Sie sind auf Seite 1von 350

THE EFFECTS OF HEAVY VEHICLE SINGLE, TANDEM

AND TRI-AXLES ON SPRAYED SEAL WEAR IN


AUSTRALIA

Kym Lawrence Neaylon


B Eng. (Civil), University of South Australia

A thesis submitted in fulfillment of the requirements for the degree


of Doctor of Philosophy

January 2017

Faculty of Science, Engineering and Technology


Swinburne University of Technology
Melbourne, Australia
ABSTRACT

Sprayed seals, also known as chip seals, are a thin and flexible pavement surfacing
comprising of a layer of bituminous material, usually a millimetre or two thick, covered
with a layer of crushed rock aggregate. These surfacings account for approximately
90% of the total length of all surfaced roads in Australia, New Zealand and South
Africa. They remain a key pillar in the provision of low cost all-weather road-transport
networks throughout these countries.

Commercial traffic loadings have changed significantly since seal designs were first
engineered in the 1930s. This is particularly so in Australia with freight efficiency
resulting in longer and heavier loads being transported by more powerful prime
movers. In other developed countries these heavier loads would be carried on heavy
duty hot mix asphalt pavements. In the less densely populated countries such as
Australia, New Zealand and South Africa such heavy-duty pavement networks are not
economically feasible, and the capacity of thin flexible pavements surfaced with
sprayed seals must be extended as far as possible.

Freight efficient vehicles today introduce significantly more tandem, triaxle axle and
quad axle group loadings than were present in the 1930s and even the 1970s when
seal designs were developed. However, the effects of differing axle groupings and axle
loadings has never been thoroughly considered in sprayed seal design, because these
effects have never been isolated and studied.

Damage to sprayed seals – as distinct to loss of shape within pavements – is


considered in terms of reduction in macrotexture, which is known in the sealing industry
as loss of surface texture. Therefore, a model for sprayed seal surface texture
deterioration is key to this research. The model must be able to separate the effects of
various axle groupings and various axle loadings. Therefore, the objective of this
research is to develop a generalised model to predict surface texture deterioration
based on the effect of axle groupings and loadings.

In meeting this objective, an additional benefit will be the determination of a specific


load damage exponent applicable to sprayed seal wear.

Such a generalised model could contribute to a refinement of sprayed seal design,


specifically in the category of heavy vehicle seal design. The results may be used to
improve the consideration of abnormal traffic affects in seal design of mine haul roads,
or national freight routes.

— i—
After reviewing the research gaps, the hypothesis of this study can be stated as Heavy
vehicle axle group loads that cause the same pavement damage do not necessarily
cause the same sprayed seal damage.

Using the model developed, this hypothesis was found to be supported under the
experimental conditions of this study. This now provides a documented argument
against adopting any existing equivalent standard axle data derived for pavement
design use, and against using equivalent standard axle data as a seal design input.

By re-defining damage as applicable to sprayed seal wear, a load damage exponent


for sprayed seals has been found to be closer to 1 than to the 4 used for unbound
granular pavements that sprayed seals generally are used on, and the 2.7 previously
assumed for sprayed seals.

It was then found that for this experimental set of conditions and for axle groups
travelling in a straight line and not applying tractive motion forces:

1. When comparing total freight task, single axles and tandem axles caused similar
surface texture reductions for a given load, but the triaxle caused less spray seal
surface wear than either of those for a given load.

2. When comparing equivalent standard axle loads, for a given freight task triaxial
groups and quad axle groups will cause less spray seal surface wear than for single
axles and tandem axles.

In summary, having more triaxles on the sprayed seal network (in straight line travel
situations) has the potential to cause less sprayed seal wear than the equivalent load
carried by single axles or tandem axles.

— ii —
ACKNOWLEDGEMENTS

Firstly, to Helen, my dear wife. Having gone without holidays for nine years that long
awaited European river cruise now looks achievable - if we are not too old.

My Swinburne supervisors Professor Kerry J. McManus and Professor Emad Gad for
providing supervision, guidance, support and understanding over the years.

My colleagues at ARRB deserve special mention: Dr Mike Moffatt, Dr Robert Urquhart,


Dr Tim Martin, and Thorolf ‘I’m not a “doctor”, I’m just the mechanic’ Thoresen; who
collectively provided the necessary support and encouragement to keep on grinding
over this tortuous and unpredictable nine-year journey, and acted as a sounding board
when I needed to discuss ideas. Mike and Tim also did their journey mature-aged and
part-time in amongst a demanding day job and know not only the dedication required,
but also the satisfaction that can be gained.

Peter Milne in his capacity as a highly skilled editor who provided his services as a
favour for a friend.

I also acknowledge Austroads, which provided the funding for experimental data
collection, albeit under a different project objective, and for granting permissions to use
much material published by them.

And finally my old canine supporter ‘Steady’ who started enthusiastically with me at the
side of the desk but finished his own lifelong journey first. His spirit remains here in my
study with me to the very end. It’s been a long nine years.

— iii —
AUTHOR’S DECLARATION

I hereby declare that this thesis contains no material which has been accepted for the
award to the candidate of any other degree or diploma, and that to the best of my
knowledge contains no material previously published or written by another person
except where due reference is made in the text of the examinable outcome.

I warrant that I have obtained, where necessary, permission from the copyright owners
to use any third party copyright material reproduced in the thesis (such as artwork,
images, unpublished documents), or to use any of my own published work (such as
Austroads Reports and conference papers) in which the copyright is held by another
party (such as publisher, co-author).

Kym L Neaylon

January 2017

Copyright of this thesis is owned by Kym L. Neaylon, except where otherwise stated
(e.g. Appendix E). Reference to or reproduction of any original part of this work may
be made provided that a proper reference is made. Authorisation to reproduce material
identified as being the copyright of a third party must be obtained from the copyright
holders concerned.

DOI: 10.13140/RG.2.2.26498.89280

— iv —
PREFACE

The central research question discussed herein arose from my 35 years of civil
engineering problem-solving, crystallised whilst consulting on causes of a major
sprayed seal stripping problem in Queensland. I asked the question ‘why is it so’ on a
number of aspects, but the existing literature remained silent on this one particular
topic. The desire to find the answer led to this thesis and ultimately to satisfaction.

Many of the photographs in this document were taken personally during years of
sprayed seal consulting work in Queensland, New South Wales, South Australia,
Western Australia, the Northern Territory and Ethiopia, and during research work in
Victoria and New Zealand.

— v—
PUBLICATIONS

Condensed versions of material in this thesis have been published in:

Neaylon, KL 2009, ‘Stereo photography as a tool to understand sprayed seal distress


mechanisms’, AAPA International Flexible Pavements Conference,13th, Gold
Coast, Australian Asphalt Pavement Association, Kew, Victoria.

Neaylon, KL 2012a, ‘The effects of heavy vehicle single, tandem and triaxle groupings
on sprayed seal wear – stage 1’, Austroads Technical Report AP-T207-12,
Austroads, Sydney. (Peer reviewed).

Neaylon, KL 2012b, ‘The effects of single, tandem and triaxles on sprayed seal wear’,
ARRB Conference 25th, Perth, ARRB, Vermont South, Victoria. (Peer reviewed).

Neaylon, KL 2015a, ‘Towards incorporating heavy vehicles into sprayed seal design –
stage 2’, Austroads Technical Report AP-T292-15, Austroads, Sydney. (Peer
reviewed).

Neaylon, KL 2015b, ‘Incorporating heavy vehicles into seal designs’, CAPSA 11th, Sun
City, South Africa. (Peer reviewed).

Neaylon, KL 2015c, ‘Are more triaxles actually good for chip seals?’ NZ Transport
Agency & NZIHT Annual Conference, 16th, Bay of Islands, New Zealand.

Neaylon, KL 2016, ‘Are more road trains good for sprayed seals?’ ARRB Conference,
27th, Melbourne, ARRB, Vermont South, Victoria. (Peer reviewed).

Other publications arising from this work but not included in this thesis are:

Millar, P, Woodward, D, McQuaid, G & Neaylon, KL 2012, ‘Investigation of aggregate


displacement in chipseal surfaces using close range digital photogrammetry’,
Maintenance and rehabilitation of pavements and technological control
(MAIREPAV) conference, 7th, Auckland. (Peer reviewed).

Neaylon, KL, Spies, R & Alderson, A 2008, ‘The equivalent heavy vehicle concept in
Australian sprayed seal design’, International sprayed sealing conference, 1st,
Adelaide, ARRB Group, Vermont South, Victoria. (Peer reviewed).

— vi —
CONTENTS

1 INTRODUCTION ................................................................................................................. 1-1

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


1.2 Theoretical framework ......................................................................................................... 1-5
1.3 Research purpose ............................................................................................................... 1-7
1.4 Method ................................................................................................................................ 1-8
1.5 Thesis structure ................................................................................................................... 1-9
1.6 Definitions.......................................................................................................................... 1-10
1.7 Acronyms .......................................................................................................................... 1-12
2 LITERATURE REVIEW....................................................................................................... 2-1

2.1 Introduction.......................................................................................................................... 2-1


2.2 Road pavements ................................................................................................................. 2-1
2.2.1 Road pavement function ..................................................................................... 2-1
2.3 Bituminous road surfacings ................................................................................................. 2-3
2.3.1 Purpose of a surfacing ........................................................................................ 2-3
2.3.2 Surfacing types ................................................................................................... 2-3
2.3.3 Selection of surfacings ....................................................................................... 2-6
2.4 Sprayed seal materials ........................................................................................................ 2-8
2.4.1 Bitumen .............................................................................................................. 2-8
2.4.2 Cutback bitumen............................................................................................... 2-19
2.4.3 Bitumen emulsion ............................................................................................. 2-21
2.4.4 Polymer modified bitumen (PMB) ..................................................................... 2-23
2.4.5 Polymer modified bituminous emulsion (PME) ................................................. 2-24
2.4.6 Aggregate ......................................................................................................... 2-25
2.5 Sprayed seal design and construction ............................................................................... 2-29
2.5.1 Seal design worldwide ...................................................................................... 2-29
2.5.2 Aggregate wetting and adhesion ...................................................................... 2-38
2.5.3 Aggregate precoating ....................................................................................... 2-40
2.5.4 Aggregate rolling and re-orientation ................................................................. 2-40
2.5.5 Aggregate embedment ..................................................................................... 2-45
2.6 Sprayed seal failure modes ............................................................................................... 2-46
2.6.1 Flushing and bleeding of seals ......................................................................... 2-47
2.7 Measurement of pavement surface texture ....................................................................... 2-49
2.7.1 Texture depth (TD) ........................................................................................... 2-52
2.7.2 Volumetric texture measurement ...................................................................... 2-53
2.7.3 Mean texture depth (MTD) ............................................................................... 2-53
2.7.4 Sand patch texture depth (SPTD)..................................................................... 2-53
2.7.5 Inaccuracies in the sand patch test .................................................................. 2-58
2.7.6 Profile depth (PD) ............................................................................................. 2-59
2.7.7 Mean profile depth (MPD) ................................................................................ 2-59
2.7.8 Sensor measured texture depth (SMTD) .......................................................... 2-60
2.7.9 Estimated texture depth (ETD) ......................................................................... 2-61
2.8 Measuring seal performance ............................................................................................. 2-62
2.8.1 Visual assessment............................................................................................ 2-62
2.8.2 Pull-out force .................................................................................................... 2-62
2.8.3 2D imaging ....................................................................................................... 2-65
2.8.4 3D imaging ....................................................................................................... 2-67
2.8.5 Photogrammetry ............................................................................................... 2-68
2.8.6 3D imaging standards....................................................................................... 2-71
2.8.7 Comment on methods of measuring seal performance .................................... 2-72
2.8.8 Summary of surface texture measurements ..................................................... 2-73
2.9 Predicting seal performance .............................................................................................. 2-74
2.9.1 Seal texture deterioration ................................................................................. 2-75
2.9.2 Finite element analysis ..................................................................................... 2-87

— vii —
2.10 Pavement design and traffic loadings ................................................................................ 2-89
2.10.1 Fundamental differences between pavements and surfacings ......................... 2-89
2.10.2 Pavement design .............................................................................................. 2-90
2.10.3 Load damage equivalencies ............................................................................. 2-91
2.10.4 Standard axle ................................................................................................... 2-92
2.10.5 Standard axle repetitions .................................................................................. 2-94
2.10.6 Load damage exponents .................................................................................. 2-95
2.10.7 The 4th power law ............................................................................................. 2-95
2.10.8 The relevance of the 4th power law in seal design ........................................... 2-99
2.10.9 Axle spacing ................................................................................................... 2-100
2.11 Conclusions ..................................................................................................................... 2-101
3 DEVELOPMENTS IN HEAVY VEHICLE SEAL DESIGN PRACTICE ................................ 3-1

3.1 Introduction.......................................................................................................................... 3-1


3.1.1 Seal designs ....................................................................................................... 3-1
3.1.2 Focus of this chapter .......................................................................................... 3-3
3.2 Tropical and sub-tropical countries ...................................................................................... 3-3
3.3 Republic of South Africa ...................................................................................................... 3-5
3.4 New Zealand ....................................................................................................................... 3-6
3.4.1 New Zealand Transport Agency ......................................................................... 3-6
3.4.2 New Zealand Forest Owners Association ........................................................... 3-7
3.5 United Kingdom ................................................................................................................... 3-7
3.6 North America ..................................................................................................................... 3-8
3.6.1 Minnesota ........................................................................................................... 3-8
3.6.2 Texas .................................................................................................................. 3-9
3.6.3 Kansas ............................................................................................................... 3-9
3.6.4 North Carolina .................................................................................................... 3-9
3.7 Australia ............................................................................................................................ 3-14
3.7.1 Current method................................................................................................. 3-14
3.7.2 Current investigations ....................................................................................... 3-18
3.8 Seal design comparisons .................................................................................................. 3-21
3.9 Summary ........................................................................................................................... 3-22
4 EXPERIMENTAL METHOD ................................................................................................ 4-1

4.1 Introduction.......................................................................................................................... 4-1


4.2 The Accelerated Loading Facility (ALF)............................................................................... 4-4
4.2.1 Equipment background ....................................................................................... 4-4
4.2.2 Experimental background ................................................................................... 4-6
4.2.3 Tyre and axle assembly characteristics .............................................................. 4-7
4.3 Pavement ............................................................................................................................ 4-8
4.4 Surfacing selection .............................................................................................................. 4-9
4.5 Sprayed seal surfacing ...................................................................................................... 4-11
4.5.1 Pre-design data collection ................................................................................ 4-11
4.5.2 Prime trials........................................................................................................ 4-11
4.5.3 Seal design ....................................................................................................... 4-12
4.5.4 Prime construction ............................................................................................ 4-13
4.5.5 Sprayed seal construction ................................................................................ 4-13
4.6 ALF experimental design ................................................................................................... 4-14
4.6.1 Definition of the problem................................................................................... 4-14
4.6.2 Response to be measured ............................................................................... 4-15
4.6.3 Selection of variables ....................................................................................... 4-17
4.6.4 Factorial design ................................................................................................ 4-20
4.6.5 Experimental layout .......................................................................................... 4-21
4.6.6 Testing regime .................................................................................................. 4-23
4.6.7 Data collection .................................................................................................. 4-24
4.7 Limitations ......................................................................................................................... 4-32
4.7.1 Surface quality of the constructed basecourse ................................................. 4-32
4.7.2 Construction variability ..................................................................................... 4-33

— viii —
4.7.3 Surface texture measurement .......................................................................... 4-33
4.7.4 Binder aging ..................................................................................................... 4-34
4.7.5 Velocity of ALF ................................................................................................. 4-36
4.7.6 Driven axles ...................................................................................................... 4-36
4.7.7 Straight line loading .......................................................................................... 4-37
4.8 Summary ........................................................................................................................... 4-38
5 DATA ANALYSIS ............................................................................................................... 5-1

5.1 Introduction.......................................................................................................................... 5-1


5.2 Sand circle data ................................................................................................................... 5-2
5.3 Raw data ............................................................................................................................. 5-2
5.3.1 Outliers ............................................................................................................... 5-2
5.3.2 Raw data scatter plots ........................................................................................ 5-3
5.3.3 Correlations in operator error............................................................................ 5-10
5.3.4 Data grouped in axle types with bedding-in removed ....................................... 5-11
5.4 Check of construction variability ........................................................................................ 5-13
5.5 Normalising the data.......................................................................................................... 5-18
5.6 Models ............................................................................................................................... 5-22
5.7 Nonlinear regression modelling ......................................................................................... 5-23
5.7.1 Principles of nonlinear regression modelling .................................................... 5-24
5.7.2 One parameter curves ...................................................................................... 5-25
5.7.3 Two parameter curves ...................................................................................... 5-26
5.7.4 Three parameter curves ................................................................................... 5-27
5.7.5 Mechanistic modelling ...................................................................................... 5-27
5.7.6 Weibull models ................................................................................................. 5-28
5.7.7 Individual model predictions ............................................................................. 5-30
5.8 Determination of a generalised model ............................................................................... 5-34
5.8.1 Statistical significance check of the generalised model .................................... 5-37
5.8.2 Results of the generalised model prediction ..................................................... 5-38
5.8.3 Check of generalised model using an independent data set ............................ 5-42
5.9 Analysis of equivalent standard axles (ESA) and standard axle repetitions (SAR) ........... 5-47
5.10 Analysis of the load damage exponent .............................................................................. 5-50
5.11 Summary ........................................................................................................................... 5-54
6 DISCUSSION OF THE MODEL .......................................................................................... 6-1

6.1 Introduction.......................................................................................................................... 6-1


6.2 Strengths and weaknesses of the model ............................................................................. 6-2
6.3 Interpretation of the model................................................................................................... 6-2
6.3.1 Bituminous viscoelastic relationship and aggregate reorientation ...................... 6-3
6.3.2 Rest periods ....................................................................................................... 6-5
6.3.3 The relationship between pavements and surfacings ......................................... 6-6
6.3.4 Tyre contact stress ............................................................................................. 6-9
6.4 Explanation of the model outcome ...................................................................................... 6-9
6.5 Using the model and the data to address research gaps .................................................. 6-14
6.5.1 Comparisons of the accumulation of surfacing wear ........................................ 6-14
6.5.2 Comparisons of Equivalent Standard Axle surface wear .................................. 6-16
6.6 Applications for road owners/managers ............................................................................ 6-18
6.6.1 Advantages of preserving surface texture ........................................................ 6-18
6.6.2 Heavy vehicle seal design ................................................................................ 6-18
6.6.3 Regulation gaps for performance-based vehicles............................................. 6-20
6.7 Conclusions ....................................................................................................................... 6-22
7 SUMMARY AND CONCLUSIONS ...................................................................................... 7-1

7.1 Introduction.......................................................................................................................... 7-1


7.2 The effect of heavy vehicle axle groupings on sprayed seal wear....................................... 7-2
7.3 The effect of heavy vehicle axle groupings on sprayed seal design .................................... 7-2
7.4 A load damage component for sprayed seal wear .............................................................. 7-3
7.5 Future work ......................................................................................................................... 7-3

— ix —
7.5.1 Improvements to the model ................................................................................ 7-3
7.5.2 Finite element modelling (FEM) analysis ............................................................ 7-4
7.5.3 The use of cutters for aggregate wetting ............................................................ 7-4
7.5.4 Seal design ......................................................................................................... 7-5
7.6 Conclusions ......................................................................................................................... 7-6
REFERENCES .................................................................................................................... 8-1

APPENDICES .................................................................................................................... A-1

A Seal design and construction detail .................................................................................... A-1


B Seal surface under ALF trafficking ..................................................................................... B-1
C Volumetric sand patch texture data .................................................................................... C-1
D Western Australian prime mover and trailer combinations ................................................. D-1
E Copyright ............................................................................................................................ E-1
F Turnitin originallity check .....................................................................................................F-1

TABLES

Table 1.1: Loads on axle groups with dual tyres which cause same damage as standard
axle ..................................................................................................................... 1-5
Table 1.2: Constraints on experimental variables .......................................................................... 1-8
Table 1.3: Conversion of some key international terms ............................................................... 1-11
Table 2.1: Elastomeric polymer modified binders ........................................................................ 2-24
Table 2.2: Example particle size distribution from a crushing plant ............................................. 2-26
Table 2.3: Australian and New Zealand aggregate size conversion ............................................ 2-27
Table 2.4: Aggregates for sprayed bituminous surfacing ............................................................. 2-28
Table 2.5: Chip seal design methods in North America ............................................................... 2-36
Table 2.6: Factors causing chipseal flushing/bleeding................................................................. 2-49
Table 2.7: Macrotexture measurement methods ......................................................................... 2-73
Table 2.8: Chip seal coefficients .................................................................................................. 2-79
Table 2.9: Macrotexture deterioration models (in chronological order) ........................................ 2-86
Table 2.10: Load damage equivalents per trip for various wheel types ....................................... 2-92
Table 2.11: Loads on axle groups (with dual tyres) which cause the same damage as the
standard axle .................................................................................................... 2-94
Table 2.12: Load damage exponents (LDE) ................................................................................ 2-95
Table 2.13: Published load damage exponents ........................................................................... 2-99
Table 2.14: Australian axle spacings obtained from WIM .......................................................... 2-101
Table 3.1: Weighting factors for surface dressing design .............................................................. 3-4
Table 3.2: Road Note 39 traffic categories .................................................................................... 3-8
Table 3.3: FHWA 13 category classification scheme ................................................................... 3-12
Table 3.4: FHWA vehicle class axle factors ................................................................................. 3-13
Table 3.5: Austroads vehicle classification system ...................................................................... 3-15
Table 3.6: Summary of Australian heavy vehicle fleet based on damage factor (EHVDF) ............ 3-21
Table 3.7: Summary of seal design traffic inputs ......................................................................... 3-22
Table 4.1: Seal design summary ................................................................................................. 4-12
Table 4.2: Composition of Emoprime........................................................................................... 4-13
Table 4.3: Composition of the Emoflex 70™ binder..................................................................... 4-13
Table 4.4: Constraints on experimental variables ........................................................................ 4-17
Table 4.5: Derivation of experimental loading regime .................................................................. 4-19
Table 4.6: ALF trial loading regimes ............................................................................................ 4-24
Table 4.7: Comparison of loading times ...................................................................................... 4-35
Table 5.1: Technicians for sand patch texture measurement ........................................... 5-10
Table 5.2: Initial SPTD readings .................................................................................................. 5-17
Table 5.3: Statistical analysis of individual Weibull model predictions ......................................... 5-32
Table 5.4: Summary of individual Weibull coefficients ................................................................. 5-34
Table 5.5: Statistical analysis of generalised Weibull model predictions.............................. 5-41

— x—
Table 5.6: Main differences between CAPTIF and ALF ............................................................... 5-43
Table 5.7: Comparison of load damage exponents ..................................................................... 5-53
Table 6.1: General effect of the main factors affecting stiffness .................................................... 6-4
Table 6.2: Viscoelastic factors affecting binder stiffness................................................................ 6-5
Table 6.3: Ranking of deterioration effect .................................................................................... 6-10
Table 6.4: Loads which cause the same pavement damage ....................................................... 6-16
Table 6.5: Reduction in texture depth for ESAs carrying the same freight task ........................... 6-17
Table 6.6: Existing equivalent heavy vehicle damage factors...................................................... 6-19

FIGURES

Figure 1.1: Dominant vehicles in each Austroads class................................................................. 1-1


Figure 1.2: Typical commercial vehicle of the 1970s ..................................................................... 1-2
Figure 1-3: Axle configuration nomenclature ................................................................................. 1-3
Figure 1.4: Australian freight efficiency in the 2010s ..................................................................... 1-3
Figure 1.5: Australian multiple axle groupings ............................................................................... 1-4
Figure 1.6: NZ quad axle groups ................................................................................................... 1-4
Figure 2.1: Components of a flexible pavement ............................................................................ 2-2
Figure 2.2: Components of a rigid pavement ................................................................................. 2-3
Figure 2.3: Types of bituminous surfacings ................................................................................... 2-4
Figure 2.4: Schematic illustration of surface dressing types (Continued on next page) ................. 2-5
Figure 2.5: Diagrammatic representation of the colloidal model of bitumen (SOL structure) ......... 2-9
Figure 2.6: Diagrammatic representation of GEL structure.......................................................... 2-10
Figure 2.7: van der Poel nomograph for determining the stiffness of bitumens ........................... 2-12
Figure 2.8: A typical dynamic shear rheometer test on a bituminous binder (at service
temperature and 0< 𝜹𝜹<90 °C) ............................................................ 2-13
Figure 2.9: Relationship between complex modulus and phase angle ................................ 2-14
Figure 2.10: Relationship between 𝛅𝛅, 𝐆𝐆 ∗, 𝐆𝐆′ and 𝐆𝐆′′ ...................................................... 2-14
Figure 2.11: Typical MSCR test results for an 80-100 bitumen at 25⁰C according to
AASHTO T 350 14.......................................................................... 2-17
Figure 2.12: MSCR test for an 80-100 bitumen at 25⁰C showing a single 3.2 kPa stress
creep cycle. ...................................................................................................... 2-18
Figure 2.13: The growth of PME use in Europe ........................................................................... 2-24
Figure 2.14: Microtexture and macrotexture ................................................................................ 2-25
Figure 2.15: Three dimensional shape showing least dimension (A) ........................................... 2-26
Figure 2.16: The void principles in seal design ............................................................................ 2-30
Figure 2.17: Components affecting wettability and adhesion....................................................... 2-38
Figure 2.18: A droplet of bitumen on glass showing the contact angle θ ..................................... 2-39
Figure 2.19: Variations of contact angle ...................................................................................... 2-39
Figure 2.20: Australian & NZ sprayed seal rollers........................................................................ 2-42
Figure 2.21: Void loss due to aggregate embedment .................................................................. 2-45
Figure 2.22: Distress modes for sprayed seals ............................................................................ 2-46
Figure 2.23: Bitumen blisters forming in hot weather ................................................................... 2-49
Figure 2.24: A typical pavement surface (vertical scale exaggerated) ......................................... 2-50
Figure 2.25: The relationship between surface characteristics and wavelength .......................... 2-51
Figure 2.26: Microtexture and macrotexture of a road surface .................................................... 2-51
Figure 2.27: Illustration of texture depth ...................................................................................... 2-52
Figure 2.28: Sprayed seal texture. ............................................................................................... 2-53
Figure 2.29: Known volume of sand poured onto surface............................................................ 2-54
Figure 2.30: Sand spread to form a circular patch ....................................................................... 2-54
Figure 2.31: Illustration of profile depth and mean profile depth .................................................. 2-59
Figure 2.32: Calculating sensor-measured texture depth ............................................................ 2-61
Figure 2.33: Limpet tester ............................................................................................................ 2-63
Figure 2.34: Mineral disc in dynamic shear rheometer ................................................................ 2-64
Figure 2.35: PATTI – ASTM D4541 adhesion test ....................................................................... 2-65

— xi —
Figure 2.36: Bitumen bond strength (BBS) test ........................................................................... 2-65
Figure 2.37: Three-dimensional acquisition systems for non-contact methods based on light
waves ............................................................................................................... 2-67
Figure 2.38: SIC/LCPC experimental image recorder.................................................................. 2-69
Figure 2.39: Photo-texture device ................................................................................................ 2-70
Figure 2.40: Reduction of texture depth with trafficking (NZ) ....................................................... 2-75
Figure 2.41: Reduction of texture depth with trafficking (UK)....................................................... 2-76
Figure 2.42: Texture depth model formats (NZ) ........................................................................... 2-79
Figure 2.43: Change of surface texture with traffic passes .......................................................... 2-82
Figure 2.44: Voids in a chipseal ................................................................................................... 2-82
Figure 2.45: Actual pavement rutting c.f. predicted, for a 80 kN load .......................................... 2-83
Figure 2.46: Actual pavement rutting c.f. predicted, for a 120 kN load ........................................ 2-84
Figure 2.47: Comparison of MPD predicted by Kinder-Lay style model with measured data
for two axle loadings ......................................................................................... 2-85
Figure 2.48: Dispersion of surface loads through a pavement structure ...................................... 2-89
Figure 2.49: Vertical tyre-pavement contact stress profiles of a single 11R22.5 tyre over a
range of loadings and inflation pressures ......................................................... 2-89
Figure 2.50: 1970 Bedford truck with standard axle at rear ......................................................... 2-92
Figure 2.51: Axle configuration nomenclature ............................................................................. 2-93
Figure 3.1: Heavy vehicles in Addis Ababa, Ethiopia, 2009........................................................... 3-4
Figure 3.2: FHWA 13 vehicle classification .................................................................................. 3-11
Figure 3.3: Typical Austroads Class 4 vehicle ............................................................................. 3-19
Figure 4.1: ALF operating indoors at Dandenong South................................................................ 4-4
Figure 4.2: Original ALF schematic................................................................................................ 4-5
Figure 4.3: Typical multiple axle groupings .................................................................................... 4-6
Figure 4.4: ALF with tri-axle assembly fitted .................................................................................. 4-7
Figure 4.5: ALF with tri-axle assembly fitted, showing dual tyres .................................................. 4-7
Figure 4.6: Structure of test pavement........................................................................................... 4-9
Figure 4.7: Ball embedment testing to estimate pavement hardness prior to seal design ........... 4-11
Figure 4.8: Setting out the prime trials ......................................................................................... 4-12
Figure 4-9: Axle configuration nomenclature ............................................................................... 4-19
Figure 4.10: Requirements for a full factorial experimental design .............................................. 4-20
Figure 4.11: Achievable experimental design .............................................................................. 4-21
Figure 4.12: ALF trial loading regimes ......................................................................................... 4-21
Figure 4.13: Original experimental layout within the shed ........................................................... 4-22
Figure 4.14: Final experimental layout ......................................................................................... 4-22
Figure 4.15: ALF profilometer and macrotexture device .............................................................. 4-25
Figure 4.16: Illustration of the ISO terms baseline, profile depth and mean profile depth ............ 4-26
Figure 4.17: Camera setup requirements .................................................................................... 4-28
Figure 4.18: Camera setup .......................................................................................................... 4-28
Figure 4.19: Prototype used to take site photographs ................................................................. 4-28
Figure 4.20: Photographic stereoscopic pairs .............................................................................. 4-30
Figure 4.21: Measurements of the sand circles ........................................................................... 4-32
Figure 4.22: Initial volumetric sand patch texture depth before bedding-in .................................. 4-33
Figure 4.23: Assigned interregional road freight 2005 ................................................................. 4-34
Figure 4.24: Victorian Commercial vehicle breakdown ................................................................ 4-35
Figure 4.25: High shear stresses can be imparted by a prime mover’s driven axles ................... 4-37
Figure 5.1: Experiment 3500 single axle 40 kN raw data............................................................... 5-3
Figure 5.2: Experiment 3501 triaxle 90 kN raw data ...................................................................... 5-4
Figure 5.3: Experiment 3502 triaxle 90 kN raw data ...................................................................... 5-4
Figure 5.4: Experiment 3503 tandem axle 60 kN raw data ............................................................ 5-5
Figure 5.5: Experiment 3504 triaxle 90 kN raw data ...................................................................... 5-5
Figure 5.6: Experiment 3505 tandem axle 80 kN raw data ............................................................ 5-6
Figure 5.7: Experiment 3506 tandem axle 80 kN raw data ............................................................ 5-6
Figure 5.8: Experiment 3507 single axle 40 kN raw data............................................................... 5-7
Figure 5.9: Experiment 3508 tandem axle 60 kN raw data ............................................................ 5-7
Figure 5.10: Experiment 3510 single axle 40 kN raw data............................................................. 5-8
Figure 5.11: Experiment 3511 single axle 40 kN raw data............................................................. 5-8

— xii —
Figure 5.12: Experiment 3512 tandem axle 80 kN raw data .......................................................... 5-9
Figure 5.13: Experiment 3514 tandem axle 80 kN raw data .......................................................... 5-9
Figure 5.14: Combined single axles 40 kN .................................................................................. 5-11
Figure 5.15: Combined tandem axles 60 kN ................................................................................ 5-11
Figure 5.16: Combined tandem axles 80 kN ................................................................................ 5-12
Figure 5.17: Combined triaxles 90 kN.......................................................................................... 5-12
Figure 5.18: All axles all loads ..................................................................................................... 5-13
Figure 5.19: Typical variation in basecourse texture before sealing ............................................ 5-14
Figure 5.20: Single axle 40 kN bedding-in only ........................................................................... 5-15
Figure 5.21: Tandem axle 60 kN bedding-in only ........................................................................ 5-15
Figure 5.22: Tandem axle 80 kN bedding-in only ........................................................................ 5-15
Figure 5.23: Triaxle 90 kN bedding-in only .................................................................................. 5-16
Figure 5.24: All axles all loads bedding-in ................................................................................... 5-16
Figure 5.25: Approximating initial SPTD for single axle 40 kN experiments ................................ 5-18
Figure 5.26: Approximating initial SPTD for tandem axle 60 kN experiments .............................. 5-19
Figure 5.27: Approximating initial SPTD for tandem axle 80 kN experiments .............................. 5-19
Figure 5.28: Approximating initial SPTD for triaxle 90 kN experiments........................................ 5-20
Figure 5.29: Approximating initial SPTD for all experiments from Excel power curve fits ............ 5-21
Figure 5.30: Estimation of initial SPTD (in mm) after 10,000 passes bedding-in ......................... 5-21
Figure 5.31: Normalised surface texture decay (Excel power curves) ......................................... 5-22
Figure 5.32: Exponential decay equation and graph.................................................................... 5-25
Figure 5.33: Exponential decay – effect of constant (k) rates ...................................................... 5-25
Figure 5.34: Examples of functions that may be described by Weibull models ........................... 5-28
Figure 5.35: The effect of the scale parameter α (shown as ŋ) in Weibull models....................... 5-29
Figure 5.36: The effect of the shape parameter β in Weibull models........................................... 5-30
Figure 5.37: Individual Weibull model prediction for single axle 40 kN ........................................ 5-30
Figure 5.38: Individual Weibull model prediction for tandem axle 80 kN...................................... 5-31
Figure 5.39: Individual Weibull model prediction for tandem axle 60 kN...................................... 5-31
Figure 5.40: Individual Weibull model prediction for triaxle 90 kN................................................ 5-32
Figure 5.41: Individual Weibull model predictions ........................................................................ 5-33
Figure 5.42: Weibull α as a function of axles in group ................................................................. 5-35
Figure 5.43: Weibull α as a function of axle load ......................................................................... 5-35
Figure 5.44: Weibull β as a function of axles in group ................................................................. 5-36
Figure 5.45: Weibull β as a function of axle load ......................................................................... 5-36
Figure 5.46: Generalised model prediction for single axle 40 kN................................................. 5-39
Figure 5.47: Generalised model prediction for tandem axle 60 kN .............................................. 5-39
Figure 5.48: Generalised model prediction for tandem axle 80 kN .............................................. 5-40
Figure 5.49: Generalised model prediction for triaxle 90 kN ........................................................ 5-40
Figure 5.50: The effect of the scale parameter α (shown as ŋ) in Weibull models....................... 5-41
Figure 5.51: Schematic of CAPTIF in side elevation.................................................................... 5-42
Figure 5.52: The Canterbury Accelerated Pavement Testing Indoor Facility (CAPTIF) ............... 5-42
Figure 5.53: CAPTIF raw data ..................................................................................................... 5-44
Figure 5.54: Generalised model using CAPTIF (8.2 t simulation) data – using 5,000 passes
bedding-in ......................................................................................................... 5-45
Figure 5.55: Generalised model using CAPTIF (12 t simulation) data – using 5,000 passes
bedding-in ......................................................................................................... 5-45
Figure 5.56: Generalised model using CAPTIF (8.2 t simulation) data – using 9,500 passes
bedding-in ......................................................................................................... 5-46
Figure 5.57: Generalised model using CAPTIF (12 t simulation) data – using 9,500 passes
bedding-in ......................................................................................................... 5-47
Figure 5.58: Equivalent standard axles (dual tyres) ..................................................................... 5-47
Figure 5.59: Comparison of dual tyre ESAs by the generalised model ........................................ 5-48
Figure 5.60: Axle passes to cause equal damage (20% reduction of texture) ............................. 5-51
Figure 5.61: Axle passes to cause equal damage (30% reduction of texture) ............................. 5-52
Figure 6.1: Stiffness vs loading times at different temperatures for two different bitumens ........... 6-3
Figure 6.2: The effect of temperature and loading time on the stiffness of a 100 pen bitumen...... 6-3
Figure 6.3: Superposition of strains under a standard axle............................................................ 6-7
Figure 6.4: Dispersion of surface loads through a pavement structure .......................................... 6-7

— xiii —
Figure 6.5: Stresses beneath a rolling road ................................................................................... 6-8
Figure 6.6: Comparison of dual tyre ESAs by the generalised model (half-axle loads shown) .... 6-10
Figure 6.7: Quad axle load pulse ................................................................................................. 6-11
Figure 6.8: Tri axle load pulse ..................................................................................................... 6-11
Figure 6.9: Single axle load pulse ................................................................................................ 6-11
Figure 6.10: Tandem axle load pulse........................................................................................... 6-12
Figure 6.11: Strain response of a polymer modified binder ......................................................... 6-13
Figure 6.12: The same total freight task carried by both single and tandem axles ...................... 6-15
Figure 6.13: The same total freight task carried by both tandem and triaxles.............................. 6-15
Figure 6.14: Result of a 300,000 t freight task carried by each equivalent standard axle………6-17

— xiv —
Chapter 1 Introduction

1 INTRODUCTION
1.1 Background
Sprayed seals, also known as chip seals, are a thin and flexible pavement surfacing
comprising a layer of bituminous material, usually sprayed at a millimetre or two thick,
and covered with a layer of crushed rock aggregate. These surfacings account for
approximately 90% of the total length of all surfaced roads in Australia, New Zealand
and South Africa. They remain a key pillar in the provision of low cost all-weather road-
transport networks throughout these countries.

Despite their ostensible simplicity, these seals are engineered, with most sprayed seal
design methods world-wide having evolved in some way from Frederick Melrose
Hanson’s landmark mathematical approach in 1935.

A key seal design input, traffic, has long been considered to consist of only two
components: passenger cars and commercial vehicles. This is despite traffic
composition being categorised as 12 classes (in Australia) as shown in Figure 1.1.

Figure 1.1: Dominant vehicles in each Austroads class

Source: Jameson (2012).

— 1-1 —
Chapter 1 Introduction

With respect to commercial vehicles, rigid or ‘body’ trucks were popular in Australia in
the 1950s and 60s, apparently due partly to the availability of heavy-duty ex-military
vehicles (Ramsay & Prem 2000). Hence early Australian sprayed seal designs were
based on the commercial vehicles of the day, as depicted in Figure 1.2.

Figure 1.2: Typical commercial vehicle of the 1970s

Source: State Library of Western Australia image 347113PD.

Prime movers towing semi-trailers (e.g. class 7, 8, 9) only became more widely
accepted later with the introduction of the quick-connect kingpin and turntable (Ramsay
& Prem 2000).

It has been reported that the Australian freight task has quadrupled over the last four
decades, and is projected to nearly double again over the next two decades, with most
of this additional freight expected to be carried by road or rail, and that a key
component of this increase is a drive towards freight efficiency (BITRE 2011). Freight
efficiency in practice has resulted in more class 10, 11 and 12 vehicles than previously
seen.

These vehicles use a variety of axle groupings. For pavement design purposes, the
heavy vehicle axle group types are identified from Jameson (2012) as:

• single axle with single tyres (SAST)

• single axle with dual tyres (SADT)

• tandem axle with single tyres (TAST)

• tandem axle with dual tyres (TADT)

— 1-2 —
Chapter 1 Introduction

• triaxle with dual tyres (TRDT)

• quad-axle with dual tyres (QADT).

This axle configuration nomenclature is shown diagrammatically in Figure 1-3.

Figure 1-3: Axle configuration nomenclature

Source: Neaylon (2012).

Figure 1.4 shows how some of these tandem axles and triaxles are configured in the
2010s to gain efficiency improvements over the 1970s type vehicles.

Figure 1.4: Australian freight efficiency in the 2010s

Source: Neaylon (2015).

It can be seen then that commercial traffic loadings have changed significantly since
seal designs were first engineered in the 1930s. This is particularly so in Australia with
freight efficiency resulting in longer and heavier loads being transported by more
powerful prime movers. A recent ministerial announcement (Nalder 2015) indicates
this trend will continue.

— 1-3 —
Chapter 1 Introduction

In other developed countries these heavier loads would be carried on heavy duty hot
mix asphalt pavements. In the less densely populated countries such as Australia,
New Zealand and South Africa such heavy-duty pavement networks are not
economically feasible, and the capacity of thin flexible pavements surfaced with
sprayed seals must be extended as far as possible.

Figure 1.5: Australian multiple axle groupings

These freight efficient vehicles introduce significantly more tandem and triaxle axle
(Figure 1.5) axle-group loadings than were present in the 1930s and even the 1970s
when seal designs were developed.

Figure 1.6: NZ quad axle groups

Additionally, in New Zealand quad axles are also commonly seen (Figure 1.6).

This then raises a question. What changes have these axle groupings and loads
brought to sprayed seal wear? Have historical sprayed seal design methods stayed
abreast of these changes?

— 1-4 —
Chapter 1 Introduction

1.2 Theoretical framework


Road network owners commonly measure wear on a road surface by the level of
reduction in surface texture. Surface texture is also a major factor in the provision of
skid resistance, and therefore important to road safety.

Therefore it is common that ‘damage’ in terms of a sprayed seal alone (when


considered independent from ‘damage’ in the pavement) is measured and monitored in
terms of change in surface texture.

It is generally accepted that wear on a sprayed seal surface is caused by traffic. The
actual rate of this wear is then moderated by a number of climatic, material property
and loading factors as discussed in chapter 2.

The legal loadings permitted on single, tandem and triaxle groupings have been
calculated such that the multiple axles cause the same pavement damage as the more
lightly loaded single axles. This calculation is undertaken by converting all traffic
loadings to a standard loading unit, called the Standard Axle. The standard axle is
defined as a single axle with dual tyres (SADT) applying an axle load of 80 kN (≈ 8.2 t)
to the pavement (Jameson 2012). When calculating what total traffic loadings a
pavement is designed for, a procedure is used that calculates the damage associated
with each axle grouping compared with the damage caused by a standard axle.
Standard axle repetitions (SARs) for unbound granular pavements with thin surfacings
are calculated with a load damage exponent of 4 (SAR4), and these are commonly
referred to as equivalent standard axles (ESAs)(Jameson 2012). As the work
described herein relates to a granular pavement with a thin bituminous surfacing,
SAR4s are described as ESAs throughout this paper.

Each axle grouping is thus expressed in terms of equivalent standard axles (ESAs), as
shown in Table 1.1. The definition of ESAs is ‘The number of standard axle loads that
are equivalent in damaging effect on a pavement to a given vehicle or axle loading.’
ESAs for unbound granular pavements, where sprayed seals are predominantly used,
use a load damage exponent of 4, i.e. a 4th power function (Jameson 2012).

Table 1.1: Loads on axle groups with dual tyres which cause same damage as standard axle

Axle group type Load (kN)


Single axle with single tyre (SAST) 53
Single axle with dual tyres (SADT) 80
Tandem axle with dual tyres (TADT) 135

— 1-5 —
Chapter 1 Introduction

Axle group type Load (kN)


Triaxle with dual tyres (TRDT) 181
Quad-axle with dual tyres (QADT) 221
Source: Jameson (2012).

But does the concept of ESAs apply to sprayed seal wear? Some practitioners believe
not. The literature review (chapters 2 and 3) finds that:

1. An investigation by an Austroads working group did not find any correlation


between ESAs using the 4th power damage exponent, and seal performance.

2. Seal damage effects are more related to the rolling action of loaded tyres, which
reorientates the aggregate and alters the voids in the aggregate mosaic that are
available to be filled with binder. This action is different to the structural load
carrying capacity of a pavement.

3. Various definitions of pavement ‘damage’ are used in deriving the 4th power law,
such as elastic deformation (deflection), plastic deformation (rutting), and
serviceability index (roughness). These are pavement performance indicators,
not sprayed seal performance indicators.

4. The definition of sprayed seal ‘damage’ is change in macrotexture, more


commonly known in the industry as change in surface texture.

5. A triaxle group and a single axle may have different damaging effects to a
sprayed seal, yet they may both be equal in Equivalent Standard Axle value.

6. A seal damage model must be sufficiently robust to evaluate the relative effects
of heavy vehicles outside of the current Austroads classes but which
predominate on some sealed roads (e.g. quarry roads with class 4 trucks
towing trailers with various types of axle groups).

7. Any value of ESA for sprayed seal use would be different in effect to pavement
design use, as sprayed seal practitioners consider traffic over the early life of a
seal (in NZ it is the first 100 days), not over 20 years.

A literature review has identified gaps in the previous research. Some of the assertions
listed above have no research – rigorous or otherwise - to support them, and no data.

— 1-6 —
Chapter 1 Introduction

Researchers who have studied related topics in the past include: Arnold, Steven.
Alabaster and Fussell (2005a, 2005b, 2005c); Henning (2008); Kinder and Lay (1988);
Pidwerbesky and Arnold (1995); Scala (1970); and Spies (2006a, 2006b, 2006c, 2007).
Scala assisted in deriving the original power relationship, Spies investigated a load
damage approach to including heavy vehicles in sprayed seal design, Pidwerbesky
investigated the seal design for heavy vehicles in forestry work, and Henning proposed
models for surface texture deterioration for use in road asset management software.
Arnold, Steven, Alabaster and Fussell came the furthest in this topic, comparing
surface texture deterioration between two loads on a single tyre single axle.

However, the effects of differing axle groupings and axle loadings has never been
thoroughly considered in sprayed seal design, because these effects have never been
isolated and studied.

The results of these effects may be significant, or they may prove in practice to be
insignificant, but that fact cannot be determined until they are first identified.

1.3 Research purpose


The study of sprayed seal surface texture deterioration is key to this research. The
study must be able to separate the effects of various axle groupings and various axle
loadings. Therefore, the objective of this research is:

To develop a generalised model to predict surface texture deterioration based on the


effect of axle groupings and loadings.

In meeting this objective, an additional benefit will be the determination of a specific


load damage exponent applicable to sprayed seal wear.

Such a generalised model could contribute to a refinement of sprayed seal design,


specifically in the category of heavy vehicle seal design. The results may be used to
improve the seal design of mine haul roads or national freight routes.

— 1-7 —
Chapter 1 Introduction

1.4 Method
Prior to this research work commencing, a program to examine the effect of multiple
axle groupings on unbound granular pavements had already been established and
funded by Austroads. That program was to collect accelerated loading facility (ALF)
data over a three year period at a running cost of about AU$1m per year. With co-
operation, it was possible to gather additional data for the objective of this thesis at no
additional operating cost. Consequently, the method already established for the
pavements study was able to be modified or supplemented to accommodate this
surfacings study.

This had one drawback. Once the accelerated loading experiments started, it was not
practicable to postpone the funded parent project whilst data collection issues that
arose in this child (parasite) project were being corrected. This was a significant
drawback in the philosophy of collecting data from an experiment funded for a different
purpose and with a different aim, and would be an important issue to address should
this work be repeated by others. The data set is imperfect but is still believed to be the
best available at this time.

The reduction of sprayed seal surface texture can be affected by many things and it
was critical to hold many of these factors constant whilst only the variables of axle
grouping and axle loads were altered. This then precluded a methodology where data
was gathered from a number of different field sites around Australia. All the ALF
experiments were conducted undercover on the same pavement with the same
sprayed seal, which enabled the extraneous variables shown in Table 1.2 to be
constrained.

Table 1.2: Constraints on experimental variables

Variable Constraint
Construction material Identical material used for each
experiment
Construction quality All experiments were constructed at the
one time by the one construction
contractor
Air temperature All experiments experienced the same
air temperatures at the same time
Pavement temperatures All experiments were kept shaded, and
all experienced the same pavement
temperatures at the same time.
Humidity All experiments experienced the same
humidity at the same time

— 1-8 —
Chapter 1 Introduction

Variable Constraint
Rainfall All experiments were shielded from
rainfall
Sunlight (UV light) All experiments were shielded from
sunlight
Tyre pressures All experiments were trafficked with tyre
pressures of 760 ±10 kPa
Loading velocity All experiments were trafficked at the
same speed (approx. 22 km/h)
No extraneous axle The exact number of axle grouping
groupings or loadings types and loadings are known for every
experiment

1.5 Thesis structure


The thesis proper commences in chapter 2 with a literature review that focusses on the
wide-ranging and sometimes interacting factors that may affect the performance of a
sprayed seal, and identifies in more detail the current state-of-the-art and the research
gaps still requiring answers.

Chapter 3 follows on from the general literature review with a specific focus on
examining the developments in heavy vehicle seal design in various countries around
the world, and discusses issues with the 4th power law.

Chapter 4 explains in more detail the large number of variables that can influence
sprayed seal wear. Details are given of the experimental plan and how data was
collected using a full scale accelerated loading facility operated for Austroads by ARRB
Group Ltd in Melbourne. This data gathering exercise was very expensive, and also
very time-consuming. Ignoring time taken in experimental establishment, the ALF
machine was operationally dedicated from October 2009 to July 2012 to collect this
and other data sets, at considerable annual Austroads expense. Due to budget and
time constraints, certain limitations were placed on the data collection that was beyond
pragmatic control, and these limitations are discussed.

— 1-9 —
Chapter 1 Introduction

Chapter 5 discusses how the data was cleaned of outliers, normalised, subjected to
multivariate non-linear regression, and analysed. Part of this data analysis includes the
calculation of a load damage exponent applicable to sprayed seal wear on an unbound
granular pavement. A modified Weibull model is then established to predict the
dependent variable for a wider extrapolation beyond the observation set taken. The
model is shown to be validated using an independent data set from a different sprayed-
seal-using country.

In Chapter 6 the new model is used to address some of the knowledge gaps found
through in the literature review.

Chapter 7 closes with some conclusions and recommendations for future work.

1.6 Definitions
At the first International Sprayed Sealing Conference held in Australia in 2008 it
became apparent early into the conference that the sharing of knowledge between
delegates was difficult due to international variations in terminology (Neaylon 2008).
The same words used with different meanings, for example ‘asphalt’, ‘bleeding’, and
‘first coat seal’ were the cause of misunderstandings during presentations and question
times. A sprayed sealing glossary for South Africa, Australia, New Zealand, the UK
and the USA would greatly clarify forays into knowledge exchange.

Gransberg and James (2005) agree, even to the point that within the USA terminology
is different (see Table 1.3). They comment ‘It is suggested that a uniform glossary of
chip seal terms be developed and distributed throughout the (USA) nation. The effect
would be to standardize the technical communication within agencies in the field’.

In Australia, a ‘thin layer of binder sprayed onto a pavement surface with a layer of
aggregate incorporated and which is impervious to water’ (Austroads 2014) is known
as a sprayed seal. In New Zealand and parts of the US they are known as chip seals,
in South Africa they are known as surfacing seals, and in the UK they are known as
surface dressing.

Throughout this thesis the terms sprayed seal, chip seal and surface dressing are often
used interchangeably, but meaning the same.

Table 1.3 lists some other terms which are likely to cause confusion between English
speaking countries.

— 1-10 —
Chapter 1 Introduction

Table 1.3: Conversion of some key international terms


Australia New Zealand North America Republic of United Kingdom
South Africa
Asphalt (asphaltic Asphalt Hot mix asphalt Asphalt Asphalt
concrete)
Bitumen Bitumen Asphalt cement Bitumen Bitumen
Bleeding Flushing Bleeding Bleeding
Industrial diesel fuel Automotive gas oil
(IDF) (AGO)
Double/double seal, Two coat chip Double seal Double seal,
comprising seal, comprising
first coat of binder, comprising tack coat of
second coat of binder first application of binder,
binder, penetration coat
second of binder
application of
binder
Flushing Bleeding Blacking up
Initial seal First coat seal
Kerosene Kerosene Kerosene Paraffin Paraffin
Primerseal First coat seal
Ravelling/stripping Scabbing Shelling Stripping
Reseal Second coat seal
Single/single seal Single seal
Sprayed seal Chip seal Asphalt surface Surfacing seal Surface dressing
treatments (North
Carolina)
Bituminous
surface treatment
(Alabama)
Chip seal
(Kansas)
Surface treatment
(ASTM)
Triaxle Tridem

— 1-11 —
Chapter 1 Introduction

1.7 Acronyms
Acronym Meaning
2D Two dimensional
3D Three dimensional
4WD Four wheel drive
AADT Annual average daily traffic
AAPA Australian Asphalt Pavement Association
AASHTO American Association of State Highway and Transportation Officials
AADT Average annual daily traffic
ADT Average daily traffic
ALF Accelerated (pavement) loading facility (Australia)
ALD Average least dimension of an aggregate particle
APT Accelerated pavement testing
ARRB Australian Road Research Board
ASTM American Society for Testing and Materials
BBS Binder bond strength
BITRE Bureau of Infrastructure, Transport and Regional Economics
(Australia)
BP British Petroleum Ltd
Caltrans California Department of Transportation
CAPTIF Canterbury accelerated pavement testing indoor facility (New
Zealand)
CCD Charge coupled device (the image sensor in a digital camera)
CIRCLY CIRCular loads (i.e. circular contact area) and LaYered systems
COLTO Committee of Land Transportation Officials (South Africa)
CR Crumbed tyre rubber
CSIR Council for Scientific and Industrial Research (South Africa)
CSIRO Commonwealth Scientific and Industrial Research Organisation
(Australia)
CT Computer aided tomography
CV Commercial vehicle (any vehicle with more than two axles or with dual
tyres on the rear axle)
DF Damage factor
DoT Department of Transport(ation)
DSL Design subgrade level
DTEI Department for Transport, Energy and Infrastructure (South Australia)
dTIMS Deighton’s total infrastructure management system
DTMR Department of Transport and Main Roads (Queensland, Australia)
ELV Equivalent light vehicles
EPDM Ethylene-propylene-diene monomer (a terpolymer of ethylene,
propylene and a diene-component)
ESA Equivalent standard axle. The number of standard axle loads that are
equivalent in damaging effect on a pavement to a given vehicle or
axle loading

— 1-12 —
Chapter 1 Introduction

Acronym Meaning
ESAL Equivalent standard axle load (USA) 18,000 lbf (80 kN) on a single
axle with dual tires
ETD Estimated texture depth
EHV Equivalent heavy vehicle
FEM Finite element method or finite element model
FEHRL Forum of European National Highway Research Laboratories
FHWA Federal Highway Administration (USA)
HCV Heavy commercial vehicle
HDM Highway development and management
HV Heavy vehicle (any vehicle with more than two axles or with dual tyres
on the rear axle)
IFSTTAR l’Institut Français des Sciences et Technologies des Transports, de
l’Aménagement et des Réseaux (The French Institute of Science and
Technology for Transport, Development and Networks)
IIR Isobutene-isoprene copolymer rubber (butyl rubber)
INRETS Institut National de Recherche sur les Transports et leur Sécurité
(National Institute for Research on Transport and Safety, France)
ISO International Standards Organisation
IPENZ Institution of Professional Engineers, New Zealand
Laboratoire Central des Ponts et Chaussées (Central Laboratory for
LCPC Roads and Bridges, France, which merged in 2010 with INRETS to
become IFSTTAR)
LDE Load damage exponent
LHV Large heavy vehicle
LTPP Long term pavement performance
MPD Mean profile depth
MSCR Multiple stress creep recovery
MTD Mean texture depth
NAASRA National Association of Australian State Road Authorities
NCAT National Centre for Asphalt Technology (USA)
NCHRP National Cooperative Highway Research Program (USA)
NHVR National Heavy Vehicle Regulator (Australia)
NITRR National Institute for Transport and Road Research (South Africa)
NSW New South Wales (Australia)
NTC National Transport Commission (Australia)
NZTA New Zealand Transport Agency
OECD Organisation for Economic Co-operation and Development
ORN Overseas road note
PAFV Polished aggregate friction value (Australia)
PBD Polybutadiene
PBS Performance based standard
PD Profile depth
Permanent International Assembly of Road Congresses, also known
PIARC as the World Road Association (France)
PMB Polymer modified binder

— 1-13 —
Chapter 1 Introduction

Acronym Meaning
PME Polymer modified emulsion binder
PSV Polished stone value
PTFE Polytetrafluoroethylene
QADT Quad axle with dual tyres
RRL Road Research Laboratory (UK)
RSTA Road Surface Treatments Association (UK)
RTA Roads and Traffic Authority (NSW, Australia)
Standard axle load. Single axle with dual wheels loaded to a total
SAL mass of 8.2 tonne (80 kN).
SABITA South African Bitumen Association
SADT Single axle with dual tyres
SANRAL South African National Roads Agency Limited
SAR Standard axle repetition
SAST Single axle with single tyres
SBR Styrene-butadiene-rubber
SBS Styrene-butadiene-styrene
SEBS Styrene-ethylene-butadiene-styrene
Service d'Études Techniques des Routes et Autoroutes (Department
SETRA of Technical Studies for Roads and Highways, France)
La Section des Fabricants d'Emulsions Routières de Bitume
SFERB (Association of French Road Bitumen Emulsion Manufacturers)
SHRP Strategic highway research program (USA)
SIS Styrene-isoprene-styrene
SLR Single lens reflex
SMTD Sensor measured texture depth
SPSS Statistical package for the social sciences
SPTD Sand patch texture depth
TADT Tandem axle with dual tyres
TNZ Transit New Zealand
TRB Transportation Research Board (USA)
TRDT Triaxle with dual tyres
TRH Technical recommendations for highways (South Africa)
TRL Transportation Research Laboratory (UK)
TRRL Transport and Road Research Laboratory (UK)
L'Union des Syndicats de l'Industrie Routière Française (Trade
USIRF Unions of the French Road Industry)
UTA Ultra-thin asphalt
UTFC Ultra-thin friction course
WIM Weigh-in-motion

— 1-14 —
Chapter 2 Literature review

2 LITERATURE REVIEW
If you want the present to be different from the past, study the past.
Baruch Spinoza, Dutch Philosopher, 1632 – 1677.

2.1 Introduction
Following a brief introduction to pavements, sprayed seals and bituminous surfacings,
this chapter discusses the current understanding of the many factors that may
influence the performance of a sprayed seal: bitumen, polymer modified binders,
aggregate, adhesion, seal construction, aggregate reorientation, and seal failure
modes. It then discusses ways of measuring sprayed seal wear, and the attempts at
predicting sprayed seal wear. It concludes with a review of basic pavement design
principles, and in particular how traffic loadings affect pavement wear.

Sprayed seals are the routine surfacing of choice in South Africa, Australia and New
Zealand, and are also now more widely accepted in the USA. However, these four
countries being great ‘nations divided by a common language 1’, clarification over
terminology is sometimes required.

With regard to the referencing method used in this thesis, whilst adhering to the
Harvard system, a decision was made to make it author-based, rather than institution-
based. For example, the Austroads pavement design method, often referenced as
Austroads (2012) is referenced by its author Jameson (2012). This is done for two
reasons: to give due acknowledgement to the intellect behind the research work; and
because referencing would have otherwise become very cumbersome (e.g. Austroads
2012b, Austroads 2012d). Where individual authors cannot be identified or the work is
substantially that of committee, then the institutional reference (e.g. Austroads 2013)
has been used.

2.2 Road pavements


2.2.1 Road pavement function
The earliest records of paved roads for wheeled traffic are believed to date from
2200 BC in Babylonia (modern Iraq), and progressed to Crete in 1500 BC and Egypt
from about 540 BC. The Romans built their 100,000 km network in Europe between
400 BC and 400 AD (SANRAL 2013).

1
The origin of this phrase is unclear - it has been attributed to both Oscar Wilde and George Bernard
Shaw.

— 2-1 —
Chapter 2 Literature review

The primary function of a road pavement is to allow natural existing ground to carry
vehicle wheel loads. The main function of a pavement is structural: to sustain the
traffic loadings and distribute them to the existing natural surface (known as the
subgrade). A pavement is therefore

a set of superimposed layers of imported materials placed on the


natural soil for the creation of a road (Nikolaides 2015).

Pavements are considered globally as being either flexible or rigid (although in practice
there can be an overlap area called semi-rigid).

Flexible pavements
The Austroads definition of a flexible pavement is

A pavement which obtains its load-spreading properties mainly by


intergranular pressure, mechanical interlock and cohesion between
the particles of the pavement material. In the case of an asphalt
pavement, this further depends on the adhesion between the bitumen
binder and the aggregate, and the cohesion of that binder. Generally,
any pavement in which high strength Portland cement concrete is not
used as a construction layer (Austroads 2015).

The typical nomenclature of the superimposed material layers of a flexible pavement is


shown in Figure 2.1, where DSL is the ‘design subgrade level’. Note also the base and
subbase may contain more than one layer.

Figure 2.1: Components of a flexible pavement

Source: Youdale & Sharp (2007).

Flexible pavements typically have an asphalt or sprayed seal surfacing and are
characterised by higher deflections or bending than is experienced by rigid pavements
(SANRAL 2013). All work in this study is related to flexible pavements.

— 2-2 —
Chapter 2 Literature review

Rigid pavements
A rigid pavement is defined by Austroads as

A pavement composed of concrete or having a concrete basecourse


(Austroads 2015).

The typical nomenclature of the superimposed material layers of a rigid pavement is


shown in Figure 2.2.

Figure 2.2: Components of a rigid pavement

Source: Youdale & Sharp (2007).

2.3 Bituminous road surfacings


2.3.1 Purpose of a surfacing
The primary functions of pavement surfacings are considered to be: to protect the
underlying pavement material from moisture ingress; minimise dust; maintain the
pavement’s ride quality; provide suitable skid resistance; and to provide for reflectivity
and line markings to ensure public safety. It is the pavement (not the surfacing) that
supplies the structural performance to withstand the various loads applied by vehicles
(Sharp 2005).

The provision of skid resistance is of key importance to governments and asset


managers, and this will be explained further in Section 2.7.

2.3.2 Surfacing types


Bituminous surfacings cover a variety of forms, and the groups and subgroups are best
explained as shown in Figure 2.3 where UTA is ultra-thin asphalt and UTFC is ultra-thin
friction course. Although this diagram originates from South Africa, van Zyl included
Australian and New Zealand comment during its early development (van Zyl, personal
communication).

— 2-3 —
Chapter 2 Literature review

Figure 2.3: Types of bituminous surfacings

Source: SABITA (2012).

A sprayed seal in its simplest form is a layer of bitumen covered with a layer of single-
sized crushed rock aggregate.

There are many variations of a sprayed seal beyond its simplest form, and these are
collectively called surface dressings. Convenient illustrations of these types of surface
dressings are found in SANRAL (2007) and are shown in Figure 2.4. The detail of the
double seal actually differs between South Africa, New Zealand and Australia, although
the principle remains similar.

— 2-4 —
Chapter 2 Literature review

Figure 2.4: Schematic illustration of surface dressing types (Continued on next page)

Source: SANRAL (2007).

