You are on page 1of 117

Technical Report

AP-T319-16

Asphalt Fatigue Damage Healing


and Endurance Limits
Guide Implementation Options

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options
Prepared by

Publisher

Geoff Jameson

Michael Pickering

Austroads Ltd.
Level 9, 287 Elizabeth Street
Sydney NSW 2000 Australia
Phone: +61 2 8265 3300
austroads@austroads.com.au
www.austroads.com.au

Abstract

About Austroads

Overseas research has suggested that asphalt fatigue endurance


limit may exist, where applied strain is sufficiently low no fatigue
damage accumulates in the asphalt as the rate of asphalt healing
exceeds the rate of damage accumulation. The existence of such
limits has implications for design of thick asphalt pavements including
potential reductions in pavement thickness under certain conditions.

Austroads is the peak organisation of Australasian road


transport and traffic agencies.

Project Manager

This report:

summarises laboratory, computer modelling and field studies


conducted to identify and validate the asphalt endurance limit
concept

summarises asphalt design procedures used overseas which


incorporate an asphalt fatigue endurance limit

discusses options for revision of the Austroads pavement design


procedures.

Keywords
asphalt, pavement design, fatigue endurance limit, long-life
pavement, perpetual pavements, healing

ISBN 978-1-925451-55-9
Austroads Project No. TT2044
Austroads Publication No. AP-T319-16
Publication date December 2016
Pages 109
Austroads 2016

Austroads purpose is to support our member organisations to


deliver an improved Australasian road transport network. To
succeed in this task, we undertake leading-edge road and
transport research which underpins our input to policy
development and published guidance on the design,
construction and management of the road network and its
associated infrastructure.
Austroads provides a collective approach that delivers value
for money, encourages shared knowledge and drives
consistency for road users.
Austroads is governed by a Board consisting of senior
executive representatives from each of its eleven member
organisations:

Roads and Maritime Services New South Wales

Department of State Growth Tasmania

Transport Canberra and City Services Directorate,


Australian Capital Territory

Australian Government Department of Infrastructure and


Regional Development

Australian Local Government Association

Roads Corporation Victoria


Queensland Department of Transport and Main Roads
Main Roads Western Australia
Department of Planning, Transport and Infrastructure
South Australia
Department of Infrastructure, Planning and Logistics
Northern Territory

New Zealand Transport Agency.

This work is copyright. Apart from any use as permitted under the
Copyright Act 1968, no part may be reproduced by any process
without the prior written permission of Austroads.
This report has been prepared for Austroads as part of its work to promote improved Australian and New Zealand transport outcomes by
providing expert technical input on road and road transport issues.
Individual road agencies will determine their response to this report following consideration of their legislative or administrative
arrangements, available funding, as well as local circumstances and priorities.
Austroads believes this publication to be correct at the time of printing and does not accept responsibility for any consequences arising from
the use of information herein. Readers should rely on their own skill and judgement to apply information to particular issues.

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Summary
The asphalt pavement thickness design process of the Guide to Pavement Technology Part 2: Pavement
Structural Design (Austroads 2012) utilises a mechanistic-empirical procedure to predict the allowable traffic
repetitions until asphalt fatigue occurs. In practice, traffic loading is a design input in an iterative process and
increases in traffic loadings necessitate increases in asphalt design thickness to ensure fatigue requirements
are met. With the current fatigue based design approach, asphalt design thickness will continue to increase
in the future due to ever increasing traffic loadings and the low asphalt moduli in some areas of Australia
(due to the warm climate). Asphalt thicknesses for some heavy duty full depth asphalt pavements in Australia
have exceeded 400 mm in hotter climates. There is concern that such designs are overly conservative.
Researchers have postulated that an asphalt fatigue endurance limit exists where applied strain is sufficiently
low no damage occurs. The premise is that the rate of asphalt healing exceeds the rate of damage
accumulation, being fatigue cracking at the base of the asphalt pavement. The corresponding thickness
would provide a limit for the pavement where additional asphalt thickness provides no additional benefit to
fatigue life.
A number of research projects have investigated field pavement performance to assess the evidence for an
endurance limit. The most comprehensive field investigation on performance of thick asphalt pavements was
undertaken in the United Kingdom (UK) by Nunn et al. (1997). They concluded that for the UK climate,
materials and traffic loading a total 270 mm thickness of asphalt on a firm foundation was sufficient for a long
and indeterminate life. It is considered that the UK data is of value in developing a design procedure for the
Guide.
To enable the UK performance data to be adapted for use in the Guide, a model needs to be developed to
predict how endurance strains estimated from UK field performance data can be extended to Australian
climates, asphalt mixes and traffic loadings. Two major sources of data for this model are the various
research projects undertaken by the University of Illinois and the recently completed Federal Highway
Administration (FHWA) research project, National Cooperative Highway Research Program (NCHRP) project
9-44A at the Arizona State University.
Against this background a number of options for inclusion of endurance limit concepts are described.
The Australian Asphalt Pavement Association (AAPA) has recently undertaken a research project titled
Asphalt Pavement Solutions for Life (APS-fL) which proposes another procedure to predict endurance limits
for mix modulus. This method is summarised and its impact on design thicknesses is illustrated.
In addition two options have been developed using laboratory endurance limit strain (ELS) relationships
derived in NCHRP project 9-44A and UK pavement performance data. Their development is detailed and
impact on design thicknesses illustrated.
A fourth option is to provide an upper limit of design traffic loading as a simplification to the above options.
Lastly an option was developed which addresses perceived over-design at high temperatures by varying the
laboratory-to-field shift factors in the asphalt fatigue relationship.
These options were discussed at Austroads Pavement Structures Working Group and Pavements Task
Force meeting and it was agreed to place limits on design traffic loadings. Based on these discussions, the
report provides text for the revision of Part 2.

Austroads 2016 | page i

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Contents
Summary ......................................................................................................................................................... i
1.

Introduction ............................................................................................................................................ 1

2.

University of Illinois Laboratory Research .......................................................................................... 2


2.1 Introduction ....................................................................................................................................... 2
2.2 Use of Rate of Dissipated Energy Change (RDEC) ......................................................................... 3
2.3 Fatigue Life Prediction for Mix Modulus and Volumetrics ................................................................ 5
2.4 Discontinuity in Laboratory Fatigue Relationship ............................................................................. 7
2.5 Use of Plateau Value to Estimate Endurance Strain Limit ............................................................... 9
2.6 Laboratory ELS Relationships ........................................................................................................ 10
2.6.1
2.6.2

Prediction of ELS from Mix Modulus ................................................................................. 10


ELS Prediction from Mix Modulus and Volumetrics .......................................................... 12

2.7 Effect of Rest Periods on ELS ........................................................................................................ 16


2.8 Summary ........................................................................................................................................ 18
3.

NCHRP Project 9-44A Laboratory Endurance Limits ....................................................................... 19


3.1 Introduction ..................................................................................................................................... 19
3.2 Overview of Methodology ............................................................................................................... 19
3.3 Materials Tested ............................................................................................................................. 21
3.4 Beam Fatigue Testing .................................................................................................................... 22
3.5 Test Program .................................................................................................................................. 22
3.6 Data Used to Develop the Stiffness Ratio Prediction Method........................................................ 24
3.7 Relationship to Predict Stiffness Ratio from Modulus .................................................................... 24
3.8 Laboratory Beam Endurance Limit Strains .................................................................................... 26
3.9 Further Analysis of Laboratory Beam Results ............................................................................... 28
3.9.1
3.9.2

Stiffness Ratio Prediction Relationship ............................................................................. 29


Laboratory Beam Endurance Limits in Terms of Volumetrics and Temperature .............. 30

3.10 Summary ........................................................................................................................................ 32


4.

Field Studies ......................................................................................................................................... 33


4.1 United Kingdom .............................................................................................................................. 33
4.2 Australia.......................................................................................................................................... 34
4.3 United States of America................................................................................................................ 37
4.3.1
4.3.2

Analysis of USA LTPP Data .............................................................................................. 37


Analysis of NCAT Accelerated Loading Data .................................................................... 38

4.4 Summary of Field Studies .............................................................................................................. 42


5.

Asphalt Pavement Solutions for Life Project .................................................................................... 43


5.1 Introduction ..................................................................................................................................... 43
5.2 Outline of the Proposed Method .................................................................................................... 44
5.3 Summary of Method Development................................................................................................. 45
5.3.1
5.3.2
5.3.3
5.3.4
5.3.5
5.3.6

Introduction ........................................................................................................................ 45
Laboratory ELS Relationship............................................................................................. 45
Selection of Pavement Sections for Field Calibration ....................................................... 48
Determination of Moduli and Asphalt Strains .................................................................... 54
Determination of In-service ELS Relationship ................................................................... 55
In-service ELS Relationship Proposed for the Guide ........................................................ 58

Austroads 2016 | page ii

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

5.4 Summary ........................................................................................................................................ 59


6.

Use of Asphalt Fatigue Endurance Limits in Overseas Design Processes ................................... 61


6.1 Introduction ..................................................................................................................................... 61
6.2 United Kingdom .............................................................................................................................. 61
6.3 United States of America................................................................................................................ 63
6.3.1
6.3.2
6.3.3

AASHTO Design Method .................................................................................................. 63


Texas Department of Transportation................................................................................. 63
Illinois Department of Transportation ................................................................................ 67

6.4 Summary ........................................................................................................................................ 69


7.

Options for Austroads Design Guide ................................................................................................ 70


7.1 The Need ........................................................................................................................................ 70
7.2 Option A: APS-fL Method to Estimate ELS from Asphalt Modulus ................................................ 70
7.3 Option B: NCHRP 9-44A Method to Calculate ELS from Modulus ................................................ 74
7.4 Option C: NCHRP 9-44A Estimating ELS from Temperature and Mix Volumetrics ...................... 76
7.5 Option D: Limiting Design Traffic Loading...................................................................................... 78
7.6 Option E: Modify Fatigue Relationship to Allow for Healing at Elevated Temperatures ................ 78
7.7 PSWG Recommendation for the Austroads Guide ........................................................................ 80

8.

Summary ............................................................................................................................................... 82

References ................................................................................................................................................... 84
Appendix A Asphalt Pavement Solutions for Life Design Examples ................................................... 87
Appendix B In-Service Endurance Strain Limits Derived from NCHRP Project 9-44A .................... 101
Appendix C Proposed Part 2 Text .......................................................................................................... 109

Tables
Table 2.1:
Table 2.2:
Table 2.3:
Table 2.4:
Table 3.1:
Table 3.2:
Table 4.1:
Table 4.2:
Table 5.1:
Table 5.2:
Table 5.3:
Table 5.4:
Table 5.5:
Table 5.6:
Table 6.1:
Table 6.2:
Table 6.3:
Table 6.4:
Table 7.1:
Table 7.2:
Table 7.3:

Range of test and mix variables used to derive fatigue relationship........................................... 5


Transition strains and fatigue life for 19 IDOT 03 mixtures .......................................................... 9
Mix composition and predicted ELS ........................................................................................... 11
Comparison of predicted ELS with those extrapolated from the low strain fatigue testing ........ 13
Factor combinations used in beam and uniaxial fatigue tests ................................................... 23
Predicted ELS for the PG64-22 mixes ....................................................................................... 26
Observations and assessment of Victorian sites investigated ................................................... 36
Summary of section performance .............................................................................................. 40
Recommended k1 values ............................................................................................................ 44
Fatigue, modulus and ELS results of Australian mixes ............................................................. 46
Sites classified as LLAP and used in the derivation of the ELS criteria .................................... 49
Sites classified as non LLAP from Roads and Maritime database ............................................ 50
VALMON sites used to develop the APS-fL method ................................................................. 51
Moduli back-calculated for the Australian LLAP sites used in the derivation of the
ELS criteria ................................................................................................................................ 55
Example long-life asphalt pavement .......................................................................................... 62
In situ asphalt moduli estimated from deflection testing of perpetual pavements ..................... 64
Typical TxDOT design moduli at a temperature of 25 C .......................................................... 66
Example perpetual pavement design using TxDOT method ..................................................... 66
Example shift factors (SF) for asphalt fatigue ............................................................................ 79
Example reliability factors (RF) for asphalt fatigue .................................................................... 79
Suggested upper limits on design traffic for asphalt fatigue ...................................................... 81

Austroads 2016 | page iii

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figures
Figure 2.1:
Figure 2.2:
Figure 2.3:
Figure 2.4:
Figure 2.5:
Figure 2.6:
Figure 2.7:
Figure 2.8:
Figure 2.9:
Figure 2.10:
Figure 2.11:
Figure 2.12:
Figure 2.13:
Figure 3.1:
Figure 3.2:
Figure 3.3:
Figure 3.4:
Figure 3.5:
Figure 3.6:
Figure 3.7:
Figure 3.8:
Figure 3.9:
Figure 3.10:
Figure 3.11:
Figure 3.12:
Figure 4.1:
Figure 4.2:
Figure 4.3:
Figure 4.4:
Figure 4.5:
Figure 4.6:
Figure 4.7:
Figure 4.8:
Figure 5.1:
Figure 5.2:
Figure 5.3:
Figure 5.4:
Figure 5.5:
Figure 5.6:
Figure 5.7:
Figure 5.8:
Figure 5.9:
Figure 5.10:
Figure 6.1:
Figure 6.2:
Figure 6.3:
Figure 6.4:
Figure 7.1:
Figure 7.2:
Figure 7.3:
Figure 7.4:
Figure 7.5:
Figure 7.6:

Typical dissipated energy ratio (DER) plot ................................................................................... 3


Examples of variation of PV with fatigue life ................................................................................ 4
Comparison of predicted and measured PV ................................................................................ 6
Examples of tolerable strains for a fatigue life of 1.1 x107 ESA ................................................... 7
Fatigue results at normal strain levels ......................................................................................... 7
Combining normal fatigue testing data with extrapolated lives at low strain levels ..................... 8
PV life relationship .................................................................................................................. 10
Relationship between ELS and flexural modulus from beam fatigue testing ............................. 11
Comparison of predicted ELS with those extrapolated from fatigue testing .............................. 14
Predicted variation in ELS with modulus and mix volumetrics................................................... 15
Variation of PV with rest period at a test temperature of 20 C ................................................. 16
PV-Nf relationship for various rest periods ................................................................................ 16
Example of the influence of rest period on ELS at a temperature of 20 C ............................... 17
Typical SR versus number of load cycles for tests with and without a rest period .................... 20
Schematic demonstration of NCHRP 9-44A methodology ........................................................ 20
Beam fatigue testing equipment................................................................................................. 22
Data points used to develop the SR prediction equation ........................................................... 24
Measured versus predicted stiffness ratios ................................................................................ 25
Laboratory endurance limit strains calculated using the modulus model .................................. 26
Predicted effect of temperature, bitumen and air voids from the ELS-modulus relationship ..... 27
Initial flexural modulus variation with temperature and mix composition ................................... 28
Comparison of measured and predicted stiffness ratios ............................................................ 29
Beam endurance limit strains for 5 seconds rest period ............................................................ 30
Effect of temperature and bitumen content ................................................................................ 31
Effect of temperature and air voids content ............................................................................... 31
Example of top-down cracking ................................................................................................... 36
Core from Bell Street Preston showing deep structural cracking .............................................. 36
Survival curve ............................................................................................................................. 37
NCAT pavement cross-sections at commencement of Phase II loading in 2003 ...................... 38
NCAT pavement cross-sections at commencement of Phase III loading in 2006 ..................... 39
Cumulative strain distributions from the three test cycles .......................................................... 40
Limiting measured strain for long-life pavements ...................................................................... 41
Refined limiting strain criteria ..................................................................................................... 41
APS-fL proposed in-service endurance limit strains .................................................................. 45
Laboratory ELS derived by combining data from Australian and Illinois mixes ......................... 47
Variation in falling weight deflectometer maximum deflections with cumulative traffic loading . 52
Assessment of VALMON sites as potential long-life pavements ............................................... 53
Strain distributions and initial design curve ................................................................................ 55
Normal score of ELS .................................................................................................................. 56
Indeterminate zone between the two solid black lines ............................................................... 57
Proposed in-service ELS for 95% and 80% confidence levels .................................................. 58
Comparison of laboratory ELS relationship with the in-service ELS relationships for 95%
confidence level ......................................................................................................................... 59
Comparison of laboratory ELS ................................................................................................... 60
UK design thicknesses for flexible pavements........................................................................... 61
General perpetual pavement design .......................................................................................... 65
IDOT asphalt design moduli ....................................................................................................... 68
Maximum asphalt thicknesses based on an endurance limit strain of 70 microstrain ............... 69
Example of variation in asphalt thickness with temperature ...................................................... 70
Option A relationship to predict ELS from modulus, 95% confidence ....................................... 71
Examples of the impact of Option A on design thicknesses calculated using the current
Guide ......................................................................................................................................... 72
Example of effect of air voids on ELS ........................................................................................ 73
Comparison of ELS variations with asphalt temperature ........................................................... 74
Comparison of Option A and Option B relationships to predict ELS from modulus................... 74

Austroads 2016 | page iv

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 7.7:
Figure 7.8:
Figure 7.9:
Figure 7.10:
Figure 7.11:

Examples of the impact of Option B on design thicknesses ...................................................... 75


Option C relationship to predict ELS from temperature and bitumen content at 5% air voids .. 76
Option C adjustment of ELS for air voids ................................................................................... 77
Examples of the impact of Option C on design thicknesses ...................................................... 77
Examples of the impact of Option E on design thicknesses ...................................................... 80

Austroads 2016 | page v

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

1. Introduction
The Austroads asphalt pavement design process of the Guide to Pavement Technology Part 2: Pavement
Structural Design (Austroads 2012) utilises a mechanistic-empirical procedure to predict the allowable traffic
repetitions until asphalt fatigue occurs. In practice, traffic loading is a design input in an iterative process and
increases in traffic loadings necessitate increases in asphalt design thickness to ensure fatigue requirements
are met. With the current fatigue based design approach, asphalt design thickness will continue to increase
in the future due to ever increasing traffic loadings and the low asphalt stiffness achieved in many areas of
Australia (due to warm climate). Asphalt thicknesses for some recently designed heavy duty full depth
asphalt pavements in Australia have exceeded 400 mm.
Overseas research has suggested that an asphalt fatigue endurance limit may assist, where applied strain is
sufficiently low no damage accumulates to the asphalt as the rate of asphalt healing exceeds the rate of
damage accumulation. The existence of such a limit has implications for design of thick asphalt pavements
including potential reductions in pavement thickness under certain conditions.
A previous Austroads report (Austroads 2009) described the endurance limit concepts and proposed several
options for implementation in the Guide. At the time of preparing this 2009 report a major research project in
the United States of America (NCHRP 9-44A) was about to commence and Austroads deferred a decision on
the implementation of endurance limits in the Guide until it was completed. In addition, in 2012 the Australian
Asphalt Pavement Association commenced a project titled Asphalt Pavement Solutions for Life (APS-fL).
In 2015, Austroads commissioned ARRB to review all relevant research findings and investigate options for
the implementation of endurance limits/healing for the anticipated 2017 edition of the Guide.
Building on the 2009 Austroads report, this report:

describes and investigates the findings of the extended programs of laboratory research undertaken by
the University of Illinois (Section 2) and by Arizona State University for NCHRP 9-44A (Section 3)

summarises research field studies and accelerated loading trials to identify long-life asphalt pavement
performance (Section 4)

summarises the findings and recommendations of the AAPA Asphalt Pavement Solutions for Life project
(Section 5)

outlines how long-life asphalt pavements have been incorporated by road agencies internationally
(Section 6)

discusses options to improve the thickness design of asphalt pavements in the Guide (Section 7)
summarises the findings (Section 8).
There are various methods of providing for asphalt fatigue endurance limits (FEL) in pavement design
methods, including limiting strains, limiting thickness and limiting design traffic loadings. In this report, the
strain-based criteria is referred to as endurance limit strain (ELS) rather than the more general term FEL.

Austroads 2016 | page 1

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

2. University of Illinois Laboratory Research


2.1

Introduction

Researchers at the University of Illinois at Urbana-Champaign have completed an extensive program of


laboratory beam flexural fatigue testing to develop a dissipated energy related fatigue criteria and endurance
limit strain (ELS).
A wide variety of asphalt mixes were included in the research (Shen 2006). Some mixes were obtained
directly from construction districts within the state of Illinois, other US states and overseas. In some cases
field-mixed samples were transported to the laboratories and reheated prior to compaction. Other mixes
were specially prepared at the Advanced Transport Research and Engineering Laboratory (ATREL) of the
University.
The preparation of beam specimens was the same regardless of the mix origin. Asphalt slabs were prepared
by heating the loose asphalt to 135 C and compacting it into a steel mould (internal dimensions
375 mm x 125 mm x 75 mm) using a laboratory rolling wheel compactor. Two test beams were then cut from
each slab. (Carpenter, Ghuzlan & Shen 2003, Shen 2006).
The mixes included those made with conventional bituminous binder (PG64-22) and also mixes made using
polymer modified binders.
Beam fatigue testing was undertaken in accordance with American Association of State Highway and
Transportation Office (AASHTO) T321-03 (AASHTO 2003) procedures, using 10 Hz haversine continuous
cyclic loading at a test temperature of 20 C. Generally controlled strain testing was used. Except for a
limited investigation of healing (Section 2.7), the haversine loading was applied continuously without rest
periods. The number of load repetitions at which the beam modulus reduced to 50% of the initial value was
defined as the fatigue life.
Section 2.2 describes the use of rate of dissipated energy change (RDEC) instead of strain to predict fatigue
life as it has the advantage of unifying the fatigue results of a wide range of asphalt mix types and modes of
loading (controlled stress and controlled strain).
Section 2.3 describes how a model was developed to predict laboratory fatigue life from mix modulus and
volumetrics.
Section 2.4 reports the significant finding that when beams were tested at low strain levels the fatigue
damage that accumulated supported the hypothesis of ELS.
Section 2.5 explains how the Plateau Value of RDEC was related to the ELS.
Section 2.6.1 describes the method of predicting laboratory ELS.
Section 2.7 summaries some limited testing to investigate how rest periods during loading increased the
ELS.
Note that the tensile strains reported in this research are those under haversine loading. As reported by
Pronk et al. (2010) and Austroads (2015a) early in the test (within the first couple of load cycles), the beam
deforms under the haversine load pulse. Due to the deformation, the haversine displacement pulse starts to
induce a sinusoidal load response. In effect, the beam is subjected to a sinusoidal strain pulse of half the
intended haversine strain amplitude. Hence the reported haversine strains are twice the actual tensile strains
being applied. The research in this section of the report is discussed without this strain correction. However,
when the results are compared to those of NCHRP 9-44A in Section 3, the strain values were halved.

Austroads 2016 | page 2

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

2.2

Use of Rate of Dissipated Energy Change (RDEC)

Carpenter and Shen (2006) outline the dissipated energy concept and the rate of dissipated energy change
(RDEC). When sustaining cyclic fatigue loading, the viscoelastic asphalt traces different paths for the
unloading and loading cycles and creates a hysteresis loop. The area inside a stress-strain loop is the
energy dissipated due to a loading cycle. The differences between the amounts of energy dissipated during
the fatigue test indicate the amount of dissipated energy that is producing damage.
To describe the percentage of dissipated energy that is producing fatigue damage, Shen and
Carpenter (2005) proposed the use of the RDEC defined in Equation 1.
RDECa =

where

DEn+1 DEn
DEn

RDECa

average ratio of dissipated energy change at cycle n compared with the next cycle

energy dissipated in cycle n

DEn+1

energy dissipated in cycle n+1

DEn

The RDEC general procedure consisted of measuring the dissipated energy for each loading cycle, curve
fitting the measured dissipated energy versus loading cycle data, to negate testing noise, and obtaining a
best-fit equation. The RDEC is then calculated as the ratio of dissipated energy change between two loading
cycles divided by the number of cycles between the two. Typically the RDEC was determined in increments
of 100 cycles however large increments were used when the dissipated energy change between every 100
cycles was too small to recognise. A regression analysis is then undertaken to determine the average RDEC
per 100 cycles.
A typical dissipated energy ratio (DER) analysis produces a curve similar to that shown in Figure 2.1.
Carpenter, Ghuzlan and Shen (2003) identified three distinct portions to the curve. Of particular interest is
Phase II, which is an extended level plateau of RDEC. This region represents a period during which there is
a constant percentage of input energy being turned into fatigue damage. Phase III represents ultimate failure
of the mix and the upturn in the curve indicates that more damage is being done per load cycle with each
subsequent load cycle.
Figure 2.1: Typical dissipated energy ratio (DER) plot

Source: Carpenter, Ghuzlan and Shen (2003).

Austroads 2016 | page 3

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

The plateau value (PV) was defined as the RDEC value at the 50% modulus reduction point. Based on
testing over 600 beams from 98 asphalt mixes Shen and Carpenter (2005) concluded that the following
relationship (Equation 2) could be used to predict the required PV for a desired fatigue life.
0.4429

PV = N 1.1102

where
PV
Nf

plateau value, RDEC at 50% initial modulus reduction

load repetitions to 50% reduction in initial modulus

Figure 2.2 illustrates some of the data measured. According to the findings by Shen and Carpenter (2005),
this unique relationship is applicable for all asphalt mixtures, loading modes, loading levels, and testing
conditions (frequency, rest periods, etc.). In contrast such a unique relationship does not exist between strain
and fatigue life.
Figure 2.2: Examples of variation of PV with fatigue life

Source: Shen (2006).

Austroads 2016 | page 4

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

2.3

Fatigue Life Prediction for Mix Modulus and Volumetrics

Shen and Carpenter (2007) developed a relationship (Equation 3) to predict PV (without rest periods) based
on detailed statistical analysis of fatigue testing of 120 asphalt mixtures at a temperature of 20 C. The mixes
tested included mixes made using conventional and polymer modified binders, with a range of characteristics
as listed in Table 2.1.
Table 2.1:

Range of test and mix variables used to derive fatigue relationship

Variable

Mean

Standard
deviation

Range

% change in Nf for
plus one standard
change in variable

% change in Nf for
minus one standard
change in variable

Strain (microstrain)

589

286

3001000

84.0%

20.7%

Flexural modulus (MPa)

5890

1658

150010000

48.8%

143.8%

Air voids AV (%)

5.2

1.7

28

27.0%

62.6%

Volume of binder (Vb) (%)

12

1.7

921

17.0%

15.9%

Volumetric parameter
(VP) = AV/(AV+Vb)

0.29

0.08

0.110.45

33.4%

71.2%

Percentage passing maximum


nominal size (PNMS) (%)

96.6

92100

1.1%

1.5%

Percentage passing the


primary control sieve
(PPCS) (%)

40

8.7

2063

5.9%

5.4%

Percentage passing No 200


(75 m) size
(P200) (%)

4.4

1.2

2.59.5

8.4%

12.4%

13.26

2.7

7.521

7%

8.0%

Gradation parameter (GP)


Source: Shen and Carpenter (2007).

