You are on page 1of 17

Carpenter, Ghuzlan, and Shen

A Fatigue Endurance Limit for Highway and


Airport Pavements
Samuel H. Carpenter
Civil and Environmental Engineering Department
University of Illinois at Urbana-Champaign
1206 Newmark CE Lab
205 N. Mathews Avenue
Urbana, IL 61801
(217) 333-4188
scarpent@ux1.cso.uiuc.edu

Khalid A. Ghuzlan
California Department of Transportation (Caltrans) District 4
9th Floor / Design North Counties
111 Grand Ave. 94612
Oakland, CA 94623-0660

(510)-622-8652
kghuzlan@dot.ca.gov

Shihui Shen
Civil and Environmental Engineering Department
University of Illinois at Urbana-Champaign
1206 Newmark CE Lab
205 N. Mathews Avenue
Urbana, IL 61801
(217) 244-6064

DRAFT - Paper No 03-3428 Not for publication


Submitted for Possible Presentation and publication at the Annual Meeting of the
Transportation Research Board, January, 2003

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

ABSTRACT
The existence of a fatigue endurance limit has been postulated for a considerable
time. With the increasing emphasis on Extended Life Hot Mix Asphalt Pavements
(ELHMAP), or perpetual pavements (PP) the verification of the existence of this
endurance limit, a strain below which none or very little fatigue damage develops, has
become a substantial consideration in the design of these new multi-layered full depth
pavements. This paper presents fatigue data collected on a surface and binder mixture
that were tested for an extended period from 5 to 48 million load repetitions at strain
levels down to 70 micro strain. The fatigue results are analyzed in the traditional manner,
and using the dissipated energy ratio (DER). This analysis shows that there is a
difference in the data at normal strain levels recommended for fatigue testing and at the
low strain levels. This difference cannot substantiate an endurance limit using traditional
analyses procedures, but the dissipated energy approach clearly shows that there is a
distinct change in material behavior at low flexural strain levels that supports the fact that
at low strain levels the damage accumulated from each load cycle is disproportionately
less than what is predicted from extrapolations of normal strain level fatigue testing
which may be attributed to the healing process. The conclusion of this study is that lab
testing can verify the existence of a fatigue endurance limit in the range of 90 to 70 micro
strain, below which the fatigue life of the mixture is significantly extended relative to
normal design considerations.

Key Words: Fatigue, Dissipated Energy, Endurance Limit, Healing, Low-Strain


Word Count: 4000
Figure Count: 8 @ 250 - 2000
Table Count: 2 @ 250 - 500
Total Count: 6500

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

A Fatigue Endurance Limit for Highway and Airport


Pavements
INTRODUCTION
Figure 1 illustrates two typical fatigue curves for a surface and binder mix as developed
using the current AASHTO procedure(1) from the SHRP program(2). Curves such as
these represent the essential design approach for asphalt pavements, given suitable field
calibration to account for pulse duration differences, rest periods between loads, and
structural interactions that cannot be duplicated efficiently in the laboratory. Fatigue data
of this nature produce the coefficients in many of the fatigue models used in thickness
design. While there are many model forms, a common representation is:
Where:

Nf = K1(1/)K2

(1)

Nf is the number of load cycles to failure


is the flexural tensile strain in the beam
K1 and K2 are the fatigue coefficients (slope and intercept)
This fatigue algorithm or any of the various formulations provide a linear relationship
between log of tensile strain and log of load repetitions shown in Figure 1.
This relationship has been validated by any number of studies and is accepted as
common for pavement design considerations when suitable field calibrations are applied.
The calibrations have varied from 6 to 70 and even higher and reflect differences between
field and lab test conditions (3). As such, this form of the fatigue model provides a
simple and consistent means of comparing the differences in fatigue behavior of mixtures
tested in the laboratory, and is used here.
The data used to prepare the relation shown in Equation 1 do not establish or
provide support for the existence of a fatigue endurance limit, a concept that has been
postulated for a considerable time. With the increasing emphasis on Extended Life Hot
Mix Asphalt Pavements (ELHMAP), or perpetual pavements, the verification of the
existence of this endurance limit, a strain below which none or very little fatigue damage
develops, has become a substantial consideration in the design of these new multi-layered
full depth pavements.
The following sections of this paper will discuss the ELHMAP concept and the
impact that a fatigue endurance limit would have on the structural design of these
pavements. Data will be provided for several typical mixtures demonstrating that the
fatigue endurance limit does exist and that further research is clearly warranted to
determine its precise value and the variables which might affect its magnitude. For this
study, mixtures that were sampled from the truck as part of QC sampling were reheated
and compacted in the laboratory for testing in flexural fatigue to provide data for this
study.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