— 2-5 —
Chapter 2 Literature review

Figure 2.5: Schematic illustration of surface dressing types

Source: SANRAL (2007).

2.3.3 Selection of surfacings


The choice of which type of surfacing is most appropriate for which condition is a wide
ranging topic, and can be influenced by available materials, personal experiences and
local cultures.

Cape seals are a good example of cultural differences. Cape seals consist of a 13 or
19mm single seal covered with a bituminous slurry seal or microsurfacing. The binder
application rate used in the single seal is the minimum necessary to hold the aggregate
during construction. In South Africa the slurry is generally applied by hand. This is for
two reasons: firstly, due to high unemployment levels labour-based methods are
preferred; and secondly it is possible to force the slurry between the sealing aggregate
so that the tops of aggregate are exposed (AAPA 2011).

— 2-6 —
Chapter 2 Literature review

The use of labour-based techniques is not considered desirable in Australia where


labour is relatively expensive and mechanised methods are preferred. However,
construction of a cape seal with an Australian mixing machine and spreader box results
in a slightly different cape seal which is then appropriate for slightly different treatment
selection conditions (Neaylon & Busuttil 2001).

To assist in documenting and recording successful experience in the selection of


surfacings, various handbooks and manuals have been created in countries around the
world that use sprayed seals to provide guidance on local practices and conditions.
However, the choice of an appropriate surfacing will generally depend on the following
factors (Austroads 2013; SANRAL 2007):

· traffic volume and composition

· traffic actions (i.e. turning, cornering, accelerating)

· vertical gradient of the road

· availability of aggregate

· maintenance capability

· surface texture required (related to skid resistance requirements)

· construction techniques available

· environmental and climatic conditions

· quality of the basecourse material

· isolation of the particular site and cost considerations.

For the experimental work described later in this study, a two coat or ‘double’ seal was
selected. This was selected primarily on the basis of the heavy traffic loadings
anticipated and also because the maintenance capability for the experiment was
limited. Two coat seals are regarded as being a robust and forgiving treatment.

— 2-7 —
Chapter 2 Literature review

2.4 Sprayed seal materials


2.4.1 Bitumen
Origins
The first uses of bitumen correspond to surface seepage of ‘natural’ bitumen, usually at
geological faults in the vicinity of subterranean crude oil deposits. Refined bitumen is
manufactured from the fractional distillation of crude oil. It is generally agreed that
crude oil has its origins in marine organisms and vegetable matter deposited onto
ocean beds. Conversion of these organisms and vegetable matter into the
hydrocarbons comprising crude oil is thought to be the result of pressure from
accumulated layers of sediment some hundreds of metres thick, and heat from within
the earth’s core (Read & Whiteoak 2003).

Chemistry
As described in Read and Whiteoak (2003) and BP Bitumen Australia (2014), bitumen
is a non-crystalline, black or dark brown viscous material, consisting essentially of
hydrocarbons (80% carbon and 15% hydrogen 2) with the remainder comprising
oxygen, sulphur, nitrogen and traces of various metals such as vanadium, nickel, iron,
magnesium and calcium.

The chemical composition of bitumen is extremely complex, with a complete and


accurate analysis considered impossible. However, it is possible to separate bitumen
into two broad chemical groups: asphaltenes and maltenes. Bitumen is traditionally
regarded as a colloidal system with high molecular weight asphaltene micelles
dispersed in a lower molecular weight oily medium called maltenes. The maltenes can
be subdivided further into saturates, aromatics and resins, although all these groupings
are not well-defined and there may be some overlap.

Asphaltenes constitute 5 to 25% of bitumen and are generally considered to be highly


polar, complex aromatic materials of high molecular weight. An increase in asphaltene
content gives a harder, more viscous bitumen with a lower penetration. Asphaltenes
can be regarded as black solids which are a colloidal dispersion in the oily yellow-
brown maltenes liquid.

Resins, composed of hydrogen and carbon, are the dispersive agent or peptisers for
the asphaltenes – they keep the asphaltenes in suspension.

2
It is not clear whether these % are w/w or v/v.

— 2-8 —
Chapter 2 Literature review

Aromatics represent the major proportion of the dispersion medium for the peptised
(digested) asphaltenes. They consist of low molecular weight, non-polar carbon chains
in which the unsaturated ring systems (aromatics) dominate. They have a high ability
to dissolve other high molecular weight hydrocarbons.

Saturates are non-polar viscous oils consisting of straight and branch chain aliphatic
hydrocarbons with alkyl-naphthenes and some alkyl-aromatics.

If there are sufficient quantities of resins and aromatics and the asphaltenes are fully
peptised, the resulting micelles have good mobility and these are known as SOL 3 type
bitumens. SOL bitumens are illustrated in Figure 2.5.

Figure 2.5: Diagrammatic representation of the colloidal model of bitumen (SOL structure)

Source: Oliver (2006).

If the aromatic/resin fraction is weak, through either insufficient quantity or insufficient


solvating power, the asphaltenes can associate further together. Alternatively, the
same result can be obtained by air blowing of bitumen at the refinery, which produces
oxidised bitumen of harder grades, by converting some of the low molecular weight
maltenes into higher molecular weight asphaltenes (Oliver 2006; Read & Whiteoak
2003). These bitumens are known as GEL 4 types, as shown in Figure 2.6.

3
SOL: A suspension of solid particles of colloidal dimensions in a liquid (Chambers 1991).
4
GEL: The apparently solid, often jellylike material formed from a colloidal solution on standing (Chambers
1991).

— 2-9 —
Chapter 2 Literature review

Figure 2.6: Diagrammatic representation of GEL structure

Source: Oliver (2006).

In practice however, most bitumens are believed to be of some intermediate type


between GEL and SOL (Read & Whiteoak 2003).

Viscoelasticity and thermoplasticity


Bitumen can be regarded as a civil engineering construction material, however its
mechanical properties are far more complex that steel or concrete (Read & Whiteoak
2003). Bitumen has also been described as a viscoelastic and thermoplastic adhesive
(Dickinson 1985).

A viscous material is a material which is able to flow. When stressed it will tend to
deform, with that deformation being unrecoverable after the loading stress is removed
(this is the same as other materials being defined as plastic). Elastic materials also
tend to deform under stress, however when the loading is removed this deformation is
fully recovered (Read & Whiteoak 2003). In the classical theory of elasticity, the stress
is always proportional to strain (deformation) but independent of strain rate. However,
bitumen deviates from this, as it shows a nonlinear behaviour between stress and
strain. In addition, the stress may also depend on the rate of strain and the length of
time the strain is applied. Such time anomalies reflect a behaviour that is both liquid-
like and solid-like, and is described as viscoelastic. The thermoplastic description of
bitumen arises because this viscoelastic behaviour is affected by temperature
(Dickinson 1985).

— 2-10 —
Chapter 2 Literature review

Thus the degree to which bitumen’s behaviour is viscous and elastic is a function of
both the loading time and the temperature. At high temperatures or long loading times
bitumen behaves as a viscous (flowing) liquid, whereas at very low temperatures or
short times of loading, bitumen behaves as an elastic (brittle) solid (Read & Whiteoak
2003).

Determination of the stiffness modulus of bitumen


Pioneering work to define the visco-elastic properties of bitumen was published by van
der Poel (1954). He introduced the concept of a stiffness modulus for bitumen, as an
analogy with the elastic modulus of solids. His work concluded that the stiffness
modulus of bitumen depends on four variables: 1) time or frequency of loading (the
strain rate); 2) temperature; 3) hardness of the bitumen; and 4) rheological type of
bitumen.

He also constructed a nomograph (Figure 2.7), still sometimes used today, in which

The stiffness of any bitumen at any temperature and time of loading


can be read if the ring-and-ball temperature and the penetration index
are given. The accuracy of this nomograph, which covers a
temperature range of 300 °C, is amply sufficient for engineering
purposes. (van der Poel 1954).

Rheology
Methods for determining the stiffness modulus of bitumen are now based on
rheological science of shear deformation, and typically use a dynamic shear rheometer
(DSR) (Read & Whiteoak 2003). Rheology can be defined as the science that deals
with the flow and deformation of fluids and solids (Austroads 2014).

Bitumen’s resistance to shear is expressed in terms of the shear modulus as shown in


equation 2.1.

𝜏𝜏 2.1
𝐺𝐺 =
𝛾𝛾

where

𝐺𝐺 = shear modulus

𝜏𝜏 = shear stress

𝛾𝛾 = shear strain.

— 2-11 —
Chapter 2 Literature review

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.7: van der Poel nomograph for determining the stiffness of bitumens

Source: van der Poel (1954), figure 6.

The elastic modulus and shear modulus are related by equation 2.2.

𝐸𝐸 = 2(1 + 𝜇𝜇)𝐺𝐺 2.2

where

𝐸𝐸 = elastic modulus

𝜇𝜇 = Poisson’s ratio

𝐺𝐺 = shear modulus

Poisson’s ratio depends on the compressibility of the material, and as bitumen is


virtually incompressible (Read & Whiteoak 2003), then 𝜇𝜇 = 0.5 and

𝐸𝐸 ≈ 3𝐺𝐺 2.3

where

𝐸𝐸 = elastic modulus

𝐺𝐺 = shear modulus.

Shear stresses can be determined statically in a creep test or dynamically using a DSR
applying a sinusoidal load at a given frequency 𝑓𝑓. Under sinusoidal loading the strain
response of bitumen will lag behind the stress that is applied, resulting in a phase angle
difference, 𝛿𝛿 (Figure 2.8).

— 2-12 —
Chapter 2 Literature review

Figure 2.8: A typical dynamic shear rheometer test on a bituminous binder (at service temperature and
0< 𝜹𝜹<90 °C)

Source: National Highways Institute (2005).

Remembering that van der Poel determined that the stiffness of bitumen depends on
the frequency of loading, then a complex modulus 𝐺𝐺 ∗ will more completely describe the
visco-elastic properties of a bituminous binder.

The complex modulus is actually a complex vector comprising real and imaginary
components (Dickinson 1985; Read & Whiteoak 2003; Tredrea 2007). The magnitude
of this vector can be calculated as

𝐺𝐺 ∗ = �(𝐺𝐺′′)2 + (𝐺𝐺′)2 2.4

where

𝐺𝐺 ∗ = complex shear modulus

the loss modulus (viscous component, dependent on


𝐺𝐺′′ =
temperature and strain rate)

the storage modulus (elastic component, dependent on


𝐺𝐺′ =
temperature and strain rate).

By illustrating this in vector form the phase angle 𝛿𝛿 can be reintroduced, resulting in
Figure 2.9 and Figure 2.10.

— 2-13 —
Chapter 2 Literature review

Figure 2.9: Relationship between complex modulus and phase angle

Source: National Highways Institute (2005).

It can be seen from Figure 2.9 and Figure 2.10 that a phase angle of 0° would describe
a purely elastic material, and a phase angle of 90° would describe a purely viscous
material.

𝐺𝐺 ∗= complex shear modulus


𝐺𝐺′′ = viscous component

phase angle 𝛿𝛿

𝐺𝐺′ = elastic component

Figure 2.10: Relationship between 𝛅𝛅, 𝐆𝐆∗ , 𝐆𝐆′ and 𝐆𝐆′′

After Read & Whiteoak (2003).

Using Figure 2.10 and relating back to phase angle 𝛿𝛿, it can be also be seen that

— 2-14 —
Chapter 2 Literature review

𝐺𝐺′′ = 𝐺𝐺 ∗ 𝑆𝑆𝑆𝑆𝑆𝑆(𝛿𝛿) 2.5

where

the loss modulus (viscous component, dependent on


𝐺𝐺′′ =
temperature and strain rate)
and

𝐺𝐺′ = 𝐺𝐺 ∗ 𝐶𝐶𝐶𝐶𝐶𝐶(𝛿𝛿) 2.6

where

the storage modulus (elastic component, dependent on


𝐺𝐺′ =
temperature and strain rate).

These two properties are used in bitumen performance grading (PG) systems. At low
service temperatures, the loss modulus has been identified as a key factor in
controlling asphalt fatigue. At high service temperatures, the storage modulus is a key
factor in controlling asphalt deformation (Tredrea 2007).

Rutting parameter
Asphalt rutting is assumed to be primarily a result of surface layer deformations, and is
considered a stress-controlled, cyclic-loading phenomena at high service temperatures.
Work has to be done to deform the asphalt layer. With each loading cycle a proportion
of this work is recovered in elastic rebound, while the remaining work is dissipated
through permanent deformation and heat. It can be shown (Pumphrey 2003) that for a
viscoelastic material the work dissipated per loading cycle is calculated using Equation
2.7

1 2.7
𝑊𝑊𝑐𝑐 = 𝜋𝜋(𝜎𝜎0 )2 � �
𝐺𝐺 ∗⁄sin 𝛿𝛿

where

𝑊𝑊𝑐𝑐 = work dissipated per loading cycle

𝜎𝜎0 = cyclic stress

𝐺𝐺 ∗ = the complex shear modulus

𝛿𝛿 = phase angle.

— 2-15 —
Chapter 2 Literature review

Note that equation 2.7 does not follow conventional engineering notations, as G* is a
shear modulus whilst σ is traditionally used to indicate normal stresses (Zaniewski and
Pumphrey 2004). Normal notation would substitute shear stress ԏ0 for cyclic stress σ0

The relationship G*/sin δ was chosen as the parameter for the US strategic highway
research program (SHRP) specifications with respect to rutting, and was specified in
AASHTO standard specification for performance–graded asphalt binder, M-320 (2005
version), as the high-temperature binder test.

However, difficulties were reported in the application of this parameter to polymer


modified binders (due to the test’s low strain levels), and a replacement test for PMBs
that was both performance based and blind to modification type, was required. The
multi-stress creep and recovery (MSCR) test (AASHTO 2014a) was developed as a
replacement for the older high-temperature binder test in the PG method (AASHTO
2010: M-320) and has been shown to be more discriminating in identifying the rutting
potential of both modified and neat binders (D’Angelo 2010; FHWA 2011).

The MSCR has been used to test and successfully rank neat, styrene butadiene
styrene (SBS), styrene butadiene (SB), elvaloy, crumb rubber, latex and chemically
modified binders. In the original draft proposal, the test used the one second creep
loading with nine second recovery over multiple stress levels, at ten cycles per stress
level. The test started with the lowest stress levels, and then increased stress in
stages up to the highest level at every ten cycles with no time lag between. The
average non-recovered strain of the ten creep and recovery cycles is divided by the
applied stress to determine the non-recoverable compliance, Jnr (D’Angelo 2010;
FHWA 2011).

However, for the updated 2014 method (AASHTO 2014b), only two stress levels,
0.1 kPa and 3.2 kPa are used. Ten creep and recovery cycles are tested at each
stress level (illustrated in Figure 2.11).

— 2-16 —
Chapter 2 Literature review

Loading portion of cycle

Recovery portion of cycle

Figure 2.11: Typical MSCR test results for an 80-100 bitumen at 25⁰C according to AASHTO T 350 14.

Source: Herrington & Wu (2016).

A simplified definition of Jnr is given in equation 2.8 (D’Angelo 2010).

𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝛾𝛾𝑢𝑢 2.8


𝐽𝐽𝑛𝑛𝑛𝑛 =
𝜏𝜏

where

𝐽𝐽𝑛𝑛𝑛𝑛 = creep compliance (non-recoverable)

unrecovered strain from the end of the 9 second recovery


𝛾𝛾𝑢𝑢 =
portion of the creep and recovery test (see Figure 2.12)

shear stress applied during the 1 second creep portion of the


𝜏𝜏 =
creep and recovery test.

— 2-17 —
Chapter 2 Literature review

Figure 2.12: MSCR test for an 80-100 bitumen at 25⁰C showing a single 3.2 kPa stress creep cycle.

Source: Herrington & Wu (2016).

It can be seen from equation 2.8 and Figure 2.12 that as Jnr is strain divided by stress,
it is analogous to the inverse of modulus (stress divided by strain).

Although the MSCR test does not appear immediately relevant to chip seals,
discussion of it has been included because Hanz and co-workers (Hanz et al. 2012;
Hanz et al. 2013), suggested that the MSCR test could be included in a chip seal
bitumen specification to control bleeding and flushing, being two failure mechanisms of
sprayed seals (refer to Section 2.6). They suggest that the test would be carried out at
52 °C and 64 °C and at three stress levels (0.1, 3.2 and 10 kPa), in contrast to the
standard two levels. Testing at two temperatures would allow the temperature
sensitivity of the bitumen to be assessed.

The MSCR test result has also recently been found to correlate to resistance to shear
loads in asphalt (White 2015a,2015b), which suggests a potential may exist of a
correlation to resistance of aggregate reorientation in sprayed seals.

— 2-18 —
Chapter 2 Literature review

Fatigue cracking parameter


Fatigue cracking of asphalt pavements is the primary distress mode at intermediate
service temperatures. For thin asphalt pavements (< 50 mm) it is considered a strain
controlled distress, and for thick asphalt pavement layers (> 150 mm) it is considered a
stress controlled distress (Pumphrey 2003). Fatigue cracking occurs primarily in thin
pavement layers, therefore it is modelled as strain-controlled, with the work dissipated
per loading cycle shown to be as per equation 2.9 (Pumphrey 2003).

𝑊𝑊𝑐𝑐 = 𝜋𝜋(𝜎𝜎0 )2 [𝐺𝐺 ∗ sin 𝛿𝛿] 2.9

where

𝑊𝑊𝑐𝑐 = work dissipated per loading cycle

𝜎𝜎0 = cyclic stress

𝐺𝐺 ∗ = the complex shear modulus

𝛿𝛿 = phase angle.

The G*sin δ relationship was therefore selected as the cracking parameter for the
strategic highway research program (SHRP) specification.

2.4.2 Cutback bitumen


The Australian Concise Oxford dictionary (Oxford 1987) gives a common meaning (i.e.
non-engineering) of ‘to cut’ as to ‘dilute or adulterate’ - which describes in simplistic
terms what ‘cutters’ do to bitumen.

Cutter oils are light petroleum solvents, typically kerosene type products (e.g. jet fuel
A1 and high flashpoint kerosene). Cutter oils are added to hot bitumen on site as an
aid to construction in quantities that are dependent upon the varying ambient
conditions of air and pavement temperature, wind, sun, shade, aggregate condition and
traffic volumes. The aim of cutting is to temporarily reduce the binder viscosity to a
level sufficient to promote wetting and adhesion of the aggregate once the binder has
been sprayed. By lowering the binder viscosity, cutters then assist with aggregate re-
orientation during the early life of the seal.

— 2-19 —
Chapter 2 Literature review

A short discussion of the in-service life of cutters is now relevant because they are one
of the compounding factors influencing aggregate orientation, and hence bear a
relationship to the hypothesis of this study.

Traditionally, cutters were assumed to evaporate quickly and therefore are not included
in residual bitumen design calculations for sprayed seals. However, this assumption
has been progressively challenged. Donbavand (1984) used thermogravimetry to
determine that in the New Zealand environment, 55% of the original amount of
kerosene cutter remained after up to eight months of service, and 45% remained after
up to 20 months. Dickinson (1989) undertook a detailed study in Australia and found
through wind tunnel testing that bitumen crude source and processing route had a
significant effect on cutter retention, and that for some trials laid under warm conditions
80% of the original cutter remained after three days. Ball (1992) and Herrington and
Ball (1994) found that in trials in NZ 85% of the cutter remained after three minutes and
20% remained after 5 years. Meydan (1997) believed that the sensitivity of the
analytical methods used by Donbavand (1984) to measure residues of hydrocarbons
from sprayed seal binder samples was low, and methods had improved since then.
Part of Meydan’s work involved a field trial, where a sprayed seal was laid on the Great
Ocean Road using 10 parts cutter:100 parts Bitumen (by volume). After 20 weeks
about 33% of this cutter remained and after 50 weeks about 13% remained.

The body of work just described indicates that cutters can remain residual in a binder
for a significant time. By doing so they can assist the reorientation of aggregate
particles when under traffic. This then has an influence on the rate of macrotexture
reduction under trafficking.

Recent research into cutter effectiveness has been based on polymer modified binders,
where the hypothesis of polymer modified binders (PMBs) requiring a higher cutter
content than Class 170 bitumen has been tested (PMBs are described in section
2.4.4). This work started with Maccarrone et al. (1997), Oliver (2011), and continued
more recently with Urquhart (2012). Urquhart concluded, however, that binder
viscosity is not the only factor that controls the wetting of aggregates - the other main
factor is likely to be binder surface tension, as this property controls the degree of
wetting of low viscosity liquids on solid surfaces. Further research is required in this
area. This point does not directly affect the work done in this research and the
resulting model, because an emulsified PMB was used rather than a cutback PMB.

— 2-20 —
Chapter 2 Literature review

Australia is ‘blessed’ with relatively high pavement temperatures, with 70 °C being


common in South Australia (authors’ personal experience), resulting in cases where
hot bitumen can be sprayed without the use of cutters, or very minimal use of cutters.

However, the use of cutback bitumen has been significantly restricted or even
prohibited in many parts of the United States (Tam 2006). A number of reasons have
been cited, such as:

· increasing environmental regulations and concerns (the release of


petrochemical hydro-carbons into the atmosphere)

· economic (use of two petroleum products and the need for additional heat)

· safety (risk of burns, injuries and low flash point explosions).

Bitumen emulsions have therefore increasingly been used in the US to substitute for
hot cutback bitumen. Bitumen emulsions are the product of choice in France (SFERB
& USIRF 2008), and are becoming more favoured in New Zealand (Waters 2013).

2.4.3 Bitumen emulsion


Bitumen emulsions 5 simply put are dispersions of bitumen in water. Hot bitumen,
water, emulsifier (a surfactant) and phosphoric acid (to control PH and hence stability)
are combined in a high-shear colloid mill to disperse the bitumen in water as small
droplets 0.1 to 5 µm in diameter. Bitumen emulsions normally comprise between 20%
and 80% (in the case of ‘inverted’ emulsions – water dispersed in bitumen) by volume.
The common concentration in Australia has been 60% bitumen. Bitumen emulsions
have a low viscosity at ambient temperatures compared to the source bitumen, and
their application involves controlled ‘breaking’ and ‘setting’ (BP Bitumen Australia 2014;
Gorman 2000; Gorman et al. 2004).

la Section des Fabricants d'Emulsions Routières de Bitume (SFERB) and l'Union des
Syndicats de l'Industrie Routière Française (USIRF)(2008), amongst other references,
quote a great many advantages of bitumen emulsions compared to hot binders
(cutback, fluxed bitumens), such as:

5
The word emulsion is derived from the Latin verb emulgere which means ‘the action of milking a cow’ –
milk itself being a complex emulsion system (Gorman 2000).

— 2-21 —
Chapter 2 Literature review

· Being applied at a temperature of 50 °C to 80 °C their use requires less energy


than hot binders.

· Storage facilities and spraying equipment are simpler.

· There are no explosion dangers or toxicity hazards during handling.

· There is less sensitivity to climatic conditions, thus enabling the chip seal
working season to be extended from the beginning of spring to the end of
autumn.

· Final properties are achieved not by evaporation of a solvent but by chemical


breaking between the bitumen and aqueous phase, making it more user-friendly
when applied during cold and wet periods.

· Emulsion does not solidify like hot and cutback binder when it comes into
contact with the substrate (some Australians see this as a disadvantage when
sealing roads with steep grades, as the emulsion proceeds to flow downhill).

· They provide maximum adhesion to aggregate due to excellent wettability


during the aqueous phase.

· They contain little or no cutter/flux, and do not soften the substrate, thus
preventing punching of the aggregates and bleeding on the surface (anecdotal
reports indicate emulsions may still contain cutter, however SFERB & USIRF
(2008) is the only written verification found in the public domain).

la Section des Fabricants d'Emulsions Routières de Bitume (SFERB) and l'Union des
Syndicats de l'Industrie Routière Française (USIRF) exist to promote emulsion use,
and accordingly are silent on any disadvantages of emulsions.

— 2-22 —
Chapter 2 Literature review

Disadvantages to emulsions appear to be rarely published. However, in Australia at


least, there must be at least some perceived disadvantages, as efforts have been
made to encourage emulsion use for some decades (e.g. Neaylon 1995) with little
impact. A detailed study of the environmental impact of emulsions was undertaken by
Austroads and the Australian Commonwealth Scientific and Industrial Research
Organisation (CSIRO) in 2000 (Leach & Beer 2000) where the disadvantages of
increased use of emulsions were found to be a) the cost and emissions generated to
transport 40% of waste product (60% bitumen emulsion contains 40% water) over long
distances, and b) the additional energy required (and therefore cost) in the production
of emulsion. Other perceived disadvantages of conventional emulsions include seal
fragility during the long break and cure times (Remtulla 2015).

In closing, it must be remembered that bitumen emulsion and cutback bitumen are in
fact simply delivery systems (Rebbechi 2008) for construction purposes - the temporary
addition of water or kerosene to either system is used to lower viscosity for the delivery
of the residual bitumen as required by the seal design.

2.4.4 Polymer modified bitumen (PMB)


Modified bitumens are those where the rheological properties are modified by the use
of one or more chemical agents. This modification alters and improves certain bitumen
properties such as temperature susceptibility, stiffness, elasticity, adhesion and aging,
and these altered properties have been related to improved performance. The need for
improved performance has arisen from rapid increases in traffic volumes and axle
loadings (Nikolaides 2015). Some governments, as the asset owner, are also
prioritising budgets 6 towards value for money and are expecting significant efficiency
gains in the delivery of services (e.g. NZ Government 2014) which in turn also
encourages improved performance.

The most common of these modifiers are polymers, and these may be grouped into
elastomeric polymers, plastomeric polymers, or thermosetting resins.

Elastomeric polymers are the most commonly used modifiers in sprayed sealing, and
examples of these are given in Table 2.1, adapted from Read & Whiteoak (2003).

6
Many budgets are insufficient in addressing the growing maintenance backlog of developed countries.

— 2-23 —
Chapter 2 Literature review

Table 2.1: Elastomeric polymer modified binders

Common
Type
name
Styrene-butadiene-styrene SBS
Styrene-butadiene-rubber SBR
Styrene-isoprene-styrene SIS
Styrene-ethylene-butadiene-styrene SEBS
Ethylene-propylene-diene monomer (a terpolymer) EPDM
Isobutene-isoprene copolymer rubber (butyl rubber) IIR
Natural rubber latex
Crumb tyre rubber CR
Polybutadiene PBD
Polyisoprene none

2.4.5 Polymer modified bituminous emulsion (PME)


Polymer modified emulsions combine the advantages of polymer modification and the
advantages of using bitumen emulsion, into the one delivery system.

Although the market for PMEs in sprayed seal is small in the Southern Hemisphere, as
is the market for emulsions in general, PMEs are better developed in the Northern
Hemisphere. In Germany 80% of the chip/sprayed seals use emulsion, with a growing
share of modified emulsions. In the UK 80% of chip/sprayed seals are made with
modified emulsions, as shown in Figure 2.13.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.13: The growth of PME use in Europe

Source: le Bouteiller (2012), slide 31.

A solution for the historical resistance to emulsion in the southern hemisphere will likely
include development of new products, such as high residue polymer modified bitumen
emulsions (Remtulla 2015) and thixotropic emulsions that exhibit low viscosity under
the high shear environments of pumping and spraying, and high viscosity under the low
shear environment of sitting on the road (Road Science 2015a, 2015b).

— 2-24 —
Chapter 2 Literature review

For the experimental work described in this study, a polymer modified emulsion was
selected. This was primarily on the basis of the heavy traffic loadings anticipated,
which required some form of modified binder, and emulsion was chosen over a hot cut-
back product because for such a relatively miniscule and specialised sealing contract
an emulsion proved to be the cheaper and more practical option.

2.4.6 Aggregate
The final fundamental ingredient of a sprayed seal is aggregate.

The aggregate or sealing chip exists to provide skid resistance for vehicles using the
road, which it does through the provision of microtexture and macrotexture as shown in
Figure 2.14.

Figure 2.14: Microtexture and macrotexture

After Flintish et al. (2003); Austroads (2005a).

Whilst being introduced here, a discussion of microtexture and macrotexture is


continued further within Section 2.7.

Aggregates also support wheel contact stresses for transmission to the pavement and
protect the waterproofing bitumen layers from traffic wear. Sealing aggregates are
usually manufactured from crushed rock, as the crushing process provides sharp
edges (giving macrotexture) and fracturing of the crystalline structure of rock (giving
microtexture) (Austroads 2005a; Transit New Zealand et al. 2005).

It is important that sealing aggregates are ‘single sized’ which then leads to the
establishment of some definitions:

— 2-25 —
Chapter 2 Literature review

Nominal size: The concept of nominal size aggregate is for convenience of reference
and ordering. The nominal size is expressed as a whole number above the smallest
sieve aperture size, through which 90% of the aggregate passes (Dickinson 1985).
Table 2.2 shows an example of this.

Table 2.2: Example particle size distribution from a crushing plant


Australian nominal size: 14 10 7 5
Australian standard sieve range (mm) % Mass
Passing Retained
19.0 13.2 4
13.2 9.5 74 13
9.5 6.7 18 59 10
6.7 4.75 2 20 48 12
4.75 2.36 1 7 40 71
2.36 1.18 - - 14
After Dickinson (1985).

Single sized aggregate: An aggregate of which at least 60% of the total material
passes a sieve (from the set) which is immediately less than the nominal size and is
retained on the sieve immediately following (Standards Australia 2009a). It can be
seen from bold figures in the example above that this quarry’s size 10 and size 7 does
not meet the Australian standard.

Average least dimension: In sprayed seal design the concept of an aggregate particle
lying with its least dimension vertical is central to volumetric seal design. The least
dimension is defined as the smallest dimension of a particle when placed in its most
stable position on a flat surface (Alderson 2006), as shown in Figure 2.15.

Figure 2.15: Three dimensional shape showing least dimension (A)

Source: Alderson (2006).

— 2-26 —
Chapter 2 Literature review

Conversion between Australian and New Zealand aggregate sizes: New Zealand uses
‘grades’ of aggregate specified by a mixture of either average least dimensions or sieve
gradings. The following conversion (Table 2.3) has been derived from information
found in Standards Australia (2009b) and Transit NZ et al. (2005).

Table 2.3: Australian and New Zealand aggregate size conversion


New Zealand Australia
ALD (mm) NZ Grade Aus nominal size ALD (mm) Sieve grading (mm)
9.5 – 12.0 2 16 Min 8 -16.0 + 9.5-
7.5 – 10.0 3 14 Min 7 -13.2 +9.5
5.5 – 8.0 4 10 Min 5 -9.5 + 6.7
Grading spec 5 7 Min 3.5 -6.7 +3.35
Grading spec 6 5 - -4.75 + 2.36

Note also that there is some overlap in the NZ system, allowing a large grade 3
aggregate to be a small grade 2, or a small grade 2 to be a large grade 3.

The requirements for sealing aggregates to provide skid resistance, be tough under
heavy axle loads, and to be a uniform shape means that they are tightly specified. The
aggregate properties commonly specified in Australia are shown in Table 2.4.

— 2-27 —
Chapter 2 Literature review

Table 2.4: Aggregates for sprayed bituminous surfacing


Purpose Property Detail Australian Standard test
Dimensional requirements Grading Particle size distribution by AS 1141.11.1
sieving
Particle size distribution by AS 1141.11.2
vision sizing systems
Specified size AS 2658.2 table 1
Tolerance on oversize AS 2658.2 table 2
Tolerance on undersize AS 2658.2 table 3
Particle shape AS 1141.14
Flakiness index AS 1141.15
Average least dimension ALD: direct measurement, AS 1141.20.1
Size 10 mm and greater
ALD: direct measurement, AS 1141.20.2
Size 5 mm & 7 mm
ALD: nomograph method AS 1141.20.3
Crushed faces AS 1141.18
Durability Wet strength and wet/dry AS 1141.22
strength variation
Abrasion Los Angeles abrasion loss AS 1141.23
Sodium sulphate soundness AS 1141.24
Unsound stone content AS 1141.30.1
Weak particles AS 1141.32
Resistance to stripping AS 1141.50
Frictional characteristics Resistance to polishing Polished aggregate friction AS 1141.40
value (PAFV): vertical wheel
Polished aggregate friction AS 1141.41
value (PAFV): horizontal bed
Pendulum friction test AS 1141.42
Adapted from Standards Australia (2009b).

— 2-28 —
Chapter 2 Literature review

2.5 Sprayed seal design and construction


2.5.1 Seal design worldwide
Origins in New Zealand
The first mathematical method for designing a sealed road surface (sprayed seal) ever
published was by the New Zealander Frederick Melrose Hanson (1935) and entitled
The bituminous surface treatment of rural highways. In this seminal work, Hanson
commented that it had been ‘almost universal practice to decide on the quantity of
bitumen for sealing or surface treatment by eye or by rule of thumb.’ This had
‘presented many objectionable features’, the most serious being slipperiness, resulting
in ‘both horse users and motorists, at times, equally venomous in their remarks’. Over
time, these un-designed seals in addition to slipperiness also produced ‘waves’ of
bitumen and ‘rhythmic corrugations’.

Hanson postulated that the bitumen content of a surface seal should be just as
carefully proportioned as it is for asphaltic concrete. The amount of bitumen required
bore a definite relationship to the percentage of voids in the covering aggregate, and
that by filling only 50% to 70% of the air spaces between the stones, the road surface
could be thoroughly waterproofed, the stone chips held securely in position, and at the
same time the top half of the stone would present a non-skid mosaic surface.

Hanson proceeded to develop the ‘void principle’ in connection with surface sealing
(Figure 2.16). By pouring water into an iron dish containing a loose layer of sealing
chips spread shoulder to shoulder, and measuring the volume of water required until
the surface of the water was at the average height of the stones, he was able to
determine that ‘the voids in a loose layer of chips on the road could be taken as 50% of
the average depth of the aggregate’. From field measurements he was also able to
conclude that these voids reduced to 30% after rolling, and reduced to 20% after
compaction by traffic.

— 2-29 —
Chapter 2 Literature review

Figure 2.16: The void principles in seal design

Hanson also identified the importance of the average least dimension of aggregate for
surface sealing. This importance was threefold: it enabled the volume of chips required
for average spreading to be identified; gave the compacted depth of chips as cover
coat; and together with an assumption of 20% compacted voids, enabled calculation of
the amount of bitumen required to rise to any height around the stone.

Hanson did not give a design formula in his paper, rather he provided an explanation of
the issues that needed to be considered followed by example calculations. In a typical
worked example he began with the basic premise that he wanted bitumen to rise to half
of the compacted height. The roller compacted depth is greater than ALD and
therefore bitumen was proportioned to rise to 2/3 of ALD, and the voids to be filled
were 20% of (2/3 of ALD), giving

2 20
𝑅𝑅 = (𝐴𝐴𝐴𝐴𝐴𝐴) 2.10
3 100

𝐴𝐴𝐴𝐴𝐴𝐴
=
7.5

where (in metric terms)

𝑅𝑅 = bitumen application rate (L/m2)

average least dimension 7 of the single sized aggregate used


𝐴𝐴𝐴𝐴𝐴𝐴 =
(mm).

7
The average height of the aggregate particles when they are spread as a single layer with their least
dimensions vertical (Austroads 2014).

— 2-30 —
Chapter 2 Literature review

According to Towler and Dawson (2008), Hanson’s theories and principles were later
adopted in New Zealand’s National Roads Board Manual of sealing and paving
practice (NRB 1968) and the basic New Zealand sealing design algorithm known as
RD 286 was developed in the 1960s from Hanson’s work.

Major and Tuohey (1974) recognised that Hanson’s theory was very much a simplified
version of the actual mechanisms occurring, and that factors other than just the ALD of
the chip needed to be taken into account when determining the binder application rate.
These factors were identified as:

· texture of the surface to be sealed, which may require additional bitumen

· traffic volume, which could change the voids in the compacted mat to either
above or below 20%

· cutback content, used to temporarily reduce the viscosity of the binder.

Because of this the RD 286:1984 Sand circle design algorithm as described in Transit
New Zealand (1989) provided the formula:

𝑅𝑅 = (0.138 × 𝐴𝐴𝐴𝐴𝐴𝐴 + 𝑒𝑒)𝑇𝑇𝑓𝑓 2.11

where

𝑅𝑅 = residual binder 8 application rate (L/m2)

𝐴𝐴𝐴𝐴𝐴𝐴 = average least dimension (mm)

texture depth constant (related to sand circle diameter)


𝑒𝑒 =
obtained from a table

𝑇𝑇𝑓𝑓 = traffic factor (related to vehicles per lane per day) obtained
from a table.

Later, the Transit New Zealand Bituminous sealing manual (1993) introduced the
concept of equivalent light vehicles (elv) based on the South African experience where
a heavy commercial vehicle was considered to have the same effect as ten light
vehicles, or cars (Towler & Dawson 2008).

8
The residual binder comprises bitumen and any additives that are expected to remain in the bitumen
throughout its service life.

— 2-31 —
Chapter 2 Literature review

The TNZ P/17 Performance based specification for bituminous reseals was developed
and trialled by Transit New Zealand (1994) with the basic concept remaining in use
ever since. By 2008 approximately 70% of all state highway reseals were
performance-based (Towler & Dawson 2008). TNZ P/17 is currently published as
Transit New Zealand (2002).

In 2005 when the 1993 Bituminous sealing manual was superseded by the book
Chipsealing in New Zealand (Transit New Zealand et al. 2005), the editors devoted a
chapter to the derivation of the 2004 chipseal design algorithm. It is interesting to note
that Hanson’s original concept of

𝐴𝐴𝐴𝐴𝐴𝐴
𝑅𝑅 =
7.5

had by then grown to:

𝑅𝑅 = (𝐴𝐴𝐴𝐴𝐴𝐴 + 0.7𝑇𝑇𝑑𝑑 )�0.291 − 0.025 × 𝑙𝑙𝑙𝑙𝑙𝑙10 (2.0 × 𝑣𝑣 ⁄𝑙𝑙 ⁄𝑑𝑑 × 100)� 2.12
+ 𝐴𝐴𝑠𝑠 + 𝑆𝑆𝑠𝑠 + 𝐺𝐺𝑠𝑠 + 𝐶𝐶𝑠𝑠 + 𝑈𝑈𝑠𝑠

where

𝑅𝑅 = residual binder application rate (L/m2)

𝑇𝑇𝑑𝑑 = texture depth derived from the sand patch test (mm)

𝑣𝑣 ⁄ 𝑙𝑙 ⁄ 𝑑𝑑 = vehicles per lane per day

𝐴𝐴𝑠𝑠 = absorptive surface allowance

𝑆𝑆𝑠𝑠 = soft substrate allowance

𝐺𝐺𝑠𝑠 = steep grade allowance

𝐶𝐶𝑠𝑠 = chip shape allowance

𝑈𝑈𝑠𝑠 = urban and/or low traffic volume allowance.

The manual is careful to add the caveat that this design algorithm provides for a basic
design only, and that site conditions and other site-specific reasons may require further
adjustments.

— 2-32 —
Chapter 2 Literature review

Under New Zealand’s performance-based contracts, any improvements to the seal


design algorithm are likely to give commercial advantage, and private correspondence
has suggested that some companies have refined their own seal design algorithms but
these remain commercial-in-confidence.

Australia
During early days in Australia, as in New Zealand, the design for binder spray rate and
aggregate spread rate was entirely empirical, resulting in variable and often poor
standard seals. When Hanson published his seminal work in 1935, his method was
adopted by the Country Roads Board in Victoria and was used successfully with
amendments made from time-to-time to adapt to changing conditions, mainly traffic-
related (VicRoads 2004).

The Australian seal design method has subsequently evolved over a number of years
as practices and materials developed to meet increasingly greater road user demands
(Khoo & Neaylon 2011). The national procedure was originally based on the work of
Hanson. The National Association of Australian State Road Authorities (NAASRA, the
forerunner to Austroads) published a series of technical reports from 1965 to 1989
(NAASRA 1965; NAASRA 1975; NAASRA 1980; NAASRA 1989) on the Australian
procedure.

In 1990 a small update was published by Austroads as an interim measure pending the
outcomes of the Austroads sprayed seal design project (Austroads 1990).

A major Austroads project, T&E3P67, the Austroads sprayed seal design project, was
established to develop improved spray seal design and construction procedures. In the
1991/92 summer sealing season, 38 trial sections were placed in Western Australia,
South Australia, New South Wales and Victoria with an additional 12 sections for
Western Australia, New South Wales and Victoria during the 1992/93 season. The
main aim of the design component of this project was to deliver comparative ranking of
three design methods: the Austroads method of the time; the individual state road
authority method of the particular state; and the South African National Institute for
Transport and Road Research (NITRR) design method (Baburamani et al. 1999).

— 2-33 —
Chapter 2 Literature review

In the assessment of these trials, an expert team consisting of two common assessors
(Dr J. Oliver and Mr W. Holtrop) and the local state road authority representative
visually assessed trial sites throughout the country. A standard reporting format and
worksheet was developed and can be found in Baburamani, Holtrop and Gaughan
(2000).

Analysis of the performance of these trials resulted in two revised philosophies


(Alderson 2004). The first was a rigorous statistical regression analysis, commonly
referred to as Method 7A, which was seen as a radical departure from existing
practices, and the second was an approach based on the existing Austroads method
with input from the regression analysis modified by practitioner experience. The latter
was the philosophy that finally reached publication, as the Austroads provisional
sprayed seal design method: Revision 2000 (Austroads 2001).

The Revision 2000 was followed the next year by a practitioners’ guide (Austroads
2002) which contained additional information, such as:

· inclusion of aggregate spread rates, including PMBs and emulsions

· detail on aggregate spread rates for seal sizes 7 and smaller

· additional text on surface texture allowance

· inclusion of binder factors for PMBs and emulsions

· minor rationalisation to the design voids factor for the first coat of a
double/double seal

· worked examples of the design method.

Another design update was published in 2006 (Alderson 2006). As before, the
philosophy remained to fill the voids in the aggregate matrix up to about ⅔ of ALD.
Allowances for aggregate shape, traffic level, embedment into the substrate, pre-
existing surface texture, and absorption of binder into either sealing aggregate or
substrate, were all refined and incorporated.

The Austroads method was supplemented in 2013 (Patrick 2013) to update double seal
designs by nationally adopting an improvement developed by one Australian state.

— 2-34 —
Chapter 2 Literature review

Following concerns from sealing practitioners that the Austroads basic voids factor for
low traffic volume single seals was too high (resulting in design bitumen application
rates being too high), this issue was studied in more detail by Urquhart et al. (2015).
Part of this review included examining the New Zealand approach and a good
correlation was found between the two, showing that sprayed seals from both countries
showed similar changes in surface texture and void reduction when they were
subjected to similar traffic. A new Austroads basic voids factor for low volume roads
has now been proposed based on the New Zealand design approach.

Republic of South Africa


Most of the seal design methods used in South Africa have evolved from Hanson’s
original (1935) work on void principles (Mackintosh 1961; SANRAL 2007; Marais
1969).

Mackintosh (1961) devised a ‘pan and cylinder’ method complete with nomograph, for
determining rates of aggregate spread and bitumen spray based on Hanson’s void
principles.

Marais (1969) focussed attention on various factors that were causing difficulties at the
time. He commenced by checking Hanson’s assertions that the voids in a loose
volume of chippings was 50% and in a compacted mat of chippings reduced to 20%,
and concluded this was not entirely true. Work carried out by the National Institute for
Transport and Road Research using actual measurements and calculations indicating
the volume of the voids in a single layer of stones is dependent on the ALD of the
stones. In a loose volume of stone, the shape of the chippings, as defined by the
flakiness index, also played a part.

Experience also showed that in order to determine the binder application rate more
accurately the following factors must be considered:

· size of chippings

· embedment of chippings

· degradation and wear of chippings

· existing surface texture

· resulting texture depth required

— 2-35 —
Chapter 2 Literature review

· minimum quantity of binder to hold the chippings

· traffic.

Marais discussed all of these issues and in conclusion noted that the ALD of the cover
aggregate is the most basic of dimensions to which all other factors are related. He
also highlighted the need for further research in the areas listed above.

In 1979 Marias published a PhD thesis entitled Advances in the design and application
of bituminous materials in road construction (Marais 1979) which has had a key
influence on South African seal design philosophy (Milne 2004).

The current official South African seal design method is technical recommendation for
highways TRH 3 (SANRAL 2007).

USA
It has been reported that in North America, the state-of-the-art in chip seal design
essentially ended in the 1960s when McLeod (1969) proposed a design method based
on Hanson’s work which was accepted by the Asphalt Institute, and most North
American Departments of Transportation (DoT). Development further stalled as road
agencies then evolved a system whereby no design was performed, and only empirical
rates were used (Gransberg & James 2005). Table 2.5 details how chip seals were
designed in North America at the time of that report.

Table 2.5: Chip seal design methods in North America


Chip seal design method USA Canada
(%) (%)
Empirical/past experience 37 33

No formal method 26 22

Own formal method 19 0

McLeod/Asphalt Institute 11 45

Kearby/Modified Kearby 7 0

Adapted from: Gransberg & James (2005).

The TRB has reported that ‘Americans and Canadians can learn from the procedures
used in Australia and New Zealand’ (Gransberg & James 2005).

— 2-36 —
Chapter 2 Literature review

Since this 2005 comment, a seal design method based on the Australian Austroads
AP-T68/06, Update of the Austroads sprayed seal design method (Alderson 2006) has
been recommended for use in the US. This resulted from NCHRP Project 14-7, which
was established to provide a rational method for the design of chip seals over hot mix
asphalt pavements. The Project 14-7 report was then endorsed and rebadged by the
American association of state highway and transportation officials in 2012 (AASHTO
2012).

UK
Sprayed seals (Australian terminology) and chip seals (New Zealand terminology) are
referred to as surface dressing in the UK.

The original design method was for tar-based surface dressing and was derived from
full scale road trials undertaken over a number of years (Nicholls & Baxter 2008).

Separate documents for tar surfacings (published in 1941) and cut-back surfacings
(published in 1968) were combined to form the first edition of road note 39 which was
published in 1972. The UK design method has continued to be documented as road
note 39, which is now in its sixth edition (Roberts & Nicholls 2008). The method
remains to be basically a catalogue-type design.

A pan-industry group chaired by the Road Surface Treatments Association (RSTA) is


updating road note 39 (sixth edition) with the aim of having it reissued in time for the
2016 surface dressing season (RSTA 2015).

France
A technical guide entitled ‘wearing surface dressings’ (found in Conception et
dimensionnement des structures des chaussées,) was published by the LCPC and
SETRA in 1994 and indicates the basic binder application rates to be adopted. This
method is reproduced in an English language publication by la Section des Fabricants
d'Emulsions Routières de Bitume (SFERB) and L'Union des Syndicats de l'Industrie
Routière Française (USIRF) (2008). The method appears to be basically a catalogue-
type design, commencing with a basic application rate obtained from a table, which is
then modified from another table for traffic/trucks, surface texture, time of year, climatic
region, etc.

— 2-37 —
Chapter 2 Literature review

2.5.2 Aggregate wetting and adhesion


Loss or lack of adhesion between aggregate and bitumen is one of the sprayed seal
failure modes, making this topic relevant.

Adhesion and adhesive technology is a specialist field. An often cited text book on this
topic (Pocius 2002) describes bonding situations that appear to be analogous to
crushed aggregate and bitumen, as illustrated in Figure 2.17. The surface of the
adherend (the microtexture of crushed aggregate) has contaminants and air voids such
that pore penetration by the adhesive (in this case cut bitumen) is not complete, leaving
voids at the interface. The adhesive bond has vacancies at the interface which then
act as a stress concentration point. Pocius is adamant that ‘for good adhesion to take
place the adhesive and the adherend must come into intimate contact’.

Figure 2.17: Components affecting wettability and adhesion

After Pocius (2002).

Hefer and Little (2005) conclude that wetting of aggregate is an important prerequisite
for good adhesion: not only does it rely on good mixing conditions, but also on the
cleanliness of the aggregate surface. Tarrer and Wagh (1991) have reported that dust
particles on aggregates have a tendency to trap air when mixed with bitumen, and
consequently weaken the bitumen-aggregate bond.

— 2-38 —
Chapter 2 Literature review

Contact angle is a term often used in wetting and adhesion technology, and the
concept is illustrated in Figure 2.18 prior to being discussed further.

Figure 2.18: A droplet of bitumen on glass showing the contact angle θ

After Wistuba et al. (2012).

A simple view of wetting and adhesion shown in Figure 2.19 is now discussed.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.19: Variations of contact angle

Source: Pocius (2002), figure 6.11.

The epoxy adhesive has a surface energy γ of 42 mJ/m2. When placed on itself, i.e. on
the same surface energy, the contact angle approaches zero and good wetting occurs.
When placed on materials of decreasing surface energy or wetting tension (γc), the
drop develops a higher profile, the contact angle increases, the wettability decreases,
and adhesion is poor. (Low surface energy materials such as polytetrafluoroethylene
(PTFE) in Figure 2.19 form the basis for non-stick cookware and are considered
abhesive or release surfaces) (Pocius 2002).

From a thermodynamic viewpoint involving the study of energy changes, wetting and
spreading depends on the competition between adhesive forces and cohesive forces.
The surface texture of an aggregate also affects its coatability or wettability, in that a
smoother surface coats easier than a rough surface (Tarrer & Wagh 1991).

Wetting is required to obtain intimate contact between the materials and thereby
establish fundamental forces ultimately responsible for adhesion. Without wetting it is
unlikely that the beneficial effects of surface texture through mechanical adhesion
could be established.

— 2-39 —
Chapter 2 Literature review

Wetting and spreading depends on the viscosity of the liquid and the roughness and
heterogeneity of the solid surface involved. In addition, wettability, including filling of
pores, also depends on the viscosity of the binder and surface chemistry of both
aggregate and bitumen. While some researchers postulate that surface texture of the
aggregate is the main factor affecting adhesion (Kiggundu & Roberts 1988) others
report that chemical and electrochemical effects dominate (Yoon & Tarrer 1988).

Hefer and Little (2005) researched adhesion in bitumen-aggregate systems. The terms
wetting, spreading, and contact angle are used frequently in discussing adhesion. The
shape of a liquid drop on a surface can provide information on intermolecular forces
and resulting interfacial forces, and these can be described through the contact angle
between solid and liquid (Figure 2.18). Contact angle measurements can also provide
a measure of wetting and spreading on a macroscopic scale.

In practice however, the wetting and adhesion of bitumen to aggregate can be assisted
by aggregate precoating.

2.5.3 Aggregate precoating


Precoating of aggregates is used mostly in ‘hot’ sprayed sealing work (i.e. not emulsion
‘cold’ sprayed sealing) to assist in the initial bond between aggregate and bitumen.
Precoats consist of predominantly a flux oil (e.g. common diesel oil), to which a small
amount of bitumen and a small amount of adhesion agent have been added (Austroads
& AAPA 2000). Adhesion agents can be regarded as a combination of wetting agent
and antistripping agent; their active components consist of amines or amine salts of
fatty acids (VicRoads 2004).

2.5.4 Aggregate rolling and re-orientation


As aggregate rolling during and after construction is also one of the compounding
factors influencing aggregate orientation and hence surface texture reduction, it bears
a relationship to the hypothesis of this study and is worthy of discussion. An example
of the different types of sprayed seal rollers used in Australia and NZ is shown in
Figure 2.20.

— 2-40 —
Chapter 2 Literature review

Aggregate rolling is undertaken in sprayed seals usually by multi-tyred pneumatic


smooth-tyred rollers to reorientate the aggregate chips from their random orientation
when first applied to the hot bitumen to a uniform orientation with their least dimensions
lying in the vertical plane. This is done for three reasons: to minimise the volume of
voids required to be filled with bitumen; to maximise the surface area covered with a
given volume (i.e. cost) of stone; and to provide the most stable layer (Neaylon 1996).

Hudson et al. (1986) investigated roller variables in New Zealand and concluded by
sand circle texture measurements (discussed in Section 2.7.4) that construction rolling
had little effect on chip re-orientation after about three passes, after which normal
public trafficking was more effective. He also undertook preliminary but inconclusive
testing of treaded tyres, ‘wobbly’ tyres and low tyre pressures.

Neaylon (1996) sought to validate Hudson’s findings in South Australian conditions (for
single seals with unmodified binders) and by using stereoscopic micro photogrammetry
concluded that public traffic was more effective than construction rollers after about
nine passes, meaning that the rolling times required by the Jurisdiction’s specified
‘rolling formula’ of that time could be halved.

— 2-41 —
Chapter 2 Literature review

11 tyred, 11 t, pneumatic tyred, water ballasted 9 tyred, pneumatic tyred, steel ballasted

Pneumatic tyred, low speed, drag broom up Pneumatic tyred, high speed, drag broom down

Combination roller, pneumatic tyred end Combination roller, steel drum end

Rubber coated vibrating steel drum 7 tyred pneumatic

Figure 2.20: Australian & NZ sprayed seal rollers

After Neaylon & Urquhart (2013).

The Oregon department of transportation investigated the advantages of hot bitumen


sprayed sealing over cold emulsion sprayed sealing (Griffith & Hunt 2000). They
compared a pneumatic roller, a single drum rigid frame roller, and a double drum rigid
roller and concluded that no one type worked more effectively than the other.

— 2-42 —
Chapter 2 Literature review

They did however observe several consistent themes: rolling operations should be
accomplished promptly; rollers should operate closely behind the chip spreaders; and
well-timed rolling is even more critical in cooler temperatures and shaded areas.
These conclusions are of no surprise to Australian and NZ practitioners where these
practices have been recommended and specified for many years. However, in
Australia pneumatic rollers are preferred as the pavement surface shape is often
uneven, and steel drum rollers are considered by many to bridge across the crest of
ruts and shape deformations leaving the sag areas un-touched. Pneumatic rollers can
more closely follow and physically contact a changing contour (Neaylon & Urquhart
2013).

Gransberg et al. (2004) confirmed for US conditions the rolling fundamentals already in
place in Australia and New Zealand, and determined another roller requirement
formula. Gransburg and James (2005) undertook a thorough international synthesis of
chip seal best practices, including the use of rollers, timing, and patterns. Typical types
of rollers used for the chip seal compaction process were the pneumatic tire, the
rubber-coated steel, a combination of the steel and the pneumatic tire, and the static
steel rollers. Pneumatic tire rollers are used universally to protect the aggregate from
crushing, but steel wheel rollers may be used on the surface to ‘tighten up 9’ the
aggregate surface.

Lee and Kim (Kim & Lee 2008, Lee & Kim 2010a, 2010b, 2010c) in North Carolina
further investigated rollers with pneumatic tyres, steel wheels, rubber coated steel
drums, and combination rollers over varying coverages and rolling patterns on
unmodified emulsion binders. They concluded that best results were obtained by
starting with a pneumatic tyred roller and finishing with a combination roller, because
the latter provided a smoother, flatter finish than the pneumatic tyres on their own.
They also concluded that the optimal number of rolling passes is three. Five passes
gave better results, but the extra time taken to achieve this was deemed impractical.
They passed no comment on the practicality of using two roller types to do three
passes combined.

9
‘Steel wheel rollers appear to orient cover aggregate particles into a flatter surface, which is important
when constructing multiple surface treatments’ (McLeod 1969).

— 2-43 —
Chapter 2 Literature review

An International best practice workshop was held in conjunction with the 25th ARRB
conference in Perth (WA) in 2012 and attended by sprayed sealing practitioners from
Australia, New Zealand, South Africa and Asia. This workshop included a discussion
on aggregate rolling practices with set questions circulated beforehand and responses
discussed at the workshop. This workshop cohort asserted that pneumatic tyred rollers
were used because they knead the aggregate chips rather than crush them, they are
effective over uneven road cross-sections, and the pneumatic tyres simulate traffic
loading on a sprayed seal. With regard to ‘how do we know when we have enough
rolling’ there was no general consensus. A ranging discussion was also held on the
advantages and disadvantages of rubber coated steel drum rollers. The workshop
consensus was that different rollers may be chosen for different purposes but the ideal
roller would be one that mimics heavy trucks at low speeds (10 – 20 km/h) and which
has pneumatic tyres (Neaylon & Urquhart 2013).

This concept of an ideal roller raises an interesting point. The model obtained later in
this research is based on chip seal data obtained from the Australian accelerated
loading facility (ALF). ALF was fitted with heavy truck pneumatic tyres and operated at
around 20 km/h, meaning that it would be considered by the workshop practitioners as
being ‘the ideal roller’. Being an ideal roller it could therefore be expected to accelerate
aggregate reorientation compared to a real-life situation.

Whilst this may accelerate the speed of texture reduction under different axle
configurations, it should not significantly affect the rankings between the axle groups.

A summation of the literature therefore suggests that controlled construction rolling is


essential to minimise the risk of aggregate stripping, but after around half a dozen
passes the public traffic (the same traffic that the seal has been designed for) is the
most cost-effective method to achieve reorientation of the aggregate surface. This has
been the case in the past, but will this remain the same with heavier vehicle loadings
and the resulting lower design bitumen application rates making fresh seals more
sensitive to unconstrained (e.g. high speed) public traffic rolling? One of the aims of
this current research is to gain a better understanding of the effect changing axle
groupings and axle loadings has on aggregate orientation and surface texture
reduction.

— 2-44 —
Chapter 2 Literature review

2.5.5 Aggregate embedment


It is commonly accepted that under traffic aggregate particles may commence
embedment into the underlying layer, e.g. the basecourse (Alderson & Oliver 2008;
Caltrans 2007; SANRAL 2007; Transit New Zealand et al. 2005). This is taken into
account in many sprayed seal design methods because it results in void loss and in
design requires less bitumen to fill the available voids, or if not designed for then in
practice causes the existing bitumen to rise towards the top of the aggregate. An
example of embedment caused void loss is shown in Figure 2.21.

Figure 2.21: Void loss due to aggregate embedment

Source: SANRAL (2007).

Embedment is discussed in this review as it can be one of the mechanisms of


reduction of surface texture under trafficking. For sprayed seal design purposes the
likelihood of aggregate embedment can be indicated through a ball penetration
(pavement hardness) test. This test was developed by Marais (1979) and adaptations
of it are in common use in Australia and Southern Africa (Alderson 2006; Austroads
2010; van Zyl 2007).

Yaacob et al. (2007) investigated the prediction of chip seal embedment and used
large, fixed-laboratory equipment that could rank materials but could not be used in the
field for seal design purposes. They did, surprisingly, conclude that embedment
increased with increasing pavement temperature. However, the basecourse used was
a sand asphalt, so this conclusion has no bearing on the unbound granular pavements
more commonly used under chip seals. A sand asphalt would be expected to be soft
at pavement operating temperatures, as bitumen exhibits time-temperature-loading
dependencies as discussed in Section 2.4.1.

— 2-45 —
Chapter 2 Literature review

It is very important to note that visual observations of experimental trials used in this
thesis (Section 4) after trenching, indicated that the deformation within the sprayed
seals in these experiments occurred mostly due to the realignment of seal aggregates
during trafficking and not by embedment into the basecourse (Moffatt 2013).

2.6 Sprayed seal failure modes


The life of a sprayed seal may be shortened by a number of factors which operate
either independently or in combination. The primary mechanisms are shown in
Figure 2.22.

Figure 2.22: Distress modes for sprayed seals

Source: Rebbechi & Alderson (2009).

In figure 2.22, traffic is involved in all four legs of seal distress:

· Traffic causes the pumping action for basecourse fines to pump through surface
cracks, leading to potholes.

· Traffic provides the mechanical leverage to commence aggregate loss.

· Traffic tyres cause the aggregate to polish.

· Traffic provides the energy to reorient the aggregate chip, punch the aggregate
into the basecourse, or grind the sealing aggregate into fines, thus causing
bleeding or flushing.

This study will focus on the bleeding or flushing modes of distress which lead to
macrotexture loss (Figure 2.22).

— 2-46 —
Chapter 2 Literature review

Factors discussed above such as bitumen’s viscoelastic behaviour, stiffness, cutter


levels, polymer content, aggregate rolling and reorientation and aggregate embedment
all influence bleeding and flushing.

2.6.1 Flushing and bleeding of seals


The Australian definitions of these are

Flushing: A pavement surface defect in which the binder is near the


top of the aggregate particles. There is minimal surface texture. In
severe cases, the bitumen covers the aggregate.

Bleeding: A surface defect in which an excess of binder completely


covers the aggregate. It results from the upward migration of the
binder, due to a combination of traffic action, warm temperatures and
other factors. It leads to a loss of surface texture (Austroads 2015).

The field interpretation of this is that flushing is where the bitumen rises such that
macrotexture is lost but the tops of the aggregate particles are still able to provide
microtexture; bleeding is when the bitumen rises to completely cover the aggregate,
such that both macrotexture and microtexture are lost.

The South African definition for bleeding is similar, but there is no differentiation for the
intermediate stage of flushing:

Bleeding: the condition which arises when excess binder is present in


the seal, causing a layer of tacky binder to appear above the
aggregate.

Flushing: see bleeding (SANRAL 2007).

As an aid to international confusion, the New Zealand definitions are the opposite way
around:

Bleeding: the extrusion of a bituminous binder onto the road surface,


generally in hot weather. It is distinguished from flushing, which is a
solid smooth surface caused by binder rise that may be the end result
of bleeding.

Flushing: loss of macrotexture, either through presence of solid binder


high up around the sealing chip, or chip embedment, or chip loss
(Transit New Zealand et al. 2005).

There have been many international investigations into flushing and bleeding of seals
over the years (Jackson et al. 1990; Bahia et al. 2008a, 2008b; Gransberg & James
2005; Lawson & Senadheera 2009). The most recent Australian and New Zealand
examples are discussed here.

— 2-47 —
Chapter 2 Literature review

Alderson and Oliver (2008) investigated the hypothesis that three possible mechanisms
were involved in Australian flushed or bleeding seals: 1) embedment of the sealing
aggregate into the underlying substrate; 2) the aggregates in the seal pack more tightly
together than is assumed; and 3) aggregates physically degrade through wear or
breakdown.

Using image analysis software on slices of bleeding seals cut from actual roads proved
to be the most useful of the various methods they tried. For the limited number of
samples studied the following mechanisms were identified:

· embedment into the underlying substrate

· disintegration of the aggregate with the resulting small particles then displacing
the binder

· changes in the aggregate mosaic (aggregate reorientation and packing)

· loss of aggregate from the seal (stripping).

In New Zealand, Herrington et al. (2012) also investigated flushing (of the NZ definition)
by examining 64 seal cores taken from State Highway sites around the country.
Volumetric analysis of the cores showed significant quantities of materials passing a
4.75 mm sieve, which was not expected. Data from cores taken both in the wheel
paths and from the shoulder of the same site indicated that breakdown of aggregate
due to over-chipping may be a significant contributor to the generation of fines. These
fines then displace the binder causing it to rise. This supports one of the findings of
Alderson and Oliver (2008), although further research was needed.