PV = 44.422 5.14 E 2.993 VP1.85 GP0.4063

where

PV

plateau value, RDEC at 50% initial modulus reduction

tensile strain applied in haversine loading (mm/mm)

flexural modulus of mix from the laboratory fatigue test (MPa)

VP

volumetric parameter VP =

GP

AV

AV+Vb

where AV = air voids of mix (%)


=

Vb = volume of binder (%)

gradation parameter GP =

PNMS PPCS
P200

where PNMS = percentage passing maximum nominal size (%)


PPCS = percentage passing the primary control sieve (%)
P200 = percentage passing No. 200 (75 m) size (%)

Austroads 2016 | page 5

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 2.3 compares the PV predicted from mix composition using Equation 3 with the values measured in
the laboratory testing.
Figure 2.3: Comparison of predicted and measured PV

Source: Shen and Carpenter (2007).

By substituting Equation 2 into Equation 3 and rearranging, a fatigue relationship of the conventional form
(Equation 4) can be obtained.
0.01575GP0.366

N = E2.696 4.63 VP1.67

where
N

load repetitions to 50% reduction in initial modulus

tensile strain applied in haversine loading (mm/mm)

flexural modulus of mix from the laboratory fatigue test (MPa)

VP

volumetric parameter VP =

GP

AV

AV+Vb

where AV = air voids of mix (%)


=

Vb = volume of binder (%)

gradation parameter GP =

PNMS PPCS
P200

where PNMS = percentage passing maximum nominal size (%)


PPCS = percentage passing the primary control sieve (%)
P200 = percentage passing No. 200 (75 m) size (%)

The fatigue life dependence on modulus (modulus exponent 2.696) is greater than the Austroads Guide
relationship (modulus exponent 1.8).
Plotted in Figure 2.4 is the variation in the tolerable strains with asphalt modulus for a fatigue life of 1.1 x 107
load repetitions.

Austroads 2016 | page 6

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 2.4: Examples of tolerable strains for a fatigue life of 1.1 x107 ESA
380
360
340

Air voids = 3%, Vb=12.5%

320

Air voids=5%, Vb=11.5%

300

Air voids = 7%, Vb=10.5%

280
260
240
Strain
(microstrain) 220
200
180
160
140
120
100
80
1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10000
Asphalt modulus (MPa)

2.4

Discontinuity in Laboratory Fatigue Relationship

Carpenter, Ghuzlan and Shen (2003) undertook asphalt laboratory fatigue testing of asphalt mixes in order
to verify the existence of a fatigue endurance limit for asphalt. Figure 2.5 shows the results of constant strain
haversine fatigue testing at a temperature of 20 C and 10 Hz frequency at normal strain levels. Most of the
testing was undertaken using applied strains between 250 and 1000 microstrain and the resulting fatigue
lives were generally less than 106 cycles.
Figure 2.5: Fatigue results at normal strain levels

Source: Carpenter, Ghuzlan and Shen (2003).

Austroads 2016 | page 7

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

As testing to fatigue at lower strain levels would require very long fatigue tests (>107 cycles), load repetitions
to fatigue at low strains needed to be extrapolated from the measured modulus reduction with loading. Shen
(2006) concluded that fatigue testing to 8 x 106 load repetitions is the minimum testing length to provide
reasonable estimates of fatigue life. To extrapolate the modulus-loading cycle data, regression analysis was
used to estimate modulus as a function of loading cycles. The form of extrapolation equation used was a
power law, with various rules to minimise prediction errors.
These extrapolated low strain results are plotted in Figure 2.6 together with the results of fatigue testing at
normal strain levels (Figure 2.5).
From this analysis Shen and Carpenter (2005) concluded there was a distinctly different fatigue behaviour
starting from about a fatigue life of 1.1 x 107 cycles or from Equation 2 a PV of 6.74E-9. It was concluded that
samples with fatigue life longer than 1.1 x 107 cycles are no longer following the same relation as seen at
normal strain-damage levels. Shen (2006) describes the process used to estimate this transition in fatigue
behaviour. The strain-loading data for each mix was regressed as two individual power law relationships: one
for normal strain/damage data, the other for the low strain/damage data. For each mix, the intersection of
these two relationships was considered to be the transition point.
Table 2.2 shows the transition points calculated using this process for the 19 IDOT03 mixes. Given the
fatigue lives at the transition point varied from 106 to 8 x 107, it is questionable whether the strains calculated
at 1.1 x 107 cycles provide a reliable method of estimating the ELS. Also the use of a loading limit of 1.1 x
107 cycles for all mixes appears to provide a non-conservative estimate of ELS as most lives in Table 2.2
exceed this value.
Figure 2.6: Combining normal fatigue testing data with extrapolated lives at low strain levels

Source: Shen and Carpenter (2005).

Austroads 2016 | page 8

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 2.2:

Transition strains and fatigue life for 19 IDOT 03 mixtures


Transition point

Mix ID

Mix description

Strain
(microstrain)

Fatigue life

Mix with conventional binders


3N70

PG64-22 limestone, 5.2% binder

100

5.23 x 107

3N70P

PG64-22 limestone, 5.6% binder (rich)

100

3.2 x 107

3N90T

PG64-22 limestone, 4.1% binder

101

8.43 x 107

3N90TP

PG64-22 limestone, 4.7% binder (rich)

100

1.98 x 107

5N90

PG64-22 dolomite, 4.9% binder 0.5% anti-strip

100

1.33 x 107

5N90P

PG64-22 dolomite, 5.5% binder 0.5% anti-strip (rich)

100

4.1 x 107

6N50

PG64-22 limestone, 5.0% binder 0.7% anti-strip

118

4.22 x 107

6N50P

PG64-22 limestone, 5.6% binder (rich)

123

1.47 x 107

8N70

PG64-22 limestone, 4.9% binder

98

1.71 x 107

8N70P

PG64-22 limestone, 5.2% binder (rich)

141

1.00 x 107

Mix with polymer modified binders


1N105

PG70-22 dolomite, 6.5% binder

159

1.53 x 107

1N105P

PG70-22 dolomite, 6.9% binder (rich)

159

1.53 x 107

1N80D

SMA, SBS PG76-28 dolomite, 8.0% binder

300

1.66 x 107

1N80DP

SMA, SBS PG76-28 dolomite, 8.5% binder (rich)

300

4.02 x 107

2N90

PG70-22 (SBS) dolomite, 5.9% binder

100

3.94 x 107

2N90P

PG70-22 (SBS) dolomite, 6.8% binder (rich)

98

2.88 x 107

3N90

PG76-22 (SBS) dolomite, 5.1% binder 0.5% anti-strip

314

1.00 x 106

3N90P

PG76-22 (SBS) dolomite, 5.4% binder 0.5% anti-strip (rich)

173

1.82 x 107

5N105

PG76-28 (SBS) limestone, 4.7% binder 0.1% lime anti-strip

350

1.10 x 107

Source: Adapted from Shen (2006).

2.5

Use of Plateau Value to Estimate Endurance Strain Limit

The unique relationship between PV and fatigue life (Equation 2, Figure 2.7) calculated from fatigue testing
at normal strain level (Section 2.2) was found (Shen & Carpenter 2005) not to be statistically different from
the PV-Nf relationship calculated from the low strain data. (Shen 2006 describes the processes used to
extrapolate the PV and Nf values at low strain levels).
Based on their assumption that the transition or break point in fatigue behaviour occurs at 1.1 x 107 cycles
for all mixes (Section 2.4), Shen and Carpenter (2005) proposed the use of a PV of 6.74E-9 as a unique
indicator below which a mix will have an extraordinarily long fatigue life. However, as discussed in
Section 2.4, the use of 1.1 x 107 cycles of loading is not a conservative value of ELS for all mixes.

Austroads 2016 | page 9

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 2.7: PV life relationship

Source: Carpenter and Shen (2009).

2.6

Laboratory ELS Relationships

2.6.1 Prediction of ELS from Mix Modulus


Thompson and Carpenter (2006) predicted the ELS of 10 mixes listed in Table 2.3. These ELS were
estimated from beam fatigue testing at a temperature of 20 C and using Equation 5. Thompson and
Carpenter plotted these ELS against the initial flexural modulus after 50 loading cycles of fatigue as shown in
Figure 2.8.
This data and the prediction relationship shown in Figure 2.8 was later extended by Shen and Carpenter
(2007) using data from a wider range of mixes as discussed in Section 2.6.2.
As discussed in Section 5.3.2, Sullivan et al. (2015) used the results in Figure 2.8 together with results of
testing Australian mixes to quantify the variation of ELS with modulus.

Austroads 2016 | page 10

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 2.3:

Mix composition and predicted ELS

Mix ID

Strains at endurance limit


(microstrain)

Mix description

Mix with conventional binders


3N70

PG64-22 limestone, 5.2% binder

134

3N90T

PG64-22 limestone, 4.1% binder

136

5N90

PG64-22 dolomite, 4.9% binder 0.5% anti-strip

162

6N50

PG64-22 limestone, 5.0% binder 0.7% anti-strip

168

8N70

PG64-22 limestone, 4.9% binder

167

Mix with polymer modified binders


1N105

PG70-22 (SBS) dolomite, 6.5% binder

139

1N80D

SMA, PG76-28 (SBS) dolomite, 8.0% binder

273

2N90

PG70-22 (SBS) dolomite, 5.9% binder

170

5N105

PG76-28 (SBS) limestone, 4.7% binder 0.1% lime anti-strip

387

Source: Adapted from Thompson and Carpenter (2006).


Figure 2.8: Relationship between ELS and flexural modulus from beam fatigue testing
400

Conventional mixes

350

PMB mixes
300

Endurance
limit
250
strain
(microstrain)
200

150

100
1000

2000

3000

4000

5000

6000

7000

Flexural modulus (MPa)

Source: Adapted from Thompson and Carpenter (2006).

Austroads 2016 | page 11

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

2.6.2 ELS Prediction from Mix Modulus and Volumetrics


To determine a strain-based fatigue endurance limit without use of measured fatigue test results, Shen and
Carpenter (2007) rearranged Equation 3 to predict strain at a PV of 6.74E-9, the value considered to be
associated with the endurance limit (Section 2.5). Equation 5 is the resulting ELS prediction equation.

= 0.0123E0.5832 VP0.3599 GP0.0790

where

endurance strain at PV = 6.74E-9 (mm/mm)

flexural modulus of mix from the laboratory fatigue test (MPa)

VP

volumetric parameter VP =

GP

AV

AV+Vb

where AV = air voids of mix (%)


=

Vb = volume of binder (%)

gradation parameter GP =

PNMS PPCS
P200

where PNMS = percentage passing maximum nominal size (%)


PPCS = percentage passing the primary control sieve (%)
P200 = percentage passing No. 200 (75 m) size (%)

Note that Equation 5 required extrapolation of Equation 3 as the PV of 6.74E-9 is well outside the range of
measured data as illustrated in Figure 2.3.
Using the results of low strain fatigue testing of 19 mixes, Shen and Carpenter (2007) compared the ELS
predicted using Equation 5 with the values roughly estimated by extrapolating the fatigue results (Table 2.2).
The results are listed in Table 2.4 and illustrated in Figure 2.9.
It seems that there are significant differences between the strains predicted from Equation 5 and those
estimated from the fatigue testing. Shen and Carpenter (2007) attributed this lack of agreement to the limited
sample testing data available which produced low precision in the strains estimated from the fatigue testing.
It is noted that for mixes made with conventional binders, Equation 5 appears to over-estimate the ELS.

Austroads 2016 | page 12

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 2.4:

Comparison of predicted ELS with those extrapolated from the low strain fatigue testing
Strains at endurance limit
(microstrain)

Mix ID

Mix description

Predicted using
Equation 5

Estimated from
fatigue testing
(Table 2.2)

Mix with conventional binders


3N70

PG64-22 limestone, 5.2% binder

142

100

3N70P

PG64-22 limestone, 5.6% binder (rich)

152

100

3N90T

PG64-22 limestone, 4.1% binder

133

100

3N90TP

PG64-22 limestone, 4.7% binder (rich)

150

101

5N90

PG64-22 dolomite, 4.9% binder 0.5% anti-strip

160

100

5N90NP

PG64-22 dolomite, 5.5% binder 0.5% anti-strip (rich)

145

100

6N50

PG64-22 limestone, 5.0% binder 0.7% anti-strip

170

118

6N50P

PG64-22 limestone, 5.6% binder (rich)

162

123

8N70

PG64-22 limestone, 4.9% binder

166

98

8N70P

PG64-22 limestone, 5.2% binder (rich)

167

141

Mix with polymer modified binders


1N105

PG70-22 (SBS) dolomite, 6.5% binder

142

159

1N105P

PG70-22 (SBS) dolomite, 6.9% binder (rich)

161

159

1N80D

SMA, PG76-28 (SBS) dolomite, 8.0% binder

263

300

1N80DP

SMA, PG76-28 (SBS) dolomite, 8.5% binder (rich)

266

300

2N90

PG70-22 (SBS) dolomite, 5.9% binder

161

100

2N90P

PG70-22 (SBS) dolomite, 6.8% binder (rich)

169

100

3N90

PG76-22 (SBS) dolomite, 5.1% binder 0.5% anti-strip

184

314

3N90P

PG76-22 (SBS) dolomite, 5.4% binder 0.5% anti-strip (rich)

173

173

5N105

PG76-28 (SBS) limestone, 4.7% binder 0.1% lime anti-strip

274

350

Source: Shen and Carpenter (2007).

Austroads 2016 | page 13

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 2.9: Comparison of predicted ELS with those extrapolated from fatigue testing

Mixes with conventional binder PG64-22

180
170
160

Line of equality
150

Endurance
limit
140
strain
from
laboratory 130
fatigue testing
(microstrain)
120
110
100
90
90

100

110

120

130

140

150

160

170

180

Predicted endurance limit strain (microstrain)

Mixes with SBS polymer modified binders

350

300

Line of equality
250

Endurance
limit
strain
200
from laboratory
fatigue testing
(microstrain)
150

100

50
50

100

150

200

250

300

350

Predicted endurance limit strain (microstrain)

Austroads 2016 | page 14

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

In this study by Shen and Carpenter (2007), the most significant mix attribute causing changes to ELS was
asphalt modulus. To illustrate the impact of modulus on the ELS, ELS were calculated for the range of
moduli investigated (150010000 MPa as shown in Table 2.1), using a gradation parameter (GP) of 13.26
(Table 2.1) and for three values of the volumetric parameter (VP):

VP = 0.40: 7% air voids and volume of binder of 10.5%


VP = 0.30: 5% air voids and volume of binder of 11.5%
VP = 0.19: 3% air voids and volume of binder of 12.5%.
Figure 2.10 illustrates the high predicted dependency of ELS with modulus and to a lesser extent mix
volumetrics. Importantly the model predicts that for a given modulus, mix volumetrics need to be considered
to estimate the ELS. This is discussed further in options for the Guide in Section 7.
Note that all the test results used in the model development were obtained at a single test temperature of
20 C and test frequency of 10 Hz. The variation in mix moduli were due to variations in binder type, mix
volumetrics, and to a less extent aggregate type and grading. It is not known whether the same dependence
on modulus would have been determined if the modulus variation was due to temperature changes rather
than binder type. This model limitation is discussed further in evaluating the options in Section 7.2.
Figure 2.10:

Predicted variation in ELS with modulus and mix volumetrics


380
360
340

Air voids = 3%, Vb=12.5%

320

Air voids=5%, Vb=11.5%

300

Air voids = 7%, Vb=10.5%

280
260
240
Strain
(microstrain) 220
200
180
160
140
120
100
80
1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500 10000
Asphalt modulus (MPa)

Austroads 2016 | page 15

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

2.7

Effect of Rest Periods on ELS

Carpenter and Shen (2006) also undertook flexural fatigue testing with rest periods of up to nine seconds
between haversine load pulses. Two asphalt mixes were tested at a temperature of 20 C: one with a
conventional binder (PG64-22), the other a polymer modified binder (PG70-22). From the testing the number
of loading cycles to 50% reduction on initial modulus (Nf) were calculated and the associated PV at Nf.
Figure 2.11 shows PV reduced with increasing rest period presumably due to the increase in healing.
Plotting the results of both mixes on the PV-Nf graph as shown in Figure 2.12, the data points appear to
follow a unique PV-Nf line, statistically no different from that previously determined (Figure 2.7) without rest
periods. The Carpenter and Shen (2006) considered this similarity provided further substantiation that the
PV-Nf relationship (Figure 2.7) is unique.
Figure 2.11:

Variation of PV with rest period at a test temperature of 20 C

Source: Carpenter and Shen (2006).


Figure 2.12:

PV-Nf relationship for various rest periods

Source: Carpenter and Shen (2006).

Austroads 2016 | page 16

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

This testing indicated Equation 6 may be used to quantify the variation in PV with rest period.

PVh = PVw/o (RP +1)n

where

PVh

plateau value with healing

PVw/o

RP

plateau value without healing, RP = 0

slope from Figure 2.11, for PG64-22 = 0.9069, for PG70-22 = 1.35

rest period, seconds

To illustrate the impact of rest period on the ELS in this report, a PV prediction equation was calculated from
the general equation (Equation 3) for example test conditions (E = 5890 MPa, air voids = 5.2%, volume of
binder 12%, GP = 13.26, type PG 64-22 binder) and including allowance made for rest periods using
Equation 6. Equation 7 is the resulting relationship.

PVh = 3.222 x 1011 5.14 (RP +1)-0.9069

where

PVh

plateau value with healing

tensile strain (mm/mm)

RP

rest period, seconds

Shen and Carpenter (2007) proposed the use of a PV of 6.74E-9 to calculate the ELS. Using Equation 7, for
a PVh value of 6.74E-9 the applied haversine strains were calculated for various rest periods. Figure 2.13
show the extent to which healing increases ELS at a temperature of 20 C.
Figure 2.13:

Example of the influence of rest period on ELS at a temperature of 20 C

240

220

200

180

ELS
(microstrain)

160

140

120

100
0

Rest period (seconds)

Austroads 2016 | page 17

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

2.8

Summary

An extensive asphalt fatigue research program has been undertaken by researchers at the University of
Illinois. This research has concluded that the use of an energy parameter (RDEC) results in a unique fatigue
relationship that is able to explain variation in fatigue characteristics between mixes, loading times and rest
periods. The researchers proposed the use of the RDEC calculated at the number of cycles to half initial
modulus, the plateau value (PV), to characterise fatigue performance.
Fatigue results at low strains have been extrapolated and compared against the trend of fatigue testing at
normal strain levels. From testing a range of mixes, the researchers concluded that after about 1.1 x 107
cycles of loading or PV of 6.74E-9, the fatigue behaviour changes markedly.
It was proposed that the fatigue endurance limit be the strain that results in a PV of 6.74E-9, at which the
fatigue life is 1.1 x 107 cycles of loading. However, the data reported casts doubt on the reliability of this
loading level for all mixes.
The research investigated the extent to which fatigue life varies with asphalt mix modulus. The modulus
dependency was substantially higher than in the fatigue relationship in the Austroads Guide. In addition, the
ELS dependence on modulus was substantially higher than recommended in another recent US research
project (NCHRP 9-44A as described in Section 3.8). The Illinois data modulus variations were due to
changes in mix composition and binder stiffness at a single test temperature (20 C) and loading frequency
(10 Hz). In contrast in the NCHRP 9-44A project moduli variations resulted from testing a limited range of
mixes over three temperatures. Differences in the causes of moduli changes maybe have contributed to the
difference in modulus dependency.
While modulus was the most important factor influencing variation in fatigue life and ELS between mixes, it
was concluded that air voids and volume of binder also need to be considered.
Note that the tensile strains reported in this research are those under haversine loading. As discussed in
Section 2.1, the reported haversine strains are twice the actual tensile strains being applied. The research in
this section of the report is discussed without this strain correction. However, when the results are compared
to those of NCHRP 9-44A in Section 3, the strain values need to be halved.

Austroads 2016 | page 18

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3. NCHRP Project 9-44A Laboratory Endurance


Limits
3.1

Introduction

In 2009 after consideration of the initial Austroads fatigue endurance limit report (Austroads 2009), it was
decided to wait for the findings of a forthcoming US research project, National Cooperative Highway
Research Program (NCHRP) project 9-44A, Validating an Endurance Limit for HMA Pavements: Laboratory
Experiment and Algorithm Development.
A previous project, NCHRP Project 9-38, Endurance Limit of Hot Mix Asphalt Mixtures to Prevent Fatigue
Cracking in Flexible Pavements, was completed in 2009. The report concluded that this project had
confirmed the existence of a hot-mixed asphalt (HMA) fatigue endurance limit through an extensive program
of laboratory testing and that endurance limits are influenced by HMA mixture and binder properties.
As stated in the final project report for NCHRP 9-44A (Witczak et al. 2013):
NCHRP Project 9-44A was designed to extend the results and findings of Project 9-38, with
particular attention to the influence of asphalt binder and mixture properties on the
endurance limit and to the relationship of the endurance limit to the phenomenon of healing
hypothesized to occur in asphalt mixtures during the rest period between load applications in
the laboratory and in pavements. The specific objectives of the project were to:
1 carry out a laboratory experiment to identify the mixture and pavement layer design
features related to an endurance limit for bottom-initiated fatigue cracking of HMA
2 develop an algorithm to incorporate this endurance limit into the Pavement ME Design
software and other selected pavement design methods
The research was performed by Arizona State University, Tempe, Arizona, in association with AMEC
(formerly MACTEC), Phoenix, Arizona. The project was completed in September 2013.
The research investigated the relationship of the ELS to factors such as asphalt binder rheology, air voids,
asphalt content, temperature, strain level, number of load cycles, and rest period between load cycles. Both
beam fatigue (AASHTO 2003) and uniaxial compression-tension testing were conducted. The project report
(Witczak et al. 2013) recommends that the beam fatigue model be used for future studies of endurance limit
and its implementation. The beam fatigue test is better established than the uniaxial test and has a larger
database of results in the literature. Hence this review is limited to the beam fatigue testing results.

3.2

Overview of Methodology

In the NCHRP 944A project, the ELS was defined as the strain level where the cumulative effect of a
loading pulse and a rest period results in no discernible reduction in modulus, e.g. where the damage
occurring during the loading time and the healing occurring during the rest periods are balanced. Fatigue
damage and healing are tracked using the stiffness ratio (SR), which is simply the modulus value at any
given load cycle divided by modulus at the 50th cycle of loading in the same test.

Austroads 2016 | page 19

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Witczak et al. (2013) describe the concept used to relate the asphalt healing phenomenon to the endurance
limit:
If the fatigue test is conducted with and without rest period between load applications, the typical
stiffness ratio (SR) (ratio between the current stiffness and the initial stiffness) versus the number
of load cycles will be as shown in Figure 3.1. Both curves start at an SR of one (no damage). The
curve for the test without a rest period is steeper than the other curve because of the continuous
deterioration during the test. The test with a rest period shows higher SR during the test because
of healing that occurs during the rest period after each load application. A larger separation
between the two curves indicates more healing, and vice versa. If the curve for the test with a rest
period remains horizontal, it indicates that full healing occurs after each load cycle.
Figure 3.1: Typical SR versus number of load cycles for tests with and without a rest period

Source: Witczak et al. (2013).

Accordingly, to estimate the ELS from the fatigue testing the strain that results in an SR of 1 needs to be
determined. As illustrated in Figure 3.2, the approach used was firstly to calculate SR values from the testing
after various cycles of loading. Regression analysis was then used to determine an equation to predict SR
from test parameters such as applied strain, rest period and mix properties such as flexural modulus, binder
content and air voids. The ELS for a given mix and rest period was then determined as the strain at which
SR = 1. When SR = 1 despite continued loading, the healing and damage are said to be balanced and the
given strain level is at or below the ELS for the particular rest period applied.
Figure 3.2: Schematic demonstration of NCHRP 9-44A methodology

Source: Zeiada, Underwood and Kaloush (2016).

Austroads 2016 | page 20

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.3

Materials Tested

A size 19 mm Superpave dense graded asphalt mix design was selected for the project that met the
requirements of typical mixtures used for paving arterial roads in Arizona. Three asphalt mixes using three
binder grades (PG 58-28, PG 64-22, and PG 76-16) were prepared to ensure that a wide range of moduli
would be encountered. The same aggregate gradation was used for all three mixtures.
Beam fatigue tests (Figure 3.3) were performed on hot mixed asphalt (HMA) mixes prepared with all three
binder grades (PG 58-28, PG 64-22, and PG 76-16). Using the dynamic shear rheometer (DSR) results
reported by Witczak et al. (2013), the viscosities of the neat binders at a temperature of 60 C were
calculated for comparison with comparable results for Australian bitumen classes. The values calculated
were:

PG 58-28: 122 Pa.s, which is slightly lower than Class 170 bitumen
PG 64-22: 209 Pa.s, which is between the values for Class 170 and Class 320 bitumen
PG 76-16: 672 Pa.s, which is similar to the values of Class 600 bitumen.
Accordingly, the results for PG64-22 and PG76-16 are relevant to Australian asphalt mixes.
For each binder grade, four mix compositions were tested with the following binder contents (by mass) and
air voids (by volume):

4.2% bitumen, 4.5% air voids


4.2% bitumen, 9.5% air voids
5.2% bitumen, 4.5% air voids
5.2% bitumen, 9.5% air voids.
The asphalt mixes were subjected to short-term ageing after mixing and prior to compaction as described by
Witczak et al. (2013):
The properly mixed HMA was then emptied into a heated metal tray, approximately
50 mm x 50 mm and 75 mm deep in size, evenly spread about 1 inch thick, and placed
uncovered into a preheated 135 C convection oven for short-term aging per the AASHTO R
30 procedure for mixture performance testing. The HMA was left uncovered in the oven for a
1 hour period, and then the door opened and the HMA hand mixed and turned over multiple
times within the tray with a heated spoon for 1520 seconds. The door was then shut and
the HMA was left to age another hour. After the second hour, the hot, aged mix was mixed
with the heated spoon again and mix sufficient to compact a specimen to a pre-determined
percentage air voids was immediately scooped into the beam mould, with the HMA placed in
the mould in two equal weighing lifts. Once the mould was filled, it was returned to an oven
for about 15 minutes to achieve the proper temperature for compaction.

Austroads 2016 | page 21

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.4

Beam Fatigue Testing

Beam flexural fatigue tests under controlled strain were performed according to the AASHTO test method
T 321/03 (AASHTO 2003). The testing were undertaken using sinusoidal loading to avoid the complications
of beam deformation which occur under haversine loading (Witczak et. al. 2013).
Figure 3.3: Beam fatigue testing equipment

Source: Witczak et al. (2013).

