10

12

14

16

Pavement age, years

Figure 3.1 Faulting Output From 2002 Design Guide.

35

18

20

22

Predicted cracking
100

Percent slabs cracked, %

90
80
70
60

Percent slabs
cracked
Limit percent
slabs cracked

50
40
30

Limit percent slabs cracked

20
10
0
0

10

12

14

16

18

20

22

Pavement age, years

Figure 3.2 Slabs cracked Output From 2002 Design Guide.

Predicted IRI
180
162

IRI Limit

IRI, in/mile, mil

144
126
108

IRI
IRI Limit

90
72
54
36
18
0
0

10

12

14

Pavement age, years

Figure 3.3 IRI Output From 2002 Design Guide.

36

16

18

20

22

These output distresses from the 2002 Design Guide were correlated to a
predicted PSI through regression analysis. Figure 3.4 shows the method by which the
2002 Design Guide output distresses were correlated to PSI.

1993
Design
Guide

Pavement Cross
Section
2002
Design
Guide
IRI

Faulting

PSI

% Slabs
Cracked
Regression
Analysis
PSI=fn(IRI, % Slabs Cracked, Faulting)

Figure 3.4 1993 PSI vs. 2002 Distress Criteria Analysis Schematic.
From Figure 3.4, it can be seen that this analysis started with a rigid pavement
section. This pavement section was input into both the 2002 and 1993 Design Guides,
respectively. Each design approach was then executed and output performance
predictions, as shown in Figure 3.4 were obtained. Once all design simulations were run
in both design guides, a regression analysis was performed relating the 1993 output to the
2002 output. The information from the regression analysis was then used to determine a
PSI from the 2002 distress output. The relation of the 1993 PSI to the 2002 PSI is
the first step in analyzing the 2002 Design Guide as it relates to current empirical design
approaches.

37

There are three main differences between the 1993 Design Guide and the 2002
Design Guide inputs. They are traffic, subgrade support, and load transfer coefficients.
As different as these two approaches may seem, a procedure to bridge the gap between
the two methodologies had to be derived through manipulation of the required inputs for
design in the 1993 and 2002 Design Guides.

Traffic
Traffic input for the 2002 Guide is radically different from the 1993 Guide.
Where the 1993 Guide uses ESALs to characterize traffic, the 2002 Design Guide uses
axle load spectra to characterize the traffic. The ESAL number used in the 1993 Guide is
a representative number of the axle load spectra. The method of obtaining design ESALs
from load spectra followed procedures outlined in the 1993 Design Guide. Specifically,
the default load spectra in the 2002 Design Guide were converted using the load
equivalency factor equations to determine the total design ESAL. Table 3.1 lists the
relevant traffic parameters used in this study while Figures 3.5 through 3.7 illustrate the
default load spectra for the single, tandem and tridem axles, respectively.
Table 3.1 Traffic Parameters.
Parameter
Average Annual Daily Traffic (AADT)
Average Annual Daily Truck Traffic (AADTT)
Lane Distribution Factor
Directional Distribution Factor
Design Period
Growth Rate

38

Value
10,000
1,500 (15% of AADT)
90%
50%
10 and 20 years
4%

20
18

Cl4
Cl5
Cl6
Cl7
Cl8
Cl9
Cl10
Cl11
Cl12
Cl13

Percentage of Axles

16
14
12
10
8
6
4
2
0
0

5000

10000

15000

20000

25000

30000

35000

40000

45000

Single Axle Weight, lb

Figure 3.5 Single Axle Weight Distribution By Vehicle Class.

40
35

Cl4
Cl5
Cl6
Cl7
Cl8
Cl9
Cl10
Cl11
Cl12
Cl13

Percentage of Axles

30
25
20
15
10
5
0
0

10000

20000

30000

40000

50000

60000

70000

Tandem Axle Weight, lb

Figure 3.6 Tandem Axle Weight Distribution By Vehicle Class.

39

80000

90000

80
Cl4
Cl5
Cl6
Cl7
Cl8
Cl9
Cl10
Cl11
Cl12
Cl13

70

Percentage of Axles

60
50
40
30
20
10
0
0

20000

40000

60000

80000

100000

120000

140000

Tridem Axle Weight, lb

Figure 3.7 Tridem Axle Weight Distribution By Vehicle Class.