This further research was wide ranging and took three years, and has since been
published (Herrington et al. 2015). Rather than investigating the hypothesis that only
three possible mechanisms were involved, seven key physical mechanisms were
investigated and a model to predict the onset of flushing was developed. It was
concluded that flushing is a complex process that involves multiple mechanisms that
operate simultaneously in different proportions that may also change over time. The
results of this study are summarised in Table 2.6.

— 2-48 —
Chapter 2 Literature review

Table 2.6: Factors causing chipseal flushing/bleeding


Major contribution Minor contribution Insufficient information
Aggregate abrasion and Bitumen thermal expansion Embedment into basecourse
breakdown
Compaction and Bitumen excess application
reorientation of the
aggregate mosaic
Water venting Bitumen viscosity

Adapted from Herrington et al. (2015).

The factor of water venting is an interesting outcome of this work. Water was first
implicated as a potential cause of flushing by Major (1972) and is obvious through the
formation of blisters or ‘volcanoes’ (Figure 2.23).

Figure 2.23: Bitumen blisters forming in hot weather

Venting of water vapour in this way is quite common and often observed where flushing
is also occurring.

This phenomenon is extremely common in NZ (Herrington et al. 2015), it is common in


Australia (where Figure 2.23 was photographed), and also is observed in South Africa
on both single and double initial seals where high moisture levels are present in the
base (van Zyl & Jenkins 2015).

2.7 Measurement of pavement surface texture


The typical horizontal profile of a pavement surface will show a range of irregularities
as shown in Figure 2.24.

— 2-49 —
Chapter 2 Literature review

Figure 2.24: A typical pavement surface (vertical scale exaggerated)

Adapted from ISO (2002a).

If a Fourier analysis was to be undertaken on the profile in Figure 2.24 the apparent
randomness of the peaks could be mathematically represented as an infinite series of
sinusoidal components of various wavelengths (ISO 2002a).

PIARC (1987) has defined these various wavelengths as being equivalent to


microtexture, macrotexture, megatexture and unevenness as shown in Figure 2.25.
This was when the term ‘megatexture’ was first proposed. Microtexture and
macrotexture are described as desirable properties, whereas megatexture and
unevenness are described as undesirable.

— 2-50 —
Chapter 2 Literature review

Figure 2.25: The relationship between surface characteristics and wavelength

Source: PIARC 1987, Chapter IX.1.2, Table 1.

Microtexture is found on the surface of aggregate particles used in road surfacing, and
results from the crystalline structure of rock being exposed during the crushing and
fragmenting process (Ledee et al. 2005).

Macrotexture is the texture caused by the size and shape and spacing of the coarse
aggregates used in surfacings, as shown in Figure 2.26.

Figure 2.26: Microtexture and macrotexture of a road surface

After Flintish et al. (2003) & Austroads (2005a).

— 2-51 —
Chapter 2 Literature review

Megatexture is related to defects in the surfacing such as rutting, potholes, patching


and major cracks (Austroads 2014).

Microtexture and macrotexture are two key components in the provision of skid
resistance. Microtexture enables chemical bonds to establish between the rubber in
tyres and the aggregate surface. It is believed that thousands of bonds are made and
broken as the tyre moves over the road surface. Macrotexture causes projections into
the tyre which causes the rubber to deform with a resulting hysteresis effect. It is
commonly believed that the effect of microtexture dominates at low traffic speed and
that the effect of macrotexture dominates at high traffic speed (Austroads 2005a; Oliver
& Halligan 2006).

2.7.1 Texture depth (TD)


Two definitions that require clarification at the beginning of this discussion are:

Repeatability: variation in a test method output from a single operator using the same
method (adapted from Austroads 2015).

Reproducibility: variation in a test method output from multiple operators or measuring


devices using the same method (adapted from Austroads 2015).

The International Standards Organisation has defined texture depth as

In the three-dimensional case, the distance between the surface and a


plane through the top of the three highest particles within a surface
area in the same order of size as that of the tyre/pavement interface
(ISO 1997).

This is shown in Figure 2.27, where ‘1’ is the pavement surface and ‘2’ is the texture
depth.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.27: Illustration of texture depth

Source: ISO (2002a) Figure 3.

To place Figure 2.27 in a chip seal /sprayed seal context, consider Figure 2.28 and an
imaginary circular plane the radius of a tyre contact area, supported as if by a tripod of
the three highest stones. Texture depth is the distance between the plane of the tyre
contact area and the underlying surface of the chip seal.

— 2-52 —
Chapter 2 Literature review

Figure 2.28: Sprayed seal texture.

Source: Herrington & Wu (2016).

2.7.2 Volumetric texture measurement


Volumetric texture measurement is when any given volume of material capable of
flowing is spread out onto a surface and distributed to form a circular patch. The
resulting diameter is measured, and by dividing the volume of the material spread by
the area that the material covers a value is obtained that represents the average depth
of the layer. The volumetric texture measurement is used not only with sand or glass
spheres but in some cases with putty or grease (ISO 1997).

2.7.3 Mean texture depth (MTD)


MTD is the texture depth obtained from using a volumetric texture method (ISO 1997).

2.7.4 Sand patch texture depth (SPTD)


The sand patch texture depth, also known as the sand circle method, has been used
worldwide for many years to give a very simple measurement describing surface
texture. It is a specific form of volumetric texture measurement.

A given volume of sand or glass beads of a specific grading are spread onto a surface
and distributed to form a circular patch and the resulting diameter is measured. By
dividing the volume of the material spread out by the area that the material covers a
value is obtained that represents the average depth of the sand or glass sphere layer.
This is a specific case of mean texture depth, and is known as the sand patch texture
depth.

— 2-53 —
Chapter 2 Literature review

Figure 2.29: Known volume of sand poured onto surface

Source: Austroads (2008).

Figure 2.30: Sand spread to form a circular patch

Source: Austroads (2008).

The resolution and hence accuracy of the method would in theory depend on the size
of the sand granules used, with smaller granules giving higher resolution, as in a
photograph where smaller (and hence greater number of) pixels give a higher
resolution. For example Road Note 27 (Road Research Laboratory, 1969) instructs
that a smaller total volume and a finer particle size be used on smoother surfaces,
where a higher resolution is required to achieve meaningful results. However in
practice it has been found that any accuracy gained by using smaller granules than
those specified in the test methods, is lost due to the inherent operator influence and
inconsistencies as discussed in Section 2.7.5

Even so, for such a fundamentally uncomplicated test a surprising number of


complications and variations have developed over time, and these will now be
discussed.

International method

When the International Standards Organisation ISO 10844:1994 Acoustics -


Specification of test tracks for the purpose of measuring noise emitted by road vehicles
Annexure A Measurement of pavement surface macrotexture depth using a volumetric
patch technique was published, it was the first time the sand patch method was
internationally standardised.

— 2-54 —
Chapter 2 Literature review

It was largely based on ASTM E965-87 Standard test method of measuring surface
macrotexture depth using a volumetric technique and like that standard it replaced
sand with glass spheres. It does not specify a volume of glass spheres, only that the
volume is greater than 25 mL and is known. The spheres are spread with a hard, flat
disc about 25 mm thick and 60 to 75 mm in diameter, with the bottom face covered with
a hard rubber material. The mean texture depth (MTD) is calculated as

𝑀𝑀𝑀𝑀𝑀𝑀 = 4𝑉𝑉 ⁄𝜋𝜋𝐷𝐷2 2.13

where

𝑀𝑀𝑀𝑀𝑀𝑀 = mean texture depth (mm)

𝑉𝑉 = sample volume (mm3)

𝐷𝐷 = average diameter of the area covered (mm).

The accompanying precision statement claims that repeatability (i.e. same operator)
can be as low as 1% of the average texture depth and that reproducibility (i.e. different
operators) can be as low as 2% of the average texture depth. However, site-to-site
variation may be as large as 27% of the average texture depth on sites randomly
selected within a nominally homogeneous pavement section.

The current version of this standard is ISO 10844:2011 (ISO 2011). In this version,
texture specification by sand patch mean texture depth (MTD) has been deleted and
replaced with texture measured by machine, to fulfil ISO 13473-3 (ISO 2002b). The
standard explains that the effect of this change is twofold: firstly, to ‘reduce variation in
measurements due to differences in hand process’; and secondly the ‘elimination of
possibility to manipulate the results’.

American method

The American method is ASTM E965-15, Standard test method for measuring
pavement macrotexture depth using a volumetric technique (ASTM 2015). This method
does not specify a volume of glass spheres, only that the volume is greater than 1.5
cubic in. and is known.

— 2-55 —
Chapter 2 Literature review

The spheres are spread with a hard, flat disc about one inch (25 mm) thick and 2.5 to 3
inch (60 to 75 mm) in diameter, with the bottom face covered with a hard rubber
material 10.

ASTM advises that

The pavement macrotexture depth values measured by this test


method, with the equipment and procedures stated herein, do not
necessarily agree or correlate directly with other techniques of surface
texture measurements (ASTM 2015).

British method

This British method is BS EN 13036-1:2010 Road and airfield surface characteristics.


Test methods. Measurement of pavement surface macrotexture depth using a
volumetric patch technique (BSI 2010). The method uses sand rather than glass
beads. The document states that it is largely based on ASTM E 965-87.

New Zealand method

The New Zealand method TNZ T/3: 1981 Standard test procedure for measurement of
texture by the sand circle method uses 45 mL of graded sand (passing 0.6 mm and
retained by 0.3 mm) spread into a circle with a straight edge of between 150 and 160
mm in length.

South African methods

The South African Draft TMH6 Method ST1 Measurement of the texture depth of a
road surface, permits two methods (CSIR Transportek 1984):

· spreading of 500 mL of sand 11 between chalked parallel lines 500 mm apart,


using a rubber squeegee. The length covered with sand is measured.

· placing 250 mL of sand inside a spreading box: the box is weighed down with a
6 kg mass and then pulled in a straight line along the test section. When the
sand has run out, the length covered with sand is measured.

10
The standard notes that an ice hockey puck is considered useful for this.
11
Why 500 mL? Mackintosh (1961) advises that the standard volume of sand used was about 534 cc,
measured by filling a one pint ‘Çonsol’ canned-fruit jar to the brim and scraping it level.

— 2-56 —
Chapter 2 Literature review

Texture depth is calculated from either method as

𝑉𝑉 2.14
𝑇𝑇 =
1000𝐴𝐴
where

𝑇𝑇 = texture depth (mm)

𝑉𝑉 = volume of sand (mL)

𝐴𝐴 = area covered (m2).

Although two methods are given, van Zyl (2007) has reported a variation in texture
depth calculations of up to 30% between the two methods.

Australian methods

There are two main Australian variants:

· Austroads Test Method AG:PT/T250 Modified surface texture depth (Pestle


method) (Austroads 2008) which uses a volume of either 50, 100 or 150 cm3
(depending on the degree of surface texture) of sand or glass beads spread
with a rubber-based pestle.

· RTA (2006) Test Method T240, Road surface texture depth. This method
specifies a 50.0 g mass of single-sized sand or glass beads spread with a 300
mm long ruler. This appears to be an unusual method that specifies mass for
what the outside world believes to be a volumetric method, and hence has the
conversion formula

1272𝑀𝑀 2.15
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ (𝑚𝑚𝑚𝑚) =
𝐷𝐷𝐷𝐷2
where

𝑀𝑀 = mass of the filling medium (g)

𝐷𝐷 = loose mass of the filling medium (t/m3)

𝑑𝑑 = mean diameter of spread of the filling medium (mm).

— 2-57 —
Chapter 2 Literature review

2.7.5 Inaccuracies in the sand patch test


ISO (1997) states in its introduction that the volumetric ‘sand patch’ method is operator
dependent.

Hillier and Soet (2009) reported that although the sand patch test is simple to conduct,
it can be vulnerable to operator influence and inconsistencies.

Patrick et al. (2000) undertook precision trials comparing the New Zealand sand circle
test with the TRL mini-texture meter and a stationary laser profilometer. They found
that ten operators who had experience in using the test, but had never observed how
the other operators did the test, gave a reproducibility of 41% of the mean. Two of
these operators could obtain texture depth measures of 1.2 mm and 1.8 mm at the
same location. In another trial four operators who had observed each other doing the
test gave a reproducibility of 24% of the mean. The authors concluded that there can
be a large operator bias, making the comparison of results amongst different operators
difficult. They also concluded that the mini-texture meter was a more reproducible test
than the sand circle and the stationary laser profilometer was the most precise method
of the three.

As discussed earlier, ISO (2011) has deleted sand patch mean texture depth and
replaced it with texture measured by machine. The standard explained that one of the
reasons for this was to ‘reduce variation in measurements due to differences in hand
process’.

However, the sand patch test continues to be used as a useful input into many seal
design methods. Although it is scientifically imprecise, most seal design methods are
based on many imprecise assumptions and generalisations, and the practicality of the
final results do not seem to be unduly hindered by the lack of precision of this one
particular input.

Given the inaccuracies of the sand patch test, should it be used for modelling
purposes? The right answer should be – preferably not. However, if circumstances
dictate that this is the only data available, the proof of the resulting model would be
whether the model can be validated using a completely different data set – even a data
set derived using a laser profilometer’s mean profile depth (MPD). This is taken further
in section 5.8.3.

— 2-58 —
Chapter 2 Literature review

2.7.6 Profile depth (PD)


Profile depth is a two-dimensional adaptation of the three-dimensional texture depth
described above, and is useful in analysing laser-derived macrotexture measurements.
Profile depth is defined by the International Standards Organisation (2002a) as the
height difference between the actual surface profile and a horizontal line through the
top of the highest particle, calculated over a distance of the same order of length as
that of a typical tyre/pavement interface (Figure 2.31).

Figure 2.31: Illustration of profile depth and mean profile depth

Adapted from ISO (2002a).

2.7.7 Mean profile depth (MPD)


ISO (1997) is devoted to determining an average depth value of pavement surface
macrotexture – the mean profile depth. In practical terms the mean profile depth is the
average value of the profile depth over a certain (baseline) distance (Figure 2.31). The
ISO definition of mean profile depth is

1𝑠𝑠𝑠𝑠 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 − 2𝑛𝑛𝑛𝑛 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 2.16


𝑀𝑀𝑀𝑀𝑀𝑀 = � � − 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
2

where

𝑀𝑀𝑀𝑀𝑀𝑀 = mean profile depth.

— 2-59 —
Chapter 2 Literature review

As mean profile depth (a two-dimensional measurement) is not the same as a sand


patch depth (a three-dimensional measurement) the resulting numbers are not directly
interchangeable. Transit New Zealand (2005) published a formula for converting sand
patch texture depths into units of laser-derived mean profile depth, as shown in
equation 2.17.

𝑀𝑀𝑀𝑀𝑀𝑀 = �0.59 × 𝑇𝑇𝑇𝑇𝑠𝑠𝑠𝑠 � + 0.27 2.17

where

𝑀𝑀𝑀𝑀𝑀𝑀 = mean profile depth

𝑇𝑇𝑇𝑇𝑆𝑆𝑆𝑆 = sand patch texture depth.

2.7.8 Sensor measured texture depth (SMTD)


Sensor measured texture depth (SMTD) does not appear to be an ISO term but is
defined in Australia as

A continuous measure of macrotexture where the pavement surface is


divided into smaller segments of a given base length, approximately
300 mm long, which are analysed, assigned a value and averaged to
give a mean value for a specific length of profile (Austroads 2014).

It is calculated from Austroads (2007) using equation 2.18.

2.18
�∑40 (𝑦𝑦𝑖𝑖 − 𝑦𝑦𝑒𝑒 )2 �
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = � 𝑖𝑖=1
40

where

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = sensor measured texture depth (mm)

the values of the residuals from a quadratic least squares


(𝑦𝑦𝑖𝑖 − 𝑦𝑦𝑒𝑒 ) = regression using a second order polynomial fit, as shown
diagrammatically in Figure 2.32.

— 2-60 —
Chapter 2 Literature review

Figure 2.32: Calculating sensor-measured texture depth

Source: Austroads (2007).

In the UK all textured depths for sprayed sealing are specified in terms of a high-speed
road monitor (HRM) which reports sensor-measured texture depth (SMTD). A
minimum specification of 1.03 mm with this method equates to approximately 1.5 mm
in sand patch terms. The UK has no precision data on the method of measuring
texture depth in sprayed seals, and the Highways Agency recommends that care
should be exercised in interpreting the results (Highways Agency et al. 1999).

2.7.9 Estimated texture depth (ETD)


MPD or SMTD may be converted to an equivalent volumetric sand patch texture depth
ETD. This will be dependent on first performing a correlation exercise under controlled
conditions to develop a relationship between the results achieved using the volumetric
test method and MPD or SMTD (Austroads 2007).

The MPD may also be transformed to an ETD by applying the following generic
transformation equation (ISO 1997):

𝐸𝐸𝐸𝐸𝐸𝐸 = (0.8 × 𝑀𝑀𝑀𝑀𝑀𝑀) + 0.2 2.19

where

𝐸𝐸𝐸𝐸𝐸𝐸 = estimated texture depth (mm)

𝑀𝑀𝑀𝑀𝑀𝑀 = mean profile depth (mm).

— 2-61 —
Chapter 2 Literature review

ISO (1997) notes that the use of this equation should give ETD values which are as
close as possible to SPTD, as the estimated error in the transformation equation is
much less than the errors due to different operators and equipment of the SPTD.
Notwithstanding, the MPD is preferred for use over the ETD.

2.8 Measuring seal performance


Measuring the change in surface texture over time is a common method for measuring
the performance of a seal. The following discussion will focus on different forms of
metrology to determine a sprayed seal’s properties and attributes.

2.8.1 Visual assessment


Historically, visual inspections using experienced assessors have been a predominant
seal assessment method (Tool & Hillier 2014). Whilst practical guidance on this in
Australia has been limited to few publications, Austroads introduced a guide to the
visual assessment of pavement condition in the late 1980s which is currently
incorporated in Austroads guidelines (Jameson 2011). VicRoads for many years has
published a Guide to surface inspection rating (the current edition is VicRoads 2009)
for assessing a seal’s performance. Visual assessment can also be supplemented by
touch, e.g. to feel with the fingertips how smooth or rough an aggregate surface is.

2.8.2 Pull-out force


One method of measuring the existence and strength of the adhesive bond between a
binder and an aggregate on a sprayed seal in the field is a pull-out test. This can be
done by hand with a screwdriver or pliers and is also used to assess the binder
condition (Patrick 2014).

There is a South African pull-out test method for surfacing aggregate, method MB-8
(Asphalt Academy 2007) which for the field version uses alligator clips, nylon line, a
spring balance and a temperature correction formula. This has also been adapted by
the DTMR (2014) for use in Queensland.

A pull-out test to measure adhesion performance of a chip seal has also been
developed at the Suleyman Demirel University, Turkey (Akýllý et al. 2012) but due to
the preparation involved prior to aggregate-bitumen contact it is likely to remain a
laboratory test for ranking.

— 2-62 —
Chapter 2 Literature review

The Limpet was originally developed to assess the strength of concrete on-site.
However, it is now a standardised test to determine the bond strength of patch repairs
and this has become its principal application, as shown in Figure 2.33 (Amphora NDT
Ltd. 2008).

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.33: Limpet tester

Source: Nanyang Technological University (2003).

The Limpet pull-off test was then further developed for use in the bitumen and surface
dressing industry to quantitatively measure the bond strength between aggregates and
binders (Read & Whiteoak 2003). It was developed for use in both the laboratory and
on the road, with a 50 mm diameter metal plate adhered to the road surface and the
peak force to achieve pull-off then determined.

Woodside and Rogan examined the use of the Limpet tester to assess the bond
strength characteristics of aggregates and bituminous binders (Woodside & Rogan
1993a). Despite finding a number of controllable and uncontrollable variables, and
poor repeatability on a spray seal site, they still concluded that this tester has potential
for considerable benefits to be gained. Some of the benefits they envisaged were
determining the adhesive bond strength of a spray seal on-site and determining the
effect of polymer modified binders on end performance.

Woodside and Rogan also developed another tensile test procedure, the INAPOT 12
test (Woodside & Rogan 1993b). Their aim was to be able to specify permissible dust
levels for single-sized aggregate rather than stripping potential due to the presence of
water. This aim is of little benefit to Australians as water supplies are becoming too
scarce to use for washing dust from quarry aggregates, and the dust issue is
addressed by using a suitable aggregate precoat. They chose an Instron laboratory
tensile testing machine and placed cut aggregate prisms in the top jaw and bitumen
trays in the lower jaws.

12
The INAPOT was a UK test based on the direct tensile test performed on the Instron test rig.

— 2-63 —
Chapter 2 Literature review

They conceded that this testing procedure is not ‘totally’ analogous to a road pavement
situation since the Instron machine used tensile forces alone to pull the aggregate
prisms from the bitumen (Woodside had previously identified a rocking effect
(Woodside & Liu 1993) that speeds up fatigue failure in the binder holding the
aggregate, which he believes may be the main contributor to stripping of the surface
aggregate. This is discussed later). They did however argue that this test was useful
in understanding the complex aggregate-bitumen adhesive bond.

The Limpet tester has been experimented with for sprayed seals under field conditions
in New Zealand (pers. comm. P Herrington 2008), but as the loading was applied
manually by turning a handle it was difficult to obtain repeatable data. Another issue
identified in the field was that bitumen temperature could not be controlled. However, it
is the only method Herrington has identified for measuring seal strength in the field.

The common binder for sprayed seals in the USA is bitumen emulsion. The University
of Wisconsin-Madison in conjunction with the University of Stellenbosch investigated
the performance grading of bitumen emulsions for sprayed seals (Bahia et al. 2008a,
2008b). With regard to a suitable adhesion test they explored whether Superpave
performance tests and concepts can be applied and investigated two options: a mineral
disk in a dynamic shear rheometer (Figure 2.34); and a piston-operated pull-stub based
on ASTM D4541 (ASTM 2009) as shown in Figure 2.35. This second test is a
pneumatic adhesion test adapted from the paint and coatings industry.

Both of these tests use a laboratory prepared rock sample obtained from saw cutting a
larger rock specimen. Whilst this is useful from a theoretical chemical affinity and
bonding point of view, they intrinsically ignore the physical effects of any microtexture
and crystalline fracture obtained during the crushing process of producing aggregate.
As discussed in section 2.5.2, the surface of a crushed aggregate has contaminants
and air voids (Figure 2.17) which interrupt the bond interface and act as stress
concentration points. With this in mind, these two test methods as they currently stand
should be regarded as giving best-case scenarios.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.34: Mineral disc in dynamic shear rheometer

Source: Bahia et al. (2008a), slide 21.

— 2-64 —
Chapter 2 Literature review

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.35: PATTI – ASTM D4541 adhesion test

Source: Bahia et al. (2008a), slide 19.

More recently at the University of Wisconsin-Madison, work on the piston-operated


pull-stub has been continued. The test has been developed into the binder bond
strength (BBS) Test (Hanz et al. 2012; Moraes et al. 2011) that has since been
accepted as an AASHTO test method TP 91-15 (AASHTO 2015). It tests for adhesive
failures at the asphalt–aggregate interface and cohesive failure within the asphalt
binder and involves subjecting a pull-stub adhered to an aggregate substrate to a
normal force created by increasing pneumatic pressure (Figure 2.36). Bond strength is
defined as the maximum pull-off pressure exerted by the machine.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.36: Bitumen bond strength (BBS) test

Source: Kim (2015), slide 62, adapted from NCHRP Project 09-­-50.

2.8.3 2D imaging
A simple camera unit for photographing road surfaces was detailed in RRL’s road note
27, as far back as 1969 (Road Research Laboratory 1969). 2D photographs were
used by skid resistance practitioners to estimate surface macrotexture in terms of
rough, medium or smooth texture.

Georgopoulos, Loizos and Fiouda (1995) developed a digital imaging algorithm to


evaluate road surface cracking. They asserted that prior to 1995 there were several
limitations concerning software capability of existing pavement imaging systems. They
developed software algorithms to recognise crack type (longitudinal, transverse,
alligator and block cracking), to estimate severity and calculate area. It is interesting to
note that the images acquired for their work were taken on a 35 mm film camera and
were then digitised using a flatbed colour scanner.

— 2-65 —
Chapter 2 Literature review

In order to evaluate the texture condition and aggregate retention performance of


sprayed seal layers, Kim and Lee (2008) trialled digital image processing using a cut
section of a sprayed seal surface, a digital scanner, Adobe Photoshop® and
MATLAB® R207a. They were able to make three clear observations on the
roughness, aggregate orientation and embedment of a seal at different stages using
this technique.

Kim and Lee (2008) also used 2D digital imaging and greyscale intensity values (GIVs)
to take bleeding measurements of sprayed seals (which they call asphalt surface
treatments, ASTs). They were able to monitor laboratory prepared samples of sprayed
seals, produced in accordance with a sweep test procedure specified by ASTM D7000
(ASTM 1988). They found photographic greyscale intensity values useful in quantifying
both bleeding seals and stripping seals, but were more focused with developing test
methods and critical specification limits for acceptance of stripping or bleeding.

Alderson used 2D photography in a preliminary study into why sprayed seals become
flushed or bleeding (Alderson & Oliver 2008). Sprayed seal samples were cut from a
road and prepared as a number of cross-sections. Their structure was reviewed using
three procedures: a visual examination by a number of experienced practitioners;
obtaining a digital image then using a printed copy to determine parameters using
manual measurement techniques; and applying a software package to the electronic
digital images to measure seal parameters.

Alderson found that the image analysis software approach provided substantial
quantities of data once the image had been prepared. However, preparation of the
samples and then the images using existing tools was labour intensive and time
consuming. He concluded that surface texture and embedment depth could be
determined relatively accurately using image analysis software, although if the
preparation of the images could be more automated then the process would be more
clearly a way forward.

— 2-66 —
Chapter 2 Literature review

2.8.4 3D imaging
X-Ray
X-Ray imaging is well established in the medical field, and the medical evaluation of 3D
shapes (e.g. the human body and its internal components) is known as computed
tomography 13 (CT) scanning. In CT scanning a series of X-ray images are taken in
cross-section, and with the use of computer technology these images can be adapted
to form a three-dimensional internal picture that can be viewed from any angle.
However, access to specialised medical equipment to examine road pavement cores is
not necessarily simple; for example the Delft University of Technology resorted to
buying their own second-hand CT Scanner from a hospital to use in asphalt research
(Nielsen 2007).

Light waves
Whilst X-rays have been developed for internal 3D modelling in the medical industry,
light waves have been developed for surface 3D modelling for terrestrial features.
Image-based modelling using light waves is a complete, economical, portable, flexible
and widely used approach (Remondino & El-Hakim 2006). The generation of a 3D
model is mainly achieved using non-contact systems and active or passive sensors as
shown in Figure 2.37.

For example, from Figure 2.37 it can be seen that lidar 14 uses lasers, time delay and
active sensors to generate a 3D model whereas photogrammetry uses simple light
waves reaching a passive sensor, photographic film, or a charged couple device (CCD)
photo-electric light sensor as found in digital cameras.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.37: Three-dimensional acquisition systems for non-contact methods based on light waves

Source: Remondino & El-Hakim (2006), Fig1.

This discussion will now examine photogrammetry. Whilst a subset of 3D imaging it is


worthy of its own section because of its relevance to this study’s topic.

13
The word tomography is derived from the Greek word tomos meaning a cutting or section and graphos
meaning written, writing or representation in a specified manner (Merriam-Webster 2011, Oxford 1987).
14
‘Lidar’ is a play on ‘Radar’. Radar uses the radio wave part of the electromagnetic spectrum, and the
word is derived from the military acronym for radio detection and ranging. Lidar similarly uses the visible
light part of the electromagnetic spectrum in the form of laser waves, hence the acronym Light detection
and ranging. Laser is yet another acronym, from light amplification by stimulated emission of radiation
(NOAA 2013, Oxford 1987).

— 2-67 —
Chapter 2 Literature review

2.8.5 Photogrammetry
In the past, aerial photographs were used in mapping to produce digital terrain models
(DTM). These models ignored above-ground details such as the tops of buildings and
trees, as they were too time-consuming to measure manually, or with available
automation methods (Chikhi & Bignone 2006).

More recently digital surface models (DSM) have been developed which are able to
calculate accurate height measurements of all objects on the urban landscape. They
require the ability to process a dramatically increased number of images.

Researchers in the discipline of pavement skid resistance have also investigated the
use of stereo-photographic surface models for the analysis of road surface
performance over many years. For example, Schonfeld in his ‘Photo-interpretation of
skid resistance in practice’ article (Schonfeld 1974) cites stereo-photographic
references dating as far back as 1971 (USA) and 1967 (England).

Schonfeld developed a somewhat cumbersome method of classifying skid resistance


of a road surfacing by measuring aggregate height, width, angularity, distribution,
harshness of projections above the aggregate matrix and harshness of the matrix itself.
He took pairs of stereo photographs using a 35 mm single lens reflex camera with a
focal length of 55 mm attached to a sliding seat within an enclosed box and with a fixed
flash, enabling photographs from two positions 95 mm apart. Although used to
generate a complex texture code number to classify skid resistance that eventually did
not excite any later skid resistance researchers, the stereo photograph concept itself
has remained of continued interest.

In the mid 1990s this author studied Hudson et al. (1986) and then carried their 2D
photographic concept further by utilising stereo photography to not only observe but
also to measure the effects of rolling of aggregate of a freshly laid seal (Neaylon 1996).
A high quality (at that time) Roliflex 6006 camera with calibrated and engraved fiducial
marks and using 6 cm by 6 cm transparency film, was used with a tripod and natural
lighting to take stereoscopic pairs of spray seal surface photographs.

The transparency pairs were then processed and analysed through a stereo digitiser
using an ADAM MPS-2 ® Microphotogrammetric System made available at a Technical
and Further Education (TAFE) college of surveying. Whilst the final outcomes were
instructional, the whole process however was slow and time-consuming.

— 2-68 —
Chapter 2 Literature review

A French researcher took another approach to 3D photography whilst researching his


PhD thesis (Benslimane 2004, Benslimane et al. 2003). He appears to have chosen
the depth from shading (DFS) approach (discussed later in section 2.8.6, 3D imaging
standards) by keeping the camera location fixed, and taking a series of photographs
utilising three separate flash guns from three different angles. He then established a
link between ‘grey levels’ of his photographs and the ‘relief variations’ of his road
surfacing subject. He finally extracted a 3D relief of the surface through a
mathematical ‘Lambertian Hypothesis’ synthesis of the three photographs. The results
were compared favourably with those of an autofocus laser sensor at the French
Laboratoire Central des Ponts et Chaussées. This procedure was then used in the
non-contact characterisation of microtexture and may find further use, again in the
discipline of skid resistance (Benslimane et al. 2007). Figure 2.38 illustrates the
apparatus developed through a joint effort between the Signal, Image and
Communications (SIC) laboratory of the Université de Poitiers and the Laboratoire
Central des Ponts et Chaussées (LCPC) (Do 2005).

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.38: SIC/LCPC experimental image recorder

Source: Do (2005), Fig 6.

In another attempt to measure microtexture Do (2005) used both a three source


photometric stereo system to reproduce the three dimensional surface of the pavement
and a laser sensor to recover two dimensional profiles of the pavement surface.
However, all three light sources were applied in the presence of ambient light, as can
be seen from Figure 2.38, which could affect the results.

In New Zealand in 2006, when searching for a rapid method for measuring surface
texture depth, Pidwerbesky et al. (2006) concluded that with the computing power and
image analysis software available at the time, the use of stereo photography remained
a relatively time-consuming method of measuring sprayed seal surface texture.

Two researchers in Canada (El Gendy & Shalaby 2007; Shalaby & El Gendy 2008)
reviewed existing work and then developed a non-contact texture device they named
‘PhotoTexture’, based on a four-sourced photometric stereo technique (Figure 2.39).
They calculated the mean profile depth of a pavement surface using photometric stereo
techniques.

— 2-69 —
Chapter 2 Literature review

In their system four images of the same pavement sample are taken with illumination
from four different directions but the same intensity. They assert that pixel intensities in
a set of images from the same viewing location will depend on light direction, and
therefore pavement surface heights can be recovered from the difference of the
intensities. Their equipment was isolated from ambient lighting to reduce data noise,
and they used a fibre optic light source manually repositioned in the four light
directions. They analysed the images using power spectrum energy, using the area
under the power spectrum function of the two dimensional Fourier transform of the
three dimensional recovered surface.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.39: Photo-texture device

Source: Shalaby & El Gendy (2008), Fig 1a.

American researchers Flintsch et al. (2008) investigated Shalaby and El Gendy’s


device whilst researching ways to measure macrotexture as an indicator of road safety.
They considered that it needed to be improved as it still required associated lane
closures and traffic inconvenience. They saw a need for a method by which road
texture measurements could be performed continuously and at traffic speed, and
developed a system using relatively low cost, off-the-shelf components for image
capture and subsequent conversion into three dimensional surface maps of a
pavement surface.

They commenced with a stationary proof-of-concept system using a commercially


available stereo system calibrated by the manufacturer. The device has two cameras
in a single enclosure with a known distance separating them, with each camera using a
1.3 megapixel CCD (charge coupled device). They found that variations in ambient
illumination adversely affected their results and so developed a lightweight housing
with artificial light. This was then developed into a vehicle-mounted system to perform
‘continuous’ texture measurements and 3D mapping at near highway speeds.

— 2-70 —
Chapter 2 Literature review

In Australia, Neaylon (2009) explored the possibility of stereo photography as a tool to


understanding sprayed seal distress mechanisms. A commercial software package
was obtained, a stereo camera framework built and field photographs of sprayed
sealing taken, as will be described later in this thesis. Eventually progress halted as it
seemed that this particular software package required more than two photographs of
an object to enable 3D calculations.

At the same time in Ireland, Millar et al. (2009) were investigating close-range
terrestrial photogrammetry to assess asphalt surfaces. Millar (experienced in land
surveying and geographical information systems) was able to adapt commercial
terrestrial photogrammetry software as a tool to model the wear of asphalt surface
mixes. From ortho-rectified digital images he was then able to superimpose a 0.1 mm
vertical contour interval. From this he was able to measure the gradual reduction of
texture depth associated with continual wearing of the asphalt surface and concluded
that mean texture depths estimated from his 3D digital models correlated well with
mean texture depth (MTD) and mean profile depth (MPD) (refer to section 2.7.7).
Stereo image pairs of photos were taken using a calibrated 10 megapixel SLR camera
with a 28 mm to 55 mm zoom lens.

Such ‘close range digital photogrammetry’ has since been found to be useful in the
investigation of chip seal surfaces (Millar 2013; Millar et al. 2012).

2.8.6 3D imaging standards


There are a large number of 3D imaging techniques available and some of these are
discussed by Remondino et al. (2009). These include: area-based matching (ABM);
depth from shading (DFS); image-based rendering (IBR); feature-based matching
(FBM); and multiphoto geometrically constrained (MPGC) matching.

With respect to Lidar discussed earlier, in 2011 the ASTM 3D data imaging systems
committee approved a new data exchange standard ASTM E2807 - 11 3D imaging
data exchange version 1.0. (ASTM 2011a).

ASTM noted that three-dimensional laser scanning is a maturing technology and


commented that ‘this new ASTM data exchange standard has the potential to move the
entire industry along the technology adoption curve by allowing the mass market
access to laser scanned data’ (ASTM 2011b).

— 2-71 —
Chapter 2 Literature review

2.8.7 Comment on methods of measuring seal performance


Measuring the change in surface texture over time is a very common component of
measuring a seal’s performance, and is generally done at a network level by measuring
macrotexture through laser survey and analysis rather than by site specific sand patch
testing. The microtexture component is contributed separately through skid resistance
measurements, which is not discussed here.

Visual assessment of a seal surface is a low cost method but relies on individual
expertise, so that in developed countries it tends to be used as a reality check for
machine readings (outputs of network survey vehicles that are fitted with laser and
cameras), or as a trigger for engaging the more accurate laser surveys and/or skid
resistance surveys.

Pull-out force methods are time consuming with the field versions possibly requiring the
additional expense of traffic control, whilst having unanswered questions over their
repeatability and reproducibility. The laboratory based method, the binder bond
strength test (AASHTO 2015), improves repeatability and reproducibility but is unable
to be undertaken in the field and requires specifically cut blocks of rock rather than
samples of crushed aggregate. As such it would be far more applicable to the
characterisation of bond development between specific binders and specific geological
sources of rock.

2D imaging is useful as a record accompanying a visual assessment but its further


development as a measurement tool is hampered by the sample preparation and
analysis of software output being very time consuming. It is currently thought that for
less effort a more complete 3D analysis could be obtained.

3D imaging using the visible light spectrum is favoured by several authors for research
and analysis of chip seal behaviour. In particular, close range digital photogrammetry
is seen as a useful research tool and can be used both in the field and in the laboratory
and provides a rich source of data.

For the experimental work undertaken for this research the aggregate to bitumen
adhesive bond proved not to be a problem. Bond strength test was not required, and
visual assessment sufficed. It is expected that bond strength testing will be utilised
more in future research in improving chip seal performance under heavy vehicles.

— 2-72 —
Chapter 2 Literature review

2.8.8 Summary of surface texture measurements


Many methods of measuring surface texture, or macrotexture, have now been
discussed. A summary of the key methods including their advantages and
disadvantages is given in Table 2.7.

Table 2.7: Macrotexture measurement methods

method advantages disadvantages conclusions


Sand patch texture depth Cheap, simple, easily portable. Operator dependant. Still used in some cases
(SPTD) Decades of history and Slow. for seal design work, but
experience with what the Poor repeatability and poor for research it is most
numbers mean. reproducibility. useful as a reality check
Used in many specifications. for the output of the more
complex systems
Concept is simple to visualise
available.
and understand.
Samples a 3D volume rather
than a 2D line (c.f. a laser
beam).
Does not require electricity.
Mean profile depth (MPD) Human variability removed. MPD texture (mm) is not Gives an accurate result
Faster than SPTD. directly interchangeable with for research but is not
SPTD texture (mm). meaningful to practitioners
or directly comparable with
historical data or research
until approximated to ETD
Sensor measured texture Suitable for high speed road Not directly interchangeable Gives an accurate result
depth (SMTD) monitoring. with MPD or SPTD. for new research but is not
meaningful to practitioners
or directly comparable with
historical data or research
until approximated to ETD.
Estimated texture depth Converts MPD and SMTD into Requires a transformation For research purposes the
(ETD) a SPTD equivalent. equation, which may generic transformation
introduce an error equation should be
established for the
experiment at hand.
2D imaging (photographs) Cheap and simple. Inaccurate. Does not provide data that
Smart phones have made this Qualitive rather than can be used for quantitive
tool ubiquitous. quantitive. research.
Simple categorisation: rough,
medium or smooth.
3D imaging (LIDAR) Human variability eliminated. Capital outlay is expensive Because of the capital
Fast. cost, for research work this
method would be more
appropriate when
collecting extensive ‘big’
data.

— 2-73 —
Chapter 2 Literature review

method advantages disadvantages conclusions


3D imaging Consumer grade camera. Data processing can be time The ideal solution for
(Photogrammetry) Small & portable. consuming. research work involving
Inexpensive capital equipment Suitable for research only a moderate sample
& software. purposes more than contact size.

Images can be captured compliance measurement.


instantaneously and analysed
safely as time permits.
Individual aggregate
movements can be tracked.

Originally three of these macrotexture measurements were chosen for use in this
study, which also gave a double redundancy should any of them fail.

Firstly photogrammetry was chosen, primarily for its ability to track and accurately
measure individual aggregate particle movements, and also for its small capital outlay.
It was seen as providing research quality data and being sufficiently independent of
operator error or bias. Photographs could be collected quickly on site without
interfering with the experimental process, and then analysed in detail in the office.

Secondly laser texture measurements converted to ETD were taken, as this system
could be automated and data collected easily. This data would not yield intelligence on
individual aggregate movements and reorientation but would give a reliable indication
of the overall and average change in surface texture.

Finally sand patch texture depth measurements were taken, as this system is accepted
practice in the sealing industry. Although the least accurate of the three methods the
results could be used as a reality check for the photogrammetry results and the ETD
calculations.

2.9 Predicting seal performance


The effect of heavy multiple-axle trucks on flexible pavement damage has been studied
before with respect to rutting (Salama et al. 2006). Various rutting prediction models
are available for asphalt (e.g. Ji et al. 2015) and for granular pavements (e.g. Arnold &
Werkmeister 2010; Henning 2008; Jannat et al. 2014; Martin 2008; Morosiuk et al.
2001; Paterson 1987).

Roughness deterioration models for pavements have also been developed (e.g.
Mamlouk & Dosa 2014; Martin 2008; Morosiuk et al. 2001; Paterson 1987).

— 2-74 —
Chapter 2 Literature review

Models of predicting seal texture deterioration are less prevalent and will now be
discussed.

2.9.1 Seal texture deterioration


The engineering disciplines of skid resistance and asset management modelling have
resulted in various endeavours to predict and model the reduction of road microtexture
when under traffic.

Skid resistance studies


Whilst initially studying high speed skid resistance, Major and Tuohey (1976) studied
changes in texture depth for typical examples of New Zealand chipseal and a hot-mix
open-graded asphalt (friction course) surfacing, and published Figure 2.40.

Figure 2.40: Reduction of texture depth with trafficking (NZ)

Source: Major & Tuohey (1976).

— 2-75 —
Chapter 2 Literature review

Again from a skid resistance point of view, the UK Design manual for roads and
bridges (Highways Agency et al. 1999) reported that for surface dressings the decay of
texture depth with time under traffic is not linear; it is rapid in the first year or two with
the initial bedding-in period being about 4 weeks. The reduction in surface texture over
the first 12 to 24 months is said to provide some indication of whether the texture depth
will remain above the required minimum for the design life of the seal.

Figure 2.41 shows the reduction in surface texture of a single seal with various
aggregate types and conventional binder on the M40 High Wickham bypass in the UK.

Figure 2.41: Reduction of texture depth with trafficking (UK)

Source: Highways Agency et al. (1999), fig 8.7.

Assuming that the traffic and basecourse hardness are the same, the fact that these 3
different aggregates have different asymptotes of surface texture is probably due to the
different hardness of the aggregates, their resistance to abrasion, and their durability.
Aggregate skid resistance is often gained at the expense of these aggregate properties
(Woodward, WDH, Woodside, AR and Jellie, JH 2005).

— 2-76 —
Chapter 2 Literature review

Asset management studies


In the discipline of road asset management, the World Bank sponsored a highway
design and maintenance (HDM) standards model for the investigation of the economic
consequences of various investments in road infrastructure (N.D. Lea International Ltd.
1995).

The Asian Development Bank supported research for an update resulting in HDM
Version 4. In this, work on prediction of pavement texture was done by Cenek and
Patrick of New Zealand (Chapter 9 of ND Lea 1995). Cenek and Patrick correlated
cumulative traffic to mean texture depth as measured by the volumetric sand patch
method, and for New Zealand single surface treatment pavements found a strong
correlation with the following expression (equation 2.20):

𝑇𝑇𝑇𝑇 = 𝐴𝐴𝐴𝐴𝐴𝐴(𝑎𝑎0 − 𝑎𝑎1 log10 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁) 2.20

where

𝑇𝑇𝑇𝑇 = sand patch texture depth (mm)

𝐴𝐴𝐴𝐴𝐴𝐴 = average least dimension of sealing aggregate (mm)

number of equivalent light vehicle (elv) passes since the


𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = sealing date, where one heavy commercial vehicle is
equivalent to ten light vehicles

𝑎𝑎0 = regression constant

𝑎𝑎1 = regression constant.

This equation shows the rate of change of macrotexture to be a function of aggregate


size and traffic loading, with a reported coefficient of determination (R2) which varied
from 0.7 to 0.9.

The authors then generalised equation 2.20 to apply to all bituminous surfacing types
in an absolute form as shown in equation 2.21.

— 2-77 —
Chapter 2 Literature review

𝑇𝑇𝑇𝑇 = 𝐼𝐼𝐼𝐼𝐼𝐼(1 − ∆𝑇𝑇𝑇𝑇𝑇𝑇 log10 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁) 2.21

where

𝑇𝑇𝑇𝑇 = sand patch texture depth (mm)

initial texture depth related to the aggregate size (mm)


𝐼𝐼𝐼𝐼𝐼𝐼 = (suggested to be 1.5 mm for fine seals and 3.5 mm for coarse
seals)

rate of change of texture with traffic - should be constant for


∆𝑇𝑇𝑇𝑇𝑇𝑇 = similar surfacing types (suggested to be 0.120 for single and
double bituminous surface dressings)

number of equivalent light vehicle (elv) passes since the


𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 = sealing date, where one heavy commercial vehicle is
equivalent to ten light vehicles.

However, the model actually adopted into HDM-4 by the International Study of Highway
Development and Management (ISOHDM) is an incremental macrotexture model
proposed by Cenek and Griffith-Jones in 1997 (Morosiuk et al. 2001; Rainsford &
Parkman 2005), as shown in equation 2.22:

DTD = Ktd {ITD - TDa - a0 ITD log10 (10 [(ITD - TDa) / (a0 ITD)] + DNELV )} 2.22

where

incremental change in sand patch derived texture depth during


DTD =
the analysis year (mm)

Ktd = calibration factor for texture depth

ITD = initial texture depth at construction of surfacing (mm)

TDa = texture depth at the beginning of the analysis year (mm)

a0 = regression constant

number of equivalent light vehicle passes during analysis year


DNELV = (one heavy truck or heavy bus is equal to 10 NELV; light vehicles
equal 1).

— 2-78 —
Chapter 2 Literature review

The initial texture depth (ITD) and regression coefficient (a0) were provided as standard
values for the entire New Zealand network. A subsequent review of the performance of
this model (Rainsford & Parkman 2005) found that the model did not align with
experience on the road network.

Henning et al. (2006) reviewed the HDM and dTIMS 15 pavement models with NZ
calibration site data. He highlighted that the classical HDM format of the texture depth
model incorporated a complex two stage process as shown in Figure 2.42.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.42: Texture depth model formats (NZ)

Source: Henning et al. (2006), fig 3.2.

He concluded that the classic two-stage deterioration model was difficult in its
application due to its various sensitivities and devised a simplified texture model as
shown in Figure 2.42, equation 2.23 and Table 2.8.

𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑎𝑎0 + 𝑎𝑎1 × log10 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 2.23

where

𝑀𝑀𝑀𝑀𝑀𝑀 = mean profile depth

𝑎𝑎0 = chip size specific coefficients (refer to Table 2.8)

𝑎𝑎1 = chip size specific coefficients (refer to Table 2.8)

total number of equivalent light vehicle passes (one heavy truck


𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 =
or heavy bus is equal to 10 NELV; light vehicles equal 1).

Table 2.8: Chip seal coefficients

Surfacing type a0 a1
AC 1.02 -0.0189
Chip G2 4.31 -0.127
Chip G3 4.16 -0.135
Chip G4 3.41 -0.109
Chip G5 3.48 -0.124
Chip G6 2.32 -0.0724
Open Grade 1.18 -0.00555

15
Deighton’s Total Infrastructure Management System (dTIMS)®.

— 2-79 —
Chapter 2 Literature review

Henning continued studying the development of pavement deterioration models on the


state highway network of New Zealand (Henning 2008) and from his analysis of long
term pavement performance (LTPP) sites in New Zealand developed a crack initiation
model and a three-stage rut progression model.

Kodippily (2013) followed on from this work and focussed on flushing, also using the
New Zealand LTPP database. For this study the flushing definition was ‘a smooth area
on the pavement surface having low skid resistance, occurring as an upward migration
of bitumen which causes full or partial covering of seal aggregates’. Flushing is a far
larger issue in New Zealand than it is in Australia, and the reasons for this are the topic
of ongoing debate.

Although not specifically surface texture, flushing can be considered as the reverse of
texture such that the two are closely related. Kodippily’s data analysis showed weak
correlations of traffic volumes and the percentage of heavy commercial vehicles (HCVs
defined as commercial vehicles with a gross mass exceeding 12 t) which was contrary
to expectations. Pavement composition characteristics (surface age, surface
thickness, aggregate grade) had a strong correlation and climatic factors (air
temperature and humidity) had weak correlations.

She developed a two stage model for the development of flushing based on the data
set available. The first stage was flushing initiation, shown in equation 2.24 and based
on the standard logistic regression model.

1
𝑃𝑃𝑓𝑓𝑓𝑓 = 2.24
1 + 𝑒𝑒 −(0.293𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠+0.046𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠ℎ−2.913)

where

𝑃𝑃𝑓𝑓𝑓𝑓 = probability flushing has initiated

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = the age of the surfacing layer (years)

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠ℎ = the thickness of the surfacing layer (mm).

Stage two - after initiation - was the progression of flushing, and modelled using a
forward stepwise linear regression model as shown in equation 2.25.

— 2-80 —
Chapter 2 Literature review

𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓ℎ𝑖𝑖𝑖𝑖𝑖𝑖 = (0.41𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 0.40𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠ℎ + 0.11𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 2.25


+ 0.170𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖)2

where

𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓ℎ𝑖𝑖𝑖𝑖𝑖𝑖 = the amount of flushing on the pavement

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = the age of the pavement surface (years)

𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠ℎ = thickness of the pavement surfacing layer (mm)

𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = average rutting on the pavement (mm)

𝑐𝑐ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 = grade of aggregate in the surfacing layer (see Table 2.3).

The model is based on pavement condition data found in the New Zealand long term
pavement performance (LTPP) data base and was tested using a separate set of the
New Zealand LTPP data (Herrington et al. 2015). The model does not apply to first
coat or initial seals, but to any subsequent reseals.

Seal design studies


The origins of the present New Zealand chip seal design algorithm (Transit New
Zealand et al. 2005) can also be related to predicting the reduction of macrotexture
over time.

The seal design algorithm is based on data from trials in Lower Hutt and the Bay of
Plenty. An analysis of this data (Patrick 1999; Patrick & Donbavand 1996) found it
followed relationships based on work originally reported by Potter and Church (1976)
and Houghton and Hallett (1987) as shown in equation 2.26:

𝑉𝑉𝑣𝑣
= 𝐴𝐴 − 𝐵𝐵 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 2.26
𝐴𝐴𝐴𝐴𝐴𝐴

where

𝑉𝑉𝑣𝑣 = total volume of voids in a chipseal (L/m2)

𝐴𝐴𝐴𝐴𝐴𝐴 = the age of the surfacing layer (years)

𝑇𝑇 = total traffic to date, i.e. cumulative traffic (elv)

equivalent light vehicles where one heavy vehicle is


𝑒𝑒𝑒𝑒𝑒𝑒 =
equivalent to ten light vehicles

— 2-81 —
Chapter 2 Literature review

𝐴𝐴, 𝐵𝐵 = constants (derived from regression analysis).

With the Lower Hutt and Bay of Plenty data sets combined (Figure 2.43) the best fit
relationship is:

𝑉𝑉𝑣𝑣
𝐴𝐴𝐴𝐴𝐴𝐴
= 0.87 − 0.074 𝐿𝐿𝐿𝐿𝐿𝐿 𝑇𝑇 with 𝑅𝑅 2 = 0.84

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.43: Change of surface texture with traffic passes

Source: Patrick & Donbavand (1996), Fig 1.

The total volume of voids in a chipseal (Vv) is reproduced again here for convenience in
Figure 2.44.

Figure 2.44: Voids in a chipseal

Source: SANRAL (2007).

Patrick assumed that for normal low traffic volume reseals in New Zealand embedment
and aggregate wear is minor such that the void volume can be defined as equation
2.27.

𝑉𝑉𝑣𝑣 = 𝑉𝑉𝑎𝑎 + 𝑉𝑉𝑏𝑏 2.27

where

𝑉𝑉𝑣𝑣 = total volume of voids in a chipseal (L/m2)

𝑉𝑉𝑎𝑎 = volume of air (L/m2)

𝑉𝑉𝑏𝑏 = volume of bitumen sprayed (L/m2).

— 2-82 —
Chapter 2 Literature review

Given that the volume of air (Va) is numerically equivalent to texture depth (Td in mm),
equations 2.26 and 2.27 can be rearranged (Patrick & Donbavand 1996) to give

𝑇𝑇𝑑𝑑 = 𝐴𝐴𝐴𝐴𝐴𝐴(𝐴𝐴 − 𝐵𝐵 log 𝑇𝑇) − 𝑉𝑉𝑏𝑏 2.28

where

𝑇𝑇𝑑𝑑 = sand patch texture depth (mm)

𝐴𝐴𝐴𝐴𝐴𝐴 = average least dimension of the sealing chip (mm)

𝐴𝐴, 𝐵𝐵 = constants (derived from regression analysis)

𝑇𝑇 = total traffic to date, i.e. cumulative traffic (elv)

𝑉𝑉𝑏𝑏 = volume of bitumen sprayed (L/m2)

equivalent light vehicles where one heavy vehicle is


𝑒𝑒𝑒𝑒𝑒𝑒 =
equivalent to ten light vehicles.

Pavement rutting studies


In the late 80s, Kinder and Lay undertook to provide ‘a definitive statement on the
fourth power law and pavement damage caused by trucks’ (Kinder & Lay 1988).

Using mechanistic computer models available at the time (CIRCLY 16, VESYS III 17, and
a finite element model) they modelled the theoretical rutting (permanent deformation) of
a pavement that had undergone a full scale accelerated loading trial at Somersby in
New South Wales. The resulting theoretical models are shown in Figure 2.45 and
Figure 2.46. Arguably however, they do not appear to give a particularly convincing
representation of the full scale testing.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.45: Actual pavement rutting c.f. predicted, for a 80 kN load

Source: Kinder & Lay (1988), Fig 3.

16
CIRCLY is an integral component of the Austroads Pavement Design Guide that is widely used in
Australia and New Zealand. CIRCLY had its genesis in software developed at CSIRO for relatively simple
loading cases. CIRCLY was first released in 1977 and handled polynomial type radial variations in contact
stress and multiple loads which provide a much closer representation of the actual loading conditions.
CIRCLY was commercialised in 1988 by MINCAD Systems (MINCAD 2012).
17
VESYS is a truck pavement interaction program developed by the FHWA’s Office of Infrastructure
Research and Development and uses mechanistic analysis to determine pavement performance (Al-
Khateeb et al. 2007).

— 2-83 —
Chapter 2 Literature review

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.46: Actual pavement rutting c.f. predicted, for a 120 kN load

Source: Kinder & Lay (1988), Fig 4.

Following the demonstration of what appears to be a very mediocre fit obtained by the
models, Kinder and Lay then used a regression of a plot of those models to provide a
performance prediction equation of the form shown in equation 2.29.

𝐷𝐷 = 𝑐𝑐𝑃𝑃𝑚𝑚 𝑁𝑁 𝑚𝑚⁄𝛼𝛼 2.29

where

𝐷𝐷 = damage, as permanent deformation (rutting) (mm)

𝑃𝑃 = load (kN)

𝑁𝑁 = number of cycles

𝑐𝑐 = a constant of proportionality

𝑚𝑚, 𝛼𝛼 = constants (derived from regression analysis)

and by further calculation they concluded that

𝛼𝛼 = a power law.

Much later De Pont, Steven, Alabaster and Fussell (2003) studied the effect of heavy
vehicles on pavement wear at the Canterbury accelerated pavement testing indoor
facility (CAPTIF) in New Zealand using a single tyre/single axle, with wander to mimic
the effect of dual tyre single axle. They used the Kinder-Lay model for vertical surface
deformation analysis and also used it under the assumption that an equation
developed for damage defined as rutting would be analogous to damage defined as
reduction in mean profile depth (MPD). They postulated that:

— 2-84 —
Chapter 2 Literature review

𝐷𝐷(𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑: 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟)


If
= 𝑀𝑀𝑀𝑀𝑀𝑀 (𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑖𝑖𝑖𝑖 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ)

then 𝐷𝐷 = 𝑐𝑐𝑃𝑃𝑚𝑚 𝑁𝑁 𝑚𝑚⁄𝛼𝛼

becomes 𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑀𝑀𝑀𝑀𝑀𝑀0 − 𝑐𝑐𝑃𝑃𝑚𝑚 𝑁𝑁 𝑚𝑚⁄𝛼𝛼 2.30

where

𝑀𝑀𝑀𝑀𝑀𝑀 = damage, as mean profile depth (mm)

𝑀𝑀𝑀𝑀𝑀𝑀0 mean profile depth at zero load (mm)

𝑃𝑃 = load (kN)

𝑁𝑁 = number of cycles

𝑐𝑐, 𝑚𝑚, 𝛼𝛼 = constants (derived from regression analysis).

This Kinder-Lay style model was attractive because it contained terms for axle mass
and a power exponent, and gave the rate of change of vertical surface deformation.

When compared with the actual rate of change of MPD this model gave a reasonable
fit as shown in Figure 2.47. Load damage exponents for surface texture wear were
also estimated using this equation. This will be discussed later in Section 2.10.7 The
4th power law.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.47: Comparison of MPD predicted by Kinder-Lay style model with measured data for two axle
loadings

Source: de Pont et al. (2003), Fig 4.8.

This model was considered an advancement in texture reduction models in that it


allows for the effect of mass on surface texture reduction, and the researchers at the
time also wanted to derive a power law exponent.

However, although possibly an advancement, this model was not taken further in this
current study because of this authors concerns over

— 2-85 —
Chapter 2 Literature review

· the poor fits of predicted vs actual performance in Figure 2.45 and Figure 2.46
used in its derivation as a rutting tool

· the assumption that the mechanism of pavement rutting ≈ the mechanism of


surface texture reduction

· the validity of the load damage exponents for MPD wear that were obtained
from this equation.

Summary of surface texture models


It is generally accepted that rate of surface texture decay under traffic is rapid at first
but then the rate of decay starts reducing with loadings, and may taper off towards a
plateau level that is dependent on the traffic loadings and the materials used.

The focus in the past has been on modelling pavement rutting and roughness, as these
are symptoms of pavement structural distress: rutting growth is of major interest to
asset owners whilst roughness growth is of major interest to the road users. There has
been very little focus on macrotexture modelling, and what there has been is based on
generic traffic data and is focused on the skid resistance aspect of asset management.

A chronological summary of these various macrotexture deterioration models is given


in Table 2.9.

Table 2.9: Macrotexture deterioration models (in chronological order)

Dependant
Reference Comment Independant variable(s)
variable
Major & Tuohey Observation Traffic: (AADT/lane)*years. Texture depth
(1976) (mm)
Cenek & Patrick Model Number of equivalent light vehicle passes since the Sand patch
(1995) sealing date, where one heavy commercial vehicle is texture depth
assumed equivalent to ten light vehicles. (mm)
Aggregate ALD.
Patrick & Model Number of equivalent light vehicle passes since the Sand patch
Donbavand sealing date, where one heavy commercial vehicle is texture depth
(1996) assumed equivalent to ten light vehicles. (mm)
Aggregate ALD
Cenek & Griffith- Model Number of equivalent light vehicle passes during Sand patch
Jones (1997) analysis year (one heavy truck or heavy bus is equal texture depth
to ten NELV; light vehicles equal one). (mm)
Highways Agency Observation Years of service. Texture depth
(1999) (mm)
De Pont et al. Model Axle loading. Mean profile
(2003) Number of axle passes. depth (mm)

— 2-86 —
Chapter 2 Literature review

Dependant
Reference Comment Independant variable(s)
variable
Henning, Costello Model Total number of equivalent light vehicle passes (one Mean profile
& Watson (2006) heavy truck or heavy bus is equal to ten NELV; light depth (mm)
vehicles equal one).
Kodippily (2013) Model Stage 1): surface age, surface depth. 1). Probability
Stage 2): surface age, surface depth, pavement of flushing
rutting, chip size. 2) amount of
flushing

Table 2.9 shows that most existing macrotexture deterioration models treat traffic very
simplistically. Those using simply years of traffic or the age of the surface have
significant weaknesses in relation to prediction of future performance under future
conditions. Those using average annual daily traffic, and assuming one heavy
commercial vehicle is equivalent to ten light vehicles, are slightly less weak.

One existing deterioration model does take into account the axle load and was derived
using a single tyre on a single axle.

There appears to be no existing deterioration model that takes into account: the
composition of the commercial vehicles; dual tyred axles; and the effect of axle
groupings such as single axles, tandem axles or triaxles. This is seen as a current
limitation in knowledge. There is an opportunity and research gap available to develop
an improved model that can accommodate future changes to the traffic fleet rather than
rely on using the traffic from the past.

2.9.2 Finite element analysis


In research aimed at progressing a performance-related sprayed seal design method
for South Africa, Milne (2004) examined the feasibility of modelling road surfacing seals
using mechanistic principles. He used the finite element method (FEM) analysis to
examine the interaction of individual seal components, i.e. stone and bitumen, at the
micro-mechanic scale. The modelling was undertaken at Delft Technical University
(The Netherlands). Simulations of different seal, environmental and traffic scenarios
(excluding seal aggregate interlock and embedment effects) were undertaken to
demonstrate the potential of the model.

Development towards a second prototype FEM model was undertaken at Delft


(Huurman 2010) which made improvements on the first model in representation of the
aggregate, the distance between aggregate particles, and basecourse embedment.

— 2-87 —
Chapter 2 Literature review

Both of these researchers developed FEM models in which the aggregate particles
were represented by regular ellipsoids arranged in a single layer and without any
stone-to-stone contact. Kathirgamanathan et al. (2012) attempted to model a more
realistic structure where stone to stone contact and aggregate orientations were
derived from an actual multiple layer seal using X-ray tomography data. The authors
considered this improvement time-consuming but well warranted.

The FEM modelling of chip seals has returned to Stellenbosch, South Africa where
Gerber and Jenkins (2015) improved the FEM models for chip seals by applying
damage transfer functions to the response model with the aim of further progressing a
performance-related sprayed seal design method for South Africa.

The current model uses failure mechanisms of adhesive failure, cohesive failure and
aggregate embedment as surrogates for the distress modes of ravelling, fatigue
cracking and surface texture loss, and still uses aggregates modelled as perfect
elliptical shapes.

Finite element analysis is a highly theoretical tool used by many academics and
researchers. Given that research is progressing in this area already in various
academic clusters around the world, this study explored a different route (research
gap) that is based on actual performance rather than assumed transfer functions
which, at this time, are based on some very crude assumptions and do not take into
account aggregate shape or aggregate reorientation under trafficking. In this different
route the variability of tyre contact pressures, cohesive and adhesive strength of the
seal, type of bitumen, etc. were all held constant, with the single focus being in
separating the effects of single, tandem and tri-axle clusters on an actual physical
pavement and surfacing.

— 2-88 —
Chapter 2 Literature review

2.10 Pavement design and traffic loadings


The last topic to be discussed in this literature review is pavement design and relevant
traffic loading calculations.

2.10.1 Fundamental differences between pavements and surfacings


As discussed in section 2.3.1 and repeated here, the primary functions of pavement
surfacings are considered to be: to protect the underlying pavement material from
moisture ingress; minimise dust; maintain the pavement’s ride quality; provide suitable
skid resistance; and to provide for reflectivity and line markings to ensure public safety.
It is the pavement (not the surfacing) that supplies the structural performance to
withstand the various loads applied by vehicles (Sharp 2005).