3.5

Test Program

Samples were tested at different asphalt and air void contents and at different temperatures and rest period
durations in a partial factorial manner. A total of 468 beams were tested for fatigue characteristics. The test
beams covered a range of properties, namely:

binder types (3 levels, PG 58-28, PG 64-22, PG76-16)


binder contents (2 levels, 4.2% and 5.2% by mass)
air voids (2 levels, 4.5% and 9.5% by volume)
strain levels (3 levels, low, medium and high)
temperature (3 levels 40 F, 70 F and 100 F, that is 4.4 C, 21.1 C and 37.8 C, respectively)
rest period between loading cycles (4 levels: 0, 1, 5 and 10 seconds).
Table 3.1 summarises the factors investigated.

Austroads 2016 | page 22

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 3.1:

Factor combinations used in beam and uniaxial fatigue tests

Binder grade

PG 7616

Binder content (%)

4.2

Air voids (%)

4.5

Temperature
(F)

Strain
level

40

Low

9.5

PG6422
5.2

4.5

9.5

4.2

PG 5828
5.2

4.5

9.5

4.5

4.2

9.5

4.5

9.5

5.2
4.5

9.5

Rest
period
(seconds)
0

1
5

10
Medium

10
High

1
5

X
X

10
70

Low

1
5
10
Medium

1
5

X
X
X

10
High

0
1

5
10
100

Low

X
X

10
Medium

0
1
5

High

X
X

10

1
5

10
Beam Fatigue Test
X

Uniaxial Fatigue Test

Source: Witczak et al. (2013).

Austroads 2016 | page 23

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.6

Data Used to Develop the Stiffness Ratio Prediction Method

As described in Section 3.2, the ELS values extrapolated from the fatigue testing are the strain levels
associated with a stiffness ratio (SR) of 1. To enable this extrapolation, SR data needed to be calculated
from the beam fatigue testing. The following data were calculated:

For beams tested without a rest period, the beams were tested until they cracked. From this testing, the
strains associated with a SR value of 0.5 were determined.

For the beams tested with rest period typically the number of loading cycles was limited to a maximum of
20 000 cycles to limit the duration of testing. For these tests three values of SR were calculated

after 10 000 cycles


after 20 000 cycles
SR extrapolated at the number of cycles corresponding to the fatigue life without a rest period.
Figure 3.4 illustrates these values of SR, which totalled 946 data points used to develop the SR prediction
equation.
Figure 3.4: Data points used to develop the SR prediction equation

Source: Witczak et al. (2013).

3.7

Relationship to Predict Stiffness Ratio from Modulus

As mentioned above, the laboratory testing program included variations in test temperature, binder type and
mix volumetrics. In an effort to develop a satisfactory relationship between the SR and the material and
testing conditions, several regression models were attempted. Originally, a regression model was developed
to relate the SR to all factors used in the study, which were binder content, air voids, binder grade,
temperature, applied tensile strain, rest period, and number of loading cycles. (This was developed further at
a later stage as discussed in Section 3.9). Then the model was simplified using the initial flexural modulus as
a surrogate for the binder content, air voids, binder grade, and temperature, as all these parameters affect
modulus. According to Witczak et al. (2013), there were two significant advantages in the model: 200410

It is simplified.
It is more compatible with the pavement design software being used with the mechanistic empirical
pavement design guide (MEPDG) used in the USA (AASHTO 2015) in which modulus is a major input to
pavement performance prediction.

Austroads 2016 | page 24

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Equation 8 is the stiffness ratio-modulus relationship obtained by regression analysis. Figure 3.5 illustrates
the degree to which Equation 8 predicts the measured stiffness ratios.

where

SR

stiffness ratio

E0

flexural modulus after 50 cycles of loading (MPa)

tensile strain (microstrain)

number of loading cycles

RP

rest period between loading cycles (seconds)

Figure 3.5: Measured versus predicted stiffness ratios

Source: Witczak et al. (2013).

Austroads 2016 | page 25

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.8

Laboratory Beam Endurance Limit Strains

Using Equation 8, the laboratory ELS can be estimated as the strain value when SR = 1 for a given modulus
and rest period. Witczak et al. (2013) recommended the ELS be calculated at N = 200 000 cycles. Figure 3.6
shows the ELS relationships so calculated.
Figure 3.6: Laboratory endurance limit strains calculated using the modulus model
170

Rest period 1 second

160

Rest period 2 seconds

150

Rest period 3 seconds

140

Rest period 5 seconds

130

Rest period 10 seconds

120
110

Endurance
limit strain 100
(microstrain)

y = 953.46x-0.247

90

y = 836.32x-0.253

80

y = 724.18x-0.262

70

y = 695.03x-0.273

60
50
40
30
1000

y = 571.85x-0.323
2000

3000

4000

5000

6000

7000

Flexural modulus (MPa)

It was of interest to investigate the extent to which endurance limits vary with temperature and mix
composition as predicted by the ELSmodulus relationships shown in Figure 3.6.
Using the data in Appendix 1 of Witczak et al. (2013), the mean initial (50 cycles) flexural modulus values
were calculated for the PG64-22 mixes as listed in Table 3.2. The ELS were then calculated using the
equation in Figure 3.6 for a rest period of 5 seconds. The ELS are plotted in Figure 3.7.
Table 3.2:

Predicted ELS for the PG64-22 mixes

Temperature
(C)

Bitumen content
(%)

Air voids
(%)

Flexural modulus
at 10 Hz after 50 loading
cycles
(MPa)

Calculated endurance
strains
5 seconds rest period
(microstrain)

4.4

4.2

4.5

13 300

74.9

4.2

9.5

11 200

78.4

5.2

4.5

12 160

76.7

5.2

9.5

10 000

80.9

4.2

4.5

5 590

94.4

4.2

9.5

3 940

103.4

5.2

4.5

5 090

96.7

5.2

9.5

3 780

104.5

4.2

4.5

1 200

138.9

4.2

9.5

901

149.1

5.2

4.5

838

151.6

5.2

9.5

723

157

21.1

37.8

Austroads 2016 | page 26

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 3.7: Predicted effect of temperature, bitumen and air voids from the ELS-modulus relationship
160

4.2% bitumen, 4.5% voids


150

5.2% bitumen, 4.5% voids


4.2% bitumen, 9.5% voids

140

5.2% bitumen, 9.5% voids


130

Endurance 120
limit strain
(microstrain) 110
100
90
80
70
4

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

Temperature (C)

As expected, the ELS increased with increasing temperature due to the modulus reducing with increasing
temperature. The calculated effect of the air voids increase from 4.5% to 9.5% is counter-intuitive as the ELS
was higher for the 9.5% air voids mixes. This is clearly inconsistent with the influence of air voids on fatigue
life. The predicted effect results from the modulus decreasing with increasing air voids.
As expected the ELS is predicted to increase with bitumen content with the effect increasing with
temperature, however the effect was marginal at temperatures of 4.4 C and 21.1 C.
The influence of bitumen content and air voids are different from:

those predicted using the PV results (Section 2.6) and the alternative analysis of beam results
(Section 3.9.2) which indicated that bitumen content had a very significant effect across a range of
temperatures

the alternative analysis of beam results (Section 3.9.2) which indicated the ELS increased with
decreasing air voids.
As seen from Figure 3.8 and Table 3.2, the major factor affecting the initial flexural modulus data was the
test temperature. This may have led to the effect of mix composition on ELS not being as accurately
quantified as the alternative method discussed in Section 3.9.2. In contrast, the variation of ELS with
modulus estimated from PV results (Section 2.6) was due to mix composition differences (e.g. binder type)
rather than changes in test temperature. The different sources of the moduli variation may have been a
factor in the difference in the observed modulus dependencies.

Austroads 2016 | page 27

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 3.8: Initial flexural modulus variation with temperature and mix composition
14000
13000
12000
PG58-28

11000

PG64-22

10000

Initial
flexural
modulus
at 10 Hz
(MPa)

PG76-16

9000
8000
7000
6000
5000

PG76-16
y = 16386e-0.066x

4000
3000

PG 64-22
y = 18191e-0.077x

2000
PG58-28
y = 14068e-0.086x

1000
0
4

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

Temperature (C)

It was concluded that the use of modulus alone to consider the combined effect of mix composition and
temperature, although simplifying the method, may result in some misleading effects that would need to be
addressed. Alternatively, the model could be used with limits on mix air voids and bitumen content.
The limitations of the use of modulus as a surrogate for binder rheology, air voids, bitumen content, and
temperature was acknowledged by Witczak et al. (2013):
It must be recognized that simplifying the models by replacing the binder grade, binder
content, air voids, and temperature with the stiffness of the material may produce a certain
level of inaccuracy. For example, mixtures with high binder content and low air voids showed
similar stiffness to those that have low binder content and high air voids, even though their
endurance limits are different.
This issue is discussed further in Section 7, in which implementation options for the Guide are presented.

3.9

Further Analysis of Laboratory Beam Results

Although Witczak et al. (2013) recommended the ELS-modulus relationship described in Section 3.8, they
also developed a relationship to predict SR from the test temperature and the mix volumetrics as detailed in
Appendix 2 of Witczak et al. (2013). The reanalysis was limited to mixes with PG 64-22 bitumen.

Austroads 2016 | page 28

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.9.1 Stiffness Ratio Prediction Relationship


Using a total of 231 data points, the SR prediction relationship shown in Equation 9 was obtained (Appendix 2,
Witczak et al. 2013). Figure 3.9 illustrates the degree to which Equation 9 explains the measured SR
variation.
9

where

SR

stiffness ratio

temperature (F)

AC

bitumen content (%)

Va

air voids content (%)

tensile strain (microstrain)

number of loading cycles

RP

rest period (seconds)

Figure 3.9: Comparison of measured and predicted stiffness ratios

Source: Witczak et al. (2013).

Austroads 2016 | page 29

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.9.2 Laboratory Beam Endurance Limits in Terms of Volumetrics and Temperature


Using N = 20 000 cycles and a rest period of 5 seconds in Equation 9, Witczak et al. (2013) calculated the
endurance limits shown in Figure 3.10 as the strain at which SR = 1. Note that these ELS are for asphalt
mixes made with the PG64-22 bitumen, which has a penetration between Australian Class 170 and Class
320 bitumens.
Figure 3.10:

Beam endurance limit strains for 5 seconds rest period

Source: Witczak et al. (2013).

As seen from Figure 3.11 the ELS predicted using Equation 9 is very sensitive to bitumen content, for
instance at 32 C for 0.5% increase in bitumen content there is a 14 microstrain increase of the ELS. The
sensitivity of ELS to air voids is considerably less (Figure 3.12). For instance at a temperature of 32 C, to
increase the ELS by 10 microstrain requires about a 3% reduction in air voids.

Austroads 2016 | page 30

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 3.11:

Effect of temperature and bitumen content


Beam data 5 seconds rest period, 5 % air voids

140

130

4.2% bitumen
4.7% bitumen
5.2% bitumen

120

110

Endurance
100
limit
(microstrain)
90

80

70

60
10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

Asphalt temperature (C)

Figure 3.12:

Effect of temperature and air voids content

Beam data 5 seconds rest period, 4.7% bitumen

140

130

4.5% voids
7.0% voids

120

9.5% voids

110

Endurance
100
limit
(microstrain)
90

80

70

60
10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

Asphalt temperature (C)

Austroads 2016 | page 31

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

3.10 Summary
The research under project NCHRP 9-44A investigated how laboratory beam fatigue damage varies with test
temperature, binder types, mix volumetrics and rest period between loading pulses. From the testing the ELS
were extrapolated.
From the relationships derived from the testing, the findings were that ELS increased with:

increasing temperature
increasing rest period up to about 5 seconds
increasing binder content
decreasing mix and binder stiffness
decreasing air voids.
Witczak et al. (2013) recommended a simple relationship (Equation 8) to predict ELS using the initial flexural
modulus as a surrogate for temperature, mix volumetrics and binder stiffness. As discussed in Section 7.3
(Option B), this laboratory ELS relationship was used to develop a possible in-service ELS for use in the
Guide. However, it needs to be acknowledged that this relationship has significant limitations in quantifying
the effects of binder content changes and is misleading in terms of the influence of air voids on ELS.
The major source of variation of the modulus data was due to temperature variations. This is a possible
reason the ELS varied with modulus (Figure 3.6) to a lesser extent than determined from the University of
Illinois research (Figure 2.10). In the University of Illinois research tests were undertaken at a single
temperature of 20 C. In this research the modulus variations were due to variation in asphalt mix
composition including the use of polymer modified binders. Hence a possible reason for the different
modulus dependencies was the different causes of the moduli variations.
Based on an analysis of the data from mixes made using the PG64-22 binder, a laboratory ELS relationship
was also developed (Equation 9) that predicts ELS from temperature and mix volumetrics rather than
modulus. As discussed in Section 7.4 (Option C), this laboratory ELS relationship was also used to develop a
possible in-service ELS for use in the Guide. A shortcoming of this relationship is that it relates to a single
binder (PG64-22) and single aggregate source.

Austroads 2016 | page 32

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

4. Field Studies
4.1

United Kingdom

Observations that thick asphalt pavements appeared to be performing better than predicted led to work in the
United Kingdom (UK) by Nunn and Smith (1997) which drew together information from full-scale
experimental pavements, studies of deterioration mechanisms on the road network, long-term deflection
monitoring of motorways and condition assessments with the aim of producing a design method for roads
expected to last at least 40 years without the need for structural maintenance, described as long-life roads.
This study found:

Pavements with less than about 180 mm of asphalt deform at a high rate but thicker pavements deform at
a rate about two orders of magnitude less; the sudden transition suggesting a threshold effect.

No correlation existed between the rate of rutting and pavement thickness for thick (180+ mm) asphalt
pavements.

The level of traffic loading was not the major factor affecting the residual fatigue life of the thick asphalt
pavements.

Deterioration of thick, well-constructed, fully-flexible pavements is not structural, and deterioration


generally occurs at the surface in the form of cracking and rutting.

Any evidence of fatigue cracking or damage in the main structural layers of the thicker, more heavily
trafficked pavements, was unable to be detected.

Deterioration, as either cracking or deformation, was far more likely to be found in the surfacing than deep
in the pavement structure.

The great majority of the thick pavements studied became stronger over time, rather than gradually
weakening with trafficking as assumed.
The research (Nunn & Smith 1997) included detailed investigation of four motorways, representing a range of
pavement age (11 to 23 years) and traffic loading (2.2 x 107 to 7.1 x 107 ESA). All the pavements examined
carried more traffic than they were originally designed to carry. Cores were cut to enable layer thickness and
structural properties of the materials to be determined. Cores cut from the wheel path of lane 1, subject to
heavy commercial traffic, were compared with cores cut from between the wheel paths in lane 3. The
laboratory-measured residual fatigue lives were then compared. Analysis of the data showed no consistent
difference between the measured residual fatigue lives with most of the difference being accounted for by
variations in binder hardness and binder content between the samples extracted from the two lanes. When
these factors were taken into account, none of the differences were statistically significant (Nunn & Ferne
2001).
Nunn and Ferne (2001) noted that all the materials tested had a residual fatigue life less than that of new
material and that traffic loading could not account for this reduction. The authors attributed this lower
laboratory residual fatigue life to aging of the materials. However, it was noted that relationships developed
as part of the study showed that the increase in elastic stiffness with age produced a reduction in the trafficinduced, tensile strain responsible for fatigue at the bottom of an asphalt pavement which more than
compensated for any reduction in the laboratory fatigue life and that the net effect was that the predicted
fatigue life increased with age (Nunn & Ferne 2001).

Austroads 2016 | page 33

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

The research (Nunn & Smith 1997, Nunn & Ferne 2001) included measurement of pavement deflection,
initially via a slow-moving standard wheel load and later by falling weight deflectometer (FWD) of the same
sites. The deflection histories of 10 heavily trafficked sections of motorway were examined to investigate
whether the strength of thick, fully flexible pavements reduced with time and traffic. The initial deflection data
studied showed considerable fluctuations which the authors attributed partly to the difficulty of applying
accurate temperature corrections, seasonal variations in the subgrade strength and variation in alignment of
successive surveys. The subsequent FWD data showed all the sections had a trend of unchanging or
decreasing deflection with age and traffic, with one exception that showed no decisive trend either way. The
authors noted that these decreases implied that the overall stiffness of the pavements was increasing over
time and that any traffic-related deterioration was more than offset by curing of the asphalt or strengthening
of the foundation.
Nunn (1997) concluded that for pavements constructed on firm foundation:
These conservative calculations suggest that a road constructed with a thickness of more
than 260 mm of asphalt material would have a long but indeterminate life for traffic flows up
to 5 x 106 per year and that 270 mm would be sufficient for any traffic flow. This thickness
would ensure long-life even if curing did not take place, provided that the effective thickness
was not reduced by deterioration due to cracking.
As a result of this research the UK procedure for design of asphalt pavements was revised in 1997 to include
a minimum asphalt thickness, corresponding to a minimum threshold pavement strength, for the most
common asphalt mixes beyond which the pavement should have a very long but indeterminate structural life.
Section 6.1 discusses the design procedure (Highways England 2006) in more detail.
An ELS corresponding to the observed long-life pavements was not determined by Nunn and Smith (1997)
who instead used the concept of threshold pavement strength (minimum pavement thickness) to explain the
long-life pavement concept.

4.2

Australia

Paul (1997) undertook a review of design and construction of deep strength asphalt pavement in Victoria
which included investigating 14 sites, constructed between 1971 and 1995, by collecting current traffic
volume data, measuring deflections using the falling weight deflectometer and undertaking pavement coring
to determine actual pavement layer thicknesses and asphalt resilient modulus for the top, middle and bottom
sections of the pavement. Paul calculated that the majority of the sites investigated would exceed or had
already exceeded the design life predicted by mechanistic analysis without further structural improvement.
From the data collected the author concluded that the performance of the majority of the deep strength
asphalt pavements studied was much better than predicted. However, this is not unexpected as the design
procedures are intentionally conservative such that there is high reliability of outlasting the design traffic
(Austroads 2012).
For 11 of the 14 sites investigated by Paul (1997), the top layers of asphalt had higher modulus than the
bottom layers, as determined by laboratory testing of core samples. The difference was quite significant on
some of the older pavements even though the lower layers were 20 mm mixes compared with 14 mm or
10 mm in the top layers. Paul noted that there could be a number of explanations for this, including what the
author considered the most obvious; that the top layers had undergone more age hardening. However, Paul
noted that another valid explanation could be that more micro-cracking had occurred in the lower layers,
which would support the traditional mechanism of bottom-up fatigue cracking. It should be noted that coring
of untrafficked areas to evaluate the effect of age hardening was not undertaken and therefore it is difficult to
draw any definitive conclusions from the modulus variation between the top and bottom layers.

Austroads 2016 | page 34

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

The AAPA Research and Development Committee (AAPA 2000, 2004) studied performance of heavy duty
flexible pavements in NSW, Queensland and Victoria by investigating 20 sites aged 3 to 28 years with
asphalt thickness primarily in the range 320 to 370 mm. The data collected included traffic data, inner and
outer wheel path deflections, rut depth, texture depth, roughness, and coring of the pavement to determine
layer thicknesses and asphalt modulus/cemented material UCS/subgrade CBR as appropriate. Analysis of
the data collected included calculating the percentage of design traffic already experienced by the pavement,
comparing the measured asphalt modulus (from coring) with the design modulus and determining binder
properties.
The data as part of these AAPA studies showed that all 20 sites were in a sound structural condition and met
functional requirements (roughness, rutting and macrotexture). Overall there did not appear to be any
definitive pattern of increased roughness with time and there was a weak correlation between rutting and
traffic. The data indicated that there was stiffening of the binder after construction but, from the data, no
definitive relationship between asphalt modulus and time was found. The research did not compare
pavement performance with design assumptions or investigate pavement cracking in detail. However, the
data did show that some pavements had performed satisfactorily for significantly longer than predicted.
Again this is not unexpected as the design procedures are intentionally conservative such that there is high
reliability of outlasting the design traffic (Austroads 2012).
Limited coring through observed surface cracking was also undertaken as part of these AAPA studies. It was
concluded that these cracks initiated at the pavement surface and progressed downwards, which supported
studies by Nunn and Smith (1997) and Sharp et al. (2000). However, no evidence was obtained on whether
micro-cracking due to fatigue occurred below the visual cracks. As discussed in Section 5, recently AAPA
commissioned further research to develop fatigue endurance limits.
Tsoumbanos (2006) undertook an investigation of performance of thick asphalt pavements in Melbourne and
the types of distress exhibited with particular focus on top-down cracking. The visual condition of the sites
investigated by Paul (1997) was assessed and four sites, where cracking and other distress was identified as
likely to be limited to the top 40 to 60 mm of asphalt, were subject to additional detailed field investigations.
Tsoumbanos found that most sites did not exhibit visible load-associated cracking, or exhibited forms of
distress such as cracking from environmental causes, which suggested performance consistent with long-life
asphalt pavements.
The four sites studied in detail ranged in age from 20 years to nearly 30 years and were investigated for
roughness, rutting, cracking, strength (deflection) and stiffness (curvature). Pavement coring was undertaken
to determine layer thicknesses and resilient modulus for the top, middle and lower asphalt layers. The results
indicated:

Roughness at all locations was increasing; however, the roughness level was below VicRoads
intervention criteria.

Rutting was generally between 3 and 10 mm, with two sites exhibiting constant average rutting depth and
two sites exhibiting an increase in average rutting depth of approximately 1 mm per year.

There was an increasing proportion of cracked pavement for all sites compared to the earlier study
(Paul 1997), though the rate of increase varied.

Very low average deflectograph deflections and curvature were obtained in all lanes and both wheel
paths.
Tsoumbanos concluded that three of the four sites selected for detailed investigations generally exhibited
behaviour expected of long-life pavements considering:

Asphalt thicknesses were 210 mm or more.


Cracking was mostly limited to the top 40 to 60 mm of the asphalt (Figure 4.1).
Deflection testing identified very strong pavement structures.
No structural interventions had occurred during the service lives to date.

Austroads 2016 | page 35

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

However, some cores from Bell Street did show deep structural cracking as shown in Figure 4.2, some other
cores crumbled during the coring process and some of the layer moduli were low (Table 4.1). So not all the
data reported by Tsoumbanos is consistent with these sites being long-life asphalt pavements.
Table 4.1:

Observations and assessment of Victorian sites investigated

Road name

Mean
asphalt
thickness
(mm)

Camp Road

210

McIntyre
Road

Bell Street

Plenty Road

275

255

210

Average Indirect tensile


modulus at 25 C
(MPa)

Asphalt core observations

Cracking in eight of nine cracked cores limited to


wearing course.
Other cores damaged during extraction or not
cracked.

Wearing course

5150

Intermediate course

1040

Base course

3760

All cracked cores cracked throughout entire core.


Much of the section investigated is built on the fill
and cracking may be a result of fill settlement.

Wearing course

3440

Intermediate course

3940

Base course

3630

Cracking in five of eight cracked cores limited to


wearing course.
Cracking in three cores extended through much of
intermediate course.
Other cores not cracked.

Wearing course

7620

Intermediate course

4200

Base course

6510

Cracking in four of seven cracked cores limited to


wearing course. Intermediate course crumbled in
three cores.
Other four cores not cracked.

Wearing course

6940

Intermediate course

4820

Base course

1970

Source: Adapted from Tsoumbanos (2006).

Figure 4.1: Example of top-down cracking

Figure 4.2: Core from Bell Street Preston showing


deep structural cracking

Source: VicRoads (2007).

Source: Tsoumbanos (2006).

Austroads 2016 | page 36

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

4.3

United States of America

4.3.1 Analysis of USA LTPP Data


Von Quintus (2006) conducted a review and analysis of field investigation data from Long Term Pavement
Performance (LTPP) asphalt test sections across North America in 1995 and again in 2006 to confirm or
reject the concept of an asphalt fatigue endurance limit. Though not formally reported, the 1995 study utilised
randomly selected LTPP data on field observations of cracking, to conduct survivability, probability of failure,
analysis for full depth asphalt pavements more than 10 years old. The survivability analysis was completed
to estimate a value for the fatigue endurance limit based on fatigue cracking observations, rather than just
using values estimated from limited laboratory test programs. The definition of failure for confirming the
endurance limit premise was nominally no fatigue cracking, which in practice was taken as 2% fatigue
cracking to account for recording errors, and linear elastic modelling was used to calculate the maximum
tensile strain at the bottom of the asphalt layer for each test section. A typical survival curve, based on
amount of fatigue cracking and probability of occurrence, is shown as Figure 4.3.
Figure 4.3: Survival curve

Source: Von Quintus (2006).

In assessing the studied data in 1995 Von Quintus (2006) concluded that an ELS appeared to exist and was
determined to be 65 microstrain at a 95% confidence level. However in 2006, using an expanded dataset
which included the original pavement sections, additional pavement sections from the same LTPP datasets
and additional more recent LTPP sections, Von Quintus found that the results did not generate the same
trends and did not support the concept of a fatigue endurance limit. The pavement structures (asphalt type,
thickness and support) were not reported by Von Quintus but all included greater than 250 mm of asphalt
and, for the updated study, half of the pavement sections were greater than 15 years but less than 20 years
of age. However, Von Quintus noted that the volume of heavy vehicles for the sections used in the updated
study was lower than would be considered heavy truck traffic and that much higher levels of heavy vehicle
traffic were needed to validate the endurance limit design premise with field observations and data.
While Von Quintus (2006) commented that the original survival curve and test results from 1995 suggested
that the asphalt fatigue endurance limit premise had some validity and conversely the survival analysis
completed with the updated LTPP performance data suggested just the opposite, he concluded that no
definite conclusion could be reached without a forensic investigation.

Austroads 2016 | page 37

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

4.3.2 Analysis of NCAT Accelerated Loading Data


Willis et al. (2009), Willis and Timm (2009, 2010) and Tran et al. (2015) analysed the results of accelerated
trafficking of test sections at the National Center for Asphalt Technology (NCAT) at Auburn in United States
of America. One of the objectives of the analysis was to develop a field-based fatigue response threshold for
long-life asphalt pavements.
Figure 4.4 and Figure 4.5 illustrate the average as-built thicknesses and the constituent materials in each
section used in the development of the field-based fatigue response.
Figure 4.4: NCAT pavement cross-sections at commencement of Phase II loading in 2003

Source: Timm et al. (2006).

Austroads 2016 | page 38

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 4.5: NCAT pavement cross-sections at commencement of Phase III loading in 2006

Source: Willis et al. (2009).

In relation to the 2003 test sections given in Figure 4.5:

Willis et al. (2009) reported that of the seven sections analysed five experienced fatigue cracking during
Phase II after less than 107 ESA of loading: namely Sections N1, N2, N5, N6 and N7. These pavements
were clearly not long-life in terms of asphalt fatigue cracking.

Sections N3 and N4 were constructed in 2003 and were subjected to 107 ESA of loading in each of the
following loading stages: 20032005 (Phase II), 20062008 (Phase III and
20092011 (Phase IV). In 2009 after a total of 2 x107 ESA of loading no cracking was observed. In 2012
after a total of 3 x107 ESA of loading some minor cracking was observed but from pavement coring this
was considered to be top-down cracking rather than bottom-up classical fatigue cracking.
In relation to the test sections constructed in 2006 (Figure 4.5):

According to Willis and Timm (2010), three of the six analysed sections fatigue cracked in Phase III after
less than 107 ESA of loading: namely N8, N10 and S11.