35%
31.3%

Percentage of AADTT

30%
24.6%

25%
20%

15.3%
15%
9.8%

10%

7.6%
5.0%

5%
1.8%

3.3%
0.5%

0.8%

Class 4 Class 5 Class 6 Class 7 Class 8 Class 9 Class

10

Class
11

0%

Figure 3.8 Vehicle Type Distribution.

40

Class
12

Class
13

Average Number of Axles Per Vehicle

4.5

Single
Tandem
Tridem

4
3.5
3
2.5
2
1.5
1
0.5

Class 13

Class 12

Class 11

Class 10

Class 9

Class 8

Class 7

Class 6

Class 5

Class 4

Figure 3.9 Average Number of Axles by Vehicle Class.

Subgrade
The subgrade support is not as complicated of a conversion technique as was the
traffic conversion. The 2002 Guide takes into account the subgrade strength by the
subgrade resilient modulus, MR (psi). The 1993 Guide accounts for subgrade strength by
the modulus of subgrade reaction, keff (pci). The following conversion was used
(AASHTO, 1993):

k eff = M R / 19.4

(3.1)

Load Transfer Between Slabs

The load transfer conversion is necessary because the 1993 Guide uses a J-factor
that takes into account the use of dowels for load transfer, edge support, slab length and
41

slab thickness. The 2002 Design Guide uses each of these inputs separately instead of
grouping them into one representative number. Table 3.2 is commonly used to convert
support conditions into a J-factor for the 1993 Design Guide. This table was used to
select the proper load transfer factor (J-factor) for use in the 1993 Design Guide for each
simulation. A predetermined slab thickness as well as predetermined subgrade resilient
modulus, which had to be converted into a subgrade reaction modulus was factored into
each simulation.
Table 3.2 J-FACTORS (After AASHTO, 1993)
Shoulder

Asphalt Shoulder

Dowels Between Slabs

JCP and JPCP

CRCP

YES
A.
B.
C.
D.
E.
F.
A.
B.
C.
D.
E.
F.

Tied PCC Shoulder

NO

3.2
3.2
3.2
3.2
3.2
3.2
2.9
3.0
3.1
2.6
2.8
2.9

A.
B.
C.
D.
E.
F.

4.5
4.1
3.9
5.2
4.8
4.5

N/A

Case A: H = 7in., k = 100pci

Case B: H = 10in., k = 100pci
Case C: H = 13 in., k = 100pci

YES
A.
B.
C.
D.
E.
F.
A.
B.
C.
D.
E.
F.

2.8
2.7
2.7
3.0
2.9
2.8
2.5
2.6
2.6
2.3
2.4
2.5

NO
A.
B.
C.
D.
E.
F.

3.7
3.6
3.6
4.0
3.8
3.7

N/A

Case D: H = 7in., k = 600pci

Case E: H = 10in., k = 600pci
Case F: H = 13in., k = 600pci

Design Simulations

Table 3.3 lists the inputs used in this investigation. The simulation matrix
consisted of seven thicknesses, eleven soil support values, two dowel conditions and two
design lives.

42

Table 3.3 Design Simulation Input Parameters

Parameter
Pavement Depth (in)
Subgrade Modulus (psi)

Values
8, 8.5, 9, 9.5, 10, 10.5, 11
970, 1940, 2910, 3880, 4850, 5820,
6970, 7760, 9700, 10670, 11640

Dowels/No Dowels 1=Dowels, 0=No

Dowels
Design Life (years)

0, 1
10, 20

Regression Analysis
Once the simulations had been executed in both the 1993 and 2002 Desing
Guides, a regression analysis was performed on the pavement distresses received from
both systems. The distress criteria set forth in the 2002 Design Guide were IRI, percent
of slabs cracked, and mean joint faulting. These three distresses were related to the
change in present serviceability index through a regression analysis by the following
equation.
PSI = 0 + 1 (IRI ) + 2 (FAULTING ) + 3 (% SLABS )