Figure 2.48: Dispersion of surface loads through a pavement structure

Source: Moffatt (2015).

It is the role of the pavement to disperse axle loads over a large area, until ultimately
the total load can be carried by the subgrade below the pavement as shown in
Figure 2.48. At the surface, the axle loads are focused over a relatively small area,
and therefore seals must be designed to withstand high contact pressures.

This contact pressure varies depending on the tyre type, tyre load and tyre inflation
pressure, as illustrated in Figure 2.49.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 2.49: Vertical tyre-pavement contact stress profiles of a single 11R22.5 tyre over a range of loadings
and inflation pressures

Source: Maina J & De Beer M (2008), Fig. 6.

— 2-89 —
Chapter 2 Literature review

In Figure 2.49, the illustrations in columns show constant inflation pressures, and
changing axle loads. It can be seen that as inflation pressure increases, peak contract
stress moves away from the centre of the tyre footprint, and becomes highest under
the tyre sidewalls. The illustrations in rows show constant axle loadings, and changing
inflation pressures. It can be seen that in general as the inflation pressure increases,
peak contact stress moves away from below the tyre sidewalls and towards the centre
of the tyre footprint.

Also, at the top left illustration it can be seen that although the tyre inflation pressure is
520 kPa, the peak contact stress under the sidewall is of the order of 1,200 kPa.

The South African National Road Agency Ltd (SANRAL 2016) is developing a web
based contact-stress information system to provide typical contact stress data based
on tyre type, axle load data and tyre inflation levels.

The fact that pavement failures can be deemed as load dependant whereas surfacing
failures can be seen as contact stress dependant, supports the proposition that a load
damage exponent for surfacings will be different than that for pavements.

2.10.2 Pavement design


The Austroads pavement design procedures (Jameson 2012) provide three different
methods for the design of pavements (Moffatt 2015).

Mechanistic method
This method is a design procedure based on stress analysis and fundamental material
behaviour in pavements (Austroads 2015) and is used for flexible pavements
containing asphalt or cemented materials.

The mechanistic method is limited to three load-associated distress types:

· rutting and shape loss (as measured at the surface but assumed to be
occurring in the subgrade)

· flexural fatigue of asphalt materials

· flexural fatigue of cemented materials.

— 2-90 —
Chapter 2 Literature review

The method uses software to predict strain responses at critical locations to a static
application of the standard axle and calculates the number of standard axle repetitions
a trial pavement can carry, which is then compared to the design traffic (converted to
standard axle repetitions). It then becomes an iterative procedure to determine the
most cost-effective trial pavement to carry the design traffic.

Empirical method
An empirical method is based on observation or experiment not on theory (Oxford
1987) and is used for unbound granular pavements typically surfaced with a sprayed
seal.

The method is based on a single empirical design chart, often referred to as ‘Figure
8.4’ (Jameson 2012).

Allowable load repetitions are expressed in equivalent standard axles, which are a
specific case of standard axle repetitions for a load damage exponent of 4 (granular
pavement).

Mechanistic - empirical method


This method is based on analytical models and field testing (Jameson 2012) and is
used for the design of rigid concrete pavements. The Australian method is based on a
procedure developed by the US Portland Cement Association and considers flexural
fatigue and pavement erosion under joints or cracks as its two causes of pavement
distress.

2.10.3 Load damage equivalencies


The concept of load damage equivalencies for road pavements is not new, and one
has even been devised to compare numbers of oxen, horses or bullocks.

Lay published a list of load damage equivalents per trip for various wheel types,
normalised for type 4, as shown in Table 2.10.

(Table 2.10)…. gives some damage equivalents for the beginning of


the 20th century, when two dramatically different transport
technologies were overlapping. The advantage of the rubber
pneumatic tyre is very obvious. The early trucks were too heavy for
the first generation of rubber tyres, which could only carry about 0.5 t.
Trucks therefore ran predominantly on solid steel tyres until reliable
solid rubber ones became widely available in 1910. The solid rubber
tyres were only marginally less damaging to the roads than were iron
tyres.

— 2-91 —
Chapter 2 Literature review

Michelins produced the first pneumatic truck tyre in 1912, but a


number of difficulties were encountered. Technological success came
about in 1916 with tyres using cord rather than canvas as reinforcing,
but these did not make a significant on-road impact until the late
1920s. Suspensions were developed mainly and initially to provide
passenger comfort, but have also reduced the dynamic and out-of-
balance loads produced by trucks (Lay 1998, as quoted in Lay 2008).

Table 2.10: Load damage equivalents per trip for various wheel types
Type Vehicle description Equivalency
based on type 4
1 Unharnessed animal 0.2

2 Single animal harnessed to a vehicle with pneumatic tyres 0.2

3 Single animal harnessed to an unloaded vehicle with steel tyres 0.5

4 Single animal harnessed to a vehicle with steel tyres 1.0

5 Car unable to travel at over 30 km/h 1.0

6 ‘n’ animals harnessed to a loaded vehicle with steel tyres n

7 Car able to travel over 30 km/h 3

8 Truck with steel tyres 25

9 Steam traction engine with steel tyres 36

Adapted from Lay (2008).

2.10.4 Standard axle


With the advent of the modern truck, a standard axle was used to compare damage
equivalencies. The standard axle is defined as a single axle with dual wheels loaded to
a total mass of 8.2 tonne (80 kN) and with tyre pressures of 750 kPa (Austroads 2015;
Moffatt 2015). This axle can be illustrated as the rear axle in Figure 2.50.

Figure 2.50: 1970 Bedford truck with standard axle at rear

Source: State library of Western Australia image 347113PD, from Stevenson et al. (1970).

— 2-92 —
Chapter 2 Literature review

For pavement design purposes, the current existing heavy vehicle axle group types
have been identified (Jameson 2012) as:

· single axle with single tyres (SAST)

· single axle with dual tyres (SADT)

· tandem axle with single tyres (TAST)

· tandem axle with dual tyres (TADT)

· triaxle 18 with dual tyres (TRDT)

· quad-axle with dual tyres (QADT).

This axle configuration nomenclature is shown in Figure 2.51.

Figure 2.51: Axle configuration nomenclature

Source: Neaylon (2015).

When designing flexible pavements in Australia and New Zealand, all traffic loadings,
i.e. axle group loadings, are converted to a standard loading unit, the standard axle.
When calculating the total traffic loadings to design for, a procedure is used that
calculates the damage associated with each axle grouping compared with the damage
caused by a standard axle. This is done in two stages. The first is to determine the
axle group equal damage load (taken from Table 2.11) which is considered to cause
the same pavement damage as the standard axle.

18
The term triaxle is used to refer to the same grouping of three axles that is termed a tridem in some
countries.

— 2-93 —
Chapter 2 Literature review

Table 2.11: Loads on axle groups (with dual tyres) which cause the same damage as the standard axle
Axle group type Load (kN)
Single axle with single tyre 53
Single axle with dual tyres 80
i.e. the standard axle
Tandem axle with dual tyres 135
Triaxle with dual tyres 181
Quad-axle with dual tyres 221
Adapted from Jameson (2012).

The second stage is to introduce a concept of standard axle repetition as the unit of
damage of a single axle.

2.10.5 Standard axle repetitions


The Austroads design method uses equation 2.31, shown here as-clarified by Moffatt
(2015), to determine standard axle repetitions.

𝑛𝑛
𝐿𝐿𝑖𝑖 𝐿𝐿𝐿𝐿𝐿𝐿 2.31
𝑆𝑆𝑆𝑆𝑆𝑆 = � � �
𝐿𝐿𝐸𝐸𝐸𝐸𝐸𝐸
𝑖𝑖=1

where

𝑆𝑆𝑆𝑆𝑆𝑆 = standard axle repetitions

𝐿𝐿𝑖𝑖 = load carried by axle group 𝑖𝑖 (KN)

𝐿𝐿𝐸𝐸𝐸𝐸𝐸𝐸 = equal damage load for that axle group (Table 2.11)

𝐿𝐿𝐿𝐿𝐿𝐿 = load damage exponent (varies with distress, from Table 2.12)

𝑛𝑛 = number of axle groups for the vehicle.

Standard axle repetitions calculated with a LDE of 4, i.e. SAR4, are commonly referred
to as equivalent standard axles, ESAs (Jameson 2012).

As the work described in this study relates to a granular pavement with a thin
bituminous surfacing, SAR4s are described as ESAs.

— 2-94 —
Chapter 2 Literature review

𝐿𝐿𝑖𝑖
This equation also illustrates that for the same traffic �𝐿𝐿 �, different materials with
𝐸𝐸𝐸𝐸𝐸𝐸

different distress mechanisms i.e. different load damage exponents, will result in a
different number of standard axle repetitions.

2.10.6 Load damage exponents


The load damage exponent used in the Austroads pavement design method varies with
the damage type being considered as shown in Table 2.12.

Table 2.12: Load damage exponents (LDE)


Design method Type of damage LDE
Mechanistic: pavement contains Fatigue of asphalt 5
one or more bound layers
Fatigue of cemented material 12
Rutting and loss of surface shape 7
Empirical: granular pavements with Overall pavement damage 4
thin bituminous surfacings
Adapted from Jameson (2012).

2.10.7 The 4th power law


The history of how the load damage exponent 4 was chosen for unbound granular
pavements, and what mode of damage was used to derive it, is an interesting one.

The American Association of State Highway Officials (AASHO) undertook an historic


series of road tests in relation to pavement research at Ottawa, Illinois in the US from
1958 to 1960 (Mamlouk 2006, Ullidtz 1987). These experiments were undertaken on
relatively thick asphalt surfacings subject to freeze and thaw cycles. After these
experiments Irick and Hudson (1964) amongst possibly others, adopted a ‘damage
exponent’ of 4. However, in Australia the AASHO road trials were not considered to be
directly applicable to the bulk of the Australian network with its chip seals and only very
thin asphalt, and not subjected to freeze and thaw in most cases (Jameson 2013;
Kinder & Lay 1988).

The topic attracted the attention of Scala (1970), who analysed the AASHO data to
form an initial premise (shown herein as Equation 2.32, being equation 2 of Scala
(1970)).

— 2-95 —
Chapter 2 Literature review

𝑊𝑊𝑑𝑑 4 2.32
𝐾𝐾𝑑𝑑 = � �
18

where

the relevant destructive effect of the test axle configuration and


𝐾𝐾𝑑𝑑 = loading compared to the standard axle configuration and
loading

𝑊𝑊𝑑𝑑 = the test loading (in kips)

18 = the standard loading of 18 kips (one kip = 1,000 lbs).

Equation 2.32 can be reconfigured as follows (equation 2.33).

𝐷𝐷𝑡𝑡 𝐿𝐿𝑡𝑡 𝑚𝑚 2.33


=� �
𝐷𝐷𝑠𝑠 𝐿𝐿𝑠𝑠

where

𝐷𝐷𝑡𝑡 = damage caused by the test axle

𝐷𝐷𝑠𝑠 = damage caused by the standard axle

𝐿𝐿𝑡𝑡 = load on the test axle

𝐿𝐿𝑠𝑠 = load on the standard axle

𝑚𝑚 = a load damage exponent.

Scala (1970) then undertook a study on a range of Australian pavements including


sprayed/chip seals and thin asphalt surfacings. Using a ‘scaled up’ version of the
Benkelman beam he was able to confirm that the load damage exponent of 4 applied
also to the Australian pavements tested. His 1970 paper states that m = 4, when
vertical elastic deflection was used as the measure for damage.

— 2-96 —
Chapter 2 Literature review

As discussed in Section 2.9.1, Kinder and Lay (1988) undertook to provide ‘a definitive
statement on the 4th power law and pavement damage caused by trucks’. They
summarised by saying that the 4th power law represented the best available tool, but it
must be used with care and within its limitations. One of the several limitations listed
was that it should not be applied without question to substandard and super standard
pavements, and this limitation continues to be raised by researchers who have followed
since, using words such as ‘high quality base materials’, ‘pavement strength’, ‘low
structural capacity’ and ‘structural number’.

The issue of axle load equivalency and load damage exponents in Australia was
revisited by Vuong et al. (2003). It was reported that when the 4th power law was first
derived from data collected in the 1950s AASHO road tests, the damage measurement
used was the present serviceability rating (PSR), later converted to the more objective
present serviceability index (PSI). In this newer work the Australian accelerated loading
facility (ALF) and the resulting net deformation (i.e. permanent deformation) were used
as the measurement of ‘damage’. Their analysis of load-damage power relationships
for the three pavements tested resulted in exponents ranging from 0.8 to 7.4. For
pavements with high quality base materials, power exponents were found to be in the
range of 3.5 to 4.1, considered very close to the 4th power law.

New Zealand, like many countries including Australia, has adopted the 4th power
relationship between axle loads and pavement wear. However, the vast majority of the
New Zealand road network consists of unbound pavements with a thin surfacing, quite
unlike the pavements from which the 4th power law was derived (de Pont et al. 2002).
de Pont et al. undertook indoor accelerated pavement testing using the University of
Canterbury accelerated pavement testing indoor facility (CAPTIF) to determine a new
empirical relationship between load and vertical pavement deformation for typical New
Zealand pavement designs of thin surfaced unbound granular pavements.

The key issue once again however, was what constitutes pavement wear. de Pont et
al. (2002) summarised the OECD DIVINE project (OECD 1998) as differentiating
between functional and structural condition for pavements. They considered that
functional condition reflects road user issues such as roughness, rutting and skid
resistance. Structural condition reflects engineering issues such as cracking and
pavement distress. The authors chose to define wear as a reduction in functional
condition and then further defined a measurement for this as permanent vertical
surface deformation (VSD).

— 2-97 —
Chapter 2 Literature review

Their CAPTIF study found that the pavement under review underwent two distinct
phases of VSD, an initial period of rapid change, named ‘compaction’, followed by a
period of linear change, named ‘wear’. They found that the power law approach was
not a useful way of modelling this wear: instead they devised a compaction-wear model
for VSD. This finding is important, but is dependent on pavement VSD. Its limitation
appears to be that it does not differentiate between mechanisms of pavement wear and
sprayed seal wear.

In the subsequent CAPTIF work conducted by Arnold et al. (2005b, 2005c), the
deterioration of sprayed seals was also investigated. Single-axle single-tyres were
loaded to 8, 10 and 12 tonne and run on a 3 coat chip-seal-surfaced pavement typical
of those used in NZ. Texture depth was measured using a laser profilometer and
reported as mean profile depth.

A load damage exponent (LDE) was calculated with the Austroads equations (now
found in Jameson 2012) using deterioration in chip-seal texture depth as the damage
indicator. For the 8 tonne versus 10 tonne test an LDE of 3.1 was calculated, while for
the 8 tonne versus 12 tonne test an LDE of 2.0 was calculated. The authors combined
both datasets to report an average LDE of 2.7.

When commenting on the other pavement-deflection-related LDEs of the project, the


authors suggested that the LDE is related to pavement strength. This suggestion was
also supported by work on Australian granular pavements (Yeo et al. 2006).

Research into the investigation of the load damage exponent of unbound granular
materials under Australian accelerated loading using ALF continued with the work
reported in Martin (2008) and Yeo et al. (2006).

Yeo et al. (2006) reported that for the high quality pavement material tested the load
damage exponent was 3.8 when using deformation as the measure of damage, and
3.4 when using roughness as the measure of damage. They considered these
numbers to be close to 4.

Martin (2008) reported that the present serviceability index (i.e. similar to that used by
AASHO) is based on some individual distresses weighted and combined and is highly
correlated to road roughness. He used data collected from ALF and derived a load
damage exponent of 3.4 (considered to be close to 4) using cumulative rutting as the
measure of damage.

— 2-98 —
Chapter 2 Literature review

With regard to thin asphalt pavements Chen et al. (2006) implemented full-scale
pavement testing inside a temperature and moisture controlled environment and used
a heavy vehicle simulator (HVS). They were able to vary the moisture content and
subgrade soil type and found a general trend where their load damage factor exponent
(LDFE) increased with increasing moisture content. They found that for materials with a
low structural capacity (structural number, SN) the 4th power law underestimated
damage and an 8th power law was more appropriate.

In summary, for unbound granular pavements (where most sprayed seals are found in
Aust, NZ and South Africa) the load damage exponent of 4 developed from the AASHO
road tests of the 1950s still appears to be roughly relevant today. It has been shown to
work for various definitions of pavement damage, such as elastic deformation
(deflection), plastic deformation (rutting), and serviceability index (roughness).

However, load damage exponents have also been found to be dependent on the
quality or type of pavement material used, by moisture ingress, and by pavement
strength. The exponent of 4 is normally applied to flexible, properly compacted
unbound granular pavements of highway quality material that are kept dry.

A summary table of different load damage exponents for different materials and
different modes of damage is shown in Table 2.13.

Table 2.13: Published load damage exponents


Material and failure mode Load damage Source
exponent
Pavement: fatigue of cemented materials 12 Jameson (2012): AGPT02-12

Pavement: thin asphalt over low structural number 8 Chen et al. (2006).
base
Pavement: rutting and loss of shape, flexible 7 Jameson (2012): AGPT02-12
pavements (Mechanistic)

Pavement: fatigue of asphalt 5 Jameson (2012): AGPT02-12

Pavement: overall granular pavement damage 4 Jameson (2012): AGPT02-12


(Empirical)
Chip seal: surface texture reduction on granular 2.7 Arnold et al. (2005b, 2005c)
pavement (single axle single tyre on 3 coat seal) (modified Kinder-Lay model)

2.10.8 The relevance of the 4th power law in seal design


The relevance or applicability of the 4th power law in seal design is an important
consideration in pursuing the objective of this thesis. This topic is discussed further in
Chapter 3, Developments in heavy seal design practice, section 3.7.

— 2-99 —
Chapter 2 Literature review

2.10.9 Axle spacing


There does not appear to be clarity regarding the effect of axle group spacing, and the
spacing between axles within an axle group, on pavement damage. The literature
search found that this topic is more widely discussed in publications related to the
bridge structural engineering field.

However, it does appear in work of Gillespie et al. (1993) and Hajek and Agarwal
(1990) who are still referred to in recent literature referring to the topic. Hajek and
Agarwal (1990) used both measured and calculated responses to axle loads to
evaluate the damage effects of dual and triple axles on flexible pavements as a
function of axle spacing. They concluded that the axle spacing had a significant
influence on pavement damage (not surfacing damage). Gillespie et al. (1993) used
analytical modelling to conclude, however, that maximum axle load and pavement
thickness have the primary influences on fatigue damage. Other truck attributes such
as number and location of axles, suspension type, and tyre type, are important but less
significant.

Gillman (1999) discussed axle spacing’s in the context of load equivalency factors. He
noted that according to AASHTO two axles spaced more than 1.0 m but not more than
2.4 m apart are considered to be a tandem axle group. Axles spaced more than this
are two separate single axles. He then calculated individual pavement load equivalency
factors at 0.1 m increments for axle spacing’s ranging from 0 to 4.0 m – which are of
little use in the sprayed seal context.

Hajek et al. (2005) state that in the US axles that are spaced more than 8 ft (2.44 m)
are called single axles. Closer than this they become either tandem, triple or quadruple
axles.

Owusu-Ababio and Schmitt (2015) state that the US definitions from AASHTO are

· single axle: an axle located at a distance greater than 8 ft (2.44 m) or at a


distance less than 3.33 ft (1.01 m) from an adjacent axle

· tandem axle: two adjacent axles with spacing 3.33 to 8 ft (1.01 to 2.44 m)

· tridem axle: three axles with spacing less than 12 ft (3.66 m) from the first to the
third axle.

— 2-100 —
Chapter 2 Literature review

Moffatt (2015) investigated whether the spacing’s of axles within a multiple axle group
affected the pavement’s strain response. He collected data from weigh-in-motion
(WIM) sites around Melbourne with the results shown in Table 2.14.

Table 2.14: Australian axle spacings obtained from WIM


Group Observations Percentiles of axle spacing (m)
50 90 95 97.5 99
Tandem 7,934,798 1.31 1.40 1.45 1.55 1.93

Triaxle 4,571,459 1.25 1.41 1.50 1.54 1.56


Adapted from Moffatt (2015).

He then modelled spacing’s of 1.5 m, 1.3 m and 1.0 m and concluded that choice of
axle spacing does not play a critical role in pavement design.

There appears to be a research gap as to what role axle spacing’s have with regard to
sprayed seal macrotexture deterioration. This current study may assist with this in a
somewhat ‘blunt’ sense by comparing deterioration from a set number of triaxle group
passes with three times that number of passes by a single axle.

2.11 Conclusions
This literature review started with the simplicity of sprayed seals being a layer of
bitumen covered with a layer of aggregate, and then introduced complexities such as
the time and temperature dependent rheological behaviour of bitumen, the theories
behind bonding and adhesion, and the ability to represent a road surface by an infinite
series of sinusoidal components of various wavelengths, which range from the
crystalline structure on the surface of each aggregate particle (microtexture) through to
the large wavelength unevenness (roughness) that affects riding comfort.

Many references are old because the basic understanding and definitions of sprayed
sealing has not changed significantly over the years, and the subject is sometimes still
referred to as a ‘black art’ rather than a science. We have also seen how sprayed
sealing has come from very pragmatic roots – for example the canned fruit jar being
the standard volume for volumetric measurement of surface texture, being no doubt
cheap, available and practical. These early roots are a long way from the analysis of
complex (being both real and imaginary) vectors from bitumen testing by the dynamic
shear rheometer. The latter illustrates some success in the progression to move away
from the ‘art’ and to focus more on the ‘science’.

— 2-101 —
Chapter 2 Literature review

It has been shown that one of the key purposes of a road surfacing is the provision of
skid resistance, and that surface texture (macrotexture) is a major contributor to skid
resistance (Cairney & Styles 2005, Oliver & Halligan 2006). The rate of surface texture
decay has been found to be rapid at first but then subsequently reduces. Attempts to
model this rate of surface texture decay have been sparse as the focus has been
instead on modelling the decay rates of pavement rutting and roughness.

A review of the various macrotexture deterioration models that have been found was
given in section 2.9.1. Some models have broken this decay into two linear stages of
decay, others have shown it to be a single logarithmic decay or an exponential decay.
Only one model was found that considered axle loading over a single-tyre single-axle,
but scrutiny raised some questions over its validity. The best of the other models
assume one ‘heavy commercial vehicle’ is about equivalent to ten cars (this is an old
South African assumption that has long since been abandoned in South Africa -
discussed in the next chapter). The worst of the models/observations simply use
surface age or years of service as a surrogate for trafficking.

This highlights a research gap: to develop an improved model that describes the
change of surface texture with time with regard to specific changes within the traffic
fleet, such as an increase in heavier axle loads or a greater use of triaxle or quad axle
groups.

An improved model would be useful in that it would no longer be necessary to


extrapolate past conditions into the future, but rather model the specific changes that
the future may bring to more accurately predict the outcome.

To fully understand the complexities of reduction in a sprayed seal macrotexture it was


necessary to introduce topics such as the complex rheology of bitumen, the use of
polymer modifiers, surfacing treatment selection and the measurement of pavement
performance. Some of these concepts will be used and developed further in Chapter 6
to assist in the interpretation of this study’s new model.

The objective of this research is to produce a model to describe the change in surface
texture as influenced by cumulative loadings. After reviewing the research gaps, the
hypothesis of this study can be stated as

Heavy vehicle axle group loads that cause the same pavement damage do not
necessarily cause the same sprayed seal damage.

— 2-102 —
Chapter 2 Literature review

Reaching the stated objective is necessary in testing this hypothesis. It will also be
necessary to determine a specific load damage exponent to predict sprayed seal wear.

As both the objective and the hypothesis of this study involves the effect of heavy
vehicles on sprayed seal performance it is appropriate that the next chapter will now
discuss how the effects of heavy vehicles are accounted for in various sprayed seal
design methods.

— 2-103 —
Chapter 3 Developments in heavy vehicle seal design practice

3 DEVELOPMENTS IN HEAVY VEHICLE SEAL


DESIGN PRACTICE
3.1 Introduction
The preceding literature review covers a very broad range of issues and topics to
provide some context for the study that follows. It is now appropriate to leave the ‘wide
angle view’ and commence zooming in to a narrower angle of view. This continuation
of the literature review begins such a transition by now considering seal design, and in
particular, seal design for heavy vehicles. Later chapters will close in to a narrower
field of view and enable examination of very fine detail, such as the effects of differing
axle groupings of heavy vehicles.

3.1.1 Seal designs


Road pavements made from crushed rock are often given a surface seal to protect the
pavement from moisture damage after rainfall, and are also sometimes sealed to
prevent loss of pavement fines through the traffic action raising dust. Such crushed
rock pavements are flexible under traffic loadings and require a flexible seal. A flexible
seal - variously known internationally as a sprayed seal, chip seal, or surface dressing
– consists of a thin coating of bitumen covered with a coating of single-sized crushed
rock aggregate. The role of the bitumen is to waterproof the pavement and to prevent
the raising of dust; the role of the crushed aggregate is to provide for grip for rubber
vehicle tyres and to provide a level of skid resistance.

The optimum thickness for the bitumen coat depends on two factors – the size of the
aggregate to be applied and the volume of traffic running over the seal. The heavier
the traffic volume the thinner the bitumen coating must be, otherwise bitumen may rise
(for a number of reasons) over the tops of the aggregate and skid resistance will be
lost. Conversely, if the bitumen coat is too thin there will not be enough bituminous
‘glue’ to hold the aggregate in place and the action of traffic in cold weather will pluck
the aggregate particles off leaving only slippery bitumen for vehicle tyres to contact.

The calculation for an optimum thickness of the bitumen coating is done through a seal
design process. This chapter will investigate the developments in the traffic component
of seal design. The composition of heavy vehicle traffic in Australia and many other
countries is now very different to when seal designs were first introduced.

— 3-1 —
Chapter 3 Developments in heavy vehicle seal design practice

It is intended for this study to add new knowledge to the area of sprayed seal design for
heavy vehicles on flexible pavements. Australia and South Africa are possibly the two
countries most interested in utilising this information: Australia in particular because of
its long freight haulage distances combined with a fragile and flexible granular
pavement road network. It would appear other countries employing sprayed seal
technology do not have the same ‘freight efficient’1 transport modes, and those
countries with heavier freight tasks than Australia seem to have economies large
enough to support a national heavy duty asphalt concrete pavement network rather
than a sprayed seal network.

However, the use of sprayed seals may change. The first international sprayed sealing
conference was held by ARRB in Adelaide in 2008. At the end of the two-day
conference, a workshop of invited experts was convened to distil the key topics and
issues identified by the conference. One such issue raised was an emerging increased
demand for sprayed seal usage (Neaylon 2008).

Unsealed roads are believed to be becoming more and more unsustainable. Unsealed
roads in South Africa are seen as wasting good ‘wearing gravels’ as they must
constantly be replaced (Ross & Field 2007) and sprayed seals are being seen
increasingly as a cost-effective pavement preservation treatment. On the other hand,
the rising cost of crude oil and preferred refinery configurations for maximising
petroleum and diesel production from crudes (Australian Institute of Petroleum 2004)
are drivers for an increased cost of bitumen. Any increased cost of bitumen has a flow-
on effect in the cost of asphaltic concrete, and this has seen a trend emerge in Texas
for the increased use of sprayed seals as a surfacing. Increased use of sprayed seals
will probably commence on the lower volume roads and then progress to the higher
volume roads (Neaylon 2008).

In the UK, both a desire to reduce input costs together with a desire for good surface
texture is leading to an interest in using sprayed seals on motorways (Neaylon 2008).

1
‘High productivity vehicles, such as B-doubles and higher mass limit vehicles are important to the
efficiency of the freight task in Australia. The larger capacity of these vehicles also reduces the number of
vehicles required to transport a given amount of freight.’ NHVR (2014).

— 3-2 —
Chapter 3 Developments in heavy vehicle seal design practice

Adams and Kim (2014) comment that in North Carolina, chips seals are ‘widely
accepted as both effective and economical’ and that ‘chip seals cover over half of the
road miles in the state’.

This then has highlighted the need for further research into improving seal designs to
more accurately accommodate the effect of heavy vehicles.

3.1.2 Focus of this chapter


The focus of this chapter is to identify gaps and opportunities in the way current seal
design practices deal with large heavy vehicles. A scan of international practices has
been undertaken of selected countries which currently utilise sprayed sealing. The
order in which the countries are listed could have been alphabetical; however, the
chapter is instead structured by commencing with the simplest approaches first,
through to the most complex. This is to enable a logical flow and introduction to the
subsequent experimental method chapter and then the data analysis chapter.

3.2 Tropical and sub-tropical countries


The UK Department for International Development (DFID) funded the international
division of TRL to devise and publish Overseas Road Note 3, A guide to surface
dressing in tropical and sub-tropical countries (TRL 2000). The project team considered
that the existing UK TRL road note 39 of the time 2 was not appropriate for most tropical
or sub-tropical countries, so they reverted to the method put forward by Jackson (1963)
as this had been validated in Kenya by Hitch (1981). It is believed the Jackson method
also incorporates concepts first put forward by Hanson in 1935 (TRL 2000) and is
discussed further in Section 2.5.1.

The Jackson method simply treats traffic as total traffic in vehicles per lane per day
(v/l/d). Should the volume of commercial vehicles exceed 20% then the next higher
traffic category and traffic factor from Table 3.1 should be adopted. Jackson worded it
as:

Until there is evidence of the effect of different axle loadings on surface dressings,
the parameter 'vehicles per day' should be interpreted literally without regard to
commercial vehicles necessarily, since these will constitute a proportion of traffic
on almost all roads. The presence of abnormal proportions of heavy vehicles
however, would clearly indicate a bias towards a heavier traffic category than that
indicated by the gross vehicle count (Jackson 1963, as reported by Hitch 1981).

2
Road note 39, the UK Design guide for road surface dressing, is now in its sixth edition. Refer to Roberts
& Nichols (2008).

— 3-3 —
Chapter 3 Developments in heavy vehicle seal design practice

Table 3.1: Weighting factors for surface dressing design

Traffic v/l/d Factor


Very light 0 – 100 +3
Light 100 – 500 +1
Medium 500 – 1000 0
Medium heavy 1000 – 3000 -1
Heavy 3000 – 6000 -3
Very heavy 6000 + -5
Source: adapted from TRL (2000).

This same table was republished in the Best practice manual for thin bituminous
surfacings (Ethiopian Roads Authority 2013). An example of Ethiopian heavy vehicles
is shown in Figure 3.1.

Figure 3.1: Heavy vehicles in Addis Ababa, Ethiopia, 2009

The factor for surface dressings in Table 3.1 is then added to other factors for existing
surface texture, climatic conditions and type of chippings to give a resultant factor F.
Using the ALD and F factor in equation 3.1 will then estimate the basic binder
application rate.

𝑅𝑅 = 0.625 + (𝐹𝐹 ∗ 0.023) + [0.0375 + (𝐹𝐹 ∗ 0.0011)]𝐴𝐴𝐴𝐴𝐴𝐴 3.1

where

𝑅𝑅 = Basic binder application rate (kg/m2).

𝐹𝐹 = The overall weighting factor

— 3-4 —
Chapter 3 Developments in heavy vehicle seal design practice

𝐴𝐴𝐴𝐴𝐴𝐴 = The average least dimension if the chippings (mm)

3.3 Republic of South Africa


In South Africa only a small percentage of roads carry more than 10,000 vpd although
there is a trend towards increased axle loads (SANRAL 2007). Heavy vehicles are
believed to make up approximately 15% of AADT. In the seal design method, heavy
vehicles are converted to equivalent light vehicles (ELV) by equation 3.1. (SANRAL
2007).

𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐸𝐸𝐸𝐸𝐸𝐸 = 𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 𝑙𝑙𝑙𝑙𝑙𝑙ℎ𝑡𝑡 𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 3.2


+ (𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 × 40)

where

𝐸𝐸𝐸𝐸𝐸𝐸 = equivalent light vehicles per lane per day.

In earlier documents an equivalency factor of ten, then 20 was used (COLTO 1998) but
increases in axle loads and tyre pressures necessitated the adoption of the higher
value of 40.

Milne mentioned this equivalency factor in his PhD thesis (Milne 2004) and has
suggested that actual equivalency factors may vary from 20 to 100, and that the factor
still requires refining.

van Zyl (2007) has reported that given the variability of vehicle types, loads and tyre
pressures there are doubts on the validity of the ‘one heavy vehicle accounts for 40
light vehicles’ equivalency factor, and more research is required to quantify individual
components that lead to this.

— 3-5 —
Chapter 3 Developments in heavy vehicle seal design practice

3.4 New Zealand


3.4.1 New Zealand Transport Agency
The concept of equivalent light vehicles (ELV) was introduced in 1993 based on South
African experience when a heavy commercial vehicle (HCV) was considered to be
equivalent to ten light vehicles or cars. However, seal design in NZ has progressed
with the adaptation of the Patrick model (Houghton & Hallett 1987; Patrick 1999) which
is published in Transit New Zealand, Road Controlling Authorities, and Roading New
Zealand (2005). A seal design method for a 100-day cumulative traffic appears unique
to New Zealand and parallels (but is not derived from) the concept of designing
pavements for a cumulative standard axle (traffic) load.

For basic cases the 2005 TNZ chip seal design algorithm still uses ELVs defined as
‘cumulative equivalent light vehicles based on the assumption that one heavy
commercial vehicle (HCV) is equivalent to 10 cars’ (from Equation 9-1 of Transit New
Zealand et al. (2005)). A heavy commercial vehicle in New Zealand is defined as any
vehicle over four tonnes gross weight (Arnold et al. 2005c).

However, for high risk design scenarios with little margin for error a more precise
algorithm can be used for converting vld traffic into ELVs as shown in equation 3.2.

𝐸𝐸𝐸𝐸𝐸𝐸 = 𝑣𝑣𝑣𝑣𝑣𝑣 × �1 + (0.09 × 𝑚𝑚)� 3.3

where

𝐸𝐸𝐸𝐸𝐸𝐸 = equivalent light vehicles per lane per day

𝑣𝑣𝑣𝑣𝑣𝑣 = vehicles per lane per day

𝑚𝑚 = percentage of heavy commercial vehicles HCVs (%).

For the typical highway case where HCV% is 11% this design algorithm calculates that
one ELV = 2 v/l/d and the factor of 2 can be considered to be a ‘heavy vehicle factor’.

If the HCV% was very high, say 40%, then the design algorithm would calculate a
heavy vehicle factor of 4.6.

The design traffic however is still reported in terms of (100 days cumulative) ELVs.

— 3-6 —
Chapter 3 Developments in heavy vehicle seal design practice

3.4.2 New Zealand Forest Owners Association


Chip seal forestry roads in New Zealand were subjected to axle loads up to 16 t per
axle in multi-axle groups and experienced severe bleeding. With the existing design
procedures proving inappropriate, Tasman Forestry Ltd and the Forestry Corporation of
New Zealand sponsored research into the problem and engaged the logging industry
research organisation and the University of Canterbury (Pidwerbesky et al. 1994;
Pidwerbesky & Arnold 1995).

Forty seal design test sections were established in January 1994 and a new theory of
chip seal design for heavy wheel loads commenced development. However, due to the
funding arrangements at the time, the findings of the research could not be released
into the public domain (Pidwerbesky et al. 1994).

Notwithstanding, some years later Pidwerbesky (2001) reported that:

• Standard penetration grade (i.e. unmodified) bitumen’s were insufficient to


handle the stresses and polymer modified binders with low application rates
showed most promise.

• Recommended traffic factors Tf for use in the RD286 NZ design algorithm were
0.66 for reseals using 180/200 pen (unmodified) bitumen and lower than 0.58
for reseals using a polymer modified binder or 80/100 pen bitumen.

The RD286 method of the late 1960s (NRB 1971) has since been superseded with the
2004 design algorithm (Transit New Zealand et al. 2005) and the RD286 traffic factors
quoted above are not directly translatable to the current method.

A recent New Zealand forest road engineering manual (NZ Forest owner’s association
2011) was sourced, but it is silent on the topic of chip seals.

3.5 United Kingdom


Road Note 39 is traditionally the method of chip seal designs in the UK. Shuler et al.
(2011) assert that the 4th edition of road note 39 of 1996 was originally based on some
of Hanson’s (1935) concepts combined with ideas of Jackson (1963). The 6th edition of
road note 39 was published in 2008 (Roberts & Nicholls 2008). In this edition no
reference is made to Hanson and instead the reader is referred to a catalogue for
design as discussed below.

— 3-7 —
Chapter 3 Developments in heavy vehicle seal design practice

The principal measure of traffic for design purposes is the number of commercial
vehicles – not light vehicles – per day travelling in the lane under consideration. A
commercial vehicle is defined as a vehicle of unladen weight greater than 1.5 tonnes
(Nicholls 1996) although it has been noted that with people carriers and off-road
vehicles becoming more popular, some of these are technically commercial vehicles.

Eight traffic categories (A to H) are then used in the design method (Roberts & Nicholls
2008). These are shown in Table 3.2.

Table 3.2: Road Note 39 traffic categories

Medium & heavy vehicles Traffic category


per lane per day
0 – 50 H
51 – 125 G
126 – 250 F
251 – 500 E
501 – 1250 D
1251 – 2000 C
2001 – 2500 B
2501 – 3250 B
3250 + A
Source: Adapted from Roberts & Nichols (2008).

The design binder application rate is then determined from a table showing traffic
category against road surface hardness (e.g. Table 9.2.1 of Roberts & Nicholls 2008).

3.6 North America


3.6.1 Minnesota
In the Minnesota Department of Transportation, the design procedure recommended is
based on McLeod (1969). This method is preferred (Wood et al. 2006) because it was
adapted for use by

• the asphalt institute (AI)

• the asphalt emulsion manufacturers association (AEMA)

• the strategic highway research program (SHRP).

— 3-8 —
Chapter 3 Developments in heavy vehicle seal design practice

For traffic input into the design, a traffic correction factor T is used, being the
percentage of the ultimate 20% void space that is to be filled. It is based only on
vehicles per day and there is no consideration given as to the composition of the traffic.

3.6.2 Texas
In the Texas Department of Transportation, chip seals are not recommended in traffic
above 10,000 vpd. The modified Kearby design method was first recommended in
1981, and is still the most commonly used method (Texas DoT 2010). Again, a traffic
correction factor T based only on vehicles per lane per day is used, with no
consideration to the composition of the traffic. The factor only applies to traffic volumes
less than 1,000 v/l/d and increases the standard (1,000 v/l/d) application rate by up to
20% for low volume roads. A version of the McLeod method is also given by Texas
DoT, developed primarily for use with emulsion binders with the qualification it has not
been verified in Texas. A different traffic correction factor T is used, this time being the
percentage of the ultimate 20% void space that is to be filled. Again, there is no
consideration given to the composition of heavy vehicles.

3.6.3 Kansas
Kansas DoT uses the Texas DoT method and modified Kearby method with traffic as
simple vehicles per day (Testa & Hossain 2014).

3.6.4 North Carolina


Adams and Kim (2014) of the North Carolina State University have commented that
one major issue with the use of chip seals in their opinion is the lack of a performance-
based seal design method. Major design methods in current use have evolved from
Hanson (1953) although in North Carolina the generally accepted chip seal asphalt
surface treatments (ASTs) are designed using Kearby’s (1953) method and McLeod’s
(1969) method, and these two existing chip seal design methods ‘are old and have not
been refined in years’.

— 3-9 —
Chapter 3 Developments in heavy vehicle seal design practice

Adams and Kim (2014) studied the change in sprayed seal macrotexture under field
and laboratory traffic loadings, with a view to progressing a new performance-based
chip seal design method that also encompasses new materials such as lightweight
aggregates and polymer modified emulsions. (Lightweight aggregates are synthetic
aggregates that can be made from clay, shale, slate etc. in a rotary kiln. They have a
low specific gravity causing less windscreen and vehicle damage and may also have
excellent skid resistance properties (Texas DoT 2010)).

To obtain laboratory samples for use in the third-scale model mobile loading simulator
(MMLS3) samples were fabricated in the field by constructing a field seal on templates
using a specified method, and the templates then transferred to the laboratory.

For the field study sections the authors used a low volume road with an average daily
traffic count (ADT) of 1,000 and a medium volume road with an ADT of 4,500.

In dealing with the type of traffic, the traffic count included the Federal Highways
Administration (FHWA) designated class of each vehicle (Figure 3.2 and Table 3.3) but
the field traffic data for each class was converted into equivalent wheel passes using
the FHWA vehicle class axle factors shown in Table 3.4.

— 3-10 —
Chapter 3 Developments in heavy vehicle seal design practice

Figure 3.2: FHWA 13 vehicle classification

Source: FHWA (2013), Figs 1.1 & C1.

— 3-11 —
Chapter 3 Developments in heavy vehicle seal design practice

Table 3.3: FHWA 13 category classification scheme

Broad Classification
categories number General description Definition Additional identifiers
1 Motorcycles 2 axles, 2 or 3 wheels. Also motor scooters, mopeds, and 3 wheel motorcycles.
2 axles. Can have 1 or 2 axle
2 Passenger cars Short-bed pickup (5-6’), no extended cab; SUVs; minivan; sedan.
Passenger trailers.
carriers 2 axles, 4 tire single units. Long-bed pickup (8’), no extended cab; short-bed and long-bed pickups with extended cab or 4 doors;
3 Pickups, panel vans
Can have 1 or 2 axle trailers. conversion van; full-size work van; limousine - regular; short-bed dually.
4 Buses 2 or 3 axles, full length. School; transit; private; commercial. Does not include compact school buses.
2 axle, 6 tire, (dual rear tires), Approx. 21’ steering to rear axles; 8’ bed dually with 4 full doors; dump or sewage truck (with or
5 Single-unit trucks
single-unit trucks. without 2 axle trailer); compact school bus or 4 full doors; extended limousines.
6 Single-unit trucks 3 axles, single-unit trucks. Dump truck; single tractor with 3 axles and no trailer; oil field equipment.
4 or more axles, single-unit
7 Single-unit trucks 4 or more axle trucks on a single frame.
trucks.
3 or 4 axle, single-trailer 2 axle truck/tractor pulling single 1 axle trailer; 2 axle pulling single 2 axle trailer; 3 axle pulling single 1
8 Single-trailer trucks
trucks. axle trailer.

Commodity 3 axle truck/tractor pulling single 2 axle trailer (18 wheeler); 2 axle pulling single 3 axle trailer; dump
9 Single-trailer trucks 5 axles, single-trailer trucks.
Carriers truck pulling 2 axle trailer.
6 or more axles, single-trailer
10 Single-trailer trucks 3 axle truck/tractor with single 3 or more axle trailer.
trucks.
5 or less axles, multi-trailer
11 Multi-trailer trucks 2 axle truck/tractor with 2 trailers, the first trailer with 1 axle, the second trailer with 2 axles.
trucks.

12 Multi-trailer trucks 6 axles, multi-trailer trucks. 2-3 axle truck/tractor with 2 trailers, the first trailer with 1-2 axles, the second trailer with 2 axles.

7 or more axles, multi-trailer


13 Multi-trailer trucks 3 axle truck/tractor with 2 or more trailers.
trucks.

Source: Adapted from Texas DoT (2012), Ch4, Sect 1.

— 3-12 —
Chapter 3 Developments in heavy vehicle seal design practice

Table 3.4: FHWA vehicle class axle factors

FHWA Class Definition Average number of axles


4 Buses 2.186
5 Two-axle, six tyre, single unit trucks 2
6 Three axle single unit trucks 3
7 Four or more axle single unit trucks 4
8 Four or fewer axle single trailer trucks 3.676
9 Five axle single trailer trucks 5
10 Six or more axle single trailer trucks 6.037
11 Five axle multi trailer trucks 5
12 Six axle multi trailer trucks 6
13 Seven or more axle multi trailer trucks 7

Source: adapted from Adams & Kim (2014) and Schmoyer et al. (1998).

They then compared surface texture reduction (measured as mean profile depth MPD)
against the cumulative individual axle passes.

This seems to be an indication of the current state-of-the-art view as to how the effect
of heavy vehicles on a chip seal is being examined in the US.

However, the concept of converting heavy vehicles into average axle passes still
leaves much room for improvement when considering the effects of heavy vehicles and
axle groupings on seal design. This will be raised again in sections 5.3.3 and 5.3.4.

A National Cooperative Highway Research Program (NCHRP) report Chip seal best
practices of 2005 concluded that ‘Americans and Canadians can learn from the
procedures used in Australia and New Zealand’ (Gransberg & James, 2005).

Since this 2005 comment a seal design method based on the Australian Austroads AP-
T68/06, Update of the Austroads sprayed seal design method (Alderson 2006) has
been recommended for use in the US. This resulted from NCHRP Project 14-7 which
was set up to provide a rational method for the design of chip seals over hot mix
asphalt pavements. In this project report (Shuler et al. 2011) the Australian approach
of a traffic adjustment for heavy vehicles was recommended and included the
Australian definitions and allowances for heavy vehicles, large heavy vehicles and
equivalent heavy vehicles. These definitions may in fact be compatible with the FHWA
13 category classification scheme (Figure 3.2 and Table 3.3).

— 3-13 —
Chapter 3 Developments in heavy vehicle seal design practice

The Project 14-7 report was then endorsed and rebadged by the American association
of state highway and transportation officials in 2012 (AASHTO 2012).

However, this progression does not appear to have been widely adopted in any recent
State DoT publications and manuals at the time of this study.

3.7 Australia
3.7.1 Current method
In the Austroads seal design method (Alderson 2006), a general mix of light and heavy
vehicles (HV) is assumed with the HV% assumed to be between 5% and 10% of the
total. If HV% is greater than 10% then the actual percentage should be determined
and the design escalates to the next level of complexity.

There have been an increasing number of large heavy vehicles (LHVs) in rural areas of
Australia, these vehicles being defined as Austroads vehicle class 10 and above (refer
to Table 3.5). Data was collected from rural areas and based on this the method for
traffic adjustment was amended and updated.

The concepts of damage factor (DF) and equivalent heavy vehicle (EHV) will be
introduced very briefly now, but are discussed in more detail in Section 3.7.2.

The current ongoing detailed analysis of the loading effect of HVs and the current
sprayed seal design method was first presented to Austroads in 2007 where the
concept of DFs and EHVs was accepted. However, the data set used within the report
was deemed outdated and the report not published (it has been updated since as
discussed later in Section 3.7.2). The concept of DFs and EHVs was also presented at
the 1st international sprayed sealing conference (Holtrop 2008; Neaylon et al. 2008)
where it met no apparent objection.

— 3-14 —
Chapter 3 Developments in heavy vehicle seal design practice

Table 3.5: Austroads vehicle classification system

Source: DTEI (2007).

— 3-15 —
Chapter 3 Developments in heavy vehicle seal design practice

Holtrop (2008) reported that during the development of the design method two options
were considered:

1. Convert heavy vehicles to light vehicles. This option was rejected as it was
decided it was important for the designer to separately consider the effect of
heavy vehicles in the design process.

2. Use the equivalent standard axles (ESA) concept as used in pavement design.
This option was rejected as it was found to not correlate well with seal
performance.

Neaylon et al. (2008) had stronger views as to why adopting the equivalent standard
axle concept as used by pavement designers was not adopted:

1. A previous investigation by the Austroads working group during the


development of the Revision 2000 method (Austroads 2001) did not find any
correlation between ESAs using the 4th power damage exponent and seal
performance.

2. Seal damage effects are more related to the rolling action of loaded tyres, which
reorientates the aggregate and alters the voids in the aggregate mosaic that are
available to be filled with binder. This action is different to the structural load
carrying capacity of a pavement.

3. A triaxle group and a single axle may have different damaging effects to a
sprayed seal yet they may both be equal in equivalent standard axle value.

4. A seal damage model must be sufficiently robust to evaluate the relative effects
of heavy vehicles outside of the current Austroads classes but which
predominate on some sealed roads (e.g. quarry roads with class 4 trucks
towing trailers with various types of axle groups).

5. Calculating a theoretically accurate ESA requires complexities that are not likely
to be mastered by many seal design practitioners.

6. Calculating any value of ESA for sprayed seal use would be different in effect to
pavement design use as sprayed seal practitioners consider traffic over the
early life of a seal, not over 20 years or more.

— 3-16 —
Chapter 3 Developments in heavy vehicle seal design practice

7. The EHV concept as already used had been well-received in training courses
and appears to be accepted and understood by most people involved in seal
design.

It is worth highlighting that a review of available literature revealed that none of these
items, in particular items 2, 3, 4, could be answered at the time and an interest in
finding answers to them (and finding a seal damage model) is the reason why this
current study commenced.

Whilst this 2007 and further work was being updated however, Austroads adopted as
an interim measure an empirical (i.e. derived by committee) relationship. Large heavy
vehicles are deemed to have four times the damaging effect of the more traditional
heavy vehicles which comprised the traffic fleet at the time the first Australian seal
design was documented in 1965 (NASRA 1965). Thus large heavy vehicles are
multiplied by a factor (4 – the 1 already included in HV% = 3) and converted into
equivalent heavy vehicles. This is expressed as

𝐸𝐸𝐸𝐸𝐸𝐸 % = 𝐻𝐻𝐻𝐻 % + (3 × 𝐿𝐿𝐿𝐿𝐿𝐿 %) 3.4

where

𝐸𝐸𝐸𝐸𝐸𝐸 = equivalent heavy vehicles (%)

heavy vehicles (%). Heavy vehicles are defined as any vehicle


𝐻𝐻𝐻𝐻 = with more than two axles or with dual tyres on the rear axle. All
other vehicles are regarded as cars (Bennett et al. 2009)

large heavy vehicles (%). Large heavy vehicles are defined as


B-doubles and other heavy truck/trailer combinations with seven
𝐿𝐿𝐿𝐿𝐿𝐿 =
or more axles, i.e. Austroads vehicle class ten and above (Table
3.6).

The EHV value calculated by this method does not alter the actual design traffic
volume but rather is applied as an additional adjustment factor to the aggregate’s basic
voids factor of the design method.

— 3-17 —
Chapter 3 Developments in heavy vehicle seal design practice

3.7.2 Current investigations


The concept of damage factors (DFs) and equivalent heavy vehicles (EHVs) was first
presented to the Austroads bituminous surfacings research reference group (BSRRG)
by Spies, circa 2005. The most recently published status of this work is described in
Khoo (2014).

In Khoo’s work, weigh-in-motion (WIM) data from 2007 to 2011 from rural regions in
Australia was collected by Australian jurisdictions. (A WIM system is: ‘a device that
measures the dynamic axle weight of a moving vehicle to estimate the corresponding
static axle mass’ (Koniditsiotis 2000)). A total of 126 million observations from 142
WIM sites were collected and analysed using IBM SPSS. This analysis was focussed
in two areas:

1. Damage Factor
The effect of increasing heavy vehicle mass limits on unbound granular pavements
with chip seal surfacings was investigated by Land Transport New Zealand (Arnold et
al. 2005c). In accordance with standard pavement design practices, a damage factor
was defined which could also be applied to sprayed seal damage.

For a standard axle grouping the seal damage factor is given by equation 3.5 (Khoo
2014).

𝐿𝐿𝐿𝐿𝐿𝐿
𝑖𝑖=𝑛𝑛 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑜𝑜𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑖𝑖 3.5
𝐷𝐷𝐷𝐷 = � � �
𝑖𝑖=1 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑓𝑓𝑓𝑓𝑓𝑓 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑖𝑖

where

𝐷𝐷𝐷𝐷 = damage factor

𝐿𝐿𝐿𝐿𝐿𝐿 = load damage exponent.

Arnold et al. calculated LDEs in terms of deterioration in chip seal texture depth of 3.1
for an 8 t to 10 t comparison, and 2.0 for an 8 t to 12 t comparison. They combined
these datasets and reported a LDE of 2.7.

Verification of this load damage exponent for Australian pavements and surfacings is
another key reason for undertaking this current study.

— 3-18 —
Chapter 3 Developments in heavy vehicle seal design practice

Pending the results of this study being published, Austroads (Khoo 2014) has
continued to use a load damage exponent of 2.7 in her work of refining the traffic
adjustment factor by updating and reanalysing the loading effect of heavy vehicles on
seal surfacings.

2. Equivalent heavy vehicle based on Damage Factor (EHVDF)


Spies was the originator of the concept of equivalent heavy vehicle based on a damage
factor, however most of his work remains unpublished and was only circulated within
the Austroads committee (Spies 2006a, 2006b, 2006c, 2007). Spies based the
approach on the premise that heavy vehicle damage (D) to sprayed seals is caused by
tyre pressure (P) × tyre contact area (A) for every tyre in unit time, i.e.

𝐷𝐷 ∝ 𝑃𝑃 × 𝐴𝐴 3.6

𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 (𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿)
and 𝑃𝑃 = 3.7
since 𝐴𝐴

then 𝐷𝐷 ∝ ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 (𝑡𝑡) 3.8

Following his work, the committee agreed to define one EHV as a class 4 truck loaded
to 50% legal payload (i.e. 14.75 t gross) proportionally shared between its axle groups
to provide 4.75 t on the first single axle with single tyre (SAST) steer axle and 10.0 t on
the rear tandem axle with dual tyres (TADT) drive axle group (Spies 2007).

Figure 3.3: Typical Austroads Class 4 vehicle

Source: Luk (2006).

The concept of converting large heavy vehicles into equivalent heavy vehicles required
a fundamental decision: what class of heavy vehicle shall be chosen as the standard
reference for all other vehicles to be converted back to?

There were two reasons for nominating Class 4 as the reference vehicle.

— 3-19 —
Chapter 3 Developments in heavy vehicle seal design practice

Firstly, the development of the Australian sprayed seal design evolved around the
prevalent commercial vehicles of the time. Rigid or ‘body’ trucks were popular in
Australia in the 1950s and 60s (Ramsay & Prem 2000). Hence, early Australian
sprayed seal designs were based on the traffic of the day which consisted of
passenger cars – Austroads class 1, and rigid commercial vehicles – Austroads
classes 3, 4 and 5 (refer to Figure 1.3). Prime movers towing semi-trailers (e.g.
classes 7, 8, 9) only became more widely accepted later with the introduction of the
quick-connect kingpin and turntable (Ramsay & Prem 2000).

The second reason class 4 was finally chosen was because, although class three was
chosen initially, subsequent studies into the WIM data (Spies 2007) indicated erratic
trends between the Australian states for class 3 vehicles due to suspected operational
inconsistencies at the time of data capture, processing and/or calibration for this
particular class (e.g. inclusion or otherwise of ‘lighter’ commercial vans). On the other
hand, the between-state variation for class 4 loadings (coefficient of variation of 8.4%)
was quite uniform.

With the reference vehicle now chosen, calculations for equivalent heavy vehicle based
on damage factors may then be undertaken using equation 3.8.

𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑓𝑓𝑓𝑓 𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 3.9


𝐸𝐸𝐸𝐸𝐸𝐸𝐷𝐷𝐷𝐷 =
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡ℎ𝑒𝑒 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 4 𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖

where

equivalent heavy vehicle value for an individual class based on


𝐸𝐸𝐸𝐸𝐸𝐸𝐷𝐷𝐷𝐷 =
damage factors.

From equation 9 damage factor EHVDF ‘s have been calculated for the typical
Australian heavy vehicle fleet as shown in Table 3.6.

— 3-20 —
Chapter 3 Developments in heavy vehicle seal design practice

Table 3.6: Summary of Australian heavy vehicle fleet based on damage factor (EHVDF)

Class National average


Fleet (%) EHVDF
3 32.7 0.6
4 7.1 1.0
5 1.1 0.8
6 1.8 0.5
7 3.4 0.9
8 3.7 1.4
9 22.7 1.7
10 19.3 2.3
11 4.5 2.9
12 4.5 3.7
Adapted from Khoo (2014).

Table 3.6 can be interpreted as follows: Class 3 vehicles form 32.7% of the national
heavy vehicle fleet, but each class 3 vehicle only causes 0.6 × the damage of the
reference class 4 vehicle. On the other hand, one class 11 vehicle causes 2.9 × the
damage of one reference class 4 vehicle.

The values of EHVDF in the table are dependent on the load damage exponent for
sprayed seal wear of 2.7 being correct. Validation of this load damage exponent is
discussed further in Chapter 5.

3.8 Seal design comparisons


A comparison of the design methodologies of the USA’s Asphalt Institute/McLeod
method, the USA’s Texas/Epps/Kearby method, Republic of South Africa TRH3 2007
method, Australian Austroads method and the UKs TRL 1996 method, was undertaken
by Shuler et al. and is detailed in Appendix J of Shuler et al. (2011). However, being
an American study, the traffic data used for the comparison was simply 1,000 total
vehicles per lane per day and the results are of little value for comparing current heavy
vehicle design.

A comparison of the USA (McLeod method) and the Australian (Austroads) method
was undertaken for a master’s degree (Einarsson 2009) in conjunction with the
University of Washington and the Washington state department of transportation
(WSDOT). Once again, being a USA led study, the traffic data used for comparison
was simply total vehicles per lane per day. Where use of the Australian method
required a percentage of commercial vehicles, an assumption was made which
appears to have been done to keep the calculations simple.

— 3-21 —
Chapter 3 Developments in heavy vehicle seal design practice

3.9 Summary
An overview of global practice shows diverging approaches to the handling of heavy
vehicles in sprayed seal design as shown in Table 3.7.

Table 3.7: Summary of seal design traffic inputs


Country Raw data used Raw data converted for design
input into units of:
Average Commercial Large
annual daily vehicles heavy
traffic vehicles
UK influenced tropics and Yes No No Total vehicles/lane/day
sub-tropics
Republic of South Africa Yes Yes No Equivalent light vehicles/lane/day
New Zealand – simple Yes Yes No Equivalent light vehicles/lane/day
designs
New Zealand – complex Yes Yes No Equivalent light vehicles/lane/first
designs 100 days accumulated
United Kingdom No Yes No Commercial vehicles only/lane/day
North America Yes No No Total vehicles/lane/day
Australia – simple Yes Yes No Total vehicles and % commercial
designs vehicles/lane/day
Australia – complex Yes Yes Yes Total vehicles and % equivalent
designs heavy vehicles/lane/day

In one extreme, heavy vehicles are considered part of the average daily traffic count
and ignored. In the other extreme, traffic is classified solely as the number of
commercial vehicles per day and the light vehicles, or passenger cars, are ignored.
Then there are countries where the medium and heavy vehicles are converted back
into light vehicles or passenger car equivalents. Australia, on the other hand, converts
medium and heavy vehicles into heavy vehicle equivalents (EHVs) and yet utilises the
overall average daily traffic count as well.

This raises some interesting philosophical questions. Should average daily traffic,
which is often swamped by the effect of passenger cars, even be retained? What
influence do passenger cars have on the deterioration of skid resistance (as measured
by macrotexture) compared to commercial vehicles? Should seal design be based
entirely on the effect of heavy vehicles? If so, could such a transition be managed?
Could the best features of each country’s seal design method be distilled and one
international method be postulated? There are a number of issues here which would
benefit from further research but are beyond the scope of this present study.

— 3-22 —
Chapter 3 Developments in heavy vehicle seal design practice

One practical solution could be to develop a two-tier seal design method (for Australia
as a first step): firstly a simple proven method for light traffic and low risk, based on
average daily traffic; and then a more complicated supplement which would add
specific detail and more complex considerations (such as converting the HV count into
EHVs) when required for higher risk designs.

The questions raised here are for pursuing at a later date. The resources of this study
are instead focused on determining whether a) the concept of heavy vehicle ESAs as
developed for pavement designers is applicable for sprayed seal design, which then
leads to a seal damage model, and b) the determination of a load damage exponent
applicable for an Australian sprayed seal. The following chapter will now discuss the
method used in examining these two questions further.

— 3-23 —
Chapter 4 Experimental Method

4 EXPERIMENTAL METHOD
4.1 Introduction
The previous chapter has described how different seal design methods cater for the
effects of heavy vehicles. Some methods ignore heavy vehicles, some convert them
into ‘passenger car equivalents’, while the most developed method converts them into
standardised ‘heavy vehicle equivalents’. This conversion was based on empirical
observation and experience. To progress this method from an empirical beginning
towards a more mechanistic direction data must be gathered and aspects of the
fundamental mechanisms interpreted.

This chapter describes the approach and methods employed for the acquisition of data
to describe the reduction in sprayed seal surface texture under various cumulative
loadings of single, tandem and tri-axle groupings.

Many potential variables (as detailed in 4.6.3) are likely to have interacted with the
response being measured if they were not held as constant as possible. For this
reason, it has not been logical to use data gathered from field sites around Australia.
For example, an early consideration was to completely use field testing and to vary the
traffic loading by varying the location study, for example one local road site, one state
highway site and one national highway site, with accurate traffic classification counts
for each site. However, this would have added significant noise to the data set, such
as

· different pavement materials and different construction quality (leading to


different aggregate embedment into the base courses)

· different seal ages (leading to a variety of binder viscosities to resist aggregate


reorientation)

· different rainfalls (leading to variable softening of basecourse near the


shoulders or to variable binder rise caused by water vapour)

· different pavement temperatures (aggregate reorientation will occur faster at


higher temperatures because of lower binder viscosities)

· unknown extraneous vehicles such as overweight vehicles (one outlier load


could swamp the damage effect of the average daily traffic count).

— 4-1 —
Chapter 4 Experimental Method

Accelerated pavement testing provides an opportunity to eliminate these variables.

Full scale accelerated pavement testing (APT) is used by pavement engineers around
the world to improve understanding of pavement behaviour and to evaluate new
pavement materials, additives and alternative materials processing, and new
construction techniques (Jones et al. 2012). It also provides the ability to validate and
calibrate new pavement models with minimal risk and a lower cost than full length
construction.

Strategic goals set by APT program planners may include: performance of new and
alternative structural designs; evaluation of new road materials and test methods;
development of performance modelling for predicting pavement deterioration;
determination of remaining life of pavements; and truck/pavement interaction (Sharp
2001).

Full scale accelerated pavement testing can take several forms, for example:

· the closed-loop test track in Alabama operated by the National Centre for
Asphalt Technology (NCAT) and the Auburn University can be loaded around-
the-clock by a fleet of full sized vehicles driven manually.

· a circular pavement test bed known as the Canterbury accelerated pavement


testing indoor facility (CAPTIF) in Christchurch New Zealand, and operated by
the NZ Transport Agency and the University of Canterbury, is loaded
automatically via a rotating beam with loaded wheels at each end.

· a circular pavement in Nantes, France, known as le manège de fatigue (the


fatigue carousel), is operated by l’Institut Français des Sciences et
Technologies des Transports, de l’Aménagement et des Réseaux (IFSTTAR,
The French institute of science and technology for transport, development and
networks) and is loaded automatically via radial rotating arms.

· the linear automated device known as the accelerated loading facility (ALF) in
Melbourne, Australia, operated by ARRB Group is loaded with a conservation of
energy system remotely similar to wheels on a pendulum but contact is made in
one direction only. The same devices also operate at the FHWA Turner
Fairbank centre near Washington, DC. The ALF will be discussed further in this
chapter while CAPTIF will be discussed further in the following chapter.