The trafficking of Sections N1 and N2 was terminated early before 107 ESA of loading because of
observed cracking. From forensic investigations, Wills and Timm concluded this was top-down cracking
rather than bottom-up fatigue cracking.

Section N9 which included more than 350 mm of asphalt, not surprisingly had not fatigue cracked at the
end of Phase III loading after 107 ESA of loading. This section was trafficked further in Phase IV and after
2 x 107 ESA of loading fatigue cracking was not observed.
Using the periodically measured asphalt tensile strains, Willis et al. (2009) calculated the cumulative strain
distributions experienced by each pavement section to either the commencement of surface cracking or the
end of loading, whichever occurred first. Figure 4.6 shows the distributions.

Austroads 2016 | page 39

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Willis et al. (2009) considered there were five sections that did not experience bottom-up fatigue cracking,
namely:

Sections N3 and N4 after a total of 3 x107 ESA of loading


Sections N1 (2006) and N2 (2006) after less than 107 ESA of loading (loading was terminated early due
to surface cracks)

Section N9 after a total of 2 x107 ESA of loading.


Despite the applied traffic loading being less than 1/10th that of design traffic of heavily trafficked roads over
40 years (about 4 x 108 ESA), Willis et al. (2009) considered these five sections long-life asphalt pavements.
Using this classification they concluded from Figure 4.6 and Table 4.2 that there was a clear distinction
between the strain distributions of pavement susceptible to fatigue cracking and long-life pavements which
are not. Willis et al. concluded that the strain distributions associated with Sections N3 and N4 were suitable
criteria for the design of long-life pavements: this was designated the limiting strain distribution.
Figure 4.6: Cumulative strain distributions from the three test cycles

Source: Willis and Timm (2009).


Table 4.2:

Summary of section performance

Section

Performance

Section

Performance

S13 2000

Did not crack

N1 2006

Top-down cracking

N1 2003

Fatigue cracking

N2 2006

Top-down cracking

N2 2003

Fatigue cracking

N3 2006

Did not crack

N3 2003

Did not crack

N4 2006

Did not crack

N4 2003

Did not crack

N8 2006

Fatigue cracking

N5 2003

Fatigue cracking

N9 2006

Did not crack

N6 2003

Fatigue cracking

N10 2006

Fatigue cracking

N7 2003

Fatigue cracking

S11 2006

Fatigue cracking

Source: Willis and Timm (2009).

Austroads 2016 | page 40

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

More recently Tran et al. (2015) refined the limiting strain distribution by:

using strains predicted using a linear elastic model (PerRoad) instead of measured strains
considering the performance of sections loaded during Phase IV commencing 2009.
This significantly reduced the limiting strain values as shown in Figure 4.8. The updated limiting strains were
well below the values obtained by Willis and Timm (2009) (Figure 4.7).
Figure 4.7: Limiting measured strain for long-life pavements

Source: Willis and Timm (2009).


Figure 4.8: Refined limiting strain criteria

Source: Tran et al. (2015).

Austroads 2016 | page 41

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

4.4

Summary of Field Studies

Investigations of the performance of thick (> 200 mm) asphalt pavements undertaken in the United Kingdom,
Australia and USA have generally found the pavements to be performing well.
Unlike other studies, the UK research sampled the pavement to measure the extent to which fatigue damage
was accumulating. The estimated residual fatigue lives of asphalt cores taken in the wheel path were
compared to lives of between wheel path cores. It was concluded that fatigue damage was not accumulating
due to loading of selected thick asphalt pavements. Pooling all the information of their study it was concluded
that for the UK climate, materials and traffic loadings a total asphalt thickness of 270 mm placed on a firm
foundation was sufficient for a long but indeterminate pavement life.
Field studies on performance of thick asphalt pavements have also been conducted in Australia (AAPA
2000, 2004, Paul 1997 and Tsoumbanos 2006). A range of pavements subject to varying traffic levels and
up to about 30 years of age were investigated. These studies found comparable roughness, rutting and
cracking trends as Nunn et al. (1997) with the vast majority of sites determined to be in a sound structural
condition and meeting functional requirements despite many having, at the time, already exceeded the
predicted design life. Unlike the UK research, these studies did not assess the difference in residual life
between trafficked and untrafficked asphalt to determine whether there was evidence of fatigue occurring in
the trafficked sections or whether the differences could be attributed to age hardening. However, given that
such pavements are designed for a structural life of 2030 years with a probability of 95% of outlasting the
design traffic the results are not inconsistent with the current Austroads design model without allowance for
an asphalt endurance limit. Therefore no reliable conclusion can be drawn as to whether these pavements
will need deep structural treatments in the future.
In addition, an analysis of the long-term pavement performance data in the USA was also inconclusive about
the existence or otherwise of an endurance limit.
In the NCAT studies some pavement sections were identified as not having fatigue cracking after
3 x 107 ESA of loading. These pavements were used to derive a field-based limiting strain threshold.
However, as this traffic loading of 3 x 107 ESA is less than 1/10th that of design traffic of heavily trafficked
roads over 40 years (about 4 x 108 ESA), it is considered the data was insufficient to conclude whether or not
the trial sections were long-life pavements i.e. pavements that would never require deep structural
treatments due to asphalt fatigue cracking under any traffic flow.
In summary a number of research projects have observed thick asphalt pavement performance not
inconsistent with the concept of an asphalt fatigue endurance limit. It is considered that the UK data is of
most value in developing a design procedure for the Guide. This is discussed further in Section 7.

Austroads 2016 | page 42

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

5. Asphalt Pavement Solutions for Life Project


5.1

Introduction

In 2011 the Australian Asphalt Pavement Association (AAPA) commenced the Asphalt Pavement Solutions
for Life (APS-fL) project. This major research project was initiated to address the concerns of clients,
consultants and industry that current pavement design procedures were producing overly conservative
asphalt thickness requirements.
As part of the project, a procedure was developed (Sullivan et al. 2015) to incorporate endurance limit strains
(ELS) into the Guide pavement structural design. The intent of the APS-fL design procedure was to
determine the maximum tensile strain(s) at which, whilst damage might occur, the asphalt will never develop
macro-cracking requiring deep structural treatment.
As described by Sullivan et al. (2015):
The long-life asphalt pavement (LLAP) concept utilises a three layered asphalt system;
a wearing course and levelling course (surfacing layers), which should be durable and rut
resistant
intermediate layer(s), which should be rut resistant and be the main structural layers
fatigue layer, which should be a fatigue resistant low air void mix.
The high rut resistant durable surface layers may consist of dense grade asphalt (It is noted
that many SRAs do not utilise dense graded asphalt (DGA) as a wearing course on freeways
due to texture requirements) or a SMA, typically a maximum size of 14 mm. Where an OGA
surfacing is proposed for noise and/or drainage considerations, an additional surface layer
consisting of a 14 mm DGA shall be provided immediately below the OGA surface. For
durability and rut resistance, the asphalt in the surfacing layer(s) may incorporate a modified
binder, if so an A35P or Multigrade is recommended over elastomeric binders such as A10
or A15E as A35P offers more structural capacity to the pavement.
The intermediate asphalt layers(s) should consist of conventional (C320, C450 or C600)
asphalt DG20 mix which is durable and rut resistant suitable for the climatic region.
It is recommended that the LLAP incorporate a low air void (< 4%) asphalt fatigue layer, in
the bottom 60mm of the pavement. The low void layer is usually achieved by the use of
higher binder content in the bottom layer of the DG20 asphalt mix. This lower level asphalt
layer should consist of no more than 30% recycled asphalt pavement, if more than 30% RAP
it is recommended that the stiffness-FEL relationship of the mix be confirmed to fall on or
above the standard stiffness-FEL mix curve.
For LLAP, Sullivan et al. (2015) recommended that a working platform be provided below the asphalt to
achieve a supporting layer of sufficient quality to be assigned a design modulus of 100 MPa (or greater).
Critical aspects of the development and implementation of the APS-fL endurance strains are described
below in relation to their possible inclusion in the Guide (Austroads 2012).

Austroads 2016 | page 43

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

5.2

Outline of the Proposed Method

Sullivan et al. (2015) outlines the procedure in detail, the critical aspects of the method are summarised
below.
To undertake a structural design, the APS-fL method requires the asphalt strains to be predicted using a
linear elastic model and compared to the ELS for the asphalt mixes. As the ELS varies with temperature, this
strain comparison is undertaken at three pavement temperatures:

midday in the hottest summer month of the year


Weighted Mean Annual Pavement Temperature (WMAPT) as currently used in the Guide
early morning in the coolest winter month.
For each temperature, the asphalt design moduli are calculated and asphalt tensile strain at the bottom of
the asphalt is calculated under an 80 kN (8.2 tonnes) Standard Axle (single axle with dual wheels) load.
The predicted asphalt strains are then compared to the ELS calculated for the asphalt design modulus and
the confidence level using Equation 10.

8.2
ELS = (k1 21625 0.65 + k 2 )

where

10

ELS

endurance limit strain (microstrain)

asphalt modulus (MPa)

upper 97.5th single load (tonnes), usually taken as 9 tonne (Sullivan et al. 2015)

k1

adjustment constants for differences in rest periods, or confidence levels, given in


Table 5.1

k2

mix adjustment factor (0 for conventional mixes)

Table 5.1:

Recommended k1 values

Confidence level

80%

85%

90%

95%

97.5%

k1

1.00

0.97

0.925

0.86

0.80

Source: Sullivan et al. (2015).

Figure 5.1 shows the variation of ELS with asphalt modulus.


The design process requires the determination, by iteration, of the minimum asphalt thickness that has
predicted strains for all three temperature conditions below the ELS values.

Austroads 2016 | page 44

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 5.1: APS-fL proposed in-service endurance limit strains


200

180
90th confidence limit

160

95th confidence limit


140

Endurance
limit
120
strain
(microstrain)

97.5th confidence limit

100

80

60

40
1000

2000

3000

4000

5000

6000

7000

8000

Asphalt modulus (MPa)

5.3

Summary of Method Development

5.3.1 Introduction
As described in Section 3, research project NCHRP 9-44A (Witczak et al. 2013) recommended that a
laboratory ELS be predicted from asphalt modulus, modulus being a surrogate for the binder content, air
voids, binder grade, and temperature. Building on this research Sullivan et al. (2015) also developed a
relationship to predict ELS from modulus, but the predicted ELS values differ from the NCHRP 9-44A project
values.
The APS-fL method was determined by:

deriving a laboratory ELS relationship, as described in Section 5.3.2


adjusting these laboratory ELS values to in-service ELS values by
identifying pavement sections to be used in the field calibration (Section 5.3.3)
determining the layer modulus for these sections over a range of temperatures and calculating the
applied strains (Section 5.3.4)

determining a relationship that relates the applied strain to the asphalt modulus (Section 5.3.5)
These processes are detailed by Sullivan et al. (2015) and are outlined below.

5.3.2 Laboratory ELS Relationship


To develop a relationship between laboratory ELS and asphalt modulus Sullivan et al. (2015) utilised the
ELS estimated from beam fatigue testing at a test temperature of 20 C. The data included:

10 Illinois mixes tested and analysed by Thompson and Carpenter (2006), as described in Section 2.6.1
Australian mixes shown in Table 5.2.

Austroads 2016 | page 45

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 5.2:

Fatigue, modulus and ELS results of Australian mixes


Fatigue equation
constants
k

Initial
modulus
(MPa)

A5E

2127

6.9

9 722

150

EME2

3295

6.1

9 021

168

5.4

APH

1243

9.1

8 580

169

4.2

C320

2740

5.6

9 178

109

14 mm Granite C320 Perth

4.5

C320

3676

5.2

7 996

110

AC20 AR450 High RAP

4.7

AR450

3801

4.7

8 519

81

C320 Geelong High RAP

4.9

C320

3413

4.8

7 947

77

14TCI Base (sample 1)

4.9

C320

3587

5.2

8 848

110

14TCI Intermediate sample 1

4.9

C320

2880

5.6

9 111

111

MSTR Bunbury

5.2

C320

3521

6.0

4 790

170

AC14 Hun 5.3% (NZ)

5.3

PG64

5163

5.0

5 681

134

AC14 5.4% C320

5.4

C320

3441

5.3

6 828

107

AC14 C600

5.4

C600

2816

5.8

7 256

122

DG10 C170 W.A. Granite

5.6

C170

4853

5.0

6 125

126

AC10 5.7% C450

5.7

AR450

3599

5.2

8 947

109

AC10 5.7% C320

5.7

C320

3266

5.3

7 730

109

DG10 Granite Perth 5.1%

5.1

C320

4578

5.0

5 738

120

10mm DG Granite (21) Perth

5.1

C320

4955

5.0

7 214

128

10mm DG Granite (22) Perth 2

5.1

C320

5012

5.0

7 218

131

AC14 Hun 4.65% Bitumen PGT64 15% RAP10

4.65

PG64

2467

7.2

5 807

195

DG10 Granite

5.1

A15E

5765

4.8

6 021

133

DG14 4.72% A15E ex-Crestmead

4.7

A15E

5570

5.0

7 510

149

AC14 5.4% Toner binder (MK II) 2 M2 B2

5.4

A10E + toner

2268

8.5

3 594

265

AC14 5.4% A10E

5.4

A10E

4012

6.7

3 504

262

AC10 5.7% A10E

5.7

A10E

6768

5.5

3 342

243

AC10 5.7% A15E

5.7

A15E

7014

5.7

3 423

284

AC10 5.7% A20E

5.7

A20E

6653

5.1

4 347

188

WA Testing PMB AC14 Ex Port Hedland

5.1

A10E

5350

5.0

6 966

142

10 mm Granite /75 blows C320 + PMB

5.0

A15E

6412

5.3

5 002

200

DG10 Granite

4.8

A15E

5765

4.8

6 021

133

14 mm MSR Basalt A20E ex-Bunbury

5.0

A20E

5490

5.0

4 826

145

A35P

6286

4.2

6 891

86

AC14 5.4% A35P

5.4

A35P

3578

5.6

7 789

140

AC10 5.7% A35P (EVA)

5.7

A35P

5115

5.0

7 120

131

14 mm Granite PMB

4.8

A35P

4648

5.0

7 010

122

20 mm Granite EVA ex-Hazelmere

4.3

A35P

5206

4.4

10 273

83

14mm Granite EVA ex-Hazelmere

4.8

A35P

3824

5.1

6 865

111

10 mm Granite /75 blows C320 + EVA

5.1

A35P

9785

3.6

8 920

63

DG10 EVA ex-WA

5.1

A35P

5834

5.1

5 507

161

Binder
%

Binder
type

AC14 5.4% A5E

5.4

EME NSW Bass Point Mix

5.6

Fulton Hogan PortPhalt


AC20 C320 Wallgrove (AS2150)

Mix Description

AC14 5.0% A35P Racetrack

ELS
()

Source: Sullivan et al. (2015).

Austroads 2016 | page 46

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

The Table 5.2 ELS values were determined by Sullivan et al. (2015) by:

assuming the endurance strain occurs at a laboratory fatigue life of 8 x 107 cycles of loading
extrapolating the fatigue relationships determined by testing at normal strain levels to determine the low
strain levels at 8 x 107 cycles.
Plotted in Figure 5.2 are the ELS of Australian mixes, together with the relationship derived by Thompson
and Carpenter (2006) from testing 10 Illinois mixes (see Section 2.6.1). Note that the Thompson and
Carpenter ELS were calculated at a fatigue life of 1.1 x 107 cycles (Section 2.6.1), whereas the Sullivan et al.
(2015) data were calculated at 8 x 107 cycles. This difference in definition should results in the Sullivan et al.
(2015). ELS being about 50% lower than calculated using the Thompson and Carpenter method. Despite the
different methods of determining ELS, Sullivan et al. (2015) concluded that the ELS of the Australian and
Illinois mixes were sufficiently similar for the data to be combined to form a single ELS-modulus relationship.
Figure 5.2: Laboratory ELS derived by combining data from Australian and Illinois mixes
310
290

C170, C320 and C600

270

A35P

250

A10E, A15E

230

Thompson &
Carpenter 2006

210

Endurance 190
limit
170
strain
(microstrain) 150
130
110
Sullivan et al. 2015
K1=1

90
70
50
30
1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

Flexural modulus (MPa)

To derive this relationship, the strains determined under haversine loading were halved to allow for beam
deformation (Section 2.8). The resulting ELS relationship (Equation 11), which is applicable to conventional,
EVA and SBS modified mixes, is plotted in Figure 5.2. Note that the Thompson and Carpenter (2006)
relationship relates the strains under haversine loading to modulus, whereas the Sullivan et al. (2015)
relationship is for strains applied under sinusoidal loading.

= 1 20200 0.65

11

where

laboratory endurance limit strain (microstrain)

initial flexural modulus (MPa)

adjustment factor for differences in rest periods and/or confidence levels

Austroads 2016 | page 47

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

5.3.3 Selection of Pavement Sections for Field Calibration


Australian sites
The intent of the APS-fL design procedure is to determine the maximum tensile strain(s) at which, whilst
damage might occur, the asphalt will never develop macro-cracking requiring deep structural treatment. This
is different from some other definitions of an ELS in which the pavement structure does not accumulate
fatigue damage.
A key need in developing the method was to identify heavily-trafficked pavements that have and have not
developed fatigue macro-cracking during their structural design lives (typically 3040 years). This task is
challenging as maintenance and rehabilitation records over such extended periods of time do not commonly
exist in Australia and many other countries.
A major source of performance data used by Sullivan et al. (2015) in the method development was a 2009
Roads and Maritime Services (Roads and Maritime) New South Wales investigation of pavement
performance (Sullivan et al. 2015). This study produced an extensive database of pavement condition,
structural capacity, layer thicknesses, materials and current visual condition of the pavements. However a
major limitation was the lack of past visual condition data to assess whether pavements had experienced
macro-cracking at any time since construction. Maintenance and rehabilitation data since construction was
also inadequate in this regard.
To develop the ELS-modulus model (Equation 10) the Roads and Maritime database was examined to find
both partial and full depth asphalt pavements which may be either a non-LLAP (susceptible to fatigue
cracking) or a LLAP (not susceptible to fatigue cracking). This subset of the Roads and Maritime database
was then re-examined to identify potential LLAP. This subset was obtained by filtering for pavements which:

had greater than 140 mm combined asphalt thickness


were greater than 20 years old
had cumulative traffic loading greater than 30 million ESA.
Additional pavement sections were added to the analysis which were used to establish the break point
between LLAP and fatigue susceptible pavements. The criteria used to select these fatigue susceptible
pavements were that they had greater than 240 mm thickness of asphalt, were fatigue cracked without
constraints on cumulative traffic loading and/or age.
To develop the ELS relationship these pavements were divided into three categories:

LLAP
possible LLAP
non-LLAP.
According to Sullivan et al:
The categorisation was undertaken by examining the results of the visual survey undertaken
at the time of testing, the current visual condition of the pavement and any comments on
asphalt condition and by examining the assessment of the remaining life of the pavement
determined as part of the visual assessment.
However, as the remaining life estimates for the sites categorised as LLAP sites were mostly less than
10 years, it seems the visual assessment was the main characteristic used in deciding whether these
pavements were LLAP. There is clearly uncertainty as to whether these are in fact LLAP as recently placed
surface treatment (such as mill and asphalt overlays) may lead to sites with fatigue cracking being
misclassified as LLAP sites.

Austroads 2016 | page 48

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

In addition to the Roads and Maritime sites, the data collected for the previous AAPA studies (AAPA 2000,
2004) was inspected to find any additional sites which could be used to establish the ELS for Australian
LLAP design. Five sites were classified as LLAP one of which (Q6, Bruce Highway) was used in the
derivation of the APS-fL ELS criteria.
The seven pavement sites categorised by Sullivan et al. (2015) as LLAP (not fatigue susceptible) and
included in the derivation of the ELS criteria are given in Table 5.3. Note that all sites had an asphalt
thickness of 300 mm or less. Note that at the time the visual inspection was conducted for the LLAP
categorisation, cumulative traffic loadings of these LLAP sites were 108 ESA or less. Such a traffic loading
can occur over a 20 year period on a heavily trafficked Australian freeway, well short of the design periods of
3040 years commonly used for such pavements. This illustrates the difficulty in identifying LLAP sites which
had no structural cracking after cumulative traffic exceeding 108 ESA.
Table 5.4 summarises the Roads and Maritime sites classified by Sullivan et al. (2015) as fatigue susceptible
(e.g. non-LLAP).
Table 5.3:

Sites classified as LLAP and used in the derivation of the ELS criteria
Cumulative traffic (ESA)
2009

2015

Year of
construction

New England
Highway Beresfield

8 x 107

1 x 108

1970

190210 mm asphalt
fine to coarse gravel
poorly graded sand
subgrade

H-S27

Pacific Motorway,
Cheero Point

5 x 107

6 x 107

1990

180205 mm asphalt
300 mm fine to coarse gravel
fine to coarse gravel

H-S29

Pacific Motorway
1.5 km south of
Ourimbah exit

5 x 107

6 x 107

1989

140160 mm asphalt
300 mm fine to coarse gravel
fine to coarse gravel

SOU-S035

Picton Road

4 x 107

5 x 107

1966

215280 mm asphalt
1000 mm sandy iron stone gravel
bedrock

Syd-S13

Hume Motorway,
St Andrews

1 x 108

1973

260270 mm asphalt
gravel
clayey gravel
subgrade

Syd-S15

Hume Motorway
St Andrews

1 x 108

1973

300 mm asphalt
175 mm gravel
clay

Q6

Bruce Highway
2.22.4 km South of
Boundary Road (SB)

1979

300 mm asphalt
working platform
subgrade

Site ID

Road/location

H-S22

3 x 107

Pavement structure

Source: Sullivan et al. (2015).

Austroads 2016 | page 49

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 5.4:

Sites classified as non LLAP from Roads and Maritime database


Cumulative traffic (ESA)

Year of
construction

New England
Highway

8 x 107

1970

220 mm asphalt
230 mm fine to coarse gravel
810 mm poorly graded sand
SG

Sou-SO86

Princess Highway,
South of Gerringong

9 x 106

1969

200 mm asphalt
200 mm unbound gravel
300 mm gravelly clay
SG

Syd-S26

Bunnerong Rd
Matraville

1.4 x 107

1960

160 mm asphalt
200 mm poorly graded gravel
300 mm poorly graded sand
SG

Syd-S23

The Grand Parade,


Monterey

4.5 x 106

2000

200 mm asphalt
150 mm poorly graded gravel
550 mm poorly graded sand
SG

Syd-S46

Rocky Point Road,


Beverly Park

1.6 x 107

1966

160 mm asphalt
240 mm poorly graded gravel
550 mm poorly graded sand
SG

H-S39(1)

Sydney Newcastle
Freeway

8.9 x 107

1970

155165 mm asphalt
440 mm fine to coarse gravel
Clay

H-S24

Sydney Newcastle
Freeway

8.9 x 107

1970

240260 mm asphalt
220330 mm fine to coarse gravel
gravel

Syd-S24

Woodville Road,
Merrylands

3.7 x 107

1970

190180 mm asphalt
300 mm fine to coarse gravel
(some poorly graded)
clay

Syd-S25

Woodville Road
Guilford

3.7 x 107

1970

150 mm asphalt
200450 mm poorly graded gravel

Syd-S33

Bunnerong Road,
Matraville

1.2 x 107

1960

150 mm asphalt
320 mm fine to coarse gravel
poorly graded sand

Sou-S089

Princess Highway,
Conjola

1.2 x 106

1985

240 mm asphalt
120 mm quartz gravel
sandy gravel

Site ID

Road/location

H-S23

Pavement structure

It is known that stripping of asphalt has occurred with the mixes placed in this location. However, as no stripping was
observed in the cores, for conservatism this site has been included in the fatigue susceptible pavements.

Source: Sullivan et al. (2015).

NCAT test sections


In addition to the Australian sites, Sullivan et al. (2015) considered that NCAT test sections (Section 4.3.2)
N3 and N4 were most likely to be LLAP, consistent with the conclusions of Willis et al. (2009). The NCAT trial
data and analysis are described in Section 4.3.2.
As described in Figure 4.4 test sections N3 and N4 comprised a total asphalt thickness of 225 mm
(9 inches), with SBS polymer modified mixes being used in Section N4.

Austroads 2016 | page 50

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Sections N3 and N4 were constructed in 2003 and were subjected to 107 ESA of loading in each of the
following loading stages: 20032005 (Phase II), 20062008 (Phase III) and 20092011 (Phase IV). In 2009
after a total of 2 x107 ESA of loading no cracking was observed. In 2012 after a total of 3 x 107 ESA of
loading some minor cracking was observed but from pavement coring this was considered to be top-down
cracking rather than bottom-up classical fatigue cracking.
Note that a cumulative traffic loading of 3 x107 ESA is a traffic loading which can occur over a
1015 year period on a heavily trafficked Australian freeway, well short of the design periods of
3040 years commonly used for such pavements. This illustrates again the difficulty in identifying sites which
had no structural cracking after cumulative traffic loading exceeding 108 ESA.

United Kingdom performance data


To add to the calibration data three sites from the UK Validation Monitoring of Design and Assessment
Methods (VALMON) (Jones et al. 2006) were included.
In the VALMON project, a total of 25 medium to heavily trafficked full-depth asphalt sites were monitored
over an 11 year period from 1994 to 2005. Each VALMON site was about 1000 m in length. From this data
Jones et al. concluded:
The predominant modes of distress for long-life, fully flexible roads having hot-rolled asphalt
surfacing were found to be rutting and chip loss/fretting. Very few sites of this type were
suffering from cracking, even after 10 years of trafficking.
Deflectograph and FWD measurements made at regular intervals on fifteen, fully flexible,
long-life pavements support the assumption that any distress that occurs on these types of
pavements is confined to the surface layers.
From these 25 VALMON sites, Sullivan et al. (2015) selected three sites to derive the ELS prediction
method: namely, Brentwood, Cole Green and Thornbury sites. Table 5.5 provides detail of the sites extracted
from the UK report (Jones et al. 2006). Note that the road A414 (Cole Green) is a moderately-trafficked road,
whereas the motorways M5 and M25 are very heavily-trafficked.
Table 5.5:

1
2

VALMON sites used to develop the APS-fL method

Road

Road/location

Motorway M25

Brentwood
ID 03

Cumulative
traffic(1)
(ESA)

Year of
construction

8.8 x 107

1982

Original
pavement
structure
310 mm
asphalt

106

1993

305 mm
asphalt

1971

360 mm
asphalt

A414

Cole Green
ID 16

4.4 x

Motorway M5

Thornbury
ID 32

6.9 x 107

Maintenance(2)
Year

Treatment

1996

Hot rolled asphalt inlay

2001

Thin asphalt overlay

No treatment during the study


period
1982

Asphalt overlay

1998

Asphalt inlay (part


length)

2001

Thin asphalt overlay

From construction until 2005.