(3.2)

where:
PSI = change in serviceability calculated from 1993 Design Guide
0,1,2,3 = regression coefficients
IRI = 2002 Design Guide predicted International Roughness Index, in/mile
FAULTING = 2002 Design Guide predicted distress of mean joint faulting, in
%SLABS = 2002 Design Guide predicted distress of percent slabs cracked

PHASE 2 SENSITIVITY ANALYSIS

Phase 2 of this study was a sensitivity analysis involving only the 2002 Design
Guide. Within the 2002 Design Guide there are many new inputs for pavement design

43

not previously used in the design of rigid pavements by the 1993 Design Guide. These
new inputs were evaluated to determine their effect on the rigid pavement distresses in
the 2002 Design Guide. When using the new software, default values are given for each
of these new inputs and so each input was varied around the default value given in the
2002 Design Guide. If the distresses encountered by the pavement structure were found
to be unchanged, then it was concluded that the input being varied had no effect on the
structural capacity of the pavement structure and thus, default values would suffice for
pavement design. Table 3.5 shows the criteria used for the new input values incorporated
into this sensitivity analysis.

44

Table 3.4. Sensitivity Analysis Parameters

Design Input
Operational Speed (mph)
Mean Wheel Location (in. from marking)
Traffic Wander SD(in)
Design Lane Width (ft)
Avg. Axle Width (ft)
Dual Tire Spacing (in)
Single Tire Pressure (psi)
Dual Tire Pressure (psi)
Tandem Axle Spacing (in)
Tridem Axle Spacing (in)
Quad Axle Spacing (in)
Short Wheelbase (ft)
Medium Wheelbase (ft)
Long Wheelbase (ft)
Percent Short Wheelbase
Percent Medium Wheelbase
Percent Long Wheelbase
Climate
Joint Sealant Type
Edge Support Selection
PCC-Base Interface
Erodobility Index
Loss of Bond Age (months)
Surface Shortwave Absorptivity
Infiltration
Drainage Path Length (ft)
Pavement Cross Slope
Coeff. Of Thermal Expansion (/Fx10-4)
Thermal Conductivity (BTU/hr-ft-F)
Heat Capacity (BTU/hr-ft-F)
PCC-Zero Stress Temp. (0F)
Ult. Shrinkage (microstrain)
Reversible Shrinkage (%)
Time to Develop 50% Shrinkage (days)
20yr/28day Compressive Strength Ratio
20yr/28day Elastic Modulus Ratio
20yr/28day Resilient Modulus Ratio
20yr/28day Tensile Strength Ratio

45

Low
40
12
10
12
6
12
80
80
40
40
40
12
12
15
20
20
20
ANC
None
None
Unbonded
Erodible
40
0.6
Low
12
0.1
6
1
0.1
100
500
25
15
1
1
1
1

Input Values
Default
High
60
80
18
20
14
20
14
20
8.5
10
20
24
120
160
120
160
51.6
60
49.2
60
49.2
60
14
20
15
18
18
21
33
60
33
60
33
60
ATL
MIA
Liquid
Silicone
Wide Slab
Tied PCC
Unbonded
Bonded
Resistant
Very Resistant
60
80
0.85
1
Medium
High
14
20
0.2
0.5
9
12
1.25
1.5
0.28
0.5
120
140
670
750
50
75
35
60
1.2
1.5
1.2
1.5
1.2
1.5
1.2
1.5

PHASE 3 DESIGN SLAB THICKNESS COMPARISON

The third and final part of this investigation was a slab thickness comparison
comparing the 2002 Design Guide slab thickness and the 1993 AASHTO Design Guide
slab thickness. A total of 125 simulations were executed and evaluated through both the
1993 Design Guide and the 2002 Design Guide. Creating a slab thickness comparison
required several more steps than Phase 1 of this investigation in which the two design
guides were compared based on predicted pavement performance. In this comparison,
only IRI and PSI were evaluated. This was justified by the fact that IRI is not a singular
pavement distress. Rather, IRI is much like the present serviceability index which is a
measure of ride quality. IRI encompasses all distresses incurred by the pavement
structure and denotes them as one factor called roughness. IRI can be empirically related
to the present serviceability index PSI index by the following equation (Darter et al.,
1992):

PSI = 5e 0.0041IRI

(3.3)

where:
PSI = Present Serviceability Index
IRI = International Roughness Index (in/mile)
While Part 1 of this analysis was to show that the two pavement design
methodologies (1993 and 2002) could be linked together, Part 3 was to expand this link
into a more practical comparison between slab thicknesses between the two
methodologies. In other words, would the new design guide result in thicker, thinner, or
about the same designed slab thicknesses?