Tensile Strain

0.01

y = 0.0037x -0.1823
R2 = 0.9828
Mix 6

0.001
y = 0.0048x -0.228
R2 = 0.9504
Mix 64
0.0001
100

1000

10000

100000 1000000 1E+07

Load Repetitions

Figure 1. Traditional Flexural Fatigue Curves for Two Mixtures Investigated in


This Study.
EXTENDED LIFE PAVEMENTS
A new concept for asphalt pavements that has generated a serious amount of interest in
recent years is the ELHMAP. This pavement is also termed a perpetual pavement (4).
The design philosophy is to construct a structural section that will not fail with a
premium surface material that can be easily rehabilitated as needed. A typical cross
section is shown in Figure 2.
The ELHMAP concept utilizes asphalt mixtures in three layers that are
engineered for the performance required of each layer. The surface layer must be highly
rut resistant and the bottom layer must be highly fatigue resistant. The inner layer is
typically a standard high quality rut resistant binder mix. Fatigue resistance is enhanced
by increasing the asphalt content of the bottom layer, termed a rich bottom layer by
increasing the design asphalt content 0.5 percent and lowering the air voids to two
percent. This combination produces an approximate no change in the modulus of the
layer. Structural reductions in thickness will result primarily from any modification to
the fatigue algorithm, the K1 and K2 coefficients, resulting from the mix changes. The
increase in asphalt content can add to stripping resistance in the bottom layer,
contributing to reliability of the design.
This layered approach produces distinctly different modulus values for each layer
which is a departure from traditional full depth designs where the surface layer is the
significantly different layer. While this raises questions regarding the structural design
relative to a traditional full depth, the only difference comes in the differing modulus
values impact on the tensile strain at the bottom of the rich bottom layer coupled with an
improved fatigue algorithm.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

Highly Rut Resistant Mixture

Typical Binder or Base


Mixture

Binder or Base Mixture with


Rich Asphalt Content

Treated Subgrade
or Working Platform
Extended Life Hot Mix Asphalt Pavement
Figure 2. Typical Cross Section for an Extended Life Hot Mix Asphalt Pavement.
MATERIALS TESTED
Ten mixtures have been tested to provide fatigue data to be used in investigating the
existence of a fatigue endurance limit. These mixtures were sampled from the truck as
part of the QC/QA sampling during construction. The mixtures were transported to the
laboratory and reheated and compacted to 7 and 4 percent air voids in a rolling wheel
compactor. The ten mixtures were part of the total of 79 mixtures tested by Ghuzlan(6).
The properties of the mixtures presented here are given in Table 1. All mixtures are
Illinois DOT mixtures using neat binders and limestone or dolomite coarse aggregates
with manufactured sands.
Testing was performed on an Industrial Process Control (IPC) fatigue machine (6) in a
temperature controlled cabinet meeting the AASHTO requirements (1).
Table 1. Mixture Properties for Fatigue Testing
MIX
Asphalt
Content
Nominal
Maximum
Size, mm
Percent
Passing
4.75 mm
Percent
Passing
0.075 mm
Air Voids

