An Analysis of the Seasonal and Short-Term Variation of

Road Pavement Skid Resistance

By:
Douglas James Wilson

Department of Civil and Environmental Engineering

A Thesis submitted in fulfilment of the requirements

for the degree of Doctor of Philosophy in Engineering,
The University of Auckland
May 2006

ABSTRACT
It has been well proven that as the skid resistance of a road surfacing decreases, the number of loss of control type
crashes increases, causing road death and injuries. However, the management of skid resistance of road surfacings
continues to be difficult due to the inherent and sometimes random variation in skid resistance levels over time.
This study is an investigation and analysis of seasonal and short term variation of measured skid resistance in two
phases.
Phase 1: Regular field monitoring was undertaken using the GripTester and the Dynamic Friction Tester
measurement devices on seven sites in the Auckland and Northland Regions of New Zealand was undertaken over a
three year period. The effects of temperature, rainfall, contaminants, new surfacings, geometric elements and
aggregate properties were analysed to investigate factors that initiate changes in the measured skid resistance of
pavement surfacings.
Phase 2: Laboratory prepared samples were constructed for accelerated polishing and skid resistance testing of four
different aggregates (two greywackes, a basalt and an artificial iron-making melter slag aggregate). The samples
were polished in an accelerated polishing machine to an equilibrium skid resistance level (Stage 1 polishing).
Contaminants were then added to the accelerated polishing process to determine the effect of varying additive,
particle size and hardness in an attempt to simulate seasonal and/or short-term variations that occur in the field.
The results have demonstrated that significant and previously unpredictable variations (greater than 30%) in
measured skid resistance can occur over short time periods. These variations cannot be explained by any one factor.
They are the result of a number of inter-related factors, including the geological properties of the aggregates and the
contaminants themselves, the previous rainfall history, the road geometry, the calendar month of the year and
(depending upon the measurement device), the temperature during testing. The laboratory tests demonstrate that
accelerated polishing tests of aggregate samples could be prepared for testing by the Dynamic Friction Tester and
that significant variations in measured skid resistance could be simulated on various aggregates in the laboratory by
the addition of contaminants. The results of the testing and addition of contaminants on various aggregates resulted
in significant behavioural differences which were related to the geological properties of the aggregates themselves,
as well as the contaminants used in the accelerated polishing process.
The findings of the research have specific relevance to three areas of industry; Road Controlling Authorities who
are primarily interested in skid resistance policy, standards and management, Road Asset Managers who operate,
maintain and manage condition level and the safety aspects of the road network and Crash Investigators who collect
and analyse crash data primarily for legal proceedings. All three of these industry organisations need to clearly
understand the inherent variability of skid resistance, the factors involved and the effects that geological and
environmental variations have on skid resistance measurement.

ACKNOWLEDGMENTS
Balancing duties of being an academic staff member whilst undertaking a doctoral programme of research with an
ever patient, although sometimes a family in-waiting is a difficult task. I am certainly pleased that the light at the
end of the tunnel has indeed turned into daylight and not an oncoming train. It is therefore, with great pleasure
that I begin with acknowledgements to those that have contributed in so many ways to the culmination of many
years of research to the thesis. I gratefully acknowledge the help of many, however to name them all in some
logical prioritised order is difficult. After much deliberation, I have decided to order acknowledgements in three
particular areas being University, industry support and personal; however these categories are not given in any
prioritised order.
University of Auckland:
This research was carried out under the supervision of Associate Professor Roger C. M. Dunn (main supervisor) and
Professor Bruce W. Melville (Head of Civil and Environmental Engineering Department) to whom I express my
sincere appreciation for all their guidance, encouragement and general advice throughout. As both Roger and Bruce
are also academic colleagues, their guidance has also been valued through the research programme.
Special acknowledgement must also go to Noel Perinpanayagam who has tirelessly worked on the data collection
side aspect of this research including looking after and calibrating the testing equipment. Noel has always been very
obliging and his happy disposition always makes him a pleasure to work with. Acknowledgement is also given to
Gary Carr for electronics expertise when it was required and other contract staff and summer students who have
helped with the data collection programme. This includes, Louis Sky Bargh, Malcolm Dunn, Subodh Gidh, Lei Wu,
Wendy Chan, Vincent Tang and Caleb Phang.
I would also like to acknowledge The University of Auckland Faculty of Engineering and the Dean of Engineering
(Professor Peter Brothers) who have supported the research programme and have invested significant internal
research funds, primarily to purchase the skid testing equipment necessary to undertake this research project.
My thanks is also given to my colleagues Dr Naresh Singhal and Dr Takis Elefsiniotis who helped advise and
supervise Wei Li (William) who undertook the bulk of the environmental analysis of the contaminant samples
collected from the road surface. This aspect formed part of a student research project in the Department of Civil and
Environmental Engineering Laboratory. Thanks is also given to Dr Bill Flinn for the numerous dicussions that we
have had regarding data analysis and statistics.
Appreciation is also given to Professor Philippa Black of the University of Auckland Geology Department, who
undertook geological tests and gave geological interpretations of the effects of accelerated polishing on the
aggregate surfaces.
Industry Support:
Acknowledgement is given to members of the Advisory Committee that were external to the University for their
advice and comments, consisting of Barry Gundersen, Dr Bryan Pidwerbesky, Dr David Hutchison, Dr John
Yeaman, Mark Owen and Peter Cenek.
Specific appreciation (as outlined below) is given to the significant industry partners/supporters that have helped
support the research in many different ways.

ii

Firstly, I must give special thanks to Works Infrastructure Limited who have supported the research both
financially and with resources almost from the outset of the research programme. The field testing data collection in
the Northland Region of New Zealand would not have been possible without the support of the Whangarei Works
Infrastructure team which included Peter Houba, Peter King, Matthew Findlay, Glen Kirk and the temporary traffic
control team headed by Derek Phillips. Thanks also go to the Auckland Works Infrastructure testing laboratory
which supplied aggregate supplies for laboratory sampling and testing and whose technicians also undertook
Polished Stone Value tests on the aggregate samples. Special thanks are also given to Michael Haydon and David
Hutchison from the technical management team of Works Infrastructure who have always been willing to listen to
my requests and the many discussions that we have had on aspects of skid resistance, policy and technical matters.
Further acknowledgement is given to David Hutchison who has willingly waded through a draft of the thesis giving
very useful technical and editing feedback.
Pavement Management Services and especially Dr John Yeaman and Daniel Rich for their support for the research
programme from the beginning, and secondly for financially contributing to the purchase of the Dynamic Friction
Tester for the static field testing and the laboratory tests.
Achnowledgement is also given to The Ports of Auckland and especially Jo Campbell of the AXIS Intermodal
group that allowed access to the rail grid site as a field-testing site outside of normal port operation hours.
Transit New Zealand head office staff (Mark Owen, David Cook and Chris Parkman) are also thanked for their
support of the research and in enabling access to historic SCRIM and RAMM data on the Transit New Zealand state
Highway network.
Land Transport New Zealand (formerly Transfund New Zealand) are acknowledged for their significant financial
support of the research, in terms of the Land Transport New Zealand Research project that allowed the field research
data collection and analysis programme to continue for a further year and for the extension to the controlled
laboratory testing.
Personal:
And to the last and most important, to my special family whom I wholeheartedly want to thank and acknowledge as
being so significantly important in the culmination of this thesis. To Puti who has advised on many occasions with
regards to statistics and who has encouraged and cared for a husband who has, at times I admit, got tired and
grumpy and who has, especially in more recent months, not always been there. For Miriama and James, who for
all of their lives have put up with a dad and his writing of a boring thesis with too many words and not enough
pictures. To Miriama, for your joy and laughter and James, for your special cuddles even throughout all your trials,
pain and ill-health; you have inspired me to persevere against all odds.
To my parents of both lineages for their encouragement and especially to my mother Elayne who worked through a
draft of the thesis and who helped edit points and rules of grammar and English that in some cases I had long
forgotten.

iii

TABLE OF CONTENTS
ABSTRACT................................................................................................................................................................... i
ACKNOWLEDGMENTS ............................................................................................................................................ii
TABLE OF CONTENTS............................................................................................................................................. iv
LIST OF FIGURES ..................................................................................................................................................... ix
LIST OF TABLES .....................................................................................................................................................xiii
ABBREVIATIONS AND ACRONYMS .................................................................................................................. xiv
1

INTRODUCTION ................................................................................................................................................ 1
1.1 Overview ........................................................................................................................................................ 1
1.2 Road Safety, Risk and Policy ......................................................................................................................... 1
1.3 The Variability of Skid Resistance ................................................................................................................. 2
1.4 Terminology ................................................................................................................................................... 3
1.5 Research Objectives........................................................................................................................................ 3
1.6 Research Methodology ................................................................................................................................... 4
1.7 Organisation of the Thesis .............................................................................................................................. 4

SKID RESISTANCE FUNDAMENTALS........................................................................................................... 7

2.1 Introduction .................................................................................................................................................... 7
2.2 Function and Characteristics of Road Surfacings ........................................................................................... 7
2.2.1
Introduction ........................................................................................................................................... 7
2.2.2
Infrastructure Asset Management Goals ............................................................................................... 7
2.2.3
Monitoring Surface Characteristics and Performance........................................................................... 8
2.2.4
Pavement Deterioration Modelling in New Zealand (NZ dTIMS System) ........................................... 8
Background..................................................................................................................................................... 8
The Need for Predictive Modelling ................................................................................................................ 9
Components of Predictive Modelling ........................................................................................................... 10
Pavement Deterioration Modelling and Road Asset Management ............................................................... 11
The NZ dTIMS System and Skid Resistance................................................................................................ 13
2.3 The Traffic System and Road Safety ............................................................................................................ 15
2.3.1
Introduction ......................................................................................................................................... 15
2.3.2
Road Crash Factors ............................................................................................................................. 15
2.3.3
Skid Resistance and Crash Risk .......................................................................................................... 18
Introduction .................................................................................................................................................. 18
Before and After Studies............................................................................................................................... 18
Comparison with Norm Studies.................................................................................................................... 18
Regression Studies........................................................................................................................................ 20
Summary of Recent Crash Risk Research .................................................................................................... 21
2.3.4
Skid Resistance Standards................................................................................................................... 23
Introduction .................................................................................................................................................. 23
UK based Standards of Skid Resistance ....................................................................................................... 24
New Zealand Standards of Skid Resistance.................................................................................................. 26
Level of Skid Resistance and the Road User ................................................................................................ 27
2.4 Vehicle System Dynamics ............................................................................................................................ 27
2.4.1
Introduction ......................................................................................................................................... 27
2.4.2
Driving Dynamic Fundamentals ......................................................................................................... 28
2.5 Definitions of Skid Resistance and Friction ................................................................................................. 29
2.5.1
Introduction ......................................................................................................................................... 29
2.5.2
Components of Friction....................................................................................................................... 29
2.5.3
The Dynamics of Friction Forces........................................................................................................ 31
2.5.4
Braking and Friction............................................................................................................................ 36
2.5.5
Crash Investigations and Braking Distances ....................................................................................... 37

FACTORS AFFECTING ROAD SURFACE SKID PROPERTIES .................................................................. 39

3.1 Introduction .................................................................................................................................................. 39
3.2 Road Pavement Surface Texture profiles...................................................................................................... 39
3.2.1
Introduction ......................................................................................................................................... 39
3.2.2
Microtexture........................................................................................................................................ 41
3.2.3
Macrotexture ....................................................................................................................................... 41
3.2.4
Surface Texture, Friction and Speed ................................................................................................... 42
3.3 Aggregate Properties .................................................................................................................................... 43
iv

3.3.1
Introduction ......................................................................................................................................... 43
3.3.2
Petrological types of Aggregates used in Pavement Construction ...................................................... 44
3.3.3
Desirable properties of Aggregates for Bituminous Surfacings .......................................................... 44
3.3.4
Abrasion of Aggregates....................................................................................................................... 45
3.3.5
Polishing of Aggregates ...................................................................................................................... 46
3.3.6
Differential Wearing of Aggregates .................................................................................................... 49
3.3.7
Weathering .......................................................................................................................................... 49
3.3.8
Summary of Aggregate Sources appropriate for Surfacings ............................................................... 49
3.3.9
Aggregate Size, Shape and Spacing .................................................................................................... 50
3.4 Loading factors ............................................................................................................................................. 50
3.4.1
Traffic Intensity and Surface Age ....................................................................................................... 50
3.4.2
Road Site and Traffic Flow Conditions............................................................................................... 53
3.5 Environmental Effects on Skid Resistance ................................................................................................... 54
3.5.1
Introduction ......................................................................................................................................... 54
3.5.2
Water Film Thickness ......................................................................................................................... 54
3.5.3
Drainage Conditions............................................................................................................................ 57
3.5.4
Surface Contamination........................................................................................................................ 57
3.5.5
Temperature ........................................................................................................................................ 60
3.5.6
Seasonal Variations ............................................................................................................................. 61
3.5.7
Short-Term Variations......................................................................................................................... 64
4

SKID RESISTANCE TESTING METHODS AND MODELS.......................................................................... 68

4.1 Introduction .................................................................................................................................................. 68
4.2 Testing Device Factors ................................................................................................................................. 68
4.2.1
Introduction ......................................................................................................................................... 68
4.2.2
Vehicle Speed...................................................................................................................................... 68
4.2.3
Tyre Angle .......................................................................................................................................... 69
4.2.4
Wheel Slip Ratio ................................................................................................................................. 69
4.2.5
Tyre Characteristics............................................................................................................................. 69
Development of Modern Tyres..................................................................................................................... 69
Tyre Construction Effects............................................................................................................................. 70
Tyre Tread Design ........................................................................................................................................ 71
Tyre Tread Pattern ........................................................................................................................................ 72
Tyre Tread Compound.................................................................................................................................. 74
Tyre Tread Wear Irregularity........................................................................................................................ 77
Developments in Tyre Design ...................................................................................................................... 77
4.2.6
Tyre Contact area and Stress ............................................................................................................... 78
Contact Area ................................................................................................................................................. 78
Distribution of Contact Stress....................................................................................................................... 78
4.3 Measuring Surface Friction .......................................................................................................................... 81
4.3.1
Introduction ......................................................................................................................................... 81
4.3.2
Sideways Force Method ...................................................................................................................... 83
4.3.3
Braked Wheel Method ........................................................................................................................ 86
GripTester..................................................................................................................................................... 86
Electronic Recording Decelerometer ............................................................................................................ 87
Norsemeter Road Analyser and Recorder - ROAR .................................................................................. 88
Locked Wheel Pavement Friction Tester...................................................................................................... 90
4.3.4
British Pendulum Method ................................................................................................................... 91
4.3.5
The Dynamic Friction Tester .............................................................................................................. 92
4.3.6
Systems for On-board Measurement of Tyre-Pavement Friction........................................................ 94
4.4 Laboratory-based Surface Friction Prediction Methods ............................................................................... 95
4.4.1
Introduction ......................................................................................................................................... 95
4.4.2
The Polished Stone Value (PSV) Test................................................................................................. 95
Introduction .................................................................................................................................................. 95
The Polished Stone Value Test Procedure.................................................................................................... 96
A critical evaluation of the PSV Test ........................................................................................................... 97
PSV Test Summary....................................................................................................................................... 99
4.4.3
Profiling and Imaging Techniques for Assessing Aggregate Microtexture....................................... 100
4.4.4
Automatic Image Analysis ................................................................................................................ 101
4.5 Measuring Macrotexture............................................................................................................................. 101
4.5.1
Introduction ....................................................................................................................................... 101
4.5.2
Volumetric Texture Method.............................................................................................................. 101
4.5.3
Laser Profile Method......................................................................................................................... 102
4.5.4
Outflow Method ................................................................................................................................ 104
v

4.6 Models of Skid Resistance.......................................................................................................................... 104

4.6.1
Introduction ....................................................................................................................................... 104
4.6.2
Polished Stone Value Empirical Model ............................................................................................ 105
4.6.3
The Penn State Model ....................................................................................................................... 108
4.6.4
The Rado Model................................................................................................................................ 110
4.6.5
Seasonal and Short-Term Variation Models ..................................................................................... 111
4.6.6
The PIARC Model and the International Friction Index ................................................................... 113
4.6.7
Goldens Model................................................................................................................................. 115
4.6.8
Neural Networks to Model Skid Resistance...................................................................................... 115
5

THE EXPERIMENTAL DESIGN.................................................................................................................... 117

5.1 Introduction ................................................................................................................................................ 117
5.2 The Hypothesis ........................................................................................................................................... 117
5.3 Research Methodology ............................................................................................................................... 118
5.4 Field Test Site Characteristics .................................................................................................................... 119
5.5 Field Test frequency ................................................................................................................................... 121
5.6 Field Test Equipment and Standard Test conditions................................................................................... 122
5.6.1
Continuous Skid Resistance (microtexture) Measurements .............................................................. 122
5.6.2
Standard Skid Resistance Field Test Procedures............................................................................... 123
5.6.3
Stationary Skid Resistance Tests....................................................................................................... 123
5.6.4
Macrotexture Measurements ............................................................................................................. 124
5.7 Field Test Sites ........................................................................................................................................... 125
5.7.1
Tamaki Campus (University of Auckland) site................................................................................. 127
5.7.2
Ports of Auckland site ....................................................................................................................... 128
5.7.3
Kaiwaka Site (SH 1N)....................................................................................................................... 130
5.7.4
Brynderwyn Site (SH 1N) ................................................................................................................. 130
5.7.5
Hikurangi Site (SH 1N)..................................................................................................................... 131
5.7.6
Kara Road Site (SH 14)..................................................................................................................... 132
5.7.7
Snooks - Tatton Site (SH 14) ............................................................................................................ 133
5.8 Surface Detritus and Skid Resistance ......................................................................................................... 134
5.8.1
Introduction ....................................................................................................................................... 134
5.8.2
Detritus Sample Collection Method .................................................................................................. 134
5.8.3
Detritus Laboratory Analysis Methods ............................................................................................. 135
5.9 Laboratory Based Polishing Methodology ................................................................................................. 136
5.9.1
Introduction ....................................................................................................................................... 136
5.9.2
The experimental variables ............................................................................................................... 137
5.9.3
Laboratory Surface Samples ............................................................................................................. 138
5.9.4
Laboratory-based Texture Measurement........................................................................................... 139
5.9.5
An Accelerated Polishing Machine ................................................................................................... 139
5.9.6
Laboratory Procedures for Polishing and Skid Resistance Measurement ......................................... 142

SKID RESISTANCE FIELD TESTING RESULTS AND DISCUSSION ...................................................... 145

6.1 Introduction ................................................................................................................................................ 145
6.2 Standard Skid Resistance Field Testing Procedures ................................................................................... 145
6.2.1
Introduction ....................................................................................................................................... 145
6.2.2
GripTester Testing Procedures.......................................................................................................... 146
6.2.3
Dynamic Friction Tester Field Procedures........................................................................................ 147
6.3 Processing of the Field Test results ............................................................................................................ 148
6.3.1
Introduction ....................................................................................................................................... 148
6.3.2
Processing of Raw GripTester Results .............................................................................................. 148
6.3.3
Example Analysis of GripTester Results .......................................................................................... 150
6.3.4
Skid resistanceTimeslice Analysis .................................................................................................... 154
6.3.5
Processing of Dynamic Friction Test Results ................................................................................... 155
6.3.6
Analysis of Dynamic Friction Test Results....................................................................................... 156
6.4 Summary Field Site Results........................................................................................................................ 157
6.4.1
Introduction ....................................................................................................................................... 157
6.4.2
Tamaki Campus (University of Auckland) Site ................................................................................ 157
6.4.3
Ports of Auckland Site....................................................................................................................... 161
6.4.4
Kaiwaka Slag Site ............................................................................................................................. 164
6.4.5
Brynderwyn South Curve Site........................................................................................................... 167
6.4.6
Hikurangi Site ................................................................................................................................... 170
6.4.7
Kara Road Site .................................................................................................................................. 172
6.4.8
Snooks Tatton Site ......................................................................................................................... 174
6.5 Aggregate PSV and Measured Skid Resistance.......................................................................................... 176
vi

6.5.1

Field Testing Summary Discussion................................................................................................... 177

REPEATABILITY OF SKID RESISTANCE TESTING DEVICES ............................................................... 180

7.1 Introduction ................................................................................................................................................ 180
7.2 GripTester................................................................................................................................................... 180
7.3 Dynamic Friction Tester ............................................................................................................................. 184
7.4 Temperature Effects on Skid resistance measurement devices................................................................... 187
7.4.1
Introduction ....................................................................................................................................... 187
7.4.2
GripTester ......................................................................................................................................... 188
7.4.3
Dynamic Friction Tester.................................................................................................................... 191
7.4.4
Summary of Temperature Effects ..................................................................................................... 194
7.5 Correlation of Skid Resistance Devices...................................................................................................... 195
7.5.1
Introduction ....................................................................................................................................... 195
7.5.2
Correlation of the GripTester with the DF Tester ............................................................................. 196
7.5.3
Correlation of GripTester and SCRIM .............................................................................................. 199
7.5.4
Correlation of GripTester and SCRIM by IFI ................................................................................... 201
7.5.5
Correlation of GripTester and Dynamic Friction Tester by IFI ........................................................ 204
7.6 Christchurch Surface Friction Device Correlation Trials ........................................................................... 204
7.6.1
Introduction ....................................................................................................................................... 204
7.6.2
Main Trial Test Site .......................................................................................................................... 205
7.6.3
Reported Results ............................................................................................................................... 206
7.6.4
Further Data Analysis........................................................................................................................ 206
7.6.5
IFI Transformations .......................................................................................................................... 210
7.7 Summary Device Discussion ...................................................................................................................... 213

SEASONAL AND SHORT-TERM VARIATIONS......................................................................................... 214

8.1 Introduction ................................................................................................................................................ 214
8.2 Seasonal Variation Results ......................................................................................................................... 214
8.2.1
Introduction ....................................................................................................................................... 214
8.2.2
Auckland Asphalt Mix Sites ............................................................................................................. 216
8.2.3
Northland State Highway Site Results .............................................................................................. 218
8.2.4
Seasonal Variation Summary discussion........................................................................................... 221
8.3 Short-Term Variations ................................................................................................................................ 222
8.3.1
Introduction ....................................................................................................................................... 222
8.3.2
Rainfall Data ..................................................................................................................................... 223
8.3.3
Rainfall Functions ............................................................................................................................. 225
8.3.4
Effects of Rainfall on GripTester Device Measurements.................................................................. 226
8.3.5
Effects of Rainfall on SCRIM Device Measurements....................................................................... 231
8.3.6
Normalised GripTester Results for Northland Sites and Effects of Rainfall..................................... 234
8.4 Effects of Geometric Elements on Skid Resistance.................................................................................... 237
8.4.1
Introduction ....................................................................................................................................... 237
8.4.2
Timeslice Analysis for Geometric Elements on the Kara Road Field Site........................................ 237
8.4.3
Summary of Timeslice Analysis on Geometric Elements................................................................. 240
8.5 Washing Trials............................................................................................................................................ 240
8.5.1
Introduction ....................................................................................................................................... 240
8.5.2
GripTester Results............................................................................................................................. 241
8.5.3
British Pendulum Results .................................................................................................................. 242
8.5.4
Washing Trial Conclusions ............................................................................................................... 243

DETRITUS RESULTS AND DISCUSSION ................................................................................................... 244

9.1 Introduction ................................................................................................................................................ 244
9.2 Sampling Site Details ................................................................................................................................. 244
9.3 Suspended Solids and Particle Size Distribution ........................................................................................ 246
9.4 Heavy Metals .............................................................................................................................................. 248
9.5 Petroleum Hydrocarbons and Organic Carbon ........................................................................................... 249
9.6 Summary of Detritus Analysis.................................................................................................................... 250

10
LABORATORY POLISHING TEST RESULTS AND DISCUSSION .................................................... 252
10.1
Introduction............................................................................................................................................ 252
10.2
Geological Properties of the Aggregates................................................................................................ 252
10.2.1
Introduction ....................................................................................................................................... 252
10.2.2
Moutohora Greywacke...................................................................................................................... 253
10.2.3
Otaika Greywacke ............................................................................................................................. 254
10.2.4
Holcim Basalt.................................................................................................................................... 254
vii

10.2.5
Melter Slag ........................................................................................................................................ 255
10.3
Polishing Aggregates to Equilibrium Levels (Stage 1) .......................................................................... 256
10.3.1
Introduction ....................................................................................................................................... 256
10.3.2
Moutohora Greywacke Laboratory Sample ...................................................................................... 258
10.3.3
Holcim Basalt Laboratory Sample .................................................................................................... 259
10.3.4
Otaika Greywacke Laboratory Sample ............................................................................................. 261
10.3.5
Melter Slag Laboratory Sample ........................................................................................................ 262
10.3.6
Summary of Stage 1 Accelerated Polishing ...................................................................................... 264
10.4
Geological Properties and the Aggregates Resistance to Polishing...................................................... 265
10.4.1
Introduction ....................................................................................................................................... 265
10.4.2
Constituents of the Aggregate Samples............................................................................................. 266
10.4.3
A Geological Interpretation of the Polishing of the Sample Aggregates........................................... 267
10.5
The Results of Polishing with Contaminants (Stage 2).......................................................................... 268
10.5.1
Introduction ....................................................................................................................................... 268
10.5.2
Contaminant Additives...................................................................................................................... 269
10.5.3
Moutohora Greywacke...................................................................................................................... 269
10.5.4
Holcim Basalt.................................................................................................................................... 272
10.5.5
Otaika Greywacke ............................................................................................................................. 274
10.5.6
Melter Slag ........................................................................................................................................ 276
10.5.7
Summary of Stage 2 Accelerated Polishing with Contaminants ....................................................... 279
10.5.8
Macro Photographs of the Effects of Polishing with Contaminants.................................................. 282
10.5.9
A Geological Explanation of the Stage 2 Polishing Experiments ..................................................... 285
10.6
Summary of Controlled Laboratory Experiments .................................................................................. 286
11
RESEARCH SUMMARY.......................................................................................................................... 288
11.1
Overview................................................................................................................................................ 288
11.2
Research Aims and Objectives............................................................................................................... 288
11.3
Seasonal and Short-Term Variations of Measured Skid Resistance ...................................................... 289
11.3.1
Outline............................................................................................................................................... 289
11.3.2
Variations in the Mean Value............................................................................................................ 289
11.3.3
Seasonal Variations by Calendar Months of Year............................................................................. 292
11.3.4
Short-Term Variations related to Rainfall ......................................................................................... 293
11.4
Skid Testing Devices ............................................................................................................................. 294
11.4.1
Outline............................................................................................................................................... 294
11.4.2
Comparative Device Conclusions ..................................................................................................... 295
11.4.3
The Correlation of Devices ............................................................................................................... 296
11.5
Timeslice Analysis and Influence of Geometry on Geometric Elements .............................................. 297
11.6
Washing Trials ....................................................................................................................................... 298
11.7
Samples of Detritus Collected off the Road........................................................................................... 298
11.8
Accelerated Polishing Experiments ....................................................................................................... 299
11.8.1
Outline............................................................................................................................................... 299
11.8.2
Stage 1 Accelerated Polishing........................................................................................................... 300
11.8.3
Stage 2 Accelerated Polishing........................................................................................................... 301
11.8.4
Summary Conclusions of the Accelerated Polishing Laboratory Experiments................................. 302
11.8.5
An Analysis of the Performance of the Polished Stone Value Test .................................................. 302
12
FINAL CONCLUSIONS............................................................................................................................ 304
12.1
Overview................................................................................................................................................ 304
In-field Skid Resistance Variation:............................................................................................................. 304
Skid Testing Devices: ................................................................................................................................. 305
Laboratory Tests: ........................................................................................................................................ 305
12.2
Relevance of the Research ..................................................................................................................... 305
Road Controlling Authorities...................................................................................................................... 306
Road Asset Managers: ................................................................................................................................ 306
Crash Investigators: .................................................................................................................................... 307
12.3
Further Research .................................................................................................................................... 307
Policy and Risk Management: .................................................................................................................... 307
Skid Testing Devices: ................................................................................................................................. 307
Laboratory Tests: ........................................................................................................................................ 307
13

REFERENCES ........................................................................................................................................... 309

viii

LIST OF FIGURES
Figure 1-1: Crash risk, Road Site Category and Standard of Skid Resistance - Idealised from (Rogers & Gargett,
1991) ............................................................................................................................................................................. 2
Figure 2-1: Road Asset Management components / process (N.K. Pradhan et al., 2001)........................................... 12
Figure 2-2: Flow Diagram of Linked Progressive Pavement and Surfacing Deterioration and Associated
Maintenance Activities (modified from RIMS, 1999). ............................................................................................... 14
Figure 2-3: Interaction between driver-vehicle-roadway /environment (Lamm et al., 1999). .................................... 15
Figure 2-4: Factors Contributing to Road Crashes (adapted from Ogden and Taylor (1996) sourced from (Sabey,
1980) - UK and Treat (1980)-US)............................................................................................................................... 16
Figure 2-5: Three Level Model of the Driver Vehicle Roadway / Environment (modified from Lamm et al
(1999)) ........................................................................................................................................................................ 17
Figure 2-6: Relative Risk of a Surface being a Crash site (sourced from Cairney, (1997))........................................ 19
Figure 2-7: Risk of accidents on wet roads and Mean Summer SCRIM Coefficient (MSSC) from Rogers and Gargett
(1991).......................................................................................................................................................................... 19
Figure 2-8: Mean Crash (accident) Risk by Skid Resistance and Site Category (Viner et al., 2005) ......................... 22
Figure 2-9: Vehicle Movements (Lamm et al., 1999) ................................................................................................. 28
Figure 2-10: Mechanisms of adhesion and hysteresis (J. W. H. Oliver, 2003b) ......................................................... 30
Figure 2-11: Factors affecting the friction potential (Lamm et al., 1999)................................................................... 31
Figure 2-12: Horizontal forces between Road Surface and Motor Vehicle (Lamm et al., 1999)................................ 32
Figure 2-13: Relationship between longitudinal (tangential) and radial (side force) under varying conditions (Lamm
et al., 1999) ................................................................................................................................................................. 32
Figure 2-14: Definition of the friction potential (Lamm et al., 1999) ......................................................................... 33
Figure 2-15: The Friction Curve (Lamm et al., 1999)................................................................................................. 35
Figure 2-16: Distribution of Skidding Values G (tangential friction factor, fT) with Speed V (Lamm et al., 1999)
originally developed by Wehner (1965)...................................................................................................................... 35
Figure 3-1: Road surface wavelength spectrum, from Huschek, (1990), source Lay, (1998)..................................... 40
Figure 3-2: Illustration of Texture Ranges (Sandberg, 1999) ..................................................................................... 40
Figure 3-3: Microtexture and Macrotexture (Highways Agency, 2001)..................................................................... 41
Figure 3-4: Idealised positive and negative macrotextures (Walsh, 2000) ................................................................. 42
Figure 3-5: Relationship between microtexture, macrotexture, friction and speed (PIARC World Road Association,
2003) ........................................................................................................................................................................... 43
Figure 3-6: Scanning Electron Microscope photographs of an Aggregate in an Unpolished and Polished state
(Austroads, 2005b)...................................................................................................................................................... 47
Figure 3-7: Increase in Skid Resistance due to a decrease in traffic on UK Trunk Road A4 (Smith & Collis, 2001) 47
Figure 3-8: Effect of traffic on skidding resistance of a typical motorway standard surfacing (Rogers & Gargett,
1991) ........................................................................................................................................................................... 51
Figure 3-9: Simplified general pavement polishing model (Prowell et al. 2003)........................................................ 52
Figure 3-10: Change in Skid Resistance with time for a section of 14mm SMA (W. D. H. Woodward et al., 2005) 52
Figure 3-11: Skid Resistance variation during a rainfall event (Bird and Scott, 1936)............................................... 54
Figure 3-12: Reductions in Coefficients of Friction due to wet surfaces -adapted from (PIARC World Road
Association, 2003) ...................................................................................................................................................... 55
Figure 3-13: Model of a Wet Tyre Footprint (Jellie, 2003) ........................................................................................ 56
Figure 3-14: Variation in Skid resistance during a Rain event (Bennis & De Witt, 2003) ......................................... 58
Figure 3-15: Measured Coefficient of Friction (SCRIM) in the United Kingdom over 11 years (J.R. Hosking, 1976)
.................................................................................................................................................................................... 61
Figure 3-16: Long term skid resistance monitoring in Whangarei, NZ using British Pendulum Tester (P. D. Cenek et
al., 1999) ..................................................................................................................................................................... 62
Figure 3-17: Estimate of the seasonal variation of SFC with the reported Mean Summer SCRIM Coefficient for the
UK (Rogers and Gargett, 1991). ................................................................................................................................. 63
Figure 3-18: Seasonal Variation of skid Resistance in Western Australia, Victoria and Queensland (J.W.H. Oliver et
al., 1988). .................................................................................................................................................................... 64
Figure 3-19: Influence of Rainfall on skid resistance - (Haydon, 2005) reproduced from (Hill & Henry, 1981)....... 65
Figure 3-20: Effects of pollutants and days since last rain on measured skid resistance (Hill & Henry, 1981).......... 66
Figure 4-1: Bias and Radial Tyre Construction (The Wheel and Tyre Bible, 2006)................................................... 71
Figure 4-2: Elements of tread design (Tyre Talk, 2006) ............................................................................................. 71
Figure 4-3: A Comparison of Tread Pattern Groove Volume ( V ) Performance on two surfaces (Veith, 1983)....... 74
Figure 4-4: Wet Friction versus Durometer Hardness for several Tread Rubbers (Veith, 1983)................................ 74
Figure 4-5: Wet Traction rating versus Tg to R ratio (Veith, 1983) .......................................................................... 75
Figure 4-6: Wet Traction versus Speed for several Tread Rubbers (Veith, 1983) ..................................................... 76
Figure 4-7: Cross Section of tyre with low inflation pressure under heavy load (Lippman, 1985) ............................ 79
ix