46

Traffic, soil stiffness, and concrete modulus of rupture were main variables
examined in this study and are summarized in 3.5. Traffic was varied by altering the
design life of the pavement. Half of the simulations were executed based on a 20-year
design life and the other half was run based on a 10 year design life. An AADTT of
1,500 vehicles was selected and applied to the default axle load distribution factors,
traffic growth factors, and vehicle class distributions described in Phase 1, above.
Soil stiffness was varied based on the resilient modulus of the subgrade according
to Equation 3.1. The PCC modulus of rupture, Sc, was varied from 400 to 900 psi and
was used to determine the PCC compressive strength, fc, as well as the PCC modulus of
elasticity, E, to be used in the 1993 Design Guide for this slab thickness comparison. The
modulus of rupture was calculated from (Huang, 1993):
S c' = 9 f c'

(3.4)

The PCC modulus of elasticity, E, was also determined from the PCC compressive
strength according to (Huang, 1993):
E = 57,000 f c'

(3.5)

Finally, the 2002 Design Guide distress criterion of IRI was varied based on terminal
serviceability in the 1993 Design Guide. The terminal PSI levels used for this analysis
were 4.0, 3.5, 3.0, and 2.5. Each design simulation was run using 50% and 90%
reliability, R. Table 3.5 summarizes the input data used for this investigation.

47

Table 3.5 Test Matrix for Slab Thickness Investigation

Parameter
PSI

Values
0.2, 0.7, 1.2

pt
IRI (in/mile)

3, 3.5, 4
54, 87, 125

MR (psi)

1000, 3000, 7000

keff (pci)
S'c (pci)
E (psi)
f'c (psi)
Reliability, %
Traffic (ESALs)
Design Life (years)

51.55, 154.64, 360.82, 515.46

400, 600, 900
3E+06, 4E+06, 6E+06
1975.3, 4444.4, 10000
50, 90
5E+06, 10E+06
10, 20

Determining optimum slab thickness in the 2002 Design Guide was an iterative
process since the 2002 Design Guides output consisted only of predicted distresses and
distress criteria set by the designer. The slab thickness was considered to be at an
optimum level when the predicted distresses were as close to the preset distress criteria as
possible without exceeding the preset distress criteria. Optimum slab thicknesses were
rounded to the nearest half-inch that would not exceed the distress criteria.

SUMMARY

This chapter detailed the three phases of this investigation that included:

A comparison between predicted pavement performance from the 1993 and 2002
Design Guides.

A senstivity analysis to investigate the relative importance of the numerous input

parameters in the 2002 Design Guide.

A slab thickness comparison between the 1993 and 2002 Design Guides.

The results of the three phases are presented and discussed in the next chapter.

48

CHAPTER 4 RESULTS AND DISCUSSION

INTRODUCTION
The results of the regression analysis relating the 1993 PSI to the 2002
performance parameters, the sensitivity analysis of the new inputs in the 2002 Design
Guide, and a slab thickness comparison comparing slab thicknesses between design
methods are presented, respectively.

PHASE 1 PSI REGRESSION ANALYSIS

The data obtained from this analysis showed that there is a definite relationship
between the change in serviceability predicted in the 1993 Design Guide and the
pavement distresses output by the 2002 Design Guide. There were 154 design
simulations executed in the two Guides, respectively. The best fit equation was, also
shown in Table 4.1, was:
PSI = 129.78(Faulting) + 0.105(% Slabs Cracked) - 0.078(IRI) + 5.16

(4.1)

R2 = 0.71
Figure 4.1 illustrates the regression equation against the actual PSI values obtained from
the 1993 Design Guide. These results suggest a strong correlation between PSI and the
predicted levels of pavement distress from the 2002 Design Guide. Further, Equation 4.1
can be used to convert results obtained from the 2002 Design Guide into PSI levels in
the 1993 Design Guide. Further discussion of Equation 4.1 is provided below.