6
5.4

64
4.6

5
4.6

7
5.3

8-7
5.3

8-4
5.3

9
5.4

17
5.7

21
4.2

11
5.7

20
5.0

9.5

19

19

19

9.5

9.5

9.5

9.5

19

12.5

12.5

58

36

36

39

53

53

63

62

40

50

49

4.5

3.9

4.0

3.9

4.6

4.6

5.2

4.2

4.2

4.7

4.0

4.0

4.0

5.4

7.0

4.0

7.0

7.0

4.0

7.0

7.0

7.0

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

ENDURANCE LIMIT
In conjunction with the introduction of the ELHMAP, the concern has been raised about
exactly what fatigue curve should be used for the structural design of these thick
pavements. Because of the thicknesses involved with these and traditional full depth
pavements used in both highways and airport pavements the strains at the bottom of the
layer are extremely low. These low tensile strain levels, typically well below 100 micro
strain are used in the structural design through a linear extrapolation of the traditional
fatigue data shown in Figure 1 which are developed above 300 micro strain. It has long
been felt that fatigue behavior at low strain levels does not follow the same relationship
as the material subjected to strains at the normal levels. Indeed, it has even been
postulated that there is a strain limit below which there is no fatigue damage.
The existence of a fatigue endurance limit has been postulated in the past by
Monismith et. al. (5). The available data indicated that a strain level in the 70 micro
strain range appeared to produce an extraordinarily long fatigue life. However, there was
not sufficient test data to substantiate this observation, and it has gone largely
uninvestigated over the years. The significance of the endurance limit is that such a value
would provide a thickness limit for the pavement, given the materials used for
construction. Increasing the thickness beyond that established by the endurance limit
would provide no increased structural resistance to fatigue damage and would represent
an unneeded expense.
The verification of a fatigue endurance limit would appear to be answerable
merely by conducting fatigue testing at the low strain levels to 70 micro strain and
comparing the results with the other strain levels. The following sections illustrate that
the traditional laboratory fatigue data, which can clearly indicate a fatigue endurance
limit, do not provide sufficient fundamental data that could be used to mechanistically
validate the observed phenomenological performance difference between normal strain,
and low strain test results.
Traditional Fatigue Analysis
A traditional fatigue analysis is performed on data collected following AASHTO
procedures (1) which require a 10 Hz haversine load with constant strain at 20 C. The
number of load repetitions when the initial modulus decreases to 50 percent of the initial
value is defined as loads to failure, Nf. This failure value is plotted against the applied
tensile strain measured during the initial 50 load cycles to develop one data point for the
fatigue curve. The test is repeated at various strain levels from 1000 to 250 micro strain,
depending on the material, to develop a complete traditional fatigue curve. The curves
shown in Figure 1 are representative of the strain levels utilized in traditional testing
programs. The fatigue equations shown are developed from the best fit to these data.
The K1 and K2 parameters for the mixtures tested at normal strain levels are given in
Table 2.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

Table 2. K1 and K2 Fatigue Parameters for Mixtures Tested


Mix 6-7
Mix 64-7
Mix 5-7
Mix 7-7
Mix 8-7
Mix 8-4
Mix 9-7
Mix 17-7
Mix 21-4
Mix 11-7
Mix 20-7

K1
2.112 E-12
3.743 E-10
8.928 E-11
5.684 E-09
3.196 E-12
1.630 E-10
3.645 E-11
2.144 E-11
1.022 E-10
3.299 E-12
1.385 E-10

K2
-4.992
-4.168
-4.411
-3.951
-5.068
-4.528
-4.603
-4.789
-4.303
-4.889
-4.303

-7, or 4 indicates air void level

The structural design of the asphalt pavement is performed using these curves
regardless of the number of load repetitions required and the resultant tensile strain level
developing in the pavement. This typically requires a linear extrapolation of the curve to
lower strain levels for higher load applications.
Dissipated Energy Ratio Analysis
The dissipated energy ratio analysis was developed by Ghuzlan and Carpenter (6, 7) that
refined work done by Carpenter and Jansen (8) and built on the dissipated energy work
done by other researchers (2, 9, 10). These other researchers have used the total
dissipated energy, or the dissipated energy curve to relate to fatigue failure. Rowe has
examined the rate of change in dissipated energy to indicate fatigue performance with
good results. Of these approaches, it is the change in dissipated energy examined by
Jansen and by Rowe that can provide a mechanistic picture of how damage accumulates
in the fatigue test.
However, the rate of change in dissipated energy by itself does not provide for a
single unified method to examine failure in different test modes. That is, the precise
same variable or procedure is not used in each mode to define failure. Thus, failure is
still determined differently for the different fatigue modes. To overcome this difficulty,
Ghuzlan and Carpenter (7) examined a ratio of the change in dissipated energy between
two cycles divided by the dissipated energy of the first cycle, represented as:
DER = [DEn+1 DEn]/DEn
Where:

DER is the Dissipated energy ratio


DEn is the dissipated energy produced in load cycle n
DEn+1 is the dissipated energy produced in load cycle n+1

This ratio provides a true indication of the damage being done to the mixture from
one cycle to another as a function of how much dissipated energy was involved in the
previous cycle.
Different loading conditions will produce different dissipated energy hysteresis
curves (stress versus strain plot). Because damage is the difference in dissipated energy,

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

Dissipated Energy Ratio

this ratio clearly illustrates the percent of the input dissipated energy that goes into
damage for a cycle. This representation of damage produces curves similar to that shown
schematically in Figure 3. There are three distinct portions to the curve. Of interest here
is portion II which is an extended level plateau in the data plot. This plateau value
represents a period where there is a constant percent of input energy being turned into
damage. This value appears to be a mixture and load/strain input related value. For any
one mixture the plateau value is a function of the load inputs, and for similar load inputs,
the plateau value is different for different mixtures (6).
Portion III of the curve in Figure 3 represents ultimate failure in the mixture. The
upturn in the curve indicates that more damage is being done per load cycle (input
energy) with each subsequent load cycle. This represents an unstable condition in the
mixture, and ultimately the mixture has no load carrying capability. This failure point,
defined by the onset of unstable damage accumulation, occurs for all fatigue modes,
providing a means of producing fatigue curves relating the plateau value of the dissipated
energy ratio to the number of loads to a true failure in the laboratory.

III

II

Plateau Value

Log Load Repetitions

Figure 3. Typical Dissipated Energy Ratio Plot with Three Behavior Zones
The work by Ghuzlan(7) clearly establishes this approach as a unifying approach
between all modes of loading, and types of loads. The analysis procedure produces one
unique fatigue curve for a mixture developed from the precise same analysis procedure
for all conditions. Additionally, the true failure point from this analysis has been
shown to be strongly related to the point of 50 percent stiffness selected by Monismith
(11) as representing a value to use that relates to field failure in asphalt pavements. For
the 79 different mixtures tested by Ghuzlan(6), the relationship is:
NTF = 21758 + 1.30727(N50)
Where:

(2)

NTF is the loads to failure defined by the DER approach


N50 is the number of loads to reduce stiffness by 50 percent.
R2 = 0.91, and SEE = 16,858

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

The importance of this relationship is that the use of an intermediate point of


damage accumulation that has been related to field failure of a pavement, such as 50
percent for unmodified binders or a higher number for modified binders, can be related to
a materials true failure point of ultimate damage accumulation. The establishment of
such a relationship supports the use of an intermediate number such as the 50 percent
stiffness reduction for failure, and eliminates the need for long-term testing to establish
the true failure point. The validation of this connection between laboratory true
failure and field failure emphasizes the importance to define the material failure point
and the field failure point for different mixtures. The mixtures tested in establishing this
relationship (6) included PG binders, pen graded binders, viscosity graded binders, and
modified binders using SBR and EVA modifiers.
LOW STRAIN FATIGUE TESTING
The dissipated energy ratio procedure provides an easy mechanistic means of
examining the energy handling capability of a mixture as it relates to fatigue behavior in
a manner not possible with traditional methods, especially important when testing at low
strain levels. Flexural fatigue testing at low strain levels is problematic because it is not
possible to devote sufficient machine time to take every sample to failure. Thus, there is
very little data showing flexural fatigue test results beyond several million load
repetitions. Mixture 6 and 64 were tested to 38 and 46 million load repetitions,
respectively and failure was extrapolated. The remaining mixtures were tested to 5 or 8
million load repetitions. Even after this abbreviated testing the plateau value is flat, and
extrapolation to establish the point of 50 percent stiffness reduction is reasonable.
This testing generated the stiffness and dissipated energy ratio vs. load repetition
data shown in Figure 4 for mix 6 which are typical of those for all mixtures. The plateau
value for the dissipated energy ratio is flat, and very low, indicating that a stable region of
damage accumulation has been achieved. The stiffness reduction trend is consistent with
normal testing, but is nowhere near reaching the 50 percent reduction defining failure.
8000