Figure 4-8: Contact stresses imposed on the road surface within the tyre contact patch: (a) static tyre and (b) free
rolling tyre (Lippman, 1985)....................................................................................................................................... 79
Figure 4-9: Classification of Skid Resistance Measuring contact Methods with common examples. ........................ 82
Figure 4-10: Relationship of Slip Speed to friction on a road surface (Austroads, 2005b)......................................... 82
Figure 4-11: Sideway Force Coefficient Method........................................................................................................ 84
Figure 4-12: WDM SCRIM Apparatus....................................................................................................................... 84
Figure 4-13: Early Skid Resistance measurement devices UK - 1930s to 1960s (P. G. Roe & Sinhal, 2005b)...... 85
Figure 4-14: SCRIM Wheel Assembly Unit (P. G. Roe & Sinhal, 2005a) ................................................................ 86
Figure 4-15: The GripTester (Findlay Irvine Ltd, 2005)............................................................................................. 87
Figure 4-16: SkidMan (Bullas, 2005) ......................................................................................................................... 88
Figure 4-17 Output from Electronic Recording Decelerometer (SkidMan) showing a comparison of ABS and non
ABS braking (Bullas, 2005)........................................................................................................................................ 88
Figure 4-18: Road Analyser and Recorder - ROAR schematic (Pavement Management Services, 2005) ................. 89
Figure 4-19: ROAR device with testing wheel and hydraulic braking mechanism demonstrated. ............................. 89
Figure 4-20: Dynatest 1295 Pavement Friction Tester (Dynatest, 2005).................................................................... 90
Figure 4-21: Locked Wheel Trailer Testing mechanism (Dynatest, 2005) ................................................................. 90
Figure 4-22: Computer Screen Test Output from Dynatest 1295 Pavement Friction Tester (Dynatest, 2005)........... 91
Figure 4-23: British Pendulum Tester (Austroads, 2005b) ......................................................................................... 91
Figure 4-24: Schematic of British Pendulum Tester (BPT) set up for Laboratory Testing (A. Woodside &
Woodward, 2002)........................................................................................................................................................ 91
Figure 4-25: Dynamic Friction Tester (DF Tester) Components (Nippo Sangyo Co Ltd., 2005) .............................. 94
Figure 4-26: Dynamic Friction Tester (DF Tester) rubber sliders .............................................................................. 94
Figure 4-27: Prepared test sample showing chips embedded in epoxy (Transit NZ & Roading NZ, 2005). .............. 96
Figure 4-28: Accelerated Polishing Machine.............................................................................................................. 97
Figure 4-29: Schematic of the Accelerated Polishing Machine (Woodside and Woodward, 2002). .......................... 97
Figure 4-30: Plot of PSV and AAV for arenaceous rocks (A. Woodside & Woodward, 2002) ................................. 98
Figure 4-31: Definitions of microtextural parameters (Forster, 1989) ...................................................................... 100
Figure 4-32: Sand patch method for measuring texture depth (A. Woodside & Woodward, 2002) ......................... 102
Figure 4-33: Laser Sensors for Texture Measurements (Millard, 1993) ................................................................... 103
Figure 4-34: Definition of Mean Profile Depth......................................................................................................... 103
Figure 4-35: Underestimate of NZ Skid Resistance based on UK PSV Equation - sourced from (Haydon, 2005)106
Figure 4-36: Penn State Model for two surfaces with differing textures (J J Henry, 2000)...................................... 109
Figure 4-37: Rado Model plot with typical values (J J Henry, 2000) ....................................................................... 110
Figure 4-38: A comparison of skid resistance measurements corrected for short-term variations (Hill & Henry, 1981)
.................................................................................................................................................................................. 113
Figure 5-1: University of Auckland GripTester and 4WD Tow Vehicle. ................................................................. 122
Figure 5-2: GripTester Automatic Watering System and Water Storage Tank in back of testing vehicle................ 122
Figure 5-3: Auckland Field Test Location Sites (source NZMS 262 Map series) .................................................... 126
Figure 5-4: Northland Field Test Location Sites (source ARC GIS) ........................................................................ 127
Figure 5-5: Tamaki Campus (University of Auckland) Test site .............................................................................. 128
Figure 5-6: Ports of Auckland Rail Grid site (source Auckland City Council aerial photo maps) ........................... 129
Figure 5-7 : Kaiwaka Slag Site (State Highway One Northland, RP 220/9.27 220/9.87)................................... 130
Figure 5-8: Brynderwyn South Curve (State Highway One Northland RP 220/0.36 220/0.532) ....................... 131
Figure 5-9 : Hikurangi Site (State Highway One Northland RP 144/6.215 144/6.766)...................................... 132
Figure 5-10 : Kara Road Site (State Highway 14 Northland RP 0/8.442 to 0/9.099) ............................................ 133
Figure 5-11 : Snooks - Tatton Site (State Highway 14 Northland RP 0/13.91 to 15/0.647) .................................. 133
Figure 5-12: Pavement surface detritus sample collection method........................................................................... 135
Figure 5-13: The UoA DFTester on a Prepared Sample ........................................................................................... 137
Figure 5-14: Rubber Sliders and Rotating Disk of the UoA DFTester ..................................................................... 137
Figure 5-15: Preparation of the Laboratory Surface Samples ................................................................................... 139
Figure 5-16: Schematic Front Elevation of the Accelerated Polishing Machine ...................................................... 141
Figure 5-17: The Accelerated Polishing Machine (left) with DF Tester (right)........................................................ 141
Figure 5-18: The Accelerated Polishing Machine wheel assembly unit in operation ............................................... 141
Figure 5-19: Stages of Accelerated Polishing modified from Prowell et al. (2003) ................................................. 142
Figure 6-1: Flow Chart of the Process developed for Skid Resistance Field Tests................................................... 146
Figure 6-2: Screen shot of the raw Access database file structure for a GripTester Survey ..................................... 149
Figure 6-3: GripTester Raw Markers Access database Table ................................................................................... 149
Figure 6-4: GripTester Raw Data Access database Table......................................................................................... 149
Figure 6-5: GripTester Pre-Survey Screen................................................................................................................ 150
Figure 6-6: Example Plot of the Coefficient of Friction (GN) in relation to Distance (chainage) at Kaiwaka in
Northland on SH 1 .................................................................................................................................................... 151
Figure 6-7: Example Statistical Plot of the GripTester survey data at Kaiwaka in Northland on SH 1 .................... 152
Figure 6-8: Statistical Plot of the GripTester survey at Kaiwaka when the New Seal was 2.5 months old .............. 153
Figure 6-9: The Variation of GripTester runs at the Kaiwaka site over 26 month period......................................... 153
x

Figure 6-10: The Variation in Average GN, SFC and Macrotexture over Time at the Kaiwaka site over 26 month
period ........................................................................................................................................................................ 154
Figure 6-11: Typical Processed Dynamic Friction Test Result showing three test runs ........................................... 156
Figure 6-12: Tamaki Campus Summary results over time (Flat section, both directions) ........................................ 158
Figure 6-13: Tamaki Campus Summary results over time (Slope section, both directions) ..................................... 159
Figure 6-14: Ports of Auckland site Summary results over time (West direction).................................................... 162
Figure 6-15: Ports of Auckland site Summary results over time (East direction)..................................................... 163
Figure 6-16: Kaiwaka Slag Site Summary Results over Time (Both directions)...................................................... 165
Figure 6-17: Brynderwyn South Curve Site Summary results over time (Both directions) ...................................... 168
Figure 6-18: Hikurangi Control Site Summary results over time (Both directions) ................................................. 171
Figure 6-19: Kara Road Site Summary results over time (Both directions).............................................................. 173
Figure 6-20: Snooks Tatton Site Summary results over time (Both directions)..................................................... 175
Figure 6-21: Measured Skid resistance over Time in Comparison to Published PSV for Northland Sites............... 177
Figure 7-1: Multiple runs of the GripTester on the same day and same site............................................................. 181
Figure 7-2: Box-plot of GripNumber (GN) showing Mean, Minimum and Maximum GripTester values in
comparison to the Age of the Seal, Kaiwaka (Increasing Direction) ........................................................................ 182
Figure 7-3: GripTester CoV between runs in relation to Age of Seal, Kaiwaka (Increasing Direction)................... 182
Figure 7-4: GripTester Coefficient of Variation between runs in relation to Number of Days since March of each
year, Kaiwaka (Increasing Direction) ....................................................................................................................... 183
Figure 7-5: GripTester Coefficient of Variation between runs for all Northland sites.............................................. 183
Figure 7-6: A comparison of run to run variability of the GripTester and DF Tester devices .................................. 185
Figure 7-7: Tamaki Campus DFT 20 and CoV by calendar month of the year. ....................................................... 185
Figure 7-8: Ports of Auckland DFT 20 and CoV by calendar month of the year...................................................... 186
Figure 7-9: Normalised DFT20 results and CoV by Calendar month of the year for Auckland Sites. ..................... 187
Figure 7-10: Normalised GripTester results and recorded Surface Temperature on Asphalt Mix surfaces.............. 189
Figure 7-11: Normalised GripTester results and recorded Surface Temperature on Northland chip seal Sites........ 189
Figure 7-12: Normalised GripTester results and recorded Ambient Temperature on Asphalt Mix surfaces ............ 190
Figure 7-13: Normalised GripTester results and recorded Ambient Temperature on Northland chip seal Sites ...... 190
Figure 7-14: GripTester Coefficient of Variation and Surface Temperature of Northland sites............................... 191
Figure 7-15: Surface Temperature effects on the DF Tester device.......................................................................... 192
Figure 7-16: A comparison of the effects of Surface Temperature on the GripTester and DF Tester devices. ........ 193
Figure 7-17: The effects of Temperature and Change of Rubber Sliders on DF Tester Results ............................... 194
Figure 7-18: Comparison of the GripTester and DF Tester results on an Asphalt Mix ............................................ 197
Figure 7-19: Correlation between the GripTester and DF Tester at 1m interval spacings ........................................ 197
Figure 7-20: Correlation between the GripTester and DF Tester for all sites ........................................................... 198
Figure 7-21: Example GripTester and SCRIM survey for the Snooks Tatton Northland site................................... 199
Figure 7-22: A Comparison of Correlation relationships for the SCRIM and GripTester ........................................ 201
Figure 7-23: International Friction Index (IFI) F60 and Sp for Kaiwaka Slag site against time ............................... 203
Figure 7-24: IFI F60 correlation between the GripTester and SCRIM devices ........................................................ 203
Figure 7-25: IFI F60 Correlation between the GripTester and Dynamic Friction Tester devices............................. 204
Figure 7-26: Schematic diagram of Christchurch Trial Test site (Austroads, 2005a)............................................... 205
Figure 7-27: GripTester results as compared with the Dynamic Friction Tester (Christchurch trials) ..................... 207
Figure 7-28: All devices compared with the Dynamic Friction Tester (Christchurch trials) .................................... 208
Figure 7-29: GripTester correlated against SCRIM on Christchurch test trial sites.................................................. 209
Figure 7-30: Dynamic Friction Tester and In-vehicle Decelerometers ..................................................................... 210
Figure 7-31: Christchurch Trial surfaces Raw CoF and Transformed IFI F60 ......................................................... 211
Figure 7-32: Correlation of the DF Tester device with other devices using IFI F60 Transformations ..................... 212
Figure 7-33: IFI F60 Correlation Results of the Australian Friction Workshop Between the DF Tester, SCRIM,
GripTester and ROAR - Dardano (2005).................................................................................................................. 213
Figure 8-1: Seasonal Variation of Auckland Asphalt Mix Surfaces over 3 Year Period .......................................... 216
Figure 8-2: The Normalised GripTester Results (NGN) for the Auckland Asphalt Mix Sites by Month of Year .... 217
Figure 8-3: Seasonal Variation of Northland Chipseal Surfaces over 2 year Period, (Increasing direction) ............ 219
Figure 8-4: Seasonal variation of Northland Chipseal Surfaces over 2 year period, (Decreasing direction) ............ 219
Figure 8-5: The Normalised GripTester Results (NGN) for the Northland Chipseal sites by Month of Year .......... 220
Figure 8-6: The Normalised GripTester Results (NGN) for the Northland Chipseal sites by Quarter of Year......... 221
Figure 8-7:Measured Decline in Skid Resistance due to Environmental Effects (P.D. Cenek & Davies, 2004)...... 223
Figure 8-8: Measured increase in skid resistance during rainfall event (Cenek et al., 2004).................................... 223
Figure 8-9: Days Since Last Rainfall and GripNumber (Hikurangi site) for Various Depths of Rainfall ................ 227
Figure 8-10: Dry Spell Factor (DSF) and GripNumber (Hikurangi site) for various depths of rainfall.................... 229
Figure 8-11: Weighted Rain Function (WRF) and GripNumber (Hikurangi site) for Various Rainfall Periods ...... 231
Figure 8-12: DSLR and DSF Analysis with the SCRIM device (Hikurangi site)..................................................... 232
Figure 8-13: Weighted Rain Function (WRF) and SCRIM SFC (Hikurangi site) for Various Rainfall Periods ...... 234
Figure 8-14: Days Since Last Rainfall (DSLR>5mm) and Normalised GripNumber (NGN) for Northland sites ... 235
Figure 8-15: Dry Spell Factor (DSF>5mm) and Normalised GripNumber (NGN) for Northland sites ................... 236
xi

Figure 8-16: Weighted Rain Function (WRF07) and Normalised GripNumber (NGN) for Northland sites ............ 236
Figure 8-17: Performance of the New Seal in Comparison to the Old Adjacent Seal (Kara Road).......................... 238
Figure 8-18: Performance of a Straight with Curved Geometrics (Kara Road) New Seal..................................... 239
Figure 8-19: Performance of a Level section in comparison with -6% Gradient (Kara Road) - New Seal............... 239
Figure 8-20: Results of Washing Trials on Hikurangi Site, Northland (Pre and Post Washing) .............................. 241
Figure 8-21: Before and After Results of the Washing Trials (decreasing direction) ............................................... 242
Figure 9-1: Mean Sediment Load at Each Site (mg/m2) ........................................................................................... 246
Figure 9-2: Mean Particle Size Distributions of Solid Samples................................................................................ 247
Figure 9-3: Distributions of road runoff sediment sizes from literature - Modified from Walker et al. (1999) and Ng
et al. (2003) ............................................................................................................................................................... 248
Figure 9-4: Metal Concentration of Each Sample..................................................................................................... 248
Figure 9-5: Carbon Load in Sediments (mg/g) ......................................................................................................... 249
Figure 9-6: TOC Concentration of each sample (mg/L) ........................................................................................... 250
Figure 10-1: Moutohora Sandstone Chips ................................................................................................................ 253
Figure 10-2: Microphoto of Typical Otaika Greywacke Sandstone Thinsection(diameter field of view is 2mm) ... 254
Figure 10-3: Microphoto of Typical Holcim Basalt Thinsection (diameter field of view is 2mm) .......................... 255
Figure 10-4: Iron and Steel Making Aggregates in New Zealand (SteelServ Ltd & Bourke, 2005)......................... 255
Figure 10-5: Microphoto of Typical Melter Slag Thinsection (diameter field of view is 2mm)............................... 256
Figure 10-6: Stage 1 Polishing of Moutohora Greywacke Aggregate ...................................................................... 258
Figure 10-7: Moutohora Greywacke Sandstone Chips (Unpolished and Polished states) ........................................ 259
Figure 10-8: Stage 1 Polishing of Holcim Basalt Aggregate .................................................................................... 260
Figure 10-9: Holcim Basalt Chips (Unpolished and Polished States)....................................................................... 260
Figure 10-10: Stage 1 Polishing of Otaika Greywacke ............................................................................................. 261
Figure 10-11: Otaika Greywacke Sandstone Chips (Unpolished and Polished states) ............................................. 262
Figure 10-12: Stage 1 Polishing of Melter Slag Aggregate ...................................................................................... 263
Figure 10-13: Melter Slag Chips (Unpolished and Polished States) ......................................................................... 263
Figure 10-14: Summary of Accelerated Polishing of all Laboratory Samples to ESR (Stage 1) .............................. 265
Figure 10-15: Moutohora Greywacke Laboratory Sample (Stages 1 and 2 Polishing Phases) ................................. 270
Figure 10-16: Moutohora Greywacke Laboratory Samples (Stage 2 Polishing with Addition of Contaminants) .... 271
Figure 10-17: Holcim Basalt Laboratory Sample (Stages 1 and 2 Polishing Phases) ............................................... 272
Figure 10-18: Holcim Basalt Laboratory Samples (Stage 2 Polishing with Addition of Contaminants) .................. 274
Figure 10-19: Otaika Greywacke Laboratory Sample (Stages 1 and 2 Polishing Phases) ........................................ 275
Figure 10-20: Otaika Greywacke Laboratory Samples (Stage 2 Polishing with Addition of Contaminants) ........... 276
Figure 10-21: Melter Slag Laboratory Sample (Stages 1 and 2 Polishing Phases) ................................................... 277
Figure 10-22: Melter Slag Laboratory Samples (Stage 2 Polishing with Addition of Contaminants) ...................... 278
Figure 10-23: Skid Resistance DFT () and the Effects of the Addition of Contaminants on the 4 Lab Samples ... 280
Figure 10-24: Skid Resistance DFT () and the Effects of the Addition of Contaminants and the % Difference from
the Initial SR value for the 4 Lab samples ................................................................................................................ 282
Figure 10-25: Moutohora - After Polishing with Various Contaminants (approx. field of view 3.5x 2mm)............ 283
Figure 10-26: Holcim basalt - After Polishing with Various Contaminants (approx. field of view 3.5x 2mm)) ...... 283
Figure 10-27: Otaika Greywacke - After Polishing with Various Contaminants (approx. field of view 3.5x 2mm))
.................................................................................................................................................................................. 284
Figure 10-28: Melter Slag - After Polishing with various contaminants (approx. field of view 3.5x 2mm)) ........... 285

xii

LIST OF TABLES
Table 2-1: Pavement Surfacing Parameters requiring Performance Monitoring (Austroads 2003).............................. 8
Table 2-2: Development Streams of Predictive Modelling - modified from (Pradhan et al. 2001) ............................ 11
Table 2-3: Diagrammatic Representation of Investigatory Levels in HD 28/04 (Haydon 2005)................................ 26
Table 2-4: Investigatory Levels (IL) for identifying sites requiring inspection (Transit New Zealand 2002a) .......... 27
Table 3-1: Factors affecting the skid properties of a road........................................................................................... 39
Table 3-2: Geological properties of aggregate origins related to resistance to abrasion ............................................. 46
Table 3-3: Geological Properties of Aggregate Origins Related to Resistance to Polishing....................................... 48
Table 4-1: Functions of Tyre tread elements (Tyre Talk, 2006) ................................................................................. 72
Table 5-1: Field Testing Sites - Location and General details .................................................................................. 120
Table 5-2: Field Testing Sites - Surface Material characteristics, properties and Loads .......................................... 120
Table 6-1: Tamaki Campus Site Descriptive Statistics (Flat Section) ...................................................................... 159
Table 6-2: Tamaki Campus Site Descriptive Statistics (Slope Section) ................................................................... 160
Table 6-3: Ports of Auckland field site Summary Statistics (both directions) .......................................................... 163
Table 6-4: Kaiwaka Slag site Summary Statistics (both directions) ......................................................................... 166
Table 6-5: Brynderwyn South curve site Summary Statistics (both directions)........................................................ 168
Table 6-6: Hikurangi site Summary Statistics (both directions) ............................................................................... 171
Table 6-7: Kara Road site Summary Statistics (both directions) .............................................................................. 174
Table 6-8: Snooks - Tatton Road site Summary Statistics (both directions)............................................................. 176
Table 6-9: Summary Coefficient of Friction Statistics of the Field Sites.................................................................. 178
Table 7-1: PIARC IFI Transformation Coefficients (Wambold et al., 1995) ........................................................... 202
Table 7-2: Coefficient of Determination (R2) between surface friction devices from Christchurch trials ................ 209
Table 8-1: Normalised GripTester results for Auckland Sites by Calendar Month of Year ..................................... 218
Table 8-2: Normalised GripTester results for Northland Sites by Calendar Month of the Year............................... 221
Table 8-3: Correlation of NIWA Rainfall Gauges with Manually Recorded Data ................................................... 224
Table 9-1: Detritus Sampling Details (2005) ............................................................................................................ 244
Table 9-2: Preliminary Study Detritus Sampling Details (2004) .............................................................................. 245
Table 9-3: Preceding 15-day Rainfall Depth (mm)................................................................................................... 245
Table 9-4: Mean Particle Size Distribution Data of Each Site .................................................................................. 247
Table 10-1: Laboratory Test Sample Details ............................................................................................................ 257
Table 10-2: Stage 1 Wet Polishing Results until '
Equilibrium Level'is reached ...................................................... 257
Table 10-3: Major Mineral Constituents of the Aggregates used in the Laboratory Experiments (Black, 2005b) ... 267
Table 10-4: DFT() and % Difference in Results of the Stage 2 Addition of Contaminants and Accelerated Polishing
.................................................................................................................................................................................. 279

xiii

ABBREVIATIONS AND ACRONYMS

95% Conf

95% Confidence value

AADT Annual Average Daily Traffic Volume

AAV

Aggregate Abrasion Value

ABS

Anti-lock Braking System

AC

Asphaltic Concrete

AC14

Asphalt Concrete 14mm maximum particle size

ADT

Average Daily Traffic

AGD

Average Greatest Dimension of the sealing chip (mm)

AIV

Aggregate Impact Value

ALD

Average Least Dimension of the sealing chip (mm)

AP

Accelerated Polishing

APM

Accelerated Polishing Machine

ARRB ARRB Group Limited (formerly ARRB Transport Research, Melbourne, Australia)
ASTM American Society for Testing & Materials
AVE

Average

BC/BCR

Benefit / Cost ratio

BFC

Braking Force Coefficient

BPN

British Pendulum Number

BS

British Standard

BPT / PBPT

Portable British Pendulum Tester

CEN

Committee for European Normalisation

CoF

Coefficient of Friction

CoV

Coefficient of Variation

CVD

Number of HCVs / lane / day

3.5 tonnes

Dec/Decreasing Linear referencing position - Decreasing direction of travel

DETR Department of the Environment Transport and Regions (UK)
DF Tester

Dynamic Friction Tester device

DFT20 Dynamic Friction Tester CoF at 20km/h slip speed

DMRB Design Manual for Roads and Bridges (UK)
dTIMS Deigtons Total Infrastructure Management System
elv

Equivalent light vehicles

EN

European Normative

ESC

Equilibrium SCRIM Coefficient

ESR

Equilibrium Skid Resistance

GN

Grip Number (CoF from GripTester device)

GN50

Grip Number at 50 km/h

GT

GripTester device

HCV

Heavy Commercial Vehicles

HDM-III

Highway Design and Maintenance Standards model version III

HDM-4 Highway Development & Management model version 4

HMA

Hot Mix Asphalt

xiv

HRA

Hot rolled asphalt (UK)

HSD

High Speed Data

HSE

Health and Safety in Employment Act 1992

HSTM High Speed Texture Meter

IPENZ NZ Institution of Professional Engineers
ISO

International Standards Organisation

Inc/Increasing Linear referencing position - Increasing direction of travel

km

kilometres

km/h

kilometres per hour

LA

Local Authority

LCV

Light Commercial Vehicles

LGA

Local Government Authorities

LTMA Land Transport Management Act 2003

LTSA

Land Transport Safety Authority

LWP

Left Wheel Path

DF Tester CoF value

MAX

Maximum value

MDE

Micro-deval test

MIN

Minimum value

MPD

Mean Profile Depth

MSSC Mean Summer SCRIM Coefficient

MTD

Mean Texture Depth

MWD

Ministry of Works and Development

NAASRA

National Association of Australian State Roads Authority

NOMAD

National Optimisation of Maintenance Activity by Decade

NRB

National Roads Board

NZ

New Zealand

NZ dTIMS

Implementation of Predictive Modelling for Road Management software

NZS

New Zealand Standard

OGA

Open Graded Asphalt

OGPA Open Graded Porous Asphalt

PCC

Portland Cement Concrete

PMS

Pavement Management Systems Ltd

PSD

Particle Size Distribution

PSMC Performance Specified Maintenance Contract

PSV

Polished Stone Value

QA

Quality Assurance

Coefficient of correlation
2

Coefficient of determination

RAMM Road Assessment & Maintenance Management

RCA

Road Controlling Authorities

RIMS

Road Infromation Management System

RMA

Resource Management Act 1991

xv

RNZ

Roading New Zealand

ROAR Road Analyser and Recorder machine

rpm

revolutions per minute

RRL

Road Research Laboratory (UK)

RRU

Road Research Unit

RS

Reference Station

RWP

Right Wheel Path

SC

Sealing Chip

SCRIM Sideway-force Coefficient Routine Investigation Machine

SFC

Sideway Friction Coefficient

SFC50 Sideway Force Coefficient at 50km/hr

SLP

Stationary Laser Profilometer

SMA

Stone Mastic Asphalt

SMTD Sensor Measured Texture Depth

SN40,SN64

Skid number at 40mph, Skid Number at 64 km/h (US locked wheel tester)

SNZ

Standards New Zealand

SRV

Skid Resistance Value

STDEV Standard Deviation

t

tonne

Td

Texture (macrotexture) depth (mm)

Tf

Traffic factor

TMP

Traffic Management Plan

TNZ

Transit New Zealand

TOC

Total Organic Carbon

TPH

Total Petroleum Hydrocarbons

TRRL

Transport and Road Research Laboratory (UK)

TRL

Transport Research Laboratory (UK)

UK

United Kingdom

US, USA

United States of America

v/l/d

vehicles per lane per day

vpd

vehicles per day

VTI

The Swedish National Road and Transport Research Institute

xvi

Chapter 1

1 INTRODUCTION
1.1

Overview

Wet pavement crashes continue to be a major concern of most road controlling agencies around the world. Research
has shown that low skid resistance results in increased numbers of wet pavement crashes (Rogers & Gargett, 1991).
There is also some evidence that reduced skid resistance may be associated with increases in dry pavement crashes
(Karmel & Gartshorne, 1982). Numerous studies have demonstrated the significant safety benefits of a targeted
approach of improving skid resistance in high risk areas where frequent braking takes place rather than attempting to
improve skid resistance to high levels over the whole network (Austroads, 2005b). In recognising the importance of
providing safe pavement surfacings for travel during wet weather, most state or national state highway controlling
authorities (such as Transit New Zealand) in developed countries have skid resistance standards. These standards
specify high levels of skid resistance at locations where there is frequent braking in response to unexpected events,
such as approaches to intersections or pedestrian crossings. This high standard of skid resistance at high risk areas
is then progressively reduced to relatively lower levels of skid resistance on straight and event free locations.
Transit New Zealand introduced a national skid resistance specification, TNZ T/10 in the mid 1990s. This was
based on a modified form of the United Kingdom skid resistance standard, itself the result of many decades of
research. Transit New Zealand, as with many state highway agencies, measures the skid resistance of their road
network with one or more skid resistance measurement devices, generally on an annual basis. Some Transit NZ
road networks are managed by external road network managers such as the Transit NZs Northland Region
Performance Specified Maintenance Contract 002 (PSMC 002). The performance of these contracts is measured
annually by Key Performance Measures (KPMs). The provision of skid resistance levels specified in TNZ T/10 is
one such KPM. The Contractors experience of this is that, unlike other more predictable pavement deterioration
indicators, skid resistance levels are very difficult to control. This has primarily been due to the diffculty in
predicting the measured skid resistance levels at various points in time (notably, at the time of the network survey).
Consequently, the level of deficiency on a specific road, let alone across the whole network, is very difficult to
predict. This leads to great difficulty in managing an appropriate level of skid resistance across a network of roads.

1.2

Road Safety, Risk and Policy

There is no doubt that as the skid resistance of a road surface decreases, the accident rate increases. This significant
road safety issue is amplified when the pavement is wet because the contact between the tyres and the road is
reduced and water acts as a lubricant.

The resistance to skidding of a road surface is therefore one of the

fundamental requirements that highway engineers must consider in pavement design to provide a safe travelled
surface. Whether the surfacing is the wearing course of an asphalt mix or a chip sealed surfacing, the skid resistance
is governed by the properties of the aggregate.
Different sites present different risks, and consequently, in an attempt to obtain a relatively constant risk of a wet
skidding crash across a road network, Road Agencies (initially in the UK) attempted to align the skid resistance of
the site with the risk or need for heavy braking. From this a number of site categories describing the situations
found on a road network and the minimum specified level of skid resistance for each category, have been developed.
Figure 1-1 demonstrates in an idealised form how the minimum desirable standards of skid resistance for various
road site categories have been established based upon the risk of a skidding crash. A high minimum standard of
skid resistance at high risk areas (e.g. the approaches to traffic signals) is specified which is then progressively
reduced to relatively low levels on straight and event free locations (e.g. on a dual carriageway).
1

Chapter 1

Notes:

* denotes the desirable minimum level of skid resistance standard for each road site category before the
crash risk significantly increases.

Figure 1-1: Crash risk, Road Site Category and Standard of Skid Resistance - Idealised from (Rogers &
Gargett, 1991)
Transit New Zealands State highway standard is specified in TNZ T/10 (Transit New Zealand, 2002) and was first
implemented in 1995. New Zealand has a high percentage of hilly to mountainous terrain and is geologically young.
Therefore, the roads are characterised by curvilinear geometrical alignments, poor local aggregates and variable
environmental/climatic conditions in comparison to many other countries. Consequently, loss of control type
crashes are over-represented. A comparative study of loss of control type crashes on NZ state highways, from the
Land Transport New Zealand Crash Analysis System (CAS) database for the five year period prior to the first full
State Highway Road Network SCRIM survey (1991 to 1995) and the five year period from 1999 to 2003, is reported
in Wilson and Hutchison (2004). The analysis was undertaken as a means of measuring the performance of the TNZ
T/10 policy that was implemented after the 1995 State Highway SCRIM road network survey and has since been
repeated on an annual basis. The analysis demonstrated that surfacing improvements to the road network resulted in
a significant reduction of up to 25% of loss of control type crashes between the two periods (i.e. for those crashes
that could be attributed to poor skid resistance). This improvement is substantial. However, the crash reduction
improvements were less than those due to all movements and road types in NZ, which during this same period,
decreased by 38%.
Transportation engineering professionals must do better as current rates of road crashes, and specifically loss of
control crashes which result in road fatalities, are at levels that are unacceptable and do not fit with current Road
Agencys road safety targets nor road users expectations. To do better, the factors that vary the measured skid
resistance over time must be better understood to enable appropriate risk management practices to be incorporated
into current policies. Road Asset Managers need to be able to better predict when the skid resistance of an existing
surfacing has failed, or is likely to fail. Furthermore, they need to have a raft of treatment options that will last an
appropriate life-cycle, which are also economically efficient and effective.

1.3

The Variability of Skid Resistance

The skid resistance available at any particular time, either to the road user or as tested dynamically by a friction
tester, depends upon many variables. It is also not constant nor easily predictable over time. The variables that can
affect the measured level of skid resistance and their inter-relationships are many and complex. They can be
grouped under four main catgeories:

Chapter 1
Surface aggregate factors, (e.g. geological properties of the aggregate, surface microtexture and
macrotexture, chip size and shape and type of surfacing);

Load factors, (e.g. surface age, traffic intensity, composition and flow conditions, and road geometry);

Environmental factors, (e.g water film thickness, surface contamination, temperature, seasonal and shortterm rainfall effects); and

Vehicle factors, (e.g. vehicle speed, angle of tyres, wheel slip ratio, tyre characteristics, tread depth and
patterns).

1.4

Terminology

Cenek et al. (2003b) considers the terms road surface friction and skid resistance are often used interchangeably
by those involved with the management and maintenance of road networks. There is a subtle difference between the
two variables.
As explained in Chapter 2, road surface friction refers to the resistive force that is developed between a specific road
surface under particular conditions. As the frictional force provided by the tyre-road combination is proportional to
the load applied, it is normalised by the load to give a Coefficient of Friction (CoF).
The CoF is affected by a large number of variables as outlined in Section 1.3 above. In general terms, when a
surface is dry, the CoF is normally high and adequate for most normal vehicle manoeuvres. However, when the
road surface is wet, the CoF decreases significantly and becomes more dependent on the condition of the tyre and
the road surface.
Skid resistance is the term used to describe the contribution that the road makes to the development of tyre-road
friction. It is essentially a measurement of the CoF obtained under standardised conditions in which various
variables are controlled so that the effects of the road surface characteristics can be isolated. Skid resistance, as with
surface friction, are high in dry conditions and so the term skid resistance is almost always used in the context of wet
road surfaces.
Drag factor is a term that is often used in accident reconstruction work and exists when two surfaces are in contact
and move relative to each other. When gravity supplies the normal force, the term is synonymous with the
coefficient of friction (CoF) if the two surfaces are horizontal. The drag factor will differ from the CoF when
measured on inclines as the drag factor alone measures the combined influence of the coefficient of friction and
the slope of the incline. By measuring on an incline, the results will reflect the effect of the incline in overcoming
the friction forces between the two surfaces. This could be important in accident reconstruction because it will
provide a direct measure of the actual drag factor at work, a slope compensated static or dynamic drag factor . It is
common practice in accident reconstructions for a CoF to be used in conjunction with a correction grade drag
factor in order to estimate the true drag factor (Pazzaglia & Nelson, 1993).

1.5

Research Objectives

The models and asset management procedures currently in use do not adequately predict the variability of skid
resistance as measured on the highway. Further research is needed to better understand and model its variability
over time and to elucidate the factors that are involved.

The research reported herein is aimed at a better

understanding of skid resistance, including the development of alternative methods, processes and models to better
manage the inherent variability of measured skid resistance.

Chapter 1
This research specifically considers the seasonal and short-term variation of skid resistance. Research has indicated
that environmental effects such as temperature and rainfall have a significant effect on the short-term variation of
skid resistance and the longer term seasonal fluctuations. Whilst the occurrence of rainfall is generally accepted as a
reason for short-term variations in skid resistance, the mechanism by which the variations are produced is not yet
sufficiently well understood to permit reliable modelling (P. D. Cenek et al., 1999). The causal relationships that
have been developed are not well understood and therefore have not been incorporated into current road asset
management practices. Yet, road asset managers have great difficulty in managing currently specified standards of
skid resistance with naturally sourced aggregates. The effects of temperature variation, the length of time since rain,
combined with the effect of the accumulation of surface contaminants on various aggregate types, are still little
understood. Furthermore, current prediction methods of an aggregate s ability to resist polishing (primarily the PSV
test) have been shown to be inadequate for predicting the long term skid resistance level that is recorded on the
highway.
The objectives of this research are therefore to quantify the:
1) Confidence limits of measured skid resistance variation that includes seasonal and shortterm variability of skid resistance on various road surfaces;
2) Variability between measurement devices (SCRIM, GripTester, and the Dynamic Friction
Tester) on various surfaces;
3) Rate of deterioration of the measured skid resistance of newly laid chipsealed surfaces on
straight sections of road in comparison to corners, for various levels of traffic;
4) Effects of temperature on the GripTester and the Dynamic Friction Tester results;
5) Concentration and particle size distribution of the types of contaminants that accumulate on the
road surface and how they vary with time and rainfall;
6) Effects of various washing treatments on the pavement surfacing to determine whether the
presence of detritus in itself reduces measured skid resistance;
7) Effects of wet and dry accelerated polishing on prepared laboratory specimens;
8) Effects of various contaminants on skid resistance due to accelerated polishing on prepared
laboratory specimens.

1.6

Research Methodology

The research methodology was designed and developed in two stages:

Field Tests regular skid resistance testing at seven carefully chosen locations using the GripTester (a
continuous friction measurement device), supplemented, where possible, by the Dynamic Friction Tester.
The aim being to quantify the extent of the skid resistance measurement and the factors involved in the
variation over a two to three year data collection period. Where possible, dependent variables would be
controlled or their variability measured; and

Laboratory based polishing tests constructing chipseal surfaces undertaking accelerated polishing of the
surfaces and periodically measuring the skid resistance using the Dynamic Friction Tester to determine the
skid resistance resulting from aggregate polishing to an equibrium level . Subsequently, quantifying the
effects of the addition of contaminants and polishing under controlled laboratory conditions.

1.7

Organisation of the Thesis

The subsequent chapters of the thesis are organised in the following manner.
4

Chapter 1
Chapter 2 provides a review of the fundamental Theory of Skid Resistance. The chapter begins with a review of the
function and characteristics of road surfacings including methods and processes used to manage the asset. It
continues with a review of the traffic system and road safety with particular attention paid to skid resistance and
crash risk that has led to the development of current standards of skid resistance. Chapter 2 concludes with a review
of vehicle system dynamics related to skid resistance and defines the components of surface friction and skid
resistance.
Chapter 3 reviews and describes the factors that affect road surface skid properties including road pavement surface
texture profiles (microtexture and macrotexture), aggregate properties (petrological properties and polishing /
abrasion effects), loading factors (traffic intensity, composition, and age) and environmental factors that affect the
seasonal and short-term measured variation of skid resistance, including the effects of water, contaminants,
temperature and drainage conditions.
Chapter 4 provides a summary of the commonly used skid resistance testing methods and models. The chapter
commences with a discussion of the factors that are relevant to testing devices, including vehicle speed, tyre angle,
slip ratio, tyre characteristics, tyre contact area and stresses. The chapter continues with a description of the various
methods of measuring surface friction both in the field and in the laboratory, including the microtexture and
macrotexture profiles. It concludes with a discussion of the models that have been developed to describe the
behaviour of skid resistance over time.
Chapter 5 describes the experimental design of the research, including the research hypotheses and an explanation of
the research methodology. The field research testing methodology is outlined, including the chosen field testing
sites and their characteristics and the standard data variables that were thought to affect the level of skid resistance,
including rainfall, temperature and the contaminants on the road surface. The chapter concludes with a detailed
description of the procedures developed to simulate field conditions in the laboratory through accelerating the
polishing of surface aggregates. The methods to control specific variables whilst testing the effect of extended
polishing of various surface aggregates and the addition of contaminants are also discussed.
Chapter 6 presents, describes and discusses the results of the skid resistance monitoring of the field test sites. It
includes a description of the procedures developed for the monitoring and the processing and analysis of the test
data .
Chapter 7 provides a discussion of the analysis of the field testing results presented in Chapter 6, with regard to the
repeatability and correlation of the various skid resistance testing devices used in this research (primarily the
GripTester, SCRIM and Dynamic Friction Tester). The chapter concludes with analysis and reporting of the data
from the Surface Friction Device Correlation Trial held in Christchurch in May of 2005, with respect to the
repeatability of the devices.
Chapter 8 provides a discussion and analysis of the field testing results initially reported in Chapter 6 with respect to
the sites and/or sections that should display both seasonal and short-term variations of measured skid resistance.
The chapter also addresses the effects of certain geometric elements on skid resistance and the results of a washing
trial that was undertaken to determine whether the presence of contaminants directly or indirectly affected the
measured skid resistance levels.
Chapter 9 presents the results of surface contaminants collected from the road surface for suspended solids, particle
size distribution, heavy metals, petroleum hydrocarbons and organic carbon with respect to how these contaminants
may affect variation in skid resistance.
5

Chapter 1
Chapter 10 gives the analysis and discussion of results for the accelerated polishing of laboratory controlled tests of
four different aggregate sources. The reported results and analysis include the polishing of the laboratory samples to
an equilibrium skid resistance level and, once an equilibrium level was chosen, the effect of accelerated polishing
with the addition of contaminants.