49

6.0
5.5

2002 Predicted PSI

5.0
4.5

R = 0.7169

4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

1993 PSI

Figure 4.1. 2002 PSI vs. 1993 Predicted PSI.

Table 4.1 PSI Regression Statistics
Regression Statistics
Multiple R
0.845311507
R Square
0.714551544
Adjusted R Square 0.708842575
Standard Error
0.712281399
Observations
154

P-value

Intercept

Coefficients
5.161042354

Standard Error t Stat

4.27817998
1.20636401

Faulting, in

129.7880078

38.78455712

3.346383649 0.00103479

Percent slabs
cracked

0.105779541

0.056079961

1.886227075 0.061197312

IRI, in/mile

-0.078695094 0.067246351

-1.170250765 0.243756006

0.229576392

Table 4.1, shows that the IRI distress criterion carries a negative regression
coefficient relating it to PSI. This is due to the fact that IRI is actually a combination of
mean joint faulting and % slabs cracked that is used to express overall roughness of a

50

rigid pavement surface and can be directly correlated to PSI. It can be said that the IRI
coefficient and the intercept are simply correction factors for the mean joint faulting and
% slabs cracked values, which influence the PSI.
The smaller p-value suggests a greater statistical importance in the regression
analysis. In this instance, the IRI and the intercept have high p-values when compared to
the faulting and percent slabs cracked p-values. While it would have been quite possible
to disregard the IRI distress in this regression analysis altogether, it was included simply
because it is a predicted distress used in the 2002 Design Guide, and therefore it is of
relevance in this situation. The correlation between PSI and IRI will be shown in the
latter sections of this chapter.

PHASE 2 SENSITIVITY ANALYSIS

Table 4.2 summarizes the sensitivity analysis. An input was considered to have a
notable influence on the rigid pavement distresses in the 2002 Design Guide output if it
affected one or more of the distresses. A considerable change in faulting was set at 0.1
inches. A considerable change in Percent Slabs Cracked was 10%, and a considerable
change in IRI was set at 10 in/mile. If any of the three distresses measured encountered a
change at or above these thresholds, then the input was listed as influential in the 2002
Design Guide. As an example, Figure 4.2 illustrates the difference between parameters
that have a large versus small impact on pavement distress. The changes in the distresses
shown in the Table 4.2 are absolute values of differences when changing from the low to
high input value.

51

Table 4.2 Sensitivity Analysis Results.

Design Input

Low

Default

High

% Slabs
Cracked

IRI

Influence

2002 Design Guide Distresses

Faulting

Input Values

Operational Speed (mph)

40

60

80

NO

Mean Wheel Location (in. from marking)

12

18

20

0.014

26.2

29

YES

Traffic Wander SD(in)

10

14

20

0.009

23.8

4.9

YES

Design Lane Width (ft)

12

14

20

NO

Avg. Axle Width (ft)

8.5

10

0.5

0.4

NO

Dual Tire Spacing (in)

12

20

24

23.5

19

YES

Single Tire Pressure (psi)

80

120

160

0.4

0.3

NO

Dual Tire Pressure (psi)

80

120

160

7.4

NO

Tandem Axle Spacing (in)

40

51.6

60

7.4

NO

Tridem Axle Spacing (in)

40

49.2

60

1.1

0.7

NO

Quad Axle Spacing (in)

40

49.2

60

NO

Short Wheelbase (ft)

12

14

20

NO

Medium Wheelbase (ft)

12

15

18

NO

Long Wheelbase (ft)

15

18

21

NO

Percent Short Wheelbase

20

33

60

NO

Percent Medium Wheelbase

20

33

60

NO

Percent Long Wheelbase

20

33

60

NO

Climate

ANC

ATL

MIA

0.25

12.4

22

YES

Joint Sealant Type

None

Liquid

Silicone
Tied
PCC

NO

Edge Support Selection

None

Wide Slab

PCC-Base Interface

Unbonded

Unbonded

Erodibility Index

Erodible

Resistant

Bonded
Very
Resistant

0.005

4.2

4.5

NO

0.5

0.6

NO

0.17

0.7

8.1

Yes

Loss of Bond Age (months)