0.0008

0.0007

7000

0.0005

DDE/DE

Stiffness, M p a

0.0006

6000

0.0004

0.0003

0.0002

5000

0.0001

0
1.0E+04

4000
0

5,000,000

10,000,000

15,000,000

20,000,000

25,000,000

30,000,000

35,000,000

40,000,000

1.0E+05

1.0E+06

1.0E+07

1.0E+08

Load Cycles

Number of Load Cycles

(a)
(b)
Figure 4. Stiffness Reduction (a), and Dissipated Energy Ratio (b) for Mix 6 at 70
Microstrain in Flexure.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

10

Figure 5 shows the complete fatigue data results for normal strain levels plotted in
the traditional fatigue analysis manner. The normal strain tests clearly indicate that the
fatigue curves are different for each mixture, and must be handled as different materials
for structural fatigue design.
Mix 8-4

Mix21-4

Mix 6-7

Mix 64.7

Mix 17-7

Mix 5-7

Mix 11-7

Mix 20-7

Mix 7-7

Mix 9-7

Mix 8-7

Tensile Strain, MIcro

10000

1000

100
100

1000

10000

100000

1000000

10000000 100000000

Load Repetitions to Nf

Figure 5. Traditional Fatigue Plot Including Low Strain Test Data.


Figure 6 presents the fatigue test data with the low strain test data included. This
plot indicates that there appears to be a distinctly different performance at the low strains
compared to the normal strain levels for all mixes tested. Some mixtures appear to reach
the extended life more slowly than others. Comparing mix 11 to mix 64 in Figure 6 it
can be seen that at strain levels around 100 micro strain mix 11 will sustain more load
repetitions to failure than mix 64. This indicates that the endurance elimit could be
different for different mixtures. There is nothing unique about the 70 micro strain level
for an endurance limit although it would appear to be a useful value that falls within what
could be considered the strain level where an endurance limit would be encountered.
More importantly, the phenomenological relationship shown in Figure 6 does not provide
a mechanistic connection between behavior of the mixtures at low and normal strain
levels. Lacking such a fundamental connection between a mechanistically determined
material property and the observed behavior at low strains a valid explanation cannot be
developed from this data.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

11

Mix 8-4

Mix 21-4

Mix 6-7

Mix 64-7

Mix 17-7

Mix 5-7

Mix 11-7

Mix 20-7

Mix 7-7

Mix 9-7

Mix 8-7

Flexural Strain, Micro

10000

1000

100

10
1.E+01 1.E+06 1.E+11 1.E+16 1.E+21 1.E+26 1.E+31 1.E+36 1.E+41

Load Reps to Nf

Figure 6. Traditional Fatigue Representation With Low Strain Behavior


Extrapolated.
The fundamental explanation needed to support a hypothesis for an endurance
limit can be obtained from the dissipated energy analysis of the fatigue data. Figure 7
presents the DER plateau value plotted against load repetitions to failure for the mixtures
studied here. The plateau values for the low strain tests fall precisely into line with the
data from all strain levels establishing a unique relationship between failure and plateau
value of the DER. From a mechanistic viewpoint this uniqueness is to be expected when
using a fundamental relationship. A change in damage accumulation results in a
consistent corresponding change in load repetitions to failure. This consistency indicates
that the energy dissipating mechanism of the mixture is consistent over all levels of
energy input, resulting in a unique relationship with loads to failure. However, this
relationship in itself does not support a fatigue endurance limit, something which requires
a discontinuity in behavior.
The discontinuity indicating a change in behavior at low strain levels becomes
evident in a plot of the DER plateau value as a function of the tensile strain as shown in
Figure 8. At normal strain levels the change in the rate of damage accumulation is
proportional to the change in the strain level. However, at low strain levels a nonlinearity is introduced, and the proportionality found at normal strain levels is not
continued. The amount of damage done per cycle changes at strains below about 100
micro strain, and is extremely evident at 70 micro strain for the mixtures tested here.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