The chapter concludes with a geological interpretation of the polishing

mechanisms of the various aggregates and the contaminants used in conjunction with accelerated polishing.
Chapter 11 summarises the research aims and objectives and the individual hypotheses for each research element.
The chapter discusses the conclusions of each of the research elements.
The final chapter, Chapter 12, presents an overview of the final conclusions reached from this research, the
relevance of the research and recommendations for issues requiring further research and investigations.

Chapter 2

2 SKID RESISTANCE FUNDAMENTALS

2.1

Introduction

This chapter reviews the published literature on the fundamentals of skid resistance. It begins with a review of the
function and characteristics of road surfacings, the traffic system and road safety and the relationship to skid
resistance and crash risk. It continues with a review of vehicle system dynamics and the definitions of skid
resistance and friction.

2.2

Function and Characteristics of Road Surfacings

2.2.1 Introduction
A wearing surface or road pavement surfacing is the uppermost layer of a pavement structure, on which traffic runs.
The principal functions of surfacings are (Austroads, 2003) to:

Provide a riding surface of suitable smoothness;

Provide a safe, economical and durable all-weather surface;

Minimise vehicle operating and maintenance costs;

Provide a dust-free surface;

Minimise the rate of pavement wear;

Reduce moisture infiltration into the pavement; and

Provide suitable properties for the local environment (eg. noise reduction and surface texture).

Leaving aside the economic optimisation and capacity / utilisation factors, Lay (1998) states that the key service
properties of a road surface are its frictional properties (level of service, road user safety) and its impermeability.
The impermeability is important as it prevents water entering the pavement structure and causing rapid structural
deterioration, whereas the frictional properties are required to maintain adequate margins of safety for the road user.
If the other factors are not adequately addressed, both in the design phase and throughout the lifecycle of the asset,
premature asset failure is possible due to one of these other factors.

2.2.2 Infrastructure Asset Management Goals

The role of the infrastructure manager, or in this case the road asset manager, is a complex one as it requires both a
clear understanding of the structural failure and deterioration mechanisms of the road asset and how its distress
modes are distinguishable, and also a clear understanding of the function and purpose of the asset in servicing the
end users (i.e. the customers or, in this case, the road users). These two important factors (condition and service
level needs) then need to be balanced economically against the life cycle costs of the asset in comparison to a
network of other road asset needs and ultimately need to be compared to other infrastructural needs within the
context of community and national resource requirements.

It is therefore important to contextualise the role and

need for road pavement skid resistance within the total Infrastructure Asset Management process.
The International Infrastructure Management Manual (Ingenium et al., 2002) states:
The goal of infrastructure asset management is to meet a required level of service in the most
cost effective way through the creation, acquisition, maintenance, operation, rehabilitation and
disposal of assets to provide for present and future customers.
It further states:
7

Chapter 2

The key elements of infrastructure asset management are:

Taking a lifecycle approach

Developing costeffective management strategies for the long-term

Providing a defined level of service and monitoring performance

Managing risks associated with asset failures

Continuous improvement in asset management practices.

However, to provide a defined and minimum level of service requires reliable and repeatable methods of monitoring
and measuring of the assets performance and / or a clear understanding of the variables that affect the measured
performance of that asset. In the case of road pavement skid resistance, this is a complex task in comparison to
other pavement and surface characteristics, as is discussed in later sections.

2.2.3 Monitoring Surface Characteristics and Performance

The monitoring of the performance of road surfacings by road asset managers requires the assessment of parameters
in two major categories, as summarised in Table 2-1:
Road User and Community parameters

Structural parameters

Pavement surface shape (including roughness)

Pavement Strength

Skid resistance

Cracking

Surface Texture

Serviceability particularly ageing effects

Noise
Rutting and shape loss
Conspicuity of markings / reflectivity
Appearance
Water spray

Table 2-1: Pavement Surfacing Parameters requiring Performance Monitoring (Austroads 2003)
In practice, many of the above factors are inter-related, for example, increasing roughness or rutting is an indicator
of deteriorating structural strength and bleeding, flushing and stripping of sprayed seal chips will lead to a loss of
surface texture, thereby causing poor skid resistance.

Rutting or poor drainage can also increase the amount of

surface water and increase the risk of skidding. Furthermore, the potential for skidding is also influenced by road
geometry, roughness and the surface macrotexture as well as the surface friction characteristics.

2.2.4 Pavement Deterioration Modelling in New Zealand (NZ dTIMS System)

Background
During 1998 the New Zealand RIMS1 Group initiated a national project called Implementation of Predictive
Modelling for Road Management commonly known as the NZ dTIMS project . The RIMS Group is comprised of
members representing local governments, Transit New Zealand and Land Transport New Zealand. It operates under
the Ingenium Association (formerly known as the Association of Local Governments Engineers of New Zealand 1

Road Information Management System (RIMS) Group.

Chapter 2
ALGENZ). The NZ dTIMS project arose from the need to implement a credible system that could be used to
predict future long-term maintenance and rehabilitation needs for road networks (N.K. Pradhan et al., 2001).
The dTIMS2 software was selected as the main analytical tool or engine for the system. The decision to use the
dTIMS software was because of its inherent flexibility and open system that allows the user to alter pavement
deterioration models or the actual analyses framework itself. The World Bank Highway Development Model
(HDM-III and HDM-4) pavement deterioration and works-effects models, together with locally developed models,
are programmed into the system for New Zealand conditions. An important aspect of the implementation process
was to provide a system that could be fully integrated with the RAMM3 inventory and road condition database
system used by almost all Road Controlling Authorities (RCAs) in NZ. The RAMM system also provided some
treatment intelligence about the condition of the road asset for decision making purposes.

The Need for Predictive Modelling

The RAMM system has been widely used by RCAs in New Zealand for recording data on road condition, and also
for short-term planning based on a simple treatment-selection algorithm since its development in 1989. Practitioners
more recently recognised that this was not enough and that more powerful tools were needed to forecast the
condition of roads in the longer term and to test the implications of different levels of investment in maintenance.
The RIMS Group therefore developed a vision for implementing a credible predictive modeling system that was also
nationally uniform (a significant benefit in terms of efficiency, economies of scale, consistency and resource
capability and availability). The system that was developed was based upon the following features:

Capability of providing treatment intelligence about the network;

Fully integratable with the RAMM inventory management system;

Capability of multi-year programming based on lifecycle analysis of the pavement, so that forward works
programmes could be prepared; and

Capability of following an evolutionary development path as new research models could be incorporated
into the system.

The proposed predictive modeling system would help asset managers to answer the following important questions
about maintaining their road-network asset:

How effective is the current network maintenance policy?4

What is the most cost effective timing for intervention on a road section?5

What is the optimal maintenance programme for an activity under ideal funding strategies or when funding
is less than ideal?

What engineering data could help substantiate annual funding requests to the Land Transport funding
agencies (e.g. Land Transport New Zealand) on the most cost effective funding levels for their network?

These questions could not be answered by the RAMM system. Furthermore, Central Government legislation
incorporated in the NZ Land Transport Management Act (New Zealand Government, 2003) and the NZ Local
Government Bill (New Zealand Government, 2002) now require that RCAs as asset managers take a long-term (20+

2
3
4

Deighton s Total Infrastructure Management System.

Road Assessment and Maintenance Management System.
The predictive modelling system helps determine the optimal funding split between resurfacing and rehabilitation
projects.
The point at which a particular treatment is performed on a section can result in major savings when considering a
network of needs through increased efficiencies.

Chapter 2
years) optimised life cycle asset management approach thereby ensuring prudent and fiscally appropriate
management of public assets.
The NZ dTIMS project sought to achieve this by focusing the implementation in four development streams of
predictive modelling, broadly demonstrated in Table 2-2. The successful implementation of a project such as this
requires strategies that target data, research, understanding and the operational adoption of the industry before true
modeling capability is achieved that best represents infield variability and deterioration. These strategies include:

Training and upskilling of users in the understanding and effective use of the predictive modelling system;

Development of best practice guidelines to improve the process for collecting, storing and processing data;

Undertaking operational research for the refinement and / or regional calibration of the models used in the
system, based on the specific needs identified during the implementation stage; and

Aiding in the operational adoption of the system.

Components of Predictive Modelling

The three main components of predictive modelling for pavement management systems are:

A methodology;

Mathematical models; and

A software platform.

The NZ dTIMS system uses a deterministic modelling approach largely based on the HDM models. This approach
ensures that certain treatments are selected, using the intervention levels and the different construction strategies for
a road section that can be generated by the model. The optimal strategy is then selected based on economic
optimisation. An important aspect of the models is that they are incremental models, meaning that the prediction of
the future condition of the pavement or surfacing begins from its existing condition. An alternative strategy is
calculated using performance criteria; in this, road sections are triggered for treatment based on their condition per
lane.

10

Chapter 2

INTELLECTUAL

DATA
APPROPRIATENESS

RESEARCH

OPERATIONAL
ADOPTION

Condition and inventory data

suitable for modelling

Refinement of models based

on appropriate data

Control mechanisms,
consumption and utilization
of outputs

Most available data are

inappropriate

Research based on
inappropriate data
(concerned with principles)

Principles understood

Box 1: Objectives
Development of the ability
to use the skills (training,
upskilling)
Box 2: Progressive Steps
Principles understood

Developing
understanding

Ability to interpret outputs

Sensitivities understood

Control mechanisms
established

Change data collection and

QA principles

Management methodologies
can absorb outputs

Appropriate data becoming

available

Model refinement capability

based on appropriate data

Outputs used in setting

policy

Box 3: Desired Outcome

True Modelling Capability

Ongoing Refinement

Table 2-2: Development Streams of Predictive Modelling - modified from (Pradhan et al. 2001)

Pavement Deterioration Modelling and Road Asset Management

As discussed previously, the NZ dTIMS system is only part of the road asset management process. The following
components are needed to manage the roading infrastructure and its elements efficiently:

Systematic and regular data collection for the various asset elements;

One central database for storing and accessing data (RAMM in NZ);

Uniform location referencing to enable consistent, reliable and accurate location of asset elements;

Analysis modules for traffic, pavement and surfacing condition, planning, bridges etc; and

Administration modules for cost control.

Figure 2-1 shows the various components of the New Zealand road asset management framework and specifically
where dTIMS and RAMM fit into the desired framework. Not all components exist at present, although the figure
shows the conceptual interactions between the various asset management information / business systems. The
RAMM system has in the past provided the basic framework for pavement management. It consists of an inventory
11

Chapter 2
of the assets and their condition. A treatment selection algorithm is used to analyse the data to establish the
maintenance needs in the short term.

The primary objective of the NZ dTIMS project was to build upon this by

providing the capability, through pavement deterioration modelling, for planning over the medium and long term.
The results would then feed into forward work programmes (e.g. using the NOMAD6 software developed by Transit
NZ).
Asset
Management

RAMM

Performance Management

Funding

Condition Achievement
Inventory of Assets and
Their Condition

Performance Achievement

(RAMM)

Trend Analysis

dTIMS
Treatment Intelligence
Treatment
Selection
Algorithm

Fund Allocation Model

Pavement
Deterioration
Modelling

Financial Management

Forward Work Programme

Annual Plan

Project Justification

Project Management

Project
Management

Figure 2-1: Road Asset Management components / process (N.K. Pradhan et al., 2001)
The risk of using tools such as NZ dTIMS and others is that network managers may become overly reliant on these
tools without an understanding of their limitations (N.K Pradhan et al., 2003). The output of the predictive
deterioration modelling and how closely that modelling represents reality is dependent upon a number of important
factors including:

Quality of the input data;

Sampling method used;

Reliability of the data collection equipment;

Repeatability of the test;

Data collection frequency;

Weather patterns; and

Statistical goodness of fit of the model and its transferability to other climatic / subsoil / pavement
mixture / material conditions.

It follows that it is imperative to understand the variability of the asset factor being measured and collected prior to
determining the sampling methodology which includes the frequency of the testing regime. This is especially
relevant to skid resistance measurement as significant and largely unpredicted variation exists and is discussed in
more detail later in Chapter 3.
6

National Optimisation of Maintenance Activity by Decade

12

Chapter 2

The NZ dTIMS System and Skid Resistance

Skid resistance was originally incorporated into the NZ dTIMS Project setup, (Henning et al., 2000) as it was
considered to be an important trigger for resealing actions.

The RIMS Group recognised that research by

Donbavand (1989) had shown a strong correlation between poor skid resistance and accident occurrence, stressing
the importance of providing reasonable skid resistance (Henning et al., 2000).

Therefore, any pavement

deterioration modeling that predicted time to interventions and intervention treatments was not complete without
considering the performance of surfacings with respect to skid resistance.
The modeling of pavement deterioration via the NZ dTIMS setup is required before Land Transport New Zealand
will approve funding for maintenance projects for Road Controlling Authorities throughout New Zealand.
Currently, there is no prediction model being used in New Zealand or internationally for the prediction of future skid
resistance for the purposes of maintenance intervention. A model is used in the NZ dTIMS system to predict
macrotexture deterioration which is an important function of the measured coefficient of friction of skid resistance,
however this is only part of the equation.

Earlier versions of the NZ dTIMS Setup used the Transit NZ T/10

relationship between Polished Stone Value (PSV) and skid resistance as measured by the SCRIM device and as a
function of Heavy Commercial Vehicles (refer Section 4.6.2). This has since been removed from the NZ dTIMS
Setup as it was recognised that the equation was not a good predictor of in-field measurements, as confirmed by
recent research by Cenek et al. (2003b); Wilson and Dunn (2005a) and Kennedy (2005).
There is still a strong desire to incorporate this capability within the NZ dTIMS system, although it was recognised
that further research was required to develop a better prediction equation. Whilst skid resistance deterioration
(including macrotexture) does not directly affect the pavement structural deterioration, it has a significant effect on
the timing of maintenance treatment actions. For instance, it would not be economically prudent to resurface a road
one year and the following year programme structural rehabilitation improvements.

The inter-relationships

between these and other factors that lead to varying maintenance treatments are recognised but as yet are little
understood due to their complexity. Figure 2-2 identifies the typical pavement and surfacing distress types, their
general relationships, and whether they result in changes to the surface macrotexture or microtexture. It also
demonstrates that skid resistance effects fall into the category of periodic maintenance activities.

13

Chapter 2

SURFACE LOADINGS (& DETERIORATION)

Traffic (Volume) and
Environment (Temperature range)

GENERAL
REMEDIAL
WORKS
REQUIREMENTS

ROUTINE
MAINTENANCE

PERIODIC
MAINTENANCE

Macrotexture
Changes

BINDER AGEING,
EMBRITTLEMENT

RUTTING

THERMAL
TRANSVERSE
CRACKING

INCREASING
SURFACE
DETERIORATION
CAN BE STOPPED
AT ALMOST ANY
POINT BY
MAINTENANCE

STRUCTURAL LOADINGS (& DETERIORATION)

Traffic (ESA's) and
Environment (Moisture ingress)

FLUSHING
POLISHING

STRUCTURAL
LONGITUDINAL
CRACKING

P
O

HUNGRY SURFACE
THRU BINDER
LOSS

THERMAL BLOCK
CRACKING

INCREASING
STRUCTURAL
DETERIORATION
CANNOT BE
STOPPED BY
MAINTENANCE,
ONLY SLOWED

T
H

STRIPPING /
RAVELLING

DELAMINATION

L
STRUCTURAL
ALLIGATOR
CRACKING

I
REHABILITATION
WORKS

Skid Resistance
Effects &
Maintenance Activity
RECONSTRUCTION

Microtexture
Changes

DELAMINATION

SHOVING
SHOVING

P P
AA
V VE EMMEENNT T

D II S
S II N
D
NT
T EE G
RR
GA
R TAI TO INO N

Figure 2-2: Flow Diagram of Linked Progressive Pavement and Surfacing Deterioration and Associated Maintenance Activities (modified from RIMS, 1999).

14

Chapter 2

2.3

The Traffic System and Road Safety

2.3.1 Introduction
As reported in many traffic engineering texts, the safety of the road-based traffic system is a complex issue (Ogden
and Taylor (1996); Lamm et al., (1999); Homburger et al., (2001)), because it involves not only the roadway
features and driving conditions but driver behaviour and vehicle characteristics. Lamm et al. (1999) state:
[Road] Safety must be approached both as a question of vehicle and roadway design and as a
question of use, driving habits, traffic regulation, law enforcement, and risk management
There is no single, simple solution to the problem of highway safety.
The highway system consists of a range of different subsystems and elements, the skid resistance of the surfacing
being only one of the many critical components of the road / environment element. The basic looped relationship
between the main elements is demonstrated in Figure 2-3 which shows that the behaviour of the driver influences
the control of the vehicle, which in turn interacts with the roadway / environment. Driver behaviour is itself
influenced by the roadway and environment.

The Driver
The Vehicle
The Roadway /
Environment
Figure 2-3: Interaction between driver-vehicle-roadway /environment (Lamm et al., 1999).

2.3.2 Road Crash Factors

Road-based crashes are very rarely able to be contributed solely to one factor and as such are defined as being rare,
random and multifactor events where the driver has failed to cope with their environment (Land Transport New
Zealand, 2004). The relative contribution between each of the factors (human, vehicle and road / environment) to
road crashes has been analysed in a number of studies. The results of two such studies, one in the UK and one in the
US, are shown in Figure 2-4. The numbers shown in the circles represent the percentage of crashes for the UK and
the US that can be attributed solely to that factor, e.g. the UK study demonstrated that 65% of the crashes could be
solely attributed to the human element whereas the US study demonstrated a level of 57%. The numbers shown in
the arrows that link two factors represent the percentage of crashes where both of the factors linked were considered
to contribute to the crash occurring, e.g. the UK study demonstrated that an additional 24% of crashes had
contributing factors that included the road user and the road environment, whereas the US study showed 27%. The
numbers in the triangle represent, for both studies, the number of crashes where all three traffic system factors could
be attributed to the crash occurring, i.e 1% and 3% respectively for the UK and US studies. It can be seen that when
the total of all combinations of factors are added together for the three road system factors, (for the UK and US
study respectively) the:

human factor contributes to 94/93 per cent;

15

Chapter 2

road factor contributes to 28/34 per cent of all crashes, and the

vehicle to 8/12 per cent of crashes.

Figure 2-4 shows the very high contribution level of the human factor in almost all crashes. To some degree this
could be expected to be even closer to 100%. For example if a driver takes an alternative action, a crash may be
avoided. However, the percentage of crashes where the road environment plays a role in the crash occurring is still
significant (approximately 30%) and provides significant incentives for engineers to aim to do better with
engineering solutions to improve road safety. It should be noted that these figures should only ever be considered
as being representative as they are obtained from traffic crash databases that are solely based upon the attending
Police Officer s summation of the events and the contributing factors to the crash occurring.

Road User
(human)
65% / 57%

4%

7%
/2

/6

%
24

1% / 3%

The Vehicle
2% / 2%

Road
Environment
2% / 3%

1% / 1 %

Note: the first percentage value is sourced from the UK study, the second from the US.
Figure 2-4: Factors Contributing to Road Crashes (adapted from Ogden and Taylor (1996) sourced from
(Sabey, 1980) - UK and Treat (1980)-US).
Lamm et al. (1999) refer to inappropriate behaviour of drivers, as well as vehicle and environmental influences, as
being the main cause of vehicle crashes.
This inappropriate behaviour results from deficiencies in human - vehicle interaction and/or
a misunderstanding of upcoming driving situations with respect to the roadway or
environment. the most effective ways to improve traffic safety are by aiding the driver in
his or her interaction with the vehicle and by designing highways in such a way that features
which are relevant to the driving task are perceptible, understandable, and soundly designed,
based on driving dynamic, driving behaviour, and psychological points of view.
The driving task, being the link that keeps the inherently unstable system in equilibrium, can be considered as
comprising three essential tasks (American Association of State Highway and Transportation Officials, 1990 cited
in Ogden and Taylor (1996)):

Navigation: trip planning and route information;

Guidance: following the road and maintaining a safe path in response to traffic conditions; and

Control: steering and speed control.

16

Chapter 2
These tasks require the driver to receive inputs (mostly visual), process them, make predictions about alternative
actions and decide which are the most appropriate, execute the actions, and observe their effects through the
reception and processing of new information (Lay, 1998).
A model originally developed by (Donges, 1978, 1982) and elaborated by Lamm et al. (1999) indicates that the
driving task levels present a hierarchy, which exhibits a decreasing complexity regarding information handling as
the driver moves from navigation to control. In terms of road safety, a reverse relationship exists as the hierarchy
exhibits increasing importance as the driver moves from navigation to control. Correspondingly, the skid resistance
level of the road pavement surface becomes critical for the safe movement and operation of the vehicle (i.e. control).
A modified version of the model by Lamm et al. (1999) is shown in Figure 2-5 and demonstrates the importance of
surface friction in vehicle control.
Environment

Driver
Navigation
Level
Route and
Time Planning

Guidance
Level
Desired Speed,
Driving Path, &
Lateral Distances

Control
Level

Road Network

Factors where the level

of pavement skid
resistance has a
significant impact in
terms of road safety

Driving Space,
Traffic Situation

Vehicle

Steering,
Acceleration,
Deceleration

Longitudinal &
Lateral Movement

Vehicle
Movement

Road Surface
characteristics,
Skid Resistance

Actual Movement, Speed

Safe Movement
Driving Task

Figure 2-5: Three Level Model of the Driver Vehicle Roadway / Environment (modified from Lamm et al
(1999))
Traffic safety measures are often discussed in terms of the three E s . These are parallel strategies that attempt to
improve road traffic safety on road networks. The three E s of road safety are:

Enforcement - by traffic regulations, rules and traffic police presence;

Education anti speed and alcohol advertising, which is intended to alter human behaviour; and

Engineering road and traffic element design and maintenance, vehicle system dynamic and control and
understanding of human factors.

Clearly the responsibility of the design and maintenance of the road lies within the engineering discipline, although
other disciplines aid in the understanding of human factors. Whilst understanding the limitations and variance of
human ability and the distribution of the population of road vehicle characteristics, the principal objective of road
and traffic designers is to design, maintain and upgrade the road environment condition to provide a road traffic
system that provides safe, consistent and efficient road networks. This objective inevitably needs to be balanced
equitably and furthermore optimised against chosen criteria in terms of the funding available to the road sector.

17

2.3.3 Skid Resistance and Crash Risk

Chapter 2

Introduction
Research into skid resistance over more than 70 years has clearly confirmed that a firm relationship exists between
the risk of a crash occurring in the wet and the slipperiness of the road surface.

It is not intended to

comprehensively review the research that has been undertaken in this area. However, a summary of the significant
early research findings linking road crashes with wet weather conditions and the level of surface friction on the road
surface follows.
There are a number of statistical techniques that have been used to examine links between road crash statistics and
road pavement surface skid resistance. They include (Jellie, 2003):

Before and After studies these assess the extent of crash reductions following resurfacing. A direct
measure of the effects of increasing skid resistance.

Comparison with the Norm studies - sites are identified where skidding crashes have occurred and a
comparison is made to randomly selected control sites that are representative of the whole road network.

Regression studies a plot of one variable (in this case the proportion of crashes that occur under wet
pavement conditions) against another (level of skid resistance) and observe how changes in that variable
relate to changes in the other variable. One issue with bivariate studies is whether they can address an
environment in which interplay may involve more than two parameters (e.g. skid resistance and
geometrics).

Before and After Studies

Giles and Sabey (1959) in a two-year UK study showed that after resurfacing, the average number of wet weather
crashes (over 55 sites) reduced from eight to two and the number of crashes involving skidding reduced from six to
one. A similar two-year study by Miller and Johnson (1973) on the UK M4 motorway in England showed that
increasing the SFC507 from 0.40 to 0.55 by resurfacing reduced total crashes by 45%, dry pavement crashes by 28%
and wet weather crashes by 63%.
In North America, Hatherly and Young (1977) examined the effects of treating 800 intersections with calcined
bauxite set in an epoxy / bitumen mix that raised SFC50 from 0.40 to 0.80 with a corresponding reduction in total
crashes of 30%. Karmel and Gartshorne (1982) demonstrated in a Canadian study that resurfacing intersections that
had been identified as crash blackspots reduced wet skidding crashes by 71% and dry road condition crashes by
21%.
An Australian study by Kinnear et al. (1984) compared three high skid resistant surface treatments with control sites
and demonstrated that calcined bauxite surfacings reduced total crashes by 28% with a subsequent benefit to cost
ratio of 84:1. In New Zealand, Hudson and Mumm (2003) demonstrated that a resurfacing with calcined bauxite on
a low standard horizontal and vertical geometric alignment reduced loss of control type crashes by greater than 50%.

Comparison with Norm Studies

An early comparison study was undertaken in the UK at the Road Research Laboratory by Giles, Sabey and Cardew
(1962) using the British Pendulum Tester (discussed in Section 4.3.4). As shown in Figure 2-6, they found that the

Skid Resistance measurement by the Sideways Force Coefficient (SFC) measured at 50km/h

18

Chapter 2
relative risk of a site being a skidding crash site was exponential in form with an extremely low risk for high values
of skid resistance and a high risk for low values of skid resistance.

Figure 2-6: Relative Risk of a Surface being a Crash site (sourced from Cairney, (1997))
Rogers and Gargett (1991) reported on a large-scale national UK skid resistance survey in 1981 of 1000kms to study
the link between skid resistance and personal injury crashes. This study provided detailed information on skid
resistance levels on the UK network with the aim of identifying the benefits of introducing a new standard. The
network was divided into different site categories such as junctions, traffic signals, pedestrian crossings and
roundabout approaches, bends, gradients, non event open road, dual / single carriageway, and speed restriction / derestriction. Skid resistance measurements were made in 1981 and 1982 using the SCRIM device and corrected for
Mean Summer SCRIM Coefficient (MSSC) (discussed in Section 4.3.2) and, along with the reported injury crashes
during 1980, 1981 and 1982 were then sectioned and allocated to their correct site category and location. These data
were analysed along with traffic flow data to produce crash rates (number of injury crashes per unit length of road)
and interrogated primarily for wet skid / wet no-skid relationships. From this analysis a number of site category
groupings and relationships were established and tabulated in ascending order of wet skidding resistance. Examples
of the relationships developed for Motorway, Minor junctions (single carriageway) and Traffic Signal approaches
are shown in Figure 2-7. Risk is defined as the number of wet skidding type crashes per year per 100m length x 106
/ Average daily traffic (one way) in thousands.

Figure 2-7: Risk of accidents on wet roads and Mean Summer SCRIM Coefficient (MSSC) from Rogers and
Gargett (1991).

The three site categories groupings shown in Figure 2-7 are summarised below:

Motorways - only minimal emergency braking is needed, thus a very gradual increase in wet skidding risk
occurs as wet skidding resistance reduces.

This low risk relationship also reflected UK policies of

providing good quality aggregates during new works and regular maintenance treatments ensured that
generally the motorways had good skidding resistance.
19

Chapter 2
Single Carriageway Minor Intersections more emergency braking was needed on these roads and a very
positive wet skidding crash / wet skidding resistance relationship was apparent. Below an MSSC level of
0.45, there was a sharp increase in risk. Due to the high incidence of crashes at junctions and as treatment
lengths are short and more clearly defined, the cost-effectiveness of these treatment categories can be often
high.

Approaches to Traffic Signals these sites are locations where good skid resistance is essential due to the
high proportion of emergency braking. Note that the scale of the y-axis for Risk (in number of crashes) in
Figure 2-7 is different for this category (0-400) in comparison to the first two categories (0-40),
demonstrating the significantly increased crash risk at signalised intersections. The study demonstrated
that levels of 0.55 MSSC would be cost effective and produce significant crash savings.

Roe et al. (1991) also undertook a comparison to the norm study to investigate the relation between the surface
texture of roads and accidents . They surmised that low skidding resistance or texture depth are unlikely to be the
first causes of a crash. However, these factors would worsen the situation where a conflict has created the need for
an emergency manoeuvre. SCRIM (MSSC) measurements were used to measure low speed skid resistance and the
level of microtexture on crash sites. Laser based sensor measured texture depths (SMTD) were used for high speed
skid resistance measurement and to determine the macrotexture at the crash sites. The crash sites were compared
against control sites with similar characteristics. At the control sites no crashes had occurred. The null hypothesis,
that texture at crash sites was distributed in the same way as texture on the general road network, was rejected. Roe
et al. (1991) concluded that:

Skidding and non-skidding crashes, in both wet and dry conditions, are less if macrotexture is high. This
was an unexpected result for dry conditions as a relation was expected only between macrotexture and wet
skidding crashes. It was suggested that this was due to high macrotexture surfaces contributing more to
dissipating energy due to the hysteresis losses of the tyre, thereby enabling drivers to slow down and
perhaps avoid a crash.

The level of macrotexture below which crash risk begins to significantly increase is approximately 0.7mm
SMTD.

Regression Studies
McCullough and Hankins (1966) and Schlosser (1977) used regression analysis to examine the relationship between
skid resistance and crashes (including the effects of traffic flow, vehicle type and weather conditions) on 3400kms
of motorway network in the Netherlands. This study found that wet pavement crashes per vehicle kilometre
travelled (VKT) reduced as skid resistance increased. The increase was found to reduce and then disappear after
which there was little variation in crash rates with respect to decreasing skid resistance. Crash rates for heavy goods
vehicles and cars were similar on access controlled roads. However, car rates had significantly higher rates when
access was not controlled.
Harewood et al. (1977), in similar research, confirmed the relationship between crash risk and the level of skid
resistance primarily on two lane rural roads. An interesting finding in this study was a higher correlation (0.67)
between wet pavement crash rates and dry weather crash rates than between wet crash rate and skid resistance
(0.23). This demonstrated that crashes are determined to a large extent by the same factors, such as road alignment
and traffic characteristics, regardless of weather conditions.
As reported by Henry (2000) early attempts to relate crash data to skid resistance in the US measured with a ribbed
tyre were unsuccessful. During the late 1970s a smooth tread tyre standard was introduced for skid resistance
20

Chapter 2
measurement and a Connecticut study in 1979 concluded that a good correspondence between low smooth-tire skid
numbers and accident experience can be seen and Ribbed-tire correspondence was quite poor. The study further
concluded that on pavements that had smooth tyre skid numbers (SN40S)8 greater than 25, there were fewer wet
skidding crashes. A further study in 1984 by the Florida Department of Transportation confirmed that the smooth
tyre skid resistance data are a better indicator of safety than data from ribbed tyre measurements. However, Henry
(2000) states that because many other factors contribute to crashes, including pavement condition, prevailing speed,
and traffic conditions, one should not expect to be able to predict crash frequency from skid resistance data alone.
Hosking (1986) in the UK recognised the importance of seasonal variation in skid resistance and estimated the likely
effect of increasing skid resistance over an entire road network, rather than relatively short stretches of road such as
those undertaken in Before and After Studies . Hosking showed that the pattern of variation on a monthly basis
was consistent, with minimum values occurring in the summer and maximum values in the winter (refer to Section
3.5.6). Hosking used the ratio of the SCRIM measured coefficient of friction for each month to the average SCRIM
value for the year. The measure of crashes used was the wet crash skidding rate (i.e. the proportion of crashes that
occurred during wet conditions and which involved skidding). The relationship between these two variables showed
that 95% of the seasonal variation in wet weather skidding crashes can be explained by seasonal variation in skid
resistance. Hosking also developed a model to predict the reduction in wet skidding crashes and therefore the
benefits from raising road network levels of skid resistance. The model results showed that increasing the SCRIM
coefficient by 0.1 units across the entire road network would decrease wet and dry road crashes by less than 2%.
His model demonstrated that attempting to increase levels of skid resistance over an entire road network may not be
a cost effective way of reducing crashes and that a more targeted approach was needed. This was seen again in the
work reported by Rogers and Gargett (1991).

Summary of Recent Crash Risk Research

As shown above, early crash studies in many countries have demonstrated that crash rates increase as the level of
skid resistance decreases. Furthermore, disproportionately high rates of crashes occur on sites with low skid
resistance or surface texture than would be expected for the site when applying the crash mean for the road or route
which includes that site. A brief summary of the recent crash studies follows (Austroads, 2005b).

An Australian study (Peter Cairney, 1997) concludes that sites where wet weather crashes occur have lower
skid resistance, on average, than sites where such crashes do not occur;

A NZ study (P D Cenek et al., 1998) found significant relationships between the wet road skidding crash
rate and skid resistance for intersections, approaches to traffic signals, and curves and gradients;

Analysis of data from the Netherlands, Germany and France (Wallman & Astrom, 2001) reaffirmed the
link between crash rates and skid resistance, stressing the road sections with the lowest skid resistance were
found to be associated with a higher proportion of crashes;

European studies (Gothie, 2002; Gothi, 2000) showed that the crash rate and the injury severity of crashes
increased as skid resistance or surface texture decreased. A particular rapid increase in crash rate was
observed at very low values of skid resistance or surface texture; and

Roe et al. (1991) in the UK also found that on high speed roads the level of surface texture (macrotexture)
had a marked effect in all crashes.

Skid number SN40S denotes skid resistance coefficient of friction as measured by a locked wheel skid tester at 40m/hr
with a smooth tyre.