40

60

80

NO

Surface Shortwave Absorptivity

0.6

0.85

0.019

33.6

35

YES

Infiltration

Low

Medium

High

NO

Drainage Path Length (ft)

12

14

20

NO

Pavement Cross Slope

0.1

0.2

0.5

NO

12

0.097

95.9

130

YES

Thermal Conductivity (BTU/hr-ft-F)

1.25

1.5

0.11

28

29

YES

Heat Capacity (BTU/hr-ft-F)

0.1

0.28

0.5

0.01

39.7

38

YES

PCC-Zero Stress Temp. ( F)

100

120

140

0.017

8.4

NO

Ult. Shrinkage (microstrain)

500

670

750

0.007

0.2

3.5

NO

-4

Coeff. Of Thermal Expansion (/ Fx10 )

Reversible Shrinkage (%)

25

50

75

NO

Time to Develop 50% Shrinkage (days)

15

35

60

NO

20yr/28day Compressive Strength Ratio

1.2

1.5

2.3

2.4

NO

20yr/28day Elastic Modulus Ratio

1.2

1.5

2.3

2.4

NO

20yr/28day Resilient Modulus Ratio

1.2

1.5

2.3

2.4

NO

20yr/28day Tensile Strength Ratio

1.2

1.5

2.3

2.4

NO

52

% Slabs Cracked with New Input Variation

40
35
% Slabs Cracked

30
25

Mean Wheel Location (in. from

lane marking)

20

Avg. Axle Width (ft)

15
10
5
0
0

10

15

20

25

New Input Values

Figure 4.2 Example of Influence of Input Parameters on Predicted Distress.

The group of new inputs that produced the greatest influence on predicted pavement
distresses included:

Coefficient of thermal expansion

Thermal conductivity

Heat capacity

Shortwave surface absorptivity

Mean wheel location

Wheel wander standard deviation

Dual tire spacing

Climate

Erodibility index

53

It makes sense that the thermal properties of the slab would influence rigid
pavement distresses, particularly the cracking of the slab. The higher coefficient of
thermal expansion, causing greater expansion in the slab, surface shortwave absorptivity,
causing a steeper thermal gradient in the slab which creates a higher degree of curling
and warping stresses in the slab, and thermal conductivity, which increases the heat
absorption in the slab, would theoretically lead to higher degrees of cracking in the slab
due to the increased expansion and contraction of the slab.
It is also logical that some of the traffic inputs, aside from amount of traffic,
would also affect the predicted pavement responses. The wheel location, wheel wander
and dual tire spacing all affect where loads are placed on the slab and the concentration of
stresses. These, in turn, affect overall performance of the pavement.
It should be noted that this sensitivity analysis was only the groundwork for a
more extensive sensitivity analysis that should be performed in the future to further
explore these new inputs. Further analyses should examine interaction of the various
input parameters.

PHASE 3 DESIGN SLAB THICKNESS COMPARISON

The final, and perhaps most important, phase of this research was to investigate
slab thickness comparisons between the 1993 and 2002 Design Guides. Figure 4.3
summarizes the slab thickness comparisons using the approach detailed in Chapter 3.
From a comparison standpoint, 78% of the designs differed by less than 1 in. between the
two approaches. 93% of the designs were within 2 in. between design approaches.
However, as seen in the figure, it did appear that the 2002 Design Guide arrived at

54

slightly thinner pavement designs. On average, thicknesses from the 2002 Design Guide
were 0.55 in. less than that from the 1993 Design Guide. It was clear from this figure
that the 2002 Design Guide, although through a very different method of pavement
design than the 1993 Design Guide, calculates optimum slab thicknesses very comparable
to the 1993 AASHTO Design Guide for Rigid Pavements.

2002 Design Guide (in)

16
15

+ - 1 inch

14
13

+ - 2 inches

12

Line of Equality

11
10
9

+- 1 INCH 78%
+- 2 INCH 93%

8
7
6

10

11

12

13

14

15

16

1993 Design Guide (in)

Figure 4.3. Slab Thickness Comparison Results.