Data until 2005 survey.

Source: Jones et al. (2006).

Austroads 2016 | page 51

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

During the 11 year VALMON study period surface deflections were regularly measured using deflectograph
and FWD equipment. As seen from Figure 5.3, most sites did not increase in deflection with loading.
However, the Cole Green site (16A, 16B) was increasing which is not consistent with the behaviour of longlife asphalt pavements (Jones et al. 2006).
Figure 5.3: Variation in falling weight deflectometer maximum deflections with cumulative traffic loading
FWD - D1 Against Traffic

250.0

FWD - D1 / microns_

200.0

16A, 16B

150.0

100.0

50.0

0.0
0

10

20

30

40
50
Traffic / MSA

60

70

80

90

Source: Jones et al. (2006).

Jones et al. (2006) classified the pavements into the following three types:

potentially long-life pavements (LLPs)


upgradeable to long-life pavements (ULLPs)
determinate life pavements (DLPs).
The classification procedure used was that in the Structural Assessment Methods HD25/94 (Highways
England 1994) which has now been withdrawn. The procedure requires the total thickness of sound
bituminous material (TTBM) to be 300 mm or more and for the deflectograph maximum deflections to not
exceed the line plotted in Figure 5.4. Jones et al. classified the VALMON sites using this method. Due to
their high deflections (Figure 5.4), Jones et al. concluded that both Brentwood (site 03) and Thornbury (site
32) were not long-life pavements, but were pavements with determinate life.
No details are provided by Sullivan et al. (2015) on the reasons these three sites were selected as LLAP for
the calibration from the full suite of 25 VALMON sites.

Austroads 2016 | page 52

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 5.4: Assessment of VALMON sites as potential long-life pavements

Sites 01 to 12
0.5

Site:

Upgradeable to LLP

01
02

LLP

03

0.4

04
05
06

Deflection (mm)

07
08

0.3

09
10
11
12E HMB35

0.2

12E DBM50
12W DBM50
12W HMB15

0.1

0
0

100

200

300

400

500

600

700

800

900

1000

TTBM (mm)

Sites 13 to 45
0.5

Site:

Upgradeable to LLP
LLP

Deflection (mm)

0.4

0.3

13

18

14

29

15

32

16A

33

16B

35

17A

45 Polegate

17B

0.2

0.1

0
0

100

200

300

400

500

600

700

800

900

1000

TTBM (mm)

Source: Jones et al. (2006).

Austroads 2016 | page 53

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

5.3.4 Determination of Moduli and Asphalt Strains


To derive the relationship between ELS and asphalt modulus for each selected pavement section
(Section 5.3.3) the following process was used by Sullivan et al. (2015):

Using the measured thicknesses of asphalt, granular base and granular subbase, and the measured
falling weight deflectometer (FWD) surface deflections, moduli were estimated for granular base, granular
subbase and subgrade. A five layer linear isotopic model was used. The back-calculation method also
considered the depth to bedrock. For example, Table 5.6 list the results for the Australian LLAP sites.

The mid-layer asphalt temperature was calculated for four seasons: summer, autumn, winter, and spring.
Additionally, the mid-layer temperature was calculated for the upper 10th percentile summer temperature
and lower 10th percentile winter temperature. These mid-depth asphalt temperatures were estimated from
the mean air temperatures using the calibrated Bells equation procedure.

The design vehicle speed for the heavy vehicle was taken as 10 km/h lower than that of the posted speed
limit.

At all sites, the asphalt layers were considered to comprise 40 mm thick dense graded asphalt surfacing
(AC14C320) and the remaining asphalt base thickness was assumed to be size 20 mm dense graded
asphalt (AC20C320). Asphalt moduli for these mixes were calculated at the pavement temperatures and
design vehicle speed described above using the measured dynamic compressive moduli measured by
Sullivan et al. (2015) for Australian mixes.

In the linear elastic strain calculations, it was decided to model the asphalt as a single layer rather than
allow for variation in mix type with depth. This required calculation of the effective modulus determined
from moduli and thicknesses of the AC14C320 and AC20C320 mixes.

Tensile strains at the base of the asphalt were calculated under a single axle with dual tyres loaded to
9.0 tonnes.
Note that the characterisations of the granular layers and subgrade differ from those used in the Guide for
the design of new pavements. For example, Table 5.6 lists the back-calculated granular base, subbase and
subgrade moduli for the LLAP sites used in the derivation of the ELS criteria. The back-calculated subgrade
moduli are generally high compared to the conservative design moduli used in the design of new pavements:
subgrade design moduli are commonly determined for laboratory soaked CBR testing. The use of these high
back-calculated subgrade moduli result in lower calculated strains than obtained in the design of new
pavements. In addition the back-calculated granular base moduli in Table 5.6 tend to be higher than those
used in the design of new pavements (Austroads 2012). Conversely, the back-calculated granular and
subgrade moduli were isotropic values compared to the anisotropic characterisation used in the Guide
(Austroads 2012). It appears that Sullivan et al. (2015) have allowed this difference in characterisation by a
7% increase in ELS.

Austroads 2016 | page 54

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 5.6:

Moduli back-calculated for the Australian LLAP sites used in the derivation of the ELS criteria

Site ID

Temperature
Asphalt
(C)

Layer thickness (mm)

Layer isotropic modulus (MPa)

Granular Granular Upper


base
subbase subgrade

Asphalt

Granular Granular Upper


Lower
base
subbase subgrade subgrade

H-S22_1

31.9

200

200

300

2933

2288

206

112

125

5000

H-S22_3

32.0

210

190

300

2978

2538

201

97

114

5000

H-S22_9

31.5

200

220

300

2700

2410

223

94

109

5000

H-S27_13

21.4

180

195

300

675

1312

345

66

55

5000

H-S27_16

21.9

200

200

300

650

1030

292

70

35

5000

H-S27_18

21.8

205

215

300

630

1124

344

49

99

5000

H-S29_1

26.5

160

190

70

930

3203

681

8139

84

5000

H-S29_10

25.1

140

150

115

945

3643

351

5760

166

5000

SOU-S035_3

21.3

280

300

700

5000

3200

150

80

150

5000

SOU-S035_7

21.3

280

300

700

5000

2636

134

154

171

5000

SYD-S13_4

30.6

260

170

400

1256

2450

1250

180

700

5000

SYD-S15_18

30.8

210

165

400

4054

1686

228

460

292

5000

V6-20

30.8

210

150

400

2232

2146

273

99

108

5000

Source: Sullivan et al. (2015).

5.3.5 Determination of In-service ELS Relationship


The calculated strains for each site and temperature are plotted against the effective asphalt modulus in
Figure 5.5. It is apparent there is substantial overlap between the LLAP sites and the Australian pavements
which are susceptible to fatigue cracking.
Figure 5.5: Strain distributions and initial design curve

Source: Sullivan et al. (2015).

Austroads 2016 | page 55

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Shown in Figure 5.5 is the in-service ELS relationship initially developed by Sullivan et al. (2015) by adjusting
the laboratory ELS (Section 5.3.2). This initial relationship, given in Equation 12, was based on the notion
that ELS values should be such that there is a low probability that pavements designed with the method
become fatigue cracked. The position of the line was adjusted until the strains calculated for almost all the
Australian cracked sites were above the ELS design curve.

ELS = 0.920200E0.65

12

where

ELS

endurance limit strain (microstrain)

asphalt modulus (MPa)

Sullivan et al. (2015) undertook further analysis to include confidence levels in the relationship. These
confidence levels were developed so that designers could select and design for desired probabilities of
achieving a LLAP. The process used to derive these confidence levels was as follows (Sullivan et al. 2015):
To establish confidence limits the predicted strain for each of the LLAP sections which fell
between the section with the lowest strain-stiffness relationship, which had shown signs of
failure, (the Sydney Newcastle Freeway section H-S39) and the section with highest
stiffness-strain relationship which showed no signs of failure (H-S29 and N4 as lines cross
over) were analysed. The analysis fitted a power relationship to the stiffness strain curve for
each of the 14 sections which fell in the indeterminate zone. These relationships were then
used to predict the strain at stiffness values near the upper summer temperature (1500
MPa), the WMAPT (3000 MPa) and winter lower (6500 MPa). For each of the stiffness
values the strains were then ranked in order of highest to lowest and the strains were then
plotted against the inverse standard normal distribution (normal (z) score) as shown in
Figure 5.6 following.
Figure 5.6: Normal score of ELS

Source: Sullivan et al. (2015).

Austroads 2016 | page 56

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 5.7 shows the indeterminate zone which included LLAP and fatigue susceptible pavements. The
assumption in defining this zone is that any pavement with strains below this zone has a high probability of
being LLAP and those pavements with strains above this zone have a high probability of being fatigue
susceptible.
From this analysis, Equation 13 was derived by Sullivan et al. (2015):

ELS = k1 20200E 0.65 + k 2

13

where

ELS

endurance limit strain (microstrain)

asphalt modulus (MPa)

k1

adjustment constants for differences in rest periods, or confidence levels as shown


in Table 5.1

k2

mix adjustment factor

Figure 5.8 shows the ELS for 95% and 80% confidence levels.
Figure 5.7: Indeterminate zone between the two solid black lines

Source: Adapted from Sullivan et al. (2015).

Austroads 2016 | page 57

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 5.8: Proposed in-service ELS for 95% and 80% confidence levels

Source: Sullivan et al. (2015).

5.3.6 In-service ELS Relationship Proposed for the Guide


Considering the Austroads elastic characterisation method is different from the procedure used to derive the
data in Figure 5.8, Equation 13 was adjusted by Sullivan et al. (2015) to obtain Equation 10 for use with the
Guide.
Figure 5.9 shows the in-service ELS relationship proposed for use in the Guide. The in-service ELS for 95%
confidence level are about 80% of the laboratory ELS values.

Austroads 2016 | page 58

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 5.9: Comparison of laboratory ELS relationship with the in-service ELS relationships for 95% confidence
level
200

180
Laboratory ELS

160

In service for 95th confidence limit


140

Endurance
limit
120
strain
(microstrain)
100

80

60

40
1000

2000

3000

4000

5000

6000

7000

8000

Asphalt modulus (MPa)

5.4

Summary

The APS-fL project has proposed a relationship to determine in-service ELS from the asphalt design
modulus for use in the Guide. The relationship is based on:

A laboratory ELS relationship obtained by analysing beam fatigue testing results at a temperature of
20 C of a range of Australian and Illinois asphalt mixes, including mixes made using polymer modified
binders.

Field performance of Australian and UK thick asphalt pavements, together with NCAT accelerated loading
data, was used to adjust the laboratory ESL relationship to an in-service ESL relationship.
The modulus dependency adopted for the in-service ELS relationship was that derived from the laboratory
beam fatigue testing. As seen from Figure 5.10, this modulus dependency is very different from that derived
in the NCHRP 9-44A laboratory testing (Section 3.9.1). This APS-fL modulus dependency was derived by
testing a wide range of mixes at one temperature (20 C). By comparison in the NHCRP 9-44A project the
data included a limited range of mixes tested at three temperatures. These differences in the sources of the
modulus variation may have been a factor in the different modulus dependencies.
This difference in ELS modulus dependency results in substantially different impacts on pavement
thicknesses as described in Section 7.2 and Section 7.3. This issue is discussed further in the development
of options for the design method for inclusion in the Guide (Section 7).

Austroads 2016 | page 59

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 5.10:

Comparison of laboratory ELS


230
210
190
170

Witczak et al. 2015 rest period 5 seconds


Sullivan et al. 2015, lab ELS

Laboratory 150
endurance
limit strain
(microstrain) 130
110
90
70
50
1000

2000

3000

4000

5000

6000

7000

Asphalt modulus (MPa)

Austroads 2016 | page 60

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

6. Use of Asphalt Fatigue Endurance Limits in


Overseas Design Processes
6.1

Introduction

Research was undertaken to identify and investigate fatigue endurance limits incorporated in road agency
pavement design manual. Only three methods have been identified to date:

UK Design Manual for Roads and Bridges (Highways England 2006)


Texas Department of Transportation Pavement Design Manual (TxDOT 2011)
Illinois Department of Transportation Pavement Design Manual (IDOT 2013).
These methods are summarised below.

6.2

United Kingdom

Section 4.1 described the research that lead to the development of the UK long-life asphalt pavement design
method in the Design Manual for Roads and Bridges (Highways England 2006).
Figure 6.1 is the design chart used to determine the total asphalt thickness for a range of asphalt mix types
and degrees of support (Foundation Classes) provided by the layers under asphalt. It is noted that for design
traffic loadings of 8 x 107 ESA or more the same asphalt thicknesses apply, these being long-life asphalt
pavements.
Figure 6.1: UK design thicknesses for flexible pavements

Source: Highways England (2006).

Austroads 2016 | page 61

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Foundation Class 2 corresponds to an effective half-space modulus of 100 MPa and commonly comprises
capping materials and unbound granular subbases, whereas Class 3 and 4 include bound subbases. Note
that the Design Manual states that if Class 2 foundation is used for design traffic in excess of 8 x 107 ESA
(long-life pavements) it needs to include 150 mm or more of bound subbase.
As discussed in Appendix B.2.2, the heavy-duty macadam using penetration grade 40/60 bitumen (labelled
as HDM50) is the asphalt type closest to the commonly used Australian base course mix, a size 20 mm
dense graded asphalt made using Class 320 bitumen. From Figure 6.1, for a long-life pavement a total
thickness of 360 mm is required on Foundation Class 2 that must include a minimum 150 mm thickness of
bound subbase. An example of the materials and thicknesses of such an LLAP pavement structure is shown
in Table 6.1.
Table 6.1:

Example long-life asphalt pavement

Material type

Austroads elastic characterisation


Vertical modulus
(MPa)

Horizontal modulus
(MPa)

Poissons ratio

Hot rolled asphalt, pen 50

40

4370

4370

0.40

Asphalt concrete (AC20), pen 50

60

7240

7240

0.40

Asphalt concrete (AC32), pen 50

190

7470

7470

0.40

Cement treated crushed rock

150

500

250

0.35

170(1)

150(1)

75(1)

0.35

Semi-infinite

50

25

0.45

Unbound granular subbase


Subgrade
1

Thickness
(mm)

For the strain calculations the subbase thickness was divided into five equi-thick sublayers. The modulus of the top
sublayer is listed.

It was of interest to calculate allowable traffic loading of this pavement configuration using the design
procedures in the Austroads Guide. In the Guide, the characteristic temperature for the calculation of asphalt
modulus is the Weighted Mean Annual Pavement Temperature (WMAPT). The WMAPT for London is
18.7 C. At this temperature and a design traffic speed of 80 km/h, the asphalt design moduli listed in
Table 6.1 were calculated as described in Appendix B.2.2. The cement bound granular material was
assumed to have consumed its fatigue life and hence was assumed to have a reduced modulus (500 MPa).
In accordance with the Guide, the critical strains under an 80 kN Standard Axle were calculated as follows:

tensile strain at bottom of the asphalt: 41 microstrain


vertical compressive strain on the top of the subgrade: 128 microstrain.
For 95% design reliability and a volume of binder for the AC32 basecourse of 10.5%, the allowable traffic
loading of this pavement is about 109 ESA. This is well above the maximum design traffic loading used in
Australia.
It is noted from Figure 6.1, that the design traffic limit of 8 x 107 ESA applies for a range of asphalt mixes and
hence range of asphalt moduli.
Note that in this case the fatigue endurance limit is a maximum design traffic loading rather than an
endurance limit strain.

Austroads 2016 | page 62

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

6.3

United States of America

6.3.1 AASHTO Design Method


The AASHTO Mechanistic-empirical Pavement Design Guide (AASHTO 2015) includes provision for the
consideration of fatigue endurance strains. The guide defines the endurance limit as the tensile strain or
stress below which no load-related fatigue damage occurs.
The limit is assumed to be independent of temperature or mixture modulus a single value is used for all
hot-mixed asphalt mixtures within a single run of the AASHTOWare software. The guide mentions that the
endurance limit was excluded from the global calibration effort and thus should not be used without recalibration of the fatigue cracking model.
If an endurance limit is used in the model, all tensile strains that are less than this limit are excluded from
calculating the fatigue damage for bottom-up or crocodile fatigue cracking.
However, this method has yet to provide guidance on the ELS values to use. As described in Section 3, the
provision of such guidance was the subject of NCHRP research project 9-44A.

6.3.2 Texas Department of Transportation


Walubita and Scullion (2010) describe the development of the perpetual pavement (long-life) design method
used in the Texas Department of Transportation (TxDOT) (2011) Pavement Design Guide.
Since 2001, the State of Texas has been designing and constructing perpetual pavements (PP) on some of
its heavily trafficked highways where the expected 20-year design traffic loading exceeds 3 x 107 ESA.
Walubita and Scullion (2010) state that prior to 2010, the typical pavement configuration of the 10 perpetual
pavements built between 2001 and 2004 was:

about 560 mm asphalt


about 200 mm of treated (lime or cement) base material
well compacted in situ subgrade material.
In 2005, a research study was initiated to validate, among other objectives, the Texas PP design concept
and make recommendations for the future design of Texas PP structures (Walubita, Liu & Scullion 2010). To
achieve these objectives, various research tasks were completed including the following:

construction monitoring and compaction quality measurements


extensive laboratory testing and material property characterization
traffic and response measurements for structural evaluations
field testing and periodic performance evaluations
comparative mix-design evaluations
modelling and software evaluations.
The above-mentioned ten PP were investigated. From the inspection and testing in 2009 after 58 years
trafficking it was concluded their performance was very satisfactory with no visible or structural defects such
as rutting or fatigue cracking. There were instances of poor construction joints and construction-related
cracking on some projects. The study made recommendations to address these issues.

Austroads 2016 | page 63

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Based on asphalt moduli measured on cores extracted from the pavements and moduli back-calculated from
Falling Weight Deflectometer deflection testing it was concluded that in general the PP structures were found
to be very stiff, with low temperature susceptibility as demonstrated by high in situ modulus values greater
than 10 000 MPa (1500 ksi) particularly the asphalt base course layers. The average back-calculated moduli
are listed in Table 6.2. These moduli were reasonably consistent with the dynamic compressive modulus of
field cores measured at a temperature of 25 C and 10 Hz frequency.
Table 6.2:

In situ asphalt moduli estimated from deflection testing of perpetual pavements

Mix type
Stone mastic asphalt

Average back-calculated modulus at 25 C


(MPa)
5 940

Size 20 mm dense graded asphalt binder course (Type C)

4 820

Size 25 mm dense graded asphalt base course (Type B)

10 300

Bitumen-rich asphalt fatigue layer

3 680

Source: Derived from Walubita, Liu and Scullion (2010).

It was concluded that:

The underestimation of the layer moduli in the original designs lead to conservative PP designs. This was
particularly the case for base course asphalt.

There is potential to optimise the PP designs down to about 300350 mm total asphalt thickness.
The TxDOT (2011) Pavement Design Guide provides procedures for the design of perpetual (long-life)
asphalt pavements. The procedures appear to have been developed from the research of Walubita and
Scullion (2010).
According to TxDOT (2011):
Structural deterioration typically occurs due to either classical bottom-up fatigue cracking, rutting
of the HMA layers, or rutting of the subgrade. Perpetual pavement is designed to withstand an
almost infinite number of axle loads without structural deterioration by limiting the level of loadinduced strain at the bottom of the HMA layers and top of the subgrade and using deformation
resistant HMA mixtures.
Perpetual pavements take into account the increased structural demands due to heavy truck
traffic, where cumulative one-direction traffic loading of more than 30 million ESALs over a 20
30 year design life is projected.
The design system is centred upon deflection measurements from the Falling Weight
Deflectometer (FWD). These measurements are then used by the MODULUS computer
program to determine the material properties of the pavement layers. The material properties of
the pavement layers are then used to compute alternative structural designs using FPS-19W.
Figure 6.2 shows the general structure of TxDOT perpetual pavements. It consists of:

50 mm thick stone mastic asphalt using PG 76-XX binder (similar stiffness to Class 600 bitumen)
surfacing or a surfacing comprising 40 mm thick open graded asphalt on 50 mm thick stone mastic
asphalt

X mm thick rut-resistant size 20 mm dense graded asphalt base made using a PG 76-XX binder (similar
stiffness to Class 600 bitumen)

Y mm thick rut-resistant dense graded asphalt base made using a PG 64-22 binder (similar stiffness to
Australian Class170/Class 320 bitumen)

a minimum 50 mm thick bitumen-rich dense graded asphalt fatigue layer using PG 64-22 binder
Based on data in Walubita, Lui and Scullion (2010), it seems binder contents of about 5.4% are used.

Austroads 2016 | page 64

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

a minimum 150 mm thick stiff working platform of either lime stabilised subgrade (minimum 6% lime) or
cement treated crushed rock (maximum 3% cement). This working platform has a design moduli of 200
300 MPa.
Figure 6.2: General perpetual pavement design

Source: Walubita and Scullion (2010).

The design process firstly involves calculating the required pavement structure for a 20 year design period
using FPS-19W. FPS-19W is a mechanistic-empirical design procedure that uses a performance model
based on degradation of the serviceability index as defined in the AASHTO road test research. The FPS19W program assumes that a smaller deflection means smaller stresses or strains and, therefore, longer life.
FPS-19W uses back-calculated modulus to characterize the pavement layer strength (stiffness) based on
FWD deflection measurements.

Austroads 2016 | page 65

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

For pavement layers not yet placed and hence which cannot be FWD tested, Table 6.3 lists a number of the
recommended design moduli. The allowable traffic loading is calculated from a number of inputs including
the initial and final pavement serviceabilities and the predicted surface curvature index (D0D300).
Table 6.3:

Typical TxDOT design moduli at a temperature of 25 C


Design modulus at 25 C
(MPa)

Material
Stone mastic asphalt using PG 76-XX binder

5860

Size 20 mm dense graded asphalt binder course using PG 76-XX binder

5860

Size 20 mm dense graded asphalt base course using PG 64-22 binder

4480

Granular bases

280480

Lime stabilised subgrade

207310

Source: TxDOT (2011).

As an example, Table 6.4 shows a heavy-duty pavement for a 20 year design traffic loading of 108 ESA. The
total thickness of asphalt is 400 mm. Note that, assuming a heavy vehicle growth rate of 4% over a 40 year
design period, the cumulative loading would be about 3 x 108 ESA.
Table 6.4:
Layer

Example perpetual pavement design using TxDOT method

Material type

Thickness
(mm)

TxDOT design
modulus at 25 C
(MPa)

Stone mastic asphalt using PG 76-22 binder

50

5860

Size 20 mm dense graded asphalt SuperPave mix using PG 76-22 binder

100

5860

Dense graded asphalt base course (Type B) using PG 64-22

250

4480

Lime stabilised subgrade

150

207

Subgrade

Semiinfinite

50

Having calculated the required pavement structure for a 20 year design period, an assessment is then made
as to whether the pavement is long-life (perpetual pavement). The following criteria are used:

Design traffic over 20 years needs to exceed 3 x 107 ESA.


A minimum total asphalt thickness of 300 mm is required.
For the pavement structure designed to carry a 20 year traffic loading, the strains are calculated under a
set of dual tyres loaded to 40 kN and compared to following limiting strain criteria

The tensile strain at the bottom of the composite hot mix asphalt layers is 70 microstrain or less
The vertical compressive strain on top of the subgrade is 200 microstrain or less.
For example, for the heavy-duty pavement structure in Table 6.4, the following strains were calculated
assuming all layers are isotropic and Table 6.3 pavement layer moduli:

tensile strain at the base of the asphalt layer, 42 microstrain


vertical compressive strain at the top of the subgrade, 121 microstrain.
Hence, according to the TxDOT method, the structure in Table 6.4 would be classified as a long-life
(perpetual) pavement. Note that all design calculations are undertaken at a single temperature of 25 C
despite the hot Texan climate.

Austroads 2016 | page 66

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

It was of interest to calculate the allowable traffic loading of the Table 6.4 pavement using the Austroads
Guide method. The following assumptions were made:

Asphalt moduli were calculated for a WMAPT of 34 C (Houston, Texas) and a heavy vehicle design
speed of 80 km/h.

The stone mastic asphalt (SMA) was assumed to be size 14 mm SMA with PG 76-XX binder. A design
modulus of 2660 MPa was calculated from the TMR design moduli (Queensland Department of Transport
and Main Roads 2013) for size 14 mm dense graded asphalt with Class 320 bitumen increased by 30%
consistent with the TxDOT modulus increase between PG64-22 to PG76-XX.

The 100 mm thick dense graded asphalt under the SMA was assumed to be manufactured using a PG76XX binder. A design modulus of 2880 MPa was calculated from the TMR design moduli for size 20 mm
dense graded asphalt with Class 320 bitumen increased by 30%.

The 250 mm thick basecourse was assumed to be a mix with PG 64-22 binder. A design modulus of 2220
MPa was calculated from the TMR design moduli for size 20 mm dense graded asphalt with Class 320
bitumen. A volume of binder of 10.5% was used consistent with TMR values.

Three to six per cent lime stabilised clay subgrade was characterised as a select subgrade material,
sublayered into five sublayers with a maxima vertical and horizontal moduli of 100 MPa and 50 MPa for
the top sublayer.

Subgrade design modulus: vertical 50 MPa and 25 MPa horizontal.


Using the Austroads Guide procedures, the calculated critical strains under an 80 kN Standard Axle were:

tensile strain at the bottom asphalt: 89 microstrain


vertical compressive strain on top of the subgrade: 200 microstrain.
Using the Guide asphalt fatigue relationship, the allowable traffic loading to asphalt fatigue was calculated to
be 3 x 108 Standard Axle Repetitions equivalent or about 2.7 x 108 ESA.
Based on recent communications with researchers in Texas, it is understood that thick asphalt pavements
are not commonly constructed on heavily-trafficked roads, continuously reinforced concrete pavements are
commonly used.

6.3.3 Illinois Department of Transportation


Illinois Department of Transportation (IDOT) pavement design manual (IDOT 2013) includes a
mechanistic-empirical method for the structural design of flexible pavements.
The climate in Illinois is very different from most locations in Australia as the winters are considerably colder.
Figure 6.3 shows the asphalt design moduli used which are similar to the values at these temperatures used
in Australia.

Austroads 2016 | page 67

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 6.3: IDOT asphalt design moduli


6000

PG64-28 & PG 70-28

5500

PG 64-22, PG70-22, PG76-22, PG76-28


5000

Asphalt
design
modulus
(MPa)

4500

4000

3500

3000

2500
22

23

24

25

26

27

28

Asphalt temperature (C)

Source: Adapted from IDOT (2013).