21

Chapter 2
In a recent review of UK trunk road skid resistance policy, Viner et al. (2005) reviewed the crash risk work of
Rogers and Gargett (1991) for the UK trunk network. The results of the research, in terms of the variation of the
mean crash risk with skid resistance are shown in Figure 2-8a-d.
100
Mean accident risk

Mean accident risk

20
15
10
5

80
60
40
20
0

0
0.3

0.35

0.4

0.45

0.5

0.55

0.3

0.6

0.35

0.4

0.45

Motorway

Dual c/way non-event

(a) Non event categories

Dual c/way minor junction

Single c/way non-event

0.55

0.6

0.65

0.7

Single c/way minor junction

Major junction

(b) Dual, single and major intersections (junctions)

100

100

Mean risk

95 percentile risk

80

80

Accident risk

Mean accident risk

0.5

Skid resistance

Skid resistance

60
40
20

60
40
20

0
0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

Roundabout approach

Signals approach

0
0.3

Skid resistance

0.35

0.4

(c) Traffic signals, roundabouts and approach

0.45

0.5

0.55

0.6

Skid resistance

Roundabout

(d) Mean and 95 percentile (single carriageway, non-event)

Figure 2-8: Mean Crash (accident) Risk by Skid Resistance and Site Category (Viner et al., 2005)
The study used similar site categories but a combination of approaches to analyse the data. The site categories were
taken individually and the standard mean and 95 percentile crash risk were calculated for the differing bands of skid
resistance specified in the UK HD 28/88 skid resistance standard. Crash risk was defined in this study as the total
number of crashes per 100 million vehicle kilometres driven. The notable points from the analyses were (Viner et
al., 2005):

There is a clear distinction in crash risk observed between the three non-event categories (refer to Figure
2-8a);

For motorways, the overall crash risk trend with skid resistance remains flat, except for the lowest band of
skid resistance level (refer to Figure 2-8a);

For dual carriageway non-event sections, there is a statistically significant trend for accident risk to
increase at locations with lower skid resistance (refer to Figure 2-8a);

For single carriageway non-event sections, the trend is both stronger and more significant (refer to Figure
2-8a);

Referring to Figure 2-8b, the crash risk for the various intersection categories is generally higher than for
the non-event sections (Figure 2-8a). However, the results are somewhat variable. Single carriageway
minor intersections exhibit the strongest and most significant trend with reducing skid resistance for all the
site categories. Conversely the trend for major intersections is not significant. For dual carriageway minor
intersections, the crash risk generally is low, except for the notable result at low skid resistance although it

22

Chapter 2
was noted by Viner et al. (2005) that this value was based on a small number of data points and is
indicative of the variability within sites;

For roundabouts and the approach to traffic signals (Figure 2-8c), the mean crash rates and the trends fall
approximately between the trends observed for minor intersections on dual and single carriageways. For
roundabout approaches, the mean crash rate was clearly higher than for the other two categories, however
the trend with skid resistance is ambiguous due largely, it was thought, to the small number of data points.

In a number of cases (single carriageway minor junctions and major junctions) the mean crash risk
increases where the skid resistance is at a particularly high level. Viner et al. (2005) stated that this
indicated that improving skid resistance does not always reduce the crash risk to the mean trend.

Figure 2-8d shows a comparison of the mean and the 95 percentile crash risk for single carriageway nonevent sections (typical of all the categories) indicating the wide range in crash risk within individual site
categories at all levels of skid resistance.

In a recent US study, Eisenberg (2003) analysed the relationship between precipitation and traffic crashes between
1975 and 2000 (over 455,000 traffic crashes and precipitation levels from over 20,000 weather stations across the
US). Eisenberg (2003) showed a surprising result that when precipitation levels are averaged over the month,
increasing monthly precipitation levels across the US demonstrated a decreasing risk of a fatality crash, which was
contrary to established research. As the severity of crashes falls (from fatal, to injury, to property damage only) the
association of crash risk and precipitation levels increasingly becomes more positive. In terms of skid resistance
research and when the data were analysed on a daily basis; the research demonstrated a positive and significant
increasing fatality crash risk with increasing days since the last significant rainfall. The analysis demonstrated that
if there were 21 or more days since the last rainfall the road is nearly three times more dangerous in terms of fatal
crashes and nearly twice as dangerous in terms of nonfatal crashes in comparison to the case where rainfall fell two
days ago. Eisenberg concluded that the results have a clear implication: drivers are at significantly elevated risk
when precipitation falls after a dry period of several days or more.

2.3.4 Skid Resistance Standards

Introduction
As shown in Section 2.3.3, a relationship has consistently been found between skid resistance and crash rates.
However, it is important to remember (as discussed in 2.3.2) that incidents on the highway rarely have a single
contributory factor and are defined as being rare, random and multifactor events where the driver has failed to
cope with their environment (Land Transport NZ, 2004). The human, vehicle and other road environment factors
and their interactions need to be considered, meaning that no simple method exists for defining, in terms of skid
resistance, a safe or a hazardous surface, or a skid resistance value at which a site automatically transforms from
safe to hazardous (Austroads, 2005b).
As research has clealy shown, crash rates fall following treatments that improve skid resistance. This established
relationship can create a desire to provide a uniformly high level of skid resistance across an entire road network.
However, this would prove to be prohibitively expensive, unsustainable in terms of resource usage, largely
unnecessary and would not generate the cost-benefits associated with a better and more targeted strategy.
Accordingly, Austroads (2005b) states:
A common theme of historical, and many current, strategies for managing skid resistance is
the equalisation of crash risk across a road network by maintaining appropriate levels of skid
23

resistance at all locations, based on site characteristics. These can be grouped into nominal

Chapter 2

site categories. This approach assists in ensuring that the provision of high friction surface
treatments can be targeted to where they are most needed, such as at intersections, the
approaches to traffic signals and on tight curves.

UK based Standards of Skid Resistance

In the UK, the trunk road network is the responsibility of the Highways Agency in England, National Assembly in
Wales, Department of Regional Development in Northern Ireland and Scottish Executive Development Department.
These bodies, along with research primarily undertaken by the Transport Research Laboratory and its predecessors,
produce a design guide and a suite of contract documents known as the Design Manual for Roads and Bridges
(DMRB) and the Manual of Contract Documents for Highway Works (MCHW). These standards are arguably the
strictest in the world and are the basis upon which much of the New Zealand Standards were developed.
Giles (1957) proposed the first comprehensive standards on skid resistance suggesting four site categories based on
requirements for wet skid resistance and a minimum coefficient of friction as measured by the sideways force
coefficient (SFC) device at 30m/hr (SFC30). The requirements were based upon comparisons between SFC data at
accident blackspots and SFC taken at random on similar sites. The site categories had SFC30 limits of 0.6, 0.5, 0.4
and no requirement.
In 1970, the earlier target values were reviewed and included in the UK Ministry of Transport s Marshall Committee
report, Ministry of Transport, (Ministry of Transport, 1970). The category for no SFC was removed and the
requirement for the highest category was reduced from 0.6 to 0.55. Based on this earlier work, Salt and Szatkowski
(1973) developed LR 510: A guide to levels of skid resistance for roads which defined desirable levels of wet
skidding resistance based on the difficulty of the site. This guide suggested values of skidding resistance for
different site categories from Most difficult (e.g. roundabouts and sharp bends); Average sites (e.g. motorways
and high speed roads; heavily trafficked urban roads) and Other sites (e.g. mainly straight roads with easy
gradients and curves and no junctions). The main difference with the Transport Road Research Laboratory (TRRL)
proposals was that they included risk ratings, which could be determined locally for the crash potential from the
geometry and location of each site.
In 1976 the UK Department of Transport (now UK Highways Agency) published Technical Memorandum H16/76
which set down requirements for aggregates and texture depth for new bituminous roads according to different types
of highway layout and traffic flow.
The UK National Skid Resistance Survey undertaken by Rogers and Gargett (1991) in 1981 and the subsequent
crash risk analysis confirmed the findings of LR 510 but took the work of Salt and Szatkowski (1973) further by
defining categories of site more precisely according to crash risk. An assessment of the benefits of monitoring the
road network and the introduction of the standard (UK HD 15/87) in 1988 demonstrated that it would be extremely
cost effective in reducing road casualties with a predicted return of 5.5 for every pound spent due to the reduction
in crashes. This mandatory standard (later contained in HD 28/88 and HD 28/94) recognised that skid resistance in
practice, due to the nature of individual sites within site categories, could vary significantly. It therefore specified
thresholds for investigation (Investigation Levels) rather than thresholds for intervention. Sites with skid resistance
and/or surface texture values at, or below, the appropriate Investigatory Level (IL) were then subject to a site
investigation to determine whether or not monitoring and/or remedial action were needed to improve skid resistance,
or if an alternative action was needed to improve the site in some other way. In attempting to equalise the risk of
skidding crashes across the UK trunk network, the varying characteristics of the network (including geometry and
24

Chapter 2
other factors) were defined in terms of 13 site categories e.g. dual carriageway no-event, single carriageway noevent, approach to major intersection, bend of less than 250m etc. These site categories were then given an
Investigatory Level that was specified in terms of the Mean Summer SCRIM Coefficient (MSSC).
The HD 28/94 standard has recently been revised to incorporate the findings of the crash risk study undertaken by
Viner et al. (2005). The main concepts of the earlier standard were proven and they remain features of the revised
standard published as HD 28/04 and reported in Sinhal (2005). The new developments that have been incorporated
into the revised standard are summarised from (Sinhal, 2005):

The incorporation of developments to the SCRIM and new updated survey procedures;

Removal of slow speed testing at sharp bends and roundabouts (due to improvements in the SCRIM
measurement system);

Rationalisation of the SCRIM site categories from 13 to 10;

Revision of some of the investigatory levels; and

Introduction of a range of investigatory levels for each site category.

Importantly, the revised standard highlights the importance of engineering judgement in setting appropriate
Investigatory Levels for each site and in conducting site investigations to determine those sites most likely to deliver
improvements in crash risk as a result of providing better skid resistance. The flexibility to undertake these
assessments was contained within the previous standard, however experience from implementation had indicated
that this part of the policy was not being applied robustly in practice. Sinhal (2005) reports that:
A major feature of the new standard is the greater range and detail of the advice included to
guide those responsible for providing adequate skid resistance in the application of the
standard. Clear advice and guidance is provided in setting investigatory levels and carrying
out investigations to determine if treatment is required.
A higher or lower Investigatory Level than indicated in the standard may be assigned if justified by the observed
crash records and local risk assessment. Table 2-3 from Haydon (2005) and prepared from data in Sinhal (2005)
summarises current Investigatory Levels on UK trunk-roads and motorways.

25

Chapter 2

!"

&&
'
"

&&

,"

'
( &&

"

)"

&
"
)"

-.

#. /

-0

12. /

" 3

4 "

" 3

Note: The dark boxes represent the recommended levels of IL for roads with normal levels of risk. The light boxes
represent a discretionary lower level of IL, to be used in low risk situations (low traffic volumes, low incidence of
crashes).
Table 2-3: Diagrammatic Representation of Investigatory Levels in HD 28/04 (Haydon 2005)

New Zealand Standards of Skid Resistance

New Zealand s skid resistance standards are based on those adopted by the UK, with some adjustments made that
were based on an analysis performed in 1997 involving wet skidding injury crashes in New Zealand over the period
1990-1994 (Transit NZ & Roading NZ, 2005). Transit New Zealand first published TNZ T/10 in 1999 Specification
for skid resistance investigation and treatment selection (from which the Investigatory Levels, ILs are shown in
Table 2-4). A revised and current specification was published in 2002 (Transit New Zealand, 2002). The T/10
specification also specifies Threshold Levels (TL s) for each site category which are set 0.10 below the level of the
Investigatory level. The IL is set at a level that should trigger an on site data investigation, and if required, treatment
should be undertaken within the next financial year. The TL is set at a level that is intended to trigger an immediate
response and some form of action at the site ranging from a minimum treatment of advisory signing (slippery when
wet signs).
The Transit NZ T/10 (2002) specification does not incorporate the latest revisions of the UK Highway s Agency HD
28/04 standard. It is understood that Transit New Zealand are currently reviewing the T/10 specification. Transit
NZ s policy defines skid resistance for five site categories of IL and TL, using Mean Summer SCRIM Coefficient
(MSSC) to standardise wet-road skidding crash risk across the network (Cook et al., 2004).

26

Chapter 2

Site Categories from TNZ T/10

Site
Category

Site Definition

Investigatory
Level (IL)

Threshold
Level (TL)

0.55

0.45

0.50

0.40

0.45

0.35

Approaches to:

Railway crossings
Traffic Lights
1

Pedestrian Crossings
Roundabouts
Stop and Give Way controlled
intersections
One Lane Bridges (incl. bridge deck)

Curve <250m radius

Down Gradients >10%
Approaches to road junctions (on the SH
or side roads).

Down Gradients 5 10%

Motorway junction area including
On/Off Ramps

Undivided carriageways (event free)*

0.40

0.30

Divided Carriageways (event free)*

0.35

0.25

Notes * Event Free, where no other geometrical constraint, or situation where vehicles may be required to brake
suddenly, may influence the skid resistance requirements.
Table 2-4: Investigatory Levels (IL) for identifying sites requiring inspection (Transit New Zealand 2002a)

Level of Skid Resistance and the Road User

One of the significant difficulties in terms of the level of the actual pavement surface skid resistance is that the road
user (the human) currently has no way of seeing and, therefore, knowing how much friction is available at any
point of time on a given surface to interact with the vehicle dynamics to allow safe negotiation of that section of
road. Furthermore, the skid resistance available, unlike many other structural pavement condition factors, does not
deteriorate in a simple linear or other predictable nature. Any method of measuring the skid resistance of a
pavement surface requires an understanding of the dynamics of the measuring vehicle and how these correlate with
the dynamics of a normal road-based vehicle while braking and / or cornering.

2.4

Vehicle System Dynamics

2.4.1 Introduction
Lamm et al. (1999) include a comprehensive discussion of the effects of the vehicle system and driving dynamics on
skid resistance, including the frictional behaviour between the tyre and the road surface and its speed dependence.
A brief summary based on Lamm et al. (1999) is given below.

27

Chapter 2

2.4.2 Driving Dynamic Fundamentals

Driving dynamics deals with the physical laws that govern the motion of a vehicle with respect to the vehicle s
properties and the properties of the road alignment and pavement (Lamm et al., 1999). To thoroughly describe the
performance of a vehicle on a road requires complex mathematical models. When including the wheels and the
chassis, a vehicle is composed of five individual parts that are connected in a moveable way. If every moveable
body is considered a rigid body with six levels of freedom (three rotational and three translatory) then the whole
vehicle is a 5 x 6 = 30 degrees of freedom object (Figure 2-9).

These movements can be described by a number of

differential equations, although the number of the differential equations increases significantly if other components
of the vehicle (such as the engine, the load, the occupants, etc) are also taken into consideration. During the motion
of a vehicle, different forces are transmitted between the tyres and the pavement as follows:

Normal forces to the road surface (vertical forces); and

Parallel forces (horizontal forces).

The horizontal forces can be projected to the tangential direction (parallel to the travel path of the vehicle) and to the
radial direction (normal to the travel path of the vehicle). The actions of the driver result in reactions of the vehicle,
which act again on the driver, who is then forced to take new actions, and so forth. Additionally, the driver and
vehicle are subject to different disturbances and road environmental factors such as:

Alignment (curves, grades, superelevation);

Pavement characteristics (road surfacing material, the macrotexture and microtexture, the strength of the
material and the underlying foundation materials (deflection); and

Weather conditions (rain, snow, ice and environmental conditions such as the presence of contaminants).

Figure 2-9: Vehicle Movements (Lamm et al., 1999)

Each tyre on the vehicle contributes individually to the transmission of forces while the vehicle is in motion. From
the point of view of vehicle mechanics, it is important to calculate the slip angle and the available tangential force
based on the tyre type if the radial and vertical forces are known. Whilst this approach is sometimes utilised for
crash reconstruction for legal proceedings and court cases, there are no known procedures or guidelines that use this
approach for modern highway geometric design guidelines.

All known driving performance analyses regard the

28

Chapter 2
path stability, acceleration and deceleration, and the forces associated with these as being based on the assumption
that the vehicle is a rigid body and that all forces act at the centre of gravity of the vehicle. From this simplification,
the vehicle is idealised as a point-mass model, and no information is given about the actual distribution of forces, for
example with respect to the individual wheels of the vehicle. From this simple point-mass model, the following
compromises are met:

The whole vehicle mass is concentrated at the centre of gravity;

All forces act at the centre of gravity;

The hypothetical centre of gravity of the vehicle lies on the road surface;

The contact area between tyres and surfaces is generalised and simplified;

Influences caused by sudden movement changes are neglected; and

The three-dimensional movement is reduced to three two-dimensional movements, which are examined
separately on a horizontal plane, on a vertical plane and on a cross sectional plane.

The frictional behaviour between the vehicle and road surface is also significantly simplified by the use of the
simple point-mass model which is then confined by the above idealisations.

These idealisations are reasonable as

long as the actual friction values (which can be measured) are significantly greater than those used in practice for
design policies, thereby providing a safety factor in the design.

2.5

Definitions of Skid Resistance and Friction

2.5.1 Introduction
Woodside and Woodward, (2002) uses the term skidding resistance for the extent to which the road surface
contributes to friction. It is more specifically characterised by the contribution that the surface microtexture and
macrotexture, (discussed in more detail in Section 3.2) make to the available level of surface friction (Austroads,
2005b). Friction is defined in PIARC World Road Association (2003) as the resistance to motion between two
surfaces in contact. Its magnitude is expressed by the coefficient of friction (f) which is a ratio of two forces, one
parallel to the surface of contact between two bodies and opposed to their motion (the friction force) and the other
perpendicular to this surface of contact (the normal force).
Skidding resistance typically refers to wet conditions and can be measured and predicted in a number of different
ways. These all involve sliding some form of rubber over a wet surface and measuring the forces developed under a
known load. Roe and Hartshorne (1998) state that frictional resistance is used to describe those forces that resist
motion when the two surfaces are in contact.

For a road surface, the force resisting motion is developed between a

vehicle tyre and the road surface and is lowest when the surface is wet.
The frictional force allows a vehicle not only to slow down, but also to accelerate and enables vehicle turning
movements. If the moving force of the vehicle between the tyre and the road surface (whether longitudinally or
transversely) is greater than the available frictional resistance of the road surface plus the longitudinal grade or
superelevation force components of the pavement surface, then a vehicle skid can occur, leading to a possible loss of
control type crash.

2.5.2 Components of Friction

Friction between the tyre and the road surface is commonly considered to consist of two main components; adhesion
and hysteresis, both of which are related to speed. These are defined by the UK Highways Agency (2001) and
Oliver (2003b) and are summarised as:
29

Chapter 2
Adhesion sliding resistance between the tyre and road surface with its magnitude determined by the
nature of the materials in contact. Adhesion is generated by the establishment of chemical bonds between
the tyre rubber and pavement aggregate; and

Hysteresis loss of energy caused by the deformation of the tyre rubber by pavement surface projections.

These components of adhesion and hysteresis are illustrated in Figure 2-10.

Adhesion is the result of molecular

bonds formed at localised areas of high pressure. The breaking of these bonds retards the tyre. The hysteresis
component is caused by deformation of the tyre rubber around the pavement surface texture. The rubber is
compressed as it approaches a surface protrusion then decompressed as it moves away. This causes a retarding
force on the tyre and energy is lost from the system in the form of heat and noise. Small-scale surface irregularities
(microtexture) of the aggregate influence adhesion while larger scale irregularities (macrotexture) of the road
surface influence hysteresis.

The tread rubber is a material capable of large, almost totally recoverable

deformations, whereas the pavement surface is considered as rigid and unyielding (Jellie, 2003).

deformation friction

hysteretic bonds
adhesion friction
Figure 2-10: Mechanisms of adhesion and hysteresis (J. W. H. Oliver, 2003b)
The extent to which either adhesion or hysteresis friction are more active depends upon the prevailing conditions.
In dry free rolling conditions, adhesion friction is the predominant component, although in the case of dry sliding at
high speeds, hysteresis friction accounts for more of the frictional mechanism. As the elasticity of the tyre rubber is
dependant on strain rate, the faster a tyre rolls or slides over a road surface, the less the asperities can penetrate the
rubber. Smaller penetrations mean smaller contact area available for adhesion and less hysteresis energy loss, which
results in less braking. Therefore, the faster a tyre rolls or slides, the less friction, because of reduced real contact
area. Furthermore, the smaller the absolute size of the asperities the lower the friction will be.
Cenek et al. (2000) also discuss the effect of the tyre tread stating that the tread pattern is often considered to have a
relatively small effect on skid resistance compared with the influence of the road surface texture. Whatever type of
tyre tread pattern is used there is always a substantial net area of rubber (>50%) in contact with the road. By
contrast, the provision of a harsh and rough pavement surface texture may change the effective area of contact by a
factor as large as 100 compared with a perfectly polished and smooth texture. Cenek et al. (2000) suggest that there
is a greater influence on both adhesion and hysteresis by varying pavement surface texture than by varying tyre tread
pattern.
In contrast, when the pavement becomes wet the presence of water prevents the molecular chemical bonds from
forming at the interface of the aggregate and the tyre and therefore the adhesion component is significantly reduced
30

Chapter 2
(almost to zero) resulting in a greatly reduced available friction. Although the tyre tends to squeeze the water out
from between the tyre and the road surface, this process is incomplete as molecular bonds are formed only where the
microtexture protrudes beyond the water film. Therefore, in wet conditions, the hysteresis component is the most
significant skid resistance component available to resist the vehicle destabilising forces. This is the principal reason
why skid resistance is almost always measured on a wetted surface. The more polished and smooth the aggregate
surface is the less able the tyre is in breaking the film of water to create the adhesion component of friction. The
effect of water on surface friction is further discussed in Section 3.5.
The macrotexture and the microtexture (discussed in more detail in Section 3.2) of the pavement surface both have
an important role in establishing the hysteresis effect and in achieving adequate wet skid resistance. This effect
decreases linearly with speed but at different rates depending upon the macrotexture of the pavement surface. In
simple terms, at low speeds microtexture primarily determines the friction available to the tyre and at high speeds
the macrotexture determines how rapidly the low speed friction falls away.

2.5.3 The Dynamics of Friction Forces

The maximum transmissible friction force depends on the characteristics of the tyre and the pavement and on the
presence of any substance on the contact area between tyres and pavement (Lamm et al., 1999). Figure 2-11 shows
the factors influencing the friction potential between the tyre and the pavement.
Road Surface

Vehicle

Type
material
macrotexture
microtexture

Condition
front/ rear / all
wheel drive
wheel load
tyre characteristics
tyre condition

Condition
dry
wet
winter conditions
slippery, mud

Driving Motion
speed
acceleration
deceleration

Friction
Force

Figure 2-11: Factors affecting the friction potential (Lamm et al., 1999)
The friction forces depend upon the friction coefficient and on the load normal to the transmitting contact patch
between the tyre and the road surface. The pressure of the contact area is not only unsymmetrical but also depends
upon the type of tyre (radial-ply or bias-ply tyre). The actual friction available is therefore not constant and depends
considerably on the tyre elasticity. The type of tyre construction and pressure distribution will therefore influence
the amount of frictional resistance available. Tyre properties including the influence of tyre tread are discussed in
detail in Section 4.2.5. However, most skid resistance testing methods (which are described in Section 4.3), utilise
as far as is possible standardised rubber compounds, smooth tyres and constant tyre pressures to minimise these
effects.
31

Chapter 2
The most important forces active in the contact area between the wheel and the pavement are the longitudinal
(tangential) and the radial or side forces. The longitudinal force in a straight vehicle path causes a driving tractive or
braking movement of the wheel, which in turn causes a difference in the velocity between the vehicle velocity itself
and the relative velocity between the wheel and the pavement.

This velocity difference or wheel slip is a

determinant of the frictional force that is available for a given pavement surface.

During cornering however, the

front wheels are at an angle to the longitudinal axis of the vehicle, thereby causing the development of a radial (side
force) that is responsible for radial (transverse) friction (Lamm et al., 1999). When cornering, and therefore when
both the longitudinal (FT) and the radial (FR) forces are present, the resultant force should not exceed an upper limit
(maximum friction force, Fmax) as shown in Figure 2-12. If this force is exceeded, loss of control (whether partial or
full) can occur which can lead to a vehicle crash occurring. For example, during panic braking on a horizontal
curve, the maximum longitudinal force is almost attained and the remaining radial component of the resultant force
is not sufficient to keep the vehicle in its desired path, thus causing the vehicle to slide off the road. The various
influencing factors of the radial force and the longitudinal force are shown in Figure 2-13 with braking forces on the
left of the ordinate and accelerating forces on the right.

Figure 2-12: Horizontal forces between Road Surface and Motor Vehicle (Lamm et al., 1999)

Figure 2-13: Relationship between longitudinal (tangential) and radial (side force) under varying conditions
(Lamm et al., 1999)
Referring to the alpha numeric letters on Figure 2-13 the following vehicle braking scenarios are discussed:
a.

A curve with a maximum available friction (in general, an elliptic curve);

b.

For a slip angle of

c.

So long as maximum longitudinal forces are generated despite wheel slip, the curves with

= 0 the radial force is not necessarily zero;

constant slip angles fall back into the characteristic curve field;
32

Chapter 2
d.

When a wheel locks up, an inner circle with a minimum friction value is generated on the
braking side;

e.

For small slip angles, i.e. the radial (side) force FR < FRmax, the tangential (longitudinal)
force (FT < FTmax) has minimal impact on the radial force;

f.

In the area of maximum radial force, the longitudinal force has a strong impact on the
radial force;

g.

Curves with constant slip angles (greater than *) fall back into the characteristic curve
field if the radial force has a maximum according to *; and

h.

Curves with a constant slip have the same form as curves with a constant slip angle ( =
constant) but rotated by 90.

In addition to the physical phenomenon of the interaction between the vehicle and the road surface, there are other
forces that can act on the vehicle such as side-wind forces. Other than in extreme conditions, these forces are
usually small in comparison to the physical phenomenon described above and are not discussed further.
When designing a highway, the designer needs to use design values that are conservative enabling a built-in factor
of safety as input to the various models considered in the design process. In the case of friction, these design
values are expressed in terms of maximum permissible values that are directly associated with the driving and
braking limits of a vehicle. The driving and braking limits of a vehicle are defined as the maximum longitudinal
(tangential) and lateral (radial / side) accelerations or decelerations that can be tolerated by a vehicle under steady
state conditions as shown in Figure 2-14.

Figure 2-14: Definition of the friction potential (Lamm et al., 1999)

The friction values arising from the interaction between tyres and pavement of a rolling vehicle in the longitudinal
direction on a wet but clean road surface and at the moment of impending skidding satisfy the following equation:

F Fmax = p Q
Where

Eq 2-1

Fmax = maximum possible (transferable) friction force

p

= peak friction factor

Q = weight force on the wheel.

This resultant horizontal force (from the longitudinal and the radial direction of travel, refer to Figure 2-12) can be
separated into the two perpendicular forces, FT and FR acting in the major directions that are of importance in
highway design. For these two forces, the following relationship is approximately valid:
33

Chapter 2
2

F = FR + FT
2

Eq 2-2

The friction factor f, is expressed as:

f =

horizontal force ( F )
vertical force (Q)

Eq 2-3

The friction between a tyre and a road surface depends on many variables making the definition of design values
difficult (discussed in detail in Section 3). These factors are shown in Figure 2-11 and summarised in Lamm et al.
(1999) as:

The pavement structure, the road surface features due to material properties, the varying traffic demands,
the time elapsed since construction and the time period of the year;

Weather variations ranging from summer dry, humid and wet with different waterfilm depths to winter dry,
wet, slush, snowy, and icy road surface conditions;

Vehicle parameters such as wheel load, slip angle, position of centre of gravity, steering wheel angle, etc;

The tyre type, width, profile, rubber synthesis, and inflation pressure; and

The speed and the driving behaviour.

In highway geometric design, the standard approach is to examine the independent friction factors for the radial
(side) and tangential direction of travel assuming approximate Newton friction values as follows:

fT =

FT
Q

Eq 2-4

fR =

FR
Q

Eq 2-5

f 2 = f R + fT

2*

f R f max
Where

Eq 2-6
Eq 2-7

* = friction circle assumed

f = friction factor
FT = tangential force (horizontal force)
FR= radial force (horizontal force)
fT = friction factor in tangential direction
fR = friction factor in radial direction
Q = weight force, (vertical force)

There is a need to distinguish between three possible values of friction:

Friction demanded, which is friction required under specific vehicle movements, e.g. during cornering, in
order to allow stable vehicle motion. Friction demanded stems from physical relationships such as the radii
of the curve and the speed of the vehicle;

Maximum friction values, being friction values available under specific and extreme conditions; and

34

Chapter 2
Permissible friction values, being the friction values assumed for highway geometric design purposes
(design values) which allow for adequate safety potential.

The relationship between the friction factor in the tangential direction, fT and wheel slip is shown in Figure 2-15.
The permissible friction factor in the tangential direction is defined at 100% wheel slip (wheels locked up), which
equals the skidding coefficient G.

In order to define the term wheel slip, the following assumptions need

consideration (Lamm et al., 1999):

When a driving or braking torque is applied to a tyre, the tread elements are compressed or stretched
accordingly, prior to entering the tyre-pavement contact area.

Consequently, the tyre rotates without the equivalent translatory progression (vehicle /movement). This
phenomenon is referred to as (longitudinal) wheel slip.

The longitudinal wheel slip is the percentage difference between the distance that the tyre travels when subject to a
torque to the distance it travels when free (0% slip). If the torque is a driving torque, the slip is identified as being
positive, whereas if the torque is a braking torque, the slip is identified as negative (Lamm et al., 1999).

Figure 2-15: The Friction Curve (Lamm et al., 1999)

The same physical conditions also apply in the radial direction where radial friction force is activated on the road
surface through a radial skidding motion of the wheel. Measurements undertaken on various wet, but clean, road
surfaces with standardised tyres developed typical distributions of skidding friction factors in relation to speed
(Figure 2-16).

Some common European design standards (e.g. German) have been based upon a percentile

distribution curve of friction which falls short by only 5 to 10% of the friction for investigated road surfaces.

Figure 2-16: Distribution of Skidding Values G (tangential friction factor, fT) with Speed V (Lamm et al.,
1999) originally developed by Wehner (1965)
35

Chapter 2

2.5.4 Braking and Friction

The road user s interest in surface friction amounts to how long will it take for the vehicle to stop or avoid an object
in an emergency situation. As discussed above the friction factor components are broken into two components
(tangential or longitudinal fT and radial or transverse friction fR) depending upon whether the vehicle is braking on a
straight or on a curved section of roadway. The coefficient of tangential friction (fT ) is a measure of friction in the
direction the vehicle is moving along a straight tangent; the smaller the fT values, the longer the acceleration and
deceleration distances.
PIARC (2003) states the tangential braking distance can be computed from the two equations below. The first
equation (Eq 2-8) assumes a constant value of fT which is a simplification that is often used in design practice. The
second equation (Eq 2-9) is more accurate as it takes into account the fact that friction decreases as speed increases
(discussed in Section 3.2).

In order to use this equation, the relationship between speed, the surface type,

macrotexture and friction needs to be known.

The differences in results from both equations depend upon the

characteristics of the road surface (microtexture and macrotexture differences).

2

V Vf
V .t
BD = i + i
3.6 254( fT G )
BD =
Where

Vi .t
+
3 .6

Vf
Vi

Eq 2-8

V
dV
127( fTV G )

Eq 2-9

BD = braking distance (m)

Vi = initial speed (km/h)
Vf = final speed (km/h)
t= driver reaction time (seconds)
fT = coefficient of tangential friction
fTV = coefficient of tangential friction at speed V
G = grade (%/100).

When both steering and braking need to be accomplished simultaneously (e.g. braking within a horizontal curve),
the available friction at the site is shared between its tangential (longitudinal) and radial (transverse) components.
Consequently the braking distance is increased (Eq 2-10).

BDcurve

Vi V f

V .t
= i +
3 .6
254

Where

f2

Vi
e
127 R

Eq 2-10

BDcurve = braking distance in curve(m)

Vi = initial speed (km/h)
Vf = final speed (km/h)
T= driver reaction time (seconds)
36

Chapter 2
f = coefficient of friction that includes the components of fT (tangential friction) and fR
(radial friction);

f 2 = fT + f R

R = curve radius (m)

e = superelevation (m/m)
G = grade (%/100).
PIARC (2003) states roads need to be designed to allow safe traffic operations even when prevailing conditions are
not ideal. As such, the values of the coefficient of tangential friction (fT) that are used at the design stage assume
poor conditions, namely wet surface and worn tyres. However, they do not assume extreme conditions such as icy,
snowy or flooded road surfaces in which a reasonable driver is expected to reduce speed. The values used in
design vary from country to country, however PIARC (2003) states that they range from around 0.45 at 30km/h to
less than 0.3 at high speeds.
The values of radial or transverse friction (fR) that are used at the design stage are based on a criterion of comfort
rather than safety, which is the speed at which the effect of the centrifugal force is sufficiently uncomfortable to
cause drivers to decelerate. PIARC (2003) states that typical design values range between 0.07 and 0.18. By using
such low (fR) values at the design stage, around 90% of the total friction remains available for braking manoeuvres
in curves which prevents excessive increases of calculated stopping distances.
Friction values that are used in design for calculation of sight distances and braking distances are absolute minimum
values. Existing road surfaces that do not meet these thresholds have obvious skid resistance problems and require
improvement. At locations with higher friction needs and where the probability of braking is increased (e.g.
approaches to intersections, horizontal curves, downhill grades) much higher friction values must be provided
(PIARC World Road Association, 2003). Various design level standards including those for New Zealand are
discussed in Section 2.3.4.

2.5.5 Crash Investigations and Braking Distances

It is important to note that the friction measured under standard skid testing cannot be directly used to calculate
vehicle-stopping distance. The following is an extract from the scope of the ASTM standard method (ASTM E27497) for a locked wheel skid testing device (refer Section 4.3.3). However, the principles remain for other devices:
The values measured represent the frictional properties obtained with the equipment and
procedures stated herein and do not necessarily agree or correlate directly with those obtained
by other pavement friction measuring methods. The values are intended for use in evaluating
the skid resistance of a pavement relative to that of other pavements or for evaluating changes
to the skid resistance of a pavement with the passage of time. The values are insufficient to
determine the distance required to stop a vehicle on either a wet or a dry pavement. They are
also insufficient for determining the speed at which control of a vehicle would be lost, because
peak and side force friction are also required for these determinations.
Henry (2000) states that the reasons for the caveat in the ASTM standard are many, including the following:

The test tyres used in skid testing are not the same as those on the crash vehicle;

The amount of water that is placed on the surface by the skid tester is not the same as that experienced by
the crash vehicle;
37

Chapter 2
A four-wheel vehicle has front tyres that displace the water so that the rear wheels encounter less water
than the front;

The skid resistance varies with speed and therefore the relationship between skid resistance and speed for
the entire range of the crash vehicle (i.e. from the initial speed at which the wheels are locked to zero speed
for a vehicle with non ABS brakes);

The crash vehicle often will not maintain a locked wheel condition in order to maintain directional control;

Seasonal and short term variations in skid resistance are difficult to predict and therefore the measurement
would be different than that at the time of the crash;

The load on the test tyre of a skid tester is not the same as that on the tyres of the crash vehicle;

The suspension characteristics of the crash vehicle and the skid tester are not the same and the load
distribution on the tyres of the crash vehicle during braking depends upon the condition of its suspension.

Determining the surface friction factors critical to a crash scene to enable calculations of vehicle speed and / or the
friction that was available at the time of the crash is a difficult task due to these above factors. Henry (2000) states
that pavement friction measurements are useful in evaluating the safety of a pavement relative to other pavements in
the system, but they should not be used for quantitative determinations of stopping distances.
Chapter 3 that follows discusses the factors that affect road surface skid properties.

38

Chapter 3

3 FACTORS AFFECTING ROAD SURFACE SKID PROPERTIES

3.1

Introduction

The surface friction available at any particular time, either to the road user or as tested dynamically by a friction
tester, depends upon many variables. These variables can be grouped together under four main surface friction
factors shown in Table 3-1:
Pavement Surface
aggregate factors

Load factors

Environmental factors

Vehicle Factors

Geological properties of the

surfacing aggregate

Age of the surface

Water film thickness and

drainage conditions

Vehicle speed

Surface texture
(microtexture and
macrotexture)

Traffic intensity and

composition - equivalent
vehicle loadings

Surface contamination

Angle of the tyre to the

direction of the moving
vehicle

Chip size and shape

Road geometry

Temperature

The wheel slip ratio

Type of surfacing
(concrete, asphalt mix and
mix design, chip seal
surface and design method)

Traffic flow conditions

The combined seasonal

effects and short-term
variations

Tyre characteristics
(structural type, hardness
and wear)

Rainfall

Tyre tread depth and pattern

Refer to Sections 3.5

Refer to Section 4.2.

Refer to Sections 3.2 and

3.3

Refer to Section 3.4

Table 3-1: Factors affecting the skid properties of a road

All of the above factors are sources of variability in the measured coefficient of friction and / or the friction
available to a road user. These factors are discussed in the sections that follow and as shown in Table 3-1. Some
factors are reasonably stable over time and do not significantly affect the variation of the measured coefficient of
friction, whereas other factors do. In practice, these variables are often difficult to consider independently especially
when measurement surfaces are exposed to a significant number of unpredictable environmental effects. Only a few
of the factors are currently regularly monitored on a road network level, which means that it is difficult to monitor
their inter-relationships.