— 4-2 —
Chapter 4 Experimental Method

Prior to this research work commencing, a program to examine the effect of these axle
groupings on unbound granular pavements had already been established and funded
by Austroads: Project TT1219, The influence of multiple axle loads on pavement
performance (Moffatt 2011). That program was to collect accelerated loading facility
data over a two-year period at a running cost of about AU$1m per year, so it was
possible to gather additional data for the research questions discussed in this thesis at
very little additional cost. Consequently, the method developed for the pavements
Project TT1219 was able to be modified to accommodate this surfacings study.

This chapter will introduce ALF, and discuss some limitations of this research work
being a ‘child’ project in relation to the larger ‘parent’ project TT1219. The chapter then
introduces pavement and sprayed seal design and construction following which the
experimental design for this particular study is discussed. As sprayed seal surface
texture deteriorates under heavy traffic loadings, the aim of the data collection is to
record this rate of surface texture decay by constraining as many unknowns as
possible and changing the axle grouping configuration and axle loadings. The resulting
data is a string of surface texture measurements in units of mm of depth, relating to a
known number of passes of a known axle group under a known loading.

The chapter then concludes with a discussion of the experimental limitations.

When this methodology using ALF was being planned in 2007, it was believed to be
the first time that a full scale accelerated pavement test device had been fitted with
single, tandem and tri-axle assemblies, potentially making the resulting data sets
unique in the engineering community. It was later found that IFSTTAR, the French
institute of science and technology for transport, development and networks, used the
pavement fatigue carrousel also with single, tandem and triaxle groups, loaded at 42.5
kN per tyre on two different asphalt pavements (Homsi 2011; Homsi et al. 2011; Moffatt
et al. 2014).

Nonetheless, it is still believed that this is the only multiple axle group data set for a
chip seal/sprayed seal on unbound granular pavements. The data sets may not be
perfect, but they are believed to be useable. They provide a first step in the journey
towards mechanistic understanding of sprayed seal behaviour and possibly a starting
point for those who may follow.

— 4-3 —
Chapter 4 Experimental Method

4.2 The Accelerated Loading Facility (ALF)


4.2.1 Equipment background
In 1978 the principal technical committee of the national association of Australian state
road authorities established a working group to ‘further develop and implement
proposals for full-scale pavement test facilities to research the factors causing
premature distress, cracking and rutting of heavily-trafficked pavements’ (Sharp et al.
1999). This resulted in the 1982 commissioning of the Department of Main Roads
(DMR) New South Wales to design and construct an accelerated loading facility (ALF),
to enable short timescale performance assessment of various road pavement
configurations. The Australian Road Research Board (ARRB) formally took possession
of this ALF on 12 July 1984, and continues to manage and operate ALF (Figure 4.1) on
behalf of Austroads (ARRB 2014; Sharp et al. 1999).

Figure 4.1: ALF operating indoors at Dandenong South

Image: Mike Moffatt.

The following schematic (Figure 4.2) and quotation taken from Moffatt et al. (2011)
describes the mechanical engineering details of ALF.

— 4-4 —
Chapter 4 Experimental Method

Figure 4.2: Original ALF schematic

Image: Moffatt et al. (2011).

ALF uses a directly driven load trolley to apply rolling wheel loads in a
single direction to pavement strips, at a constant speed, using a
constant mass. The load is applied to the test pavement through a
load assembly trolley, consisting of a standard heavy vehicle hub and
wheel assembly, chassis and weight bed. This trolley tracks linearly,
guided by rails mounted on a stationary main frame and is driven by
electric motors mounted directly onto the wheel hub.

The wheel is lifted off the pavement at the end of each cycle and
supported by the main frame on its return. By loading in a single
direction only, the ALF machine is able to simulate real-world
trafficking conditions, and the raised rails at the ends of the machine
allow for the conservation of the majority of the trolley’s kinetic energy.

The trolley applies a sprung, via air springs, static load, and a
hydraulic lift cylinder on the loading trolley is used to control the
raising and lowering of the test wheel during each loading cycle. The
hydraulic cylinder is not used to apply load to the pavement. The front
end of the trolley is fixed to the lower chassis axle, and the trolley is
able to pivot at this location when the rear end of the trolley is lifted by
the hydraulic cylinder.

The load applied to the pavement can be varied from 40 kN to 90 kN


in 10 kN increments, by adding ballast weights to the trolley above the
axle assembly.

The cycle time for each load is about 10 seconds, which corresponds
to approximately 350 load cycles per hour or, depending on the
percentage of operating time, about 50,000 cycles per week (based
on 22 hours per day operation). ALF can be set up to apply loads
using dual, single or super-single wheel types. The loading can be
channelised or applied over any transverse distribution pattern within
a 1.2 m width. A normal distribution of transverse locations covering a
1.0 m wide trafficked area is commonly used to simulate the typical
traffic wander occurring on a road (Moffatt et al. 2011).

— 4-5 —
Chapter 4 Experimental Method

4.2.2 Experimental background


In 2006 Austroads initiated a research project to improve the understanding of the
relationship between different multiple axle group loads (single, tandem, and triaxles,
as shown in Figure 4.3) on Australian flexible pavement types and road pavement
performance (Moffatt et al. 2011).

Triaxles
Tandem
axles Single
axle

Figure 4.3: Typical multiple axle groupings

One component of this project was the trafficking of a typical unbound granular
pavement with a sprayed seal surfacing by the accelerated loading facility (ALF)
using single, tandem and triaxle axle groups (note that only half axles were fitted and
thus half of the total axle loads required).

It is not common for this machine to test a sprayed seal surfacing over an unbound
granular pavement - asphaltic concrete pavements are more common - nor for
multiple axle groups to be fitted (Figure 4.4 and Figure 4.5) - and the opportunity was
taken to collect time-series surface texture data from the sprayed seal in addition to
the broader experiment being undertaken by others.

— 4-6 —
Chapter 4 Experimental Method

Figure 4.4: ALF with tri-axle assembly fitted

Figure 4.5: ALF with tri-axle assembly fitted, showing dual tyres

4.2.3 Tyre and axle assembly characteristics


The tyres fitted were representative of what the pavement designers believed were the
most commonly used truck tyres in Australia, Michelin 11R22.5 configured as dual
tyres at 330 mm centres. The tyre pressures were checked daily. The target pressure
was 750 kPa when cold and under load and 760 kPa when hot and under load.
Pressures were corrected when the deviation was found to be greater than 10 kPa.

— 4-7 —
Chapter 4 Experimental Method

Each axle configuration was individually sprung and damped using airbags and
standard heavy vehicle shock absorbers. The detachable axle configurations were
mounted through a pivoting plate designed to ensure that the overall self-weight of the
assembly was shared evenly across each axle. When operating in tandem or triaxle
configurations the drive motor – a 22 kW bevel gear/motor powered by a three-phase
514 V variable speed drive – was attached to one axle only, leaving the remaining
axles to free wheel (Moffatt et al. 2011; Moffatt 2013).

The axle spacing within the multiple axle groups was 1.25 m.

The speed of the rig could be varied from zero to reduced speed (16 km/h) to full speed
(22 km/h). It was operated at reduced speed during the seal bedding-in process to
minimise risk of damage to the seal and then operated at full speed for the work under
study to minimise the cycle time of the testing.

4.3 Pavement
Due to the nature of the primary parent project TT1219 the pavement design was
undertaken by others. Full details of the pavement material and design are therefore
published in Moffatt (2011); Moffatt et al. (2011); and Moffatt (2013).

In summary, the pavement chosen for the experiment was an unbound granular
pavement with a thin bituminous surfacing. The pavement was designed such that the
materials and thickness were representative of a typical (i.e. not heavy duty) Australian
state road. A low to mid-strength clay subgrade was used to enable permanent
deformation of the pavement to be measurable. The basecourse consisted of a 20 mm
VicRoads Class 2 crushed rock (VicRoads 2007). Class 2 is a high quality pavement
base material for unbound flexible pavements in locations where a very high standard
of surface preparation may not be required. Class 2 crushed rock does not have a
minimum Plasticity Index or a maximum permeability requirement. The basecourse
rock is a Hornfels from Boral Lysterfield, and can be described as a metamorphic rock
formed by the intrusion of granite into the overlaying sediments. It is regarded as
essentially unweathered, finely crystalline, hard, and strong (Boral 2016). This material
was used as it was deemed representative of the crushed rock quality for the
Australian rural network (Moffatt 2011). A schematic of the pavement configuration is
shown in Figure 4.6.

— 4-8 —
Chapter 4 Experimental Method

Figure 4.6: Structure of test pavement

Source: Moffatt (2015).

The test pavement was screened by use of a falling weight deflectometer (FWD) prior
to experimental work proceeding and again after trafficking. These deflections are
recorded in Appendix G of Moffatt (2013). For project TT1219 permanent deformation
of the pavement was to be the performance measure monitored and uniformity of
material quality was considered important. Deflections were one way of measuring the
as-constructed uniformity of the material. The pavement at location 3509 was rejected
due to the large variations in deflections along the length of the experiment location.

4.4 Surfacing selection


Due to the nature of the parent project the non-elastic deformation of the pavement
was being measured as a key performance indicator. The surfacing selected was
therefore required to be completely flexible and to offer no component of structural
support to that provided by the underlying pavement. This then excluded the use of
concrete or a thick (≥ 40 mm) dense graded asphalt. In Australia stone mastic asphalt
and open graded asphalt are not recommended or used as surfacings directly over
unbound granular pavements and these were not considered. Thin (≤ 40 mm) dense
grade asphalt, bituminous slurry seals and microsurfacings were excluded as they
would have still provided too much rigidity, and possibly masked any non-elastic
deformation occurring.

— 4-9 —
Chapter 4 Experimental Method

Sprayed seal surfacings over unbound granular pavements are the major pavement
type in Australia and comprise some 90% of the length of all surfaced roads (Rebbechi
& Sharp 2009). It is known that fresh sprayed seals by their very nature are completely
flexible, although bitumen films may lose their flexibility through oxidative aging and
over time become brittle (Read & Whiteoak 2003).

Sprayed seals are commonly ‘single/single’ (single layer of bituminous binder followed
by a single layer of aggregate) however ‘double/double’ (two applications of bituminous
layers and two applications of aggregate layers) 1 are known to provide a robust and
durable surfacing (Rebbechi & Sharp 2009). Experienced practitioners believe that a
14/7 double/double (14 mm aggregate for the bottom layer, 7 mm aggregate for the top
layer) provides a more robust seal than a 10/5 (10 mm aggregate for the bottom layer,
5 mm aggregate for the top layer) double/double seal.

Austroads has recently introduced a concept of an extreme stress seal (XSS) for
pavements where a particularly robust seal is required. XSS seals are double/double
seals where a polymer modified binder is used for both layers of bituminous binder
application (Patrick 2013). XSS seals were not public knowledge at the time of this
experimental work in 2008, and the design was able to demonstrate a successful proof-
of-concept of such a seal prior to a revised design method being published.

The sprayed seal treatment selected was a 14 mm/7 mm, double/double, 3% latex
emulsion seal.

Whilst emulsion seals are not common in Australia, emulsion was the binder of choice
by the only sealing contractor who was willing to tender on the small 972 m2 sealing
job. The only polymer modified emulsion binder available in Melbourne was modified
with latex (a natural polymer) rather than the more common Styrene-Butadiene-Styrene
(SBS, a synthetic polymer) used in hot bituminous binders. The choice of latex over
SBS was not considered important given the loading conditions to be applied.

1
The terminology of ‘single/single’ and ‘double/double’ are Austroads terms and are solely used in
Australia.

— 4-10 —
Chapter 4 Experimental Method

4.5 Sprayed seal surfacing


4.5.1 Pre-design data collection
To gain an indication of the pavement surface hardness, ball embedment testing of the
pavement in accordance with Austroads test method AGPT-T251 (Austroads 2010)
was undertaken prior to brooming (Figure 4.7). This resulted in an average
embedment of 2 to 3 mm (sample size = 39). Experience has shown the ball
embedment test to be imprecise, and the results are used in the seal design as an
indicator without any further refining of accuracy. It is also noted that there is no
precision statement accompanying the test method.

Figure 4.7: Ball embedment testing to estimate pavement hardness prior to seal design

4.5.2 Prime trials


The function and acceptance of priming a basecourse prior to sealing has been
discussed in the literature review. As this study was considered high risk/high profile,
the use of a prime was mandated as part of reducing risk of seal failure and also as it
was regarded as simply good practice. The sealing contractor chose to use an
emulsion product rather than a hot cut-back bitumen, as emulsion primes in Australia
are more commonplace than emulsion seals.

There is no recognised theoretical method for designing prime application rates and in
practice either a nominal 1 L/m2 is used or experience with the local materials is called
upon (Alderson 2006). As this author had no previous experience with this prime
material in combination with this basecourse material, a small trial was undertaken
using a range of application rates. A one square metre area was marked on the
pavement and volumes of 0.8 L, 1.0 L and 1.2 L of prime were poured with a
measuring container and spread with a paint brush.

— 4-11 —
Chapter 4 Experimental Method

Figure 4.8: Setting out the prime trials

The performance indicators for a successful prime are purely visual and usually
experienced-based. A visual inspection is held 24 hours after laying. If the surface
shows excess bitumen which would likely adhere to the rubber tyres of the sealing
equipment, the application rate is deemed too high. If the pavement has adsorbed all
bitumen leaving no free bitumen at the surface, the application rate is deemed too low.

Using the above criteria, a prime application rate of 1.0 L/m² was nominated for this
pavement material in moistened, swept and as-constructed condition.

4.5.3 Seal design


The sprayed seal was designed by this author specifically for this study using the
Austroads design method AP-T68-06 (Alderson 2006) as the basis. Judgement was
required as the design traffic loadings exceed the scale on the published design charts.
The design incorporated as an initial starting point the seal design assumptions made
by Holtrop & Moffatt (2008) when designing the successful double/double sprayed seal
used in the previous 2007 ALF experiment.

The detail of the seal design is included as Appendix A.

In summary, the resulting seal design is shown in Table 4.1.

Table 4.1: Seal design summary

Application Material Application rate


SBR 3% latex emulsion binder 1.2 L/m2 residual
First application
14 mm aggregate 100 m2/m3
SBR 3% latex emulsion binder 0.5 L/m2 residual
Second application
7 mm aggregate 340 m2/m3

— 4-12 —
Chapter 4 Experimental Method

4.5.4 Prime construction


The basecourse was primed on 16 September 2008 by Primal Surfacings using
Emoprime™ at 1.0 L/m² of product at an air temperature ≈ 12°C.

The Emoprime was manufactured by Primal Surfacings at Laverton North in


Melbourne. Its composition as detailed by the product data sheet was typical of
emulsion primes and is shown in Table 4.2.

Table 4.2: Composition of Emoprime

Component Proportion (% v/v)


Bitumen ≤ 50
Solvent ≤ 20
Fatty amines & derivatives ≤3
Hydrochloric acid ≤1
Water ≥ 26
Adapted from Primal Surfacing (2008a).

The steps in the priming process are shown in Appendix A.

4.5.5 Sprayed seal construction


Sprayed sealing was undertaken on 18 September 2008.

The Emoflex™ 70 polymer modified emulsion (PME) was manufactured that morning.
Its composition is detailed by the product data sheet as shown in Table 4.3.

Table 4.3: Composition of the Emoflex 70™ binder


Component Proportion (% v/v)
Bitumen 68
Solvent <1
Latex 3
Fatty amines and derivatives <2
Hydrochloric acid <1
Water > 25
Adapted from Primal Surfacing (2008b).

For details of the sprayed seal construction see Appendix A.

This then marked the completion of the civil construction components of the study.

— 4-13 —
Chapter 4 Experimental Method

4.6 ALF experimental design


The work of Box et al. (1978) was cited by several of the references in the literature
review undertaken for this study. Box et al. introduced empirical models and
theoretical/mechanistic models. Empirical models are used when the underlying
mechanism of a process is too complicated or not understood, and where the response
needs to be approximated over a limited range of variables. By contrast, a mechanistic
model is justified when a basic understanding of the system exists, i.e. the state-of-the-
art is sufficiently advanced. If a mechanistic model is well-founded it can give a much
better representation of the response being measured.

The model developed through this experimental design is an empirical model.

4.6.1 Definition of the problem.


Austroads funding of approximately $1m/y was secured for project TT1219 to:

investigate – using a combination of current research, laboratory


characterisation and field trials – improved methods for assessing the
pavement damage caused by different multiple axle group loads, and
to develop a framework that can be used to quantify this pavement
damage for use in the Austroads flexible pavement design processes.
Recognising that a severe lack of material pavement performance
data was preventing the development of a defendable framework, the
project ambitiously sought to obtain performance data for common
flexible pavement materials (Moffatt 2011).

As discussed earlier, the opportunity was taken to leverage upon the TT1219 study to
pursue two additional research questions, which are now introduced.

Firstly, with the increasing freight task imposed on Australian roads there appears to be
an economic expectation that roads will be able to continue meeting this demand at
little extra cost. Unlike much of Europe and the USA, much of Australia’s vast road
network comprises unbound granular pavements made from materials available nearby
(that is, not necessarily chosen because they are ideal), and then surfaced with a thin
flexible sprayed seal. The Australian economy does not appear to be able to support
the construction of vast networks of dense mix asphalt and heavy duty pavements.

This creates an incentive to maximise the utilisation of existing materials, and in turn
maximise the heavy vehicle capacity of sprayed seals.

— 4-14 —
Chapter 4 Experimental Method

For seal design purposes the approach of converting heavy vehicles into 20 equivalent
light vehicles or passenger car equivalents, then later 40, then possibly 80 or 100 (refer
to Section 3.3) as the heavy vehicle fleet changes, is not deemed a defendable
response and does not address the cause of the problem.

Discussions have been held amongst practitioners regarding the possibility of using
equivalent standard axles (ESAs) rather than passenger car equivalents, as ESA
calculations for many pavements are already in existence or can easily be obtained.

However, investigations could not find any correlation between ESAs using the 4th
power damage exponent and seal performance (Neaylon et al. 2008).

It was also believed that pavements, being a structural layer, respond primarily to
vertical loads whereas sprayed seals, being a non-structural layer, are less affected by
vertical loads than by tyre pressures and horizontal loads. This belief can be inferred
from comments such as (Jameson 2012, p.107):

While the effect of this (tyre) stress variation has relatively little impact
on the performance of thick asphalt pavements, it can have a
pronounced effect on thin surfacings.

and

During certain manoeuvres, such as braking, turning and travelling


uphill, heavy vehicles apply horizontal loads to the pavement, which
are currently not taken into account in the (pavement) design model.
For thin surfaced granular pavements, the stresses generated by
these loads are concentrated in the upper pavement layers and can
have a significant impact on the performance of the surfacing. In thick
asphalt pavements, constructed in a continuous operation, the
interface bonding and the strength/modulus of the layer allows these
stresses to be dissipated throughout the pavement and, hence, to
have relatively little effect on (pavement) performance. (Jameson
2012).

4.6.2 Response to be measured


Decay of surface texture (macrotexture, as described in the literature review) was
chosen as the response to be measured.

From the literature review, it can be seen that deterioration in macrotexture is a


common measure to determine loss of service life in a seal, as macrotexture is
important in providing skid resistance at highway speeds. Macrotexture reduces as a
result of aggregate particle reorientation under cumulative traffic, resulting in a loss of
voids in the total aggregate mat.

— 4-15 —
Chapter 4 Experimental Method

This is succinctly summarised in Transit New Zealand et al. (2005) as:

Under traffic, the voids decrease in volume as the chip becomes re-
oriented and embedded in the substrate and leads to reduced texture.
On heavier trafficked roads this loss of texture, which ultimately leads
to flushing, is the most common failure mechanism.

Decay in surface texture (macrotexture) was also the response measured by Arnold et
al. (2005c) when they were undertaking a similar study.

The reasons for loss of texture will often vary from case to case, and the influencing
factors may include chip re-orientation; a too high binder application rate; chip
embedment into the basecourse; and binder rise because of internal moisture vapour
(Transit New Zealand et al. 2005). Recently it was suggested that breakdown of
aggregate into fine particles, when encapsulated into the bitumen, also caused the
binder volume to rise (Herrington et al. 2012).

Decay of microtexture was not measured, as this is a function of cumulative traffic vs.
the polishing resistance of the aggregate, which in turn is a function of the geological
origins of the source rock, and therefore is totally independent of any seal design
method used.

Other surfacing issues such as rutting, roughness and cracking are defects instigated
by the performance of the pavement material below the sprayed seal, and again are
independent of the seal design.

Rutting was being measured under the parent project, but as it is a function of the
pavement, it was not used in this sealing study. Roughness and cracking have no
history of occurring in an ALF trial. Too high a binder content was not expected as
significant attention to detail was given to the seal design.

This left aggregate reorientation and embedment into the basecourse as the expected
mechanisms for the decay in macrotexture. It is not considered that the precise cause
of texture decay is significant for this study, rather, whatever the cause or combination
of causes were, they remained the same for each experiment, and the relative rates of
decay for the different loading scenarios were the significant response to measure.

— 4-16 —
Chapter 4 Experimental Method

4.6.3 Selection of variables


The key variable required to answer the research question was multiple axle groupings
– being single axles, tandem axles and tri-axles. The secondary variable required was
axle loading.

Constrained variables
Although all experiments were conducted over a two-year period, the experiments were
undertaken adjacent to each other at one location under cover of an industrial shed
(Figure 4.1). This enabled the following potential variables (from experiment to
experiment) to be constrained as much as practicable.

Table 4.4: Constraints on experimental variables


Variable Constraint
Construction material Identical material used for each
experiment
Construction quality All experiments were constructed at the
one time by the one construction
contractor
Air temperature All experiments experienced the same
air temperatures at the same time
Pavement temperatures All experiments were kept shaded, and
all experienced the same pavement
temperatures at the same time.
Humidity All experiments experienced the same
humidity at the same time
Rainfall All experiments were shielded from
rainfall
Sunlight (UV light) All experiments were shielded from
sunlight
Tyre pressures All experiments were trafficked with tyre
pressures of 760 ±10 kPa
Loading velocity All experiments were trafficked at the
same speed (approx. 22 km/h)
No extraneous axle The exact number of axle grouping
groupings or loadings types and loadings are known for every
experiment

— 4-17 —
Chapter 4 Experimental Method

A weakness in this constraint of variables is that whilst all experiments were subject to
the same humidity and air and pavement temperatures, they were tested at different
stages of their accumulation of these regimes. It was regarded that the consequence
of this variation was minor, compared to the expense of implementing a climate control
system to fine tolerances within the shed2.

Controlled variables
Axle load and axle groupings were the variables that were chosen to be systematically
altered. As illustrated in Figure 4.4, single axles, tandem axles and tri-axles are very
common in Australia and hence these axle groupings were selected.

As discussed in Chapter 1, Australian heavy vehicle axle group types are identified
from Jameson (2012) as:

• single axle with single tyres (SAST)

• single axle with dual tyres (SADT)

• tandem axle with single tyres (TAST)

• tandem axle with dual tyres (TADT)

• triaxle with dual tyres (TRDT)

• quad-axle with dual tyres (QADT).

This axle configuration nomenclature is shown visually in Figure 4-9.

2
The University of California Pavement Research Centre at Davis use climate control (heaters, chillers,
temperature controlled housing) in their heavy vehicle simulator (HVS) shed. The FHWAs Turner Fairbank
Highway Research Centre near Washington DC, uses a climate control housing under their (outdoor)
ALFs. However, the budget for this Australian work did not extend to temperature control housing – other
than doing the experiments ‘indoors’, as is done in NZ (personal observation).

— 4-18 —
Chapter 4 Experimental Method

Figure 4-9: Axle configuration nomenclature

As noted earlier, ALF, like most automated pavement testers does not use complete
axles but rather half axles, i.e.

rather than

Therefore, the Standard Axle load of 80 kN on a SADT is replicated in ALF by a 40 kN


load on a half axle. Table 1.1: Loads on axle groups with dual tyres which cause the
same damage as standard axle from Chapter 1 can now be expanded as shown in
Table 4.5 to explain the ALF experimental loading regime selected.

Table 4.5: Derivation of experimental loading regime

Loads which cause the Half axle loads which ALF


same damage as the standard cause the same experimental
Axle group type
load damage loading (kN)
(kN) (kN)
Single axle with dual tyres (SADT) 80 40 40
Tandem axle with dual tyres 135 67.5 60, 80
(TADT)
Triaxle with dual tyres (TRDT) 181 90.5 90
Quad-axle with dual tyres (QADT) 221 110.5 nil

Rather than load the tandem axle with 67.5 kN (say 70 kN), there were two reasons for
loadings of 60 kN and 80 kN.

Firstly, this enabled a linkage for a common load per axle between groups, i.e 40 kN
per axle and 30 kN per axle, and secondly, two noticeably different loadings on one of
the axle groupings was essential to determine a load damage factor (LDE) as will be
discussed further in chapter 5.

— 4-19 —
Chapter 4 Experimental Method

4.6.4 Factorial design


Factorial experiments can be defined as ‘experiments where two or more independent
variables are manipulated systematically’ (Francis 2012). Advantages of factorial
designs are that information on the effects of different variables can be collected more
efficiently in a fewer number of experiments, and that any interactions between the
independent variables can be investigated. This is particularly important in this work,
as it is subsequently found that the variables (axle loadings and axle groupings) when
combined have an effect upon the response being measured (surface texture decay).

A good experimental design for an experimental model should allow contemplation and
examination of all the various conditions possible. Thus, an ideal experimental design
would have examined all axle loading combinations with all axle group combinations,
i.e. have been a full 3 × 4 factorial design (and repeated in triplicate, thus requiring 36
separate and expensive APT experiments) as shown in Figure 4.10.

Figure 4.10: Requirements for a full factorial experimental design

However, in practice two problems prevented this from being achieved. Firstly, with
ALF the loading forces are applied using Force = mass × gravity. A single axle
applying a load of 30 kN was not possible due to the mass reductions needed to the
axle assembly construction to create such a light axle. A triaxle grouping with a total
load of 120 kN was not possible because the mass required to achieve this would have
overloaded the mechanical structure and operational ability of ALF. Also, the budget
for data collection was limited to AU$2m over two years – a full factorial design would
have cost significantly more in equipment redesign and taken substantially longer
operating time to run.

— 4-20 —
Chapter 4 Experimental Method

Therefore, the practical and achievable experimental design was limited to that as
shown in Figure 4.11.

Figure 4.11: Achievable experimental design

The resulting four experimental configurations are shown diagrammatically in


Figure 4.12.

Single axle dual tyre Tandem axle dual tyre Tandem axle dual tyre Triaxle dual tyre
40 kN load 60 kN Load 80 kN Load 90 kN load

40 kN 60 kN 80 kN 90 kN

Figure 4.12: ALF trial loading regimes

4.6.5 Experimental layout


The pavement constructed and sealed within the shed was marked into 12 separate
experiments (Figure 4.1 and Figure 4.13), with space between the longitudinal run of
experiments available should an emergency occur and a replacement location needed
unexpectedly.

— 4-21 —
Chapter 4 Experimental Method

North

Figure 4.13: Original experimental layout within the shed

The degree of freedom afforded by provision of emergency locations proved useful,


and by the end of the two-year program the experimental locations were as shown in
Figure 4.14.

North

Figure 4.14: Final experimental layout

— 4-22 —
Chapter 4 Experimental Method

4.6.6 Testing regime


Bedding in
All ALF experiments underwent a preconditioning (bedding-in) of 10,000 passes at
16 km/h of a tandem axle loaded at 60 kN. 10,000 passes were considered to be
appropriate bedding-in based on years of experience in running previous ALF
experiments (Yeo et al. 2008). A tandem axle loading was chosen as it was
considered to be more representative of in-service conditions.

If the pavement surface temperature was below 20 °C it was heated to this temperature
by infrared radiators and this was maintained for 24 hours to ensure the seal binder
was sufficiently pliable to allow the aggregate in the seal to reorientate rather than have
the seal break up by brittle failure, during initial trafficking. It is considered that the
effect that this had on binder properties was negligible, and any effect, if at all, would
aid in overcoming the issue of accelerated loading without corresponding accelerated
aging.

The travelled path of the axle groupings was also programmed to mimic the transverse
wander that is found in real-life situations.

Once bedding-in was completed the appropriate axle assembly and axle load was
installed prior to the loading experiments commencing.

Loading
The pavement was then loaded with a number of cycles of the axle assembly
configurations as shown in Figure 4.12, until the testing program (shown in Table 4.6)
was completed. The ALF assembly was fitted with driving wheels and limited steering,
and could be repositioned between experimental locations.

— 4-23 —
Chapter 4 Experimental Method

Table 4.6: ALF trial loading regimes


Axle Load Experiment Total cycles Date commenced Date completed
(kN)
Single 40 3511 298,400 23 May 2010 7 July 2010
3507 342,500 20 Sep 2010 16 Nov 2010
3510 75,000 23 Jun 2011 17 July 2011
Abandoned due to
hydraulic oil spill
3500 365,680 3 Aug 2011 7 Oct 2011
Discard all data from site
Ch 0 – 4 m
Tandem 60 3508 278,232 7 Dec 2009 6 Feb 2010
3503 322,600 22 Nov 2010 10 Jan 2011
80 3505 291,000 15 Mar 2010 10 May 2010
3506 370,000 19 Feb 2011 12 Apr 2011
3512 22,000 7 Feb 2012 2 Mar 2012
Abandoned due to
hydraulic oil spill
3514 234,500 19 Mar 2012 9 Jul 2012
Tri 90 3502 210,000 15 Oct 2009 19 Nov 2009
3504 310,413 13 July 2010 13 Sep 2010
3501 390,000 18 Apr 2011 14 Jun 2011
Nil Nil 3509 Rejected Test pavement failed to
meet screening criteria of
the falling weight
deflectometer
3513 Not proposed End of funding

The decision on when to cease trafficking was based on a combination of seal and
pavement condition, pavement deformation achieved, budget constraints and the
remaining time available before funding ceased.

4.6.7 Data collection


After bedding-in the ALF was then operated at full speed (approx. 22 km/h) with
programmed transverse movements every 50 cycles to mimic traffic wander. In
addition to logging the number of cycles, at each transverse move the following data
was logged: air-spring pressures; hydraulic oil pressure and temperature; ambient air
temperature; pavement temperature at depth; and seal surface temperature. Also
during the operation of each experiment, the loading cycles were regularly paused and
manual sand patch texture measurements and laser texture cross-sections were taken
at specified chainages along each of the 12 m experimental sections.

— 4-24 —
Chapter 4 Experimental Method

Laser profiling
The laser used was a Baumer OADM 13│6480/S35A with a 650 nm wavelength laser
generated by a pulsed red laser diode and a point beam of 1 mm diameter.

Prior to each experiment the laser logger was calibrated for: longitudinal conversion
(counts to mm); trigger intervals in the transverse and longitudinal directions; and start
and end points for the longitudinal passes and three transverse offsets (Martin &
Howard 2010).

The laser logger measured three longitudinal passes along the length of each
experiment: two within the trafficked area and usually 10 to 30 mm apart; and one as a
reference offset outside the trafficked area. The logger collected data at approximately
every 0.6 mm in the longitudinal direction, equating to around 17,000 data points for
each of the three passes. It was attached to a four wheeled device and used a 12-volt
DC system to drive itself along a pair of longitudinal tracks, as shown in Figure 4.15.

Figure 4.15: ALF profilometer and macrotexture device

Source: Moffatt et al. (2011).

The raw data extracted from the loggers was then further calibrated by Martin and
Howard using: a transverse rotopulser (rotary transducer) calibration factor; a
longitudinal rotopulser calibration factor; a laser height factor; and slope suppression.
The high resolution data was then averaged over a 100 mm baseline before the mean
profile depth (MPD) was calculated using the International Standards Organisation
(1997) method (equation 4.1).

— 4-25 —
Chapter 4 Experimental Method

(𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 1𝑠𝑠𝑠𝑠 ℎ𝑎𝑎𝑎𝑎𝑎𝑎) + (𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 2𝑛𝑛𝑛𝑛 ℎ𝑎𝑎𝑎𝑎𝑎𝑎)


𝑀𝑀𝑀𝑀𝑀𝑀 (𝑚𝑚𝑚𝑚) = − (𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙) 4.1
2

where the inputs are defined as in Figure 4.16.

Figure 4.16: Illustration of the ISO terms baseline, profile depth and mean profile depth

Adapted from ISO (2002a).

The mean profile depth (MPD) was then converted to estimated texture depth (ETD) to
enable direct comparison to the volumetric sand patch test using the ISO (1997)
transformation equation

𝐸𝐸𝐸𝐸𝐸𝐸 = 0.2 𝑚𝑚𝑚𝑚 + 0.8 × 𝑀𝑀𝑀𝑀𝐷𝐷 4.2

ISO (1997) attributes the derivation of this transformation equation to Wambold et al.
(1995).

The laser texture data resulted in an ETD that, apart from the first few hundred cycles,
remained relatively constant with increasing loading cycles. This was contrary to visual
inspections and to the sand patch testing results.

This indicates an error in the calibration, collection, processing and conversion of laser
data. This had important implications which are discussed later.

— 4-26 —
Chapter 4 Experimental Method

The transform equation was developed by Wambold et al. (1995) during data analysis
for the PIARC International experiment to compare and harmonize skid resistance and
texture measurements. Data was obtained from 16 different texture measurement
devices, and every measure (Y) was correlated to all the others (X) by a linear
regression Y = AX + B. In their conclusions the authors mention that the signal
processing method is critical, and also that different bandwidths have different
sensitivities to profile shape and profile wavelength.

The Austroads test method AG:AM/T013 Pavement surface texture measurement with
a laser profilometer (Wix & Moffatt 2007) advises that, as an alternative to the ISO
(1997) transformation equation, MPD can be converted to an equivalent volumetric
sand patch texture depth. This must be done by first performing a correlation exercise
under controlled conditions to develop a specific relationship.

The resources available for this study were not sufficient to cover a correlation exercise
under controlled conditions to develop a specific relationship, and the decision was
made not to pursue with this method.

Microphotogrammetry
Initially during the loading cycle’s regular pauses, pairs of stereo photographs were
also taken in an effort to quantify the progression of aggregate reorientation over time.

Commercial 3D photographic software based on dense digital surface modeling (Eos


Systems Inc 2009) was obtained at relatively low cost, and a prototype measurement
apparatus designed, built, calibrated and used to collect photographic data.

The apparatus consists of two entry-level digital single lens reflex cameras connected
to a single flash light source, enclosed in thin aluminum cladding to exclude any
ambient light. Two cameras are mounted facing down and side-by-side (Figure 4.18)
and arranged such that their field of view overlapped by about 200 mm (Figure 4.17).
The overall dimensions of the metal frame (Figure 4.19) arose from calculations based
on the length of overlap required to give an adequate sample size and the fixed focal
length of the lenses used.

— 4-27 —
Chapter 4 Experimental Method

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 4.17: Camera setup requirements

Source: Eos Systems Incorporated (2009).

Figure 4.18: Camera setup

Source: Neaylon (2009).

Figure 4.19: Prototype used to take site photographs

Source: Neaylon (2009).

— 4-28 —
Chapter 4 Experimental Method

The development of the prototype is described in Neaylon (2009). The stereoscopic


images (Figure 4.20) were then analysed using PhotoModler Scanner ™ (Eos Systems
Inc. 2009).

During the development of this analytical method serious difficulties were encountered
in obtaining an accurate micro-contour plan of the sprayed seal surface. There
appeared to be issues with using a stereo photographic pair from two different cameras
whereas the software could only retain one camera calibration file.

Collaboration with a colleague at the University of Ulster pointed to a change in data


gathering using a single camera in two locations, and different software. This work
(Millar et al. 2011) was undertaken within the School of Surveying, and used Image
Master™ software (Topcon 2007) which was much more appropriate for micro-
photogrammetry and the production of topographic contour maps. It appears the
software used in this study up until that point was more appropriate for forensic crash
scene recording and architectural imaging.

This collaboration was fruitful and resulted in a joint publication (Millar et al. 2012) but it
was too late to rephotograph past experiments in two images with a single camera for
use with the more successful software, and the micro-photogrammetric method was
not pursued any further in this work.

— 4-29 —
Chapter 4 Experimental Method

Left hand image

Right hand image

Figure 4.20: Photographic stereoscopic pairs

— 4-30 —
Chapter 4 Experimental Method

It is worth noting that the ALF experiments were funded for the primary study and were
continuing independently of this work, and therefore preparation time was severely
constrained. Close range digital photogrammetry has subsequently been pursued with
better success in later non-related experiments.

3D photography has potential as it holds the possibility of being able to measure the
specific reorientation of individual aggregate particles, rather than simply taking a
blanket approach with averaged surface texture over any given area. Once aggregate
reorientation can be measured then engineering measures could be devised to hinder
this orientation – for example the development of binders that reach a higher viscosity
over a shorter length of time – and their performances measured against a
reproducible failure criterion. If aggregate reorientation can be halted at a chosen point
in time, then the onset of flushing and bleeding should be slowed. This should then
extend the life of the seal, and save the road agency a significant sum in its periodic
maintenance spend.

This was another possibility considered early for improving seal designs for heavy
traffic, but has become a topic for further research at another time.

Sand patch data


After every nominated number of passes, the ALF was stopped and sand patch tests
were taken at chainages 11, 13, 15, and 17 m along the test section at the centreline of
the wheel path. At each of these chainages, the sand circle diameter from 24.6 cm3 of
graded single sized sand was measured as per Figure 4.21, the diameters averaged,
and the total diameter measurements of all four chainages averaged and then input in
the conversion equation in AG:PT/T250 Modified surface texture depth (Pestle method)
(Austroads 2008). Thus the average surface texture depth (STD) was reported as sand
patch texture depth (SPTD) in mm and calculated using an average from four
measurements at each of four chainages over the trial length, after each nominated
number of passes.

— 4-31 —
Chapter 4 Experimental Method

Figure 4.21: Measurements of the sand circles

Inaccuracies with the sand patch method were discussed in Section 2.7.1

4.7 Limitations
4.7.1 Surface quality of the constructed basecourse
The surface quality of the constructed basecourse was disappointingly marginal. The
pavement material at some stage had been tyned and reworked, and the tyne marks
were still very visible at the time of sealing (Figure A.1). The author considers that the
standard of finish was much lower than what would have been acceptable had this
been part of a larger construction project for a rural highway. However, as this
construction project was so small relative to usual construction contracts, the pool of
experienced road builders and their associated equipment was not available (nor
interested). Time and budget constraints did not permit any further reworking of the
surface.

The tyne marks in the pavement surface were considered acceptable for the pavement
part of the project as this focussed on pavement deflections over time, but they would
have added unnecessary variation to surface texture (as shown in Figure 5.19) and
sand circle data collected for the surfacings component of the project.

— 4-32 —
Chapter 4 Experimental Method

4.7.2 Construction variability


As with all unbound granular pavement constructions, there are local variabilities in
materials and placement. Figure 4.22 shows that after seal construction the initial sand
patch texture readings were not the same in different experiment locations, as ideally
they should have been.

Figure 4.22: Initial volumetric sand patch texture depth before bedding-in

It is suggested that the random locations of tyne markings of the pavement surface are
the primary cause of this variation. This issue will be investigated further in the
following chapter.

A review of the FWD results (see appendix G of Moffatt 2013) suggested that the
pavements for the trials along the western wall (experiments 3509, 3510, 3511) were
softer than the others. Fortunately, out of these three pavement 3509 was rejected;
pavement 3510 was abandoned due to a hydraulic spill, leaving pavement 3511 used
for a single axle 40 kN load experiment. Experiment 3511 was found later to have the
most rapid surface texture decay of the triplicate single 40 kN experiments, and added
to the variability in the single 40 kN data.

4.7.3 Surface texture measurement


Measurement of surface texture depth started with double redundancy, being 1) laser
(quick, automated, accurate); 2) photogrammetry (slow, manual, potential for
accuracy); 3) volumetric sand patch test (slow, manual, scientifically inaccurate), with
the third being intended as a reality check for the first two.

— 4-33 —
Chapter 4 Experimental Method

As already discussed, the laser surface texture data was flawed. There were serious
resolution deficiencies in the micro-contour mapping obtained from the
photogrammetry software package employed, and the testing program could not be
stopped to enable remedial action to be taken.

The surviving method of collecting surface texture data was therefore volumetric sand
patch, which was originally intended to be used as a reality check of the more
sophisticated processes. In order to manage the risk associated with this, a review of
the sand patch data and the operators who collected it was undertaken and is reported
in the data analysis chapter.

4.7.4 Binder aging


Originally, the fact that loading was accelerated but binder aging was not was
considered to be a significant factor. However, the loading cycles of each experiment
have been compared to traffic in one direction on the Hume Freeway (the main road
freight route between Sydney and Melbourne) near Wangaratta (Figure 4.23), which
has one of the highest freight tasks in Australia. The northbound carriageway and the
southbound carriageway in 2013 each carried 1,800 commercial vehicles per day
(VicRoads 2014).

Hume Freeway
near Wangaratta

Figure 4.23: Assigned interregional road freight 2005

Map source: BITRE (2011).

— 4-34 —
Chapter 4 Experimental Method

Given that the typical Victorian commercial vehicle breakdown is as shown in


Figure 4.24,

45% 30% 25%

Figure 4.24: Victorian Commercial vehicle breakdown

Adapted from Jameson (2012); Khoo (2014).

and assuming that 98% of the commercial vehicles are in the left hand lane (2% are in
the right hand lane passing) and that all commercial vehicles are fully loaded (not
entirely true), then in terms of pavement design and standard axle repetitions (SARs,
as discussed in Section 2.10.4) this freeway traffic can be approximated as

𝑆𝑆𝑆𝑆𝑆𝑆 ≅ 0.98 × 1,800 × �(0.45 × 2) + (0.30 × 3) + (0.25 × 4)� = 4,939 ≈ 5,000

Thus the 250,000 cycles of single axle 40 kN (1.0 SAR) equates to about 7 weeks; the
300,000 cycles of tandem axle 60 kN (≈ 0.9 SAR) equates to about 8 weeks; the
380,000 cycles of tandem axle 80 kN (≈ 1.2 SAR) equates to about 13 weeks; and the
400,000 cycles of triaxle 90 kN (1.0 SAR) equates to about 20 weeks. This is
compared to the actual experimental time in Table 4.7

Table 4.7: Comparison of loading times


Experiment Laboratory time Hume Fwy time
(Wks) (Wks)
Single axle 40 kN 8 7
Tandem axle 60 kN 9 8
Tandem axle 80 kN 8 13
Triaxle 90 kN 8 11

It could be argued that the loading, although carefully controlled, was not accelerated
significantly and the lack of associated binder aging can be ignored.

One of the key advantages of accelerated loading facilities not yet mentioned is that
the trial pavement can be deliberately under-designed to accelerate failures. The
pavement for these experiments, however, was a full-depth structure and not designed
for premature failure.

— 4-35 —
Chapter 4 Experimental Method

4.7.5 Velocity of ALF


The longitudinal velocity of ALF is around 20 km/h, whilst truck traffic on the major
freeways travel at around 100 km/h.

Bitumen and polymer modified bituminous binders are visco-elastic materials, and their
properties vary with rate of loading and the temperature at which the rate of loading
applies. At low temperatures or short loading times the response tends to be elastic.
At high temperatures or long loading times the response tends to be viscous (Read &
Whiteoak 2003). This could possibly be an important limitation, as at lower loading
rates bituminous binders tend to be more viscous (i.e. plastic and flow) and at the
higher loading rates they tend to be more elastic (i.e. bounce back).

This suggests that at the slower loading rates (ALF travelling longitudinally slower than
highway traffic) the accompanying more plastic response would mean the aggregate
could reorientate in shorter time - which would assist the ‘accelerated’ part in the
accelerated loading facility experiment.

4.7.6 Driven axles


Anecdotal reports suggest that heavy vehicle damage to sprayed seals may be caused
by the driving axles of the prime moving unit, especially during uphill grades when
maximum torque must be applied to the driving axles to lift the combined mass of the
prime-mover and loaded trailers up an incline.

For this experimental method, whilst one axle group was always a ‘driven’ group, it was
driven by one 11 kW (15 hp) electric geared motor per tyre. This provided energy to
overcome friction losses in the system, but much less torque (generating horizontal
shear forces) than that required by commercial trucks and prime movers when
transporting payloads, for example the driven axles hauling a freight task of cement up
a long steep hill in Figure 4.25. For this reason, the shear effect of the driven wheel
has been ignored.

— 4-36 —
Chapter 4 Experimental Method

Figure 4.25: High shear stresses can be imparted by a prime mover’s driven axles

So although driven, none of the axle groupings in this study are regarded as being
analogous to a prime-mover’s driven axle/s. This would involve the use of different
experimental equipment - which has been developed in Australia and is described in
Patrick and Wright (2014).

For this experiment the axles, for practical reasons, are regarded as being trailing axles
and not driven axles.

4.7.7 Straight line loading


In addition to the absence of any significant acceleration or deceleration shear stresses
being applied to the seal, as this experimental layout was straight- line only, there were
also no cornering shear stresses applied.

It should be noted also that in this study lateral axle-wander was programmed into the
experimental condition to mimic normal highway conditions. Some practitioners
suggest that axle-wander in the trailing axles of a road train is in fact minimal, causing
accelerated bleeding of the seal, whilst other practitioners suggest that axle-wander in
the trailing axles are greater, with the highest wander occurring at the end trailer.
Neither of these assertion have been verified. Possible variables to be considered
would be driver habits and the degree of rutting, and what level of pavement rutting
causes channelisation of trailing axles (if at all).

— 4-37 —
Chapter 4 Experimental Method

4.8 Summary
There was a unique opportunity to be seized to collect surface texture decay data
under controlled multiple axle groupings and loadings, in conjunction with another
(parent) study, funded and undertaken for other purposes.

However, all did not go to plan.

The pavement surface finish was rough and introduced unwanted surface texture
variation which was not an issue to the parent project. The next chapter will explain
how the data was normalised to mitigate the problem and enable meaningful
comparison.

Planned spare redundancies in the surface texture measurement failed. This left only
the least accurate measurement method, the sand circle or sand patch, although this
method is the industry standard. The next chapter will describe how this was managed
by combining duplicates, and in one case triplicates, of individual experiments. Should
this work be repeated by others, the purchase of an off-the-shelf texture measuring
device such as the circular texture meter at the commencement of the experiments
would be preferable to using a bespoke (in retrospect called ‘experimental’) laser
device and being unable to transform irretrievable experimental data part way through
an experiment that could not be halted.

Once the accelerated loading experiments started, it was not possible to postpone the
parent project to correct data collection issues that arose from time to time, such as the
laser texture and micro-photogrammetry issues. This was the one drawback in
collecting data from an experiment funded for a different purpose and with a different
aim, and would be the single most important issue to address should this work be
repeated by others.

The resulting data set, although flawed, is believed to be still useful and to be the only
data set in existence for multiple axle groupings on thin sprayed seal flexible
pavements.

It is believed useful because, although precision (refinement in the measurement) has


been affected, the accuracy of the data (conformity with the truth) is believed to be still
very plausible as shown later in the successful validation - using independent data - of
the resulting model derived from this experiment.

— 4-38 —
Chapter 4 Experimental Method

The next chapter will describe the analysis stages of this data, from the assessment of
the raw data and a discussion on outliers, through to nonlinear regression modelling of
the data and finally through to development of a generalised model which is able to
predict the outcome of a data set independent of the data set on which it was based.

The model is used to generate an outcome for the hypothesis – do axles loaded to the
pavement designers’ ESA and which cause equal damage to a pavement, cause equal
decay to a sprayed seal surface texture?

Finally, from the data set a load damage exponent for a sprayed seal is derived,
providing an outcome for the question – does the fourth power law apply to sprayed
seals?

— 4-39 —
Chapter 5 Data Analysis

5 DATA ANALYSIS
5.1 Introduction
The several limitations mentioned in the previous chapter, in particular the variable
surface roughness built into the pavement prior to sealing, introduced a significant
amount of noise to the data. Extracting statistically significant intelligence from
amongst the noise was one of the greatest challenges in this study, and this chapter
describes how this was undertaken.

The chapter begins by examination of the raw data from the volumetric sand circle test,
with outliers, operator error and construction variability being discussed. In order to
overcome the variability caused by the built-in surface roughness, the data is then
normalised so that rates of texture decay from the initial starting points can be
extracted and compared. Multivariate nonlinear regression modelling is then explored
to obtain a mathematical description of these rates of decay. An individual regression
model is then established for each case of axle groupings and axle loadings. The
ultimate goal of this chapter, however, is the determination of a general model: a single
model that can describe surface texture decay for any axle grouping and any axle
loading. This general model is then validated using an independent data set obtained
from a similar experiment undertaken in another country.

The chapter concludes by calculating a load damage exponent that is applicable to


surface texture reduction of sprayed seals on unbound granular pavements.

For the work described herein, the raw data was first prepared using Microsoft™
Excel™ (Microsoft Corporation 2010), then imported to IBM™ SPSS™ Statistics (IBM
Corporation 2012) for the multivariate nonlinear regressions analyses and exported
back into Excel for presentation.

— 5-1 —
Chapter 5 Data Analysis

5.2 Sand circle data


Volumetric sand circle tests, also referred to as sand patch tests, were taken at
chainages 11, 13, 15, and 17 m along all test sections at the centreline of the wheel
path. At each of these chainages, the sand circle diameter from a volume of 24.6 cm3
of graded single sized sand was measured, the diameters averaged and the total
diameter measurements of all four chainages averaged to and input in the conversion
equation in AG:PT/T250 Modified surface texture depth (Pestle method) (Austroads
2008). Thus the average surface texture depth, termed sand patch texture depth
(SPTD) in mm was calculated using an average from four measurements at each of
four chainages over each trial length, after each nominated number of assembly
passes.

5.3 Raw data


All raw data was plotted as a scatterplot and checked visually for consistency and
pattern as shown in Section 5.3.2, but before examining these scatter plots a
discussion on outliers is appropriate.

5.3.1 Outliers
Outliers can be defined in various ways.

Ratkowsky’s (1990) simplified definition is ‘points that are, in some sense, atypical of
the remaining points’. Mendenhall and Sincich (1996) however, suggest that outliers
can be defined as where regression residuals lie more than three standard deviations
from the mean. Field (2009) prefers 3.29 standard deviations from the mean, and
suggests using z scores as one way of expressing outliers in terms of mean and
standard deviation. Mendenhall and Sinich also suggests that although some analysts
support the elimination of outliers, others would not do so unless they could trace the
outlier to a specific cause. Thoresen (2012) suggested care when removing outliers as
they may be part of the relationship, or even a variable not yet considered.

It also appears that determining outliers for a nonlinear regression is more complicated
than determining outliers in linear relationships.

Motulsky and Christopoulos (2004) comment that smoothing the data could violate the
assumptions of nonlinear regression as the residuals are no longer independent.

— 5-2 —
Chapter 5 Data Analysis

Motulsky and Brown (2006) have determined outliers for nonlinear regressions by
setting probabilities for false discovery rates (FDRs) after assuming the data follows a
Lorentzian distribution.

Tabachnick and Fidell (2007) advised that the influence of possible outliers in terms of
a multivariate analysis can be checked during each regression run by calculating the
Mahalanobis distance of each independent variable data point.

In consideration of this discussion, it was decided to let all but one outlier remain in the
data set (refer to the point thickly circled in red in Figure 5.1). Later, once a suitable
model function had been selected, removal of this point was then checked for statistical
correctness. It was intended to check the Mahalanobis distance of suspected outlier
data points using SPSS however, although SPSS does calculate this for linear
regressions, a way was not found to calculate this for nonlinear regressions.
Therefore, the residuals (difference in SPTD between the recorded experimental point
and the nonlinear-model derived predicted point) were examined, and it was confirmed
that the residual of the point in question was 3.28 standard deviations from the mean
residual, this point was statistically concluded to be an outlier.

5.3.2 Raw data scatter plots

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.1: Experiment 3500 single axle 40 kN raw data

For the experiment 3500 data (Figure 5.1), the data point shown in the red ring is the
only outlier that was removed from any of the data sets.

— 5-3 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passess (incl. bedding-in)

Figure 5.2: Experiment 3501 triaxle 90 kN raw data

The experiment 3501 data (Figure 5.2) was acceptable.

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.3: Experiment 3502 triaxle 90 kN raw data

The experiment 3502 data (Figure 5.3) looks insufficient and unrealistic, and the
decision was made to exclude the whole experiment from analysis. The reasons for
this are explained in detail in Section 5.3.3. A photographic record of the surface
texture reduction during this experiment can be found in Appendix B. The
photographic record supports a quite different and more realistic outcome to the one
provided by this data.

— 5-4 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passess (incl. bedding-in)

Figure 5.4: Experiment 3503 tandem axle 60 kN raw data

The experiment 3503 data (Figure 5.4) was acceptable.

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passess (incl. bedding-in)

Figure 5.5: Experiment 3504 triaxle 90 kN raw data

There is a possible outlier in Experiment 3504 (Figure 5.5) data, however, given the
discussion earlier regarding outliers in non-linear regressions, it was retained.

— 5-5 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.6: Experiment 3505 tandem axle 80 kN raw data

The experiment 3505 data (Figure 5.6) was acceptable.

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.7: Experiment 3506 tandem axle 80 kN raw data

The experiment 3506 data (Figure 5.7) was acceptable.

— 5-6 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.8: Experiment 3507 single axle 40 kN raw data

The experiment 3507 data (Figure 5.8) was acceptable.

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.9: Experiment 3508 tandem axle 60 kN raw data

The experiment 3508 data (Figure 5.9) was acceptable.

Experiment 3509 did not proceed after it failed to meet the screening criteria of the
falling weight deflectometer.

— 5-7 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.10: Experiment 3510 single axle 40 kN raw data

The experiment 3510 (Figure 5.10) was discarded due to hydraulic oil leaks followed by
an hydraulic oil spill. The oil spill had the effect of softening the bitumen in the seal,
allowing artificially excessive aggregate reorientation and subsequent rapid decay of
surface texture. It most likely also softened the basecourse material, enabling
accelerated aggregate embedment into the basecourse.

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.11: Experiment 3511 single axle 40 kN raw data

The experiment 3511 data (Figure 5.11) was acceptable.

— 5-8 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.12: Experiment 3512 tandem axle 80 kN raw data

The experiment 3512 data (Figure 5.12) looks odd. This experiment was then
(ironically) abandoned due to another hydraulic oil spill.

2.5

2.0

1.5
SPTD (mm)

1.0

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes (incl. bedding-in)

Figure 5.13: Experiment 3514 tandem axle 80 kN raw data

The experiment 3514 data (Figure 5.13) looks unrealistic and unreliable, and the
decision was made to exclude the whole experiment from analysis. This is discussed
further in section 5.3.3.

— 5-9 —
Chapter 5 Data Analysis

5.3.3 Correlations in operator error


A perusal of the raw data scatter plots above led to a review of individual involvement
of the seven sand patch texture technicians (coded A, B, C, D, E, F, G) used over the
two-year period (Figure 5.1). It became apparent that there was a strong correlation
between any data that E and G collected, and that data being unrealistic or not
believable. Fortunately, having had experiments run in triplicate or quadruplicate
mitigated in some way the effects of excluding all data collected by E & G.

Table 5.1: Technicians for sand patch texture measurement


Axle Load Experiment Total cycles Texture technicians Comment
(kN)
Single 40 3511 298,400 B,D,F Data used
3507 342,500 A,B,C,D,F Data used
3510 75,000 No record Abandoned due to
hydraulic oil spill
3500 365,680 B,D,F Data used
Tandem 60 3508 278,232 B Data used
3503 322,600 B,D,F Data used
80 3505 291,000 B,D,F Data used
3506 370,000 B,D,F Data used
3512 22,000 D,E,G Trial excluded. Data is
unrealistic. Was there an
oil leak all along prior to
the final system burst?
Exp. abandoned due to
hydraulic oil spill
3514 234,500 D,E,G Sand patch data
unrealistic

Tri 90 3502 210,000 E,G Sand patch data


insufficient and serious
operator error
3504 310,413 B,F Data used
3501 390,000 B,D,F Data used
Nil Nil 3509 Nil Test pavement failed to
meet screening criteria of
the falling weight
deflectometer
3513 Nil End of funding

— 5-10 —
Chapter 5 Data Analysis

5.3.4 Data grouped in axle types with bedding-in removed


Figure 5.14 to Figure 5.17 show the data with bedding in removed and then the data
regrouped for comparison into similar axle types and axle loadings.

2.5

2.0

1.5
SPTD (mm)

Exp3500_Sing40
1.0 Exp3507_Sing40
Exp3511_Sing40
0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes

Figure 5.14: Combined single axles 40 kN

The outlier discussed earlier is shown within the red ring. This outlier has been
removed from all subsequent analysis.

2.5

2.0

1.5
SPTD (mm)

Exp3508_Tand60
1.0
Exp3503_Tand60

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes

Figure 5.15: Combined tandem axles 60 kN

— 5-11 —
Chapter 5 Data Analysis

2.5

2.0

1.5
SPTD (mm)
Exp3505_Tand80
1.0
Exp3506_Tand80

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes

Figure 5.16: Combined tandem axles 80 kN

2.5

2.0

1.5
SPTD (mm)

Exp3501_Trix90
1.0
Exp3504_Trix90

0.5

0.0
0 100,000 200,000 300,000 400,000
Assembly passes

Figure 5.17: Combined triaxles 90 kN

The combined data is shown in Figure 5.18

— 5-12 —
Chapter 5 Data Analysis

2.0
1.8
1.6
1.4
1.2
SPTD (mm)

Single axle 40 kN
1.0
Tandem axle 60 kN
0.8
Tandem axle 80 kN
0.6
0.4 Triaxle 90 kN
0.2
0.0
0 100,000 200,000 300,000 400,000
Assembly passes

Figure 5.18: All axles all loads

The raw data in Figure 5.18 shows a consistent trend of reduction in sand patch texture
depth with accumulation of loaded axle assembly passes. However, further analysis is
clearly required to extract any intelligence from this ‘chaos’ of data.

5.4 Check of construction variability


It is immediately apparent from Figure 5.18 that after bedding-in there was not a
common starting point for the initial surface texture (in sand patch texture depth, mm).

Did this variability occur during the bedding-in process (i.e. a variation in the seal
response to standard loading), or did this variation occur during construction (before
the seal was first loaded)?

Photographs taken at the time of construction suggest there was definitely a potential
for variation to be built into the project (Figure 5.19) even before sealing.

— 5-13 —
Chapter 5 Data Analysis

Figure 5.19: Typical variation in basecourse texture before sealing

The dimensions of the phone in Figure 5.19 are approximately 60 mm by 120 mm.

The question that remained was whether any further variation was introduced during
the bedding-in process.

The sand patch texture depths taken during the bedding-in process were examined in
further detail. In order to smooth out the large variations (variations that were to be
expected – which is why the bedding-in process was done first), a rolling average of
two data points was overlayed on the raw data. This is shown in Figure 5.20 to
Figure 5.24.

— 5-14 —
Chapter 5 Data Analysis

2.50
Exp3500_Sing40
2.00
Exp3507_Sing40

SPTD (mm)
1.50

1.00 Exp3511_Sing40

0.50
2 per. Mov. Avg.
(Exp3500_Sing40)
0.00
0 2000 4000 6000 8000 10000 2 per. Mov. Avg.
(Exp3507_Sing40)
Bedding-in assembly passes

Figure 5.20: Single axle 40 kN bedding-in only

2.50

2.00 3503_tand60
SPTD (mm)

1.50
Exp3508_tand60
1.00
2 per. Mov. Avg.
0.50 (3503_tand60)

0.00 2 per. Mov. Avg.


0 2000 4000 6000 8000 10000 (Exp3508_tand60)
Bedding-in assembly passes

Figure 5.21: Tandem axle 60 kN bedding-in only

2.50

2.00 Exp3505_Tand80
SPTD (mm)

1.50
exp3506_Tand80
1.00
2 per. Mov. Avg.
0.50 (Exp3505_Tand80)
2 per. Mov. Avg.
0.00
(exp3506_Tand80)
0 2000 4000 6000 8000 10000
Bedding-in assembly passes

Figure 5.22: Tandem axle 80 kN bedding-in only

— 5-15 —
Chapter 5 Data Analysis

2.50

2.00
Exp3501_Trix90

SPTD (mm)
1.50
Exp3504_Trix90
1.00
2 per. Mov. Avg.
0.50 (Exp3501_Trix90)
2 per. Mov. Avg.
0.00 (Exp3504_Trix90)
0 2000 4000 6000 8000 10000
Bedding-in assembly passes

Figure 5.23: Triaxle 90 kN bedding-in only

Figure 5.24: All axles all loads bedding-in

These data suggest that in general the seal’s response to the standard bedding-in
loading trended with similar slopes, and the observed variation was in fact the variation
in existence prior to loading. This is logical, as the bedding-in process was identical for
all experiments, the sprayed seal was the same, and therefore the rate of attenuation
of the sprayed seal surface texture should have been the same. What seems to have
occurred is that this rate commenced from different starting conditions. This conclusion
was also supported by visual observations throughout the experiments.

— 5-16 —
Chapter 5 Data Analysis

As discussed in Chapter 3, it is strongly suspected that the tyne marks in the finished
pavement surface prior to sealing had a major influence on the initial surface texture.

However, due to the nature of this research the rate of decay of surface texture due to
the axle groupings and axle loadings can be considered of more importance than the
initial surface texture starting point.

A view of sand patch texture depths (excluding any measurements taken by E and G)
prior to any assembly passes being applied is shown in Table 5.2.

Table 5.2: Initial SPTD readings


Experiment Initial SPTD reading
(mm)
3500 2.36
3501 1.87
3503 2.23
3504 2.09
3505 2.10
3506 1.95
3507 2.27
3508 2.29
3510 1.87
3511 2.29
3512 2.27
3514 1.91
Mean 2.12
Standard deviation 0.18

The largest variations in the data occurred in experiment 3500 (+11% of the mean) and
experiments 3510 and 3501 (-12 % of the mean). The 95% confidence range of the
data is the mean ± 1.64*std dev, which calculates to ± 13% of the mean.

As discussed in Chapter 3, Patrick et al. (2000) compared some of the various


methods that are used to measure macrotexture. With regard to volumetric sand patch
they found the reproducibility of the test in terms of texture depth in a first trial to be
24% of the mean and in a second trial 41% of the mean. Therefore, the data in this
study is considered reasonable and well within acceptable levels of accuracy.

— 5-17 —
Chapter 5 Data Analysis

5.5 Normalising the data


It is far easier to compare the various decay rates of surface texture if the data is
normalised, such that all starting points are the same.

Normalising data involves dividing the y values by a constant, to make all the y values
range between 0% and 100%. Doing this will not affect the results of nonlinear
regression. Normalising the data makes it easier to see what is happening in the
experiment, and to compare one experiment with another (Motulsky & Christopoulos
2004).