The design method includes an asphalt ELS of 70 microstrain calculated at a mean monthly pavement
temperature for summer (July), which appears to be based on the method proposed by Thompson and
Carpenter (2006). This ELS results in maximum asphalt thickness as illustrated in Figure 6.4. These
thicknesses were conservatively calculated assuming a very low strength subgrade: subgrade design
modulus was 14 MPa (2 kis). No provision is made for adjusting these thicknesses when an improved
subgrade or granular subbase is provided or for pavement constructed on higher strength subgrades.
For Chicago, the largest city in Illinois, the average July pavement temperature at 100 mm depth is about
29 C. At this temperature the asphalt design modulus of mix with PG64-22 binder is about 2850 MPa (413
ksi). For this climate, the IDOT Manual requires a total asphalt of about 370 mm to reduce the strain to 70
microstrain. For the warmer climates in southern Illinois the maximum thicknesses can exceed 400 mm of
asphalt.

Austroads 2016 | page 68

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 6.4: Maximum asphalt thicknesses based on an endurance limit strain of 70 microstrain

Source: IDOT (2013).

6.4

Summary

Only three overseas road agencies have been found to have endurance limit concepts in their pavement
design manual.
The Highways England method provides a limiting design traffic loading (8 x 107 ESA) beyond which no
further increase in asphalt thickness is required provided a stiff foundation is provided, including a minimum
150 mm thickness of cement bound subbase. The concept of a limiting design traffic loading is also used as
an option for the Guide (Section 7.5).
In the USA, only the states of Texas and Illinois consider asphalt fatigue endurance concepts in their pavement
design manuals. In the case of Texas, pavements are firstly designed to be structurally adequate for a 20 year
design period. For this 20 year design structure, the asphalt tensile strain and the subgrade vertical
compressive strain are calculated at a temperature of 25 C. If the calculated asphalt tensile strain is 70
microstrain or less and the subgrade strain is 200 microstrain or less, the pavement is considered to be a
perpetual pavement and hence no structural treatments need to be provided to extend the life beyond 20 years.
The Illinois pavement design manual includes maximum total asphalt thicknesses which vary with average
summer pavement temperatures. The maximum asphalt thicknesses vary from about 350 mm in the colder
northern Illinois to 425 mm in the warmer southern Illinois. These maximum thicknesses were calculated
using an asphalt ELS of 70 microstrain.

Austroads 2016 | page 69

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

7. Options for Austroads Design Guide


7.1

The Need

This research project was principally generated due to concerns by practitioners about the perceived
over-design of thick asphalt pavements in the hotter climates of Australia.
In relation to the design of thick asphalt pavements, the current Austroads design method results in the
design asphalt thickness increasing significantly with temperature. For instance, for a design traffic loading of
3 x 108 ESA, the required thickness of a full depth asphalt pavement is 55 mm greater in Brisbane (WMAPT
= 32 C) than Melbourne (WMAPT = 24 C) as illustrated in Figure 7.1.
Figure 7.1: Example of variation in asphalt thickness with temperature
480

430

24C
27C

380

Total
asphalt
330
thickness
(mm)

32C
37C
42C

280

230

180
1.0E+07

1.0E+08

Design traffic (SAR5)

Discussed below are options for modifying the Guide to address this perceived overdesign for
implementation in the 2017 edition of the Guide.
It should be noted that in selecting an option consideration needs to be given to compatibility with the current
Austroads Guide asphalt fatigue relationship or alternatively a modified version of this relationship for the
2017 edition (Section 7.6).

7.2

Option A: APS-fL Method to Estimate ELS from Asphalt Modulus

As described in Section 5, in this method the ELS criteria is applied at three pavement temperatures:
summer, WMAPT and winter. The long-life asphalt pavement is designed such that asphalt strains predicted
at all three temperatures are below the limiting strains. The 95% confidence level relationship is illustrated in
Figure 7.2.

Austroads 2016 | page 70

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 7.2: Option A relationship to predict ELS from modulus, 95% confidence
200

180

160

140

Endurance
limit
120
strain
(microstrain)
100

80
ELS = 16530 E-0.65
60

40
1000

2000

3000

4000

5000

6000

7000

8000

Asphalt modulus (MPa)

To calculate the strains at the summer and winter temperatures a new process is proposed to determine the
asphalt temperature from the air temperature. As the asphalt temperature varies the depth of asphalt, the
asphalt temperature needs to be calculated for each trial asphalt thickness. This process is different from the
current method in the Guide to estimate WMAPT from air temperature.
In this method the dependence of ELS on modulus was determined from the results of Australian and
University of Illinois laboratory fatigue testing (Section 5.3.2). This modulus dependency was obtained by
testing a wide range of asphalt mix compositions at a temperature of 20 C, including mixes with polymer
modified binders. The effect of temperature on ELS was not investigated. This method assumes that the
effect of temperature on ELS can be estimated from:

the effect of temperature on modulus


ELS dependence on modulus measured by varying asphalt mix composition.
The modulus dependency is very different from that estimated from the NCHRP 9-44A project (see Option B
below) which was derived from testing a limited range of mixes at three temperatures.
The impacts of these APS-fL limiting strains on design thicknesses calculated using the current Austroads
asphalt fatigue relationship and the Guide procedures are illustrated in Appendix A and Figure 7.3. The
example design thicknesses in Figure 7.3 relate to the following pavement composition:

wearing course: 40 mm thickness of size 14 mm dense graded asphalt (Class 320 bitumen)
base course: X mm thickness of size 20 mm dense graded asphalt (Class 600 bitumen)
fatigue layer: 70 mm thickness of size 20 mm dense graded asphalt (Class 320 bitumen), volume of
binder 10.5%

300 mm thickness of unbound granular subbase


subgrade with design modulus of 50 MPa.
The adoption of Option A results in similar maximum asphalt thicknesses across a range of asphalt design
modulus and pavement temperatures. Thus the method has more impact on design thicknesses the higher
the pavement temperature.

Austroads 2016 | page 71

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

It should be noted that, for the 2017 edition of the Guide, consideration is being given to a new method of
assessing the damage due to axle loads (Austroads 2015b). The new method is based on calculating the
strains of each isolated axle load. If this method is adopted, the fatigue lives of thick asphalt layers will
increase by about 60%. Applying this factor to the fatigue lives in Figure 7.3, for WMAPT of 24 C or less
Option A will not affect design thicknesses up to a design traffic loading of 4 x 108 SAR5. For Sydney and
Adelaide (WMAPT = 27 C), Option A will not affect thicknesses for loading up to 3 x 108 SAR5. The effect is
more significant for Brisbane and warmer climates.
Figure 7.3: Examples of the impact of Option A on design thicknesses calculated using the current Guide
400
380
360
340

21C
24C
27C
32C
37C
42C

320

Total 300
asphalt
thickness
280
(mm)
260
240
220
200
180
1.0E+07

1.0E+08

Design traffic (SAR5)

The use of modulus to predict ELS is simple and versatile as it could be used for a wide range of asphalt
mixes. However, changes in mix volumetrics are not adequately addressed. For instance, an increase in air
voids decreases the modulus and hence leads to an increase in ELS, contrary to the findings of fatigue
testing. An example of this is shown in Figure 7.4, in which the modulus values were adjusted for air voids
using the equation 8 in the Guide (Austroads 2012). In this example, the use of Option A results in the ELS
at 7% air voids being about 9 microstrain higher than at 5% air voids. Hence Option A results in a 10 mm
reduction in maximum asphalt thickness for mixes with 7% air voids compared to 5% air voids rather than 5
10 mm increase provided by Option C discussed below.
Note that, to partly address this issue, Sullivan et al. (2015) recommended that LLAP include a low air voids
(< 4%) fatigue layer. The limitations of the use of modulus as a surrogate for binder rheology, air voids,
bitumen content, and temperature was acknowledged by Witczak et al. (2013) as discussed in Section 3.8.

Austroads 2016 | page 72

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 7.4: Example of effect of air voids on ELS


Effect of voids content on endurance limit strain at a temperature of 32C
10
8
Option A
6

Option C

4
2

ELS
adjustment 0
3.5
(microstrain)

4.5

5.5

6.5

-2
-4
-6
-8
-10

Voids content (%)

For use in the Guide, Option A could be modified as follows:

simplified to only requiring the strain assessment to be undertaken at the WMAPT


modified to include mix volumetrics as provided in Equation 5 or limits placed on maximum air voids and
minimum bitumen contents.
However, in light of Options B and C discussed below some doubt remains as to how well the relationship
reflects the variation in ELS with temperature: the effect of temperature on ELS was not included in the
laboratory testing that forms the basis of modulus dependency.
To illustrate this issue Figure 7.5 shows an example of how Option A ELS vary with asphalt temperature. At
each temperature the asphalt modulus was calculated consistent with the Guide (Austroads 2012) and using
this modulus the ELS calculated using the equation in Figure 7.2. The variation with temperature greatly
exceeds that of Option B and Option C ELS discussed below. For instance, for Option A the ELS increases
about 50% between 24 C (Melbourne) and 32 C (Brisbane) whereas for Option C the increase is only
about 20%. As pavement temperatures vary widely throughout Australia it is important that the ELS are
appropriate across a range of climates.
It should be noted that the ELS modulus dependency (modulus exponent 0.65) proposed by Sullivan et al.
(2015) reflects how their measured laboratory fatigue lives varied with asphalt modulus. This laboratory
fatigue data has a very different dependence on modulus than the Austroads Guide fatigue relationship
(modulus exponent 0.36). This incompatibility of modulus dependency is reflected in Figure 7.3 which uses
the Sullivan et al. (2015) modulus dependency to calculate the ELS and the Austroads Guide modulus
dependency to calculate fatigue life. Although the laboratory ELS (Section 5.3.2) were based on the
supposition that the endurance limit was reached at the same number of loading cycles irrespective of mix
modulus, this is not reflected in Figure 7.3 when the Option A ELS is used with the Guide procedures,
including the Guide fatigue relationship. Section 7.6 discusses modifications to the Guide fatigue relationship
to allow for increase in healing with temperature and gives an example of such a modification with Option A
ELS.

Austroads 2016 | page 73

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 7.5: Comparison of ELS variations with asphalt temperature


160
150

Option A

140

Option B

130

Option C

120

Endurance 110
limit
strain
100
(microstrain)
90
80
70
60
50
20

22

24

26

28

30

32

34

36

38

40

Asphalt temperature (C)

7.3

Option B: NCHRP 9-44A Method to Calculate ELS from Modulus

Like Option A, this method estimates ELS from asphalt design modulus. However, the ELS dependence on
modulus is different from Option A being based on the findings of NCHRP 9-44A project (Witczak et al. 2013)
at the Arizona State University (Section 3) rather than University of Illinois research findings (Section 2).
Figure 7.6 shows the Option B relationship.
Figure 7.6: Comparison of Option A and Option B relationships to predict ELS from modulus
180
170
160
150
140
130
120

Endurance
limit strain 110
(microstrain)

Option A
ELS = 16530E-0.65

100
90
80
70

Option B
ELS = 604E-0.254

60
50
40
1000

2000

3000

4000

5000

6000

7000

8000

Asphalt modulus (MPa)

Austroads 2016 | page 74

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

As described in Appendix B.2 this relationship was developed from the laboratory ELS obtained by Witczak
et al. (2013) adjusted to an in-service ELS by taking into consideration the performance of UK long-life
asphalt pavements. The ELS dependence on modulus is significantly less than Option A (Figure 7.2). Note
that the Option B modulus dependency was estimated from testing an asphalt mix at combinations of binder
contents and air voids and three binder types. In all 12 mixes were tested, each at three temperatures
(4.4 C, 21.1 C, 37.8 C). This range of asphalt mixes was significantly less than used in the University of
Illinois research (120 mixes, Section 2.3). However, the testing included the effect on temperature which is
an important factor in modulus variations in applying the Guide throughout Australia.
The effect of considering the Option B ELS in the Guide procedures is illustrated in Figure 7.7 for the same
example pavements that were used in the Option A evaluation.
Figure 7.7: Examples of the impact of Option B on design thicknesses
480
460

21C
24C

440

27C

420

32C

400

37C
42C

380

Total 360
asphalt
340
thickness
(mm) 320
300
280
260
240
220
200
1.0E+07

1.0E+08

Design traffic (SAR5)

In this case, the traffic loadings at which the thickness plateau increases as the modulus reduces is in
contrast to Option A (Figure 7.3). Note that the impact on design thicknesses is greatest at lower
temperatures. Hence Option B does not adequately address the perceived over-design at high temperatures.
As mentioned above, for the 2017 edition of the Guide, consideration is being given to a new method of
assessing the damage due to axle loads. If this method is adopted, the fatigue lives of thick (> 300 mm)
asphalt layers will increase by about 60%. Applying this factor to the fatigue lives in Figure 7.7, Option B
would not affect design thicknesses for Brisbane (WMAPT = 32 C and warmer climates, again reinforcing
the conclusion that this option does not address concerns about over-design at high temperatures.
Note that because this ELS relationship does not directly consider mix volumetrics, differences between
Australian and UK asphalt base course volumetrics was not considered fully in the calibration
(Appendix B.2.2). This is one of the reasons Option B is more conservative than Option C discussed in
Section 7.4.

Austroads 2016 | page 75

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

7.4

Option C: NCHRP 9-44A Estimating ELS from Temperature and


Mix Volumetrics

Unlike Options A and B, this method estimates ELS from asphalt temperatures and mix volumetrics based
on the beam fatigue testing undertaken in NCHRP project 9-44A (Witczak et al. 2013). Its major limitation is
that the laboratory testing results used to derive the method was limited to one mix at four combinations of
bitumen content and air voids and one binder type (i.e. PG 64-22). However the testing did include laboratory
testing at three temperatures (4.4 C, 21.1 C, 37.8 C).
The development of the in-service ELS relationship is described in Appendix B.2. The relationship was
developed from the laboratory ELS obtained by Witczak et al. (2013) adjusted to an in-service ELS by taking
into consideration the performance of UK long-life asphalt pavements.
Figure 7.8 shows the Option C ELS values for bitumen contents of 4.2%, 4.7% and 5.2% for mixes with 5%
air voids and Figure 7.9 provides ELS adjustments for air void variations.
Figure 7.8: Option C relationship to predict ELS from temperature and bitumen content at 5% air voids

5 % air voids

140
130

4.2% bitumen
4.7% bitumen

120

5.2% bitumen

110

Endurance 100
limit
strain
(microstrain) 90
80
70
60
50
14

16

18

20

22

24

26

28

30

32

34

36

38

40

Asphalt temperature (C)

Austroads 2016 | page 76

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure 7.9: Option C adjustment of ELS for air voids


10
8

21C
24C
27C
32C
37C

6
4
2
ELS
adjustment 0
(microstrain) 3.5

4.5

5.5

6.5

-2
-4
-6
-8
-10
Air voids (%)

The effects of considering the Option C endurance strains in the Guide are illustrated in Figure 7.10 for the
same example pavements as used for Options A and B. The assumed bitumen and air voids contents of the
basecourse mix were 4.6% and 5.0% respectively.
In this case, the traffic loadings at which asphalt thicknesses plateau are similar across a range of asphalt
moduli and pavement temperatures in contrast to the Option A (Figure 7.3) and Option B (Figure 7.6). The
maximum required asphalt thickness is reached at about 1.5 x 108 Standard Axle Repetitions or just above
108 ESA. Note that Option C has a similar impact on design thicknesses across a range of climates.
Figure 7.10:
440
420
400
380
360

Examples of the impact of Option C on design thicknesses


Example Option C with Vb=10.5%, bitumen content 4.6% and air voids 5%
21C
24C
27C
32C
37C
42C

340

Total
asphalt
320
thickness
(mm) 300
280
260
240
220
200
1.0E+07

1.0E+08

Design traffic (SAR5)

Austroads 2016 | page 77

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

As mentioned above, for the 2017 edition of the Guide, consideration is being given to a new method of
assessing the damage due to axle loads (Austroads 2015b). If this method is adopted, the fatigue lives of
thick asphalt layers will increase by about 60%. Applying this factor to the fatigue lives in Figure 7.10, the
impact of Option C on design thicknesses will be reduced.
The approach could be readily implemented in the Guide. However, a major limitation of the method is that it
is based around the performance of UK mixes made using penetration grade 40/60 bitumen which is similar
to the penetration of a Class 320 bitumen.
To address this limitation Option C would benefit by the inclusion of an adjustment for binder type. The
University of Illinois research (Equation 5) may be useful in developing an adjustment for binder type based
on its effect on modulus. Preliminary calculations indicate that if Class 600 bitumen is used instead of Class
320 bitumen, the ELS would reduce by about 15%. Alternatively, it could be assumed, consistent with the UK
design charts (Figure 6.1) that the traffic loading at which the maximum thickness is reached is the same for
all commonly used binders.

7.5

Option D: Limiting Design Traffic Loading

As shown in Figure 6.1, the UK pavement design procedures include the same limiting design traffic loading
(8 x 107 ESA) for all asphalt mixes types and design moduli.
Considering the Option C thickness impacts (Figure 7.10), Option D does not use an ELS but simply includes
a limiting design traffic loading for structural design for asphalt fatigue in a similar manner to the UK method.
For instance a maximum design traffic loading in terms of asphalt fatigue of say 2 x 108 SAR5 could be used
with the current Austroads Guide. This option is simpler than Options A, B and C.
It may be necessary to supplement the loading limit with limits on the asphalt mix properties such as dense
graded asphalt with conventional binders, maximum air voids (say 5%) and minimum binder content (say
5%).
As mentioned above, for the 2017 edition of the Guide consideration is being given to including a new
method of assessing the damage due to axle loads (Austroads 2015b). In this case the design traffic is
expressed in terms of cumulative number of heavy vehicle axle groups (NHVAG). If this method is adopted, the
fatigue lives of thick asphalt layers will increase by about 60%. In this case the maximum NHVAG would be
used in assessing asphalt fatigue rather than maximum Standard Axle Repetitions.
After discussions with Austroads Pavements Structures Group, this option was further developed as
discussed in Section 7.7.

7.6

Option E: Modify Fatigue Relationship to Allow for Healing at


Elevated Temperatures

If it is agreed that the key need is limited to the perceived overdesign of asphalt thickness at high
temperatures, the most direct means of addressing this is to modify the current Austroads fatigue
relationship to reduce the rate at which asphalt thicknesses increase with temperature.
The current Austroads fatigue relationship was developed from the Shell laboratory fatigue relationship. This
fatigue relationship was developed from laboratory testing at temperatures mostly between 5 C to 30 C
without specific allowance for the beneficial effect of rest periods between load applications (Bonnaure,
Gravois & Urdon 1980), particularly the increase in fatigue life due to crack healing at elevated temperatures.
Note that this modulus dependency was determined by varying the test temperature rather than mix
composition.

Austroads 2016 | page 78

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

In this option, the focus of the project would now be directed to reviewing available information to develop a
fatigue relationship that separately allows for:

a shift factor (SF) relating mean laboratory fatigue life to mean in-service fatigue life, taking account of the
differences between laboratory test conditions and conditions applying to the in-service pavement,
including the increase in healing with WMAPT

a reliability factor (RF) relating mean in-service fatigue life to the in-service life predicted at a desired
project reliability, taking into account those factors (e.g. materials, construction and environment
variabilities) discussed in Section 2 of the Guide.
Equation 14 shows a possible form of the revised asphalt fatigue relationship. Table 7.1 and Table 7.2 are
examples of possible SF and RF values. In determining these example SF and RF values, it was assumed
that the current Guide processes adequately addresses healing for climates with a WMAPT of 27 C and
less and hence the fatigue lives are the same as the current Guide. For WMAPTs above 27 C the SF
increases results in higher fatigue lives to address the perceived over-design of thicknesses. The SFs above
27 C were estimated by adjusting the SF at 27 C (SF = 6) using the relative laboratory fatigue lives of
Bonnaure, Gravois and Urdon (1980).

SF 6918(0.856 Vb + 1.08)
N =

RF
E 0.36

14

where

allowable number of repetitions of the load

tensile strain produced by the load (microstrain)

Vb

percentage by volume of bitumen in the asphalt (%)

asphalt modulus (MPa)

SF

shift factor between laboratory and in-service fatigue lives (Table 7.1)

RF

reliability factor for asphalt fatigue (Table 7.2)

Table 7.1:

Example shift factors (SF) for asphalt fatigue


Weighted Mean Annual Pavement Temperature (C)

27

2830

3133

3436

3739

40

6.0

7.0

9.55

12

16

20

Table 7.2:

Example reliability factors (RF) for asphalt fatigue


Desired project reliability

50%

80%

85%

90%

95%

97.5%

1.0

2.4

3.0

3.9

6.0

9.0

Austroads 2016 | page 79

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

This option has the effect of reducing asphalt thicknesses across a range of traffic loadings at WMAPT in
excess of 27 C, whereas the adoption of Options A to D would restrict the impact to the most heavily
trafficked roads. Using the example SF and RF factors, the impact is illustrated in Figure 7.11. Consideration
could also be given to limiting the WMAPT to a maximum say of 35 C.
Figure 7.11:

Examples of the impact of Option E on design thicknesses


Example Option E

400
21C
380
360
340

24C
27C
32C
37C

320

Total
asphalt
300
thickness
(mm)
280
260
240
220
200
1.0E+07

1.0E+08

Design traffic (SAR5)

Laboratory fatigue testing with rest periods over a range of temperatures currently being undertaken for TMR
and MRWA may assist in this approach.
For heavy-duty full-depth asphalt pavements in Brisbane (WMAPT = 32 C) the combined effect of this
modified fatigue relationship together with the proposed changes to a strain-based method of assessing the
damage due to axle group loads (Austroads 2015b), results in the predicted fatigue lives increasing by about
a factor of 2.5. For a design traffic loading of 4 x 108 SAR5, the combined impact on the pavement
configuration used in the examples (e.g. Figure 7.11) would be about a 50 mm reduction in asphalt
thickness, similar to the Option A impact.

7.7

PSWG Recommendation for the Austroads Guide

In June 2016, the above options were discussed at meetings of the Austroads Pavement Structures Working
Group and Pavements Task Force. It was agreed that although Option E is the preferred approach there is
currently insufficient information to revise the asphalt fatigue relationship to more appropriately reflect the
increase in healing with temperature. As an interim measure for the 2017 edition of the Guide, it was agreed
to simply place limits on the design traffic loadings used in the asphalt damage calculations.
In consultations with road agencies, the Table 7.3 upper limits of the design traffic in ESA were agreed for
use with the revised design procedures in the 2017 edition of Part 2. In the asphalt damage calculations
these ESA limits are converted to limits of the cumulative number of heavy vehicle axle groups (HVAG) using
Equation 15.

Austroads 2016 | page 80

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table 7.3:

Suggested upper limits on design traffic for asphalt fatigue


25C

WMAPT
Design traffic loading limit (ESA)

where

4x

2634C

108

2x

35C

108

108

(/)

15

= upper limit on the design traffic expressed as heavy vehicle axle groups (HVAG)
for use in the asphalt fatigue damage calculations
= upper limit on the design traffic expressed as Equivalent Standard Axles (ESA)
for use in the asphalt fatigue damage calculations
= average number of ESA per HVAG from the project traffic load distribution

Appendix C contains the proposed text for inclusion in the 2017 edition of Part 2.

Austroads 2016 | page 81

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

8. Summary
The asphalt pavement design process of the Guide to Pavement Technology Part 2: Pavement Structural
Design (Austroads 2012) utilises a mechanistic-empirical procedure to predict the allowable traffic repetitions
until asphalt fatigue occurs. In practice, traffic loading is a design input in an iterative process and increases
in traffic loadings necessitate increases in asphalt design thickness to ensure fatigue requirements are met.
With the current fatigue-based design approach, asphalt design thickness will continue to increase in the
future due to ever increasing traffic loadings and the low asphalt moduli in some areas of Australia (due to
warm climate). In some cases heavy duty full-depth asphalt pavements have been designed with total
asphalt thickness exceeding 400 mm.
Researchers have postulated that an asphalt fatigue endurance limit exists where applied strain is sufficiently
low that fatigue damage no longer accumulates to the extent that deep structural treatments are required
due to fatigue cracking. The premise is that the rate of asphalt healing exceeds the rate of damage
accumulation, being fatigue cracking at the base of the asphalt pavement. The corresponding thickness
would provide a limit for the pavement where additional asphalt thickness provides no additional benefit to
fatigue life.
Currently three overseas road agencies have adopted asphalt fatigue endurance concepts in their pavement
design manuals: Highway England, Texas Department of Transportation (TxDOT) and Illinois Department of
Transportation (IDOT). In the UK, the design method provides for long-life asphalt pavements by placing
minimum requirements on foundation support and a limiting design traffic loading (8 x 107 ESA). TxDOT and
IDOT methods include consideration of asphalt endurance limit strains (ELS). The AASHTO (2015) design
guide also recognise the existence of ELS does not provide values for design.
In terms of quantifying how the properties of asphalt mixes and temperatures affect ELS, extensive research
projects undertaken by the University of Illinois and the NCHRP 9-44A project undertaken by Arizona State
University are described. The report describes and compares the laboratory ELS relationships derived from
these projects. Of particular note are the very different rates of variation of ELS with asphalt modulus.
To derive an in-service ELS for use in the Guide, data is required on the long term performance of thick
asphalt pavements. The most comprehensive field investigation on performance of thick asphalt pavements
was undertaken in the United Kingdom by Nunn et al. (1997), including estimating the residual life of
pavements by testing field cores. They concluded that for the UK climate, materials and traffic loading a total
270 mm thickness of asphalt on a firm foundation was sufficient for a long and indeterminate life. This
research lead to the inclusion of the fatigue endurance concepts in the Highways England design method.
The performance of Australian thick asphalt pavements has also been investigated a number of times over
the last 20 years but not as comprehensively as the UK research. While these studies have reported good
pavement performance, the data is insufficient to reliably conclude these pavements will never need deep
structural treatments due to fatigue of the thick asphalt layers. A similar conclusion has also been reached
from an analysis of USA long-term pavement performance data. It was concluded that the UK performance
data is of most value in developing endurance limit options for the Guide.
The Australian Asphalt Pavement Association has recently completed a research project titled Asphalt
Pavement Solutions for Life (APS-fL) which proposes a procedure to predict endurance limit strains for mix
modulus. In this case, the method took into consideration a laboratory ELS relationship derived from fatigue
testing Illinois and Australian mixes using an analysis method developed from University of Illinois research.
Considering performance data from accelerated loading trials and in-service performance of Australian and
UK thick asphalt pavements, an in-service ELS relationship was developed. The development of the APS-fL
method is described and its impact on design thicknesses is illustrated by example designs.