3.2

Road Pavement Surface Texture profiles

3.2.1 Introduction
Vehicle forces due to accelerating, braking and cornering are transmitted between the tyres and the pavement due to
friction. The maximum transmissible friction force depends upon the characteristics of the tyre and the pavement
surface and on the presence of any substance on the contact area between tyres and the pavement surface (Lamm et
al., 1999). Pavement texture is the feature of the road surface that ultimately determines most tyre/road interactions,
including wet friction, noise, splash and spray, rolling resistance and tyre wear. The surface of a road can be
considered from a number of perspective views as illustrated in Figure 3-1 and Figure 3-2 that describe the surface
texture wavelengths, the factors that influence that wavelength and the possible consequence to the road user. Lay
(1998) gives the following summary of the descriptive scales and their wavelength spectrum related to surface
characteristics, ordered from the largest to the smallest size:

39

Chapter 3
roughness or uneveness which provides a broad description of the longitudinal profile of the road and is
the facet of surface felt by the passing motorist (at least 500mm wavelength);

megatexture or bumps which are corrugations and faulting (between 5 and 500mm wavelength);

macrotexture which is the surface characteristic perceived by someone standing or squatting on the road
surface. It is the picture of the road surface that is presented to the naked eye and includes the individual
pieces of aggregate. Macrotexture can be felt by the human hand (between 0.5 and 50mm wavelength and
a texture depth range commonly between 0.2 and 3mm); and

microtexture which is the characteristic perceived by someone lying on the road and using a magnifying
glass or their finger tips to detect the roughness of the surface of the exposed pieces of aggregate. This
texture has a wavelength of less than or equal to 0.5mm and is important at a pitch down to about 25 m
(which means that it may need to be assessed using electron microscopy).

Figure 3-1: Road surface wavelength spectrum, from Huschek, (1990), source Lay, (1998)

Figure 3-2: Illustration of Texture Ranges (Sandberg, 1999)

Sandberg, (1999) illustrates the texture ranges in terms of a cross section of the road surface. This is shown in
Figure 3-2.

Roughness (unevenness) and megatexture are important texture wavelengths when considering

pavement deterioration. However, their larger scale surface irregularities do not have the same influence on skid
resistance when compared with microtexture and macrotexture.
40

3.2.2 Microtexture

Chapter 3

Microtexture is defined as the amplitude of deviations from the surface plane with wavelengths less than or equal to
0.5mm in length and depth. It is considered as a first order surface irregularity measured at the micron level of
harshness. Microtexture is the micron scale harshness of the surface aggregate and results from irregularities at the
surface of the pavement aggregates. Texture on this scale is a function of the crystalline structure of the aggregate
mineralogy / petrology in the upper layer of the surfacing material and its interaction with the climate and traffic
loading (Rogers & Gargett, 1991). Microtexture can also be affected by the fine particles such as sand in asphaltic
concrete and concrete surfacings. The UK Highways Agency (2001) refers to the fine scale microtexture of the
surface aggregate as the main contributor to sliding resistance and the dominant factor in determining wet skid
resistance at low speeds. Figure 3-3 details the relative texture depths and lengths.

Figure 3-3: Microtexture and Macrotexture (Highways Agency, 2001)

3.2.3 Macrotexture
Macrotexture is the coarser texture formed by the general shape of, and the spaces between, the particles or
grooves in the surfacing (P. Roe et al., 1998). Macrotexture, a second order surface irregularity measured in
millimetres (often incorrectly termed by practitioners as surface texture or texture depth) can be defined as the
amplitude of deviations (irregularities) from the plane with wavelengths ranging between 0.5 and 50 mm in length
(PIARC, (2003) and Austroads, (2005b)). Macrotexture as shown in Figure 3-3, results from the presence of coarse
aggregate particles at the road surface and is determined by the chip size, aggregate grading and spacing of the
aggregate for chip seal and asphalt layers, respectively. It can be increased by the presence of grooves that have
been purposefully manufactured in concrete or asphalt mixes.
Macrotextures greater than 0.8mm are desirable (PIARC World Road Association, 2003), because this provides
drainage channels in wet conditions for the water to escape from the tyre-pavement interface and allows the vehicle
tyre tread to deform, thereby creating the hysteresis forces discussed in Section 2.5.2. The greater the capacity of
the drainage channels, the more reduction in water spray that is generated from the road surface, which has a
significant benefit in terms of driver visibility in wet conditions.
Macrotexture takes the following two forms (Jellie, 2003), as shown in idealised form in Figure 3-4:
41

Chapter 3
Positive texture which is formed by aggregate particles protruding above the plane of the surface.
Typically, such textures are formed by the application of aggregate chippings and rolling them into a
smooth surface, e.g. AC40 and chip seal surface dressings. Positive texture may also be formed by removal
of the matrix to expose aggregate particles due to ravelling; and

Negative texture which is formed when the texture of a material largely comprises voids between particles
whose upper surfaces form the generally flat plane of the road surface. Negative texture is found in Porous
Asphalt and Stone Mastic Asphalt (SMA) surfaces.

Figure 3-4: Idealised positive and negative macrotextures (Walsh, 2000)

In addition to describing the form of texture in the vertical plane (positive or negative texture), the internal structure
of the surfacing layer can be described as either:

Porous, which allows water to pass through the surface layer, e.g. Open Graded Porous Asphalt (OGPA);

Impermeable, which prevents the ingress or flow of water through the texture, e.g. Chip Seal or AC40; or

Containing sub-surface voids, which allow water to flow through their texture but prevent ingress of water
to the lower layers, e.g. Stone Mastic Asphalt (SMA).

Too much macrotexture is undesirable as it causes greater rolling resistance, greater tyre wear and excess car tyre
noise. Therefore its use is limited in urban areas.

3.2.4 Surface Texture, Friction and Speed

Microtexture controls the maximum skid resistance attainable at low speed and is often described as the low-speed
friction parameter. However, it affects the level of skid resistance at all speeds as shown in Figure 3-5. The
macrotexture determines the rate at which skid resistance is lost as speed increases (PIARC World Road
Association, 2003). It is often described as the friction-speed gradient parameter. Figure 3-5 shows the extremes of
pavement surface texture and demonstrates the relationship between microtexture, macrotexture, speed and friction.
The diagram demonstrates that whilst good microtexture may be sufficient to provide adequate friction at low
speeds, good micro and macrotextures are necessary at higher speeds.

Whilst all four surfaces decrease in skid

resistance as speed increases, the decrease is not as rapid when the pavement surface has a rough macrotexture.
42

Chapter 3
It is generally well known that wet friction reduces as vehicle speeds increase and that macrotexture depth governs
the extent of the reduction. As a consequence, minimum levels of macrotexture depth are specified for most
countries including on State Highways in New Zealand (Transit NZ & Roading NZ, 2005).

Roe et al. (1998)

identified from coefficient of friction measurements made with a locked-wheel skid tester over a speed range from
20km/h to 130km/h that macrotexture depth has an impact on friction loss at speeds as low as 50km/h. This can be
seen by the difference in coefficients of friction between Cases B and D in Figure 3-5 from PIARC (2003).

Figure 3-5: Relationship between microtexture, macrotexture, friction and speed (PIARC World Road
Association, 2003)
This finding that low-textured surfacings lose friction at a faster rate than high-textured surfaces, even at
comparatively low speeds, was very significant as previously this had not been taken account of in design and
maintenance standards for lower speed environments (e.g. urban environments). However, Cenek et al. (2000)
argued that this represented an unlikely and extreme situation, because the analysis was based upon results obtained
with smooth testing tyres. Cenek et al. (2000) then investigated actual dry and wet braking performance of
commercial car tyres on road surfaces having the same coefficient of friction at 50 km/h (as measured by SCRIM
discussed in Section 4.3.2), but different texture depths, to determine whether this effect was significant for normal
operating conditions. Cenek et al. (2000) then concluded from their research that reported texture depth influences
on low and high speed skid resistance are negated by vehicles fitted with anti-lock braking systems (ABS) and tyres
with adequate tread depth. Under such operating conditions, wet skid resistance appears to be insensitive to both
speed range (over a 25 to 100km/h) and macrotexture depth.

Under locked-wheel braking, wet skid resistance

reduces with increasing speed but not at the rate indicated by previous research.

It was conjectured that the

difference in the research was due to the tyre tread providing extra texture depth in addition to that provided by the
road surface.

3.3

Aggregate Properties

3.3.1 Introduction
The term aggregate refers to granular mineral particles that are widely used for highway pavement construction
including bases, subbases and backfill. Aggregates are also used in combination with a cementing material to form
concretes for bases, subbases, wearing surfaces, and drainage structures. Sources of aggregates include natural
43

Chapter 3
deposits of sand and gravel, pulverized concrete and asphalt pavements, crushed rock, and blast furnace slag. The
most important properties of aggregates used for highway construction, summarised from Wright, (2004) are:

Particle size and gradation. The blend of particle sizes in the mix affects the density, strength and
economy of the pavement structure;

Hardness and resistance to wear. Aggregates should be hard and resist wear due to the loading from
compaction equipment, the polishing effects of traffic, and the internal abrasive effects of repeated
loadings;

Particle shape and surface texture.

The shape, size and texture of the aggregate is important in

determining the thickness of the pavement or surfacing layer, the micro and macrotexture of the surfacing;

Durability or resistance to weathering. Aggregates should be resistant to disintegration by weathering;

Specific gravity and absorption. These are important properties required for the design of concrete and
bituminous mixes;

Chemical stability. The effects of the chemical composition of the aggregate particles, especially in regard
to the aggregates affinity to water; and

Freedom from deleterious particles or substances. Specifications for aggregates used in bituminous mixes
usually require that the aggregates be clean, tough, durable in nature and free of excess amounts of flat or
elongated pieces, dust, clay balls, and other objectionable material .

3.3.2 Petrological types of Aggregates used in Pavement Construction

A wide range of rocks of igneous, sedimentary and metamorphic origins and a number of artificial aggregates are
suitable for use in bituminous and unbound granular mixes. The general requirement is for rocks which will be
durable (i.e. resistant to abrasive wear), strong (i.e resistant to slow crushing loads and to rapid impact loads) and in
the case of wearing courses and chip seal applications, resistant to the polishing action of traffic. Where applicable,
they should demonstrate good adhesion with bituminous binders (Smith & Collis, 2001).
As early as 1935, B.H. and R.G. Knight carefully described the desirable qualities of rocks for the purposes of road
pavement aggregates, relating mineralogical and textural details to strength and, to a slight degree, their polishing
characteristics (Smith & Collis, 2001). Considerable emphasis was given to defining the best type of rock for a
particular purpose. However, the best economic use of available raw materials has proven to be of far greater
significance than the search for exactness of matching raw materials to a particular purpose, due to:

High costs of haulage of aggregates; and

Depletion of some natural resources (sustainability issues).

3.3.3 Desirable properties of Aggregates for Bituminous Surfacings

Smith and Collis (2001) further defines the main engineering qualities required from an aggregate to be used in a
pavement surfacing from works of Hosking (1970), Hartley (1974) and Lees and Kennedy (1975) as:

Toughness, which is resistance to slow crushing and resistance to rapid loading;

Hardness, which is resistance to abrasion / attrition;

Resistance to polishing (for wearing course materials and chip sealed surfaces);

Resistance to stripping, which is the ability to maintain adhesion to any bituminous binder with which the
aggregate is used;

Resistance to weathering effects in the pavement (e.g. resistance to frost action, swelling and softening by
water); and
44

Chapter 3
Ability to contribute to strength and stiffness of total mix by intrinsic aggregate strength and shape
properties.

The above listed engineering properties applicable to surfacing aggregates are documented in Smith and Collis
(2001), and are not discussed in detail here other than the engineering properties most directly applicable to skid
resistance.
Neville (1974) states that the topography of a stone exposed in the surface of a road is affected by four main factors:

Abrasion;

Polishing;

Differential wear; and

Weathering.

The actual texture at any moment in time is determined by the balance between these factors which are described in
the sections following.

3.3.4 Abrasion of Aggregates

The abrasion of road surfacing aggregates takes place by the action of vehicle tyres carrying detritus and grinding
away material from the exposed aggregate. This detritus is in part material worn from the road surface and in part
naturally derived sand and clay materials and artificial debris carried or blown onto the road surface (Maclean &
Shergold, 1958). The severity of the abrasive action would be expected to be proportional to the density of traffic
and the softness of the aggregate (Neville, 1974).

Hartley (1974) considers the important petrographical

characteristics as being:

Degree of hardness and proportion of hard minerals;

Proportion, orientation and distribution of cleaved minerals;

Grain size, the nature of intergranular bond; and

Degree of liability to chemical alteration of the mineral content.

Neville (1974) further states that abrasion by traffic tends to smooth the projections of the aggregate, reducing the
angularity and micro-texture.

The pressures set up between the tyre and the aggregate are therefore decreased,

leading to a lowering in the level of resistance to skidding. In extreme cases, the macrotexture is reduced by planing
of the surface leading, additionally, to reduced skidding resistance at higher speeds (as was discussed in Section
3.2).
The abrasion characteristics of aggregates ordered by geological category and origin are summarised in Table 3-2
from Smith and Collis (2001):

45

Chapter 3

Geological
Category
Igneous rocks

Geological properties related to resistance to abrasion

High free silica content tends to resist abrasion better than the basic rocks which have a high
ferromagnesium content due to the hardness and lack of cleavage of the quartz in the former;
Chemical decomposition generally results in increased abrasion (especially with high degree of
peripheral alteration as this can destroy the intergranular bond and allows the harder cores to be
plucked from the rock surface);
Vesicular texture generally reduces the resistance to abrasion.

Sedimentary
rocks

The low hardness and cleavage facility of minerals contained in the younger and more porous
limestones and dolomites renders them liable to rapid wear;
Abrasion resistance of siliceous rocks is almost entirely dependent upon the nature of the intergranular
bond (e.g. flint and quartzite are highly resistant, whilst poorly cemented ferruginous and calcareous
sandstones are soon abraded due to plucking of the grains;
Mixed mineral sedimentary rocks such as greywacke tend to have poor abrasion resistance, although
this is not always the case.

Metamorphic
rocks

Of those used in road surfacings, gneisses tend to behave similarly to igneous rocks of the same
mineralogy while hornfels and quartzites have a high abrasion resistance due to their hard mineral
content and dense interlocking texture.

Artificial
aggregates

Steel and iron making slags (e.g. melter slag) in general exhibit a higher abrasion resistance than blast
furnace slags.

Table 3-2: Geological properties of aggregate origins related to resistance to abrasion

3.3.5 Polishing of Aggregates

The importance of aggregate microtexture in tyre / road friction has been recognised since research by Knill (1960)
drew attention to the significance of detailed aggregate petrology in promoting or opposing development of a state
of polish on particle surfaces. Rocks which contained minerals of sufficiently different hardness or which were
friable, i.e. consisted of grains rather insecurely cemented together, were found to give high polishing resistance
(Smith & Collis, 2001).
Neville (1974) states that the term polishing has been commonly used to describe any general smoothing of an
aggregate including the rounding that takes place by abrasion. However, in a more strict sense, polishing should be
used to describe the process that leads to the highly glossy condition that is achieved when an actual flow of the
surface takes place. Under these conditions, an amorphous layer is formed called the Bielby layer (named after
Beilby (1921) who published illustrations of calcite before and after polishing, which showed the flowed material
covering over scratches incurred by the earlier abrasion process). The resulting layer of polished calcite was much
more resistant to the abrasive powders, which had previously been used to grind the surface.
This effect has been shown in the polishing of all gemstones, except for diamond, where the final polish is caused by
a momentary fusion and flow of submicroscopic surface projections, spreading a liquid-like Beilby layer over the
polishing plane (Neville, 1974). In highway or airfield pavements, the state of polish attained results from the
opposing effects of traffic (tending to increase the state of polish) and by weathering where frost action, wetting and
drying and temperature changes all tend to remove or reduce the polished state (Smith & Collis, 2001). Whilst this
effect is most clearly demonstrated in the case of carbonate rocks, Szatkowski and Hosking (1972) have shown that
weathering effects are still significant with other rock types. Figure 3-6 shows the polishing of an aggregate at the
46

Chapter 3
microscopic level using a Scanning Electron Microscope (SEM) and the resultant loss of microtexture. Figure 3-7
demonstrates the increased weathering or abrasion of an initially polished aggregate, which resulted in a subsequent
increase in skid resistance due to a reduction in traffic intensity on the route, following the construction of a new
bypass.

Figure 3-6: Scanning Electron Microscope photographs of an Aggregate in an Unpolished and Polished state
(Austroads, 2005b)

Figure 3-7: Increase in Skid Resistance due to a decrease in traffic on UK Trunk Road A4 (Smith & Collis,
2001)
The polishing characteristics of aggregates ordered by geological category and origin are summarised in Table 3-3
from Smith and Collis (2001):

47

Chapter 3

Geological
Category
Igneous and
Metamorphic
rocks

Geological properties related to resistance to polishing

Rocks with a moderate degree of decomposition by weathering effects have been shown to be more
resistant to polishing1 than fresh unweathered rocks due to the lower hardness of the secondary
minerals (kaolinite, chlorite, serpentine, sericite etc) compared with the primary minerals (quartz and
unaltered feldspar, augite, hornblende etc);
If metallic ore inclusions are disseminated through the rock, the rock has been found to have an
increased resistance to polishing due to the differential hardness of the minerals as with the case of
fractured grains of medium to large size;
Grain size in general appears to have only a slight effect on polish resistance, a coarser grain size
leading to a small increase in polish resistance;
A well dispersed system of vesicles generally increases the resistance to polishing;
Of the metamorphic rocks, only gneisses, hornfels and metaquartzites are generally considered as
likely aggregates for road surfacings it has been observed that the orthogneisses behave in a similar
manner to the equivalent igneous rock whilst the hornfels and quartzites tend to exhibit a high polish
since they generally have a fine-grained interlocking texture and are either monomineralic or
composed of uniformly relatively hard minerals (>5 on Moh s scale).

Sedimentary
rocks
(sandstones)

Clastic sandstones of variable mineral content and certain sedimentary quartzites tend to be polishresistant;
Gritstones have been described2 as demonstrating a retention of a rough non-polished texture in this
category of rocks due primarily to the dislodgement of weakly cemented grains before they can attain a
high state of polish this results in a general inverse relationship between polishing resistance and
abrasion resistance;
Well cemented quartzites (crypto-crystalline flints, cherts and opaline gravels), due to their low
porosity and uniform hardness and in contrast to the more friable clastic sedimentary rocks described
above, demonstrate lower resistance to polishing.

Sedimentary
rocks
(limestones)

A study of carbonate rocks 3 demonstrated that a degree of impurity (as measured by the insoluble
residue and chiefly consisting of quartz sand grains, mica and clay mineral matter) significantly
improves the polishing resistance (especially if the insoluble grain sizes were >0.5mm i.e. sand size
that eliminates the mineral clay and silt sizes);
Thus, a carbonate matrix with a difference in hardness of the matrix minerals (differential solubility)
leads to an aggregate that is resistant to polishing e.g. between quartz (7 on Moh s scale) and calcite
and dolomite (3 and 3.5 to 4 respectively on Moh s scale) and furthermore sand sized mica flakes
(with lower hardness than the carbonate minerals);

Artificial
aggregates

Industrial by-products in general, steel slags (i.e. melter slags) have a lower polishing resistance

than blast furnace slags principally due to the blast furnace slags having a higher vesicularity which
leads to higher angularity on crushing and to the possession of pressure-relieving internal drainage
channels 4.

Special products with superior polishing resistance (e.g. Calcined bauxite).

Notes: 1.
2.
3
4

As measured by the Polished Stone Value (PSV) test refer to Section 4.4.4
Hawkes and Hosking (1972) and Hosking (1970) who extended the work of Knill (1960).
Shupe & Lounsbury (1959) and Gray and Renninger (1963; 1965)
Hosking (1973)

Table 3-3: Geological Properties of Aggregate Origins Related to Resistance to Polishing

48

Chapter 3

3.3.6 Differential Wearing of Aggregates

Where aggregate particles consist of agglomerations of several minerals with different resistances to wear, rough
texture can remain as the particles are worn. Some minerals remain in high relief, whilst others are worn down to a
lower level. This phenomenon can be expected where the minerals are of different hardness or toughness.

This

process of plucking will tend to offset the general smoothing that takes place during abrasion of the particles by
traffic and can also prevent polishing (Neville, 1974).

3.3.7 Weathering
Neville, (1974) states that some minerals are subject to chemical and / or physical changes that result from
weathering. An example is the attack on calcite by rainwater containing dissolved carbon dioxide and / or sulphur
dioxide from the atmosphere. Other examples are the rupture of some minerals by frost action or by the stresses set
up during cycles of wet and drying.

In extreme cases the effect of weathering can lead to a substantial loss of

aggregate and possibly to loss of resistance to skidding, but in general, the effect would be expected to improve
resistance to skidding by helping to maintain a rough texture in a similar way to differential wear.

3.3.8 Summary of Aggregate Sources appropriate for Surfacings

The best source of rocks for the purposes of surfacing aggregates with a high resistance to polishing tend to be those
that (Austroads, 2005b):

Contain minerals of sufficiently different hardness; or are inherently friable (to a degree); and

Have grains that are insecurely cemented together.

In these types of rocks, either different rates of wear will maintain microtexture, or weakly cemented grains will
break away to reveal fresh, unpolished material. The rate of differential wear must however be balanced against the
need to maintain macrotexture.
For the above reasons the source rocks that have the highest resistance to polishing tend to be those with a
sedimentary origin, or with a similar mode of deposition such as the pyroclastic and volcaniclastic rocks. However,
sedimentary rocks can produce aggregates with low or high resistance to polishing, depending upon the variability
and hardness of the grain matrix.
Some rocks within the igneous rock group (including basalt, dolerite, granite, andesite, porphyry, rhyolite and
diorite) can also produce aggregate with a reasonably high resistance to polishing. However, these materials have
usually undergone a moderate degree of weathering. Due to their process of formation, igneous rocks usually have a
very close interlocking crystal structure from which individual grains are quite difficult to remove. This does not
promote the inherent friability conducive to the production of aggregates having the highest resistance to polishing.
However, the more weathered an igneous rock becomes, the greater the reduction in performance in its other
mechanical properties (e.g. durability and strength). As a consequence, igneous rocks which are sufficiently
weathered to produce a very high skid resistance to polishing tend to be unsuitable as a road aggregate due to failing
other specified engineering criteria.
Most metamorphic rocks have been found to be unsuitable for road aggregates.

However, some types of

metamorphic rocks (e.g. quartzite) can produce aggregates possessing extremely good skid resistance properties.

49

Chapter 3

3.3.9 Aggregate Size, Shape and Spacing

Macrotexture is controlled to a certain degree by the size of the aggregate particles. On chipseal surfaces and most
asphaltic mix surfaces, the macrotexture is determined by the space between particles. This is a factor of the size
and evenness of particles on the surface (Highways Agency, 2001). Cubic particles may not embed properly and
flaky particles may be crushed under the weight of the roller. On porous asphalt and Stone Mastic Asphalt (SMA)
thin surfacings, macrotexture is produced by voids between adjacent particles and is controlled by the way the
particles pack together in the mix. Cubic shaped aggregate may pack more closely together and produce smaller
voids (Highways Agency, 2001). In a high stone contact mix such as SMA, flaky shaped aggregate will tend to
crush with trafficking because of the high point to point contact. Aggregate shape and the way particles pack
contributes to macrotexture. In the UK (Highways Agency, 2001), a typical limit for the amount of flaky aggregate
in a bituminous material would be <25%. In NZ, the chip shape is controlled by specifying the maximum ratio of
the average greatest dimension (AGD) to the average least dimension (ALD). For chipsealing purposes, the
maximum chip shape ratio AGD:ALD is 1:2.25 and has been found to give acceptable performance (Transit NZ &
Roading NZ, 2005).
Cenek et al. (1998) state that wet skid resistance of chipseal road surfaces is not only a function of the polishing
resistance of the aggregate, but also of its size, shape and spacing.

Cenek et al. (1998) showed a positive

correlation (R = 0.46) of increasing measured skid resistance with increasing percentage of crushed faces.
Furthermore, as the slip speed increased, the correlations between wet skid resistance and percentage crushed faces
increased while the influence of chip grade (size) decreased. Cenek et al. (1998) suggested that a high percentage of
crushed faces is important for high speed skid resistance while chip grade is more important at low speeds. Cenek
et al. (1998) also demonstrated that a better predictive model could be developed by deriving horizontal and vertical
forces directly from road texture profiles as measured by laser profilometers (refer to section 4.5.3). Using the
model developed by Cenek et al. (1998), significant texture wavelengths influencing wet skid resistance were found
to range from 1 to 50mm, with the highest correlations occurring at 1.3, 1.7, 2.1, 2.7, 27 and 43 mm. Cenek et al.
(1998) recommended that an assessment of the processing characteristics of commonly used impact and gyratory
crushers in use in New Zealand should be undertaken to identify those which generate significant surface texture on
crushed aggregates in the 1-5mm wavelength range.

The texture in this wavelength was therefore shown to be a

significant contributor to the hysteretic component of tyre friction, thereby contributing to an improvement in wet
road skid resistance.

3.4

Loading factors

The measured skid resistance of a pavement surface varies according to the demand (load) placed upon the surface.
Almost all new road surfaces constructed with exposed crushed aggregate have a high skid resistance because the
exposed aggregate particles have good microtexture and sharp edges.

However, under the polishing action of

vehicle tyres, microtexture is reduced, the edges become worn and the skid resistance falls. The loading factors that
influence skid resistance are discussed in the following sections.

3.4.1 Traffic Intensity and Surface Age

The extent to which a road surface becomes polished is directly related to the traffic intensity. Consequently, a
transverse profile of skid resistance across a carriageway will reveal lower levels in the wheel tracks.

The skid

resistance of the surface of both flexible bituminous mix surfaces and concrete roads varies significantly with traffic
volumes and especially with the commercial traffic density. Kennedy et al. (1990) state that:
50

 other conditions being equal, a road with the highest commercial traffic flow will have the

Chapter 3

lowest skid resistance. Under the action of traffic, a new surfacing is polished rapidly until
after about one year [although this depends upon the number of equivalent single axle loads,
the type of surfacing and the geological makeup of the aggregate] it reaches its equilibrium
value. After that time, apart from seasonal variation, it maintains that value until the surfacing
deteriorates or the commercial traffic density changes.
Rogers and Gargett (1991) state that heavy traffic tends to polish away the fine-scale microtexture and for all
aggregates, the heavier the traffic, the lower the wet skidding resistance. Figure 3-8 demonstrates this effect for a
standard UK motorway surfacing consisting of rolled asphalt with pre-coated chipping (PSV 58-60).

Figure 3-8: Effect of traffic on skidding resistance of a typical motorway standard surfacing (Rogers &
Gargett, 1991)
In practice, if environmental factors are removed, because commercial traffic density generally increases each year,
the mean skid resistance gradually becomes lower. As discussed in Section 3.3.5, the state of polish of a surface
results from the opposing effects of traffic and traffic density (tending to increase the state of polish) and by
weathering where frost action, wetting and drying and temperature changes all tend to remove or reduce the polished
state.

The action of traffic on the pavement surface causes it to abrade and / or polish, and thus the skid resistance

measured at any time (if environmental conditions are controlled) is primarily a function of the geological properties
of the aggregates and the traffic loads placed upon them.
Figure 3-9 is a generalised model of how skid resistance has been traditionally considered to decline with time
(Prowell et al., 2003). After construction, there can be an initial rise in skid resistance if the surface aggregate has
been contaminated by a film of bitumen. As the film of bitumen is worn away, the aggregate microtexture is
exposed and therefore skid resistance increases. Figure 3-10 demonstrates this effect, showing the measured skid
resistance during the first year of an SMA surface that continued to increase in skid resistance as the binder wore,
thus exposing the aggregate surface. This effect, the degree of change and rate of change of skid resistance will
depend upon the type of surfacing material (Jellie, 2003). For example, for well constructed chipseal surfaces, this
initial effect is not really evident. However, with stone mastic asphalt (SMA), there may be a greater time lag due to
the thicker coating of binder on the aggregate. Woodward et al. (2002; 2005) demonstrated the significance of this
initial effect on binder-rich SMA surfaces that largely dominated the whole of life skid resistance performance.
They reported on a polymer modified surfacing that, after four years of wearing, still had no exposed aggregate and
51

Chapter 3
therefore the measured skid resistance remained low. This illustrates that the trafficking and climatic conditions
were not severe enough to remove the bitumen from the surface. In this case, the polymer modified binder was too
good in its elastic and cohesive properties. The measured wet skid resistance would therefore remain low due to
the excess bitumen coating the aggregate and could pose a significant early life road safety risk and may also be
susceptible to dry skidding events.
>6 Million Equivalent
single axle loads
Initial Roughening

Equilibrium Phase

Pendulum Value

Polishing Phase

Seasonal Variations

After 1M standard Axles or

Time / ESAs
approximatley
2 years

Figure 3-9: Simplified general pavement polishing model (Prowell et al. 2003)
0.9
Mean GripNumber

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0

50

100

150

200

250

300

350

Time since laid (days)

Figure 3-10: Change in Skid Resistance with time for a section of 14mm SMA (W. D. H. Woodward et al.,
2005)
Bullas (2005) states that negative textured road surfacings may produce surface friction characteristics that
challenge the widely held belief that uncontaminated bituminous road surfacings cannot be slippery when dry . He
quotes Zipkes (1944) who postulated:
On some road surfaces, the melting point of the binder (the bituminous layer) may be
reached before that of the tread rubber, in which case the slipping co-efficient will have a
different value from that on which the rubber melts first
Bullas, (2005) describes this phenomenon as bituplaning which is the generation of low levels of surface friction
as a result of the presence of a thick bituminous layer between the tyre and the road surface melting under fully
locked wheel braking (100% slippage). Bullas gives examples where crashes that occurred in the UK and the
Netherlands on negative textured surfaces led crash investigators to believe that this phenomenon was a significant
52

and contributing factor in the crashes.

Chapter 3
Although the skid resistance values measured were no worse than wet

friction coefficients at intermediate speeds, this is a significant finding. Typical drivers are not prepared for low
skid resistance on dry roads where under extreme braking the melting of the bitumen beneath the tyres can produce
coefficients of friction similar to wet friction measurements. Due to this reason the Netherlands Road Controlling
Authorities (Bullas, 2005) now use short term signing (e.g. New road surface, longer stopping distance ) designed
to alert the road user to the specific risk of bituplaning and / or low wet skid resistance on newly constructed
negative textured surfaces.
However, with more traditional positive texture asphalt surfaces and, more importantly, for chipseal surfaces like
those more commonly used in New Zealand, the surfacing aggregate is usually not fully coated with bitumen.
Therefore, skid resistance reduces with time, under the action of traffic and environmental factors. Additionally, as
the aggregates are exposed, bituplaning will not occur unless the surfacings have a layer of binder coating the
surface of the aggregate and masking the microtexture. Cenek et. al. (2003b) state that this coating can occur either
on newly-laid asphalt surfaces, as discussed above, or as a pre-coating for chipseal surfacings, or as a result of
binder tracked from bleeding road sections during hot weather spells (i.e. a contaminant, discussed further in
Section 3.5.4). This coating is gradually worn away by traffic until the aggregate is exposed and the normal
polishing process begins. Therefore, once equilibrium polishing is attained and provided there is consistent traffic
flow, the consensus of research has been that fluctuations are the result of seasonal and/or short-term
environmental variations. These seasonal environmental effects will be discussed further in Section 3.5 and can be
summarised as a general roughening or abrasion of the surface by traffic with heavy rain and/or freeze-thaw
conditions causing higher skid resistance during winter. A moderate reduction in skid resistance is then experienced
in summer months due to polishing of the aggregates with the finer contaminants that build up on the road surface.
The term equilibrium skid resistance level is to some degree a misnomer as the skid resistance measured at any
time is a function of the traffic levels, the aggregate properties, the type of surfacing and the environmental effects
during and preceding the measurement of skid resistance. This effect was shown in Figure 3-7, which demonstrates
the result of significantly reducing traffic on a highway route due to a bypass being constructed. The bypass resulted
in a decrease in the state of polish, thereby increasing the level of measured skid resistance.

3.4.2 Road Site and Traffic Flow Conditions

A relatively lower level of measured skid resistance is encountered at sites where horizontal forces are developed
and longitudinal skid resistance profiles on otherwise uniform surfaces demonstrate this. Examples include those
featuring additional stresses, such as where vehicles are braking, accelerating or turning at bends. The measured
skid resistance can be expected to be 10% or more lower at these types of sites in comparison to straight, level and
uninterrupted traffic flow condition sites (C. K. Kennedy et al., 1990). These sites tend to occur where the demand
is higher and therefore the potential of a crash occurring is increased as the level of skid resistance is reduced.
Understanding that crash risk increased when the level of skid resistance reduced and that crash risk varied for
different types of road site, led the UK to develop standards that attempted to equalise the crash risk for the various
road sites. These standards varied the minimum level of skid resistance for each site category (Rogers & Gargett,
1991). Standards of skid resistance were described in more detail in Section 2.3.4.

53

3.5

Environmental Effects on Skid Resistance

Chapter 3

3.5.1 Introduction
The skid resistance of a road surface has been shown to depend upon many factors including the aggregate
properties and loading factors. This section discusses the environmental factors that are significant in terms of
affecting the measured value of coefficient of friction. The combined effect is often referred to as the seasonal
effect of skid resistance. As shown in Table 3-1, the known environmental factors are:

Water film thickness and drainage conditions;

Surface contamination;

Temperature; and

The seasonal and short-term rainfall variation effects.

3.5.2 Water Film Thickness

The most basic environmental factor is whether the surface is wet or dry. In dry conditions, clean surfaced roads
have high skidding resistance because the vehicle tyres can keep in good contact with the road surface. When a road
surface transitions from dry to being slightly wet, there is a sharp reduction in the coefficient of friction due to the
presence of the water film. This acts as a lubricant between the tyre and road surface and furthermore reduces the
contact area between the tyre and the road, thereby reducing skid resistance. As long ago as the 1930s, Bird and
Scott (1936) proposed an idealised representation of the loss of skid resistance with time during a rain event from a
dry surface to wet and then dry again (refer to Figure 3-11).

Figure 3-11: Skid Resistance variation during a rainfall event (Bird and Scott, 1936)
Research undertaken by Fricke, (1990) and reported by PIARC (2003) demonstrated that, depending upon the state
of the surface, asphalt surfaces reduce their measured coefficient of friction from dry to wet conditions by up to
37.5% for speeds less than 50km/h and 31% for speeds greater than 50km/h. Figure 3-12 demonstrates the range of
reductions seen for the various states of asphalt surfacing from a newly constructed clean and sharp angular surface
to a travelled surface and then a traffic-polished surface.

54

Chapter 3

% reduction from Dry to Wet

% Reduction in CoF from Dry to Wet for

Asphalt Surfaces <50km/h
45%
40%
35%
30%
25%

Max Range

20%

Min Range

15%
10%
5%
0%
New , Sharp

Travelled

Traffic polished

Excess Bitumen

State of Surface

% reduction from Dry to Wet

% Reduction in CoF from Dry to Wet for

Asphalt Surfaces >50km/h
35%
30%
25%
20%

Max Range

15%

Min Range

10%
5%
0%
New , Sharp

Travelled

Traffic polished

Excess Bitumen

State of Surface

Figure 3-12: Reductions in Coefficients of Friction due to wet surfaces -adapted from (PIARC World Road
Association, 2003)
When rain intensity exceeds 0.1mm/hr, the water film formed may have a depth varying from microns to
millimetres (Jellie, 2003). Harwood et al. (1989) reported that even as little as 0.025mm depth of water on the
pavement can reduce the tyrepavement friction by as much as 75% on surfaces having poor skid resistance
characteristics.