In normalising the data, the constant chosen in this case is the initial starting point, and
all subsequent readings become a percentage of that initial point.

However, due to the variability in the data, using the first measurement as recorded
may not give an accurate representation of the starting point as can be seen in the
following figures, particularly Figure 5.26 to Figure 5.28. To overcome this some very
simple Excel linear, exponential and power curve fits were plotted for all cases in an
attempt to find a common, repeatable and objective estimation method for a starting
point that is better than a subjective fit by eye.

2.0

1.8

1.6

1.4

1.2
SPTD (mm)

Single 40 kN
1.0
Linear (Single 40 kN)
0.8 Expon. (Single 40 kN)

0.6 Power (Single 40 kN)

0.4 y = -2E-06x + 1.8148


R² = 0.6824
0.2 y = 1.8214e-1E-06x
R² = 0.6815
0.0
0 100,000 200,000 300,000 400,000 y = 4.1673x-0.085
R² = 0.6266
Assembly passes

Figure 5.25: Approximating initial SPTD for single axle 40 kN experiments

— 5-18 —
Chapter 5 Data Analysis

2.0

1.8

1.6

1.4

1.2 Tandem 60 kN
SPTD (mm)

Linear (Tandem 60 kN)


1.0
Expon. (Tandem 60 kN)
0.8
Power (Tandem 60 kN)
0.6
y = -2E-06x + 1.6312
0.4 R² = 0.7354
y = 1.6347e-1E-06x
0.2 R² = 0.7791
y = 4.8122x-0.109
0.0
R² = 0.8082
0 100,000 200,000 300,000 400,000
Assembly passes

Figure 5.26: Approximating initial SPTD for tandem axle 60 kN experiments

2.0
1.8
1.6
1.4
Tandem 80 kN
1.2
SPTD (mm)

Linear (Tandem 80 kN)


1.0 Expon. (Tandem 80 kN)
0.8 Power (Tandem 80 kN)
0.6
y = -1E-06x + 1.3911
0.4 R² = 0.6168
y = 1.3892e-1E-06x
0.2
R² = 0.6329
0.0 y = 4.6198x-0.118
0 100,000 200,000 300,000 400,000 R² = 0.8132
Assembly passes

Figure 5.27: Approximating initial SPTD for tandem axle 80 kN experiments

— 5-19 —
Chapter 5 Data Analysis

2.0

1.8

1.6

1.4 Triaxle 90 kN

1.2 Linear (Triaxle 90 kN)


SPTD (mm)

1.0 Expon. (Triaxle 90


kN)
0.8

0.6 y = -1E-06x + 1.3895


R² = 0.5222
0.4
y = 1.3844e-8E-07x
0.2 R² = 0.5423
0.0
y = 3.5615x-0.094
0 100,000 200,000 300,000 400,000
R² = 0.7326
Assembly passes

Figure 5.28: Approximating initial SPTD for triaxle 90 kN experiments

The best coefficients of determination (R2) were obtained with the power curve fits, so
these were then selected as the most objective way of estimating the starting positions
(Figure 5.29 and Figure 5.30). It was interesting to note that the power curves also
gave the best visual fits, when compared with the work by Major and Tuohey (1976)
and Highways Agency (1999) as discussed in the literature review, section 2.9.1.

— 5-20 —
Chapter 5 Data Analysis

2.0 Single 40 kN

1.8
Tandem 60 kN
1.6

1.4 Tandem 80 kN

1.2
Triaxle 90 kN
SPTD (mm)

1.0
Power (Single
0.8
40 kN)
0.6 Power
(Tandem 60
0.4 kN)
0.2
y = 4.1673x-0.085 y = 4.8122x
-0.109
0.0 R² = 0.8082
R² = 0.6266
0 100,000 200,000 300,000 400,000
y = 3.7511x-0.101 y = 3.5615x
-0.094
Assembly passes R² = 0.7326
R² = 0.6455

Figure 5.29: Approximating initial SPTD for all experiments from Excel power curve fits

Using the equations fitted in Figure 5.29, the calculated SPTD at assembly passes
10,000 (end of bedding-in) are shown in Figure 5.30.

Sing40 Tand60 Tand80 Trix90


1.9 1.8
1.6 1.5

estimation of SPTD (10,000)

Figure 5.30: Estimation of initial SPTD (in mm) after 10,000 passes bedding-in

The variation in the SPTD is thought to be caused by a combination of the variation in


initial surface condition prior to spraying, and the inherent lack of accuracy of the sand
patch test.

— 5-21 —
Chapter 5 Data Analysis

The mean SPTD value is 1.7 mm, with a standard deviation of 0.2 mm. This gives an
estimation of 1.7 mm ± 0.3 mm (95% confidence range) which is acceptable. As all
bedding in was undertaken with one constant load, the subsequent loading chosen for
each section is purely coincidental.

Having established this, it was possible to normalise the data as a percentage of each
trial’s starting point, as shown in Figure 5.31.

SPTD % reduction for axle assemblies


120%
Single 40 kN

Tandem 60 kN
100%
Tandem 80 kN
Reduction in SPTD from initial

80%
Triaxle 90 kN

Power (Single 40
60%
kN)
Power (Tandem
60 kN)
40%
Power (Tandem
80 kN)
Power (Triaxle
20%
90 kN)

y = 2.1887x-0.085 y = 2.7296x-0.109
0%
R² = 0.6266 R² = 0.8082
0 100,000 200,000 300,000 400,000
Assembly passes y = 2.9652x-0.118 y = 2.3775x-0.094
R² = 0.8132 R² = 0.7326

Figure 5.31: Normalised surface texture decay (Excel power curves)

5.6 Models
Mathematical models are commonly used for problem solving. There are two different
approaches to building mathematical models:

• Mechanistic modelling. The model is based on established physical theories


relevant to the problem, such as stress analysis, elasticity, modulus, etc. This
model can be called a physics based model, as the underlying mechanisms are
the starting point for the model building.

— 5-22 —
Chapter 5 Data Analysis

• Empirical modelling. The observed data available forms the basis for the model
building, and it does not require a theoretical understanding of the underlying
mechanisms involved. This kind of model can also be called a data-dependent
model (Murthy et al. 2004).

For modelling road deterioration and maintenance effects (RDME) there may be a third
approach:

• Mechanistic-empirical, or structured empirical, identifying functional form and


primary variables from external sources and then using various statistical
techniques to quantify their impact. The resulting models combine both the
theoretical and experimental basis of mechanistic models with the behaviour
observed in empirical studies. (ND Lea 1995).

In empirical modelling, if the analysis indicates a high degree of variability then models
that can capture this variability are required and these are probabilistic or stochastic
models (Murthey et al. 2004).

• Stochastic or probabilistic models provide one or more reasonable solutions


based on their probabilities.

• Deterministic models produce a single mathematically exact solution (Austroads


2015).

The modelling philosophy adopted for this work is deterministic empirical.

5.7 Nonlinear regression modelling


Power curve fits such as 𝑦𝑦 = 𝑎𝑎𝑥𝑥 𝑏𝑏 using Microsoft Excel are statistically of limited use in
resolving more complex research questions, and more detailed and powerful models
were sought.

A reference cited frequently by other authors is Ratkowsky (1990). This reference was
used at length throughout this component of this study. At first, however, Ratkowsky in
another publication (1993) has outlined some principles of nonlinear regression
modelling which are summarised as follows.

— 5-23 —
Chapter 5 Data Analysis

5.7.1 Principles of nonlinear regression modelling


Parsimony
Models should contain as few parameters as possible. William of Ockham embodied a
basic philosophical principle that a simple solution or answer is far more desirable than
a complex one. This philosophy is now commonly referred to as ‘Ockham’s Razor’.
The view expressed is that a complicated model is not necessarily superior to a simple
model, and simplicity should be preferred.

Parameterisation
Many models can be re-parameterised, i.e. one function in the model may be
expressed in terms of a different model. An example of this is

𝑦𝑦 =∝ 𝛽𝛽 𝑥𝑥 5.1

which can be re-parameterised with

𝛽𝛽 = 𝑒𝑒𝑒𝑒𝑒𝑒(𝛾𝛾) 5.2

to become

𝑦𝑦 = 𝛼𝛼 𝑒𝑒𝑒𝑒𝑒𝑒(𝛾𝛾𝛾𝛾) 5.3

These models are identical, in as much as the fitted value for x will be the same.
However, one model will have better estimation properties over the other.

Range of applicability
The data set used to generate the model should cover the full range over which the
model is intended to apply.

Stochastic specification
All models need an error term є, also called the stochastic term, to enable the
independent term in the model to completely describe the dependent term. An
estimation of є needs to be included with the model. As stated by Field (2009),
outcomei = (model) + errori.

Interpretability
All parameters generated or used in the model need to have an actual meaning that
makes scientific or engineering sense. Can the parameters be interpreted as
something plausible?

— 5-24 —
Chapter 5 Data Analysis

5.7.2 One parameter curves


The first step in model fitting was to observe the scatter plots, and find a model that
approximated the data. In the interests of parsimony, a very simple one parameter
curve was tried at first.

Exponential decay model


Figure 5.32 illustrates a simple exponential decay model.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.32: Exponential decay equation and graph

Source: Motulsky & Christopoulos (2004).

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.33: Exponential decay – effect of constant (k) rates

Source: Motulsky & Christopoulos (2004).

Figure 5.33 shows the effect of four different rate constants ‘k’ (left to right: 0.7, 0.3,
0.1, 0.04) on an exponential decay plot.

An exponential decay model seemed to mimic the behaviour of the scatter plots in this
study very well. In fact, had the purpose of the model been to simply describe the
general shape of the data sets and to interpolate data points between observations, a
simple exponential decay model would have sufficed (as previous researchers have
used).

However, the purpose of the model was to provide some insight into physical
phenomena under study. What does the factor ‘k’ relate to – axle loadings or axle
groupings? Although the principle of parsimony was met, interpretability was not.

To achieve interpretability more was required – at least a two parameter model.

— 5-25 —
Chapter 5 Data Analysis

5.7.3 Two parameter curves


In the phenomena under study, it is assumed there are two parameters contributing to
the outcome – axle groupings and axle force – and therefore two parameter models are
more appropriate. Ratkowsky (1990) was employed and every model that graphically
replicated the data scatter plot shapes was used in SPSS, and found to either not have
solutions or to have standard errors of the parameter estimates that were not
acceptable. Models from other sources were also trialled.

One of these other sources was the British Ports Authority heavy duty pavement
design manual (Knapton 2007).

The damaging effect to pavements due to forklift trucks and cranes handling shipping
containers is of the form:

𝑊𝑊 3.75 𝑃𝑃 1.25 5.4


𝐷𝐷 = � � ×� � × 𝑁𝑁
12000 0.8

where

𝐷𝐷 = damaging effect

𝑊𝑊 = wheel load corresponding to specific container weight (kg)

𝑃𝑃 = tyre pressure (N/mm2)

% figure from a table in Knapton (2007) showing container


𝑁𝑁 =
weights (kg) and the proportions of 40 ft and 20 ft containers.

Model parameterisations were tried of the form

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝛼𝛼𝐺𝐺𝛽𝛽 × 𝛾𝛾𝐹𝐹 𝛿𝛿 5.5

where

𝐺𝐺 = axles in group (i.e. 1, 2, or 3)

𝐹𝐹 = assembly force (i.e. 40, 60, 80 or 90 kN)

𝛼𝛼, 𝛽𝛽, 𝛾𝛾, 𝛿𝛿 = constants.

however acceptable solutions were not found.

— 5-26 —
Chapter 5 Data Analysis

5.7.4 Three parameter curves


At one stage, before the data scatter plots were normalised, the initial starting point
was also a variable. Because of this, initially three parameter curves were also
investigated including every applicable model in Ratkowsky (1990). Again, acceptable
solutions were not found. This failure then finalised the decision to normalise the data,
with the outcome being a reduction in the total number of variables from three to two.

5.7.5 Mechanistic modelling


A brief attempt was made at mechanistic modelling, where the mechanism of the
phenomena is thought through, and a model then devised that mimics the mechanism.

One such model devised was

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆0 5.6
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 =
�1 + 𝑓𝑓(𝐺𝐺, 𝐹𝐹, 𝑖𝑖)�
which became

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆0 5.7
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 =
�1 + (𝛼𝛼𝛼𝛼 × 𝛽𝛽𝐹𝐹 𝛾𝛾 × 𝑖𝑖)�

where

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖 = sand patch texture depth at axle assembly pass 𝑖𝑖

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆0 = sand patch texture depth (initial)

𝐺𝐺 = axles in group (i.e. 1, 2, or 3)

𝐹𝐹 = assembly force (i.e. 40, 60, 80 or 90 kN)

𝑖𝑖 = number of axle assembly passes

𝛼𝛼, 𝛽𝛽, 𝛾𝛾 = constants.

A solution to this model was found in SPSS, but the standard errors (SE) of the
estimates were too high for the model to be acceptable.

— 5-27 —
Chapter 5 Data Analysis

5.7.6 Weibull models


The search for a suitable decay model led outside of engineering towards biology, food
microbiology and bacterial inactivation modelling, where it was discovered that various
parameterisations of Weibull models (Weibull 1951) have been used with success
(Albert & Mafart 2005; Khoo 2006; van Boekel 2002).

Murthy et al. (2004) state that Weibull models have been used in a large variety of
applications and in a range of disciplines.

Some more recent uses of Weibull models in the civil engineering and pavements
discipline have been found to be: the analysis of aging on binder properties of porous
asphaltic concrete (Hagos 2008); the development of a predictive model for the
reduction in skid resistance on an asphaltic concrete (AC) pavement (Kassem et al.
2013); the relationship of aggregate properties to AC skid resistance (Masad et al.
2009); a field evaluation of asphalt mixture skid resistance and its relationship to
aggregate characteristics (Masad et al. 2010); the modelling of road deterioration and
maintenance effects in HDM-4 (N.D. Lea International Ltd 1995, although here there
are some qualifications to its success); validating the fatigue endurance limit for hot mix
asphalt (Prowell et al. 2010), and also predictive modelling for healing and fatigue
endurance limit for asphalt concrete (Souliman 2012).

Weibull models can be used to model a variety of different behaviours as illustrated in


Figure 5.34.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.34: Examples of functions that may be described by Weibull models

Source: Murthy et al (2004).

Weibull (1951) presented a statistical distribution function of:

(𝑥𝑥− 𝑥𝑥𝑢𝑢 )𝑚𝑚 5.8



𝐹𝐹(𝑥𝑥) = 1 − 𝑒𝑒 𝑥𝑥0

This function is now widely given (Murthy et al. 2004) as:

— 5-28 —
Chapter 5 Data Analysis

𝑡𝑡 − 𝜏𝜏 𝛽𝛽 5.9
𝐹𝐹(𝑡𝑡) = 1 − 𝑒𝑒𝑒𝑒𝑒𝑒 �− �� � ��
𝛼𝛼

where

𝑡𝑡 ≥ 𝜏𝜏

𝛼𝛼 = the scale parameter of the distribution

𝛽𝛽 = the shape parameter of the distribution

𝜏𝜏 the location parameter of the distribution.

Equation 5.9 is the three parameter Weibull distribution. The two parameter
distribution is the simple case of this where the location or shift parameter 𝜏𝜏 = 0, and is
shown as Equation 5.10.

𝑡𝑡 𝛽𝛽 5.10
𝐹𝐹(𝑡𝑡) = 1 − 𝑒𝑒𝑒𝑒𝑒𝑒 �− �� � ��
𝛼𝛼

where

𝑡𝑡 ≥ 0

𝛼𝛼 = the scale parameter of the distribution

𝛽𝛽 = the shape parameter of the distribution.

The meanings of scale parameters and shape parameters are not immediately
apparent, and are best described pictorially in the probability distribution functions (pdf)
shown in Figure 5.35 and Figure 5.36 (but note in these figures that their author has
shown α as ŋ).

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.35: The effect of the scale parameter α (shown as ŋ) in Weibull models

Source: Reliasoft Corporation (2002).

— 5-29 —
Chapter 5 Data Analysis

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.36: The effect of the shape parameter β in Weibull models

Source: Reliasoft Corporation (2002).

5.7.7 Individual model predictions


𝑡𝑡 𝛽𝛽
−� �
For this study a Weibull model of the form 𝐹𝐹(𝑡𝑡)=1-𝑒𝑒 𝛼𝛼 was then applied to each
loading regime using IBM SPSS statistics version 21, with the results being illustrated
in Figure 5.37 to Figure 5.40. The critical values of t for given probability levels to test
for goodness of fit were taken from tables in Cooper and Schindler (2003). The
coefficient of determination (R2), the independent variable coefficients (α and β),
standard errors (SE), degrees of freedom (df), test values (t), and probability levels (p),
are tabulated in Table 5.3 for ease of comparison.

120%

100%
SPTD as % of original

80%

60%
Experimental data
40% Individual prediction

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.37: Individual Weibull model prediction for single axle 40 kN

— 5-30 —
Chapter 5 Data Analysis

120%

100%
SPTD as % of original
80%

60%
Experimental data
40% Individual prediction

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.38: Individual Weibull model prediction for tandem axle 80 kN

120%

100%
SPTD as % of original

80%

60% Experimental data


Individual prediction
40%

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.39: Individual Weibull model prediction for tandem axle 60 kN

— 5-31 —
Chapter 5 Data Analysis

120%

100%
SPTD as % of original

80%

60%
Experimental data
Individual predrediction
40%

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.40: Individual Weibull model prediction for triaxle 90 kN

The statistical analysis (summarised in Table 5.3) shows that the coefficients of
determinations (R2) are reasonable, the standard errors (SE) are all less than 50% of
the estimate, and they do not cross zero (i.e. zero is not a likely outcome). The t test
values are all > 2, and the p values are all << 0.05. This combines to give a high level
of confidence in the overall models.

Table 5.3: Statistical analysis of individual Weibull model predictions


Curve R2 Regression Coefficient Standard SE as a % Degree of t test p value
coefficient value error (SE) of freedom value
estimate
Single axle α 628,432 130,545 21% 57-2=55 4.81 < 0.0005
40 kN 0.688
β -0.324 0.037 11% 57-2=55 8.76 < 0.0005

Tandem α 406,180 59,588 15% 41-2=39 6.82 < 0.0005


0.801
axle 60 kN β -0.292 0.028 9% 41-2=39 10.4 < 0.0005
Tandem α 325,682 38,800 12% 45-2=43 8.39 < 0.0005
axle 80 kN 0.813
β -0.284 0.025 9% 45-2=43 11.4 < 0.0005
Triaxle α 711,027 216,267 30% 34-2=32 3.28 < 0.0005
90 kN 0.686
β -0.276 0.042 15% 34-2=32 6.57 < 0.0005

— 5-32 —
Chapter 5 Data Analysis

120%

100%
SPTD as % of original

80%

Single 40 kN
60%
Tandem 80 kN
Tandem 60 kN
40%
Triaxle 90 kN

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.41: Individual Weibull model predictions

The experimental data originally portrayed as chaos in Figure 5.18 can now be more
clearly portrayed by Figure 5.41. The model still does not have the individual
regressions commencing from the same starting point, and this is most likely because
of the errors within the estimation of initial SPTD as shown graphically in Figure 5.25 to
Figure 5.28.

The interpretation of Figure 5.41 now requires some consideration.

If surface texture decay was dominated simply by the axle loading, the expected
ranking would be: (single axle 40 kN & tandem axle 80 kN) < (triaxle 90 kN & tandem
axle 60 kN).

If surface texture decay was dominated simply by the axle groupings, the expected
ranking would be: single axle 40 kN < (tandem axle 60 kN & tandem axle 80 kN) <
triaxle 90 kN.

However, the actual ranking seems to be an interrelationship of both:


single axle 40 kN < triaxle 90 kN < tandem axle 60 kN < tandem axle 80 kN.

This outcome is of particular interest, as it suggests that the effect of the number of
axles in the group does not dominate or swamp the effect of axle load; nor does the
effect of axle load dominate or swamp the effect of the number of axles in the group.

— 5-33 —
Chapter 5 Data Analysis

The decay in surface texture must be some form of function that involves a relationship
between both.

The next stage is to explore what form this function may take.

5.8 Determination of a generalised model


A generalised model is a model that says something beyond the set of observations
that spawned it (Field 2009). It can be applied not just to the sample on which it is
based, but to a wider population. In this case it must be a single expression that is able
to accommodate all axle cluster combinations and all axle loadings, yet also be
capable of reproducing any observed experimental results.

𝑡𝑡 𝛽𝛽
−� �
In Weibull models such as 𝐹𝐹(𝑡𝑡)=1-𝑒𝑒 𝛼𝛼 , there can be cases where α and β are of
such complexity that they are no longer constants but variables within their own right
(for example van Boekel 2002).

In order to calculate what form any variables for the scale parameter α and shape
parameter β may be, the individual Weibull variables as discussed in Table 5.3 and the
paragraph above it, were re-presented in Table 5.4 to allow further investigation.

Table 5.4: Summary of individual Weibull coefficients


Axles in
Experiment Axle force (kN) α β
group
Single axle 40 kN 1 40 628, 432 -0.32426
Tandem axle 60 kN 2 30 406,180 -0.29240
Tandem axle 80 kN 2 40 325,682 -0.28435
Triaxle 90 kN 3 30 711,027 -0.27554

Possible relationships between α and either axle groupings or axle force, and between
β and either axle groupings and axle force, were next investigated by plotting the data
points and inspecting for relationships as shown in Figure 5.42 to Figure 5.45.

— 5-34 —
Chapter 5 Data Analysis

y = 303,798.89x2 - 1,173,898.13x +
α 1,498,531.51
R² = 0.97
800,000
700,000
600,000
Variable a 500,000
400,000
a
300,000
Poly. (a)
200,000
100,000
-
0 1 2 3 4
Axles in group (G)

Figure 5.42: Weibull α as a function of axles in group

Having a parabolic function is not ideal. The experimental program was far removed
from being full-factorial, i.e. in having enough experiments to cover the full combination
of variables (see section 4.6.4 factorial design), which is a stated limitation to the work.
Further data points would see the function of the fitted curve change, but the current
data set suggests that a strongly non-linear fit would still be likely. One interpretation of
this is that the addition of axles produces a definite nonlinear effect in surface texture
wear, and that further work is needed to understand this fully.

α y = -8,154.64x + 803,242.66
R² = 0.07
800,000
700,000
600,000
500,000
Variable a

400,000
a
300,000
Linear (a)
200,000
100,000
-
0 10 20 30 40 50
Axle load (F)

Figure 5.43: Weibull α as a function of axle load

— 5-35 —
Chapter 5 Data Analysis

β y = -0.0115x2 + 0.0705x - 0.3832


R² = 0.9760
0.00
-0.05
-0.10
Variable b -0.15
-0.20
b
-0.25
Poly. (b)
-0.30
-0.35
-0.40
0 1 2 3 4
Axles in group (G)

Figure 5.44: Weibull β as a function of axles in group

β y = -0.0020x - 0.2230
R² = 0.3057
0.00
-0.05
-0.10
-0.15
Variable b

-0.20
b
-0.25
Linear (b)
-0.30
-0.35
-0.40
0 10 20 30 40 50
Axle load (F)

Figure 5.45: Weibull β as a function of axle load

Clearly, the worst model fit of the four is α as a linear function of axle load, with an R2
of 0.07 (Figure 5.43). This was therefore excluded early, leaving the determination of α
as a parabolic function of number of axles in the group, with an R2 of 0.97
(Figure 5.42).

By excluding the relationship with the worst R2 and making α a function of G, then β as
a function of F is the only remaining option, with an R2 of 0.3.

— 5-36 —
Chapter 5 Data Analysis

This, although not an overly strong argument, was considered the only available option
within the budget in maximising the available data, as Figure 5.42 to Figure 5.45 in
reality contain very few data points. It was considered the real test of assumptions
made would be in the analysis of the final model derived from these data.

These relationships were used to generate a trial generalised model of the form:

𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑓𝑓(𝐹𝐹) 5.11


�−�� 𝑓𝑓(𝐺𝐺) � ��
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 (%) = 1 − 𝑒𝑒

where

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = sand patch texture depth (% of original)

𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = number of axle group assembly passes

𝐺𝐺 = number of axles in the grouping

𝑓𝑓(𝐺𝐺) = scale parameter 𝛼𝛼 = 303,799𝐺𝐺 2 − 1,173,898𝐺𝐺 + 1,498,531

𝐹𝐹 force on the halfaxle (kN)

𝑓𝑓(𝐹𝐹) = shape parameter 𝛽𝛽 = −0.0016𝐹𝐹 − 0.2230.

5.8.1 Statistical significance check of the generalised model


To reiterate, whether or not these assumptions for 𝑓𝑓(𝐺𝐺) and 𝑓𝑓(𝐹𝐹) would be sufficient
would depend on testing the statistical significance of outputs from the resulting
generalised model.

The Likelihood Ratio test is used to test statistical significance of non-linear models
(Greene 1990; Martin 2008; Ratkowsky 1990; Thoresen 2006) and uses a sum of
squares of [observed result – predicted result]. It can be easily calculated using an
Excel spreadsheet and equations 5.12.

— 5-37 —
Chapter 5 Data Analysis

𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 = 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿ℎ𝑜𝑜𝑜𝑜𝑜𝑜 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 = 2 × (𝐿𝐿𝑢𝑢 − 𝐿𝐿𝑟𝑟 ) 5.12

where

𝑛𝑛
2
𝑅𝑅𝑅𝑅𝑅𝑅𝑢𝑢 = residual sum of squares, unrestricted case = ��𝑌𝑌𝑜𝑜𝑜𝑜𝑜𝑜 − 𝑌𝑌𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 �
𝑖𝑖=0

𝑛𝑛
2
𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟 = residual sum of squares, restricted case = ��𝑌𝑌𝑜𝑜𝑜𝑜𝑜𝑜 − 𝑌𝑌𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 �
𝑖𝑖=0

log likelihood, unrestricted case


𝐿𝐿𝑢𝑢 = 𝑛𝑛 𝑅𝑅𝑅𝑅𝑅𝑅𝑢𝑢
= − � � × �1 + 𝑙𝑙𝑙𝑙(2𝜋𝜋) + 𝑙𝑙𝑙𝑙 � ��
2 𝑛𝑛

log likelihood, restricted case


𝐿𝐿𝑟𝑟 = 𝑛𝑛 𝑅𝑅𝑅𝑅𝑅𝑅𝑟𝑟
= − � � × �1 + 𝑙𝑙𝑙𝑙(2𝜋𝜋) + 𝑙𝑙𝑙𝑙 � ��
2 𝑛𝑛

and unrestricted case = data as predicted by the model

and restricted case = simple mean of actual data.

The LRTEST value is a Chi-square distribution, with the degree of freedom being the
difference of number of parameters in the model equation and the equation based on
the mean. Chi-square tables are then used to estimate the probability ‘p’ value of the
fitted equation being statistically significant.

5.8.2 Results of the generalised model prediction


The experimental conditions were then input into the generalised model (Equation 11),
and the output of the generalised model compared to the experimental data, as shown
in Figure 5.46 to Figure 5.49. The statistical result of the likelihood ratio analysis is
summarised in Table 5.5.

— 5-38 —
Chapter 5 Data Analysis

120%

100%
SPTD as % of original
80%

60%
Experimental data
40% Generalised model prediction

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.46: Generalised model prediction for single axle 40 kN

120%

100%
SPTD as % of original

80%

60% Experimental data

40% Generalised model


prediction
20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.47: Generalised model prediction for tandem axle 60 kN

— 5-39 —
Chapter 5 Data Analysis

120%

100%

SPTD as % of original
80%

60% Experimental data

40% Generalised model


prediction
20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.48: Generalised model prediction for tandem axle 80 kN

120%

100%
SPTD as % of original

80%

60% Experimental data

40% Generalised model


prediction

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passess

Figure 5.49: Generalised model prediction for triaxle 90 kN

— 5-40 —
Chapter 5 Data Analysis

The statistical analysis is tabulated in Table 5.5 and shows: the log likelihood for the
unrestricted case, data as predicted by the model, (Lu); the log likelihood for the
restricted case, simple mean of actual data, (Lr); likelihood ratio test value (LRTEST);
the degrees of freedom (df); and the resulting probability (p) from a Chi-square
analysis. It is seen that for all cases p < 0.05, indicating the generalised model
predictions are statistically significant.

Table 5.5: Statistical analysis of generalised Weibull model predictions


Curve Lu Lr LRTEST df p
Single axle
93.43 61.78 63.30 1 <0.001
40 kN
Tandem
66.41 34.85 63.11 1 <0.001
axle 60 kN
Tandem
70.82 36.25 69.14 1 <0.001
axle 80 kN
Triaxle
47.21 27.63 39.19 1 <0.001
90 kN

Next, the general model is checked for interpretability, one of the principles of nonlinear
regression modelling discussed in Section 5.7.1.

With 𝑓𝑓(𝐺𝐺) =𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝛼𝛼, then changing the number of axles in the grouping
would have the following effect (Figure 5.50),

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.50: The effect of the scale parameter α (shown as ŋ) in Weibull models

Source: Reliasoft Corporation (2002).

This can be interpreted as suggesting that for a constant load (constant β), the
magnitude of the function under study decreases with increasing number of axles in the
group, and the peak effect occurs further in time. In this case, the rate of deterioration
in surface texture would decrease as the same load is spread over more axles, which
can be interpreted as being plausible.

It is considered that the generalised model is plausible. It does reproduce the data that
spawned it, and the fit is statistically significant.

One final important check on the generalised model is – can it reproduce independently
derived data?

— 5-41 —
Chapter 5 Data Analysis

5.8.3 Check of generalised model using an independent data set


As discussed earlier, the Canterbury accelerated pavement testing indoor facility
(CAPTIF) is the accelerated pavement test (APT) device in use in New Zealand. A
side elevation schematic of it is shown in Figure 5.51. It runs in a continuous circle, as
shown in Figure 5.52.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 5.51: Schematic of CAPTIF in side elevation

Source: Arnold, Steven, Alabaster, & Fussell (2005a), Fig. 1.2.

Figure 5.52: The Canterbury Accelerated Pavement Testing Indoor Facility (CAPTIF)

Transit New Zealand has in the past funded research on the effect of increased mass
limits on pavement wear (Arnold et al. 2005c).

The CAPTIF test pavement was surfaced with a three coat cutback chip seal whilst the
Australian ALF study was surfaced with a two coat seal, as three coat chip seals are
very rare on Australian roads. Two single-tyred ‘slave’ (simulated loading and vehicle
emulator) units were run in offset wheelpaths. The inside wheelpath had a single wheel
load of 20 kN (to simulate the 8.2-tonne dual-tyred axle) while the outside wheel had a
load of 30 kN (to simulate the 12-tonne dual-tyred axle). A conditioning 5,000 loads
were applied at a single wheel load of 20 kN over the whole pavement. A stationary
laser profilometer was used to measure the surface texture, reported in mm of mean
profile depth (MPD).

A summary of the salient differences and similarities between the CAPTIF study and
this study may be helpful, and is given in Table 5.6.

— 5-42 —
Chapter 5 Data Analysis

Table 5.6: Main differences between CAPTIF and ALF


Item CAPTF ALF
Loading action Circular Linear
Loading speed 20 km/h 20 km/h
Tyres Single (similar to 11R22.5) Dual 11R22.5
Axles Single Single, tandem & tri
Loadings 21 &26 kN on single tyres (in lieu 40, 60, 80 & 90 kN on tandem tyres
of to 40 & 50 kN on tandem tyres) over individual virgin sites
over concentric offset wheel paths
Pavement material Replicated typical NZ pavement Representative of crushed rock
with low levels of rutting quality typical of the Australian rural
network
Pavement Pavement in concrete lined tank Not controlled
moisture control
Seal coat Cutback prime Emulsion prime
10 mm single seal with 80/100 pen 14/7 two coat seal with 5% latex in
bitumen emulsified bitumen
Precondition rolling, then
14/7 double seal with
80/100 pen bitumen

Bedding-in 5,000 cycles of the single axle 10,000 cycles of tandem axle dual
single tyre loaded at 21 kN tyre loaded at 60 kN
(15 kN per tyre)

A three coat seal for CAPTIF was used because of the circular loading motion, and the
desire for a robust seal able to withstand turning stress (de Pont et al. 2003).

A two coat seal was used at ALF because of the linear loading motion, and was made
more robust by the use of a polymer modified binder.

The differences are not considered major, and both use representative crushed rock
typical of their respective countries unbound granular networks, both use similar tyres
and similar tyre loadings at similar speeds. The different definitions/requirements of
bedding-in, however, will be discussed further.

— 5-43 —
Chapter 5 Data Analysis

The raw data from the CAPTIF experiment is available in the public domain, and is
published in the appendix of Arnold et al. (2005b), amongst other places. This data is
now portrayed as accumulated axle assembly passes vs surface texture reduction, in
Figure 5.53. It would appear that after the conditioning loading of 5,000 passes the
surface was not yet ‘bedded-in’ compared to what was achieved under ALF by
bedding-in. Under the ALF definition, by reviewing the CAPTIF graph (Figure 5.53)
bedding-in would not have been deemed to have occurred until about 8,000 passes.

4.0
3.5
3.0
2.5
MPD (mm)

2.0
8.2 t simulation data
1.5
12 t simulation data
1.0
0.5
0.0
0 20,000 40,000 60,000 80,000 100,000 120,000
Axle assembly passes

Figure 5.53: CAPTIF raw data

Source: adapted from Arnold, Steven, Alabaster, & Fussell (2005b).

The CAPTIF raw data was input to the new generalised model, and the results are
shown in Figure 5.54 and Figure 5.55.

— 5-44 —
Chapter 5 Data Analysis

120%

100%
MPD % reduction from original

80%

60% 8.2 t CAPTIF data

Generalised model
40%
prediction

20%

0%
0 20,000 40,000 60,000 80,000 100,000 120,000
Axle assembly passes

Figure 5.54: Generalised model using CAPTIF (8.2 t simulation) data – using 5,000 passes bedding-in

120%

100%
MPD % reduction from original

80%

60% 12 t CAPTIF data

Generalised model
40% prediction

20%

0%
0 10,000 20,000 30,000 40,000 50,000 60,000
Axle assembly passes

Figure 5.55: Generalised model using CAPTIF (12 t simulation) data – using 5,000 passes bedding-in

— 5-45 —
Chapter 5 Data Analysis

Aside from the issue of the definition of bedding-in, the rates of decay after bedding-in
appear very similar (the slopes of the decay curves appear effectively parallel), and
confirm the generality of the model.

If bedding-in is redefined and recalculated to the first 9,500 passes of data (the closest
data point to the 10,000 passes used at ALF) and then with this amount of bedding-in
excluded, the model output matches with the CAPTIF data well, as shown in
Figure 5.56 and Figure 5.57.

120%

100%
MPD % r4eduction from original

80%

60% 8.2 t CAPTIF data

Generalised model
40%
prediction

20%

0%
0 20,000 40,000 60,000 80,000100,000120,000
Axle assembly passes

Figure 5.56: Generalised model using CAPTIF (8.2 t simulation) data – using 9,500 passes bedding-in

— 5-46 —
Chapter 5 Data Analysis

120%

MPD % reduction from original


100%

80%

60% 12 t CAPTIF data

40% Generalised model


prediction
20%

0%
0 10,000 20,000 30,000 40,000 50,000 60,000
Axle assembly passes

Figure 5.57: Generalised model using CAPTIF (12 t simulation) data – using 9,500 passes bedding-in

Thus it can be concluded that the generalised model is validated by using data that is
totally independent to the derivation of the model. These CAPTIF tests were
undertaken on unbound granular road pavements surfaced with a sprayed seal, and
the resulting data is freely available in the public domain. This raises additional
confidence in the validity of all the assumptions made during the data collection and the
model formulation.

5.9 Analysis of equivalent standard axles (ESA) and


standard axle repetitions (SAR)
With a generalised model now established to predict rate of decay of surface texture
from the number of axle passes of G axle clusters and F axle loads, it then becomes
possible to investigate the initial research question: do the equivalent standard axles
that cause the same damage to a pavement, cause different damage to a sprayed
seal?

The experimental data was collected with dual tyre axles only. Equivalent standard
axles for dual tyres are reproduced in Figure 5.58.

Figure 5.58: Equivalent standard axles (dual tyres)

— 5-47 —
Chapter 5 Data Analysis

These axle groupings and axle loads can then be input into the generalised model
(equation 5-11), noting the model uses half axles, thus half the axle loads. The result
is shown in Figure 5.59.

120%

100%
SPTD as % of original

80%

SADT 40
60%
TADT 67.5 kN
TRDT 90.5 kN
40%
QADT 110.5 kN

20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 5.59: Comparison of dual tyre ESAs by the generalised model

It is immediately apparent from Figure 5.59 that axle groupings and loadings that cause
the same damage in an unbound granular pavement can cause different damage to the
sprayed seal surfacing covering this pavement.

It must be noted that this is an empirical model based on experimental observations of


the dependent variable (macrotexture) after altering the independent variables (axle
groupings and axle loadings). It is not a mechanistic model based on theoretical
equations of material behaviour and pavement response functions. The main
advantage that mechanistic models may claim is their ability to extrapolate predictions
beyond the original data range.

— 5-48 —
Chapter 5 Data Analysis

This is an empirical model, and it is worth remembering that predicting a quad axle
outcome is an extrapolation beyond the data used to create the model, where single,
tandem and tri axles were used. However, although this was understood to be a
potential weakness from the outset, it was always intended to use the model as a first
indication of what may happen. It is surprising then to note that the model predicts that
a quad axle (under this ‘standard’ axle loading) causes the least reduction in surface
texture wear of all axle groups. Under different loading scenarios, this may not always
be the case, and this is an opportunity for further work.

It can also be seen that the rate of wear from a single axle is initially rapid but then
becomes slower over time when the rate of wear becomes slower than that of a
tandem and triaxle.

These results are applicable to pavement as constructed for this experiment, and
sealed with a two coat emulsion PMB that was not significantly aged. The CAPTIF
experiments were also undertaken with a seal that had not significantly aged.

However, it is most important to note that this work has identified a case where
equivalent standard axle loadings, believed to cause the same pavement damage
(where damage may be deflection, rutting, or roughness as introduced in Section
2.10.6) have not caused equivalent surface texture damage.

Consider the statement ‘All axle types when loaded to an equivalent standard axle, will
cause the same surface damage’. This statement is testable. If a single counter
example is seen, then it can be concluded that the statement is not true. The
statement also falsifiable, i.e. if it is false this can be shown by observation or
experiment (Clem 2010). This study has found the statement to be false, and has now
provided a documented argument against adopting any existing ESA database for
heavy vehicle sprayed seal design.

— 5-49 —
Chapter 5 Data Analysis

5.10 Analysis of the load damage exponent


Scala’s equation for load damage exponent, as described in section 2.10.6 is
reproduced below.

𝐷𝐷𝑡𝑡 𝐿𝐿𝑡𝑡 𝑚𝑚 5.13


=� �
𝐷𝐷𝑠𝑠 𝐿𝐿𝑠𝑠

where

𝐷𝐷𝑡𝑡 = damage caused by the test axle

𝐷𝐷𝑠𝑠 = damage caused by the standard axle

𝐿𝐿𝑡𝑡 = load on the test axle

𝐿𝐿𝑠𝑠 = load on the standard axle

𝑚𝑚 = a load damage exponent.

Using this equation, a load damage exponent for sprayed sealing, with damage being
defined as a percentage reduction in surface texture (sand patch texture depth) can be
calculated from the experimental data.

As a check on the sensitivity of the damage definition, a load damage exponent has
been calculated for two damage definitions: a 20% reduction in surface texture, and a
30% reduction in surface texture.

These calculations are shown below, and are derived for a dual tyre tandem axle.

— 5-50 —
Chapter 5 Data Analysis

Figure 5.60: Axle passes to cause equal damage (20% reduction of texture)

The calculation for the load damage exponent (for 20% reduction of texture) uses data
from Figure 5.60 and is as follows:

𝐷𝐷𝑡𝑡 𝐿𝐿𝑡𝑡 𝑚𝑚
=� �
𝐷𝐷𝑠𝑠 𝐿𝐿𝑠𝑠

(62,375 − 10,000) 80 𝑚𝑚
=� �
(81,000 − 10,000) 60

0.737 = (1.333)𝑚𝑚

𝑙𝑙𝑙𝑙𝑙𝑙(1.290)
𝑚𝑚 = = 1.06
𝑙𝑙𝑙𝑙𝑙𝑙(1.333)

— 5-51 —
Chapter 5 Data Analysis

Figure 5.61: Axle passes to cause equal damage (30% reduction of texture)

The calculation for the load damage exponent (30% reduction of texture) uses data
from Figure 5.61 and is as follows:

𝐷𝐷𝑡𝑡 𝐿𝐿𝑡𝑡 𝑚𝑚
=� �
𝐷𝐷𝑠𝑠 𝐿𝐿𝑠𝑠

(168,000 − 10,000) 80 𝑚𝑚
=� �
(220,000 − 10,000) 60

0.752 = (1.333)𝑚𝑚

𝑙𝑙𝑙𝑙𝑙𝑙(0.752)
𝑚𝑚 = = 0.989
𝑙𝑙𝑙𝑙𝑙𝑙(1.333)

From these experiments it can be concluded that the load damage exponent for
tandem axle dual tyres causing a reduction in surface texture is

— 5-52 —
Chapter 5 Data Analysis

𝐿𝐿𝐿𝐿𝐿𝐿 ≅ 1 5.14

where

𝐿𝐿𝐿𝐿𝐿𝐿 = load damage exponent

and

𝑚𝑚 = 𝐿𝐿𝐿𝐿𝐿𝐿.

This can then be compared with other published load damage exponents, as shown in
Table 5.7.

Table 5.7: Comparison of load damage exponents


Material Load Source/Method Reference
Damage
Exponent
Fatigue of cemented materials 12 Austroads design guide Jameson (2012)
Rutting and loss of shape, 7 Austroads design guide Jameson (2012)
flexible pavements (Mechanistic)
Fatigue of asphalt 5 Austroads design guide Jameson (2012)
Overall granular pavement 4 Austroads design guide Jameson (2012)
damage (empirical)
Chip seal surface texture wear 3.1 Modified Patrick 1 Arnold et al.
–derived from CAPTIF single model (2005c)
axle single tyre 8 t vs 10 t
Chip seal surface texture wear 2.0 Modified Patrick model Arnold et al.
–derived from CAPTIF single (2005b)
axle single tyre 8 t vs 12 t
Chip seal surface texture wear 1.7 Modified Kinder-Lay 2 De Pont et al.
–derived from CAPTIF single model (2003)
axle single tyre 8 t vs 10 t
Chip seal surface texture wear 1 Basic Scala method This study
–derived from ALF tandem axle (Austroads method)
dual tyre 12 t vs 16 t

1
For the original Patrick model refer to Patrick et al. (1988)
2
For original Kinder-Lay model refer to Kinder and Lay (1988)

— 5-53 —
Chapter 5 Data Analysis

This study’s method of calculation of LDE, and the resulting exponent, have been
discussed with Arnold (the lead author of Arnold et al. 2005 b and 2005c). Arnold
commented that his more recent experience would support a LDE lower than that
published in 2005.

5.11 Summary
From beginning with data that at first appeared chaotic, this chapter has described how
the data was checked for outliers and for operator error, normalised to allow
comparison, and how a multivariate nonlinear regression model passing statistical
scrutiny was identified for each individual experiment.

Next a generalised model – ‘one that says something beyond the set of observations
that spawned it’ – was developed. This generalised model passed the tests of
parsimony, interpretability, and statistical significance, and was used to reproduce data
from a independent experiment conducted by researchers in a different country.

The model now invites the questions: is there a practical application to road
authorities? Can it be used to improve heavy vehicle seal design? How? What are the
strengths and weakness? These questions are quite valid and warrant a small chapter
of their own, as will be discussed next.

— 5-54 —
Chapter 6 Discussion of the model

6 DISCUSSION OF THE MODEL


Essentially, all models are wrong, but some models are useful.
(Box & Draper 1987).

6.1 Introduction
The previous chapter described how the data of several hundred sand patch texture
depth readings were filtered and transformed into a form that was able to yield some
intelligence, followed by the development of a general model that is able to predict the
different rate of surface texture attenuation for a variety of axle loadings over a variety
of axle groupings. Another outcome of this work was identification of a load damage
exponent for sprayed seal wear, and an assertion that axle groupings and loadings that
cause the equal damage to a pavement, do not necessarily cause equal damage to a
sprayed seal.

The generalised model met the fundamental principles of nonlinear regression


modelling, and was validated using independent data.

The resulting model is reproduced again here for reference, as

𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑓𝑓(𝐹𝐹) 6.1


�−�� 𝑓𝑓(𝐺𝐺) � ��
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 (%) = 1 − 𝑒𝑒

where

𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = sand patch texture depth

𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = axle assembly passes

𝑓𝑓(𝐺𝐺) = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝛼𝛼 = 303,799𝐺𝐺 2 − 1,173,898𝐺𝐺 + 1,498,531

𝑓𝑓(𝐹𝐹) = 𝑠𝑠ℎ𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝛽𝛽 = −0.0016𝐹𝐹 − 0.2230

𝐺𝐺 number of axles in the axle assembly (1,2,3 or 4)

𝐹𝐹 force on the half axle (kN).

This chapter will interpret the model and discuss its strengths and weakness, it’s
practical application to road authorities, and how it could be used to answer some of
the identified research gaps and to improve heavy vehicle seal design.

— 6-1 —
Chapter 6 Discussion of the model

6.2 Strengths and weaknesses of the model


This model is but one interpretation of the data.

One way of describing multivariate analysis is that it is a way of


creating a “model” of the data. This ‘model’ is an artificial, simplified
version of reality. For example, the mean is a ‘model’ of data. Many
different models of the same data may be possible (Cunningham
2012).

On first appearance, the model is cumbersome, consisting of functions within a


function.

Further work could be done to improve upon the model, and this is discussed in
Section 7.5.

In balance, however, an analogy is that this model derived using a Weibull model is the
basic ‘chassis’ from which to start building a horseless carriage. There is no doubt that
the next model built will be faster and have better handling. This model in its current
form, however, is still useful in testing the hypothesis ‘axle types when loaded to an
equivalent standard axle will not cause the same surface damage’ and finding that
hypothesis to be true (Section 5.9).

It is considered a useful model because as stated in an earlier chapter but in a different


context, that although precision (refinement in the measurement) has been affected
because of the inherent variations in the sand patch test, the accuracy (conformity with
the truth) has been preserved by a sound methodical process of analysis, an
acknowledgment of the limitations, and by validation through totally independent data
from another country.

6.3 Interpretation of the model


There are several methods by which the surface texture of sprayed seals may
attenuate or decay: void loss due to aggregate embedment into the basecourse; void
loss due to aggregate wear; aggregate reorientation and changes to the aggregate
mosaic; and loss of aggregate from the seal (Alderson & Oliver 2008; Neaylon 1996;
SANRAL 2007).

— 6-2 —
Chapter 6 Discussion of the model

Visual observations of the ALF trials after trenching (as described in Moffatt 2013) This
trenching also enabled a visual comparison between the trafficked aggregate mosaic
and the non-trafficked aggregate mosaic. Visual observations indicate that the
deformation within the sprayed seals in this experiment occurred mostly due to the
realignment of seal aggregates during trafficking, and not by embedment into the
basecourse. Aggregate wear was visually assessed as being no different than
normally expected, and as the aggregate source used throughout the experiments was
held constant, this variable, although it may have had a minor contribution, was
constant.

It is postulated that the reorientation of aggregate is strongly influenced by the


horizontal component of traffic loadings and the rotation of shear stresses (Figure 6.5)
in addition to the vertical loading component as solely considered in pavement design.
It is also postulated that other influencing factors exist, such as the viscoelastic
behaviour of bitumen, which does not play a part in unbound granular pavement
design.

6.3.1 Bituminous viscoelastic relationship and aggregate reorientation


The viscoelastic properties of bitumen become evident in the application of different
shear rates of loadings on sprayed seals and at various temperatures. It is known that
the stiffness of bitumen exhibits an elastic response under conditions of low
temperatures or short loading times, but shows a viscous (i.e. plastic) response at high
temperatures or long loading times (Read & Whiteoak 2003; van der Poel 1954) as
illustrated in Figure 6.1 and Figure 6.2.

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 6.1: Stiffness vs loading times at different temperatures for two different bitumens

Source: van der Poel (1954).

This image is unable to be reproduced online. Please consult the print copy held in the Swinburne Library.

Figure 6.2: The effect of temperature and loading time on the stiffness of a 100 pen bitumen

Source: Read & Whiteoak (2003).

— 6-3 —
Chapter 6 Discussion of the model

Figure 6.2 shows the temperature – loading time – stiffness relationship for a 100 pen
bitumen with softening point of 43 °C and penetration index of -1.0. At a loading time
of 0.02 s, which equates to a vehicle speed of around 50 km/h when using 𝑡𝑡(𝑠𝑠) =
1
from Denneman (2013), then the stiffness modulus at 25 °C is about 107 Pa.
𝑉𝑉(𝑘𝑘𝑘𝑘/ℎ)

It has also been shown that at intermediate temperatures, where bitumen exhibits both
viscous and elastic stiffness (i.e. is visco-elastic), the stiffness is stress dependant.
Higher stress results in lower stiffness and lower stress results in higher stiffness (Read
1996).

Table 6.1 shows factors that Read (1996) identified as affecting the stiffness modulus
of bituminous materials (as measured by the indirect tensile stiffness modulus (ITSM)
test).

Table 6.1: General effect of the main factors affecting stiffness

Factor General effect on stiffness


high temperature → low stiffness modulus
temperature
low temperature → high stiffness modulus
low frequency → low stiffness modulus
loading frequency
high frequency → high stiffness modulus
high stress → low stiffness modulus
stress amplitude
low stress → high stiffness modulus
Adapted from Read (1996).

Considering the above, the loading stresses, loading duration, and binder temperature
can each affect the binder stiffness, both in magnitude and in recovery (in terms of
elastic or plastic stiffness). These will then affect the binders’ resistance to any
reorientation of aggregate particles which would resist a reduction in surface texture.

This concept is developed in Table 6.2, which illustrates the effects of the various
possible combinations of loading stress, loading duration and binder temperature.

— 6-4 —
Chapter 6 Discussion of the model

Table 6.2: Viscoelastic factors affecting binder stiffness

Loading Loading duration Binder Resulting binder stiffness Expected texture


stresses (Read & Whiteoak temperature (Read response deterioration rate
(Read 1996) 2003) & Whiteoak 2003)
high temp → very high
low in magnitude, strongly plastic
long duration → plastic
plastic low in magnitude, both plastic and high
high stress low temp → elastic
elastic
→low stiffness
high temp → low in magnitude, both elastic and high
short duration → plastic plastic
elastic
low temp → elastic low in magnitude, strongly elastic medium
high temp → medium
high in magnitude, strongly plastic
long duration → plastic
plastic high in magnitude, both plastic low
low stress → low temp → elastic
and elastic
high stiffness
high temp → high in magnitude, both elastic low
short duration → plastic and plastic
elastic
low temp → elastic high in magnitude strongly elastic very low

Table 6.2 will be used as background for further explanations developed in section 6.4.

6.3.2 Rest periods


Read (1996) found that increasing the rest period between loads increased the
observed stiffness. He defined a rest period as the time between successive
applications of wheel loads. In the study of asphalt fatigue, rest periods are important
as they allow time for cracks to heal and for the relaxation of stresses and strains
through the viscous flow of bitumen. Self-healing was also noted on asphalt samples
that were stored for periods of time. Self-healing was seen to improve with high
temperatures, long rest times, high bitumen content and the use of low viscosity
binders, all of which indicated that healing is a function of the viscous flow of bitumen.

Read found that the effect of rest periods on the fatigue response of a bituminous
paving mixture had been well researched, and identified at least 6 other researchers
who concluded that inclusion of a rest period can give an increase in asphalt fatigue life
of between 5 and 25 times (Read 1996), depending on the ratios between the length of
load duration and the length of the rest period. Read also equates an increase in
fatigue life with an increase in stiffness.

— 6-5 —
Chapter 6 Discussion of the model

In this study the chip seal has a bitumen content of 31% v/v, an order of magnitude 10
times higher than typical asphalt, and that the viscosity grade of bitumen is generally
lower for chip seals than for asphalt, and therefore the action of self-healing could be
even faster in chip seals.

Rebbechi and Petho (2014) confirm that shift factors are normally applied when
converting the asphalt fatigue life in a laboratory test to the higher fatigue life found in
the field. The increase in field fatigue is attributed to rest periods, traffic wander and
self-healing. The magnitude of this shift factor in Australasia ranges from 10 to 100
times.

It can be inferred from this that rest periods have a large influence of the field
performance of bituminous surfacings.

When comparing triaxle groupings with single axle groupings, it could be considered
that passes of a triaxle grouping would lead to shorter rest periods. A triaxle grouping
gives three pulses in rapid succession followed by a longer pause, whereas a single
axle gives only the longer pause between each individual pulse. Under a triaxle
grouping pass there is less time for self-healing, meaning lower dissipation of strain
through the viscous flow of bitumen. Lower dissipation of strain means that the
aggregate particles would tend to stay where they are rather than show a tendency to
reorientate. An outcome of this would be that triaxle groupings cause a lower reduction
in surface texture, which is predicted by the model.

6.3.3 The relationship between pavements and surfacings


a) Superposition of pavement strains
When considering pavement behaviour, a concept of superposition of strains can be
applied. In some modelling and design software it is common to model multiple loads,
such as the four contact points of a standard axle (being single axle dual tyre) as a
single load at the various offsets.

For example, if the horizontal strains at a particular depth below tyre B in Figure 6.3
need to be calculated, the effect of the other loaded tyres are equivalent to the effect of
load B at the other three locations. Load B will cause a strain under load locations A,
B, C and D. The superposition concept enables the total strain under B to be equal to
the summation of the strains caused by load B at locations A+B+C+D (for further detail
see Gonzales et al. 2012).

— 6-6 —
Chapter 6 Discussion of the model

Figure 6.3: Superposition of strains under a standard axle

Adapted from Gonzales et al. (2012).

This concept relies on the dispersion of surface loads through the pavement structure.

Moffatt (2015) has applied a similar concept to moving multiple axle groups, saying that
at the surface the axles can be considered to act independently, but at depth a
pavement element will still be reacting to a passing load as the approaching load nears
it, as shown in Figure 6.4.

Figure 6.4: Dispersion of surface loads through a pavement structure

Source: Moffatt (2015).

A conclusion from this is that the effect of loads at a pavement depth can be
significantly different to the effects of loads at a pavement surfacing. For a surfacing
the discrete individual axle loads are important, but at depth the combined axle
grouping loads play an increased role. It would appear (Moffatt 2015) that the physical
spacing between the axles within the axle grouping is not particularly critical.

— 6-7 —
Chapter 6 Discussion of the model

The actual calculation of stresses, however, is more complex than simple superposition
theory.

b) Rotation of shear stress


As road pavements and surfacings are loaded by moving wheel loads, the magnitudes
of the vertical, horizontal and shear stresses will vary as the applied principal stress
rotates. This can be shown pictorially in Figure 6.5, where the principle stresses act
purely vertically (σz) and horizontally (σr) only when the centre of the wheel load is
directly above (Arnold 2004; Shaw 1980; Werkmeister 2003).

Figure 6.5: Stresses beneath a rolling road

Source: Werkmeister (2003).

One possibility is that although the rotation mechanism reduces in magnitude closer to
the surface, it may actually be a contributing factor towards surface aggregate
reorientation. It can also be suggested that the time and temperature dependant
viscoelastic properties of bitumen would play a significant role in the final outcome,
along with the packing density of the surface aggregate (expressed as aggregate
spread rate in m3/m2) which would prove some resistance to movement depending on
the degree of mechanical interlock between the aggregate particles.

To what degree it is possible to take this pavement theory of stress rotation at depth to
the surfacing theory and its application to chip seal aggregate rotation, requires further
investigation.

— 6-8 —
Chapter 6 Discussion of the model

6.3.4 Tyre contact stress


It has been well established that for a given tyre the vertical contact stress distribution
below a tyre changes in location and intensity as a function of tyre loading and tyre
inflation pressures. De Beer and Maina (2011) have shown that for overloaded and/or
underinflated tyres, the vertical contact stresses are highest towards the edges of the
tyre footprint. For higher inflation pressures at lower loads, the vertical contact
stresses are highest towards the centre of the tyre footprint.

The discussion at section 2.10.1 illustrates that the peak surface stress directly below a
tyre (at the sprayed seal surface) can be much higher (as much as double) than the
calculated average surface stress, which for pavement design purposes is often
approximated as the tyre inflation pressure (De Beer 2015; De Beer et al. 2012). The
actual tyre footprint of the triaxle assembly on carbon paper can be found in the
appendix (Figure B.1).

Peak localised stresses under the tyre footprint may can be expected to cause
variations in aggregate embedment of a sprayed seal. In this study, any variations in
tyre contact stresses between the centre of the contact patches and beneath the
sidewall of the tyres were not considered, as traffic wander was programmed into the
ALF trafficking so that any such affect would be smoothed and these higher stresses
would be distributed more evenly over the wheelpath as they are in normal traffic.

6.4 Explanation of the model outcome


Consider Figure 6.6, comparison of dual tyre ESAs by the generalised model (half-axle
loads shown), first shown in Chapter 5 Data analysis.

— 6-9 —
Chapter 6 Discussion of the model

Figure 6.6: Comparison of dual tyre ESAs by the generalised model (half-axle loads shown)

An explanation of why the model shows this ranking requires a combination of the
viscoelastic nature of bitumen with the difference between the superposition of
pavement strains compared to surface strains.

The ranking of the axle groups with regard to their deterioration effects on surface
macrotexture is shown in Table 6.3

Table 6.3: Ranking of deterioration effect

Effect Axle types (when loaded as ESAs)


Benign Quad axle dual tyre
↓ Tri axle dual tyre
↓ Single axle dual tyre
Aggressive Tandem axle dual tyre

Given that the tyre footprint is approximately 150 mm diameter (Appendix B Fig B1),
the individual axles are at 330 mm centres, speed is 22 km/h and cycle time is 10
seconds (Section 4.2), the loading pulse of the first 1 second are shown in Figure 6.7 to
Figure 6.10. The figures are ranked in order of most benign to most aggressive. For
the remaining nine seconds of the cycle the load remains at zero.

— 6-10 —
Chapter 6 Discussion of the model

40
35
30

Load (kN)
25
20
15
10
5
0
0 0.2 0.4 0.6 0.8 1
loading time (s)

Figure 6.7: Quad axle load pulse

40
35
30
Load (kN)

25
20
15
10
5
0
0 0.2 0.4 0.6 0.8 1
loading time (s)

Figure 6.8: Tri axle load pulse

40
35
30
Load (kN)

25
20
15
10
5
0
0 0.2 0.4 0.6 0.8 1
loading time (s)

Figure 6.9: Single axle load pulse

— 6-11 —
Chapter 6 Discussion of the model

40
35
30

Load (kN)
25
20
15
10
5
0
0 0.2 0.4 0.6 0.8 1
loading time (s)

Figure 6.10: Tandem axle load pulse

One hypothesis arising from these graphs is that a lighter load applied in a greater
number of pulses, causes less plastic deformation (in this case in the form of aggregate
reorientation) whilst a higher load with a smaller number of pulses causes greater
plastic deformation. However, this research lacks the necessary data to test this
hypothesis.

Sprayed seal performance is highly dependent on tyre contact pressure rather than
axle load, and whilst the tyre inflation pressures were all held constant, individual
contact pressures were not measured in the belief that the wandering of the wheel path
would average this effect out. However it is now apparent that measuring the
maximum contact pressure, which can be up to twice the inflation pressure, would have
provided useful data. The rheological properties of the binder (a polymer modified
emulsion), which was not available at the time is also required to characterise the
elastic and plastic response of the binder at different loading rates and frequencies.

This is now an ideal opportunity for further research, given the advances made in finite
element modelling (FEM), and a recommendation for this is made in Section 7.5, future
work.

Thus it can be suggested that the viscoelastic response of a bituminous binder could
contribute to the reduction of surface texture through its ability to effect the plastic or
elastic response of bitumen, and therefore assist or hinder aggregate reorientation.
The effect of aggregate wear would also assist in texture reduction, but as the same
aggregate was used for all experiments this effect has been held constant.

— 6-12 —
Chapter 6 Discussion of the model

It may be considered that given this, a triaxle would cause less decay than three single
axles. However, the model suggests that for the same total freight tasks on ESA
loadings they are the same (Figure 6.14).

This suggests therefore, that the actual outcome is somewhat more complex than
simple additions of discrete properties. For example, some stiffness relationships are
not linear, but rather log-log as shown in Figure 6.1 and Figure 6.2. The elastic
response of bitumen under transient loading is also not linear (Dickinson 1985; Read
1996) as shown in Figure 6.11.

Figure 6.11: Strain response of a polymer modified binder

Source: Dickinson (1985).

It is therefore suggested that the interrelationship is more complex than simple


consideration of loading times, and that the individual axle loadings in the triaxle
cluster, and the loadings on the individual axle within the axle cluster are a contributing
factor that also come into play.

The model output shown in Figure 6.14 suggests that the rate of texture decay caused
by single and triaxle ESAs are similar. Utilising Figure 6.7 to Figure 6.10 suggests that
the combination of three closely spaced axles of 30.2 kN load cause the same total
decay as three isolated axles of 40 kN load.

— 6-13 —
Chapter 6 Discussion of the model

Put another way, an increase in the rate of surface texture decay caused by the plastic
response increasing when the load frequency increases (loading goes from one
isolated loading event to three loads in very quick succession) is matched in magnitude
by a reduction in the rate of surface texture decay caused by the lower stress (the
stress under the lower 30.2 kN load causing a higher stiffness than under the 40 kN
load).

It must be kept in mind that for this research the visco-elastic relationship is in the
context of:

1. the ambient pavement temperatures of the experiments, which were constantly


in shade, and therefore lower than maximum temperatures commonly
experienced on the road (and because of the shelter provided, the minimum
ambient temperatures were probably higher than the minimum temperatures
experienced on the road)

2. the travel velocity of ALF, which was 20 km/h.

The inter-relationship between these two contributing factors – decreasing loading


stresses and increasing loading rate – appears to be incorporated and functioning
within the model developed, as evidenced by trialling different input parameters as
done in Section 6.5.2.

6.5 Using the model and the data to address research gaps
6.5.1 Comparisons of the accumulation of surfacing wear
Consider a freight task of say 95,000 t. Which axle group would cause less surface
texture reduction in the transporting of this freight; single axles or tandem axles?

To rephrase this question in terms of the experimental data, did 240,000 passes of the
single 40 kN axle (a total of 9.6 ×106 kN, or 94,176 t) cause the same amount of
surface texture deterioration as 120,000 passes of the tandem 80 kN axle (a total of
9.6 ×106 kN, or 94,176 t)? The results are shown in Figure 6.12.