Austroads 2016 | page 82

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Using the laboratory ELS relationships derived in the NCHRP 9-44A project and taking account of the UK
studies of long-life asphalt pavements, an alternative procedure for determining in-service ELS from asphalt
modulus was developed. In this method, the ELS vary less with modulus than in the APS-fL project as a
consequence of different laboratory ELS modulus dependencies.
The use of asphalt modulus to predict ELS has the advantage of simplicity. However, the limitations of using
modulus as a surrogate for the many factors that can influence ELS, including temperature, binder type,
binder content and air voids need to be considered. Of particular note is that use of modulus to determine
ELS is misleading in terms of the influence of air voids: it can result in a reduction in required asphalt
thickness with increasing air voids. If a modulus relationship is adopted by Austroads, supplementary
procedures will need to be provided to limit this air voids effect.
As an alternative to a modulus-based model, a procedure was developed to determine ELS from asphalt
temperature and mix volumetrics based on the NCHRP 9-44A laboratory ELS relationship.
The development of these two options is described and their impact on design thicknesses is illustrated by
example designs.
A fourth option is to provide an upper limit of design traffic loading rather than an ELS. Like the Highways
England maximum design traffic loading this method has the advantage of simplicity.
Lastly a fifth option was mentioned which would address the perceived over-design at high temperatures by
varying the laboratory-to-field shift factors used in the Austroads asphalt fatigue relationship.
In June 2016, the above options were discussed at meetings of the Austroads Pavement Structures Working
Group and Pavements Task Force. It was agreed that although Option E is the preferred approach there is
currently insufficient information to revise the asphalt fatigue relationship to more appropriately reflect the
increase in healing with temperature. As an interim measure for the 2017 edition of the Guide, it was agreed
to simply place limits on the design traffic loadings used in the asphalt damage calculations.

Austroads 2016 | page 83

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

References
AAPA 2000, Heavy duty flexible pavements performance (HeavyFlex), project report 3 PD 1.3, Australian
Asphalt Pavement Association, Kew, Vic.
AAPA 2004, Heavy duty asphalt pavement performance study: stage 2, project report 3 PD 1.6, Australian
Asphalt Pavement Association, Kew, Vic.
AASHTO 2003, Test standard for determining the fatigue life of compacted Hot-Mix Asphalt (HMA) subjected
to repeated flexural bending, test method T321-03, American Association of State Highway and
Transportation Officials, Washington, DC, USA.
AASHTO 2015, Mechanistic-empirical pavement design guide: a manual of practice, 2nd edn, American
Association of State Highway and Transportation Officials, Washington, DC, USA.
Austroads 2009, Asphalt fatigue endurance limit, technical report, AP-T131-09, Austroads, Sydney, NSW.
Austroads 2012, Guide to pavement technology part 2: pavement structural design, AGPT02-12, Austroads,
Sydney, NSW.
Austroads 2015a, Improved design procedures for asphalt pavements: outcome for year 2 of 3, technical
report, AP-T296-15, Austroads, Sydney, NSW.
Austroads 2015b, The influence of multiple axle loads on flexible pavement design, AP-R486-15, Austroads,
Sydney, NSW.
Bonnaure, F, Gravois, A & Urdon, J 1980, A new method for predicting the fatigue life of bituminous mixes,
Association of Asphalt Paving Technologists proceedings, Association of Asphalt Paving Technologists,
Minnesota, USA, pp. 499-529.
British Standards Institution 2015, Guidance on the use of BS EN13108-1 bituminous mixtures: materials
specifications, PD 6691:2015, BSI, London, UK.
Carpenter, SH, Ghuzlan, KA & Shen, S 2003, Fatigue endurance limit for highway and airport pavements,
Transportation Research Record, no. 1832, pp. 131-8.
Carpenter, SH & Shen, S 2006, Dissipated energy approach to study hot-mix asphalt healing in fatigue,
Transportation Research Record, no. 1970, pp. 178-85.
Carpenter, SH & Shen, S 2009, Effect of mixture variables on the fatigue endurance limit for perpetual
pavement design, Proceedings of the international conference on perpetual pavement, Columbus, Ohio,
Ohio University, Athens, OH, USA, 13 pp.
Highways England 1994, Structural assessment methods: volume 7: section 3: part 2, HD 25/94, Highways
Agency, Department for Transport, UK, (withdrawn).
Highways England 2006, Design manual for roads and bridges: volume 7: section 2: part 3: pavement
design, HD 26/06, Highways Agency, Department for Transport, UK.
Highways England 2009, Design guidance for road pavement foundations (draft HD25), interim advice note
73/06, revision 1 (2009), Highways Agency, Department for Transport, UK.

Austroads 2016 | page 84

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Jones, CR, Chaddock, BJC, Bradford, J, Read, C & Rolt, J 2006, Validation monitoring of design and
assessment methods (3), package 1, final report, unpublished project report UPR/IE/172/06, Transport
Research Laboratory (TRL), Crowthorne, UK.
Illinois Department of Transportation 2013, Bureau of Design and Environment manual: part VI: other
highway design elements: chapter 54: pavement design, IDOT Division of Highways, Illinois, USA.
Nunn, N 1997, Long-life flexible roads, International conference on asphalt pavements, 8th, 1997, Seattle,
Washington, USA, University of Washington, Seattle, USA, vol. 1, pp. 3-16.
Nunn, M 2004, Development of a more versatile approach to flexible and flexible composite pavement
design, report 615, Transport Research Laboratory (TRL), Crowthorne, UK.
Nunn, M, Brown, A, Weston, D & Nicholls, JC 1997, Design of long-life flexible pavements for heavy traffic,
report 250, Transport Research Laboratory (TRL), Crowthorne, UK.
Nunn, M & Ferne, BW 2001, Design and assessment of long-life flexible pavements, Transportation
Research Circular, no. 503, pp. 32-49.
Nunn, ME & Smith, T 1997, Road trials of high modulus base for heavily trafficked roads, report 231,
Transport Research Laboratory (TRL), Crowthorne, UK.
Paul, R 1997, Developments in the design and construction of deep strength asphalt pavements in Victoria
over the past 25 years, AAPA international flexible pavements conference, 10th, 1997, Perth, Western
Australia, Australian Asphalt Pavement Association, Kew, Vic, vol. 2, 21 pp.
Pronk, AC, Poot, MR, Jacobs, MMJ & Gelpke, RF 2010, Haversine fatigue testing in controlled deflection
mode: is it really possible?, Transportation Research Board annual meeting, 89th, Washington, DC, TRB,
Washington, DC, USA, paper no. 10-0485.
Queensland Department of Transport and Main Roads 2013, Pavement design supplement: supplement to
Part 2: Pavement Structural Design of the Austroads Guide to Pavement Technology, QDMR, Brisbane,
Qld.
Roberts, J 2009, Modelling of asphalt temperature and strength variations with time of day and depth in
pavement, unpublished, ARRB Group, Vermont South, Vic.
Sharp, KG, Mangan, DPA, Armstrong, PJ & Tepper, SB 2000, Performance of heavy duty flexible
pavements, AAPA international flexible pavements conference, 11th, 2000, Sydney, New South Wales,
Australian Asphalt Pavement Association, Kew, Vic, pp. 73-103.
Shen, S 2006, Dissipated energy concepts for HMA performance: fatigue and healing, PhD thesis, Civil
Engineering Department, University of Illinois, Urbana-Champaign, USA.
Shen, S & Carpenter, SH 2005, Application of the dissipated energy concept in fatigue endurance limit
testing, Transportation Research Record, no. 1929, pp. 165-73.
Shen, S & Carpenter, SH 2007, Development of an asphalt fatigue model based on energy principles,
Journal of the Association of Asphalt Paving Technologists, vol. 76, pp. 525 -73.
Sullivan, B, Youdale, G, Rickards, I & Yousefdoost, S 2015, Technical background to the development of the
AAPA long life asphalt pavement design procedure, draft Australian Asphalt Pavement Association
report.
Texas Department of Transportation 2011, Pavement design guide, TxDOT, Austin, TX, USA.

Austroads 2016 | page 85

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Thompson, MR & Carpenter, SH 2006, Considering hot-mix-asphalt fatigue endurance limit in full-depth
mechanistic-empirical pavement design, International conference on perpetual pavement, Columbus,
Ohio, USA, Ohio University, Athens, Ohio, USA.
Timm, DH, West, R, Priest, A, Powell, B, Selvaraj, I, Zhang, J & Brown, R 2006, Phase II NCAT test track
results, NCAT report 06-05, National Center for Asphalt Technology, Auburn, USA.
Tran, NE, Robins, MM, Timm, DH, Willis, RJ & Rodezno, C 2015, Refined limiting strain criteria and
approximate ranges of maximum thicknesses for designing long-life asphalt pavements, NCAT report 1505, National Center for Asphalt Technology, Auburn, USA.
Tsoumbanos, B 2006, Top-down cracking in Melbourne pavements, ARRB conference, 22nd, 2006,
Canberra, ACT, Australia, ARRB Group, Vermont South, Vic, 20 pp.
VicRoads 2007, Top down cracking in Melbourne pavements (maintenance of thick asphalt pavements),
technical note 82, VicRoads, Kew, Vic.
Von Quintus, HL 2006, Application of the endurance limit premise in mechanistic-empirical based pavement
design procedures, International conference on perpetual pavement, Columbus, Ohio, USA, Ohio
University, Athens, Ohio, USA.
Walubita, LF, Liu, W & Scullion, T 2010, The Texas perpetual pavements: experience overview and way
forward, report 0-4822-3, Texas Transportation Institute, College Station, TX, USA.
Walubita, LF & Scullion, T 2010, Texas perpetual pavements: new design guidelines, Texas Transportation
Institute, College Station, TX, USA.
Willis, R & Timm, D 2009, Field-based strain thresholds for flexible perpetual pavement design, NCAT report
09-09, National Center for Asphalt Technology, Auburn, USA.
Willis, R & Timm, D 2010, Development of stochastic perpetual pavement design criteria, Journal of the
Association of Asphalt Paving Technologists, vol. 79, pp. 561-96.
Willis, R, Timm, D, West, R, Powell, B, Robins, M, Taylor, A, Smit, A, Tran, N, Heitzman, M & Bianchini, A
2009, Phase III NCAT test track findings, NCAT report 09-08, National Center for Asphalt Technology,
Auburn, USA.
Witczak, M, Mamlouk, M, Souliman, M & Zeiada, W 2013, Laboratory validation of an endurance limit for
asphalt pavements, NCHRP report 762, Transportation Research Board, Washington, DC, USA.
Zeiada, WA, Underwood, BS & Kaloush, KE 2016, Impact of asphalt concrete fatigue endurance limit
definition on pavement performance prediction, International Journal of Pavement Engineering, DOI:
10.1080/10298436.2015.1127372.

Austroads 2016 | page 86

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Appendix A Asphalt Pavement Solutions for Life


Design Examples
A.1

Introduction

In 2011 the Australian Asphalt Pavement Association (AAPA) commenced the Asphalt Pavement Solutions
for Life (APS-fL) project. This project was initiated to address the concerns of their clients, consultants and
industry that current pavement design procedures were producing overly conservative asphalt thickness
requirements (Sullivan et al. 2015).
The project included the development of a new method of determining the endurance limit strains (ELS) for
asphalt strain below which fatigue damage does not accumulate to the extent that macrocracking develops.
(Note that Sullivan et al. (2015) used the general term Fatigue Endurance Limit (FEL) instead of the more
specific term ELS.)
In October 2015 a draft Technical Report (Sullivan et al. 2015) was submitted to the Austroads Pavement
Task Force for their consideration in the forthcoming revision of the Guide to Pavement Technology Part 2:
Pavement Structural Design (Austroads 2012).
Task 1 of Austroads project TT2044 was to prepare design examples using the APS-fL method. These are
discussed below.

A.2

Design Inputs

A.2.1 Introduction
Chapter 9 of Sullivan et.al. (2015) describes the proposed method for inclusion in the Guide
(Austroads 2012). Critical inputs to the design examples are summarised below.

A.2.2 Working Platform


Section 9.2.1 describes the minimum support condition under the asphalt layers for a pavement to be
considered a long life asphalt pavement:
As compaction of asphalt layers on subgrades with stiffness of less than 100 MPa
(CBR 10%), is difficult, for a subgrade with a stiffness of less than a 100 MPa, it is
recommended that a working platform be established to achieve a supporting layer of
sufficient quality to be assigned a design modulus of 100 MPa (or greater) in CIRCLY should
be placed directly below the bottom asphalt layer to support this bottom layer and assist with
achieving compaction.
The thickness of the layer with a modulus of 100 MPa is not clear from this description. The design examples
consider three subgrade design CBR (3%, 5% and 15%) each covered with 250 mm thickness of unbound
granular material. In accordance with Table 6.5 of the Guide, the top granular sublayer was limited to a
maximum vertical modulus of 210 MPa. The elastic characterisation of the semi-infinite subgrade and the
five granular sublayers is given in Table A 1. For subgrade CBR = 3%, although the top sublayer has a
design modulus exceeding 100 MPa there may be an argument for requiring at least 100 mm thickness of
material with a design modulus of 100 MPa or more under the asphalt.

Austroads 2016 | page 87

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table A 1: Elastic characterisation of subgrade and unbound granular layers


Thickness
(mm)

Vertical modulus
(MPa)

Horizontal modulus
(MPa)

Poissons ratio

Granular sublayer 1

50

120.0

60.0

0.35

Granular sublayer 2

50

90.9

45.5

0.35

Granular sublayer 3

50

68.9

34.5

0.35

Granular sublayer 4

50

52.2

26.1

0.35

Granular sublayer 5

50

39.6

19.8

0.35

Semi-infinite

30

15

0.45

Granular sublayer 1

50

200.0

100.0

0.35

Granular sublayer 2

50

151.6

75.8

0.35

Granular sublayer 3

50

114.9

57.4

0.35

Granular sublayer 4

50

87.1

43.5

0.35

50

66.0

33.0

0.35

Semi-infinite

50

25

0.45

Material
Subgrade design CBR=3%

Subgrade
Subgrade design CBR=5%

Granular sublayer 5
Subgrade

Subgrade design CBR=15%


Granular sublayer 1

50

210.0

105.0

0.35

Granular sublayer 2

50

196.3

98.2

0.35

Granular sublayer 3

50

183.6

91.8

0.35

Granular sublayer 4

50

171.6

85.8

0.35

Granular sublayer 5

50

160.4

80.2

0.35

Semi-infinite

150

75

0.35

Subgrade

A.2.3 Asphalt Mix Types


According to Sullivan et al. (2015) the LLAP concept utilises a three layered asphalt system:

a wearing course and levelling course (surfacing layers), which should be durable and rut resistant
intermediate layer(s), which should be rut resistant and be the main structural layers
fatigue layer, which should be a fatigue resistant low air void mix.
A 50 mm thick size 14 mm Class 320 bitumen dense graded asphalt wearing course was used in all design
examples.
The design examples included two asphalt mixes for the basecourse:

either a size 20 mm Class 320 bitumen dense graded asphalt, or


a size 20 mm Class 600 bitumen dense graded asphalt.
Sullivan et al. (2015) recommend that:
the LLAP incorporate a low air void (<4%) asphalt fatigue layer, in the bottom 60 mm of the
pavement. The low void layer is usually achieved by the use of higher binder content in the
bottom layer of the DG 20 asphalt mix. This lower level asphalt layer should consist of no
more than 30% RAP, if more than 30% RAP it is recommended that the stiffness-FEL
relationship of the mix be confirmed to fall on or above the standard stiffness-FEL mix curve.
However as dynamic compressive moduli for such a mix is not provided in Sullivan et al. (2015), in each
design example the asphalt mix for fatigue layer was the same as the intermediate layer mix.

Austroads 2016 | page 88

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

A.2.4 Temperatures
As the proposed endurance strains varies with asphalt modulus (see Appendix A.2.6), the design examples
were undertaken for a range climates, namely Brisbane, Perth, Sydney, Adelaide and Melbourne.
For each climate the APS-fL method requires strain calculations at three asphalt temperatures:

The effective asphalt layer temperature at midday in the hottest month


The Weighted Mean Annual Pavement Temperature (WMAPT) as used in the Guide
The effective asphalt layer temperature early morning in the coolest winter month.
These asphalt temperatures were estimate from the Bureau of Metrology climate average maximum and
minimum air temperatures. Table A 2 lists the mean air temperatures used to estimate the asphalt
temperatures and the effective asphalt temperatures calculated using the APS-fL method for a total asphalt
thickness of 330 mm.
Table A 2: Air temperatures and effective asphalt temperatures

Location

Brisbane
(station 040214)

Perth
(station 009034)

Adelaide
(station 023000)

Sydney
(station 066062)

Melbourne
(station 086071)

Temperature
characteristic
Summer maximum
WMAPT

Mean air temperature


(C)
Mean daily
maximum

Mean daily
minimum

Effective asphalt
temperature for 330 mm
asphalt thickness
(C)

29.4

20.7

41.9

31.7

Winter minimum

20.4

9.5

16.5

Summer maximum

30.0

17.9

40.8

29.0

WMAPT
Winter minimum

17.4

9.0

15.2

Summer maximum

28.5

16.8

38.8

27.0

Winter minimum

14.9

7.5

13.5

Summer maximum

25.9

18.7

37.4

27.0

Winter minimum

17.0

9.3

15.3

Summer maximum

26.0

14.3

35.1

24.0

13.5

6.0

12.2

WMAPT

WMAPT

WMAPT
Winter minimum

A.2.5 Asphalt Moduli


The average dynamic compressive moduli measured in the APS-fL project were used to derive the asphalt
design moduli for the calculations. Sullivan et al. (2015) fitted a master curve to the measured mix moduli.
The resulting temperature shift factors are given in Table A 3 and the sigmoidal master curve fitting
parameters are listed in Table A 4.
Table A 3: Temperature shift factors

Conventional Binders

0.001

0.116

Source: Sullivan et al. (2015).

Austroads 2016 | page 89

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table A 4: Master curve fitting parameters

DG14-C320

2.379

1.878

0.043

0.706

DG20-C320

2.569

1.715

0.157

0.818

DG20-C600

1.985

2.363

0.465

0.658

Mix

Source: Sullivan et al. (2015).

Using the parameters to the master curve the asphalt design moduli were calculated for a heavy vehicle
design speed of 80 km/h. For this speed Sullivan et al. (2015) recommend the use of compressive moduli
measured at a frequency of 2 Hz.
Table A 5 and Table A 6 list the asphalt design moduli calculated for the three mix types at the relevant three
asphalt temperatures for each city. Note that these asphalt moduli tend to be higher than values specified by
some Australian road agencies (see Appendix A.3.4).
Table A 5: Asphalt design moduli and ELS for bases made using Class 320 bitumen

City

Temperature
characteristic

Asphalt design modulus at a


vehicle speed of 80 km/h (MPa)

Effective
asphalt
temperature
(C)

DG14C320

DG20C320
DG20C320 (Low voids)

Effective
asphalt
modulus
(MPa)

ELS
(microstrain)

Brisbane
(station
040214)

Summer maximum

41.9

1 280

1 460

1 460

1 430

150

WMAPT

31.7

2 610

3 060

3 060

2 980

93

Winter minimum

16.5

8 060

9 900

9 900

9 630

44

Perth
(station
009034)

Summer maximum

40.8

1 370

1 560

1 560

1 530

144

WMAPT

29.0

3 220

3 830

3 830

3 730

81

Winter minimum

15.2

8 710

10 670

10 670

10 350

42

Adelaide
(station
023000)

Summer maximum

38.8

1 560

1 780

1 780

1 750

132

WMAPT

27.0

3 780

4 530

4 530

4 530

72

Winter minimum

13.5

9 580

11 670

11 670

11 340

39

Sydney
(station
066062)

Summer maximum

37.4

1 720

1 960

1 960

1 920

124

WMAPT

27.0

3 780

4 530

4 530

4 530

72

Winter minimum

15.3

8 680

10 620

10 620

10 310

42

35.1

2 010

2 320

2 320

2 270

112

Melbourne Summer maximum


(station
WMAPT
086071)
Winter minimum

24.0

4 770

5 800

5 800

5 640

62

12.2

10 250

12 400

12 400

11 730

38

Austroads 2016 | page 90

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Table A 6: Asphalt design moduli and ELS for bases made using Class 600 bitumen

Temperature
characteristic

City

Asphalt design modulus at a


vehicle speed of 80 km/h (MPa)

Effective
asphalt
temperature
(C)

DG14C320

DG20C600

DG20C600
(LV)

Effective
asphalt
modulus
(MPa)

ELS
(microstrain)

Brisbane
(station
040214)

Summer maximum

41.9

1280

1640

1640

1590

141

WMAPT

31.7

2610

3670

3670

3490

84

Perth
(station
009034)

Summer maximum

40.8

1370

1770

1770

1710

134

WMAPT

29.0

3220

3220

3220

4340

73

Adelaide
(station
023000)

Summer maximum

38.8

1560

2080

2080

1990

121

WMAPT

27.0

3780

5340

5340

5080

66

Sydney
(station
066062)

Summer maximum

37.4

1720

2300

2300

2220

113

WMAPT

27.0

3780

5340

5340

5080

66

35.1

2010

2780

2780

2650

101

24.0

4770

6700

6700

6380

57

Melbourne Summer maximum


(station
WMAPT
086071)

A.2.6 Asphalt Strain at Fatigue Endurance Limit


The APS-fL draft Technical report (Sullivan et al. 2015) proposes the use of the following relationship
(Equation A1) to determine the endurance limit strains.
ELS =

where

8.2
[k 21625E 0.65 + k 2 ]
1

A1

ELS

endurance limit strain, asphalt tensile strain calculated under an 80 kN single axle
load (microstrain)

asphalt modulus (MPa)

upper 97.5th axle load in the traffic load distribution, usually 9 tonnes

k1

adjustment factor for confidence limits (see Table A 7)

k2

asphalt mix adjustment factor (k 2 =0 for conventional mixes)

Table A 7: Recommended k1 factors


Confidence limit

80th

85th

90th

95th

97.5th

k1

1.0

0.97

0.925

0.86

0.80

Source: Sullivan et al. (2015).

Using an Ul value of 9 tonnes as recommended by Sullivan et al. (2015), the endurance strains for a range of
asphalt moduli are plotted in Figure A 1. The strains for 95th confidence level were used in the design
examples.

Austroads 2016 | page 91

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure A 1: APS-fL asphalt strains at endurance limit


200

180
80th confidence limit
160

85th confidence limit


90th confidence limit

140

95th confidence limit

Asphalt
120
strain
(microstrain)

97.5th confidence limit

100

80

60

40
1000

2000

3000

4000

5000

6000

7000

8000

Asphalt modulus (MPa)

A.3

Design Examples

A.3.1 Effect of Endurance Strains for Selected Subgrade Strengths


The Guide (Austroads 2012) procedures were used to calculate the required asphalt thicknesses for a range
of design traffic loadings for the pavement configurations listed below using the WMAPT for Brisbane of
31.7 C:

50 mm size 14 mm dense graded asphalt (Class 320 bitumen)


X mm size 20 mm dense graded asphalt (Class 320 bitumen)
250 mm unbound granular material as described in Appendix A.2.2
subgrade, with vertical moduli of 30 MPa, 50 MPa and 150 MPa.
In terms of asphalt fatigue damage, the allowable traffic loadings in ESA were calculated using:

Table A 3 asphalt moduli at the WMAPT for Brisbane (31.7C)


a volume of binder of 10.5%
a factor of 1.1 to convert from allowable traffic loading in Standard Axle Repetitions to ESA.
The resulting total asphalt thicknesses are plotted in Figure A 2, without allowance for construction
tolerances.

Austroads 2016 | page 92

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure A 2: Variation in limiting thickness with subgrade modulus

Example ELS for Brisbane: 50 mm DG14Class320 on x mm DG20Class320


on 250 mm granular , subgrade modulus varies
380
370
360
350
340
330
320
310

Total 300
asphalt
290
thickness
(mm) 280
270
260
250
240
230
220
210
200
1.00E+07

1.00E+08
Design traffic (ESA)

Subgrade modulus 30 Mpa, Summer ELS


Subgrade modulus 30MPa, WMAPT ELS
Subgrade modulus 30 Mpa, Winter ELS
Subgrade modulus 50 Mpa, Summer ELS
Subgrade modulus 50 Mpa, WMAPT ELS
Subgrade modulus 50 Mpa, Winter ELS
Subgrade modulus 150 Mpa,Summer ELS
Subgrade modulus 150 Mpa, WMAPT ELS
Subgrade modulus 150 Mpa, Winter ELS

1.00E+09

As shown in Figure A 2, the required asphalt thicknesses have been limited by the APS-fL endurance limit
strains for each of the following temperature:

the effective asphalt layer temperature at midday in the hottest month


the Weighted Mean Annual Pavement Temperature (WMAPT) as used in the Guide
the effective asphalt layer temperature at early morning in the coolest winter month.
The limiting asphalt thickness for each subgrade modulus is the highest thicknesses required for these three
temperature conditions.
As seen from Figure A 2, the variation in limiting thickness with temperature differed between the three
pavement configurations. To understand why this occurs the load-induced strains under a Standard Axle
load were calculated over a range of asphalt moduli for the three subgrade strengths. The load-induced
strains are compared to the APS-fL endurance strains in Figure A 3.
For the 270 mm asphalt thickness pavement configuration, the load-induced strains at the WMAPT are
higher than the APS-fL endurance strains hence a slightly higher limiting thickness is required at this
temperature. Conversely, at the winter temperature the load-induced strains are lower than the endurance
strains hence a limiting thickness is lower. Similarly, the differences in load-induced and endurance strains
explain the variations in limiting thicknesses for the other two pavement configurations.
The following observations were made:

The thickness required at the winter temperatures was the clearly the lowest in all three subgrade moduli.
If this finding can be validated by more extensive testing, the method could be simplified by deleting the
requirement to assess the thickness at winter temperatures.

The thickness required at the WMAPT and at the summer temperature agreed within 10 mm. Again it may be
possible to simplify the method, to only require the limiting thickness at a single temperature to be determined.

The limiting thickness increased as the subgrade modulus decreased.

Austroads 2016 | page 93

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

For Brisbane, the maximum required thicknesses occur for allowable traffic loadings of about
10 to 15 million ESA.
Figure A 3: Comparison of the load-induced strains with the APS-fL endurance limit strains

270 mm asphalt, subgrade modulus 150 MPa


190
180
170
160
150

APS-fL ELS

summer

Load-induced strain, 270 mm asphalt subgrade modulus 150 MPa

140
130
120

Asphalt
110
strain
(microstrain) 100

WMAPT

90
80

270 mm AC
y = 20165x-0.672

70
60
50
40
30
1000

winter
2000

3000

4000

5000

6000

7000

8000

9000

10000

Asphalt modulus (MPa)


330 mm asphalt, subgrade modulus 50 MPa

190
180
170
160
150

APS-fL ELS
Load-induced strain 330 mm asphalt ,Subgrade modulus 50 MPa

summer

140
130
120

Asphalt
110
strain
(microstrain) 100

WMAPT

90
80
70
60
50

330 mm AC
y = 25213x-0.703

40
30
1000

2000

3000

4000

5000

winter
6000

7000

8000

9000

10000

Asphalt modulus (MPa)

Austroads 2016 | page 94

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

375 mm asphalt, subgrade modulus 30 MPa


190
180
170
APS-fL ELS

160
150

summer

Load-induced strain, 375 mm asphalt, subgrade modulus 30 MPa

140
130
120

Asphalt
110
strain
(microstrain) 100
WMAPT

90
80
70
60
50

375 mm AC
y = 34423x-0.746

40
30
1000

2000

3000

4000

5000

winter
6000

7000

8000

9000

10000

Asphalt modulus (MPa)

A.3.2 Effect of Endurance Limit Strains for Selected Cities


The variation in limiting asphalt thickness between selected Australian cities was investigated for the 330 mm
asphalt/subgrade modulus 50 MPa pavement configuration described in Appendix A.3.1.
The resulting total asphalt thicknesses are plotted in Figure A 4, without allowance for construction
tolerances.
As shown in Figure A 4, the required asphalt thicknesses have been limited by the endurance limit strains
proposed in APS-fL project for two asphalt temperature:

the effective asphalt layer temperature at midday in the hottest month


the Weighted Mean Annual Pavement Temperature (WMAPT) as used in the Guide.
(Based on the findings of Appendix A.3.1, the thicknesses for winter temperatures were not calculated).
The following observations were made:

The limiting asphalt thickness was 330 mm for all five climates investigated.
The design traffic loading at which the limiting asphalt thickness was reached deceased as the
temperature increased. For example, for Brisbane the thickness plateau was reached at 110 MESA
whereas for Melbourne the traffic loading was much greater 360 MESA.