Thin films of water have also been shown to be sufficient to produce hydroplaning.

The micro

drainage routes provided by the surface texture roughness (macrotexture) together with the tyre tread help to
eliminate the bulk of the water, an especially important requirement as speed increases. However, the penetration of
the remaining water film can only be achieved if there are sufficient fine-scale sharp edges (microtexture) on which
high pressures can build up as the tyre passes. These high pressures are needed to break through the water film to
establish dry contact between road and tyre (Rogers & Gargett, 1991).
Allbert and Walker (1965) studied the effect of tread pattern design and road surface texture on skidding resistance
of tyres in wet conditions. They developed the concept of the contact area between the tyre and the road being
divided into three zones. Figure 3-13 illustrates the different tyre-pavement main contact zones.

55

Chapter 3
Travel
Tyre
Water Film

1.
Sinkage
Zone

2.
Transition
Zone

3.
Tractive
Zone

Pavement Surface
Figure 3-13: Model of a Wet Tyre Footprint (Jellie, 2003)
The Sinkage Zone (1), which occurs at the leading edge of the tyre where there is no contact between the tyre and
the pavement and a water wedge in front of the tyre can be created. The sinkage zone is also termed the zone of
hydrodynamic lubrication. The tyre floats on a thin film of water, the thickness of which decreases progressively as
it is squeezed out by hydrostatic pressure.
The Transition Zone (2), where the water film is locally interrupted by some aggregate protrusions allowing the
water film to be penetrated. This zone is also termed the partial or mixed hydrodynamic lubrication zone. Friction
forces can develop due to the microtexture existing on the top of the aggregate protrusions and the development of
the hysteresis effect.
The Tractive Zone (3), where water is expelled making substantially dry tyre-pavement contact area possible. This
zone is also termed the boundary layer lubrication zone. The harshness of the microtexture is required for effective
breaking of the thin water films to achieve dry contact and allowing the whole surface microtexture of the
aggregate to become active in promoting friction.
The size of these three areas, and therefore the degree of contact that can be established between the vehicle tyre and
the road surface, is determined by (Austroads, 2005b; Jellie, 2003):

Water film thickness, which depends upon pavement surface drainage capabilities and the rain intensity and
duration;

Surface macrotexture depth and the capacity of the surface to shed water, which is also a function of the
efficacy of any positive drainage provided at the location;

Tread depth, width and tyre pressure; and

Vehicle speed.

At slow vehicle speeds, the bulk water is readily dispersed, the tractive zone (3) is large and the maximum adhesive
forces can be developed. Typically the tractive zone may cover 50% of the contact area. However, as speeds
increase, the time available for expelling water from the tyre-pavement interface decreases and therefore the contact
area decreases with a consequent reduction in adhesive force. This reduction can be minimised by providing
drainage grooves or more effectively by providing high macrotexture surfaces that not only provides drainage paths
but also produce greater tyre deformations resulting in dissipation of energy through hysteresis losses.
Kokkalis (1998) determined that an increase in vehicle speed from 50 to 100km/h results in a 30 to 70% decrease in
the tractive zone (3). If this tractive zone is reduced to zero and therefore all contact with the road surface is lost,
56

Chapter 3
and there is an uninterrupted film of water under the tyre, then hydroplaning (or aquaplaning) occurs. Austroads
(2005b) defines hydroplaning as:
the build-up of hydro-pressure beneath a tyre that partially or fully exceeds the capacity of the
tyre to absorb it, and thus reduces the contact area between the tyre and the pavement.
The vehicle then slides over the water which, as a fluid, is incapable of resisting shear forces.

There are two types

of hydroplaning (Austroads, 2005b):

Viscous hydroplaning, that can occur at low speeds where the macrotexture of the road surface is low (i.e.
tending towards smooth). It requires minimal water depth to occur, as the mechanism is dependent upon the
viscosity of the water, which prevents it from escaping from under the tyre footprint. This type of
hydroplaning is most likely to occur during braking and can be exacerbated if the vehicle is running on
tyres with limited or no tread; and

Dynamic hydroplaning, that occurs once the vehicle has exceeded a critical speed (which is a function of
the tyre pressure), at which time surface water in front of the tyre (that acts as a wedge) can penetrate the
tyre footprint and reduce the surface contact area. Under total dynamic hydroplaning virtually no part of
the tyre will have any contact with the surface (i.e. the tractive zone (3) is zero) and the tyre can be fully
locked. Once hydroplaning has occurred, the vehicle speed must be reduced significantly, and to well
below the critical speed for the tractive zone to be created again and dynamic hydroplaning to cease. Even
with dynamic hydroplaning, retardation forces may be significant as the vehicle is effectively ploughing
through the surface water.

In the sinkage section, zone (1), of the contact footprint, microtexture has a negligible contribution to skid resistance
and macrotexture has a reduced contribution. In the transition section, zone (2), microtexture has a reduced
contribution but macrotexture is fully active and in the tractive section, zone (3), both microtexture and
macrotexture are fully active (Jellie, 2003).

3.5.3 Drainage Conditions

Poor road condition such as rutting and local depressions (due to pavement and / or subgrade consolidation or poor
road construction) can lead to a loss of shape of the carriageway and therefore to ponding or accumulation of water
in wet conditions. This can lead to dynamic hydroplaning at high speeds as the depth of water in front of the tyre
wedges the tyre off the road surface, thereby significantly reducing the available frictional resistance. Furthermore,
water accumulation in a single wheel track can lead to differential skid resistance between the two wheel tracks that
can increase the transverse frictional demand. If the difference between the two wheel tracks is large enough, this
may result in a vehicle losing control due to rotation.

3.5.4 Surface Contamination

Road surface contamination can interfere with the two main components of surface friction available at the tyre road surface contact area (i.e. adhesion and hysteresis effects). The contamination can originate from various
sources such as (Austroads, 2005b; PIARC World Road Association, 2003):

Water (discussed separately above), snow, frost and ice;

Dust, clay, sand, loose gravel, tracked bitumen;

Vehicle residues (e.g. oil, fuel, rubber, brake pad linings); and

Farm waste (e.g. stock effluent), fallen leaves, grass clippings, lichen and moss.
57

Chapter 3
The effects of individual contaminants on surface friction is as yet little understood. Contaminants can cover the
pavement texture and accelerate its ageing. Contaminants that build up on the road surface during dry periods (that
include any of the above factors other than water, snow, frost and ice) can also create particularly hazardous
situations for road users at the beginning of a first rain as the contaminants on the surface form a slick coating over
the pavement texture. Figure 3-14, from Bennis and De Wit (2003), that is very similar to Bird and Scott s (1936)
much earlier interpretation (Figure 3-11), illustrates how the surface friction varies over a five minute period with a
short shower after a reasonable period of no rain. The measured skid resistance significantly reduces immediately
after rainfall and then recovers to a more typical wet skid resistance. Furthermore, the diagram demonstrates that as
the rain stops and the surface dries, the measured skid resistance increases to a dry skid resistance level similar to
that measured prior to the rain event.

Figure 3-14: Variation in Skid resistance during a Rain event (Bennis & De Witt, 2003)
The presence of frost, ice or snow can also mask the microtexture and /or macrotexture of the road surface, thereby
reducing the hysteresis effect, and provide a lubricant between a vehicle tyre and the road surface, leading to a loss
of surface friction.

VTI9 (1981) identifies and describes in detail the formation of three categories of winter

condition on roads. They are:

Hoar frost (white frost);

Ice (thick, ice glaze thin ice and ground icing); and

Snow (loose, compacted, slush and sleet).

Extreme winter conditions in cold alpine or semi-alpine areas can produce any of the above conditions. The most
extreme conditions (e.g. thick ice and snow) are most often visually more apparent to the driver resulting in a more
conservative and compensating driver behaviour (e.g. the use of tyre chains or studded tyres in Northern
Hemisphere countries) to compensate for the lack of surface friction. However, the conditions that are most likely
to catch the driver unaware of the loss of skid resistance and that are common in New Zealand are the formation of
hoar frost or thin ice often termed black ice . Dravitzki et al. (2003) summarises the work of the VTI (1981) for
these two conditions:
Hoar frost (white frost) forms when the temperature of the road is both less than 0C and less than the dew point of
the air immediately above. The dew point is the temperature at which the water-holding capacity of the air is at its
limit, i.e. its relative humidity is 100%. Frost can form in three main types of weather:

Clear calm nights when radiative cooling rapidly cools the road but the air is close to 0C;

The Swedish National Road and Transport Research Institute.

58

Chapter 3
After a cold clear clear night when the rising sun causes air circulation and moist air moves over the
already cold road; or

After a cold but cloudy night, the sky clears and the road rapidly cools but the air is still moist.

Thin ice (ice glaze) is a layer of ice that coats the macrotexture of the road. The VTI (1981) describes this condition
resulting from:

Super-cooled rain landing on a road surface that is close to 0C, where it instantly freezes;

Light rain on a road surface which is already less than 0C, where it quickly freezes;

Moisture on a cooling road from earlier dew or light rain, which then freezes when the road cools to below
0C; and

Frost or light snow compacted by trafficking.

The conditions which allow the road to reach these conditions are complex and include the thermal properties of the
road surface and base layers, the prevailing day time conditions, the adjacent land type, terrain and the presence of
moisture.
In New Zealand, Dravitzki et al. (2003) formed frost and ice on samples of road surfaces in controlled laboratory
climate rooms. The surfaces were tested using the British Pendulum Tester (BPT this method is discussed in
Section 4.3.4) both before and after freezing, to determine the coefficient of friction of the surfaces and their
subsequent loss of friction due to the frost / ice conditions. The authors concluded that:

Frost first forms on chipseal on the tips of the chips, then spreads over the chips and bitumen layer as the
frost intensifies. In general this frost is only loosely adhered to the chips;

Because frost forms in this way, the effect of macrotexture in maintaining skid resistance in these
conditions appears to be small, although under weighted rotated tyres, this may be greater;

Macrotexture showed only a small effect in maintaining skid resistance for the laboratory samples when ice
layers were present as the ice tightly adhered to the chip;

Smooth surfaces appear to be very vulnerable to the effects of frost and ice, i.e. have low skid resistance,
because the ice covers the entire surface, and it cannot be easily displaced into the voids between the chips;

The loss of skid resistance is considerable, being 30% to 60% of the wet skid resistance value. This loss is
significant for road management as it makes policies providing higher skid resistant surfaces for higher risk
road sections difficult to achieve for frost or ice conditions;

The influence of microtexture was not clear. With some road surfaces, those samples that had higher skid
resistance in wet conditions (indicating a harsher microtexture) also had higher skid resistance in the frost
or ice conditions. However, the pattern was not consistent;

A small improvement in the ability of a surface to retain skid resistance was demonstrated for surfaces that
have a macrotexture greater than 1.5mm as measured in terms of Mean Profile Depth (MPD); and

The use of additional strategies was recommended for managing frost and light ice conditions for mild or
moderate New Zealand winter conditions, (other than keeping macrotexture greater than 1.5 mm) such as
the use of a de-icing salt (e.g. CMA).

Tracked bitumen from road sections that are bleeding can also cause contamination of an aggregate surfacing by
masking the microtexture. Cenek et al. (2003b) state that it is likely that different friction-generating mechanisms
apply when the aggregates are in this blackened state because of the reduced influence of microtexture. The skid
resistance characteristics will more likely depend firstly upon the shape characteristics of the aggregate, which helps
promote the hysteretic friction, and secondly on the adhesive or viscous properties of the binder. Cenek et al.
59

Chapter 3
(2003b) undertook a study to demonstrate the influence of a surface film of bitumen on British Pendulum Tester
(BPT) measurements of skid resistance. A summary of the findings of that study follows:

The skid resistance of fine textured road surfaces is more affected by the presence of bitumen film than it is
for coarse textured surfaces. This is because the relative contribution of aggregate microtexture to skid
resistance is greater for fine textured surfaces than coarse textured surfaces;

The presence of a bitumen film thickness of 100 microns on chip seal road surfaces (i.e. severe
contamination) results in a reduction in skid resistance of between 20% and 30%. However, the resulting
level of skid resistance was still significantly greater than that provided by a smooth, bitumen-only surface.
This finding suggests that chipseal surfaces in a flushed condition pose a greater safety hazard to motorists
in wet conditions than chipseal surfaces that have been blackened by tracked bitumen;

A smooth, bitumen-only, surface provides skid resistance of between 20 BPN to 35 BPN (coefficient of
friction as measured by the British Pendulum Tester refer to Section 4.3.4 for a method description) that
is comparable to the level of skid resistance provided by chipseal surfaces exposed to medium to heavy
icing (Dravitzki & Wood, 2001).

3.5.5 Temperature
Except in extreme climatic conditions, temperature does not adversely affect the skid resistance properties of the
crushed aggregate materials. However, as both tyre rubber and bituminous materials are visco-elastic materials, it
follows that these materials are sensitive to change in temperature due to the hysteretic component of surface
friction, which will subsequently affect the skid resistance of the pavement surface. Nevertheless, all common skid
resistance measurement tools employ rubber contact elements (to best simulate the in-field reality of rubber tyres on
road vehicles), which are susceptible to changes in temperature. As discussed by Jayawickrama and Thomas
(1998), it is generally believed that, as the rubber temperature increases, the hysteresis losses in the rubber become
less and the friction correspondingly reduces. There is some disagreement regarding the size of this effect over
normal operating conditions. An Australian study by Oliver (1980) found that surfaces with high initial levels of
skid resistance experienced the greatest decrease in the measured skid resistance with temperature. Furthermore,
Oliver (1980) found that correction factors that had been developed in more temperate climates (such as in the UK)
may not cover the range of normal operating temperatures experienced in more tropical and / or hot climates and
therefore the correction factor applied to the raw test data may not be appropriate.
A summary of the effect of temperature on the measured coefficient of surface friction primarily from Oliver (1980),
Hill and Henry (1981) and Hosking (1992) is as follows:

The measured coefficient of friction tends to decrease with increasing air temperature;

Temperature change has more effect on the frictional properties of the tyre, leading to an indirect effect on
skid resistance as measured by testers and that is available to road users;

Tyre temperature tends to be proportional to air and pavement temperature, with higher tyre temperatures
leading to decreased measured coefficients of friction;

Water temperature has negligible effect on measured coefficients of friction; and

Increased pavement temperatures lead to reduced coefficients of friction.

Some skid resistance measurement tools are potentially more susceptible to temperature changes than others and this
has led to the derivation of temperature correction factors for most pieces of skid testing equipment (although not
all) used to measure the coefficient of friction. Therefore, care needs to be taken when comparing reported
coefficients of friction obtained from various devices as some may be corrected for temperature effects and others
60

Chapter 3
may not (refer to Section 4.3 for a method description of the commonly used skid resistance measurement devices
used).

3.5.6 Seasonal Variations

The level of measured skidding resistance of a road surfacing has been shown to vary throughout the year (after the
initial phase of traffic polishing and wearing has been completed, as shown earlier in Figure 3-9. This variation has
been shown to be highest in the winter and lowest in the summer for most northern hemisphere climates. This effect
has been attributed to the combined effect of traffic and weather on the surface aggregate. When roads are dry, the
polishing effect of traffic tends to dominate, but when they are wet for prolonged periods, they recover some of their
former harshness (Rogers & Gargett, 1991). The scale of the seasonal and traffic effects depends largely upon the
geological history and petrography of the aggregates. Hosking (1976) reported this seasonal effect in a study
undertaken for the UK Highways Agency where the coefficient of friction was measured by the SCRIM device at a
group of highway sites (that had polished to an equilibrium or long term skid resistance level), monthly over an 11
year period from 1958 to 1968. The results of this study are shown in Figure 3-15 and clearly show an annual
cyclical effect with a significant seasonal variation.

Figure 3-15: Measured Coefficient of Friction (SCRIM) in the United Kingdom over 11 years (J.R. Hosking,
1976)
Other researchers in the UK (J.R. Hosking, 1976; R. Hosking, 1992; Rogers & Gargett, 1991; G F Salt, 1977) and in
the US (S.H. Dahir & Henry, 1979; Henry & Saito, 1983; Jayawickrama & Thomas, 1998; Rice, 1977) also indicate
a seasonal variation that is approximately sinusoidal with seasons of the year. This variation has been found to be as
high as 0.15 to 0.20 SFC (Sideway Force Coefficient as measured by the SCRIM device refer Section 4.3.2 for a
description of this method) between winter and summer months.
In New Zealand, Cenek et al. (1999) reported on a study of the long term variation of measured skid resistance
undertaken in Northland over a ten-year period (1988 to 1998) on a series of seven newly surfaced chipseal surfaces
south of Whangarei. The chipseal surfaces were constructed in 1988 on a geometrically straight and flat grade using
different sources of greywacke aggregate with similar PSVs. The skid resistance measurements were undertaken
with a British Pendulum Tester (BPT) approximately every three months. The results of this study are shown in
Figure 3-16. The study results clearly demonstrate the three stages of variation in skid resistance as suggested by
61

Chapter 3
Prowell et al. (2003) and as shown in Figure 3-9. The three phases are described for the NZ research study, as
follows:
Phase 1: Initial Roughening phase. Initial values of measured coefficient of friction are low, most likely due to
bitumen coating the aggregate and/or the stone microtexture being roughened by trafficking. The bitumen is
subsequently worn away which exposes and roughens the natural aggregate surface, thereby providing increasing
surface friction. This initial roughening phase lasted approximately one year.
Phase 2: Polishing phase. The vehicle trafficking polishes the microtexture of the aggregate reducing the harshness
of stone texture, thereby reducing the measured coefficient of friction. The general shape of this polishing (after
smoothing out yearly cyclical seasonal variations) follows a negative exponential shape and took some four to five
years to complete.
Phase 3: Equilibrium phase. The polishing phase has reached a stable or near equilibrium phase for the level of
traffic action and the variation of measured coefficient of friction follows a roughly annual cyclical seasonal pattern
with low skid resistance in the summer months and high skid resistance over winter months.

110

AVERAGE SKID RESISTANCE (BPN)

100

90

80

70

60
60

50

40
1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

DATE OF TEST

Figure 3-16: Long term skid resistance monitoring in Whangarei, NZ using British Pendulum Tester (P. D.
Cenek et al., 1999)
The third seasonal variation phase (the equilibrium phase ) has been idealised in the UK and forms the basis for
measuring skid resistance surveys in the summer months in the UK and NZ. Network surveys are adjusted for a
mean summer coefficient of friction, to determine the lowest level of measured skid resistance in any one year
(being the worst condition for the road user). This variation in wet skidding resistance in the UK (northern
hemisphere) throughout a year is shown in Figure 3-17 for a mean summer value of 0.50 SFC as measured by
SCRIM and is characterised by a clear sinusoidal pattern. The degree of variation of this effect (0.15 to 0.20 SFC)
demonstrates the degree of unpredictability that could be encountered unless some form of correction factor is
applied.

62

Chapter 3

Figure 3-17: Estimate of the seasonal variation of SFC with the reported Mean Summer SCRIM Coefficient
for the UK (Rogers and Gargett, 1991).
For this reason, controlling authorities such as the UK Highways Agency (2004) and Transit NZ (2002) have put in
place a methodology to determine network and project level skid resistance performance based on the Mean
Summer SCRIM Coefficient (MSSC). The MSSC is based upon collecting at least three samples across the summer
period for each region on carefully chosen seasonal skid resistance sites , being sections that should be in an
equilibrium polished state . The mean of the three measurements is then used to factor the network surveys up or
down based upon the network measurement on the seasonal sites. This is then expected to give the worst case skid
resistance values over the summer period. The NZ policy has recently been extended (Transit New Zealand, 2002)
to include a term called Equilibrium Skid Coefficient (ESC) which factors yearly network surveys by a rolling four
year average of the seasonal sites . The procedure is meant to take account of annual fluctuations in climatic
patterns by region, where some summers are much wetter or dryer than others.
Whilst the shape in Figure 3-17 is idealised and needs to be shifted by six months for the southern hemisphere, it has
been recognised that the minimum values and the shape of the sine curve of this seasonal variation, vary from year
to year and at different times of the year, depending upon the predominant weather patterns (R. Hosking, 1992).
These differences are also evident in the NZ study that demonstrates that whilst most summers from 1992 onwards
(refer to Figure 3-16) reached a similar minimum summer value, one summer (1996) did not reach the same level of
polishing. Furthermore, a comparison of the highest winter levels of skid resistance from year to year demonstrate
that the winter prior to the 1996 higher summer values, were higher also. This could perhaps suggest that the lower
summer values in 1996 were linked to weather patterns in the winter period that caused greater rejuvenation of skid
resistance some six months prior to the summer values being recorded. A close examination shows the amplitude of
the sine wave length for the year of 1996 is very similar to previous years it just moved slightly upwards. Similar
examples can also be seen in the UK study (refer to Figure 3-15) between winter in 1962/1963 and summer in 1963.
As discussed by Jayawickrama and Thomas (1998), there is general consensus among researchers regarding the
following mechanism causing seasonal variation:
Prolonged periods of dry weather in the summer allow the accumulation of fine particles that
assist in polishing of the pavement surface. The combination of polishing and particle
accumulation, together with the contamination from vehicles such as oil drippings and grease,
results in a loss of microtexture and macrotexture during the summer months. In winter, the
aggregate surface is rejuvenated with chemical reactions from the rainwater exposing new
particles. The increased rain flushes out the finer particles responsible for polishing and other
63

debris increasing macrotexture.

The coarser aggregate surface and the increased

Chapter 3

macrotexture in turn lead to an increased skid resistance of the pavement.

Some researchers also suggest that the water film covering the pavement for longer periods in winter acts as a
lubricant and reduces the polishing effect of vehicles on the surface aggregate.
However, not all climates or regions have been shown to act in this same relatively predictable and cyclical pattern
after an equilibrium level of polishing has been attained.

In Australia, Oliver et al. (1988) indicated different

seasonal patterns and variations. Regular testing (almost weekly) was carried out on a group of sites with a British
Pendulum Tester (BPT) in the capital cities of each of the six States in Australia and examples of the results are
shown in Figure 3-18. The measured BPT coefficient of friction values have been normalised to 20C to eliminate
seasonal temperature effects and the resultant values shown represent the mean of six sites (three sprayed seal and
three asphalt surfaces) in each State of Australia. The results show the seasonal variation is not only substantial and
varies from State to State, but is largely unpredictable. The variation for Queensland (a sub tropical climate) is quite
unique in that it shows very little variation in amplitude and no real dependency on seasonal variation in comparison
to Victoria and Western Australia.

Figure 3-18: Seasonal Variation of skid Resistance in Western Australia, Victoria and Queensland (J.W.H.
Oliver et al., 1988).

3.5.7 Short-Term Variations

Bird and Scott (1936) took measurements on the UK Transport and Road Research Laboratory (TRRL) test track in
Crowthorne, UK using a side-ways force method tester (refer section 4.3.2 for test methods). They found that skid
resistance levels varied significantly (as shown in Figure 3-11) with changing weather conditions from prior to a
rainfall event with an immediate drop in skid resistance as rain fell and then increasing as contaminants were
washed away and then subsequently as the pavement dried out.
Hill and Henry (1981) discuss the results of a three-year research program that undertook regular skid resistance
measurements with the locked wheel ASTM Test Method E 274 (refer section 4.3.3 for test method) on public
highways in Pennsylvania and other US States that demonstrated that there were both long-term cyclical seasonal (a
primary effect) and short-term skid resistance variations (a secondary effect). They recognised that the combination
of the two variations makes the establishment of a surfacing maintenance programme a difficult task. Henry (2000)
64

Chapter 3
discusses the secondary effect that is not limited to typical annual seasonal variations such as seen in northern
climates, but is a short-term variation caused by the combined effect of rainfall and contaminants on the road
surface. During dry periods dust and oil accumulate on the pavement surfacing. When a skid resistance test is made
during the dry period, the water that is applied mixes with dust and oil, which reduces the measured friction. When
the measurements are made shortly after periods of rain, the pavements are less contaminated and this effect is
reduced.
Hill and Henry (1981) state that the variations from day to day (short term), seemingly due to rainfall pattern and
local weather conditions, are superimposed on the seasonal annual cycle. Frequent tests during the summer periods
indicated that pavement skid resistance may vary as much as 25 percent during a single week and they and others
(Hegmon, 1978) concluded that these variations were real skid resistance changes related to changing conditions.
Dahir and Henry (1979) as part of the same research reported this effect with regular testing only days apart for a
period of two summer months that resulted in low skid numbers (coefficient of friction as measured by the ASTM E
274 Locked Wheel tester) after a dry period and rejuvenated high skid numbers after a rainy period (refer to Figure
3-19).
Rejuvenation
Rejuvenation

Rejuvenation
Rejuvenation

Figure 3-19: Influence of Rainfall on skid resistance - (Haydon, 2005) reproduced from (Hill & Henry, 1981)
The effect of rainfall on skid resistance in terms of dry skid resistance to wet skid resistance has been discussed
previously. However, there is a further combined effect of contaminants and periods of prolonged dry periods that
also reduces measured wet skid resistance. Cenek et al. (2003a; 2003b) describe this effect that occurs during
periods of prolonged dry weather where a fine film of contamination builds up on road surfaces, especially heavily
trafficked highways. The contamination consists of a mixture of rubber particles, road aggregate and bitumen
particles, soil and dust and other detritus. The contamination is washed off by regular rain. However, after
prolonged dry periods, a shower or rainfall can create a very slippery road surface (often described as summer ice ).
This effect is commonly known by vehicle drivers who describe the road as being most slippery when it first gets
wet. If the rainfall continues some washing and cleaning occurs that increases surface friction.
This short-term influence of pollutants at the beginning of precipitation causing lower measured wet skid resistance
that then recovers to a cleaner wet surface friction, was also confirmed by Bennis and De Wit (2003) and is shown
in Figure 3-14. The short-term pollutants effect was measured by Hill and Henry (1981) as a reduction in measured
wet skid resistance (as measured by the E274 Locked Wheel Tester) over a period of ten days and is shown in
65

Chapter 3
Figure 3-20 for one site in Pennsylvania, USA. Hill and Henry (1981) did not quantify the nature of the pollutants.
Nevertheless, they reported similar relationships for a number of sites and generally the decrease in measured skid
resistance reached a maximum after approximately seven days of no rainfall. It was observed that this lower value
of measured skid resistance then remained at that low level until the next significant rain.

Figure 3-20: Effects of pollutants and days since last rain on measured skid resistance (Hill & Henry, 1981)
Hill and Henry (1981) assumed that the long-term variations in skid resistance (after an equilibrium polishing level
is reached) are a function of pavement aggregate properties and traffic density, whereas the short term residuals are
a result of:

Rainfall effects;

Temperature effects (discussed previously); and

Errors in the measurement of skid resistance by the testing equipment or vehicle.

Hill and Henry (1981) reported upon the work of Dahir et al. (1979) who attempted to predict the measured skid
resistance from rainfall records and suggested using a Weighted Rain Function (WRF) over the five days prior to a
skid test as a measure of the effect of rain and pollutants on skid resistance.

WRF =

5
i =1

where:

Ri
i

Eq 3-1

WRF = Weighted Rain Function

Ri = Rainfall in mm on the ith day prior to the test
i = number of days prior to the test ranging from 0 to 5 days

Dickinson (1989) in a study in Australia also suggested that it is the proportion of rainy days during the preceeding
two weeks that has the most effect, rather than the number of days since the rainfall. However, Hill and Henry
(1981) found when using the WRF function with data collected from a number of Pennsylvania sites that the
correlation coefficients (r) were consistently low (0.08 to 0.13) and concluded that if a correlation exists between
precipitation and skid resistance, a parameter other than WRF must be used. They subsequently proposed the
following dry-spell factor (DSF) in place of the WRF function:

DSF = ln(t R + 1)

Eq 3-2
66

Chapter 3
where:

DSF = Dry Spell Factor

tR = the number of days since the last rainfall of 2.5mm or more (up to a maximum of 7
days, hence 0<tR<7)

The DSF rainfall relationship was then tested on the same data as used to test the WRF and significant
improvements were obtained for the correlation coefficients (r) over the five sites ranging from 0.16 to 0.56 with an
average value of 0.38. Hill and Henry (1981) concluded that although (r) is still small, the length of time since the
last significant rainfall (DSF) is a better measure of the effect of rainfall on the skid resistance of pavements when
compared to a weighted rain function (WRF).
Cenek et al. (1999) reported that although the occurrence of rainfall is generally accepted as the reason for short
variations in skid resistance, the mechanism by which the variations are produced is not yet sufficiently well
understood to permit reliable modelling.

This view regarding rainfall as a contributory cause of short-term

variations in skid resistance is not universally agreed in the research literature. Research undertaken by the
Australian Road Research Board reported by Oliver et al. (1988) indicated that contamination of surfaces by vehicle
oil droppings and tyre detritus could be eliminated as a cause of the variation observed in skid resistance.
Chapter 4 will review and discuss the methods, devices and models that have been developed to enable the better
management of road surfacings and attempts to predict requirements and or changes for future surfacings.

67

Chapter 4

4 SKID RESISTANCE TESTING METHODS AND MODELS

4.1

Introduction

This chapter reviews existing methods and the subsequent equipment devices for measuring the skid resistance
(including the microtexture and macrotexture) of roads and airports that are commonly used in New Zealand and
internationally. Furthermore, a review of attempts to harmonise the various measuring devices and subsequently the
models that have been developed to describe the behaviour of skid resistance is discussed. The factors have been
grouped under the following sections:

Testing device factors;

Measuring skid resistance (microtexture);

Laboratory based skid resistance prediction methods (microtexture);

Measuring macrotexture;

Models that have been developed to describe the behaviour of skid resistance; and

Harmonisation of various testing equipment.

4.2

Testing Device Factors

4.2.1 Introduction
As shown in Chapter 3, Table 3-1, there are many factors affecting the measured surface friction of a road surfacing.
The factors that are related to a vehicle being used by a road user or for the purposes of measuring the surface
friction of a surfacing are the same and can be summarised under the following categories:

Vehicle speed;

The tyre angle to the direction of the moving vehicle;

The wheel slip ratio;

Tyre characteristics; and

Tyre contact area and stress.

4.2.2 Vehicle Speed

As has been discussed in Chapter 3, Section 3.2, the removal of water between the tyre and the road is a timedependent process and therefore the area of dry contact decreases as the vehicle speed increases, with a consequent
decrease in the effective skid resistance. The rate at which the skid resistance reduces with increasing speed is a
function primarily of the macrotexture of the surfacing due to the hysteretic component of friction but also the type
of surfacing (e.g. chips seal, asphalt mix or concrete).

It is most influenced at speeds above 50km/h when

macrotexture is lacking (refer to Figure 3-5). It is therefore important to consider speed when measuring skid
resistance, as the loads on the testing tyre vary for differing types of testers (affecting the hysteresis component).
Furthermore, some surfaces are more sensitive to differences between the actual testing speed in comparison to the
target speed of the surveys. Most road pavement surveys in rural areas in NZ, Australia and the UK are undertaken
at 50km/h. In Europe and the USA, it is common practice to have these surveys undertaken at 60km/h and 64km/h
respectively. There are, as yet, no set standards for urban areas, although 30 or 40 km/h are common testing speeds
depending upon the level of congestion encountered. Some skid resistance testing methods (e.g. SCRIM) modify
their raw measured data by factors to correct the measured value to the target speed.

However, these correction

68

Chapter 4
factors are generalised and do not take into account the change in macrotexture of the various surface sections upon
which the speed relationship depends.

4.2.3 Tyre Angle

The measured surface friction or the friction available to a road user is the resultant friction of the tangential friction
component and longitudinal friction squared (as shown earlier in Section 2.5.4 and Eq 2-5). It therefore follows that
if the tyre angle is at a different angle from that longitudinally in line with the vehicle (i.e. braking whilst negotiating
a corner, or a tyre is set at a different angle) then the measured surface friction will be different from that measured
in line.

Most skid testing equipment currently available have their measuring tyre set in line with the vehicle.

However, the earliest surface friction testing method, the Sideways Force Coefficient method (as measured by the
SCRIM vehicle) is set at an angle of 20 to the line of the vehicle. This method is the most commonly used testing
method for road networks in the UK, Australia and NZ (refer to Section 4.3.2 for a review of the SCRIM testing
device).

4.2.4 Wheel Slip Ratio

As discussed in Section 2.5.3 and shown in Figure 2-15, the surface friction is not constant through a braking cycle.
It varies as a function of the wheel slip speed (defined as the difference between the angular speed of a wheel and
the vehicle speed) during a braking manoeuvre. A free-rolling wheel has 0% slip as the wheel speed is the same as
the vehicle speed, whereas at the other extreme a locked wheel is operating at 100% slip as it has no angular velocity
(tyre slides along the road surface).

The measured coefficient of friction value, as measured by a device, varies

according to the percent slip. The measured coefficient of friction increases rapidly with percent slip to a peak value
that typically occurs between 10% and 20% slip.

The shape of the curve prior to the peak slip value is

predominantly a function of the characteristics and properties of the testing tyre. The coefficient of friction then
slowly decreases from the peak to reach, at 100%, a value called the coefficient of sliding friction. This second
reducing friction phase is predominantly influenced by the road surface type and texture. Antilock brake systems
(ABS) on road vehicles prevent the vehicle wheels from locking and are designed to keep and try to keep the
coefficient of friction close to the peak value, which significantly reduces the braking distances. Without ABS
brakes, it is generally assumed that the driver locks the vehicle wheels during a panic brake that has the result of
increasing braking distances.

4.2.5 Tyre Characteristics

Veith (1983) states that a meaningful assessment of tyre wet traction performance requires an understanding of the
tyre use environment. The important factors are tyres, rainfall conditions, road surfaces, the vehicles and vehicle
operation. This section will discuss the development of modern tyres and their characteristics in terms of skid
resistance properties.

Development of Modern Tyres

As highway engineers have not been directly involved in the design or manufacture of vehicle tyres, they have
tended to regard skid resistance as only a property of the road surface. However, as Jellie (2003) points out, the
study of skid resistance should include a discussion on the contribution the tyre makes to the frictional interaction of
the tyre and the pavement.

69

Chapter 4
The development of the modern pneumatic tyre followed from John Boyd Dunlop s patent of the first cost effective
pneumatic tyre in 1888 (Bells, 2006). The modern pneumatic tyre allowed faster and more comfortable travel by
providing better shock absorption, load transfer and traction. Modern tyres are made from the following materials
(Rubber Manufacturers Association, 2006):

Typically a combination of several types of rubber;

Carbon black fillers;

Pigments, antioxidants, silica and other additives;

Steel and polyester.

A summary of the main points follows.

Tyre Construction Effects

All modern tyres are constructed in a series of layered materials (Rubber Manufacturers Association, 2006) and
include:

Tread, which provides traction especially in wet conditions;

Belts which are layers of steel cord wires that stabilise and strengthen the tread;

Sidewalls, which protects the side of the tyre from road and kerb damage, and also transfers load especially
at low inflation pressures;

Body ply that gives the tyre strength and flexibility;

Bead that assures an air-tight fit with the wheel; and

Inner liner that maintains inflation pressure.

There are three main types of modern tyre construction methods as shown in Figure 4-1 and summarised below:

Bias tyres, also known as conventional tyres. The carcass or body plies are laid at angles between 30 and
60 to the centreline or direction of rotation of the tyre. Succeeding plies are laid opposite to each other
with cords running diagonally (cross ply) to provide balanced strength;

Belted-bias tyres have a body similar to that of bias tyres, plus two or more steel belts under the tread to
strengthen and stabilise the tread. The belts improve tread by reducing tread movement during contact with
the road;

Radial tyres where body cords run across the tyre perpendicular to the beads (90 to the centreline or
direction of rotation of the tyre). Radial tyres have steel belt plies that run circumferentially around the tyre
under the tread to stabilise the tread and strengthen the tread area. The belts restrict tread movement during
contact with the road surface thereby improving tread life.