DER Plateau Value

Carpenter, Ghuzlan, and Shen

12

Mix 8-4

Mix 21-7

Mix 6-7

Mix 64-7

Mix 17-7

Mix 5-7

Mix 11-7

Mix 20-7

Mix 7-7

Mix 9-7

Mix 8-7

0.01
1E-05
1E-08
1E-11
1E-14
1E-17
1E-20
1E-23
1E-26
1E-29
1E-32
1E-35
1E-38
1E-41
1

100000

1E+10

1E+15

1E+20

1E+25

1E+30

1E+35

1E+40

Load Repetitions to Nf

Figure 7. Dissipated Energy Ratio Plateau Value Plotted Against Load Repetitions
to Failure.
Mix 8-4

Mix 21-4

Mix 6-7

Mix 64-7

Mix 17-7

Mix 5-7

Mix 11-7

Mix 20-7

Mix 7-7

Mix 9-7

Mix 8-7

Flexural Strain, Micro

10000

1000

100

10
1E-41

1E-36

1E-31

1E-26

1E-21

1E-16

1E-11

1E-06

0.1

Plateau Value, DER

Figure 8. Flexural Strain Plotted Against Plateau Value Illustrating the


Differentiated Behavior at Low-Strain Values.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

13

Figure 8 clearly establishes a different energy into damage response in the


mixtures at the lower strain levels compared to normal levels. There is a fundamental
change in material behavior at low strain levels where the energy going into damage after
a load cycle is decreased substantially in comparison to normal strain results. The data
presented support the long held understanding that the strain response is not a
fundamental property in relation to failure, but only a convenient phenomenological
relationship.
Healing A Source of the Fatigue Endurance Limit
At low strain levels there is a decidedly reduced amount of damage being done
per cycle compared to normal strain level testing. The data support a gradual trend
toward the endurance limit, and not a distinct break point. The recognition of healing and
the resultant property changes it produces in a mixture can be proposed to explain this
observed non-linearity, lending credence to a physical rationale for a fatigue endurance
limit. The work being performed at Texas A&M University provides insight into this
phenomenon (17). Although not specifically addressed in this study, healing provides the
connection required to have a fatigue endurance limit.
Healing is a continual process that can be thought of as a process that returns
energy into the HMA, increasing the load carrying ability of the mixture, and in effect
repairing a portion of the damage done by the previous loads. Healing becomes most
evident when a rest period is imposed between load cycles and the healing can be seen in
the increased modulus after the rest period. In actuality this repair process is continual,
and occurs to some extent even during load cycling. At high strain levels the amount of
healing energy is relatively small in relation to the damage energy, but at low strain levels
the proportion of damage energy is smaller and could approach the energy returned to the
mixture by the healing process. Given that an asphalt aggregate combination produces a
specific amount of healing potential, there could exist a strain level at which the damage
energy was equal to the healing energy, and no damage would accumulate if the load
cycles were slow enough and total healing was allowed to occur. Even if loading was
continual, there would exist a point at which the kinetics of the healing process would
offset the load cycle damage, and little or no damage would accumulate in the HMA,
producing an extended fatigue life, a fatigue endurance limit similar to what is shown in
the testing presented here.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