— 6-14 —
Chapter 6 Discussion of the model

120%

100%

SPTD as % of original
80%

60%
Sing 40 kN
40%
Tand 80 KN
20%

0%
0 100,000 200,000 300,000
Axle assembly passes

Figure 6.12: The same total freight task carried by both single and tandem axles

Figure 6.12 shows that in the experimental study, for this combination it did not matter
whether the freight task was carried by single axles or tandem axles, the reduction in
surface texture of 25% (or down to 75% of original) was the same for both.

Similarly then, did 300,000 passes of the tandem 60 kN axle (a total of 18 ×106 kN, or
176,580 t) cause the same amount of surface texture deterioration as 200,000 passes
of the triaxle 90 kN axle (a total of 18 ×106 kN or 176,580 t)? The results are shown in
Figure 6.13.

120%
100%
SPTD as % of original

80%
60%
Tand 60 kN
40%
Tri 90 kN
20%
0%
0 100,000 200,000 300,000 400,000
Axle assembly passess

Figure 6.13: The same total freight task carried by both tandem and triaxles

Figure 6.13 shows that a freight task of around 177,000 t carried by a tandem axle
caused a surface texture reduction of 34% (to 66% of the original), whilst the same task
carried by a triaxle caused a surface texture reduction of 24% (to 76% of the original).

— 6-15 —
Chapter 6 Discussion of the model

Therefore under the experimental conditions, single axles and tandem axles caused
similar surface texture reductions for a given freight task, but a triaxle caused less
surface texture damage for a given freight task.

The experimental conditions, however, did not include significantly overloaded axles
and the associated peak vertical surface stresses that may occur under the tyre
sidewalls, which may occur in practice.

6.5.2 Comparisons of Equivalent Standard Axle surface wear


It is then of interest to move on to the general model, to see the results regarding
equivalent standard axles. Table 6.4 shows loads which cause the same pavement
damage under the 4th power equivalent standard axle concept, as discussed in Section
2.10.

Table 6.4: Loads which cause the same pavement damage

Loads which cause the Half axle loads which


Axle group type same damage as the standard load cause the same damage
(kN) (kN)
Single axle with dual tyres (SADT) 80 40
Tandem axle with dual tyres (TADT) 135 67.5
Triaxle with dual tyres (TRDT) 181 90.5
Quad-axle with dual tyres (QADT) 221 110.5

Adapted from Jameson (2012).

If a total freight task of 300,000 t is chosen, and the axles are loaded in accordance
with Table 6.4, then the SADT will require 382,268 passes, the TADT 226,519 passes,
the TRDT 168.950 passes and the QADT 138,371 passes to each carry the same load.

The general model can then be used to suggest the outcome shown in Figure 6.14.

— 6-16 —
Chapter 6 Discussion of the model

120%

100%

SPTD as % of original
80%

Sing 40 kN
60%
Tand 67.5 kN

40% Tri 90.5 kN


Quad 110.5 kN
20%

0%
0 100,000 200,000 300,000 400,000
Axle assembly passes

Figure 6.14: Result of a 300,000 t freight task carried by each equivalent standard axle

In this case it is again suggested that single axles and tandem axles cause the same
damage as each other, and that triaxles cause less damage than single and tandem
axles, and that quad axles cause less damage than single, tandem and triaxles.

The actual values at the end points where the 300,000 t has been accumulated, are
shown in Table 6.5, but expressed in terms of a percentage reduction rather than as a
percentage remaining of original.

Table 6.5: Reduction in texture depth for ESAs carrying the same freight task

Axle group type Reduction in SPTD (%))


Single axle with dual tyres (SADT) 32
Tandem axle with dual tyres (TADT) 32
Triaxle with dual tyres (TRDT) 23
Quad-axle with dual tyres (QADT) 14

This is an interesting outcome. It suggests that for axle groups travelling in a straight
line and not applying tractive motion forces, for a given freight task triaxial groups and
quad axle groups will cause less surface wear than for single axles and tandem axles.

This outcome will be of significant interest in the development of performance based


standards for the freight industry, as described in 6.6.3.

— 6-17 —
Chapter 6 Discussion of the model

This finding, however, is in direct opposition to the conclusions of Adams and Kim
(2014), who concluded that ‘a short rest period in a high-traffic volume road retards the
recovery of the binder and, therefore, leads to more permanent changes in the MPD
(macrotexture)’.

Upon closer inspection it can be seen that Adams and Kim (2014) used American
‘equivalent wheel passes’ in their work, as discussed in 3.6.4. The FHWA vehicle class
axle factors (Table 3.5) appear to be based on the work of Schmoyer et al. (1998),
where American weigh-in-motion (WIM) data were used to convert average annual
daily traffic into average annual daily loads for the American fleet. These loads are
expressed in equivalent standard axle loads (ESALs) and this implies the application of
a load damage exponent, most likely a 4th power. However, this Australian study using
ALF provides data to strongly suggest that ESALs relating only to pavement design,
and thus loadings at depth, do not relate to the surface texture deterioration of a chip
seal.

6.6 Applications for road owners/managers


6.6.1 Advantages of preserving surface texture
It is well accepted amongst road owners and managers that surface texture
(macrotexture) is vital in providing resistance to skidding and a safe riding surface for
traffic (for example AASHTO 2008; PIARC 1987; Roe et al. 1998).

Skid resistance policies vary around the world (Neaylon 2011), with the New Zealand
national road jurisdiction being one that is highly focused on skid resistance. In New
Zealand, road maintenance activities are strongly driven by the results of skid
resistance testing and skid resistance renewal (NZTA 2010; NZTA 2013). However,
even for jurisdictions that have different weightings to the various inputs that trigger a
maintenance reseal, surface texture (macrotexture) weights very highly.

The advantage of prolonging the useful life of surface texture would be either to gain a
better coverage of the network for a given budget, or to maintain the existing coverage
for a reduced budget.

6.6.2 Heavy vehicle seal design


The work undertaken previously towards incorporating heavy vehicles into the
Austroads sprayed seal design method (Khoo 2014) can now be reviewed.

— 6-18 —
Chapter 6 Discussion of the model

In that document, a damage factor DF was calculated for each class of vehicle using
equation 6.2 and a load damage exponent of 2.7.

𝑛𝑛 2.7
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑜𝑜𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 𝑖𝑖 6.2
𝐷𝐷𝐷𝐷 = � 𝑖𝑖 � �
𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑓𝑓𝑓𝑓𝑓𝑓 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑖𝑖
1

where

𝐷𝐷𝐷𝐷 = 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑓𝑓𝑓𝑓𝑓𝑓 𝑒𝑒𝑒𝑒𝑒𝑒ℎ 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑜𝑜𝑜𝑜 ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖.

Once the damage factor for each individual class of vehicle has been calculated, the
equivalent heavy vehicle damage factor can be calculated from equation 6.3.

𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑓𝑓𝑓𝑓 𝑎𝑎𝑎𝑎 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 6.3


𝐸𝐸𝐸𝐸𝐸𝐸𝐷𝐷𝐷𝐷 =
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡ℎ𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 4 (𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟)𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖

where

𝐸𝐸𝐸𝐸𝐸𝐸𝐷𝐷𝐷𝐷 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑣𝑣𝑣𝑣ℎ𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑜𝑜𝑜𝑜 𝑡𝑡ℎ𝑒𝑒 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝐷𝐷𝐷𝐷.

The load damage exponent of 2.7 was taken from Arnold et al. (2005c) as the best
available information at the time. A new damage factor for each class of heavy vehicle
can now be calculated by replacing the exponent of 2.7 with 1. This will result in the
seal design method being revised with new values of EHVDF, replacing the existing
ones shown in Table 6.6.

Table 6.6: Existing equivalent heavy vehicle damage factors


Austroads Class Australian national average
Fleet composition EHVDF
(%)
3 25.9 1.0
4 9.1 1.0
5 1.3 1.3
6 1.5 0.8
7 4.3 1.1
8 6.8 1.8
9 36.3 2.6
10 7.6 4.1
11 4.5 3.6
12 2.8 4.4

— 6-19 —
Chapter 6 Discussion of the model

An example of the interpretation of Table 6.6 is that by using the old exponent of 2.7, a
class 6 vehicle causes 0.8 times the damage of the reference class 4, whilst a class 12
vehicle causes 4.4 times the damage.

A desk top review of several different heavy vehicle design scenarios should be
undertaken as future work to determine what difference a new damage exponent would
make to the current design method, and whether any such differences would be
significantly important enough to instigate a change in design practice. In other words,
is the heavy vehicle seal design method sensitive to this new load damage exponent?

If so, it may also be necessary to undertake a field validation trial to compare the old
method with the revised design method, on a road where there are significant numbers
of equivalent heavy vehicles.

6.6.3 Regulation gaps for performance-based vehicles


In May 2001 the Australian Transport Council endorsed a proposal for the development
of a performance based standard (PBS) approach to heavy vehicle regulation (NTC
2005). The aim of PBS is to promote innovation through the adoption of a performance
based system of heavy vehicle regulation, whilst at the same time protecting road
infrastructure. As with all performance based specifications, the PBS approach does
not prescribe inputs (such as number of axles or the length of trailers) but rather
focusses on an innovative vehicle that meets prescribed outcomes.

The initial discussion document included comment on pavement horizontal loading,


including:

The horizontal forces imposed on a pavement are the longitudinal tyre


forces caused by the drive axle or axles when a vehicle starts,
accelerates or climbs an uphill grade and the lateral or side forces that
are generated by multi-axle groups during a turning manoeuvre.

Traditionally, wear of surfacings has not been treated separately from


pavement wear, and has been implicitly assumed to be part of the
wear attributed to 4th power Equivalent Standard Axles (ESAs).
However, there are growing concerns that the horizontal forces
generated by heavy vehicles could be causing excessive wear for
bituminous chip-seal pavements. Horizontal forces have a direct
influence on pavement wear and are particularly important for
bituminous chip-seal pavements in areas of road-train operation.

— 6-20 —
Chapter 6 Discussion of the model

It is the intention that the standard will cover the principle that
pavement wear for PBS vehicles for a particular freight task will be no
greater than for the same task being performed by conventional
vehicles. However, until further work has been undertaken to establish
an acceptable method of limiting horizontal tyre forces, a prescriptive
requirement has been adopted (NTC 2005).

The NTC published vehicle assessment rules in 2007 (NTC 2007) which were
approved by the Australian Transport Council. Due to lack of work in the area of
pavement horizontal loading, a performance standard could not be applied and a
prescriptive requirement was adopted ‘until further work has been undertaken’.

The take-up of the PBS option was slow for a number of reasons, and by June 2009
there were only 52 PBS approved vehicles (NTC 2009). The NTC review (2009)
addressed many of these issues, and whilst there was ongoing debate on the topic of
vertical loading with the Austroads pavement task force, the topic of horizontal loading
appears to have been left for another time.

A National Heavy Vehicle Regulator (NHVR) has since been established to administer
rules under the heavy vehicle national law, applying to heavy vehicles over 4.5 t gross
mass (NHVR 2014). It still appears that only heavy vehicle vertical loadings have been
addressed. In the meantime, Austroads has a research project to gather data on
heavy vehicle horizontal loadings. This involves the use of purpose built experimental
equipment developed in Australia, and is described in Patrick and Wright (2014).

Whilst this study does not address the issue of

forces caused by the drive axle or axles when a vehicle starts,


accelerates or climbs an uphill grade and the lateral or side forces that
are generated by multi-axle groups during a turning manoeuvre (NTC
2005)

this study does add value as it is relevant to the debate over sprayed seal wear being
implicitly assumed to be part of the wear attributed to 4th power equivalent standard
axles (ESAs).

This study is also relevant in that it suggests that for axle groups travelling in a straight
line and not applying tractive motion forces, for any given freight task triaxial groups
and quad axle groups will cause less sprayed seal wear than single axles and tandem
axles.

— 6-21 —
Chapter 6 Discussion of the model

It is hoped that this work can further assist the understanding and development of
performance based specifications.

6.7 Conclusions
This chapter has taken the model developed in the previous chapter, and checked
whether the outputs from it could be explained and pass the test of ‘reasonableness’.

This then led to a consideration of the time-temperature-frequency of loading


dependant viscoelastic properties of bitumen, and how this may either hinder or assist
– or both – aggregate reorientation.

The issue of rotation of shear stresses at the surface was identified as a future
research topic.

Once the model could be assessed as being plausible, it was then used to ‘say
something beyond the set of observations that spawned it’ (Section 5.8) in the final
stage of the process in testing the research hypothesis.

It was found that for this experimental set of conditions and for axle groups travelling in
a straight line and not applying tractive motion forces:

1. When comparing total freight task, single axles and tandem axles caused
similar surface texture reductions for a given load, but the triaxle caused less
spray seal surface wear than either of those for a given load.

2. When comparing equivalent standard axle loads for a given freight task, triaxial
groups and quad axle groups will cause less spray seal surface wear than
single axles and tandem axles.

In summary then, having more triaxles on the sprayed seal network, in straight line
travel situations, has the potential to cause less sprayed seal wear than the equivalent
load carried by single axles or tandem axles.

Whether all these axles and loading combinations would ultimately merge on a
common asymptote for surface texture over a very long period of time, would require
extrapolation well beyond the range of data currently collected, or further
experimentation.

— 6-22 —
Chapter 7 Summary and conclusions

7 SUMMARY AND CONCLUSIONS


7.1 Introduction
Questions raised during sprayed sealing design and consultancy revealed no known
answers, and led to a hypothesis that had never been tested:

Heavy vehicle axle group loads that cause the same pavement damage do not
necessarily cause the same sprayed seal damage.

This hypothesis has spurred an interesting journey, and much has been learned along
the way. Firstly, data was needed. A number of difficulties had to be overcome in the
collection of the data, such as:

• The experimental design was associated with a larger test and this work was not
the main objective.

• Preparation of the test pavement surface was not ideal.

• The design of the experimental factorial matrix was limited by time and cost
considerations.

• Although data was to be collected in triplicate, in many cases only duplicate data
survived the experimental process.

• Of the methods established to measure macrotexture, both the


photogrammetric method and the laser system adopted were found wanting,
but testing could not be paused due to the experiment being part of a larger
test. This left the least accurate - but industry standard of the time - volumetric
sand patch testing.

In spite of these difficulties - which upon reflection are part of daily life in the world of
construction engineering - steps were taken to filter and manage the data so that what
remained was both unique to researchers, and useable.

Skills in statistical analysis and multivariate non-linear regression were then developed,
resulting in the determination of a generalised model that was able to be convincingly
verified using data from an independent experiment from a different sprayed-seal-using
country.

— 7-1 —
Chapter 7 Summary and conclusions

The resulting generalised model was checked for interpretability and also plausibility,
and then finally used to address the research hypothesis.

7.2 The effect of heavy vehicle axle groupings on sprayed


seal wear
It was found that for this experimental set of conditions, and for axle groups travelling in
a straight line and not applying tractive motion forces, that

1. Heavy vehicle axle groupings, loaded so that they cause the same pavement
damage (defined as vertical deformation), cause disproportional sprayed seal
damage (defined as surface texture reduction).

2. When comparing total freight task, single axles and tandem axles cause similar
surface texture reductions for a given load, but a triaxle causes less sprayed
seal surface wear than either of those two for a given total payload.

3. When comparing equivalent standard axle loadings and for a given freight task,
triaxial groups and quad axle groups will cause less sprayed seal surface wear
than for single axles and tandem axles.

In summary then, having more triaxles on the sprayed seal network in straight line
travel situations has the potential to cause less sprayed seal wear than the equivalent
payload load carried by single axles or tandem axles.

7.3 The effect of heavy vehicle axle groupings on sprayed


seal design
Most sprayed seal design methods have evolved over many years from Frederick
Melrose Hanson’s landmark mathematical approach in 1935 (Hanson 1935; IPENZ
1979).

However, traffic loadings have increased significantly since then, particularly in


Australia with freight efficiency resulting in longer and heavier loads with more triaxial
groupings being transported by more powerful prime movers.

The literature review has shown that the Australian seal design method for heavy
vehicles is currently more detailed and more focused on heavy vehicles than any other
country.

— 7-2 —
Chapter 7 Summary and conclusions

Using the outcomes of this study the traffic component of seal design can now be
refined to the same extent that the traffic component of pavement design is refined, as
the relative damage caused by the design fleet axle loads and axle groupings can be
more confidently calculated.

Whether or not this refinement will result in the same seal design outcome or a different
seal design outcome, requires future work.

7.4 A load damage component for sprayed seal wear


The derivation of load damage exponents has been examined in the literature review,
and again in the data analysis. The legend behind the 4th power law has been
researched, and found to be becoming increasingly at risk of fitting the definition of
legend – ‘a traditional story popularly regarded as historical’ (Oxford 1987). Current
research work underway world-wide is supporting the need for more scrutiny.

A load damage exponent of approximately 1 has been determined for the seal and
pavement construction used in this study. This implies a linear relationship meaning
when the load is doubled, the resulting wear is doubled. This is very different to the
unbound granular pavements that lie beneath the seals, where if the load is doubled
the damage is increased 24 = 16 times (again, if the 4th power law is indeed applicable).

However, the literature review suggests that load damage exponents are very sensitive
to the types of materials used and the quality of construction (i.e. compaction).

It can now be concluded that the load damage exponent for sprayed seals is not the
same as for pavements - it is much lower.

7.5 Future work


This study has highlighted several areas that would benefit from further research.

7.5.1 Improvements to the model


A fresh analysis of the data by a specialist mathematician could well extract a more
direct interpretation of the data into a single, streamlined function. However, a
mathematician would still require access to engineering experience for an
understanding of the impact that any mathematical assumptions made may have on
the engineering understanding of reality.

— 7-3 —
Chapter 7 Summary and conclusions

The model could also be improved by a mathematician able to assess data for outliers
in a more sophisticated manner, for example a) by determining data outliers by setting
probabilities for false discovery rates after assuming the data follows a Lorentzian
distribution (as per. Motulsky & Brown 2006), or b) by calculating the Mahalanobis
distances of suspected outlier data points specifically in nonlinear regressions.

The model is based on data that is derived from the operator dependant sand patch
test, a test first formalised nearly 50 years ago by the Road Research Laboratory
(1969) and now withdrawn from the International Standards Organisation (Sections
2.7.4 and 2.7.5). An improved model should be based on data sourced by state-of-the-
art, repeatable and operator-independent laser technology.

7.5.2 Finite element modelling (FEM) analysis


When this experimental work commenced in 2009, Finite element models of sprayed
seals at the time were based on some very crude assumptions and did not take into
account aggregate shape or aggregate reorientation under trafficking. A decision was
made for this work to measure actual performance under an accelerated pavement
tester, rather than modelled performance. In this route the variability of tyre contact
pressures, cohesive and adhesive strength of the seal, type of bitumen, etc where all
held constant. However finite element models have advanced significantly in since
then. For example Xiao (2013) developed finite element models of an antiskid
surfacing for an airfield runway that included the viscoelastic properties of the binder,
maximum principle stress and shear strain levels, recovery after loading, and
aggregate shapes that more closely matched reality.

This type of analysis would be strongly beneficial in addressing the research gaps
identified in Section 6.4 explanation of the model outcome.

7.5.3 The use of cutters for aggregate wetting


More recent research into cutter effectiveness has been based on polymer modified
binders, where the hypothesis of polymer modified binders (PMBs) requiring a higher
cutter content than Class 170 bitumen has been tested (PMBs are described in section
2.4.4). This work started with Maccarrone et al. (1997), Oliver (2011), and more
recently with Urquhart (2012). Urquhart concluded however that binder viscosity is not
the only factor that controls the wetting of aggregates - the other main factor is likely to
be binder surface tension, as this property controls the degree of wetting of low
viscosity liquids on solid surfaces. Further research is required in this area.

— 7-4 —
Chapter 7 Summary and conclusions

7.5.4 Seal design


Short term
a) At the detailed level, a new damage factor for each class of heavy vehicle can
now be calculated by replacing the currently assumed exponent of 2.7 with 1.
This will result in the seal design method being revised with new values of
EHVDF. A desk top review of several different heavy vehicle design scenarios
should be undertaken to determine what difference a new damage exponent
would make to the current design method, and whether any such differences
would be significantly important enough to instigate a change in design practice.
If so, it may also be necessary to undertake a field validation trial to compare
the old method with the revised design method, on a road where there are
significant numbers of equivalent heavy vehicles.
b) The MSCR test result has also recently been found to correlate well to
resistance to shear loads in asphalt (White 2015a, 2015b), which suggests a
potential may exist of a correlation to resistance of aggregate reorientation in
sprayed seals. This could be examined relatively easily.

Long term
a) Another opportunity to repeat accelerated loading facility work of this kind, with
the surface texture measured by a laser device, would provide data that is more
precise. The involvement of a specialist mathematician/modeller could also
produce a model with greater refinement.

b) At a philosophical level, there remain some interesting questions. Should the


traditional use of average daily traffic, which is often swamped by the effect of
passenger cars, be retained? What influence do passenger cars have on the
deterioration of skid resistance (as measured by macrotexture) compared to
commercial vehicles? Should seal design be based entirely on the effect of
heavy vehicles? If so, could such a transition be managed?

c) One practical solution for this would be to develop a two-tier seal design
method (for Australia as a first step): firstly, a simple, proven method for light
traffic and low risk, based on average daily traffic; and then a more complicated
supplement, which would add specific detail and more complex considerations
(such as converting the HV count into EHVs) when required, for higher risk
designs. This type of approach has already been adopted for asphalt design
(Rebbechi & Petho 2014).

— 7-5 —
Chapter 7 Summary and conclusions

d) There are many different seal design methods in the world. Could the best
features of each country’s seal design method be distilled, and one international
method be postulated? What effects do environmental and cultural differences
have in seal design? There are a number of issues here which would benefit
from further research but are beyond the scope of this present study.

e) As road pavements and surfacings are loaded by moving wheel loads, the
magnitudes of the vertical, horizontal and shear stresses will vary as the applied
principal stress rotates (Refer Fig. 6.5). The degree that it is possible to take
this pavement theory of stress rotation at depth, to a surfacing application and
chip seal aggregate rotation, requires further investigation.

f) The results of this study were obtained from the experimental set of conditions
of axle groups travelling in a straight line and not applying tractive motion forces
nor the shear forces of axle groupings ‘scuffing’ the surface in short radii turns.
This is the final half of the complete picture, and would benefit from continued
research, particularly in the context of performance based specifications (PBS).

7.6 Conclusions
The hypothesis of this study, that

heavy vehicle axle group loads that cause the same pavement damage do not
necessarily cause the same sprayed seal damage,

has been found to be supported under the experimental conditions of this study.

This study has now provided a documented argument against adopting any existing
equivalent standard axle data derived for pavement design use, and using this as a
seal design input.

By re-defining damage to being more applicable to sprayed seal wear, a load damage
exponent for sprayed seals has been found to be closer to 1 than to the 4 used for
unbound granular pavements that sprayed seals generally are used on, and the 2.7
previously assumed for sprayed seals.

— 7-6 —
Chapter 7 Summary and conclusions

The study has found several gaps that would benefit from further research. The growth
of freight efficient road trains is likely to continue in Australia: for example, the Western
Australian Minister of Transport announced that even longer road trains have been
approved for the Pilbara (Nalder 2015). A deeper understanding of their effects on
sprayed seals will enable such seals to be engineered to withstand these effects.

This study was limited to straight line loading with no cornering involved, and with no
significant tractive effort applied as happens when a prime mover hauls a loaded road
train up an incline. It would be very interesting to research these other effects further.

— 7-7 —
8 References

8 REFERENCES
AAPA 2011, ‘Study tour 2011 - South Africa and CAPSA'11 - group report’. v5-1,
Australian Asphalt Pavement Association, Kew, Victoria.

AASHTO 2005, ‘Performance-graded asphalt binder’, Standard Specification M 320,


American Association of State Highway and Transportation Officials, Washington.

AASHTO 2008, ‘Guide for pavement friction’, Guide, Joint technical committee on
pavements, American Association of State Highway and Transportation Officials,
Washington.

AASHTO 2010, ‘Performance-graded asphalt binder’, Standard Specification M 320,


American Association of State Highway and Transportation Officials, Washington.

AASHTO 2012, ‘Manual for emulsion-based chip seals for pavement preservation’,
Single user digital publication MEBC-1-UL, American Association of State
Highway and Transportation Officials, Washington.

AASHTO 2014a, ‘Multiple stress creep recovery (MSCR) test of asphalt binder using a
dynamic shear rheometer (DSR)’, Standard Method of Test T 350-14, American
Association of State Highway and Transportation Officials, Washington.

AASHTO 2014b, ‘Performance graded asphalt binder using multiple stress creep
recovery (MSCR) test’, Standard Specification M 332-14, American Association
of State Highway and Transportation Officials, Washington.

AASHTO 2015, ‘Determining asphalt binder bond strength by means of the binder
bond strength (BBS) test’, Standard method of test TP 91-15, American
Association of State Highway and Transportation Officials, Washington.

Adams, JM, & Kim, YR 2014, ‘Mean profile depth analysis of field and laboratory traffic-
loaded chip seal surface treatments’, International Journal of Pavement
Engineering, Vol 15 (Nos 7-8).

Akýllý, A, Saltan, M, Karaþahin M, & Gürer, C 2012, ‘Investigation of adhesion


properties in chip seals with pull out test’, A5EE-325, Eurasphalt & Eurobitume
Congress, 5th, Istanbul.

— 8-1 —
8 References

Albert, I & Mafart, P 2005, ‘A modified Weibull model for bacterial inactivation’,
International Journal of Food Microbiology, 100(1–3).

Alderson, A 2004, ‘Sprayed seal design - 2003/2004: Summary’, Austroads Research


Report AP-R258-04, Austroads, Sydney.

Alderson, A 2006, ‘Update of the Austroads sprayed seal design method’, Austroads
Technical Report AP-T68-06, Austroads, Sydney.

Alderson, A, & Oliver, JHW 2008, ‘Seal distress mechanisms: an initial study into
flushing’, Austroads Technical Report AP-T108-08, Austroads, Sydney.

Al-Khateeb, G, Shenoy, A & Gibson, N 2007, ‘Mechanistic performance analyses of the


FHWA's Accelerated Loading Facility pavements’, Association of Asphalt Paving
Technologists, proceedings of the technical sessions May 2007, AAPT.

Amphora NDT Ltd. (2008). ‘Limpet’, http://www.amphorandt.com/cgi-


bin/display/products?s_category=&s_id=10 ,viewed 16 November 2008.

Arnold, G 2004, ‘Rutting of granular pavements’, PhD Thesis, University of Nottingham.

Arnold, G, Steven, B, Alabaster, D, & Fussell, A 2005a, ‘Effect on pavement wear of


increased mass limits for heavy vehicles – Stage 3’, Research Report 279, Land
Transport New Zealand, Wellington.

Arnold, G, Steven, B, Alabaster, D, & Fussell, A 2005b, ‘Effect on pavement wear of


increased mass limits for heavy vehicles – Stage 4’, Research Report 280, Land
Transport New Zealand, Wellington.

Arnold, G, Steven, B, Alabaster, D, & Fussell, A 2005c, ‘Effect on pavement wear of


increased mass limits for heavy vehicles – concluding report’, Research Report
281, Land Transport New Zealand. Wellington.

Arnold, G, & Werkmeister S 2010, ‘Pavement thickness design charts derived from a
rut depth finite element model’, Research Report 427, NZ Transport Agency,
Wellington.

ARRB Group, 2014, ‘Austroads-ARRB partnership agreement: programs and


achievements 2013-14’, ARRB, Vermont South, Victoria.

— 8-2 —
8 References

Asphalt Academy 2007, ‘The use of modified bituminous binders in road construction;
2nd ed.’ TG1, Pretoria, South Africa.

ASTM 1987, ‘Standard test method of measuring surface macrotexture depth using a
volumetric technique’, ASTM E965-87, American society for Testing and
Materials, West Conshohocken, PA.

ASTM 1988, ‘Specification for phenolic molding compounds’ (Withdrawn 1991), D700-
81, American Society for Testing and Materials, West Conshohocken, PA.

ASTM, 2009, ‘Pull-off strength of coatings using portable adhesion testers’, Standard
Test Method D4541-09e1, American Society for Testing and Materials, West
Conshohocken, PA.

ASTM 2011a, ‘3D imaging data exchange Version 1.0’, Standard Specification E2807-
11, American Society for Testing and Materials, West Conshohocken, PA.

ASTM 2011b, ‘ASTM 3D data imaging systems committee approves new data
exchange standard’, News release #8856, American Society for Testing and
Materials, West Conshohocken, PA.

ASTM 2015, ‘Measuring pavement macrotexture depth using a volumetric technique’,


Standard Test Method E965-15, American Society for Testing and Materials,
West Conshohocken, PA.

Australian Institute of Petroleum, 2004, ‘Supply security’,


http://www.aip.com.au/industry/supplysecurity.htm#lfss , viewed 19 Aug 2004.

Austroads 1990, ‘Design of sprayed seals’, Austroads, Sydney.

Austroads 2001, ‘Austroads provisional sprayed seal design method: Revision 2000’,
Austroads Technical Report, AP-T09-01, Austroads, Sydney.

Austroads 2002, ‘Practitioner's guide to design of sprayed seals: revision 2000


method’, Austroads Technical Report AP-T17-02, Austroads, Sydney.

Austroads 2005a, ‘Guidelines for the management of road surface skid resistance’,
Austroads Guideline AP-G83-05, Austroads, Sydney.

— 8-3 —
8 References

Austroads 2005b, ‘Sprayed seal cutting practice’, Austroads Technical Report AP-T39-
05, Austroads, Sydney.

Austroads 2007, ‘Pavement surface texture measurement with a laser profilometer’,


Austroads Test Method AGAM-T013R, Austroads, Sydney.

Austroads 2008, ‘Modified surface texture depth (pestle method)’, Austroads Test
Method AGPT-T250, Austroads, Sydney.

Austroads 2010, ‘Ball penetration test’, Austroads Test Method AGPT-T251,


Austroads, Sydney

Austroads 2013, ‘Guide to the Selection and use of Polymer Modified Binders and
Multigrade Bitumens’, Austroads Technical Report AP-T235-13, Austroads,
Sydney.

Austroads 2014, ‘Austroads glossary of terms (5th edition)’, AP-C87-14, Austroads,


Sydney.

Austroads 2015, ‘Austroads glossary of terms (6th edition)’, AP-C87-15, Austroads,


Sydney.

Austroads & AAPA 2000, ‘Sprayed sealing - aggregate precoating’, Pavement worktip
No 23, http://www.aapa.asn.au/documents/item/76 viewed 14/01/2016.

Ball, GFA 1992, ‘Weathering of New Zealand chipseal binders: investigations from
June 1991 to May 1992’, Central Laboratories Report 92-26323.06, Opus,
Wellington.

Baburamani, PS, Gaughan, RL, Leach, RD, & Holtrop, WP 1999, ‘A summary of
Austroads sprayed seal design method draft 7A – October 1997’, APRG 99/09,
ARRB Transport Research, Vermont South, Victoria.

Baburamani, PS, Holtrop, W, & Gaughan, R 2000, ‘Strategy for monitoring sprayed
seal field trials’, Contract report RC91024-B, ARRB, Melbourne.

Bahia, HU, Jenkins, KJ, & Hanz, A 2008a, ‘Performance grading of bitumen emulsions
for sprayed seals’, Mid-continent transportation research forum 2008, Wisconsin
Highway Research Program, University of Wisconsin Maddison, Wisconsin, USA.

— 8-4 —
8 References

Bahia, HU, Jenkins, KJ, & Hanz, A 2008b, ‘Performance grading of bitumen emulsions
for sprayed seals’, 1st Sprayed Sealing Conference, Adelaide, ARRB, Vermont
South, Victoria.

Bennett, D, Troutbeck, R, Croft, P, McTiernan, D, & Shah, A 2009, ‘Guide to traffic


management part 3: traffic studies and analysis’, Guide to Traffic Management,
AGTM03-09, Austroads, Sydney.

Benslimane, A 2004, ‘Caractérisation de textures rugueuses par traitement d'images:


application aux revêtements routiers’, PhD thesis, Université de Poitiers.

Benslimane, A, Khoudeir, M, Brochard, J, & Do, M.-T 2007, ‘Characterization of road


microtexture by means of image analysis’, Wear, Vol 264, Issue 5-6.

Benslimane, A, Khoudeir, M, Brochard, J, Legeay, V, & Do, M.-T 2003, ‘Relief


reconstruction of rough textured surface through image analysis’, Proceedings of
SPIE-IS&T Electronic Imaging, SPIE Vol 5011.

BITRE 2011, ‘Truck productivity: sources, trends and future prospects’, Research
Report 123, Bureau of Infrastructure Transport and Regional Economics,
Canberra.

Boral Resources 2008a, ‘Material test report BR3622200814749’, dated 17/09/08,


Montrose field laboratory, Boral Resources, Montrose, Victoria.

Boral Resources 2008b, ‘Material test report BR3622200814750’, dated 17/09/08,


Montrose field laboratory, Boral Resources, Montrose, Victoria.

Boral Resources 2008c, ‘Material test report BR3622200814751’, dated 17/09/08,


Montrose field laboratory, Boral Resources, Montrose, Victoria.

Boral resources 2016, ‘Boral Lysterfield Quarry’,


http://www.boral.com/article/Lysterfield_Resources_Products.asp viewed
12/11/16

le Bouteiller, É 2012, ‘Asphalt emulsions: why we could fail, why we will succeed’,
Asphalt Emulsion Manufacturers Association annual meeting 2012, Bonita
Springs, Florida.

— 8-5 —
8 References

Box, GEP, & Draper, NR 1987, ‘Empirical Model Building and Response Surfaces’,
John Wiley & Sons, New York, NY.

Box, GEP, Hunter, WG, & Hunter, JS 1978, ‘Statistics for experimenters - an
introduction to design, data analysis, and model building’, John Wiley & sons,
New York, NY.

BP Bitumen Australia 2014, ‘Bitumen basics’, 2nd Ed, BP Bitumen, Melbourne.

British Standards Institution 2010, ‘Measurement of pavement surface macrotexture


depth using a volumetric patch technique’, Road and airfield surface
characteristics, BS EN 13036-1:2010. BSI.

Cairney, P & Styles, E 2005. A pilot study of the relationship between macrotexture and
crash occurrence, Report No: CR 223, Australian Transport Safety Bureau,
Canberra.

Caltrans 2007, ‘Chipseal maintenance technical advisory guide’. Vol. 1 Ch.7. California
Department of Transportation, CA.

Cenek, PD & Patrick, JE 1995, ‘Ch. 9, Prediction of pavement texture’, Modelling road
deterioration and maintenance effects in HDM-4, Ltd., ND Lea International Ltd,
Asian Development Bank.

Cenek, PD & Griffith-Jones, G 1997, ‘Incremental forms for modelling texture depth
and friction parameters in HDM-4’, Communication to the ISOHDM, University of
Birmingham, Works Central Laboratories, Lower Hutt, New Zealand.

Chambers 1991, ‘Chambers Science and Technology Dictionary’, Walker PMB ed.,
Chambers Ltd, Edinburgh.

Chen, D-H, Zhou, F & Cortez, ER 2006, ‘Determination of load damage relationships
through accelerated pavement testing’, Journal of Testing and Evaluation, vol.
34, no. 4.

Chikhi, M &. Bignone F 2006, ‘Dense digital surface model and true ortho image allow
supporting versatile and innovative applications’, Asian conference on remote
sensing, 27th, Ulaanbaatar, Mongolia, Asian Association on Remote Sensing.

— 8-6 —
8 References

Clem, K 2010, ‘A short course on scientific writing’, Swinburne University of


Technology, Hawthorn, Victoria.

COLTO 1998, ‘Surfacing seals for rural and urban roads’, draft TRH3, technical
recommendations for highways, Committee of Land Transportation Officials,
Pretoria.

Cooper DR & Schindler PS 2003, ‘Business research methods’, 8th Ed, McGraw-Hill
Irwin, New York, NY.

CSIR Transportek 1984, ‘Measurement of the texture depth of a road surface’, TMH6
(draft) Method ST1, Special methods for testing roads, Council for Scientific and
Industrial research, Pretoria.

Cunningham, E. 2012, ‘Multivariate statistics: multiple regression and factor analysis


using SPSS’, Swinburne University of Technology, Hawthorn, Victoria.

D'Angelo, J. 2010, ‘New high-temperature binder specification using multistress creep


and recovery’, Developments in asphalt binder specifications, Transportation
Research Circular E-C147, Transportation Research Board, Washington, DC.

Denneman, E 2013, ‘Improved design procedures for asphalt pavements: pavement


temperature and load frequency estimation’, Austroads Technical Report AP-
T248-13, Austroads, Sydney.

de Beer, M 2015, ‘Road traffic: flow, capacity, loading, tyre contract stress’, South
African pavement design method revision - SARDS CAPSA Workshop, Sun City,
South African National Roads Agency Limited, South Africa.

de Beer, M & Maina, JW 2011, ‘Using tire-road contact stresses in road pavement
design and analysis’, 2011 Annual Review, Tire technology INTERNATIONAL,
UK.

de Beer, M, Maina, JW, van Rensburg, Y & Greben, JM 2012, ‘Toward using tire-road
contact stresses in pavement design and analysis’, Tire Science and Technology,
Vol 40, No. 4.

— 8-7 —
8 References

de Pont, J, Steven, B & Alabaster, D 2002, ‘The effect of mass limit changes on thin-
surface pavement performance’, International symposium on heavy vehicle
weights & dimensions, 7th, Delft, The Netherlands, Delft Technical University,
The Netherlands.

de Pont, J, Steven, B, Alabaster, D, Fussell, A 2003, ‘Effect on pavement wear of an


increase in mass limits for heavy vehicles – stage 2’, Transfund New Zealand
Research Report No. 231, Transfund New Zealand, Wellington.

Dickinson, EJ 1985, ‘Bituminous Roads in Australia’, Australian Road Research Board,


Vermont South, Victoria.

Dickinson, EJ 1989, ‘The rate of evaporation of volatile diluents (cutter) from cutback
bitumen sprayed seals’, ARRB Research Report, ARR 157, Australian Road
Research Board, Vermont South, Victoria.

Do, M-T 2005, ‘Relationship between microtexture and skid resistance’, Bulletin des
Laboratories des Ponts et Chaussees, Vol 255, Laboratoire Central des Ponts et
Chaussees, Nantes.

Donbavand, J. 1984, ‘An investigation into the use of thermogravimetry and derivative
thermogravimetry for the analysis of bituminous materials’, ARRB Conference,
12th, Hobart, ARRB, Vermont South, Victoria.

DTEI 2007, ‘Austroads vehicle classification system, culway executive summary’,


Department for Transport Energy and Infrastructure, Adelaide.

DTMR 2014, ‘Pull-out force for surfacing aggregate’, Test Method Q227, Department of
Transport and Main Roads, Brisbane.

Einarsson, IT 2009, ‘Chip Seals - Examination of design and construction in two


countries’, Master of Science in Civil Engineering Thesis, University of
Washington.

El Gendy, A. & Shalaby, A 2007, ‘Mean Profile depth of pavement surface


macrotexture using photometric stereo techniques’, Journal of transportation
engineering, Vol 133, No. 7.

Eos Systems Inc. 2009, ‘PhotoModler Scanner’, Vancouver, Canada.

— 8-8 —
8 References

Ethiopian Roads Authority 2013, ‘Best practice manual for thin bituminous surfacings’,
Roughton International, Addis Ababa.

FHWA 2011, ‘The multiple stress creep recovery (MSCR) procedure’, TechBrief
FHWA‐HIF‐11‐038, Office of pavement technology, Federal Highway
Administration, Washington, DC.

FHWA 2013, ‘Traffic monitoring guide’, FHWA-PL-13-105, Office of highway policy


information, Federal Highway Administration, Washington, DC.

Field, AP 2009, ‘Discovering statistics using SPSS’, SAGE, London.

Flintsch, GW, de Leon, E, McGhee, KK, & Al-Qadi, IL 2003, ‘Pavement surface macro-
texture measurement and application’, Transportation Research Record 1860,
TRB, Washington, DC.

Flintsch, GW, de León Izeppi, ED, Saleh, MI, & McGhee, KK 2008, ‘Area-based
macrotexture measurements using stereo vision’, SURF 2008. Slovenia, World
Road Association, Paris.

Francis, G 2012, ‘Analysis of variance and interaction’, Pearson Australia, Frenchs


Forest, NSW.

Georgopoulos, A, Loizos, A & Fiouda, A 1995, ‘Digital Image processing as a tool for
pavement distress evaluation’, ISPRS Journal of Photogrammetry and remote
sensing, Vol 50, Issue 1.

Gerber, J & Jenkins, K 2015, ‘Damage quantification of bituminous surfacing seals with
a finite element modelling approach’, International Society for Asphalt Pavements
(ISAP) Asphalt Pavement and Environment (APE) Symposium, 3rd, Sun City,
South Africa.

Gillespie, TD, Karamihas, SM, Sayers, MW, Nasim, MA, Hansen, W, Ehsan, N, &
Cebon, D 1993, ‘Effects of heavy-vehicle characteristics on pavement response
and performance’, NCHRP Report 353, Transportation Research Board,
Washington, DC.

— 8-9 —
8 References

Gillmann, R 1999, ‘Axle spacing and load equivalency factors’, Transportation


Research Record: Journal of the Transportation Research Board, Vol 1655,
(Pavement Management Systems and Other Pavement Design Issues),
Transportation Research Board, Washington, DC.

Gonzales, A, Bodin, D, Jameson, G, Oeser, M, & Vuong, B 2012, ‘Development of a


Nonlinear Finite Element Pavement Response to Load Model’, Austroads
Technical Report, AP-T199-12, Austroads, Sydney.

Gorman, JL 2000, ‘The role of colloid science in bitumen emulsion-mineral aggregate


interactions’, PhD thesis, Swinburne University of Technology, Melbourne.

Gorman, JL, Crawford, RJ & Harding, IH 2004, ‘Bitumen emulsions in road


construction: a review’, Road and Transport Research, Vol13, Issue1.

Gransberg, DD, & James, DBM 2005, ‘Chip seal best practices’, NCHRP Synthesis
342, Transportation Research Board, Washington, DC.

Gransberg, DD, Karaca, I, & Senadheera, S 2004, ‘Calculating Roller Requirements for
Chip Seal Projects’, Journal of construction engineering and management,
Volume 130 Issue 3.

Greene, WH 1990, ‘Econometric analysis’, Macmillan Publishing Company, New York,


NY.

Griffith, A, & Hunt, E 2000, ‘Asphalt cement chip seals in Oregon’, State research
project, Oregon Department of Transportation, Salem.

Hagos, ET 2008, ‘The Effect of Aging on Binder Properties of Porous Asphalt


Concrete’, Master of science thesis, Technische Universiteit, Delft.

Hajek, JJ, & Agarwal, A, 1990, ‘Influence of axle group spacing on pavement damage’,
Transportation Research Record: Journal of the Transportation Research Board,
No 1286.

Hajek, JJ, Selezneva, OI, Mladenovic, G, & Jiang, YJ 2005, ‘Estimating cumulative
traffic loads, Volume II: Traffic Data Assessment and Axle Load Projection for the
Sites with Acceptable Axle Weight Data, Final Report for Phase 2’, FHWA-RD-
03-094, Federal Highway Administration. McLean, VA.

— 8-10 —
8 References

Hanson, FM, 1935, ‘The bituminous surface treatment of rural highways’, NZ Society of
Civil Engineers, Vol 21: pp 89-178.Wellington, NZ.

Hanz, AJ, Johannes, P, & Bahia, H 2012, ‘Development of emulsion residue testing
framework for improved chip seal performance’, Transportation Research
Record: Journal of the Transportation Research Board, No. 2293, Washington,
DC.

Hanz, AJ, Johannes, P, Chaturabong, P, & Bahia, H 2013, ‘Development of a more


rational system for selecting emulsions for surface treatments’, Transportation
Research Circular E-C182, Transportation Research Board, Washington D.C.

Hefer, AW, & Little DN 2005, ‘Adhesion in bitumen-aggregate systems and


quantification of the effects of water on the adhesive bond’, ICAR Research
report. ICAR –505-1, International centre for aggregate research, Alexandria, VA.

Henning, TFP 2008, ‘The development of pavement deterioration models on the state
highway network of New Zealand’, PhD Thesis, University of Auckland, Auckland.

Henning, TFP, Costello, SB, & Watson, TG, 2006, ‘A review of the HDM/dTIMS
pavement models based on calibration site data’, Land Transport New Zealand
Research Report 303, Wellington.

Herrington, PR, & Ball, GFA 1994,’ Weathering of New Zealand chipseal binders:
investigation from June 1993 to April 1994’, Transit New Zealand Research
Report PR3-0105, Wellington.

Herrington, PR, Ball, GFA, Patrick, J, & Towler, J 2012, ‘Aggregate breakdown as a
cause of chip seal flushing’, ARRB Conference, 25th, Perth, ARRB Group,
Vermont South, Victoria.

Herrington, PR, Kodippily, S, & Henning, TFP 2015, ‘Flushing in Chip seals’, New
Zealand Transport Agency Research Report 576, NZ Transport Agency,
Wellington.

Herrington, PR & Wu, J 2016, ‘Performance based specification for chip seal bitumen’,
Discussion Paper. DRAFT 14 Jan 2016, Opus Research, Lower Hutt, NZ.

— 8-11 —
8 References

Highways Agency, Scottish Office Development Department, & Welsh Office and
Department of the Environment for Northern Ireland 1999, ‘Bituminous surfacings
materials and techniques. Design Manual for Roads and Bridges’, Vol 7 Sect 5
Part 2 HD 37/99 Ammdt. 1, Department for Transport, London, UK.

Hillier, P, & Soet, W 2009, ‘Guide to asset management part 5G: texture’, Austroads
Guide to Asset Management, AGAM05G-09, Austroads, Sydney.

Hitch, L 1981, ‘Surface dressing in developing countries: research in Kenya’, TRRL


Laboratory Report 1019, Transport and Road Research Laboratory, Crowthorne,
UK.

Holtrop, W 2008, ‘Sprayed sealing practice in Australia’, International sprayed sealing


conference, 1st, Adelaide, ARRB Group, Vermont South.

Holtrop, W, & Moffatt, MA 2008, ‘Design of sprayed seal suitable for accelerated
pavement testing’, International Sprayed Sealing Conference, 1st, Adelaide,
ARRB, Vermont South.

Homsi, F 2011, ‘Endommagement de chaussées sous chargement multi-essieux


(Damage to roads by multi-axle loads)’, PhD Thesis de Doctorat, University de
Nantes, Nantes, France.

Homsi, F, Bodin, D, Yotte, S, Breysse, D, & Balaya, J-M 2011, ‘Multiple axle loadings:
Shape parameters and their effect on the fatigue life of asphalt mixtures’,
European Journal of Environmental and Civil Engineering, Vol 15, issue 5.

Houghton, L, & Hallett, J 1987, ‘An analysis of single coat seal designs’, New Zealand
Roading Symposium, Vol. 2, Pavement Design and Maintenance, National
Roads Board, Wellington.

Hudson, KC, Saunders, LR, Nicholls, KF, & Hambleton, PH 1986, ‘Rolling of chip
seals’, ARRB Conference, 13th, Adelaide, Australian Road Research Board,
Vermont South, Victoria.

Huurman, M 2010, ‘Developments in 3D surfacing seals FE modelling’, International


Journal of Pavement Engineering, Vol. 11, No. 1, February 2010.

IBM Corporation, 2012, ‘SPSS Statistics’, version 21, IBM Software, Armonk, New
York.

— 8-12 —
8 References

IPENZ 1979, FM Hanson obituary, Institute of Professional Engineers New Zealand,


Wellington.

Irick, PE & Hudson, WR 1964, ‘Guidelines for satellite studies of pavement


performance’, NCHRP Report No 2, Highway Research Board, Washington DC.

ISO 1994, ‘Acoustics - Specification of test tracks for measuring noise emitted by road
vehicles and their tyres’, ISO 10844:1994, International Standards Organisation,
Geneva, Switzerland.

ISO 1997, ‘Part 1: Determination of mean profile depth’, Characterization of pavement


texture by use of surface profiles, ISO 13473-1:1997, International Standards
Organisation, Geneva, Switzerland.

ISO, 2002a, ‘Part 2: Terminology and basic requirements related to pavement texture
profile analysis’, Characterisation of pavement texture by use of surface profiles,
ISO 13473-2:2002, International Standards Organisation, Geneva, Switzerland.

ISO 2002b, ‘Part 3: Specification and classification of profilometers’, Characterisation


of pavement texture by use of surface profiles, ISO 13473-3:2002, International
Standards Organisation, Geneva, Switzerland.

ISO 2011, ‘Specification of test tracks for measuring noise emitted by road vehicles
and their tyres’, ISO 10844:2011, International Standards Organisation Geneva,
Switzerland.

Jackson, DC, Jackson, NC & Mahoney JP 1990, ‘Washington State chipseal study’,
Transportation Research Record: Journal of the Transportation Research Board
1259: 1–10, Washington, DC..

Jackson, G 1963, ‘Surface Dressing’, Shell International Petroleum Company Ltd,


London.

Jannat, GE, Yuan, XX & Shehata, M 2014, ‘Development of regression equations for
local calibration of rutting and IRI as predicted by the MEPDG models for flexible
pavements using Ontario's long-term PMS data’, International Journal of
Pavement Engineering, Vol 17, Issue 2.

Jameson, G 2011, ‘Pavement evaluation and treatment design’, AGPT05-11,


Austroads guide to pavement technology, Austroads, Sydney.

— 8-13 —
8 References

Jameson, G, 2012, ‘Pavement structural design’, AGPT02-12, Austroads Guide to


pavement technology, Austroads, Sydney.

Jameson, G, 2013, ‘Technical basis of Austroads guide to pavement technology Part 2:


pavement structural design’, Research Report ARR 384, ARRB Group Ltd,
Vermont South, Victoria.

Ji, X, Zheng, N, Niu, S, Meng, S, & Xu, Q 2015, ‘Development of a rutting prediction
model for asphalt pavements with the use of an accelerated loading facility’,
Road Materials and Pavement Design, DOI 10.1080/14680629.2015.1055337.

Jones, D, Harvey, J, Al-Qadi, IL, & Mateos, A, 2012, ‘Advances in pavement design
through full-scale accelerated pavement testing’, Proceedings of the 4th
international conference on accelerated pavement testing, Davis, CA, USA, 19–
21 September 2012, CRC Press, Leiden, The Netherlands.

Kassem, E, Awed, A, Masad EA, & Little DN 2013, ‘Development of predictive model
for skid loss of asphalt pavements’, Transportation Research Record: Journal of
the Transportation Research Board, No. 2372, (Asphalt Materials and Mixtures
2013, Vol. 3).

Kathirgamanathan, P, Herrington, PR & McIver, I 2012, ‘Chip seal finite element


model’, International conference on maintenance and rehabilitation of pavements
and technological control (MAIREPAV), 7th, Auckland.

Kearby, J. 1953, ‘Tests and theories on penetration surfaces’, Proceedings of the 32nd
annual meeting of the Highway Research Board, Washington DC.

Khoo, KY 2006, ‘Predictive modelling and experimental studies of thermal inactivation


of bacteria as affected by combined temperature and pH in liquid’, PhD thesis,
University of Adelaide, Adelaide.

Khoo, KY, 2014, ‘Towards incorporating heavy vehicles into the Austroads sprayed
seal design method’, Austroads Technical Report AP-T260-14, Austroads,
Sydney.

Khoo, KY & Neaylon KL 2011, ‘An Indicative Review of Austroads Aggregate Spread
Rates’, BSRRG committee report, ARRB, Vermont South, Victoria.

— 8-14 —
8 References

Kiggundu, BM & Roberts, FL, 1988, ‘The success/failure of methods used to predict the
stripping propensity in the performance of bituminous pavement mixtures’, NCAT
Report 88-03, National Centre for Asphalt Technology, Auburn, AL.

Kim, YR, 2015, ‘Performance‑Related Specifications for Chip Seals’, TRB Webinar on
Flexible Pavement Preservation, 18 Aug 2015.

Kim, YR & Lee, J 2008, ‘Quantifying the Benefits of Improved Rolling of Chip Seals’,
FHWA, North Carolina Department of Transportation.

Kinder, DF, & Lay MG 1988, ‘Review of the fourth power law’, Australian Road
Research Board Internal Report AIR 000-248, Australian Road Research Board,
Vermont South, Victoria.

Knapton, J 2007, ‘The structural design of heavy duty pavements for ports and other
industries’, Edition 4, Interpave, Leicester, UK.

Kodippily, S 2013, ‘Modelling the flushing mechanism of thin flexible surface


pavements’, PhD Thesis, the University of Auckland, Auckland.

Koniditsiotis, C 2000, ‘Weigh-In-Motion Technology’, Austroads Research Report AP-


R168-00, Austroads, Sydney.

Lawson, WD, & Senadheera, S 2009, ‘Chipseal maintenance’, Transportation


Research Record: Journal of the Transportation Research Board, Vol 2108: 61–
68, Washington, DC

Lay, MG 1998, ‘The history of vehicle suspensions’, Routes/Roads (PIARC) 298-II: 56-
63, April, Paris, France.

Lay, MG 2008, ‘A history of freight transport prior to the modern truck’, International
Symposium on Heavy Vehicle Transportation Technology, 10th, International
forum for road transport technology, France.

Leach, R & Beer, T 2000, ‘Environmental assessment of emulsions’, Austroads


Research Report, AP-R153-00, Austroads, Sydney.

Ledee, V, Delalande, G, & Dupont, P 2005, ‘Skid resistance and aggregates’,


Laboratorie Central des Ponts et Chaussees, Cedex, France.

— 8-15 —
8 References

Lee, J, & Kim, YR 2010a, ‘Optimal distribution of rolling coverage in multiple chip
seals’, Transportation Research Record: Journal of the Transportation Research
Board, No. 2150, Washington, DC.

Lee, J &. Kim, YR, 2010b, ‘Optimizing rolling patterns for chip seals using laboratory
aggregate loss performance tests on field fabricated samples’, Journal of
Performance of Constructed Facilities, Vol. 24, Issue 3.

Lee, J & Kim, YR 2010c, ‘Determination of the optimal number of coverages for the
rolling of chip seals’, Canadian Journal of Civil Engineering, Vol. 37, Issue 1.

Luk, J 2006, ‘Automatic vehicle classification by vehicle length’, Austroads Technical


Report AP-T60-06, Austroads, Sydney.

Maccarrone, S, Neaylon, KL, Clark, T, & Gnanaseelan, G 1997, ‘Effect of cutter on the
performance of polymer modified sealing binders’, AAPA International flexible
pavements conference, 10th, Perth, Australian Asphalt Pavement Association,
Kew, Vic.

Mackintosh, CS 1961, ‘Rates of spread and spray in bituminous surface dressing of


roads’, Transactions of the South African Institution of Civil Engineers, Vol 3 (Oct
1961).

Major, NG 1972, Letter [concerning water vapour venting in New Zealand Roads], NZ
National Roads Board research unit newsletter, no.36: 3, Wellington.

Major, NG & Tuohey, GJ 1974, ‘New Zealand practice for surface treatments and
friction courses’, Conference on Asphalt Pavements for Southern Africa
(CAPSA), 2nd, Durban, Republic of South Africa.

Major, NG, & Tuohey, GJ 1976, ‘Surface treatments for asphaltic concrete for high
speed traffic’, ARRB Conference, 8th, Perth, Australian Road Research Board,
Vermont South, Victoria.

Mamlouk, MS 2006, ‘Ch. 8, Design of Flexible Pavements’, The handbook of highway


engineering, Fwa, TF, CRC Press, Boca Raton, FL.

Mamlouk, MS & and Dosa M 2014, ‘Verification of effectiveness of chip seal as a


pavement preventive maintenance treatment using LTPP data’, International
Journal of Pavement Engineering Vol 15, No 10.

— 8-16 —
8 References

Maina, J and De Beer, M 2008, Improved performance evaluation of road pavements


by using measured tyre loading, CSIR Biennial Conference, 2nd, Council for
Scientific and Industrial Research, Pretoria.

Marais, CP 1969, ‘A fresh outlook on the design of single surface treatments’,


Conference on Asphalt Pavements for Southern Africa (CAPSA), 1st, Durban,
Republic of South Africa.

Marais, CP, 1979, ‘Advances in the design and application of bituminous materials in
road construction’, PhD thesis, University of Natal, Republic of South Africa.

Martin, A, & Howard A 2010, ‘Laser profiling for ALF Trial 35 – discussion of results’,
ARRB Report TT1219, Draft, ARRB, Vermont South.

Martin, T, 2008, ‘Predicting sealed granular pavement deterioration at a road network


level’, PhD thesis, Monash University, Melbourne.

Masad, E, & Rezaei, A 2009, ‘Relationship of asphalt pavement skid resistance to


aggregate properties’, Chapter 50, Efficient transportation and pavement systems
characterization, mechanisms, simulation, and modelling, edited by Masad, E,
Alnuaimi, NA, Sayed, T, & Al-Qadi, IL, CRC Press, Boca Raton, FL.

Masad, E, Rezaei, A, Chowdhury, A, & Freeman, T 2010, ‘Field evaluation of asphalt


mixture skid resistance and its relationship to aggregate characteristics’,
FHWA/TX-10/0-5627-2, Texas Transportation Institute, College Station, Texas.

McLeod, NW 1969, ‘A general method of design for seal coats and surface treatments’,
Proceedings of the Association of Asphalt Paving Technologists, Volume 38. St.
Paul, MN.

Mendenhall, W & Sincich, T 1996, ‘A second course in statistics: regression analysis’,


Prentice-Hall, New Jersey.

Merriam-Webster 2011, on-line dictionary http://www.merriam-webster.com/ , viewed 4


Oct 2015.

Meydan, A 1997, ‘Evaporation of Cutters from a Chip Seal’, AAPA International


Flexible Pavements Conference, 10th, Perth, Australian Asphalt Pavement
Association, Kew, Victoria.

— 8-17 —
8 References

Microsoft Corporation, 2010, Microsoft Office Professional 2010, Redmond, WA.

Millar, PA 2013, ‘Non-contact evaluation of the geometric properties of highway


surfacing textures using close range photogrammetry’, PhD thesis, University of
Ulster, Ulster.

Millar, PA, Woodward, D, & Friel, S 2011, ‘Mapping interfacial stress distributions to
digital surface microphotography’, International surface friction conference, 3rd.
Gold Coast, Australia, ARRB, Vermont South, Victoria.

Millar, PA, Woodward, D, McQuaid, G, & Neaylon, K 2012, ‘Investigation of aggregate


displacement in chipseal surfaces using close range digital photogrammetry’,
Maintenance and rehabilitation of pavements and technological control
(MAIREPAV) conference, 7th, Auckland.

Millar, PA, Woodward, D & Woodside, AR 2009, ‘Use of close range terrestrial
photogrammetry to assess accelerated wear of asphalt concrete surface course
mixes’, Mairepav 6, The sixth international conference on maintenance and
rehabilitation of pavements and technological control, Politecnico di Torino,
Torino, Italy.

Milne, TI 2004, ‘Towards a performance related seal design method for bitumen and
modified road seal binders’, PhD Thesis (Civil Engineering), University of
Stellenbosch, South Africa.

Mincad Systems 2012, ‘Circly 5 user manual’, Mincad Systems Pty Ltd, Richmond
South, Victoria.

Moffatt, MA 2011, ‘The influence of multiple axle loads on the performance of an


unbound granular pavement under accelerated loading: construction of test
pavements’, Austroads Technical Report AP-T187-11, Austroads, Sydney.

Moffatt, MA 2013, ‘The influence of multiple axle loads on the performance of an


unbound granular pavement under accelerated loading: interim data report’,
Austroads Technical Report AP-T232-13, Austroads, Sydney

Moffatt, MA 2015, ‘The effect of multiple axle loads on the design of flexible
pavements’, PhD Thesis, Department of Civil Engineering, Monash University,
Melbourne.

— 8-18 —
8 References

Moffatt, MA, Jameson, GW, & Martin, A 2011, ‘The influence of multiple axle loads on
pavement performance: interim findings’, Austroads Technical Report AP-T184-
11, Austroads, Sydney.

Moffatt, MA, Jameson, GW, & Young, W 2014, ‘Influence of multiple axle group loads
on mechanistic design of asphalt pavements’, ARRB Conference, 26th, Sydney,
ARRB Group, Vermont South, Victoria.

Moraes, R, Velasquez, R & Bahia HU 2011, ‘Measuring the effect of moisture on


asphalt–aggregate bond with the bitumen bond strength test’, Transportation
Research Record: Journal of the Transportation Research Board Asphalt
Materials and Mixtures 2011, Volume 3, Issue 2209.

Morosiuk, G, Riley, MJ, & Odoki, JB 2001, ‘HDM-4 Modelling road deterioration and
works effects’, The Highway development and management series, Vol 6,
Highway Development and Management, The World Bank, Washington, DC.

Motulsky, HJ & Brown, RE 2006, ‘Detecting outliers when fitting data with nonlinear
regression – a new method based on robust nonlinear regression and the false
discovery rate’, BMC Bioinformatics 7.

Motulsky, HJ & Christopoulos, A 2004, ‘Fitting models to biological data using linear
and non-linear regression’, Oxford University Press, New York.

Murthy, DNP, Xie, M & Jiang R 2004, ‘Weibull Models’, John Wiley and Sons,
Hoboken, NJ.

NAASRA 1965, ‘Principles and practice of bituminous surfacing: volume 1: sprayed


work’, National Association of Australian State Road Authorities, Sydney, NSW.

NAASRA 1975, ‘Principles and practice of bituminous surfacing: volume 1: sprayed


work’, National Association of Australian State Road Authorities, Sydney, NSW.

NAASRA 1980, ‘Principles and practice of bituminous surfacing: volume 1: sprayed


work’, National Association of Australian State Road Authorities, Sydney, NSW.

NAASRA 1989, ‘Bituminous surfacing: sprayed work’, Report No. NTR-07, National
Association of Australian State Road Authorities, Sydney, NSW.

— 8-19 —
8 References

Nalder, D 2015, ‘Longer road trains approved for the Pilbara’, Media Statement,
Minister for Transport, Govt. of Western Australia.
https://www.mediastatements.wa.gov.au/Pages/Barnett/2015/10/Longer-road-
trains-approved-for-the-Pilbara.aspx viewed 12 Oct 2015.

Nanyang Technological University 2003, ‘Pull off test and limpet’,


http://www3.ntu.edu.sg/home/cchtui/webequip.htm , viewed 16 November 2008.

National Highways Institute 2005, ‘Superpave fundamentals’, National Highway


Institute reference manual. NHI Course #131053, Federal Highway
Administration, Washington, DC.

N.D. Lea International Ltd 1995, ‘Modelling road deterioration and maintenance effects
in HDM—4’, Final Report RETA 5549—REG, Highway Development and
Management Research, Asian Development Bank.