70

Chapter 4

Figure 4-1: Bias and Radial Tyre Construction (The Wheel and Tyre Bible, 2006)
Speyer and Kelley (1977) as reported in Veith (1983), established that on both high and low skid resistance surfaces,
that there is no significant difference between radial, belted-bias, and bias tyres in peak friction (at approximately
15-20% slip speed) or sliding (100% slip speed with fully locked wheels) in wet braking conditions at 64km/h and
96 km/h test speeds. Importantly, these tests were undertaken with equivalent tread patterns and with the same tread
compound and therefore any differences could only be attributed to the generic tyre construction type.
Most modern tyres are of radial construction, as radial tyres wear better than cross ply tyres and have greater rigidity
for cornering (Tyre Talk, 2006).

Tyre Tread Design

Analysis by Segel (1977) of the tyre-road-rainfall-vehicle system indicates that the probability of skidding is
greatest at high velocities, high water film thickness, low micro / macrotexture and for abrupt manoeuvres. Two
important features of tyre design which help to prevent the loss of traction in wet conditions are tread pattern and
tread compound (Veith, 1983). These and other features are shown in Figure 4-2. Other factors include inflation
pressure and vertical load.

Figure 4-2: Elements of tread design (Tyre Talk, 2006)

71

Chapter 4

The functions of the tread elements are briefly described in Table 4-1.
Tyre Component function

Picture

Sipes:
Small, slit-like grooves in tread blocks that allow the blocks to move more. This
added flexibility increases traction by creating an additional biting edge. Sipes are
especially helpful on ice, light snow, and loose dirt.

Blocks:
Those segments making up a tyre'
s tread. The primary function of tread blocks is to
provide traction.

Ribs:
The straight-lined row of blocks that create a circumferential contact "band" to
assist manoeuvring.
Dimples:
Indentations in the tread that improves cooling.
Shoulders:
Provide continuous contact with the road while manoeuvring. Shoulders wrap
slightly over the inner and outer sidewall of a tyre.

Void Ratio:
The amount of open space in the tread. A low void ratio means more rubber in
contact with the road. A high void ratio increases the ability to drain water.
Whether a tyre has a high or low void ratio depends on the tyre'
s intended use.
Grooving:
Used to create voids for better water channelling on wet road surfaces. It is the
most efficient means of channelling water from in front of to behind the tyre. By
designing grooves circumferentially, water has less distance to be channelled.
Circumferential grooves provide the shortest distance from the front to the rear
edges of the contact patch.
Table 4-1: Functions of Tyre tread elements (Tyre Talk, 2006)

Tyre Tread Pattern

The tyre tread patterns are designed to suit the vehicle (e.g. commercial or passenger car) and the running surface
(e.g. on road or off road, asphalt mix or snow) on which they will be operating. The tread pattern design (that
72

Chapter 4
includes rib, lug, rib-lug, block shape, asymmetric and directional pattern shapes) is a balance of performance
properties that include:

Traction and braking power;

Directional steering control;

Rolling resistance;

Low heat generation at high speeds;

Water shedding properties;

Noise generation; and

Low wear properties.

As shown in Figure 4-2 and described in Table 4-1, most tread patterns are based on circumferential ribs separated
by grooves which must be wide enough to accommodate the displaced water without developing a hydrostatic
pressure between the rib surface and the road surface. It is generally accepted that the rib width should be as small
as practical and the adjacent groove preferably straight so as not to impede the flow of water (Veith, 1983).
The fractional groove volume ( V) concept was developed by Veith (1983) to relate the effects of the number of
ribs, grooves and tread depth. It is defined as the fraction of the total volume occupied by the grooves referenced to
the volume encompassed by the developed or in-road-contact tread width (dTW) and an arbitrary tread reference
depth dimension (ho) which is 10mm. It can also be calculated in practice by measuring the fraction groove area
( A) of a footprint impression and the tread depth (hG) in millimetres:

V = A
Where:

hG
10

Eq 4-1

= fractional groove volume

= fraction groove area

hG = tread depth (mm)

Veith (1983) states that tyre designs that will perform well under critical traction (friction) conditions require open
tread patterns where

is in the range of 0.30 to 0.35. The importance of tread pattern groove volume in terms of

measured coefficient of friction is illustrated by Veith (1983) and is shown in Figure 4-3. He compared the
performance of two selected tyres, both nominally identical, with the exception of the groove void volume (Tyre A
with

= 0.18 and Tyre B with

= 0.41), on two surfaces (Surface S with good macrotexture and good

microtexture and Surface H with low macrotexture and fairly low microtexture) at various speeds (V) and tyre slip
angles ( ) in degrees (i.e. braking and cornering). The superiority of Tyre B with the higher groove void volume at
either 32km/h or 96 km/h is clearly evident, but even more evident on the poor surface friction surface Surface H.
This surface demonstrates a reducing cornering friction coefficient ( c) at high speeds (96 km/h) and greater than 4
degrees slip angle ( ) for Tyre A whereas Tyre B is still rising.

73

Chapter 4

(a) Good macrotexture and microtexture.

(b) Low macrotexture and microtexture.

Figure 4-3: A Comparison of Tread Pattern Groove Volume (

) Performance on two surfaces (Veith, 1983)

Tyre Tread Compound

The influence of the tread compound on wet traction (friction) can be assessed in two ways (Veith, 1983):

On the influence of compound modulus or hardness and hysteresis properties; and

On the basis of the nature of the materials that constitute the tread.

Veith (1983) found that the hardness of the tyre rubber influenced the traction (friction) performance. If there was
very little or no lubrication by water (e.g. in the Tractive Zone 3 of Figure 3-13), the softer rubbers yield higher
friction because the actual tyre contact area with the surface increases. However, at high speeds and thick water
films (e.g. in the Sinkage Zone 1 of Figure 3-13), when hydrodynamic lubrication conditions predominate, higher
rubber hardness produces increased traction (friction) levels. In this case, the harder rubber resists the elastic
deformation and this yields improved performance in comparison to the softer rubbers. Figure 4-4 demonstrates the
effects of warmed-up rubber compound hardness on wet cornering friction coefficient ( c).

The tests were

undertaken at a speed of 80km/h measured on smooth pebbles with external water application.

Figure 4-4: Wet Friction versus Durometer Hardness for several Tread Rubbers (Veith, 1983)
74

Chapter 4
Veith (1983) concludes from his results that the types of rubber used in tyre production fall into three essential
groups (refer to Figure 4-4):

Butyl or isobutene-isoprene rubber (IIR);

Styrene-butadiene rubber (SBR) with natural rubber (NR); and

Cis-polybutadiene rubber (BR).

Each of these three groups has a different but direct dependence of traction on hardness and therefore one cannot
generalise and use hardness alone as a tread parameter for wet traction (friction) performance. For example, a tread
of 55-durometer hardness can have varying traction performance depending upon which rubber is used. Each
rubber has a characteristic performance level which is a function of its chemical structure and properties (Veith,
1983).
Veith (1983) found that the hysteresis or its inverse, resilience, also affects wet traction performance. As tread
elements are deformed in the contact, mechanical energy is converted into heat by hysteresis mechanisms.
Hysteresis may be inversely measured by a rebound or resilience test. Veith (1983) summarises the research of
Peterson et al. (1974) that reported that SBR, NR, and BR rubber in terms of wet traction performance is directly
correlated with the glass transition temperature10 (Tg) with one marked exception, butyl rubber. In general the
hysteresis of a rubber follows the Tg value. Veith (1983) further analysed the work of Peterson et al. (1974)
demonstrating that the performance of butyl and other rubbers may be brought together by the use of the Tg/R ratio
where R is the percent rebound under standard 0 C conditions. The Tg/R ratio may be thought of as the product of
Tg and hysteresis. Figure 4-5 shows a plot of traction rating of Peterson et al (1974) for two sets of tyre rubbers in
terms of the Tg /R ratio as calculated by Veith (1983) for an intermediate speed (approximately 60km/h).

Figure 4-5: Wet Traction rating versus Tg to R ratio (Veith, 1983)

The data indicate that both the hysteresis and chemical makeup of the rubber and therefore the friction properties, as
represented by Tg, are correctly indicated (average correlation coefficient of 0.96) by the Tg/R ratio.

Furthermore,

the measured wet friction is dependent upon the rubber compound (chemical makeup) and the speed of the vehicle
(refer to Figure 4-6). For example, butyl is superior to SBR only at low-speeds and becomes inferior at high speeds.
Veith (1983) explains this reversal is due to the lower hardness of the buytl tread. At high speeds and in Zone 1 of
Figure 3-13, elastohydrodynamic lubrication (EHL) forms, which is a special case of hydrodynamic lubrication,

75

Chapter 4
where an accumulation of displaced water results in rubber tyre indentation and an upward pressure that lifts the tyre
off the surface resulting in a drastic reduction in friction forces. The EHL term derives its name from the elastic
indentation of the rubber tyre in the frictional contact region. At high speeds, EHL increases, the boundary layer
adhesion that occurs in Zone 3 of Figure 3-13 is reduced, and the nominally high hysteresis component of frictional
energy loss for butyl rubber is reduced, due mainly to increased tyre temperature at higher speeds.

Figure 4-6: Wet Traction versus Speed for several Tread Rubbers (Veith, 1983)
In summary, the amount of friction (traction) loss will depend upon the balance between hysteresis and hardness;
high hardness together with high hysteresis will reduce the friction loss at high speeds and in wet conditions. The
influence of hardness, hysteresis and Tg will also depend upon the percentage of wheel slip of the tyre. Peak friction
coefficients during braking traction and cornering traction occur with a rolling tyre (10%-20% wheel slip as
explained in the previous Section 4.2.4) whereas slide braking friction (100% wheel slip) is developed with a locked
or non-rotating tyre. The distinction between these two modes, rolling versus locked, is important in regard to the
amplitude and rate of tread element motions that are much larger for a locked wheel tyre.
Veith (1983) indicates that the probability of skidding is greatest at high velocities, in deep water conditions, at low
pavement micro-macro texture levels, and for abrupt maneouvers. Under these critical conditions, the demand
friction is quite likely to be greater than the available friction and loss of vehicle control is the probable outcome.
Veith (1983) states that under low traction (friction) demand, all tyres perform in a relatively satisfactory manner.
As the friction demand rises, substantial differences may develop. Poor performing tyres lose their control abilities
most quickly on smooth, slippery low-texture pavements, at high speeds, and in deep water (i.e. surfaces that are
wet, have low macrotexture and microtexture levels and are contaminated with detritus).
Veith (1983) concludes that tyre designs that will perform well under critical conditions require:

Open tread patterns (

Selecting tread compounds with reasonably high hardness and high hysteresis (although these will need to

in the 0.30 to 0.35 range);

be balanced against other performance criteria such as endurance, tread wear, and rolling resistance); and
10

The glass transition (Tg) is defined as the temperature below which molecules have little relative mobility, the
point at which the transition manifests itself as a tremendous change in viscosity over a small temperature
range.
76

Chapter 4
Any choice of radial, belted-bias, or bias constructions as these make little difference.

Tyre Tread Wear Irregularity

It is well known that the degree of tread wear can significantly reduce the inherent friction properties of tyres.
However, Williams and Evans (1983) indicate that the type of wear regularity can govern the rate of loss of wet
friction with increasing speed or water depth. They showed that vehicle stability under wet conditions, which is a
factor in vehicle crashes, depends upon the state and type of tyre wear with tread depths greater than the generally
accepted legal limit likely to cause stability problems. For example, the interchanging of worn tyres front to rear or
the fitting of new front tyres with shoulder-worn tyres fitted to the rear can be particularly dangerous under wet
conditions even at moderate speeds.
In a study into the cause of crashes by Fox et al. (1979) the tread depth of tyres involved in crashes was measured.
The results indicated that the relative risk of a crash decreases and remains unchanged if the remaining tread pattern
depth is at least 3mm, but below 3mm the risk rises exponentially. However, most countries including New
Zealand, have a minimum legal Warrant of Vehicle Fitness tread depth of 1.5mm.

Developments in Tyre Design

Tyres Online (2005) state a major problem facing tyre designers has traditionally been solving the compromise
between low rolling resistance and wet surface friction. The rolling resistance is the amount of energy a tyre
absorbs as it revolves and deflects. The lower the rolling resistance, the less fuel is required to propel the vehicle
forward. Lowering the rolling resistance, however, has traditionally meant a reduction in wet surface friction
performance, which for road safety reasons was unacceptable.
The Michelin Energy Tyre (Tyres-Online, 2005) solved this problem in 1992 by the replacement of certain
quantities of carbon black with silica in the tyre'
s tread compound which has enabled manufacturers to produce tyres
which provide improved wet skid properties, better winter performance and lower rolling resistance all at the same
time.

As explained above, friction (traction) is affected by the degree to which a tyre is distorted at high

frequencies, that is, the degree to which it hits small stones and unevenness in the road surface. Surface friction is
also best served by rubber compounds which absorb high levels of energy (high hysteresis compounds).
Rolling resistance, on the other hand, is affected by low frequency distortion, the deflection of the tyre as it revolves.
Rolling resistance requires compounds which absorb low quantities of energy (low hysteresis compounds). This
contrast explains why it has been impossible in the past to provide tyres which both reduce rolling resistance and
increase wet surface friction.
With the addition of silica, however, tyre manufacturing engineers have been able to produce compounds which
have high hysteresis at high frequencies but low hysteresis at low frequencies, thereby reducing vehicle operating
costs whilst retaining adequate levels of skid resistance performance. Michelin (Tyres-Online, 2005) claim that the
use of silica can result in a reduction in rolling resistance of 20%, which equates to a 5% fuel saving when assuming
correct tyre pressures are maintained and making allowances for varying speeds and different driving characteristics.
The use of silica can also improve wet skid performance, by up to 15% according to Vredestein (Tyres-Online,
2005), thereby substantially improving braking distances. Silica also provides substantial benefits in winter tyres
and all-season tyres. Compounds using silica are more elastic and flexible at lower temperatures, allowing better
grip and braking during wintry weather.

77

Chapter 4
However, from the highway engineer s perspective, the use of silica in tyres is likely to cause accelerated wear and
polishing of the road surface depending upon the particle size distribution. Silica is an abrasive substance (Mohs
hardness scale of 7) and it is used as part of the PSV test as a substance to accelerate the polishing of the aggregate
samples and to simulate the effect of detritus on the road surface. It is also used in the Aggregate Abrasion Value
tests to help measure aggregate resistance to wear from abrasion. Jellie (2003) suggests that the introduction of
silica in tyre tread compound may be a factor in the anomalous and more variable patterns of skid resistance
reported by Roe and Hartshorne (1998) in comparison to earlier research that demonstrated more regular and
predictable patterns.

4.2.6 Tyre Contact area and Stress

The contact area between a vehicle tyre and the road surface controls the traction and load transfer of the vehicle.
To understand skid resistance that is available to a road user, it is necessary to explain the influence of the contact
area and contact stress between the tyre and the road surface.

Contact Area
Traditionally, researchers described the contact area by assuming that the average contact stress is equal to the tyre
inflation pressure and therefore the ratio of load to inflation pressure gives the contact area. However, Jellie (2003),
reporting on early work on contact areas of commercial vehicles by Lister and Nunn (1968), showed that the actual
area enclosed within the contact perimeter may be only 60% of the area calculated by dividing the wheel load by the
tyre pressure. As discussed above, the groove area is significant and should be subtracted from the area of the
contact patch to give the true contact area between the tyre rubber and pavement surface. This can be measured by
covering the tyre with ink and traversing the tyre across white paper, thereby leaving an imprint of the contact area.
This contact area can vary according to the type of road surface construction. Woodside and Woodward (2002)
showed that the actual contact for a negatively textured Stone Mastic Asphalt (SMA) surface was 56% of the contact
patch area whilst the actual contact for a positively textured Hot Rolled Asphalt (HRA) Mix was 35% of the contact
patch area.
The shape of an automotive vehicle contact patch is generally rectangular due to the thick side walls and tread. The
width remains fairly uniform but, as would be expected, the length increases with either increased loading or
decreasing tyre pressure (Douglas et al., 2000) as well as when braking or accelerating.
Jellie (2003) summarises the main factors that affect actual shape and size of the contact patch:

Type of tyre;

Tread pattern and depth;

Inflation pressure;

Wheel load; and

Type of road surface, i.e. aggregate size and texture depth.

Distribution of Contact Stress

Various researchers have approached the concept of contact stresses beneath tyres from a number of points of view,
including terramechanics, tyre design and pavement design.

Lippmann (1985) indicated that vertical contact

stresses under tyres should be expected to be non-uniform, especially with low tyre inflation pressures as illustrated
in Figure 4-7. In this case, the vertical component of the inflation pressure acting on the bulging sidewalls beyond
the edge of the tread is supported by increased contact stress under the outer tread rib and by bending stresses in the
78

Chapter 4
tyre carcass. Towards the centre of the tread, the tyre acts more like a membrane, with the inflation pressure being
resisted by an approximately equal contact stress beneath the tyre.
Static testing and dynamic tests have been carried out to measure and understand the forces and stresses within the
contact patch. The measured stress has been found to vary throughout the contact patch and is influenced by many
factors including load, tyre inflation pressure, speed and surface macrotexture, with dynamic stress existing in three
directions, vertical, longitudinal and lateral (A. Woodside & Woodward, 2002). Tielking and Roberts (1987)
reported that the vehicle speed has almost no impact on vertical contact stresses and therefore static vertical contact
stresses are about equal to dynamic vertical contact stresses. However, longitudinal contact stresses are affected by
vehicle speed. The formerly curved tread surface of a stationary tyre sitting on pavement is forced to become planar
within the contact patch. Friction between the tread face and the road surface develops tangential stresses (Figure
4-8a) directed on the road surface toward the centre of the contact patch.

When the tyre is rolling freely,

longitudinal shear stresses develop in the middle of the contact patch (Figure 4-8b), resulting in a double reversal of
the longitudinal stresses from the leading edge to the trailing edge of the contact patch (Douglas et al., 2000;
Lippman, 1985; Siegfried, 1998; Tielking & Roberts, 1987). As a point on the outer rib of the tyre rolls into the
contact patch, it is pushed toward the centre, and therefore transverse stresses also develop within the contact patch.
Tielking and Roberts (1987) also found that longitudinal and transverse contact stresses are affected by tyre
construction (radial or bias ply) with transverse shear stresses being less for radial tyres.

Figure 4-7: Cross Section of tyre with low

inflation pressure under heavy load
(Lippman, 1985)

Figure 4-8: Contact stresses imposed on the road surface

within the tyre contact patch: (a) static tyre and (b) free
rolling tyre (Lippman, 1985)

Liu (1993) and subsequently Siegfried (1998) demonstrated on static and dynamic tests on passenger car and truck
tyres that maximum contact stresses are higher than the average contact stresses assumed by many design methods.
Liu (1993) found that surface texture depth was the most important parameter in influencing contact stresses.
During a tyre contact action, the loading from the tyre is transferred to the road structure through the surface
chippings. If the height of the chips is small (a smooth road surface) the loading results in a low contact stress.
However, for greater heights (high macrotexture surface), the loading is concentrated on the tops of the aggregate
protrusions resulting in high levels of contact stresses. Liu found that a twisting effect occurs on an individual
chipping top during contact with a moving tyre due to a continual change in direction of the tangential contact stress.
Liu proposed that high contact stresses and this twisting effect cause the pavement surface to wear, lose chippings
and polish.
Siegfried (1998) furthered the research of Liu, demonstrating that the maximum vertical contact stress within the
contact patch occurs for approximately 75% of the contact duration. Siegfried (1998) also demonstrated that the
vertical contact stress increased significantly (approximately doubling) towards the edge of the contact patch. This
79

Chapter 4
effect indicates that the edge of a contact patch is subjected to significantly greater interfacial stress than the centre,
which has significant implications for premature surfacing failure through loss of skid resistance or permanent
deformation with roads that carry heavy channelised traffic.
The work was furthered by Douglas et al. (2000) on the effects of very low inflation pressure on the dynamic
contact stresses between heavily loaded truck tyres and road surfaces. His interest was in the use of commercial
central tyre inflation (CTI) systems that have been developed to permit haul truck operators to increase or decrease
the inflation pressure of their tyres as the truck drives along, to suit the road conditions varying from the highest
standard paved roads to the lowest standard gravelsurfaced haul roads. Trials by other researchers had shown that
with reduced tyre inflation pressure haul trucks have increased grade climbing ability, require less maintenance, do
far less damage to access roads and may even heal roads damaged by previous trafficking by vehicles equipped
with tyres at higher inflation pressures. Douglas et al. (2000) concluded that:

Vertical contact stresses are non-uniform under a tyre load and inflation pressure typical of real trucks are
extremely non-uniform when the load is heavy and the inflation pressure is low;

The orientation of the measured transverse and longitudinal contact stresses within the contact patch
generally confirmed the orientations indicated by Tielking and Roberts (1987); and

Longitudinal contact stresses at the trailing edge of the contact patch were significantly greater when the
inflation pressure was low, compared to those for high inflation pressure.

Douglas et al. (2000) postulated from his research that the benefits of CTI system implementations may be due to
differences in transverse and longitudinal contact stresses between the low and high inflation pressure cases.
In summary, Woodside and Woodward (2002) reported that longitudinal contact stresses change direction during a
contact action. This suggests that a single chipping suffers two types of stressing as a tyre passes over it, ie.
compressive and tensile. Combining the different directions of stressing indicates that a screwing type of action
occurs. This will have a greater effect on surfaces with higher texture (e.g. large sized chip seal surfaces in New
Zealand) causing the harder aggregates to polish and softer aggregates to wear away, thereby significantly reducing
surface friction. It may also influence the shearing mechanism of the aggregate-bitumen bond, thus causing a loss of
integral strength and cohesion.
Furthermore, Woodside and Woodward (2002) warned that the use of bituminous mixtures to create surfacings with
high macrotexture depths will result in aggregates experiencing higher levels of dynamic vertical contact stress than
smooth or negative-textured surfacings. The level of stresses was found to be greater than expected and this is
closely related to the axle loading, tyre inflation pressure and the macrotexture depth. They have suggested that
these (load, inflation pressure, texture) are the principal factors which contribute towards the loss of skid resistance
and other types of premature failures.
Woodside and Woodward (2002) recommended that consideration of the contact stresses to be induced on the road
surface should be incorporated into the surfacing design process, whether for a simple surface dressing (chip seal
surface) or a high performance thin surfacing. They have recommended that in order to minimise contact stress,
road surfacings should be designed to have a smooth-textured surface or one possessing a negative texture. Whilst
this offers challenges in terms of bulk water removal, especially at high operating speeds, they have shown that
these surface textures will suffer less dynamic vertical contact stressing and consequently, will be less susceptible to
premature loss of skid resistance and permanent deformation.

80

4.3 Measuring Surface Friction

Chapter 4

4.3.1 Introduction
In order to provide a safer road, it is necessary for the highway engineer to be capable of measuring and predicting
the wet skid resistance of the road surface (A. Woodside & Woodward, 2002). There is a variety of skid testing
equipment / methods available for skid resistance measurement that have been developed by different countries over
the last 50 or more years. However, all of the commercially available test equipment developed essentially use the
same principle, that is, to measure the resistance of a rubber slider or tyre being forced to slide across a wetted road
surface, under an applied load (Austroads, 2005b). The horizontal friction, traction or force resisting the sliding of
the tyre or slider is measured, and the vertical load is either measured or assumed to be constant.
As discussed earlier in Chapter 2, Section 2.5.3 and shown in Equation 2-3, the frictional force measured depends
upon the load that is applied, and therefore the coefficient of surface friction (f or ) is the ratio of the frictional
force resisting motion divided by the applied vertical load.
The range of wet skid resistance measuring techniques available are categorised into either:

In-situ road surface based devices, or

Laboratory based predictive devices.

The in situ road surface (on site) devices can be further divided into two basic categories; those which are capable of
measuring continuously over a long stretch of pavement surfacing (Continuous Friction Measurement Equipment,
CFME devices) and those which measure skid resistance at specific sites (static devices).

The continuous

techniques can also be categorised as either angled test wheel methods or the braked wheel method. The braked
wheel method can be further subdivided into locked-wheel, variable slip and fixed slip methods. The most common
static devices can either be used at a specific point on a road surface or in the laboratory. These techniques all
measure friction, although with different weightings on the variables that surface friction depends upon. Figure 4-9
shows a classification of the different skid resistance measurement methods and subsequent examples of commonly
used measurement devices that all use some form of contact between rubber and the road surface.

81

Chapter 4
Measuring Friction
(microtexture)

Static / Portable
Methods - Rubber slider

British
Pendulum
Tester (BPT)

Dynamic
Friction
Tester (DFT)

Dynamic / Continuous
Measurement Methods

Angled Test
Wheel

Braked in-line
Test Wheel

Sideways
Force
Method

Fixed Slip
Methods

Locked Wheel
Method
(100% Slip)

Variable Slip
Method

Sideway force
Coefficient
Routine
Investigation
Machine
(SCRIM) - UK

GripTester
(GT)
Scotland

ASTM E274
Locked
Wheel
Tester - US

Road Analyser
and Recorder
(ROAR) in
Variable Slip
Mode
Norway

Road Analyser
and Recorder
(ROAR) in
Fixed Slip
Mode

LCPC
Adhera
France

Figure 4-9: Classification of Skid Resistance Measuring contact Methods with common examples.
Due to their significant differences, a direct comparison between the different device results is not generally
possible. As discussed earlier in Section 4.2, measured friction is dependent upon variable parameters such as slip
ratio, testing speed, vertical load, tyre-rubber composition, tyre tread and inflation pressure and the amount of
surface water present. Some of the systems detect the peak friction and some vary the slip in an attempt to operate
around the peak friction level. Henry (2000) states that each method of measuring friction has advantages. The
direct use of the values produced by one type of measurement relates to a different testing scenario. The locked
wheel method simulates emergency braking without anti-lock (ABS) brakes, the sideway force method measures the
ability to maintain control in curves, and the fixed slip and variable slip methods relate to braking with antilock
brakes. The variation of friction with slip speed is shown in Figure 4-10.

Figure 4-10: Relationship of Slip Speed to friction on a road surface (Austroads, 2005b)
82

Chapter 4
The PIARC World Road Association International Harmonisation study (Wambold et al., 1995) discussed in Section
4.6.6 tested 51 various skid testing and texture devices (37 friction measurement and 14 texture measurement
devices). In New Zealand, the common skid testing measurement devices comprise the SCRIM, GripTester, ROAR,
British Pendulum Tester and the Dynamic Friction Tester. They represent five of the six groups identified in Figure
4-9: Classification of Skid Resistance Measuring contact Methods with common examples.
Due to their significant differences, a direct comparison between the different device results is not generally
possible. As discussed earlier in Section 4.2, measured friction is dependent upon variable parameters such as slip
ratio, testing speed, vertical load, tyre-rubber composition, tyre tread and inflation pressure and the amount of
surface water present. Some of the systems detect the peak friction and some vary the slip in an attempt to operate
around the peak friction level. Henry (2000) states that each method of measuring friction has advantages. The
direct use of the values produced by one type of measurement relates to a different testing scenario. The locked
wheel method simulates emergency braking without anti-lock (ABS) brakes, the sideway force method measures the
ability to maintain control in curves, and the fixed slip and variable slip methods relate to braking with antilock
brakes. The variation of friction with slip speed is shown in Figure 4-10.
. These and the ASTM E274 Locked Wheel Tester and on-board measurement systems will be discussed in the
following sections as examples of the main category types of skid resistance measurement devices.

4.3.2 Sideways Force Method

Sideways force devices are an angled test wheel method that use a freely rotating test wheel maintained at an angle
(the yaw angle) to the direction of motion. The sideways force (cornering force) is measured perpendicular to the
plane of rotation as shown in Figure 4-11. The relative velocity between the rubber and the pavement surface for
these devices is approximately:

VR = V sin
Where:

Eq 4-2

VR = Relative velocity (slip speed)

= yaw angle
V = vehicle speed

These systems are CFME devices and produce a low slip speed measurement even though the vehicle velocity is
high (usually 50km/h). As the surface friction depends upon both microtexture and macrotexture, the CFME skid
resistance measurements are usually complemented with macrotexture measurements. The most frequently used
side force devices are the Sideways force Coefficient Routine Investigation Machine (SCRIM shown in Figure
4-12) and the MuMeter (originally designed for use at airports, however is also used for roads) where the yaw angle
is 20 degrees giving a relative velocity slip speed of 17 km/h (34% wheel slip) when the vehicle speed is 50km/h.
Other side force devices are the Belgian Odoliograph and the Danish Stradograph (retired in 1998). The sideways
force coefficient (Sfc) is determined by this method and is defined as the force at right angles to the plane of the
angled wheel, expressed as a fraction of the vertical force acting on the wheel (refer to Eq 4-3).

Sfc =
Where:

Fp
Q

Eq 4-3

Sfc= Sideways force coefficient

83

Chapter 4
Fp= horizontal force at right angles to the plane of the angled wheel
Q = weight force (vertical force)

Figure 4-11: Sideway Force Coefficient Method

Figure 4-12: WDM SCRIM Apparatus

The Sideways-force Coefficient Routine Investigation Machine (SCRIM) apparatus originally evolved from
motorcycle sidecar based testing machines in the 1930s in the UK (shown in Figure 4-13a). Research undertaken by
the UK Road Research Laboratory (later to become TRL), discovered that the force exerted on a wheel that is
angled to the direction of travel, and held in a vertical plane with the tyre in contact with the road surface, could be
correlated with the wet skid resistance of that surface. The concept was further developed through a number of
angled tyre fifth wheel vehicle configurations (as shown in Figure 4-13b and c) and various other machines,
including a small-wheeled braking force trailer (refer to Figure 4-13d) to the present WDM SCRIM machine (as
shown in Figure 4-12). These original devices used analogue technology such as chart recorders to record the data
and all needed an external water tanker to wet the road in advance. Their primary use was for research rather than
for monitoring skid resistance on a network scale (P. G. Roe & Sinhal, 2005b).
In the late 1960s in the UK, the Road Controlling Authorities decided to monitor skid resistance on a regular basis.
This led to the production of the SCRIM device commercially under license by WDM Limited (since 1971), as a
tool for monitoring continuous skid resistance on a network scale (J. R Hosking & Woodford, 1976b). The SCRIM
device was designed to incorporate large-capacity on-board water supply and electronic data recording (initially
using a simple printer and, later punched paper-tape). The operation of SCRIMs in the UK is kept under close
review by the Transport Research Laboratory (TRL) with minimum annual correlation and calibration equipment
trials.
Many incremental developments to the equipment have occurred over the last 35 years in the light of operating
experience and changing electronic technology, but the main Sideways Force principles have remained the same.
Some developments have been to details of the mechanism, to make it more robust and reliable, whilst others have
led to improvements in the data recording system, which now utilises an onboard PC and touch-sensitive screen.

84

Chapter 4

(a)

(c)

(b)

(d)

Figure 4-13: Early Skid Resistance measurement devices UK - 1930s to 1960s (P. G. Roe & Sinhal, 2005b)
Roe and Sinhal (2005a) report that the test wheel on a SCRIM device is mounted on an angled beam known as the
swinging arm . This is fixed at one end through a flexible bearing to a large, approximately triangular, steel plate
(the back plate) that is mounted on vertical shafts attached to the chassis through a rigid frame. Two bearing units
on the forward edge of the back plate and one on the rearward edge allow the plate to move up and down shafts,
which are protected from the elements by flexible gaiters. A spring and damper (shock-absorber) system links the
swinging arm and back plate to provide a simple suspension for the test wheel. The combined mass of the back
plate, the swinging arm, the wheel hub and the test wheel provides a 200kg static vertical load.
In 2004, all UK SCRIM devices were fitted with a Dynamic Vertical Load Measuring (DVLM) system that enabled
the vertical load applied to the test wheel to be measured dynamically. This enhancement allowed errors in the
reported SFC (sideways force coefficient) to be automatically corrected. The errors could occur as a result of
changes induced in the vertical load on uneven roads or around curves. The vertical load is measured by a straingauge system that is integral with the upper fixing point of the shock absorber. From this, the vertical load under the
wheel is determined by measuring the reaction force between the swinging arm and the back-plate transmitted
through the damper system. The various features of the SCRIM wheel hub system are shown in Figure 4-14. The
side force is measured with a load cell mounted behind the wheel hub.

85

Chapter 4
Wheel lift/lower
actuator

Vertical shaft

Flexible gaiter

Vertical load cell

Front upper
bearing

Shock absorber
and spring

Back plate

Rear bearing

Front lower
bearing

Test wheel

Flexible gaiter

Wheel hub

Swinging-arm
pivot bearing

Water feed pipe

Figure 4-14: SCRIM Wheel Assembly Unit (P. G. Roe & Sinhal, 2005a)
The SCRIM machine uses a smooth tyre fabricated from natural rubber (within a specified resilience range) to
reduce errors through tyre wear. For a standard test at 50km/h, water is delivered in front of the test wheel at a rate
that provides a nominal water film thickness of 1mm.

The WDM Limited SCRIM device from the UK is

contracted to test the Transit New Zealand State Highway network (approximately 10,500kms) once every year in
the summer months. This has been undertaken since 1995. The SCRIM device specified by Transit NZ requires a
test wheel in both left hand and right hand wheel tracks. The SCRIM device collects much more than just skid
resistance condition data. It also collects macrotexture and longitudinal roughness, transverse profiles (for rutting)
by the use of laser profilometers, road centreline curvature by differential GPS, video capture and other road
inventory data at in-highway speeds.

4.3.3 Braked Wheel Method

The braked wheel methods comprise a range of devices that use similar techniques of varying sophistication to
measure the braked surface friction.

GripTester
The GripTester, shown in Figure 4-15a is a three wheeled CFME skid resistance tester that was originally designed
by Findlay Irvine Limited, Scotland in 1987 for airport runway operations. It is now in use in over 30 countries
around the world with over 300 units sold. It can be towed behind a vehicle with an automatic water delivery
system or pushed by hand by the operator. The GripTester has one test wheel that is braked and two bogey driving
wheels as shown in Figure 4-15b). Its mode of operation is the simultaneous measurement of drag force (Fd) and
load force (Q) on a single treadless ASTM specified rubber test tyre (standard specification) of 254mm diameter,
during braked skidding at approximately 15% of the survey speed. The fixed slip of the test tyre is achieved by
means of a chain and sprocket transmission. The test wheel axle features strain gauges to measure the tractive
longitudinal and dynamic vertical forces acting on the test tyre.

The GripTester reports the surface friction

coefficient as a Grip Number (GN) which is the fraction of tractive drag force (Fd) over the load force (Q) as given
in Eq 4-4. The GN ranges from 0 to 1.2.

86

Chapter 4

F
GN = d
Q
Where:

Eq 4-4

GN = Braked friction force coefficient by means of a GripTester

Fd = drag force (horizontal)
Q = weight force (vertical)

GripTester surveys typically use an automated (or manual) water delivery system that provides a 0.25mm water film
depth beneath the testing tyre. The water film depth can be varied if required. The Grip Number (GN) is measured
every 1m but is typically reported as an average over every 10 metre length.

(a) The GripTester (side view)

(b) GripTester measuring wheel and

chain transmission

Figure 4-15: The GripTester (Findlay Irvine Ltd, 2005)

The GripTester is commonly used for friction measurements on airport pavements, for research purposes and more
recently for monitoring road networks in the UK, Europe, Australia and New Zealand. The braked wheel device can
be operated in push mode of 5km/h up to a maximum speed of 130km/h. It is relatively easy to transport (at
approximately 85kg weight) and can be used with any towing vehicle. The GripTester has only one test wheel, and
therefore results are typically obtained for the left wheel path of a road lane only, which is usually the wheel path
with the lowest measured surface friction. However, results can be obtained from the right wheel path separately,
given appropriate on-road traffic management. Recent improvements to the axle system in the Type D GripTester
have resolved the earlier problems that were experienced, associated with the measurement of skid resistance on
bends.
The GripTester is compact and highly manoeuverable and is a flexible tool that allows testing on both road, air
pavement and footway surfaces and is relatively inexpensive to operate. Experience in Australia and New Zealand
has shown that the equipment is more suited to testing over targeted road lengths, rather than being used on large
network wide surveys (Donbavand, 2004).