14

CONCLUSIONS
The data generated from the mechanistic analysis of damage accumulation through the
dissipated energy approach clearly provides support for the existence of a fatigue
endurance limit. The normal and low strain data can be considered as two distinctly
different processes that can be represented by their individual fatigue curves as related to
tensile strain, something that cannot be substantiated from the traditional analysis.
Although the data set is currently limited to low strain testing at 70 micro strain,
the data shows that the trend is toward an extraordinarily extended fatigue life. While the
change may be a continuous function rather than a precise lower limit it would appear
that an asymptote is being approached at 70 micro strain. The exact limit is very likely
mixture/binder specific. Whether or not the 70 micro strain level is accepted as an
endurance limit it is apparent that this level is capable of providing a significantly longer
fatigue life than would be predicted from normal testing. For practical design
considerations this could be considered a limit beyond which life extension becomes
extremely long in comparison to traditional designs and load repetitions used.
The existence of the fatigue endurance limit proposed here has serious
ramification for design of asphalt pavements. There must still be a positive production
of sound MHA mixtures with adequate fatigue performance. To date, all mixtures tested
have produced appropriate K1 and K2 values which satisfy the unique phenomenological
relationship first shown by Myre(12), and verified by Ghuzlan and Carpenter(6,13).
Such mixtures have sufficient internal strength to provide satisfactory fatigue resistance
to support the existence of an endurance limit.
Given a suitable mixture the structural design for fatigue in an ELHMAP does not
require either traffic or a fatigue algorithm. When the materials and the pavement
structural section are sufficient to produce a tensile strain of around 70 micro strain, there
is no effect of traffic on fatigue life. If the rich bottom layer produces a different
modulus, this will change the strain response allowing thickness variations. Material
variability will not impact fatigue life as long as the tensile strain remains around 70
micro strain.
Thicker asphalt sections to reduce the strain below 70 micro strain are not
necessary to provide any increased factor of safety against fatigue. This concept has a
significant value to the practitioner as it provides a very simple means of selecting asphalt
layer thickness based only on modulus testing combined with the use of a suitable
response model, both of which are becoming standard elements of pavement design.
RECOMMENDATIONS
This research provides a fundamental mechanistic approach that clearly establishes a
difference in performance in asphalt mixtures at low strain levels that indicates a fatigue
endurance limit. This endurance limit requires more diverse testing, and should be
combined with emerging research that is establishing a tensile strength/strain limit (14,
15) below which fatigue damage does not appear to develop. This is an especially
intriguing relationship given the work by Maupin which previously established
relationships between tensile strength and fatigue life (16).
The testing conducted for this study was performed on standard flexural beam
equipment. Further testing at low strain levels using other sample configurations and

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

15

strain gages mounted on the specimens should be conducted. This will provide more
accurate values of the dissipated energy at low strain levels, and allow monitoring of the
specimen to ensure accurate strain at low levels over long times to reduce possibility of
creep induced because of sample geometry affecting results.
There must be intermediate testing at strain levels below the 250 micro strain
level and below the 70 micro strain level to define the nature of the transition from
normal behavior to the modified behavior at the low strain levels. The testing in this
study was limited primarily to the 70 micro strain level.
The existing database of pavements already constructed that produced a design
strain level at the bottom of the asphalt layer in the range of 70 micro strain should be
carefully examined to establish performance trends from in service pavements. This
must be done to establish any complicating factors present in the field that would
compromise the endurance limit behavior.
The impact of overloads on fatigue life of a pavement designed to the endurance
limit approach are currently under investigation to determine any detrimental impact on
the extended life performance of the pavement section of brief overload situations.
Given that healing has been demonstrated by others as a valid consideration in
fatigue life extensions, the impact of healing and asphalt composition on the energy level
required for damage accumulation should be investigated. The healing of HMA should
be tied to energy which can then be tied to mixture performance. It is likely that
fundamental studies of this nature are necessary to define and validate if and how the
fatigue endurance limit develops as has been proposed here.
ACKNOWLEDGEMENTS
The authors would like to gratefully acknowledge the funding for this study through the
Federal Aviation Administration Center of Excellence for Airport Technology at the
University of Illinois. The center is funded under Research Grant Number 95-C-001,
through the Pavement Center at the FAA. Ms. Patricia Watts is the FAA Program
Manger for Air Transportation Centers of Excellence, and Dr. Satish Agrawal is the FAA
Technical Director for the Pavement Center.
DISCLAIMER
The contents of this paper reflect the views of the authors who are responsible for the
facts and accuracy of the data presented within. The contents do not necessarily reflect
the official views and policies of the Federal aviation Administration. This paper does
not constitute a standard, specification, or regulation.
REFERENCES
1.