SUMMARY
This chapter detailed the results of the three phases of the investigation. A
regression equation was developed to determine PSI, in the context of the 1993 Design
Guide, from predicted distresses from the 2002 Design Guide. The sensitivity analysis
showed that many parameters did not influence the predicted distresses, while some slab
and traffic characteristics proved to be more influential. Finally, through slab thickness
comparisons, it was found that the two approaches yielded very similar slab thicknesses
with the 2002 Design Guide thickness slightly thinner.

55

CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS

NCHRP 1-37A has released the beta version of the new M-E Pavement Design
Guide, also called the 2002 Design Guide. Before it gains widespread acceptance and
use, there is a need for studies into how the new guide compares to the existing design
methodology currently used by many state agencies. This report documented a study
comparing the 2002 versus the 1993 Pavement Design Guides. Specifically, the
investigation was divided into three phases. Phase 1 compared performance predictions
made by the 1993 and 2002 Design Guides. Phase 2 determined the sensitivity of the
2002 Design Guide to its many inputs. Finally, Phase 3 compared slab thicknesses
between the two design approaches. Based upon the findings presented in this report, the
following conclusions and recommendations can be made:
1. A strong relationship exists between performance predictions made by the 1993 and
2002 Design Guides. The developed regression equation, presented in Chapter 4, can
translate results obtained from the 2002 Design Guide (i.e., IRI, faulting and
cracking) into performance predicted by the 1993 Design Guide (PSI).
2. The sensitivity analysis identified slab thermal properties and traffic characteristics as
critical design inputs. Emphasis should be placed on obtaining these inputs to
increase accuracy of performance predictions. However, further studies are
warranted to examine interactions between the various input parameters.
3. The two design approaches produce very similar design thicknesses, though the 2002
Design Guide typically was somewhat thinner; 0.55 in. on average. Therefore, it is
not expected that future designs using the 2002 Design Guide will dramatically

56

change PCC designs in terms of slab thickness. However, the 2002 Design Guide
will give the designer a more complete understanding of what kinds of distress may
appear over the design life of the pavement, in addition to overall ride quality.
4. To more fully understand the 2002 Design Guide, further comparison designs are
warranted. New designs being developed by ALDOT for rigid pavements, using the
1993 Design Guide, could also be run through the 2002 Design Guide software to
build a more extensive basis of comparison.

57

REFERENCES

1.

Basma, A. and Adli Al-Balbissi, Probabilistic Design of Flexible and Rigid

Pavements Using AASHTO Equations, Transportation Research Record No.
1227, 1989, pp. 34-43.

2.

Hall, K.T., State of the Art and Practice in Rigid Pavement Design, AB202:
Committee on Rigid Pavement Design, Transportation Research Board,
Washington D.C., 2000.

3.

Huang, Y.H., Pavement Analysis and Design, Prentice Hall, New Jersey, 1993.

4.

Lee, Y-H, TKUPAV: Stress Analysis and Thickness Design Program for Rigid
Pavements, Journal of Transportation Engineering, July/August 1999, pp. 338398.

5.

McAuliffe, D., L. J. Bendana, and Wie-Shih Yang, Mechanistic-Empirical Rigid

Pavement Design for New York State, Transportation Research Record No.
1449, 1994, pp. 141-147.

6.

Packard, R.G., Thickness Design for Concrete Highway and Street Pavements,
Portland Cement Association, 1984.

7.

Portland Cement Association, Thickness Design for Concrete Highway and Street
Pavements, Skokie, IL, 1984.

8.

Roesler, J.R., J.T. Harvey, J.Farver and F. Long, Investigation of Design and
Construction Issues for Long Life Concrete Pavement Strategies, California
Department of Transportation, February, 2000.

58

9.

Wu, Chung-Lung and Mang Tia, Development of a Rational Thickness Design

Method for Rigid Pavements, Transportation Research Record No. 1227, 1989,
pp. 205-218.

10.

Yoder, E.J. and M.W. Witczak, Principles of Pavement Design, Second Edition,
Wiley-Interscience Publications, New York, 1975.

11.

Calibrated Mechanistic Structural Analysis Procedures for Pavements; NCHRP

Project 1-26; TRB National Research Council, Washington D.C.; 1992.

12.

Guide for Design of Pavement Structures; American Association of State and

Highway Transportation Officials; Washington D.C.; 1993.

59