Neaylon, KL 1995, ‘Development and use of bitumen emulsions in road surfacings’,


Materials Technology Research and Development Program, MTRD Report No.
40-1, Department of Transport, Adelaide.

Neaylon, KL 1996, ‘The measurement of aggregate reorientation in sprayed seals


using stereoscopic micro-photogrammetry’, Combined 18th ARRB Transport
Research Conference and Transit New Zealand Land Transport Symposium,
Christchurch, NZ, ARRB, Vermont South, Vic.

Neaylon, KL 2008, ‘Sprayed Sealing 2008 - the conference findings’, Summary


communiqué to conference delegates, ARRB Group, Vermont South, Victoria.

Neaylon, KL 2009, ‘Stereo photography as a tool to understand sprayed seal distress


mechanisms’, AAPA International Flexible Pavements Conference,13th, Gold
Coast, Australian Asphalt Pavement Association, Kew, Victoria.

Neaylon, KL 2011, ‘Guidance for the Development of Policy to Manage Skid


Resistance’, Austroads Research Report, AP-R374-11, Austroads, Sydney.

Neaylon, KL 2012, ‘The Effects of Heavy Vehicle Single, Tandem and Triaxle
Groupings on Sprayed Seal Wear – Stage 1’, Austroads Technical Report AP-
T207-12, Austroads, Sydney.

— 8-20 —
8 References

Neaylon, KL 2015, ‘Towards Incorporating Heavy Vehicles into Sprayed Seal Design –
Stage 2’, Austroads Technical Report AP-T292-15, Austroads, Sydney.

Neaylon, KL & Busuttil, R 2001, ‘Adapting South African solutions to South Australia’,
AAPA Pavements Industry Conference, Gold Coast, AAPA, Kew, Victoria.

Neaylon, KL, Spies, R, & Alderson A 2008, ‘The equivalent heavy vehicle concept in
Australian sprayed seal design’, International sprayed sealing conference, 1st,
Adelaide, ARRB Group, Vermont South, Victoria.

Neaylon, KL & Urquhart, R 2013, ‘The 1st Sprayed Sealing Alliance Workshop:
International Best Practice (Perth)’, ARRB Research Report, ARR 381, ARRB,
Vermont South, Victoria.

NHVR 2014, ‘National heavy vehicle mass and dimension limits’, Fact sheet, National
Heavy Vehicle Regulator, Fortitude Valley, Qld.

Nicholls, JC 1996 ‘Design guide for road surface dressing’, 4th Edition, Road Note 39,
TRL, Wokingham, Berkshire, UK.

Nicholls, C, & Baxter J 2008, ‘The design of surface dressings (chip seals) in the
United Kingdom’, International sprayed sealing conference, 1st, Adelaide, ARRB
Group, Vermont South, Vic.

Nielsen, CB 2007, ‘Assessment of porous pavements - How to look inside’, Report 157,
Danish Road Institute, København, Denmark.

Nikolaides, A 2015, ‘Highway Engineering: pavements, materials and control of quality’,


CRC Press, London.

NOAA, 2013, ‘LIDAR’, http://www.webcitation.org/6H82i1Gfx viewed 10/01/2016


National Oceanic and Atmospheric Administration, Silver Spring, MD.

NRB 1968, ‘Manual of sealing and paving practice’, National Roads Board, Wellington.

NRB 1971, ‘Manual of sealing and paving practice’, prepared by roading division
ministry of works for the road research unit, National Roads Board, Wellington.

NTC 2005, ‘Rules for assessment of potential performance based vehicles - Discussion
paper’, National Transport Commission, Melbourne.

— 8-21 —
8 References

NTC 2007, ‘PBS scheme - the standards and vehicle assessment rules’, National
Transport Commission, Melbourne.

NTC 2009, ‘Performance based standards – review’, National Transport Commission,


Melbourne.

NZ Forest Owners Association 2011, ‘New Zealand forest road engineering manual’,
Wellington.

NZ Government 2014, ‘Government policy statement on land transport, 2015/16 -


2024/25’, Wellington.

NZTA 2010, ‘Specification for state highway skid resistance management’, T10: 2010,
NZ Transport Agency, Wellington.

NZTA 2013, ‘Maintenance specification, network outcomes contract’, Vol 4, NZ


Transport Agency, Wellington

OECD 1998, ‘Dynamic interaction between vehicles and infrastructure experiment


(DIVINE)’, technical report, DSTI/DOT/RTR/IR6(98)1/FINAL, Organisation for
Economic Co-operation and Development, Paris.

Oliver, JWH 2006, ‘Changes in the chemical composition of Australian Bitumens’,


ARRB Conference, 22nd, Canberra, ARRB, Vermont South, Victoria.

Oliver, JWH 2011, ‘Cutter contents for PMB seals’, Austroads Technical Report AP-
T180-11, Austroads, Sydney.

Oliver, JWH & Halligan, S 2006, ‘A targeted approach to the measurement of skid
resistance’. ARRB Conference, 22nd, Canberra, ARRB, Vermont South, Victoria.

Owusu-Ababio, S, & Schmitt, R 2015, ‘Analysis of Heavier Truck Weight Data: Case
Study of Logging Trucks’, Transportation Research Record: Journal of the
Transportation Research Board, Volume 2478: Freight Systems, Vol. 2,
Transportation Research Board, Washington DC.

Oxford 1987, The Australian concise oxford dictionary, ed. Turner, GW, Oxford
University Press, Melbourne.

— 8-22 —
8 References

Paterson, WDO 1987, ‘Road deterioration and maintenance effects: models for
planning and management’, The Highway Design and Maintenance Standards
Series, The World Bank.

Patrick, JE 1999, ‘Background to the development of the transit New Zealand


performance based chipseal specification P/17 texture requirements’, Central
Laboratories Report 99-526242.04, Opus International Consultants Limited,
Lower Hutt, New Zealand.

Patrick, JE, Alabaster, DJ, & Dongol, DMS 1988, ‘Pavement density’, Transfund New
Zealand research report 100, Transfund New Zealand, Wellington.

Patrick, JE, Cenek, PD, & Owen, MT, 2000, ‘Comparison of methods to measure
macrotexture’, AAPA International Flexible Pavements Conference, 11th, Sydney,
Australian Asphalt Pavement Association, Kew Victoria.

Patrick, JE & Donbavand J 1996, ‘Development of a performance based chipseal


specification’, Combined 18th ARRB Transport Research Conference and Transit
New Zealand Land Transport Symposium, 1996, Christchurch, New Zealand.
ARRB, Vermont South, Victoria.

Patrick, SA 2013, ‘Update of double / double design for austroads sprayed seal design
method’, Austroads Technical Report, AP-T236-13, Austroads, Sydney.

Patrick, SA 2014, ‘Inspections of sprayed seal trials’, Austroads Technical Report, AP-
T277-14, Austroads, Sydney.

Patrick, SA & Wright, R, 2014, ‘Initial field trials with surface wear rig’, Austroads
Technical Report AP-T274-14, Austroads, Sydney.

PIARC 1987, ‘Report of the committee on surface characteristics’, Technical


Committee 1 Report, XVIII World Road Congress, Permanent International
Association of Road Congresses, Brussels, Belgium.

Pidwerbesky, B 2001, ‘Chip seal design for high stress conditions on forestry roads’,
Low volume roads workshop, Rotorua.

Pidwerbesky, B & Arnold, G 1995, ‘Seal coat design for unbound granular pavements
carrying heavy axle loads’, International Journal of Forest Engineering, Vol 7,
No.1.

— 8-23 —
8 References

Pidwerbesky, B, Arnold, G, & Pollard, J 1994, ‘Can chip seals carry even heavier axle
loads?’ Land Transport Symposium, Transit New Zealand, Wellington.

Pidwerbesky, B, Waters, J, Gransberg D, & Stemprok, R 2006, ‘Road surface texture


measurement using digital image processing and information theory’, Research
Report 290, Land Transport New Zealand, Wellington.

Pocius, AV 2002, ‘Adhesion and Adhesives Technology’, Hanser, Munich.

Potter, JL, & Church, M 1976, ‘The design of sprayed single seals’. ARRB Conference,
8th, Perth. Vol 8, ARRB, Vermont South, Victoria.

Primal Surfacing 2008a, ‘Material safety data sheet: Emoprime emulsion primer,
01/02/2008’, Primal Surfacing, Laverton North, Victoria.

Primal Surfacing 2008b, ‘Material safety data sheet: Emoflex 70, 01/02/2008’, Primal
Surfacing, Laverton North, Victoria.

Prowell, BD, Brown, ER, Anderson, RM, Swamy, JSDAK, Von Quintus, H, Shen, S,
Carpenter, SH, Bhattacharjee, S, & Maghsoodloo, S 2010, ‘Validating the fatigue
endurance limit for hot mix asphalt’, NCHRP Report 646, Transportation
Research Board, Washington DC.

Pumphrey, ME 2003, ‘Evaluation of performance graded asphalt binder equipment and


testing protocol’, Master of Science in Civil Engineering Thesis, West Virginia
University.

Rainsford, S, & Parkman, C 2005, ‘Predicting texture deficiency in pavement


management’, International Surface Friction Conference, 1st, Christchurch.

Ramsay, E & Prem, H 2000, ‘Development of an Austroads heavy vehicle


nomenclature system — discussion paper’, Austroads Research Report, AP-
R174-00, Austroads, Sydney.

Ratkowsky, DA 1990, ‘Handbook of nonlinear regression models’, Marcel Dekker, New


York NY.

Ratkowsky, DA 1993, ‘Principles of nonlinear regression modelling’, Journal of


Industrial Microbiology, Vol 12.

— 8-24 —
8 References

Read, JM 1996, ‘Fatigue cracking of bituminous pavement mixtures’, PhD Thesis,


University of Nottingham, Nottingham, UK.

Read, JM & Whiteoak, D 2003, ‘The shell bitumen handbook’, 5th edition, Thomas
Telford, London.

Rebbechi, J 2008, ‘Guide to Pavement Technology Part 4F: Bituminous Binders’,


Austroads Guide to Pavement Technology, AGPT04F-08, Austroads, Sydney.

Rebbechi, J & Alderson, A 2009, ‘Guide to Pavement Technology Part 4K: Seals’,
Austroads Guide to Pavement Technology, AGPT04K-09, Austroads, Sydney.

Rebbechi, J, & Petho, L 2014, ‘Guide to Pavement Technology Part 4B: Asphalt’,
Austroads Guide to Pavement Technology, AGPT04B-14, Austroads, Sydney.

Rebbechi, J, & Sharp, K 2009, ‘Guide to pavement technology part 3: Pavement


surfacings’, Austroads Guide to Pavement Technology, AGPT03-09, Austroads,
Sydney.

Reliasoft Corporation 2002, ‘Characteristics of the Weibull distribution’, Reliability Hot


Wire, Issue 14, April 2002, Tuscon AZ.

Remondino, F. & El-Hakim, SF 2006, ‘Image based 3D modelling: a review’, The


Photogrammetric Record, Vol. 21, Issue 115.

Remondino, F, El-Hakim SF, Gruen, A, & Zhang, L 2009, ‘Turning images into 3D
models’, IEEE Signal processing magazine, Vol 56.

Remtulla, A 2015, ‘Bituminous binder technologies of the future’, AAPA International


Flexible Pavements Conference, 16th, Gold Coast, Australian Asphalt Pavement
Association, Kew, Victoria.

Road Research Laboratory 1969, ‘Instructions for using the portable skid-resistance
tester’, Road Note 27, Her Majesty's Stationary Office, Crowthorne UK.

Road Science 2015a, ‘New generation emulsions’, Fact sheet, Mt Maunganui, NZ

Road Science 2015b, personal communication with John Vercoe.

Roberts, C, & Nicholls, JC, 2008, ‘Design guide for road surface dressing’, Road Note
39, 6th Edition, TRL, Crowthorne UK.

— 8-25 —
8 References

Roe, PG, Parry AR, & Viner HE 1998, ‘High and low speed skidding resistance: the
influence of texture depth’, TRL Report 367, TRL, Berkshire, UK.

Ross, D, & Field K 2007, ‘South African road surfacing policy, international oil price
changes, and the shadow pricing of costs and benefits’, Contract Report,
Southern African Bitumen Association (SABITA), Cape Town.

RSTA, 2015, ‘Renew magazine’, issue 3 autumn 2015, Road Surface Treatments
Association, Wolverhampton, UK.

RTA 2006, ‘Road surface texture depth’, Test Method T240, Roads and Traffic
Authority, Sydney.

SABITA 2012, ‘Bituminous surfacings for low volume roads and temporary deviations’.
Manual 10, South African Bitumen Association, Howard Place, South Africa.

Salama, HK, Chatti, K, & Lyles, RW 2006, ‘Effect of heavy multiple axle trucks on
flexible pavement damage using in-service pavement performance data’, Journal
of Transportation Engineering, Vol 132, Issue 10.

SANRAL 2007, ‘TRH3 Design and construction of surfacing seals’, Technical


Recommendations for Highways No.3, South African National Roads Agency
Limited, Pretoria.

SANRAL 2013, ‘South African pavement engineering manual’, South African National
Roads Agency Ltd, Pretoria, South Africa.

SANRAL 2016, South African pavement design method – end product’


http://sapdm.co.za/pgcontent.php?UID=512 viewed 28 Dec 2016

Scala, A 1970, ‘Comparison of the response of pavements to single and tandem axle
loads’, ARRB Conference, 5th, Canberra, ARRB Group Vermont South, Victoria.

Schmoyer, R, Hu, PS, & Swank, P 1998, ‘Analysis of vehicle classification and truck
weight data of the New England states’, Final report, Oak Ridge National
Laboratory. Oak Ridge, Tennessee.

Schonfeld, R 1974, ‘Photo-interpretation of skid resistance in practice’, Transportation


Research Record Vol. 523.

— 8-26 —
8 References

SFERB & USIRF 2008, ‘Bitumen Emulsions’. La Section des Fabricants d'Emulsions
Routières de Bitume (SFERB) and L'Union des Syndicats de l'Industrie Routière
Française (USIRF), Paris.

Shalaby, A & El Gendy, A 2008, ‘Three-dimensional pavement surface macrotexture


measurements using the photometric stereo technique and applications’, SURF
2008, Slovenia, World Road Association, Paris.

Sharp, KG, 2001, ‘Report on accelerated pavement testing 1999 international


conference’, Research Report, ARR 352, ARRB Group, Vermont South, Victoria.

Sharp, KG, 2005 ‘Guide to Pavement Technology Part 1: Introduction to Pavement


Technology’, Austroads Guide to Pavement Technology, AGPT01-05, Austroads,
Sydney.

Sharp, KG, Johnson-Clark, JR & Fossey, DW 1999, ‘A review of the Australian ALF
program’, International conference on accelerated pavement testing, 1st, Reno,
University of Nevada, Nevada.

Shaw, P 1980, ‘Stress-strain relationships for granular materials under repeated


loading’, PhD Thesis, University of Nottingham.

Shuler, S, Lord, A, Epps-Martin, A, & Hoyt, D 2011, ‘Manual for emulsion-based chip
seals for pavement preservation’, NCHRP report 680, National Cooperative
Highway Research Program, Transportation Research Board, Washington DC.

Souliman, M 2012, ‘Integrated predictive model for healing and fatigue endurance limit
for asphalt concrete’, PhD Thesis, Arizona State University.

Spies, RE 2006a, ‘A suggested approach to incorporating heavy vehicles into sprayed


seal design’, Working Group Committee Report for Austroads Bituminous
Surfacings Research Reference Group, unpublished, Queensland Department of
Main Roads, Brisbane.

Spies, RE 2006b, ‘Considerations in developing seal design factors for various types of
heavy vehicles’, Working group committee report for Austroads bituminous
surfacings research reference group, unpublished, Queensland Department of
Main Roads, Brisbane.

— 8-27 —
8 References

Spies, RE 2006c, ‘An approach to incorporating heavy vehicles into sprayed seal
design’, Working group committee report for austroads bituminous surfacings
research reference group, unpublished, Queensland Department of Main Roads,
Brisbane.

Spies, RE 2007, ‘Potential impacts on seal design of heavy vehicle loadings across
Australia’, Working group committee report for austroads bituminous surfacings
research reference group, unpublished, Queensland Department of Main Roads,
Brisbane.

Standards Australia 2009a, ‘Method 11.1: Particle size distribution—Sieving method’,


Methods for sampling and testing aggregates, AS 1141.11.1:2009, SAI Global,
Sydney.

Standards Australia 2009b, ‘Part 2: Aggregate for sprayed bituminous surfacing’,


Aggregates and rock for engineering purposes, AS 2758.2-2009, SAI Global,
Sydney.

Stevenson, Kinder & Scott 1970, ‘Bedford Trucks’, Historical records rescue
consortium, State Library of Western Australia, Perth.

Tabachnick, BG & Fidell, LS 2007, ‘Using multivariate statistics’, Pearson, San


Francisco.

Tam, WO 2006, ‘Ch. 7 Highway Materials’, The handbook of highway engineering, Fwa
TF (ed.), CRC Press. New York.

Tarrer, AR & Wagh, VP 1991, ‘The effect of the physical and chemical characteristics
of the aggregate on bonding’, Strategic Highway Research Program Report
SHRP-A/UIR-91-507, Washington, D.C.

Testa, DM & Hossain, M 2014, ‘Kansas Department of Transportation 2014 chip seal
manual’, final report, K-TRAN: KSU-09-8, Kansas Department of Transportation,
Topeka, Kansas.

Texas Department of Transportation 2010, ‘Seal coat and surface treatment manual’,
Texas Department of Transportation, Austin, TX.

Texas Department of Transportation 2012, Traffic recorder instruction manual, online


manuals, Austin, TX.

— 8-28 —
8 References

Thoresen, T 2006, ‘Basic statistical tools: likelihood ratio tests’, training presentation,
ARRB, Vermont South, Victoria.

Thoresen, T 2012, ‘Statistics discussion’, Unpublished, ARRB, Vermont South,


Victoria.

Toole, T, & Hillier, P 2014, ‘Good practice in reseal programming’, Austroads Technical
Report AP-T273-14T, Austroads, Sydney.

Topcon 2007, ‘Image master’, operation manual, Tokyo.

Towler, J & Dawson, J. 2008, ‘History of chipsealing in NZ - Hanson to P/17’,


International Sprayed Sealing Conference, 1st, Adelaide, ARRB, Vermont South,
Victoria.

Transit New Zealand 1981, ‘Standard test procedure for measurement of texture by the
sand circle method’, TNZ T/3, Transit New Zealand, Wellington.

Transit New Zealand 1989, ‘Notes for resealing’, TNZ P/4 notes, Transit New Zealand,
Wellington.

Transit New Zealand 1993, ‘Bituminous sealing manual’, Transit New Zealand,
Wellington.

Transit New Zealand 1994, ‘Performance based specification for bituminous reseals’,
TNZ P/17:1994, Transit New Zealand, Wellington.

Transit New Zealand 2002, ‘Performance based specification for reseals’, TNZ
P/17:2002, Transit New Zealand, Wellington.

Transit New Zealand 2005, ‘Performance based specification for hotmix asphalt
wearing course surfacing’, TNZ P/23: 2005, Transit New Zealand, Wellington.

Transit New Zealand, Road Controlling Authorities, & Roading New Zealand 2005,
‘Chip sealing in New Zealand’, NZ Transport Agency, Wellington.

Tredrea PF 2006,’ Specification framework for polymer modified binders and


multigrade bitumens’, Austroads Technical Report AP-T41-06, Austroads,
Sydney.

— 8-29 —
8 References

Tredrea, PF 2007, ‘Superpave binder properties and the role of viscosity’, AAPA
pavements industry conference, Sydney, Australian Asphalt Pavements
Association, Kew, Victoria.

TRL 2000, ‘A guide to surface dressing in tropical and subtropical countries’, Overseas
Road Note, ORN 3, Transport Research Laboratory, Berkshire, UK.

Ullidtz, P 1987, ‘Pavement analysis’, Elsevier, Amsterdam.

Urquhart, R 2012, ‘Aggregate wetting and initial adhesion of sprayed sealing binders’,
Austroads Technical Report AP-T210-12, Austroads, Sydney.

Urquhart, R, Aguiar, L & Khoo, KY 2015, ‘Seal design improvement for low volume
roads’, Austroads Technical Report, AP-T302-15, Austroads, Sydney.

van Boekel, MAJS 2002, ‘On the use of the Weibull model to describe thermal
inactivation of microbial vegetative cells’, International Journal of Food
Microbiology, Vol 74, Issue (1–2).

van der Poel, C. 1954, ‘A general system describing the viscoelastic properties of
bitumens and its relation to routine test data’, Journal of Applied Chemistry, Vol 4,
Issue 5.

van Zyl, GD 2007, ‘Measurement and interpretation of input parameters used in the
South African surface seal design method’, Conference on Asphalt Pavements
for Southern Africa, 9th, Gaborone, Botswana.

van Zyl, GD & Jenkins, K 2015, ‘Overview of long term seal performance’, Conference
on Asphalt Pavements for South Africa, 11th, Sun City, South Africa.

VicRoads 2004, ‘Bituminous sprayed surfacing manual’, Technical Bulletin No 45,


VicRoads, Ktew, Victoria.

VicRoads 2007, ‘Standard specification for roadworks: Section 812: crushed rock for
base and subbase pavement’, VicRoads, Kew, Victoria.

VicRoads 2009, ‘Guide to surface inspection rating’, Technical Bulletin 50, VicRoads,
Kew, Victoria.

— 8-30 —
8 References

VicRoads 2014, ‘Traffic volume data for Victoria’, May 2014, Information management
and technology, VicRoads, Kew, Victoria.

Vuong, B, Yeo, R & Choummanivong, L 2003, ‘Axle load equivalency ALF Trial:
measured versus predicted performance’, Contract Report RC2485-2, ARRB,
Vermont South, Victoria.

Wambold, JC, Antle, CE, Henry, JJ, Rado, Z, Descornet, G, Sandburg, U, Gothié, M, &
Huschek, S 1995, ‘International experiment to compare and harmonize skid
resistance and texture measurements’, PIARC Report 01.04.T, PIARC, Paris.

Waters, JC 2013, ‘Sealing with emulsion in New Zealand’, Southern Hemisphere


Sprayed Sealing Alliance workshop, AAPA international flexible pavements
conference, 15th, Brisbane, Australian Asphalt Pavement Association, Kew,
Victoria.

Weibull, W 1951, ‘A statistical distribution function of wide applicability’, ASME Journal


of applied mechanics, September 1951.

Werkmeister, S 2003, ‘Permanent deformation behaviour of unbound granular


materials in pavement constructions’, PhD Thesis, Dresden University of
Technology.

White, G 2015a, ‘Systematic diagnosis of factors leading to cyclic shear creep of airport
asphalt surfaces’, PhD Thesis, University of the Sunshine Coast, Queensland.

White, G 2015b, ‘A National specification for airport sealing’, AAPA International


flexible pavements conference, 16th, Gold Coast, Queensland, Australian Asphalt
Pavement Association, Kew, Victoria.

Wistuba, MP, Grothe, H, Grönniger, J, & Handle F 2012, ‘Adhesion of bitumen:


screening and evaluating laboratory testing techniques’, Eurasphalt & Eurobitume
Congress, 5th, Istanbul.

Wix, R & Moffatt, MA 2007, ‘Pavement surface texture measurement with a laser
profilometer’, Austroads Test Method AGAM-T013R, Austroads, Sydney

Wood, TJ, Janisch DW & Gaillard, FS 2006, ‘Minnesota seal coat handbook 2006’,
Minnesota Department of Transportation.

— 8-31 —
8 References

Woodside, AR &. Liu, GX 1993, ‘The assessment of interfacial stresses between


individual surface chippings and bituminous binder’, Eurobitume Congress, 5th,
Stockholm, Sweden.

Woodside, AR & Rogan, C 1993a, ‘The use of the 'Limpet' tester to assess the bond
strength characteristics of bituminous materials’. Eurobitume Congress, 5th,
Stockholm, Sweden.

Woodside, AR & Rogan, C 1993b, ‘The development of a tensile tenacity test to predict
the performance of a bituminous binder surfacing’, Eurobitume Congress, 5th,
Stockholm, Sweden.

Woodward, WDH, Woodside, AR & Jellie, JH 2005, Higher PSV and other aggregate
properties, International Surface Friction Conference, 1st Christchurch, New
Zealand, Transit New Zealand.

Xiao, Y 2013, Towards a Performance Evaluation Method for Durable and Sustainable
Thin Surfacings, PhD Thesis, Delft University of Technology.

Yaacob, H, Woodward, D & Woodside, A, 2007, ‘Predicting chip seal embedment’.


International conference on maintenance and rehabilitation of pavements and
technological control (MAIREPAV), 5th, Park City, Utah.

Yeo, R, Koh, SL, & Sharp, K 2008, ‘Relative pavement wear of an unbound granular
pavement due to dual tyres and single tyres’, Austroads Technical Report, AP-
T104-08, Austroads, Sydney.

Yeo, R, Martin, T, & Koh, SL, 2006, ‘Investigation of the load damage exponent of
unbound granular materials under accelerated loading’, Austroads Technical
Report AP-T73-06, Austroads, Sydney.

Yoon, HH & Tarrer, AR 1988, ‘Effect of Aggregate Properties on Stripping’,


Transportation Research Record, Issue 1171, Asphalt Materials and Mixtures.

Youdale, G, & Sharp K 2007, ‘Guide to Pavement Technology Part 4: Pavement


Materials’, Austroads Guide to Pavement Technology AGPT04-07, Austroads,
Sydney.

— 8-32 —
8 References

Zaniewski, JP and Pumphrey, ME 2004, Evaluation of Performance Graded Asphalt


Binder Equipment and Testing Protocol, Department of Civil and Environmental
Engineering, Morgantown, West Virginia.

— 8-33 —
Appendix A Seal design and construction detail

A. SEAL DESIGN AND CONSTRUCTION DETAIL


This appendix provides more detailed information regarding the seal construction and
design used in this study.

A.1 Seal design detail


Acronyms used
Acronym Meaning
ALD Average least dimension of aggregate particles
EHV Equivalent heavy vehicle
EHV (%) HV% + (3 x LHV%)
FI Flakiness index
HV Heavy vehicle (Class 2 – 9)
LHV Large heavy vehicle (class 10 – 12 & above)
PME Polymer modified emulsion
SPTD Sand patch texture depth
v/l/d Vehicles/lane/day (the traffic unit used in seal design)

Design Method used


Largely based on Austroads AP-T68-06 (Alderson 2006).

Design Traffic (Austroads AP-T68-06, Section 1.5.4)


The estimated passes of ALF were 8,000 axle passes/day (24 hr). The expected axle
loading 60 ± 20 kN, compared to standard axle loading of 40 kN.

If all were Class 9 (six axles) then this would equate to 1,333 v/l/d (HV).

If all were Class 10 (seven axles, i.e. B-doubles) then this would equate to 1,142 v/l/d
(LHV).

Assuming 70:30 split for HV:LHV (as per Holtrop et al. 2008)

this equates to 8,000 ÷ 6 x 0.7 + 8,000 ÷ 7 x 0.3 = 1,275 trucks/l/d

or in terms of EHV:

EHV = 1,333 x 0.7 + (1,142 x 0.3) x 3

= 1,960 EHV/l/d.

If say trucks constitute 15% of v/l/d (my assumption),

— A-1 —
Appendix A Seal design and construction detail

then 0.15 x v/l/d = 1,960

v/l/d = 13,066.

If say trucks constitute 10% of v/l/d (as per Holtrop et al. 2008)

then 0.10 x v/l/d = 1,960

v/l/d = 19,600.

In actual fact it does not significantly matter whether the traffic is 13,066 or 19,600 as
the design chart goes off-scale after 10,000 (AP-T68-06 Figure 2.3).

Voids Factor and Adjustments


Basic Voids Factor (Vf) (AP-T68-06, Figure 2.3)

The minimum voids factor believed to be physically possible is 10% (Holtrop, personal
discussion), so a Vf of 0.10 was assumed (note: the facts, folk lore and history behind
the derivation of the Austroads voids factors is tangled and interesting, and worthy of a
conference paper in its own right).

Aggregate adjustment (Va) (AP-T68-06 Table 2.1)

For good quality, angular aggregate, with a Flakiness Index of 15 – 20 %, Va = 0.0.

For the 14 mm aggregate to be used in the design, the supplied Flakiness Index (see
above) is 14, therefore:

Va ≈ 0.0.

Traffic Adjustment (Vt) (AP-T68-06 Table 2.2)

Holtrop et al. (2008) noted:

since the fact that all ALF loading corresponds to heavy vehicle
loading, and that this was incorporated in the determination of the
design traffic, an additional adjustment for traffic is not applicable, and
a traffic adjustment of 0.00 is applied.

Accepting this therefore results in:

Vt = 0.0.

— A-2 —
Appendix A Seal design and construction detail

Design Voids Factor (VF)

VF = Vf + Va + Vt

= 0.10 + 0.00 + 0.00

= 0.10.

Voids Factor adjustment as a first coat of a double/double seal (AP-T68-06 see Section
6.2.2)

Reduction Factor, RF = 0.90 (see AP-T68-06 Table 6.1).

Holtrop et al. (2008) noted:

However as voids factor is already at the minimum voids factor


believed physically possible, i.e. 10%, no further reduction is possible
so a Vf of 0.10 is assumed.

This 2008 assumption was not accepted in this design, as it was considered that
further void reduction by infilling the uppermost voids with the next coat is possible, i.e.
topmost air voids reduced.

VF1 = VF x RF (see AP-T68-06 Section 6.2.2[d])

= 0.10 x 0.90

= 0.09.

First Binder Application Rate


Basic binder application rate (Bb) (AP-T68-06 Section 6.2.2[e])

Bp = VF1 x ALD

= 0.09 x 8.8

= 0.79 L/m2.

Surface texture allowance (As) (AP-T68-06 Section 2.1.7)

— A-3 —
Appendix A Seal design and construction detail

The texture allowance for a primed surface is generally in the order of +0.0 to +0.30
L/m2, as newly constructed basecourses of highway standard are typically quite
smooth. However, this base as presented was extremely rough (Figure A.1), and
appeared to have tyne marks from when the pavement surface was tyned and
reworked. Depth of the tyne marks was not measured, but estimated to be 25 - 30 mm.
This was far from ideal for a sprayed seal study, but was considered acceptable for the
primary (pavement) study.

Basecourse after brooming Basecourse after priming

Figure A.1: Rough final basecourse

From AP-T68-06 Table 2.3, visual assessment, and judgement, a surface texture
allowance was finally selected as:

As = +0.4 L/m2.

Binder Embedment allowance (Ae) (AP-T68-06 Section 2.1.7[b])

As discussed before, the ball embedment (prior to brooming) was 2 to 3 mm.

This was checked again after brooming, when it was found to be closer to 2 mm. This
subtlety is beyond the scope of the design method at this traffic level, therefore more
accurate data was not critical.

Ae = 0.1 L/m2 (see AP-T68-06 Figure 2.4).

— A-4 —
Appendix A Seal design and construction detail

Binder adsorption allowance (Aba) (AP-T68-06 Section 2.1.7[c])

The new surface was primed, the base material had no known history of being
unusually adsorptive, and PMB is more viscous than unmodified bitumen, so the
assumed allowance was

Aba = 0.0.

PMB Factor (PF) (AP-T68-06 Table 4.1)

It was assumed from experience that that PME modified with 3% latex ≈ S10E (S10E is
an Austroads PMB grade, and the specification at the time was AP-T41-06 (Tredrea
2006)).

Assume that the ‘type of treatment’ is ‘aggregate retention on high to very high traffic
application’, then

PF = 1.0.

Emulsion Factor (AP-T68-06 Section 8)

As using High Bitumen Content Emulsion (HBCE) then:

EF = 1.1.

Modified basic binder rate (Bmb) (AP-T68-06, Section 5.2.1[b])

Bmb = Bb x PF x EF

= 0.79 x 1.0 x 1.1

= 0.87 L/m2 residual.

Design binder application rate (Bd) (AP-T68-06 Section 5.2.1[c])

Bd = Bmb + allowances

= Bmb + As + Ae

= 0.87 + 0.4 - 0.1

= 1.17

— A-5 —
Appendix A Seal design and construction detail

= 1.2 L/m2 residual.

First Aggregate Spread Rate


(ASR1, AP-T68-06, Section 6.2.2[h])
ASR1 = 900 / ALD

= 900 / 8.8

= 102

= 100 m2/m3.

Second Binder Application Rate (Bb)


Basic binder application rate (Bb) (AP-T68-06 Section 6.2.3[b])

Design voids factor VF2 = VF = 0.10

Bb = VF x ALD

= 0.10 x 3.2

= 0.32 L/m2 residual.

for a 7 mm aggregate without a measured ALD (see AP-T68-06 Table 3.1):

Bb = 0.5 L/m2 residual.

Design second binder application rate (Bd).

Bd = Bb

= 0.5 L/m2 residual.

Second Aggregate Spread Rate (ASR2)


(AP-T68-06, Table 6.3)

ASR (known ALD) = 1100/ALD to 1150/ALD

= 1100/3.2 to 1150/3.2

= 343 to 359 m2/m3.

— A-6 —
Appendix A Seal design and construction detail

ASR (unknown ALD) = 250 to 300 m2/m3.

Select ASR2 ≈ 340 m2/m3.

Design Summary
The resulting seal design is shown in Table A.1.

Table A.1: Constraints on experimental variables

Application Material Application rate


SBR latex emulsion binder 1.2 L/m2 residual
First application
14 mm aggregate 100 m2/m3
SBR latex emulsion binder 0.5 L/m2 residual
Second application
7 mm aggregate 340 m2/m3

A.2 Basecourse priming


The steps in the priming process are shown in Figure A.2 to Figure A.7.

Figure A.2: Base broomed back to hard material

Figure A.3: Basecourse after brooming

— A-7 —
Appendix A Seal design and construction detail

Figure A.4: Base moistened with water

Figure A.5: Basecourse primed with Emoprime at 1.0 L/m² of product

Figure A.6: Prime ponding in the hollows

The hollows shown in Figure A.6 are roughly 200 mm in diameter.

— A-8 —
Appendix A Seal design and construction detail

Figure A.7: Finished primed surface (facing north)

A.3 Sprayed seal construction


Pre-commencement
Sprayed sealing was undertaken on 18th September 2008. Spraying commenced at
approximately 1:25 pm with air temperature approximately 16 °C and pavement
temperature approximately 15 °C.

The Emoflex™ 70 polymer modified emulsion (PME) was manufactured that morning,
and arrived to site at 85 °C. The whole job was completed by about 3:30 pm, by which
stage the PME was at 75 °C. The composition of the Emoflex™ as detailed by the
product data sheet is shown in Table A.2.

Table A.2: Composition of the Emoflex 70™ binder

Component Proportion (% v/v)


Bitumen 68
Solvent <1
Latex 3
Fatty amines and derivatives <2
Hydrochloric acid <1
Water > 25
Source: Primal Surfacing (2008b).

The aggregate was dusty, and very lightly precoated with Shell Mexicoat™ 2B (precoat
rate approximately 1 L/m³ for the 14 mm). On observation, the 7 mm aggregate
appeared like an asphalt 7 mm ( -7 + 2 mm) rather than a sprayed seal 7 mm ( -7 +5
mm). Test properties for the aggregate are shown in Table A.3.

— A-9 —
Appendix A Seal design and construction detail

Table A.3: Properties of Boral (Montrose) seal aggregate

Component First application Second application


Nominal size 14 mm 7 mm
ALD 8.8 mm 3.1 to 3.4 mm
FI 14 34 to 36
Source: Adapted from Boral Resources (2008a), (2008b) & (2008c).

First binder application


All spraying ran from south to north, commencing against the eastern wall. The actual
(as opposed to design) application rate for the first binder application was calculated
from tank dipping as 1.0 L/m² (compared to 1.2 L/m2 specified).

First aggregate spread (14 mm)


All spreading ran from north to south, commencing against the eastern wall using
reversing tip trucks. The emulsion was left for approximately 5 minutes before
spreading of the cover aggregate.

Aggregate spread rate (ASR) = 972 m² ÷ (8 to 8.5 m³) ≈ 115 to 120 m²/m³.

Figure A.8: First aggregate spread

— A-10 —
Appendix A Seal design and construction detail

Second binder application

Figure A.9: Second binder application

The actual second binder application rate was calculated after dipping as 0.5 L/m²,
compared to 0.5 L/m2 specified.

Second aggregate spread rate (7mm)


ASR = 972 m² ÷ 5.5 m³ ≈ 175 m²/m³.

Rolling
One multi-tyred, pneumatic tyred roller was used (Figure A.10) and operated for
approximately 45 minutes.

Figure A.10: The multi-tyre roller

Finished seal
There appeared to be significant undersize 14 mm and over-spreading on the first run,
against the eastern wall (Figure A.11).

— A-11 —
Appendix A Seal design and construction detail

Figure A.11: Undersize 14 mm together with 14 mm aggregate over-spreading

Note that in Figure A.11 and Figure A.12, the Canon lens cap is 60 mm in diameter and
the AU$1 coin shown is 25 mm diameter 1.

It was also observed that there were 14 mm (with undersize) aggregate filling the tyne
mark valleys in the pavement, with tyne mark crests more open (Figure A.12, taken
10 m from the western wall).

Figure A.12: 14 mm (with undersize) aggregate filling the tyne mark valleys

The shed side panels remained off for several weeks to enable air to circulate and any
volatiles to escape, and final brooming and suction sweeping occurred on 22nd October
2008.

1
The AU$1 coin shown is exactly the same colour and diameter as the NZ$2 coin. The use of coins and
lens caps for scale is confusingly parochial and has since been abandoned.

— A-12 —
Appendix B Seal surface under ALF trafficking

B. SEAL SURFACE UNDER ALF TRAFFICKING


B.1 Tyre footprints through carbon paper

Figure B.1: Exp 3502 Triaxle 90 kN tyre footprints (front, middle, rear axles)

— B-1 —
Appendix B Seal surface under ALF trafficking

B.2 Typical initial texture

Figure B.2: Exp 3502, at Ch 5, 7, & 9 m at 0 assembly passes

B.3 Progression of surface texture reduction

Figure B.3: Exp 3502 Triaxle 90 kN, Ch 7 m at 0 assembly passes

— B-2 —
Appendix B Seal surface under ALF trafficking

Figure B.4: Exp 3502 Triaxle 90 kN, Ch 7 m at 100 assembly passes (of bedding-in load)

Figure B.5: Exp 3502 Triaxle 90 kN, Ch 7 m at 200 assembly passes (of bedding-in load)

— B-3 —
Appendix B Seal surface under ALF trafficking

Figure B.6: Exp 3502 Triaxle 90 kN, Ch 7 m at 3000 assembly passes (of bedding-in load)

Figure B.7: Exp 3502 Triaxle 90 kN, Ch 7 m at 500 assembly passes (of bedding-in load)

— B-4 —
Appendix B Seal surface under ALF trafficking

Figure B.8: Exp 3502 Triaxle 90 kN, Ch 7 m at 1,000 assembly passes (of bedding-in load)

Figure B.9: Exp 3502 Triaxle 90 kN, Ch 7 m at 5,000 assembly passes (of bedding-in load)

— B-5 —
Appendix B Seal surface under ALF trafficking

Figure B.10: Exp 3502 Triaxle 90 kN, Ch 7 m at 8,500 assembly passes (of bedding-in load)

Figure B.11: Exp 3502 Triaxle 90 kN, Ch 7 m at 13,000 assembly passes

— B-6 —
Appendix B Seal surface under ALF trafficking

Figure B.12: Exp 3502 Triaxle 90 kN, Ch 7 m at 25,000 assembly passes

Figure B.13: Exp 3502 Triaxle 90 kN, Ch 7 m at 130,000 assembly passes

— B-7 —
Appendix B Seal surface under ALF trafficking

Figure B.14: Exp 3502 Triaxle 90 kN, Ch 7 m at 168,845 assembly passes

Figure B.15: Exp 3502 Triaxle 90 kN, Ch 7 m at 197,000 assembly passes

— B-8 —
Appendix C Volumetric sand patch texture data

C. VOLUMETRIC SAND PATCH TEXTURE DATA


After every nominated number of passes, the ALF was stopped and volumetric sand
patch tests were taken at chainages 11, 13, 15, and 17 m along all test sections, at the
centreline of the wheel path. At each of these chainages, the sand circle diameter from
24.6 cm3 of graded single sized sand was measured as per Figure 4.28, the diameters
averaged, and the total diameter measurements of all four chainages averaged and
then input in the conversion equation in AG:PT/T250 Modified surface texture depth
(Pestle method) (Austroads 2008).

Figure 4.28: Measurements of the sand circles

Thus the average surface texture depth (STD) was reported as sand patch texture
depth (SPTD) in mm and calculated using a simple average from four measurements
at each of four chainages over the trial length, after each nominated number of passes.

The resulting data, shown below, were used in this analysis.

— C-1 —
Appendix C Volumetric sand patch texture data

Experiment No and Axle Axle SPTD (mm)


axle type assembly assembly
passes passes (after
(total) bedding-in)

Exp3500_sing40 10000 0 1.73


Exp3500_Sing40 22500 12500 1.80
Exp3500_sing40 30000 20000 1.73
Exp3500_sing40 40000 30000 1.73
Exp3500_sing40 50000 40000 1.68
Exp3500_sing40 60610 50610 1.68
Exp3500_sing40 66750 56750 1.86
(outlier &
removed)
Exp3500_sing40 74611 64611 1.64
Exp3500_sing40 81250 71250 1.63
Exp3500_sing40 89000 79000 1.59
Exp3500_sing40 100000 90000 1.68
Exp3500_sing40 122000 112000 1.46
Exp3500_sing40 175000 165000 1.47
Exp3500_sing40 200000 190000 1.47
Exp3500_sing40 225000 215000 1.44
Exp3500_sing40 250000 240000 1.44
Exp3501_trix90 10000 0 1.46
Exp3501_trix90 12500 2500 1.48
Exp3501_trix90 15000 5000 1.48
Exp3501_trix90 18200 8200 1.35
Exp3501_trix90 25500 15500 1.30
Exp3501_trix90 61500 51500 1.22
Exp3501_trix90 76000 66000 1.15
Exp3501_trix90 100000 90000 1.24
Exp3501_trix90 130000 120000 1.09
Exp3501_trix90 143000 133000 1.18
Exp3501_trix90 158117 148117 1.11
Exp3501_trix90 171303 161303 1.14
Exp3501_trix90 186500 176500 1.23
Exp3501_trix90 201000 191000 1.19
Exp3501_trix90 225000 215000 1.15
Exp3501_trix90 249000 239000 1.12
Exp3501_trix90 283000 273000 1.05
Exp3501_trix90 300000 290000 1.18
Exp3501_trix90 325000 315000 1.19
Exp3501_trix90 348000 338000 1.05

— C-2 —
Appendix C Volumetric sand patch texture data

Experiment No and Axle Axle SPTD (mm)


axle type assembly assembly
passes passes (after
(total) bedding-in)

Exp3501_trix90 390000 380000 1.03


Exp3503_tand60 10000 0 1.61
Exp3503_tand60 12000 2000 1.62
Exp3503_tand60 15500 5500 1.61
Exp3503_tand60 17000 7000 1.65
Exp3503_tand60 18500 8500 1.60
Exp3503_tand60 23500 13500 1.53
Exp3503_tand60 25000 15000 1.49
Exp3503_tand60 33000 23000 1.51
Exp3503_tand60 40000 30000 1.46
Exp3503_tand60 48000 38000 1.44
Exp3503_tand60 53300 43300 1.46
Exp3503_tand60 60000 50000 1.51
Exp3503_tand60 75000 65000 1.39
Exp3503_tand60 88000 78000 1.35
Exp3503_tand60 112000 102000 1.32
Exp3503_tand60 140000 130000 1.33
Exp3503_tand60 143000 133000 1.30
Exp3503_tand60 171850 161850 1.29
Exp3503_tand60 192000 182000 1.32
Exp3503_tand60 230000 220000 1.25
Exp3503_tand60 263100 253100 1.27
Exp3503_tand60 289111 279111 1.22
Exp3503_tand60 322600 312600 1.26
Exp3504_trix90 10000 0 1.58
Exp3504_trix90 11000 1000 1.50
Exp3504_trix90 16000 6000 1.45
Exp3504_trix90 21750 11750 1.54
Exp3504_trix90 30000 20000 1.53
Exp3504_trix90 35750 25750 1.34
Exp3504_trix90 39000 29000 1.26
Exp3504_trix90 45000 35000 1.17
Exp3504_trix90 51000 41000 1.24
Exp3504_trix90 65000 55000 1.48
Exp3504_trix90 80000 70000 1.15
Exp3504_trix90 87000 77000 1.08
Exp3505_tand80 10000 0 1.68
Exp3505_tand80 15000 5000 1.58

— C-3 —
Appendix C Volumetric sand patch texture data

Experiment No and Axle Axle SPTD (mm)


axle type assembly assembly
passes passes (after
(total) bedding-in)

Exp3505_tand80 20800 10800 1.53


Exp3505_tand80 21000 11000 1.42
Exp3505_tand80 23000 13000 1.43
Exp3505_tand80 35900 25900 1.37
Exp3505_tand80 43030 33030 1.30
Exp3505_tand80 54000 44000 1.34
Exp3505_tand80 62752 52752 1.26
Exp3505_tand80 76700 66700 1.20
Exp3505_tand80 89960 79960 1.17
Exp3505_tand80 100000 90000 1.16
Exp3505_tand80 123000 113000 1.11
Exp3505_tand80 139066 129066 1.11
Exp3505_tand80 150000 140000 1.08
Exp3505_tand80 153475 143475 1.10
Exp3505_tand80 159500 149500 1.06
Exp3505_tand80 168000 158000 1.11
Exp3505_tand80 221629 211629 0.99
Exp3505_tand80 232656 222656 0.98
Exp3505_tand80 248900 238900 0.95
Exp3505_tand80 270000 260000 0.95
Exp3505_tand80 291000 281000 0.91
Exp3506_tand80 10000 0 1.47
Exp3506_tand80 13000 3000 1.46
Exp3506_tand80 15650 5650 1.41
Exp3506_tand80 18280 8280 1.44
Exp3506_tand80 24000 14000 1.36
Exp3506_tand80 40210 30210 1.21
Exp3506_tand80 54000 44000 1.24
Exp3506_tand80 62000 52000 1.26
Exp3506_tand80 78200 68200 1.25
Exp3506_tand80 99000 89000 1.24
Exp3506_tand80 113000 103000 1.21
Exp3506_tand80 130000 120000 1.20
Exp3506_tand80 160000 150000 1.26
Exp3506_tand80 180000 170000 1.16
Exp3506_tand80 220000 210000 1.10
Exp3506_tand80 243200 233200 1.16
Exp3506_tand80 264600 254600 1.10

— C-4 —
Appendix C Volumetric sand patch texture data

Experiment No and Axle Axle SPTD (mm)


axle type assembly assembly
passes passes (after
(total) bedding-in)

Exp3506_tand80 322223 312223 1.16


Exp3506_tand80 344700 334700 1.13
Exp3506_tand80 370000 360000 1.12
Exp3506_tand80 376000 366000 1.01
Exp3507_sing40 10000 0 1.87
Exp3507_sing40 11000 1000 1.83
Exp3507_sing40 14675 4675 1.87
Exp3507_sing40 17650 7650 1.88
Exp3507_sing40 23300 13300 1.79
Exp3507_sing40 31150 21150 1.77
Exp3507_sing40 40000 30000 1.87
Exp3507_sing40 50000 40000 1.74
Exp3507_sing40 55296 45296 1.62
Exp3507_sing40 64000 54000 1.71
Exp3507_sing40 80000 70000 1.61
Exp3507_sing40 87549 77549 1.69
Exp3507_sing40 98200 88200 1.60
Exp3507_sing40 111350 101350 1.69
Exp3507_sing40 118000 108000 1.68
Exp3507_sing40 126000 116000 1.68
Exp3507_sing40 134000 124000 1.68
Exp3507_sing40 149200 139200 1.59
Exp3507_sing40 165000 155000 1.68
Exp3507_sing40 180000 170000 1.60
Exp3507_sing40 196000 186000 1.59
Exp3508_tand60 10000 0 1.92
Exp3508_tand60 18200 8200 1.89
Exp3508_tand60 25325 15325 1.77
Exp3508_tand60 46000 36000 1.58
Exp3508_tand60 61499 51499 1.52
Exp3508_tand60 76366 66366 1.49
Exp3508_tand60 100000 90000 1.46
Exp3508_tand60 122850 112850 1.50
Exp3508_tand60 152286 142286 1.40
Exp3508_tand60 164357 154357 1.37
Exp3508_tand60 184939 174939 1.33
Exp3508_tand60 190639 180639 1.33
Exp3508_tand60 222000 212000 1.27

— C-5 —
Appendix C Volumetric sand patch texture data

Experiment No and Axle Axle SPTD (mm)


axle type assembly assembly
passes passes (after
(total) bedding-in)

Exp3508_tand60 253000 243000 1.17


Exp3508_tand60 263580 253580 1.15
Exp3508_tand60 278232 268232 1.12
Exp3508_tand60 290991 280991 1.11
Exp3511_sing40 10000 0 1.78
Exp3511_sing40 17000 7000 1.72
Exp3511_sing40 22900 12900 1.89
Exp3511_sing40 28418 18418 1.67
Exp3511_sing40 38033 28033 1.68
Exp3511_sing40 61000 51000 1.69
Exp3511_sing40 82300 72300 1.61
Exp3511_sing40 91700 81700 1.52
Exp3511_sing40 104657 94657 1.48
Exp3511_sing40 112257 102257 1.48
Exp3511_sing40 119181 109181 1.40
Exp3511_sing40 138272 128272 1.48
Exp3511_sing40 150000 140000 1.39
Exp3511_sing40 163658 153658 1.39
Exp3511_sing40 170000 160000 1.40
Exp3511_sing40 179050 169050 1.41
Exp3511_sing40 188000 178000 1.30
Exp3511_sing40 200426 190426 1.44
Exp3511_sing40 207000 197000 1.34
Exp3511_sing40 214400 204400 1.36

— C-6 —
Appendix D Western Australian prime Mover and Trailer combinations

D. WESTERN AUSTRALIAN PRIME MOVER AND TRAILER COMBINATIONS

— D-1 —
Appendix D Western Australian prime Mover and Trailer combinations

Source: Main Roads Western Australia (2012). Prime mover, trailer combinations - Operating conditions. Restricted Access Vehicles (RAV). Heavy Vehicle Operations. Perth.

— D-2 —
Appendix E Copyright

E. COPYRIGHT
This appendix contains two sections of the Copyright Act (1968) that are believed to be
relevant to this thesis. It then lists relevant copyright statements and/or copyright
permissions, relating to illustrations and text used in this thesis.

E.1 Commonwealth of Australia Copyright Act 1968


Compilation No. 47 dated 27 November 2015
(http://www.austlii.edu.au/au/legis/cth/consol_act/ca1968133/ viewed 23/1/16).

E.1.1 SECTION 33 Duration of copyright in original works


Subject to this section, copyright that subsists in a literary, dramatic,
musical or artistic work by virtue of this Part continues to subsist until
the end of 70 years after the end of the calendar year in which the
author of the work died.

E.1.2 SECTION 41 Fair dealing for purpose of criticism or review


A fair dealing with a literary, dramatic, musical or artistic work, or with
an adaptation of a literary, dramatic or musical work, does not
constitute an infringement of the copyright in the work if it is for the
purpose of criticism or review, whether of that work or of another work,
and a sufficient acknowledgement of the work is made.

Section 41 is invoked in certain situations as detailed in D.2.

E.2 Relevant copyright statements


E.2.1 ARRB
From: Andrew Meier

Sent: Wednesday, 27 January 2016 7:00 p.m.


To: Kym Neaylon
Subject: RE: Permission to use copyright material in a PhD thesis.

Kym

ARRB Group is happy to grant permission for use of the images as


outlined in your letter of 27 January 2016, for the rights it holds.

ARRB Group is the sole copyright holder of the 8th ARRB Conference
proceedings, and is happy to grant permission for use of the image in
your thesis with due acknowledgement.

The 2006 ARRB conference proceedings however were published as


copyright held by ARRB and authors. As part of our national interest
services, ARRB Library will seek to contact the author and obtain
permission for reproduction in a PhD thesis.

Trust this will not delay proceedings for you and please feel free to
contact me to discuss if you have any questions.

— E-1 —
Appendix E Copyright

Andrew

Andrew Meier

Library Manager
National Interest Services
ARRB Group Ltd
500 Burwood Hwy
Vermont South VIC 3133
Australia
P: +61 3 9881 1603 | F: +61 3 9802 5502

From: John Oliver


Sent: Thursday, 28 January 2016 9:35 AM
To: Andrew Meier
Subject: Re: Copyright permission request

Andrew,

I remember figures 2 and 3 – these represent one of my earliest


attempts at mastering (or more realistically making some use of)
Photoshop.

I am happy to give permission for the images to be used and to thus


support, in a small way, the free exchange of information among
researchers.

Kind regards,

John Oliver

E.2.2 Austroads
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.

‘From: Nick Koukoulas


Sent: Wednesday, 27 January 2016 10:45 a.m.
To: Kym Neaylon
Cc:; David Francis ; Elaena Gardner
Subject: RE: Permission to use copyright material in a PhD thesis

Dear Kym

Thank you for your email.

Austroads does own the copyright to all of its publications and their
contents. We are very happy for you to reference the content of our
publications and acknowledge them accordingly as you have stated.

— E-2 —
Appendix E Copyright

I trust you will be well rewarded for the completion of your PhD and I
congratulate you on the work in anticipation.

Best regards

Nick Koukoulas

Chief Executive
Austroads’

E.2.3 BITRE
Apart from any fair dealing or other statutory use permitted under the
Copyright Act 1968, the terms and conditions of reuse of the material
on this website are governed by the following:

Commonwealth of Australia permission. You are free to use


information available on the Department's website provided you
comply with the Copyright Act 1968.

(https://infrastructure.gov.au/utilities/copyright.aspx viewed 27/01/16).

Bureau of Infrastructure Transport and Regional Economics 2011, Truck productivity:


sources, trends and future prospects, report 123, Canberra, ACT.

The report contains no statement reserving copyright, and SECTION 41 Fair dealing
for purpose of criticism or review is applied.

E.2.4 Dickinson 1985


Dickinson, EJ 1985, ‘Bituminous Roads in Australia’, Australian Road Research Board,
Vermont South.

Reference to, or reproduction of, any part or this book may be made
provided a precise reference is quoted.

E.2.5 FHWA
Under government-wide regulations for grants and cooperative
agreements with state and local governments, institutions of higher
education or non-profit organizations, recipients and subrecipients and
their contractors are free to copyright any work produced with Federal
funds. As a result, "State DOTs and their subrecipients may copyright
any books, publications, or other copyrightable materials developed in
the course of the FHWA planning and research funded project. The
FHWA reserves a royalty-free, nonexclusive and irrevocable right to
reproduce, publish, or otherwise use, and to authorize others to use,
the work for Government purposes." 23 C.F.R. §420.121(b).

(http://www.fhwa.dot.gov/hep/guidance/sprpat.cfm viewed 25/01/16).

— E-3 —
Appendix E Copyright

FWHA documents used herein contain no statements reserving copyright, and


SECTION 41 Fair dealing for purpose of criticism or review is applied.

E.2.6 Herrington & Wu (2016)


Herrington, PR & Wu, J 2016, ‘Performance based specification for chip seal bitumen’,
Discussion Paper. DRAFT 14 Jan 2016, Opus Research, Lower Hutt, NZ.

From: Phil Herrington


Sent: Monday, 1 February 2016 10:46 AM
To: Kym Neaylon
Subject: RE: Request to use your material in my PhD Lit review

Hi Kym,

I confirm that I am the copyright holder for the material detailed in your
letter of the 25 January 2016 and I consent to your use of the material
as requested.

Regards

Phil

E.2.7 Highways Agency (UK)


All of the documents on this website are now available to use under
the Open Government Licence. The Open Government Licence is a
simple set of terms and conditions to enable the free re-use of
government and public sector information.

(http://www.standardsforhighways.co.uk/ha/standards/copyright.htm viewed 27/01/16).

E.2.8 Moffatt 2015


Moffatt, MA 2015, ‘The effect of multiple axle loads on the design of flexible
pavements’, PhD Thesis, Department of Civil Engineering, Monash University,
Melbourne.

Under the Copyright Act 1968, this thesis must be used only under the
normal conditions of scholarly fair dealing. In particular no results or
conclusions should be extracted from it, nor should it be copied or
closely paraphrased in whole or in part without the written consent of
the author. Proper written acknowledgement should be made for any
assistance obtained from this thesis.

From: "Michael Moffatt (Dr)"


Date: 11/02/2016 17:50 (GMT+12:00)
To: Kym Neaylon

Subject: RE: Copyright request

— E-4 —
Appendix E Copyright

Dear Kym Neaylon,

You have my permission to use the figures 2.4 and 4.7 from my PhD
thesis in your thesis in the manner in which you described in your
letter of 11 February.

Yours faithfully,

Dr Michael Moffatt

Principal Engineer, Pavements and Surfacings (PhD)


Pavements and Surfacings
ARRB Group Ltd
500 Burwood Hwy
Vermont South VIC 3133
Australia

E.2.9 NZ Transport Agency


Except where stated, the information on this website is the copyright
of the NZ Transport Agency.

The material protected by NZ Transport Agency copyright may be


reproduced for personal or in-house use without formal permission or
charge.

Requests and enquiries concerning the reproduction of information on


this website for any purpose other than personal or in house use
should be directed to the Information Manager, NZ Transport Agency,
Private Bag 6995, Wellington 6141, New Zealand, or
email info@nzta.govt.nz.

The permission to reproduce material protected by NZ Transport


Agency copyright does not extend to any material on this site that is
identified as being the copyright of a third party. Authorisation to
reproduce such material must be obtained from the copyright holders
concerned.

http://www.nzta.govt.nz/about-us/about-this-site/ viewed 27/1/16.

Permission requested 27 Jan 2016

Awaiting a reply

— E-5 —
Appendix E Copyright

E.2.10 PIARC
PIARC have reserved all rights on the ROAD NETWORK OPERATIONS &
INTELLIGENT TRANSPORT SYSTEMS publication, but have not reserved any rights
on PIARC 1987, Report of the Committee on Surface Characteristics, Technical
Committee 1 Report, World Road Congress, XVIII Brussels, Belgium.

(http://www.piarc.org/en/search.htm?q=copyright&l=all viewed 27/1/16).


(http://www.piarc.org/en/publications/Congress-Proceedings/Brussels-1987/technical-
committee-report.htm viewed 27/1/16).

PIARC 1987 and the PIARC web site contains no statement reserving copyright, and
SECTION 41 Fair dealing for purpose of criticism or review is applied.

E.2.11 Read (1996)


Read, J 1996, ‘Fatigue cracking of bituminous pavement mixtures’. PhD Thesis,
University of Nottingham.

The thesis contains no statement reserving copyright, and SECTION 41 Fair dealing
for purpose of criticism or review is applied.

E.2.12 SABITA
No article, extracts, photographs or other elements of this publication
may be reproduced in any form whatsoever without the written
permission of the Southern African Bitumen Association.

From: Saied Solomons


Sent: Thursday, 28 January 2016 11:17 PM
To: Kym Neaylon
Subject: RE: copyright request from SABITA Manual 10

Dear Kym

I confirm that the copyright vests in Sabita and that you are welcome
to use the material for the purpose requested.

All the very best.

Kind regards

M Saied Solomons

— E-6 —
Appendix E Copyright

Sabita | Postnet Suite 56 | Private Bag X21 | Howard Place 7450


Tel +27 (0)21 531 2718 | Fax +27 (0)86 263 7067 |
Co Reg No 1989/000047/08, 031-993-NPO | www.sabita.co.za

E.2.13 SANRAL
Copyright of material presented on the SANRAL web site is owned by
SANRAL. Unless otherwise stated, the reproduction, publication,
performance, communication or adaptation of the material presented
on this web site is permitted, provided that SANRAL is acknowledge
as the copyright owner. In addition, that the material is not modified or
used in a manner prejudicial to the purposes and/or reputation of
SANRAL, there are no specific overriding copyright conditions relating
to the material, and no charge, claim or encumbrance is made upon
any recipient of the material.

(http://www.nra.co.za/live/content.php?Session_ID=53892b7948d303161fd3e2d3f57bd
1c5&Item_ID=6 viewed 25/01/16).

E.2.14 State Library of Western Australia


‘From: Zofia Carter
Sent: Monday, 9 July 2012 11:11 AM
To: Amy Walsh
Subject: RE: Copyright Request - Images: Bedford Trucks

Good morning Amy

Pleas find attached Reproduction Permit as requested.

You can also order high resolution copies of the image at a cost of
$25.00 per image.

I have attached an order form should you choose to proceed.

Kind regards

Zofia Carter

Public Orders Coordinator


T (08) 9427 3450

— E-7 —
Appendix E Copyright

From: Amy Walsh


Sent: Monday, 2 July 2012 1:16 PM
To: Pictorial Orders
Subject: Copyright Request - Images: Bedford Trucks

Good afternoon,

I am contacting you on behalf of Mr Kym Neaylon, National Technical


Leader, Pavements and Surfacings, with ARRB Group in Victoria,
Australia.

ARRB Group (www.arrb.com.au) is a non-profit organisation with


more than 50 years of service providing trusted advice, technical
input, and solutions to Australian road authorities.

Mr Neaylon is currently researching the influence of heavy vehicle


single, tandem and triaxle groupings on road surfacing (sprayed seal)
wear. This has become necessary partly because of the way the
composition of Australia’s truck fleet has changed over the years.

This research will result in a) a report published soon by Austroads


and available for free download from their website, b) a conference
paper to be presented at the ARRB Conference, 25th, Perth,
September 2012, and c) PhD thesis (in progress).

Mr Neaylon requests permission to use three images (please refer to


attached SLWA Online Catalogue pdf) in a) a PowerPoint
presentation at the ARRB conference in September, b) his PhD thesis,
and c) a future Austroads technical report on this topic.

Please don’t hesitate to contact me should require any additional


information.

Thanks and regards,

Amy.

Amy Walsh

Divisional Administrator
General Management
ARRB Group Ltd
Vermont South VIC 3133
AUSTRALIA’

— E-8 —
Appendix E Copyright

— E-9 —
Appendix E Copyright

— E-10 —
Appendix E Copyright

E.2.15 Werkmeister (2003)


Werkmeister, S 2003, ‘Permanent Deformation Behaviour of Unbound Granular
Materials in Pavement Constructions’, PhD thesis, Dresden University of
Technology.

The thesis contains no statement reserving copyright, and SECTION 41 Fair dealing
for purpose of criticism or review is applied.

— E-11 —
Appendix F Turnitin originality check

F. TURNITIN ORIGINALITY CHECK


This thesis has undergone an originality check using Turnitin.

— F-1 —

Das könnte Ihnen auch gefallen