Standard Test AASHTO Provisional Standards, Standard Test Method for


Determination of the Fatigue Life of Compacted Hot Mix Asphalt (HMA)
Subjected to Repeated Flexural Bending. TP8-94, September, 1994

2.

SHRP, A-404, Fatigue Response of Asphalt-Aggregate Mixes. Strategic


Highway Research Program, National Research Council, 1994.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

16

3.

Thompson, M. R, and F. Hugo, Design Methods, Workshop 2, Proceedings of


the Sixth International Conference on the Structural Design of Asphalt
Pavements, Vol. 2, Ann Arbor, Michigan, 1987

4.

Perpetual Bituminous Pavements, Transportation Research Board Circular No.


503, January, 2001.

5.

Monismith,C. L., Epps, J. A., Kasianchuk, D. A., and McLean, D. B., Asphalt
Mixture Behavior in Repeated Flexure, Report No. TE 70-5, Institute of
Transportation and Traffic Engineering, University of California, Berkeley, 1970

6.

Ghuzlan, Khalid, Fatigue Damage Analysis in Asphalt Concrete Mixtures Based


Upon Dissipated Energy Concepts, Ph.D. Thesis, University of Illinois at UbanaChampaign, August, 2001.

7.

GHuzlan, K, and S. H. Carpenter, An Energy-Derived/Damage-Based Failure


Criteria for Fatigue Testing, Transportation Research Record No. 1723, p 131141

8.

Carpenter, S. H., and M Jansen, Fatigue Behavior Under New Aircraft Loading
Conditions, Aircraft/Pavement Technology, ASCE, 1997

9.

Pronk, A. C., and Hopman, P. C., Energy Dissipation: The Leading Factor of
Fatigue, Highway Research: Sharing the Benefits, Proceedings of the United
States Strategic Highway Research Program, London, 1990, pp. 255-267.

10.

Rowe, G. M., Performance of Asphalt Mixtures in the Trapezoidal Fatigue Test,


Proceedings, Association of Asphalt Paving Technologists, Vol. 62, 1993, pp.
344-384

11.

Tayebali, A. A., Deacon, J. A., Coplantz, J. S., and Monismith, C.L., Modeling
Fatigue Response of Asphalt Aggregate Mixtures, Proceedings of Association of
Asphalt Paving Technologists, Vol. 62, 1993, pp. 385-421.

12.

Myre, Josten, Fatigue of Asphalt Pavements, Third International Conference on


Bearing Capacity of Roads and Airfields, The Norwegian Institute of Technology,
Trondheim, Norway, July, 1990, pp. 703-714

13.

Ghuzlan, K. A., and S. H. Carpenter, Traditional Fatigue Analysis of Asphalt


Concrete Mixtures, Paper submitted for possible presentation and publication by
the Transportation Research Board at Annual Meeting, January 2003.

14.

B. E. Ruth, R. Roque, and B. Nununya, Aggregate Gradation Characterization


Factors and Their Relationships to Fracture Energy and Failure Strain of Asphalt
Mixtures, Paper Presented at the 2002 meeting of the Association of Asphalt
Paving Technologists, March 2002.

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

Carpenter, Ghuzlan, and Shen

17

15. Daniel, J. S, and Y. R. Kim, Development of a Simplified Fatigue Test and


Analysis Procedure Using a Viscoelastic Continuum Damage Model, Paper
Presented at the 2002 Meeting of the Association of Asphalt Paving
Technologists, March 2002
16. Maupin, G. W., and Freeman, J. R., Simple Procedure for Fatigue
Characterization of Bituminous Concrete, FHWA-RD-76-102, 1976
17. Kim, Yong-Rak, D. Little, and R. Lytton, Use of Dynamic Mechanical Analysis
(DMA) to Evaluate the Fatigue and Healing Potential of Asphalt Binders in Sand
Asphalt Mixtures, Paper presented at the Annual Meeting of the Association of
Asphalt Paving Technologists, 2002

TRB 2003 Annual Meeting CD-ROM

Paper revised from original submittal.

You might also like