The limiting asphalt thickness required at the summer temperatures equalled or exceeded the limiting
thickness calculated to the WMAPT.

Austroads 2016 | page 95

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure A 4: Variation in limiting thickness with climate

Example using Summer ELS: 50 mm DG14Class320 on x mm DG20Class320


on 250 mm granular , subgrade modulus 50 MPa
340
330
320
310
300
290

Total 280
asphalt
270
thickness
(mm) 260

Melbourne
Sydney + Adelaide
Perth
Brisbane

250
240
230
220
210
200
1.00E+07

1.00E+08
Design traffic (ESA)

1.00E+09

Example using ELS at WMAPT: 50 mm DG14Class320 on x mm DG20Class320


on 250 mm granular , subgrade modulus 50 MPa
340
330
320
310
300
290

Total 280
asphalt
270
thickness
(mm) 260

Melbourne
Sydney + Adelaide
Perth
Brisbane

250
240
230
220
210
200
1.00E+07

1.00E+08
Design traffic (ESA)

1.00E+09

Austroads 2016 | page 96

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

A.3.3 Effect of Endurance Limit Strains for Variations of Asphalt Mix Type
The calculations in Appendix A.3.1 and Appendix A.3.2 were undertaken for pavements with size 20 mm
asphalt base made using Class 320 bitumen. It was of interest to investigate the extent to which the limiting
asphalt thickness and associated traffic loading varied due to the use of a harder bitumen in the base. Hence
the calculations in Appendix A.3.2 were repeated except that the base course was changed to a size 20 mm
Class 600 asphalt mix.
The presumptive asphalt moduli in Table 6.13 of the Guide indicate that the design modulus of a Class 600
asphalt mix is about 30% higher a Class 320 asphalt mix. However, there were lesser differences between
these mix types in the compressive modulus values use in these design examples. As seen from Table A 5
and Table A 6, the Class 600 mix was about a 1020% higher than the Class 320 mix at the WMAPT.
The total asphalt thicknesses for the two base mix types are plotted in Figure A 5, together with the limiting
thicknesses at summer temperatures. The thicknesses relate to asphalt layers are placed on top on of 250
mm granular on subgrade modulus of 50 MPa.
As expected due the higher modulus of the Class 600 mix, for a given asphalt thickness the applied
Standard Axle strains are lower and hence the allowable traffic loading is greater.
The limiting asphalt thicknesses are similar (330 mm) for the example pavements with Class 600 and
Class 320 bitumen mixes. Although the Standard Axle strains for a 330 mm thickness are lower when the
higher modulus Class 600 mix is used, the endurance strains for the Class 600 mix are also lower. The net
effect is the limiting thicknesses are the same.
This observation that the limiting thickness does not vary markedly with mix modulus is expected as it
reflects the method used to derive the endurance strains in the APS-fL project. In Equation A1 the
endurance strains are inversely proportional to the modulus to the power of 0.65. This dependency on
modulus was obtained by calculating the Standard Axle strains at individual road pavements for a range of
temperatures and hence asphalt moduli. For example, Figure A 3 shows the variation in Standard Axle
strains with modulus for 330 mm of asphalt on 250 mm unbound granular on subgrade modulus of 50 MPa.
The modulus dependency is close to the 0.65 power of Equation A1. Hence for this pavement configuration
the limiting thickness does not vary significantly with modulus.
The above analysis was taken for pavements in which the asphalt layers were placed on top of 250 mm
granular subbase on subgrade modulus of 50 MPa. It is noted from Figure A 3, that for pavement
configurations with 250 mm granular on subgrade modulus of 30 MPa, there is a greater divergence between
the rate at which the endurance strains varies with modulus and the rate at which the load-induced strains
vary with modulus. For such pavements, the limiting asphalt thicknesses may well vary with modulus
particularly if the modulus difference was greater than
1020% difference investigated above.

Austroads 2016 | page 97

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure A 5: Variation in limiting asphalt thickness with asphalt mix modulus

Example using Summer ELS: 50 mm DG14Class320 on x mm DG20Class320


on 250 mm granular , subgrade modulus 50 MPa
340
330
320
310
300
290

Total 280
asphalt
270
thickness
(mm) 260

Melbourne
Sydney + Adelaide
Perth
Brisbane

250
240
230
220
210
200
1.00E+07

1.00E+08
Design traffic (ESA)

1.00E+09

Example using Summer ELS: 50 mm DG14Class320 on x mm DG20Class600


on 250 mm granular , subgrade modulus 50 MPa
340
330
320
310
300
290

Total 280
asphalt
270
thickness
(mm) 260

Melbourne
Sydney + Adelaide
Perth
Brisbane

250
240
230
220
210
200
1.00E+07

1.00E+08
Design traffic (ESA)

1.00E+09

Austroads 2016 | page 98

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

A.3.4 Effect of Endurance Limit Strains when Current Road Agency Asphalt Moduli
are Used
In Appendix A.3.2 the effect of the APS-fL endurance strains was investigated using the dynamic
compressive moduli measured in the APS-fL project. This section describes the influence of these ELS on
pavement designs in Brisbane, Adelaide and Melbourne using the asphalt design moduli, volume of binders
and design reliabilities in the design supplements of TMR, DPTI and VicRoads respectively. The design
inputs are listed in Table A 8. The asphalt moduli at the summer maximum temperatures were calculated
from the values at the WMAPT using the temperature adjustment formula in the Guide. In all three cases
APS-fL endurance strains at the 95% confidence level were used.
Table A 8: Design inputs
City

Temperature
characteristic

DG14

Brisbane

Summer maximum

41.9*

1090

WMAPT

31.7

2400

2600

Summer maximum

38.8*

1350

1480

WMAPT

27.0

3480

3800

Summer maximum

35.1*

1600

1740

WMAPT

24.0

3600

3900

Adelaide
Melbourne

Temperature
(C)

Asphalt modulus (MPa)


DG20

Volume of
binder (%)

Design reliability
(%)

1180

10.5

95

11.0

95

10.4

97.5

* For a total asphalt thickness of 330 mm.

The resulting total asphalt thicknesses are plotted in Figure A 6 without allowance for construction
tolerances.
Figure A 6: Variation in limiting asphalt thickness using current road agency asphalt moduli

Example using Summer ELS: 50 mm DG14Class320 on x mm DG20Class320


on 250 mm granular , subgrade modulus 50 MPa
340
330
320
310
300
290

Total 280
asphalt
270
thickness
(mm) 260

Melbourne, reliability 97.5%


Adelaide, reliability 95%
Brisbane, reliability 95%

250
240
230
220
210
200
1.00E+07

1.00E+08
Design traffic (ESA)

1.00E+09

Austroads 2016 | page 99

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Comparing the results of Figure A 4 and Figure A 6, there are very substantial differences in asphalt
thicknesses for Melbourne as:

VicRoads asphalt moduli are about 70% of values estimated from the APS-fL method.
The design reliability in Figure A 6 was 97.5% consistent with the VicRoads method, compared to 95% in
Figure A 4.
Note also the allowable traffic loading at which the Melbourne thickness reached its maximum reduced
significantly to a value similar to the other two cities.

Austroads 2016 | page 100

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Appendix B In-Service Endurance Strain Limits


Derived from NCHRP Project 9-44A
B.1

Introduction

Section 3 describes a major research project (NCHRP 9-44A) undertaken to develop laboratory endurance
limit strains (ELS).
This Appendix describes how the results were utilised to develop possible in-service ELS relationships for
use in the Guide (Austroads 2012).
Two in-service ELS relationships were developed:

ELS predicted from the asphalt design modulus (Appendix B.2), which is listed as Option B in Section 7
ELS predicted from asphalt temperature with adjustments for mix volumetrics (Appendix B.3), which is
listed as Option C in Section 7.

B.2

Limiting Strain-asphalt Modulus Relationship

B.2.1 Relationship Derived from Laboratory Beam Fatigue Testing


The final project report (Witczak et al. 2013) proposed the use of Equation 8 to further develop endurance
limits for inclusion in the United States mechanistic-empirical pavement design method (MEPDG). The report
recommends the endurance limits be calculated using N = 200 000. Using this value, the extrapolated strain
at a stiffness ratio (SR) of 1.0 is the ELS.
Figure 3.6 illustrates how the laboratory ELS varies with flexural modulus and rest period. It is apparent that
the duration of the rest between loading pulses has a major effect on FEL.
Witczak et al. (2013) have proposed the variation in ELS with rest period be included in the MEPDG. A
suggested process is to calculate the average time between individual axles in each two hour interval of a
typical day each month. For each of the 144 analysis periods in a year, the MEPDG provides a process for
estimating the pavement temperature and hence asphalt design modulus to enable the allowable loading to
be calculated and compare the applied traffic loading (Figure B 1).
The Guide determines a single asphalt modulus from the WMAPT and truck vehicle speed rather than allow
for hourly and monthly variations in modulus. Accordingly, for the Guide, allowance for the variation in rest
period with traffic loadings is not considered appropriate.
The beam fatigue testing was undertaken using rest periods of 0, 1, 5 and 10 seconds, with 5 seconds being
the predominant period. Consequently, in developing a potential method for inclusion in the Guide, the
laboratory ELS for a rest period of 5 seconds were utilised and adjusted to in-service values as described in
Appendix B.2.2.

Austroads 2016 | page 101

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure B 1: Example of the truck axle distribution during a 24 hour period

Source: Appendix 2 of Witczak et al. (2013).

B.2.2 Field Adjustment Factor


To provide endurance limits suitable for inclusion in the Guide information is required on the pavement
composition of long-life asphalt pavements. The most comprehensive investigation to identify long-life
asphalt pavements was the United Kingdom (UK) study described in Section 4.1. Consequently, in the
absence of a comparable Australian study, it was decided to utilise this performance data to adjust the
laboratory ELS relationship (Equation 8) to an in-service ELS relationship.
The UK study lead to the current UK pavement design manual contained in the Highways England (2006)
and Highways England (2009). Figure B 2 is UK design chart which can be used to determine the specified
thickness for a range of asphalt mix types and degrees of support (Foundation Classes) provided by the
layers under asphalt. It is noted that for design traffic loadings of 8 x 107 ESA or more the same asphalt
thicknesses apply, these being long-life asphalt pavements.
Foundation Class 2 corresponds to an effective half-space modulus of 100 MPa and commonly comprises
capping materials and unbound granular subbases, whereas Class 3 and 4 include bound subbases. As the
performance data used in the development of their long-life pavement design method was largely based on
pavements with foundations similar to Class 2 it was decided to utilise designs with this foundation support in
developing the field adjustment factor.
Note that the UK Design Manual states that if Class 2 foundation is used for design traffic in excess of 8 x
107 ESA (long-life pavements) 150 mm or more of bound subbase must be used. Such bound subbases
were not used in the pavements which were researched to develop the method (Nunn et al. 1997, Nunn
2004). It seems this bound layer was added later, possibly to provide additional reliability in the design of
long-life pavements. In developing the field adjustment factor, it was considered inappropriate to require
long-life pavements to have bound subbase, this being overly conservative. As seen from Figure B 3 for a
subgrade with design modulus of 50 MPa, 320 mm thickness granular subbase is required to provide a
Class 2 foundation. This foundation was assumed in the calculation of the field adjustment factor.

Austroads 2016 | page 102

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure B 2: UK design thicknesses for flexible pavements

Source: Highways England (2006).


Figure B 3: Design chart for foundations

Source: Highway England (2009).

Austroads 2016 | page 103

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

The next issue to address in calculating the field adjustment factor was which of the bituminous mixes shown
on the design chart (Figure B 2) is most closely aligned with Australian basecourse. As Class 320 and Class
600 binders are commonly used in Australia, the designs for the dense bituminous macadam with 125
penetration grade bituminous binders (DBM125) and EME2 were considered the least relevant. In addition,
as dense graded asphalt is commonly used in Australia, the gap-graded hot-rolled asphalt (HRA50) design
thicknesses were not utilised.
Consequently, it was decided to develop the adjustment factor from the design thicknesses for heavy duty
mixture (HDM) (formerly called heavy duty macadam) with 40/60 (designated 50) penetration bitumen.
Penetration grade 40/60 binder has a similar penetration to Australian Class 320. According to the latest
standard (British Standard (BS) 2015), HDM is now commonly referred to as asphalt concrete (AC). When
used as binder course it can have a nominal maximum size of 20 mm (designated AC20) or 32 mm
(designed AC32).
As seen from Figure B 2, for Foundation Class 2 and a design traffic loading of 8 x 107 ESA or more, the
total asphalt thickness of HDM50 is 360 mm. In determining these design thicknesses the following two
conservative adjustments were made by the developers of the UK method to thicknesses of the field long-life
pavements:

In the field investigations of long-life asphalt pavements (Section 4.1) it was observed that surface
cracking could extend to 100 mm below the surface without associated deeper structural cracking (Nunn
et al. 1997). Consequently, an additional 100 mm was added to the thickness of these long-life
pavements based on the conservative assumption that the material down to the depth of the crack
penetration does not contribute to load-spreading.

The long-life pavements had been subjected to truck traffic operating under maximum legal axles of
10.5 tonnes. Nunn et al. reported that the legal load limit would increase to 11.5 tonnes in 1997. An
increase of asphalt thickness of 20 mm was included which more than allowed for this 1 tonne load
increase.
As the current Australian legal limit for a single axle with dual tyres is 9 tonnes, it is unnecessary to include
the 20 mm additional thickness for long-life performance on Australian roads.
In addition, the assumption that at no time during a 3040 year design period would the top 100 mm of
asphalt contribute structurally is considered overly conservative. When Australian heavy-duty pavements are
rehabilitated the cracking distress is such that it is commonly not necessary to remove more than the top 40
50 mm thickness. It is anticipated that this cracked top 50 mm still contributes structurally together with the
underlying 50 mm which may be micro-cracked. Rather than 100 mm allowance for this surface distress, it
was assumed that a less conservative 40 mm thickness adjustment would be appropriate for the
development of the field adjustment factor.
Consequently, a total asphalt thickness of 280 mm of HDM50 on Class 2 foundation would be an appropriate
structure for the development of the field adjustment factor. As the UK long-life pavements investigated by
Nunn et al. appear to have included hot rolled asphalt (HRA) surfacing, this surfacing was assumed.
Table B 1 lists the assumed long-life pavement structure.
Table B 1: Assumed configuration of a long-life asphalt pavement for model calibration
Thickness
(mm)

Vertical modulus
(MPa)

Horizontal modulus
(MPa)

Poissons ratio

Hot rolled asphalt, pen 50

40

4370

4370

0.40

Asphalt concrete (AC20), pen 50

60

7240

7240

0.40

Asphalt concrete (AC32), pen 50

180

7470

7470

0.40

320(1)

150(1)

75(1)

0.35

Semi-infinite

50

25

0.45

Material type

Unbound granular subbase


Subgrade
1

The total granular thickness was divided into five equi-thick sublayers. The modulus of the top sublayer is listed.

Austroads 2016 | page 104

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

To adjust the laboratory ELS (Equation 8) to reflect field performance, the asphalt strain under a 80 kN
Standard Axle load needed to be predicted for the assumed long-life asphalt pavement (Table B 1) and
compared to the laboratory ELS. The asphalt moduli used in these strain predictions need to be calculated
using the Guide procedures but using UK pavement temperatures and mix characteristics. The Guide uses the
WMAPT as the pavement temperature at which to calculate asphalt design modulus. It was decided to use the
WMAPT for London to calculate the moduli, this temperature is 18.7 C. As the UK long-life pavements were
generally motorways, the asphalt moduli needed to be calculated using a design heavy vehicle speed
consistent with Australian practice for freeways: a heavy vehicle design speed of 80 km/h was used.
In predicting the strains applied to the long-life asphalt pavement, the asphalt moduli needed to be calculated
using the Guide processes. The most reliable method of obtaining the asphalt moduli in this manner would
be to test UK HDM mixes using the Australian mix design processes for sample preparation and
measurement of indirect tensile modulus. Unfortunately this test information is not available. Instead, the
design moduli listed by three Australian road agencies (i.e. VicRoads, TMR and DPTI) were utilised. For a
WMAPT of 18.7 C and a design speed of 80 km/h the following design moduli were calculated as being
typical for an Australian size 20 mm dense graded asphalt base course:

E = 7240 MPa at WMAPT = 18.7 C, speed of 80 km/h and 5% air voids


E = 6790 MPa at WMAPT = 18.7 C, speed of 80 km/h and 6% air voids.
As explained below these moduli were adjusted to determine the design moduli for three mixes in Table B 1.
In the calculation of the field adjustment factor, it was assumed that a typical UK pavement (Table B 1) would
include a 60 mm thick binder course of AC20 under the surfacing. In the examples in Appendix B of the
guide to the BS EN 13108-1 (BS 2015), the target binder content of AC20 binder course is about 4.64.7%.
In addition, the target particle size distribution for AC20 is within the typical specification range for an
Australian AC20 mix. Accordingly, the UK AC20 binder course was assumed to have similar modulus to an
Australian AC20 base course. A design modulus of 7240 MPa was used for the AC20 binder course
consistent with the value calculated above.
The assumed long-life pavement (Table B 1) included an AC32 basecourse. This mix differs from Australia
size 20 mm basecourse in at least two ways:

Nominal size: AC32 has a nominal maximum particle size of 32 mm compared to 20 mm. Based on the
indirect tensile moduli in Table 6.13 of the Guide (Austroads 2012), it would be expected that this larger
size mix would have a modulus about 10% above Australian size 20 mm basecourse.

Mix volumetrics: In the examples in Appendix B of the guide to the BS EN 13108-1 (BS 2015), the
binder contents of AC32 for mixes made using crushed rock are 3.94.0%. In addition, the binder
contents of size 28 mm HDM basecourse mixes appear to be 4.0-4.5% based on the mixes reported by
Nunn et al. (1997) and Nunn and Smith (1997). In the calculation of the field adjustment factor it was
decided to assume a bitumen content of 4.1% and 6% air voids for AC32 mix. Although data on typical
Australian basecourses are not readily available, based on the volume of binder content being used in
structural design, it was assumed that Australian size 20 mm basecourses have higher bitumen contents
(similar to UK AC20) and slightly lower air voids, about 5%. This difference in mix volumetrics results in
AC32 being about 7% lower than the Australian basecourse, hence almost cancelling the effect of the
high nominal size.
With these adjustments, an AC32 design modulus of 7470 MPa was calculated. It is noted a more
conservative design modulus at a higher temperature (20 C) and lower loading rate is used in the
UK (Highways England 2006): namely, 4700 MPa.
As indicated in Table B 1, the example UK heavy-duty pavement includes a hot rolled asphalt surfacing (HRA).
Such a gap-graded asphalt is not commonly used as base in Australia and so a different approach was used to
estimate its design modulus. The UK design modulus for HRA50 is 66% of the value for HDM50. Applying this
factor to the value determined for AC32 (7470 MPa), a design modulus of 4930 MPa was calculated for HRA
surfacing. Again, the design modulus (3100 MPa) used in the UK design processes is lower.
The moduli for the granular subbase and the subgrade were calculated in accordance with the Guide
(Austroads 2012), these are listed in Table B 1.

Austroads 2016 | page 105

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

For the pavement listed in Table B 1, the maximum tensile asphalt strain under an 80 kN standard axle load
was calculated to be 63 microstrain.
To derive the field adjustment factor, this value needs to be compared to the laboratory ELS calculated from
the NCHRP 9-44A project. For this basecourse asphalt design modulus of 7470 MPa, a rest period of 5
seconds and using the equation in Figure 3.6 the laboratory ELS is 87 microstrain, well above 63 microstrain
calculated from the UK field performance data.
Consequently it was estimated that the laboratory ELS at a rest period of 5 seconds need to be multiplied by
an adjustment factor of 0.72 to provide a value consistent with the performance of UK long-life pavements.
This NCHRP 9-44A design process assumes any differences in ELS due to differences between asphalt mix
characteristics of UK mixes, Australian mixes and the mixes used in the NCHRP research, can be explained
in terms of the effect of these mix characteristics on asphalt design modulus.

B.2.3 Limiting Strain vs ModulusE Relationship


Applying this adjustment factor to the laboratory ELS for 5 seconds (see equation in Figure 3.6), the resulting
in-service ELS relationship was calculated and is given in Figure B 4.
Figure B 4: Endurance limit strains after adjustment to in-service values using UK long-life pavements
110

100

90

Endurance
limit strain 80
(microstrain)
70

ELS = 604E-0.254

60

50
1000

2000

3000

4000

5000

6000

7000

8000

Asphalt modulus (MPa)

B.3

Limiting Strain Temperature Relationship

B.3.1 Relationship Derived from Laboratory Testing


As discussed in Section 3.9, an alternative procedure was derived in NCHRP 9-44A to determine laboratory
ELS from test temperature, bitumen content and air voids. The laboratory ELS are determined from
Equation 9 by calculating the strain at which the stiffness ratio (SR) is 1 and N = 20 000 for a given
temperature, bitumen content, air voids content and rest period.
Figure 3.10 shows some examples of the laboratory ELS so determined.

Austroads 2016 | page 106

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

B.3.2 Field Adjustment Factor


The laboratory ELS determined using Equation 9, need to be adjusted to in-service values. The process
used to determine the adjustment was similar to that described in Appendix B.2.2 using UK pavement
performance data.
As mentioned in Appendix B.2.2, for the example long-life pavement in Table B 1 the applied strain under a
Standard Axle load is 63 microstrain.
To derive the field adjustment factor, this applied strain needs to be compared to the laboratory ELS
calculated from Equation 9. Equation 9 requires the following inputs to enable that strain at SR = 1 to be
determined:

asphalt temperature
binder content, percentage by mass
volume of air voids
rest period between loading pulses
number of loading cycles.
In the Guide, the asphalt temperature is characterised as the WMAPT. Consistent with Appendix B.2.2, it
was assumed the UK performance data relates a WMAPT of 18.7 C. Hence this temperature was used to
calculate the laboratory ELS for the AC32 basecourse mix (Table B 1).
As discussed in Appendix B.2.2, the assumed bitumen content was 4.1% and the air voids content 6%.
Consistent with Appendix B.2.2, the laboratory ELS was calculated using a 5 second rest period, as most of
the laboratory beam testing with rest periods was obtained at this value.
As recommended by Witczak et al (2013), the ELS was calculated using N = 20 000 cycles.
Using these inputs in Equation 9, a laboratory ELS of 68 microstrain was calculated for this
AC32 basecourse. Note that this value is well below the 87 microstrain estimated by the modulus prediction
method (Appendix B.2.2).
Consequently it was estimated that the laboratory ELS at a rest period of 5 seconds of 68 microstrain need
to be multiplied by a factor of 0.93 to provide a value consistent with the field performance of UK long-life
pavements (63 microstrain).

B.3.3 Limiting Strain vs Temperature Relationship


Applying the adjustment factor to the laboratory ESL for 5 seconds estimated using Equation 9, the resulting
in-service ELS are given in Figure B 5 for mixes with 5% air voids and a range of bitumen contents.
A significant limitation of this method for use in the Guide is that the limiting strains relate to dense graded
asphalt basecourses with penetration grade 4060 bitumen. A procedure would need to be developed to
cater for a wider range of mixes, possibly using mix moduli relative to the mix used in the calibration. This is
discussed further in Section 7.4.

Austroads 2016 | page 107

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Figure B 5: Example endurance limits after adjustment to in-service values using UK long-life pavements

5 % air voids

140
130

4.2% bitumen
4.7% bitumen

120

5.2% bitumen

110

Endurance 100
limit
strain
(microstrain) 90
80
70
60
50
14

16

18

20

22

24

26

28

30

32

34

36

38

40

Asphalt temperature (C)

Austroads 2016 | page 108

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Appendix C Proposed Part 2 Text


The following text is proposed for the 2017 Edition of the Guide to Pavement Technology Part 2: Pavement
Structural Design (Austroads 2012).

Section 6.5.6
Replace the last paragraph in the 2012 Edition of Part 2 with the following:
As an interim measure and pending further research to quantify the increase in crack healing with increasing
temperature, limits may be placed on the design traffic used in the asphalt fatigue damage calculations
(Section 8.2.5).
Add the following new sub-section:

8.2.5 Traffic loading limits due to asphalt crack healing


As discussed in Section 6.5.6, there is increasing recognition of the notion that asphalt mixes experience
increased crack healing with increasing temperature to a greater extent than provided in the asphalt fatigue
relationship (Equation 11).
As an interim measure, pending further research to quantify the increase in crack healing with increasing
temperature, an upper limit may be used in the design traffic loading used in the asphalt fatigue damage
calculations. Table 8.4 provides suggested limiting design traffic loadings in Equivalent Standard Axles
(ESA). The limit reduces with increasing temperature due to the increase in crack healing. Based on the ESA
limits and using the number ESA per heavy vehicle axle groups (HVAG), the upper limit in terms of
cumulative HVAG used in the asphalt fatigue damage calculations is determined as shown in Equation 25.
Table 8.4 Suggested upper limits on design traffic for asphalt fatigue
WMAPT

25 C

26-34 C

35 C

Design traffic loading limit (ESA)

4 x 108

2 x 108

108

where

(/)

25

upper limit on the design traffic expressed as heavy vehicle axle groups
(HVAG) for use in the asphalt fatigue damage calculations

upper limit on the design traffic expressed as Equivalent Standard Axles


(ESA) for use in the asphalt fatigue damage calculations (Table 8.4)

average number of ESA per HVAG from the project traffic load distribution

Note that such design traffic loading limits are only used in the asphalt fatigue calculations; they are not
applicable to the design traffic used for cemented materials fatigue or permanent deformation damage
calculations.

Austroads 2016 | page 109

Asphalt Fatigue Damage Healing and Endurance Limits: Guide Implementation Options

Appendix D

Austroads 2016 | page 110

You might also like