Electronic Recording Decelerometer

These devices do not measure a coefficient of friction but the Drag Factor and includes the effects of any road
inclines. Examples of these devices are the Bowmonk AFM2, used extensively on runways in North America and
Europe (Bowmonk Ltd, 2005) and the SkidMan (Figure 4-16), manufactured by Turnkey Instruments Ltd, UK
87

Chapter 4
(2005) and the Vericom System. The Bowmonk AFM2, SkidMan and the Vericom devices are compact, portable
electronic instruments for measuring and recording the maximum deceleration possible on an airport runway or road
surface. They are a modern day equivalent of a mechanical decelerometer. They contain a solid state accelerometer
and an accurate crystal-controlled clock. During a locked wheel (100% slip) surface friction test, the accelerometer
is positioned close to the centre of gravity in a host vehicle and measures the deceleration (g-force) experienced by
the transporting vehicle from the initial measuring speed to a stopped position. These devices can measure the
deceleration (g-force) 400 times per second with an accuracy of better then 2% (Bowmonk Ltd, 2005; Turnkey
Instruments Ltd, 2005) and convert the deceleration force to a peak friction coefficient (fp or p) for the average of
the four wheels. The output is read by a microprocessor and stored in the 128K byte memory for automatic analysis
at the end of the test. The host vehicle can be anything from a 1.5 tonne utility vehicle to a standard car. A fifth
wheel may be used to accurately measure the equally important measured stopping distances. The measuring
procedure is non-continuous, and is repeated several times usually using commercial tyres fitted to the host vehicle.
Figure 4-17 shows the output of a SkidMan test demonstrating the difference between the significantly shorter
braking times, and therefore distances of ABS brake systems, when compared to non-ABS brake systems at 100%
slip or fully locked brakes (Bullas, 2005).

Figure 4-16: SkidMan

(Bullas, 2005)

Figure 4-17 Output from Electronic Recording Decelerometer (SkidMan)

showing a comparison of ABS and non ABS braking (Bullas, 2005)

Norsemeter Road Analyser and Recorder - ROAR

The Road Analyser and Recorder (ROAR) is a surface friction measuring device originally developed by
Norsemeter in Norway as shown in Figure 4-18 and Figure 4-19. It is a hydraulically braked friction measuring unit
that is trailer mounted and utilises an ASTM 1551 smooth rubber test tyre of 400mm diameter.

88

Chapter 4

Figure 4-18: Road Analyser and Recorder - ROAR

schematic (Pavement Management Services, 2005)

Figure 4-19: ROAR device with testing wheel and

hydraulic braking mechanism demonstrated.

The ROAR device can operate in two different measuring modes:

Standard fixed slippage CFME mode, where the constant slip ratio of the measuring wheel can be varied
between 1 % and 100 %; and

Variable slip measuring mode where the full friction - wheel slip curve is determined by progressively
braking the test wheel at small intervals from a free rolling tyre (0% slip) to a fully locked wheel (100%
slip). A variable slip controller continuously measures the braking friction force that the road surface
exerts against the braking wheel (via the torque on the wheel axle) which is then converted to a friction
coefficient whilst a braking action linearly decreases the rotational wheel speed from free rolling to a
locked wheel. Each brake test lasts 0.5 seconds and is controlled via the brake actuating force. The nature
of the variable slip measuring test method means that tests are undertaken at spot locations and not
continuously and an assumption that the road surface is largely similar in characteristics and condition is
required.

The test method has been incorporated into an ASTM Standard test method (ASTM

International, 2001).
A vertical static load of 1.5kN is applied to the test wheel which has an inflation pressure of 207kPa (30psi). The
test tyre is an ASTM E 1551 tyre manufactured to the same rubber specification as the GripTester. The variable slip
mode allows the peak friction coefficient ( p) and its corresponding slip speed to be directly determined over an
approximate 7m section of pavement for a standard test speed of 50km/h. The slope of the friction curve after the
peak has occurred can be also be calculated from these data as the full friction curve, as shown in Figure 4-10, has
been determined. This allows the sensitivity of that road surface as a function of vehicle speed to be determined
directly with one test run. Furthermore, as the ROAR can perform variable slip measurements, it can report directly
the International Friction Index (IFI), discussed in Section 4.6.6, when combined with macrotexture measurements.
The Mark II and III ROAR models are compact units with a single test wheel that can be mounted to almost any
kind of host vehicle or trailer and incorporate air and surface temperature probes. The test wheel unit can be moved
along the back of the trailer as shown in Figure 4-18 to enable skid resistance testing in the left, right or centre wheel
track and, if required, at any point in between. The test speed range is 20km/h to 130km/h and software can
automatically adjust the water flow rate to supply a constant 0.5mm or 1.0mm nominal depth over this speed range.
The unit is often towed by a vehicle that also has laser texture measurement sensors for the determination of
macrotexture, road roughness and the transverse profile of the road lane.
89

Chapter 4

Locked Wheel Pavement Friction Tester

The locked wheel systems such as the Dynatest 1295 Pavement Friction Tester (PFT) or the LCPC Adhera system
produce a 100% slip condition (i.e. locked wheel) and as such simulate the emergency braking process for non-ABS
vehicle brake systems more closely than other methods. The relative velocity between the surface of the tyre and
the pavement surface (the slip speed) is equal to the vehicle speed. The Dynatest 1295 PFT tester consists of a fully
instrumented tow vehicle and test trailer as shown in Figure 4-20. It uses a two axis force transducer to provide
dynamic vertical load and horizontal drag (tractive) force measurements (Dynatest, 2005). The locked wheel trailer
testing mechanism is shown in Figure 4-21.

Figure 4-20: Dynatest 1295 Pavement Friction Tester

(Dynatest, 2005)

Figure 4-21: Locked Wheel Trailer Testing

mechanism (Dynatest, 2005)

The brake is applied and the force is measured and averaged for approximately one second after the test wheel is
fully locked. The time whilst fully locked is minimised due to the high rate of tyre wear. The test wheel brakes are
then released and the wheel spins up to the vehicle speed. The operating test speed ranges between 30km/h to
110km/h. Because the force measurement is continuous during the braking process, these systems can usually also
detect the peak friction ( p). An example of a computer screen friction test output from the Dynatest 1295 PFT is
shown in Figure 4-22. The nature of this test method means that tests are undertaken at spot locations and not
continuously. Hence, it is assumed that the road surface is largely homogeneous in characteristics and condition.
This method is extensively used in the US but has not been used to date in New Zealand.
The braked locked wheel testers are usually fitted with a self watering laminar flow nozzle system for wet-testing
that can be raised and lowered for each test set at a common nominal water film depth of 0.5mm, distributed 25mm
wider than the tyre width. ASTM International have developed an ASTM Standard Test Method E274-97 (ASTM E
274-97, 1997) for the locked wheel test method that reports the coefficient of friction () as the skid number (SN)
which is the measured value of friction multiplied by 100. The ASTM E-274 test method provides for reporting
results using a standardised ribbed test tyre as follows: SN{Test Speed} followed by R for the ribbed tyre or S for
the smooth tread tyre and if the test speed is expressed in kilometres per hour it is enclosed in parenthesis. For
example, the value of SN40R is equivalent to SN(64)R.

90

Chapter 4

Figure 4-22: Computer Screen Test Output from Dynatest 1295 Pavement Friction Tester (Dynatest, 2005)

4.3.4 British Pendulum Method

The British Pendulum method uses a portable skid tester with a rubber slider mounted on a pendulum arm and is
often referred to as the British Pendulum Tester (BPT) or the Portable Pendulum Tester (PPT). The BPT which
is shown in Figure 4-23 and Figure 4-24 was developed in the 1950s. The equipment and testing methodology was
first specified by the UK Road Research Laboratory (1960). The British Standard BS 7976-1 (2002) details the
configuration and properties of the slider and BS 7976-2 (2002) describes the test method. It is also specified in the
US as ASTM Standard E-303 (2003). It is a method that can be used either in the laboratory or the field and has
been widely used since the early 1960s.

Figure 4-23: British Pendulum Tester (Austroads,

2005b)

Figure 4-24: Schematic of British Pendulum

Tester (BPT) set up for Laboratory Testing
(A. Woodside & Woodward, 2002)

91

Chapter 4
The tester works using the Izod principle by releasing a pendulum that is mounted on a soft spring from a height that
is adjusted so that a rubber slider that is inclined by approximately 20 contacts the pavement surface over a fixed
length at the bottom of its arc for a distance of 125mm to 127mm. When the pendulum reaches the surface, the
rubber slider moves across the surface and the friction reduces the kinetic energy of the pendulum in proportion to
the level of friction. When the slider breaks contact with the surface, the reduced kinetic energy is converted to
potential energy as the pendulum swings and reaches its maximum height. A pointer on the tester marks the
recoverable height of the pendulum swing after the surface contact. The scale is calibrated in units of Skid
Resistance Value (SRV), and tables are available to allow friction coefficients to be derived from the measured SRV
values in terms of a British Pendulum Number (BPN) or Portable Pendulum Number (PPN) over a range of zero to
140. The difference between the height before the release and the height recovered (the amount the pendulum was
retarded) is equal to the loss of kinetic energy due to the friction between the slider and the pavement or surface
sample.
As the average velocity of the slider relative to the pavement is also a function of the friction, the average slip speed
is not constant throughout the surface contact and therefore decreases with increasing friction. The typical slip
speed of the BPT is usually assumed to be about 10 to 11km/h (P. Cairney, 1997; J J Henry, 2000). As the slip
speed of the BPT is very low, the BPN is mainly dependent on microtexture and, therefore, is used as a surrogate for
microtexture. This is very useful, because the direct measurement of microtexture is difficult, although the BPT
does not measure the impact that macrotexture has on skid resistance as vehicle speeds increase. Therefore,
measurements of surface macrotexture by other means should be undertaken in conjunction with this equipment.
If the rubber sliders are made from natural rubber which is temperature dependent (UK, European, NZ and
Australian practice), the SRV readings obtained are typically corrected for slider temperature (which is determined
from the road surface temperature) to a standard 20 Celcius. However, in the US, it has been the practice to use the
rubber compound specified for the ASTM standard test tyres (ASTM E274-97, 1997; ASTM E501-94, 2000). This
test uses ASTMspecified synthetic rubber, which is formulated to be independent of temperature, and therefore no
temperature correction is required (J. J. Henry, H. Abe, S. Kameyama, A, Tamai, A, Kasahara, K, Saito, 2000).
The rubber slider has dimensions of 25mm x 76mm and makes contact with the surface over approximately 125mm.
The same skid tester (BPT) is used in the polished-stone value (PSV) test BS EN 1097-8 (2000) to assess the skid
resistance of aggregates after accelerated polishing (refer section 4.4.2).

As the device is a static portable test

method, it cannot be used for continuous measurement. The results have also been shown to be operator-dependent.
It is best suited to measuring site-specific skid resistance such as at intersections or accident sites. Tests are
normally carried out on a pre-wetted pavement.

As with any sensitive equipment, regular maintenance and

calibration of the device is important. Detailed procedures of the methodology and the calibration of the BPT are
described in BS 7976-1, (2002), Oliver and Witt (1978), ASTM E-303 (2003) and Transit New Zealand
Specification T/2 (Transit New Zealand, n.d.).

4.3.5 The Dynamic Friction Tester

The Dynamic Friction Tester (DF Tester) is a stationary skid testing device developed by Nippo Sangyo Co. Ltd
(2005b). The DFTester device was designed in Japan mainly to measure the dynamic coefficient of friction on road
surfaces. However, it can also be used as a static device to determine the friction on laboratory-prepared samples,
paved surfaces of footpaths, promenades, amusement parks and on floor surfaces of buildings and gymnasiums. It
has been found to be very stable with time and to give highly repeatible measurements, and has recently been chosen
92

Chapter 4
as the standard reference in the recently revised IFI ASTM International Standard (Henry, 2003). The testing
procedure and methodology is described in ASTM Standard Test Method E-1911 (2002).
The DF Tester consists of a horizontal spinning disk that spins with its plane parallel to the test surface. The
spinning disk is fitted with three spring-loaded rubber sliders centred on a diameter of 284mm. These contact the
paved surface as the disk rotational speed decreases due to the friction generated between the sliders and the paved
surface. The DF Tester can be used for laboratory investigations and in the field on actual paved surfaces. The disk
is brought to the desired rotational velocity, corresponding to the maximum tangential velocity of the sliders (V up
to a maximum of 90km/h). Water is introduced in front of the sliders and the disk is lowered to contact the test
surface so that it bears the full velocity of the disk and model assembly. The torque is monitored continuously as the
disk rotational velocity reduces due to the friction between the sliders and the test surface. The torque signal is
reduced to a measurement of friction by converting the torque to the force on the sliders and dividing by the weight
of the disk and motor assembly. The coefficient of friction () is then calculated as follows:

=
Where:

F
Q

Eq 4-5

= coefficient of friction as measured by the DF Tester

F= Torque force (horizontal)
Q = weight force on the three rubber sliders (vertical)

By holding Q constant and substituting K (a constant of proportionality) for 1/Q,

= K.F

Eq 4-6

Thus, the coefficient varies in direct proportion to F. The DF Tester has a main motor-driven unit (refer Figure
4-25a), that consists of a fly-wheel and disc with three rubber sliders (refer Figure 4-25b) attached by leaf springs,
and a control unit. The sliders are pressed on the test surface by the weight of the device through three rollers. Each
slider is loaded to 11.8N by the leaf springs. The disc and the fly wheel are connected by a spring balance mounted
along a circle on which the rubber sliders are fixed. Due to the forces on the rubber sliders, displacement occurs in a
spring balance. This displacement is converted to an electrical signal attached to the opposite side of the disc. The
signal is output through a slip ring and brush, both of which are mounted on a driving shaft. The speed of rubber
sliders is measured from the output of a rotational speed dynamo. The friction at 20, 40, 60 and 80km/h is recorded
and the friction speed relationship is plotted as shown in Figure 4-25c.
The slider assembly consists of a steel backing plate to which is bonded a 6 x 16 x 20mm rubber shaped as shown in
Figure 4-26. This shape provides a contact pressure of 150 kPa. The rubber compound is synthetic rubber as
specified by ASTM E 501 (2000) and is required to have a Shore Hardness of 58 2.

93

Chapter 4

(a) DF Tester Main unit

(b) DF Tester spinning disk

and rubber sliders

(c) DF Tester typical result output.

Figure 4-25: Dynamic Friction Tester (DF Tester) Components (Nippo Sangyo Co Ltd., 2005)

Figure 4-26: Dynamic Friction Tester (DF Tester) rubber sliders

The DF Tester has the advantage of being able to measure the friction as a function of speed over the range of zero
to 80km/h. A significant benefit of the DFTester is that whilst being a stationary device, it has a significantly larger
contact area than the British Pendulum Tester, and is less affected by individual aggregate chips. It has been found
to produce stable and highly repeatable measurements over time. These benefits enable the use of this device as a
calibration device for other Continuous Friction Measurement devices such as the GripTester and ROAR (Wambold
et al., 1995).

4.3.6 Systems for On-board Measurement of Tyre-Pavement Friction

Significant research has been undertaken in recent years on vehicle-dynamic systems like the anti-lock braking
systems (ABS), traction control and driver assistance systems such as autonomous intelligent cruise control and
active driving systems.

All these systems require information about friction and as such have on-board

measurement methods to develop, in real-time, information about the roadtyre friction interface. Jellie (2003)
summarises the European PROMETHEUS research project from Breuer et al. (1994) that developed two different
methods of determining the surface friction for these types of systems:

The first method utilises an optical sensor based model system that calculated descriptors of water film
thickness and asperity height to determine the wetness of the road and the surface profile, which together
with vehicle speed form the main inputs to a friction model. The system allows prediction of friction
without applying forces to the wheels. However, the system only functioned accurately under conditions
where its sensors could detect (i.e. not in heavy rainfall) and for which the model has been trained.
94

Chapter 4
The second method gathers information on tyre-pavement friction by using the deformation of tread

elements in the tyre to infer the local frictional behaviour between the tyre and the road. This effect based
approach uses a passenger car fitted with an fzd tyre sensor. The sensor records information about the
distribution of longitudinal, lateral and vertical deformation of the tread element, which provides
information on localised sliding and sideways forces.
In the future, it is expected that these vehicle systems, coupled with advances in automated control and intelligent
transport systems, will be developed to a point where vehicles will be almost completely prevented from getting into
a loss of control situation. The vehicles could thus automatically prevent driver behaviour that could lead to a loss
of control by skidding to occur. This will, however, require significant changes to the balance of the driver / vehicle
/ road environment traffic system, where equilibrium and control is held in balance by the human element (the
driver).

4.4

Laboratory-based Surface Friction Prediction Methods

4.4.1 Introduction
As discussed in Section 4.3.1 most common skid resistance measuring devices are based upon sliding some form of
rubber over a paved surface. Some devices were developed specifically for in-field measurement and others
originally for the laboratory that were sometimes adapted for in-field use. The following section reports on
laboratory based methods for:

the Polished Stone Value method of prediction and measurement of skid resistance by accelerated polishing
of aggregates;

the direct measurement of microtextural profiles; and

automatic image analysis.

4.4.2 The Polished Stone Value (PSV) Test

Introduction
Early research by the UK Transport Research laboratory into the polishing of the microtexture of surfacing
aggregates led to laboratory tests being developed to assess the resistance of different aggregate sources and
geological types to such action. The tests included the Polished Stone Value (PSV) test (originally BS 812: Part 114
(1989) and Road Note 27(Road Research Laboratory, 1960); now superseded by BS EN 1097-8 (2000); ASTM D3319 (2000); AS1141.40 (1999) the Polished Aggregate Friction Value (PAFV) test (AS 1141.41-1999), the British
Flat Bed Test and the French / Canadian GRAP test. Of these tests, the main laboratory method used throughout the
world is the UK PSV test (A. Woodside & Woodward, 2002). The test was developed after World War II and
developed into a British Standard in 1960. At the time there was a large increase in vehicles and speeds on UK
roads and this gave impetus to the need to predict the ability or resistance of an aggregate chip to polish under heavy
traffic loads.

Understanding this relationship between surfacing material and road safety would enable the

specifying of appropriate surfacing aggregates to perform over the intended surfacing lifecycle (R. Hosking, 1992).
Apart from some minor modifications, the original method as devised by Maclean and Shergold (1958) was recently
accepted as the CEN European Norm (BS EN 1097-8) for measuring the polishing characteristics of an aggregate.
As discussed in Chapter 3, Section 3.3, to provide good microtexture over the desired life of the surfacing; the
aggregate chip must be resistant to polishing. The PSV tests allowed empirical relationships to be established
95

Chapter 4
between traffic levels, resistance to polishing and skid resistance, and as such are used in the specification of
surfacing materials by Road Controlling authorities such as the UK Highways Agency and Transit New Zealand
(refer to Section 2.3.4). The PSV test has been used to determine the PSV of a chip source (quarry, river or pit)
rather than that of the chip from single stock piles. PSV testing has been used for new sources of aggregates, ongoing annual testing of quarry sources, or existing sources where crashes or skid testing on road surfaces indicate
possible problems (Transit New Zealand & Roading New Zealand, 2005).

The Polished Stone Value Test Procedure

The PSV test procedure is a two-stage test; the first stage comprises accelerated polishing of the aggregate, followed
by the determination of the resulting coefficient of friction using the British Pendulum Tester (BPT), described
earlier in Section 4.3.4. The state of polish is measured in the second stage.
In the first stage, slightly curved test specimens of cubic-shaped 10mm sized aggregate chippings (Transit NZ Grade
4) are fixed to the curved moulds using an epoxy binder (refer to Figure 4-27) and then 14 samples are bolted
around the road wheel of the accelerated polishing machine (refer to Figure 4-28 and Figure 4-29). The samples,
including a sample prepared with a control chip from a particular quarry reputedly always having the same PSV, are
then subjected to wet polishing action with the road wheel rotated at a set speed using the accelerated polishing
machine. The wet polishing action lasts for six hours during which time each specimen is subjected to 115,200
passes of a solid rubber tyre held against the face of the wheel under a force of 725N. The polishing stage has two
phases of three hours polishing each, the first as a roughening phase where a coarse emery abrasive is used as the
polishing medium. This is followed by a polishing medium of fine emery flour and water for the remaining three
hours to complete the first stage of the PSV test.
The second stage of the test measures the state of polish reached by each of the specimens using the BPT. This
results in a laboratory assessed Polished Stone Value (PSV), the higher the value obtained (between 30 and 80) the
better an aggregate s resistance to polishing.

Figure 4-27: Prepared test sample showing chips embedded in epoxy (Transit NZ & Roading NZ, 2005).

96

Chapter 4

Figure 4-28: Accelerated Polishing Machine.

Figure 4-29: Schematic of the Accelerated Polishing

Machine (Woodside and Woodward, 2002).

Jellie (2003) states the factors that can affect the variability of the PSV test can be summarised as test equipment,
the operators and the aggregate source. The control stone sample is obtained from a single source that has been
shown to give a consistent PSV. The PSV test compares the difference between test specimens and control
specimens to minimise differences between tests within individual laboratories and between different laboratories.

A critical evaluation of the PSV Test

As discussed above, the UK PSV test is widely used throughout the world to predict aggregate skid resistance in the
laboratory. However, critical evaluation shows that, in contrast with modern traffic conditions, which have changed
significantly over this period, this 50 year old equipment has changed relatively little during its lifetime (A.
Woodside & Woodward, 2002).
Significant research has been undertaken in recent years on the polishing and skid resistance characteristics of road
surfacing aggregates, together with the variability and limitations of the PSV test by researchers at the Transport and
Road Assessment Centre (TRAC) at the University of Ulster in Northern Ireland (Jellie, 2003; Perry, 1996; A. R.
Woodside, 1981; W. D. H. Woodward, 1995) and Bean and Pidwerbesky (1994) in New Zealand. A summary of
the research findings in terms of the use of the Polished Stone Value test for the prediction of the long-term
performance of aggregate skid resistance follows.
Woodside (1981) polished PSV test specimens under non-standard test conditions to examine the effects of varying
load, contact area, contact length and contact stress with tyre pressure. He found that the contact stress and contact
area were directly related to the rate of polishing and thus the frictional resistance of a PSV test sample. The PSV
test initially used a pneumatic tyre to cause polishing. This has since been replaced by a solid rubber tyre to
improve the distribution of applied loading and so reduce the problems of variable loading within the contact patch
due to increases in tyre pressure during testing. However, in practice (as discussed earlier in Section 4.2.6) the
rubber of a pneumatic tyre will mould itself around aggregate particles and stresses are increased at the tyre
97

Chapter 4
sidewalls. Therefore the use of a solid tyre does not accurately simulate the conditions in the contact area between a
vehicle tyre and the pavement surface.
Woodward (1995) related the PSV to other aggregate test properties. His results showed poor linear correlations for
PSV with standard aggregate tests to assess strength, durability and physical properties, but better linear correlations
with tests to assess wear, particularly micro-Deval. The micro-Deval test (BS EN 1097-1, 1996) determines the
resistance to wear between a sample of aggregate and abrasion by 10mm ball bearings in a rotating drum containing
water. With many highly trafficked roads in the UK and worldwide (being significantly greater than 50 years ago),
there is now significant demand for very high PSV aggregates. This has favoured the use of certain types of
aggregate (e.g. gritstone or greywackes) as the principal type of surfacing aggregate. However, small increases for
higher PSV aggregates correspond to large decreases in terms of wear and other engineering properties. This is
demonstrated by Woodside and Woodward (2002) and shown in Figure 4-30 where the PSV and Aggregate
Abrasion Value (AAV) for arenaceous rocks clearly show that, with small increases in PSV, this type of aggregate is
susceptible to an increasing amount of wear. The Aggregate Abrasion Value (AAV) is a measure of the resistance
of an aggregate to surface wear by abrasion through a horizontal steel grinding lap using Leighton Buzzard sand as
an abrasive BS EN 1097-8, (2000).

Figure 4-30: Plot of PSV and AAV for arenaceous rocks (A. Woodside & Woodward, 2002)
Woodward (1995) also examined the variation of PSV within a single aggregate source and found a significant
variation (PSV 60 to 69) between samples from different localities within a single quarry. Tests on extended
polishing on PSV test specimens for a total period of 12 hours also showed that almost all specimens typically
reduced in PSV by a further 10%.
Perry (1996) investigated size and shape factors of gritstone aggregates and the effect on the PSV result. The results
showed that, whilst the size of aggregate used in the PSV test samples was not found to have a significant effect on
polishing (except when 3mm grit was used), the aggregate shape was found to affect the BPT values (the second
stage of the PSV test). PSV test specimens prepared using flaky aggregate were shown to give higher pendulum
values.

Perry (1996) attributed this to the flaky aggregate being more angular, thereby having a rougher

macrotexture and presenting a greater surface area. The 3mm grit sized aggregate also gave higher BPT values, and
this was also attributed to the samples presenting a rougher macrotexture sample surface. Cenek et al. (1998) also
found (as discussed in Chapter 3, Section 3.3) that small sized aggregate and shape, as measured by increasing
percentage of crushed faces, increased skid resistance.
Perry (1996) also investigated the effects of modifying the accelerated polishing machine to induce higher contact
stresses on PSV test specimens by angling the rubber wheel by 20 and reducing the standard load by 50% to 360N.
98

Chapter 4
High specified PSV aggregates were used with the angled rubber wheel and polishing extended to 15hours. The
result was 10-24% reduction of PSV below the standard PSV.
Bean and Pidwerbesky (1994) determined from their research the following sources of variability and/or the
repeatability in between PSV test runs:

Wearing of the solid rubber tyre used to polish the specimens. There was a tendency for the measured
coefficient of friction of the control samples to drop as the rubber tyre wore.

Operator variability is most likely to occur at the specimen preparation stage, especially for the aggregate
particle selection and placement of aggregate chips in the mould. Samples are supposed to be placed in a
random manner. However, it is difficult not to line them up in distinct rows and columns. It was suggested
this may result in higher PSVs due to increased hysteresis losses.

Operator variability between two experienced laboratory operatives varied between a reproduceability of
1.6 PSV units for one operator and 3.0 for the other.

Variation within the aggregate source itself. Aggregate sources can show significant variation in their
petrology, making it difficult to obtain a representative sample, e.g. a source may be representative of only
one quarry face. A blended sample from a number of quarry faces may be more representative, but may
not represent any particular stockpile.

Sampling and stone selection of a sample source by preparing samples using aggregate with rounded faces,
aggregate with rough faces, aggregate with high percentage of quartz and random aggregate chips. The
PSV Test results demonstrated that the rough faces gave results 5 PSV point higher than the random
selected sample. Aggregate with high quartz content on the surface was not significantly different from the
random selected sample.

Effects of aggregate crushing process. Tests that reduced the normal three-stage crushing process (primary,
secondary and tertiary crushing) to a one stage crushing process obtained from the same quarry face were
shown to increase the PSV tests results by up to 7 PSV units. This was thought to be due to the process of
crushing removing angularity from the aggregate particles.

PSV Test Summary

Various stone polishing tests are available to assess the propensity of a stone to polish. The most frequently used
test worldwide and the one primarily used in New Zealand is the Accelerated Polishing Test which produces
Polished Stone Values. The PSV of an aggregate has been related to its in-field skid resistance by a prediction
equation that includes the expected commercial traffic volumes since the 1970s. This is discussed later in Section
4.6.2. However, although the PSV test attempts to simulate in-service factors such as water, traffic loading and the
presence of detritus, it cannot adequately simulate real in-service conditions where there are large variations in
traffic conditions, stresses, types of surface treatments, etc. The standard six hour PSV test therefore is not a true
reflection of the performance of an aggregate, but rather it offers a ranking mechanism under simplified laboratory
conditions where end-use, climate and trafficking conditions are not considered. This must be borne in mind when
considering the use of the values obtained by the PSV test (A. Woodside & Woodward, 2002).
There have been significant attempts to improve the accuracy of predicting an aggregate s in-situ skid resistance.
The TRAC laboratories in Northern Ireland have developed modified PSV testing protocols (A. Woodside &
Woodward, 2002) that have utilised extended polishing cycles, the repetitive usage of coarse and fine emery
polishing cycles, different types of polishing agents and angled polishing to increase the amount of polish
experienced during testing, thereby better simulating heavy trafficked routes. The results of these test methods have
99

Chapter 4
been compared to others such as the French GRAP and the British Flat Bed test methods. However, the overall
conclusion by TRAC was that the standard PSV test remains the best laboratory test to measure wet skid
resistance (Woodside and Woodward, 2002). Although, modified PSV test methods, as described above, can
however better simulate an ultimate state of polish with higher simulated traffic loads and stresses.

4.4.3 Profiling and Imaging Techniques for Assessing Aggregate Microtexture

Forster (1989) used microtexture profiling techniques to predict the polishing performance of surface aggregate by
applying a thin white opaque coat to the sample surface so that when the surface is illuminated, light is uniformly
reflected from it and does not penetrate. The sample is mounted on a viewing stage and a semi-circle of light
projected onto it at a 45 angle. The straight side of this semicircle is sharply focused and, where the straight edge
hits the sample surface, it produces a profile which duplicates the vertical profile through the sample because of the
incidence and viewing angles. The profile image can be viewed through a microscope attached to a television
camera.
Microtexture profiles are described by two parameters that are measured directly and a third that is derived from the
first two. These parameters are defined in Figure 4-31. The number of microtextural peaks per unit length is
defined below as the average asperity density and the height is the average of the number of peaks profiled. Both
the height and the density of microtextural peaks are necessary under wet conditions as these determine the
penetration of the surface water film and the amount of adhesion the pavement surfacing will develop (directly
related to the number of contact points between the tyre and surfacing, i.e. density and height of peaks). Forster
(1989) suggests optimal peak height as being in the range of 0.01 to 0.1mm. Forster (1989) also defines an average
shape factor as a measure of the height to width ratio of asperities which was also found to affect the tyre pavement adhesion in wet conditions.
L

Average asperity density = number of peaks / length of profile

3/L
Average asperity height = sum of heights / number of peaks
(a + b+ c) / 3
Average shape factor = average height / average width
{ (a + b + c) / 3 } / (L / 3)

Figure 4-31: Definitions of microtextural parameters (Forster, 1989)

The research used 100 microprofiles each 2mm long to characterise the microtexture of an aggregate sample.
Forster (1989) correlated microtextural profiles from 87 primarily asphaltic concrete (AC) cores in the US to BPT
data from the same cores and SN(64) skid resistance measurements measured in the field where the cores were
obtained from. A correlation coefficient of R2 = 0.68 was obtained between the British Pendulum Number (BPN)
and the shape factor, although this was shown to increase if the percentage of the tyre-pavement contact area was
also measured for the various core samples.
100

Chapter 4
The use of microtextural profiles by Forster (1989) represents a different approach to predicting the polishing
performance of surfacing aggregates in the field. It shows some promise, although further refinement is required to
try to improve the correlation with on site measures of skid resistance.

4.4.4 Automatic Image Analysis

Jellie (2003) summarises an automatic image analysis technique to determine the effect of grain size distribution and
grain shape on the PSV of greywacke aggregates that was developed by Perry (1996). Samples were assessed by
thin section using a video camera attached to a biological microscope under plane polarised light. Image processing
and analysis were carried out using software. Perry (1996) carried out geometrical analysis by converting the image
to a binary colour scale making the feldspar and quartz grains light against a dark background. The grain shape was
then quantified using form factor and Wadells measured sphericity. Grain sizes between 0.03 and 0.5mm were
compared to PSV and a positive correlation of up to R2 = 0.74 was obtained between measures of grain roundness
and PSV. Perry (1996) suggests that higher PSV samples contained a greater proportion of more circular quartz and
feldspar grains. An aggregate that contains more circular grains is more prone to having such hard grains plucked
from the softer matrix and thereby retain microtexture.

Furthermore, the research found that, as the sub-

microtexture of the grain is reduced, the microtexture as measured by PSV rises.

4.5

Measuring Macrotexture

4.5.1 Introduction
As discussed in Chapter 3, Section 3.2, the macrotexture of the road pavement is a measure of the texture of the
surface and is defined by the longitudinal wavelength between 0.5mm and 10mm.

Both macrotexture and

microtexture are needed to ensure adequate surface friction, especially at high speeds. There are three principal
methods of measuring macrotexture which involve obtaining an average surface texture depth. The three methods
are (Austroads, 2005b):

Volumetric;

Laser profile, and

Outflow.

The first two methods measure the variation in height (profile) of the surface, whilst the third method uses the
passage of water through the surface as a surrogate for the effect of macrotexture (texture depth) in removing water
from the contact patch between the tyre and the road surface. As the three methods are quite different in their
approach, they do not necessarily relate directly to one another, and the relationships vary depending upon the type
of surfacing (e.g. negative or positive texture profiles).

4.5.2 Volumetric Texture Method

The volumetric method is often called the sand circle or sand patch method. The method is based upon a known
volume of sand that is poured onto the surface and spread in a circular fashion to physically fill the hollows in the
road surface below the highest peaks. The roughly circular patch area covered is then measured in terms of its mean
diameter (generally 90mm to 250mm diameter) and the Mean Texture Depth (MTD) is calculated from the known
volume of sand and the estimate of its area. The sand (or glass spheres - used in the US) should be dry and finegrained or and the testing surface should also be dry and swept free from detritus. The test method is specified by
BS 598-105 (2000) ASTM E-965 (2001) and Transit New Zealand T/03 (1981). Transit NZ T/03 specifies using 45
101

Chapter 4
0.5ml of sand where 100% must be passed through the 600m sieve and 100% retained on the 300m sieve. The
mean texture depth for the 45ml of sand specified in Transit NZ T/03 is determined by:

MTD =
where:

57300
D2

Eq 4-7

MTD = mean texture depth (mm)

D = diameter of the sand patch (mm)

Figure 4-32: Sand patch method for measuring texture depth (A. Woodside & Woodward, 2002)
The technique, although simple, is vulnerable to operator inconsistencies and can be distorted by individual
aggregate particles forming atypically high peaks. Therefore, repeat measurements at regular intervals are required
to characterise a particular section of road. The results obtained vary from wheel path to wheel path, and also in the
areas of carriageway outside the normal wheel paths. The method is slow to execute and traffic management is
required to protect the operator. Therefore, the technique is unsuitable for routine monitoring of road surface texture
over a large road network (Austroads, 2005b).
In Japan, another variation of the method uses glass spheres but spreads them in a linear track using a spreader that
is maintained at a small fixed distance above the surface in a fixture of constant width. A further variation of the
volumetric method used by NASA is the Grease Patch Method in which the material is grease (J J Henry, 2000).

4.5.3 Laser Profile Method

For efficient and safe asset management and international standardisation purposes, it was important that a mobilemeasured value of pavement texture (macrotexture) depth be developed to replace the volumetric patch-measured
texture depth.

From this need, various non-contact sensor-measured texture depth (SMTD) methods were

developed. The most common methods that were developed use either laser or ultrasound texture meters mounted
on either stationary apparatus or vehicle profilometers for continuous in-highway vehicle operation-based texture
measurement. Well known systems were developed in the late 1980s and early 1990s by various road management
data collection specialists such as the ARAN, WDM and Pavement Management Systems profilometers.
The vehicle mounted profilometers make a sequence of displacement measurements along the line of the surface
profile. A focused laser beam is directed towards the surface and the scattered light is detected with an array of
light-sensitive cells. These are arranged in such a way that the distance between the sensor and the road surface can
102

Chapter 4
be measured. The laser pulses at high frequency and, as the sensor moves along the surface, a two dimensional
profile of the surface along a narrow line is built up. From the sequence of displacement measurements, calculations
are made to determine the Mean Profile Depth (MPD) of the texture, which is a value that represents the mean
texture depth and is specified by ASTM E1845-01 (2001). The technology of using lasers for macrotexture
measurements has been in use for over 20 years. However, few instruments have the capability of measuring the
complete spectrum of texture wavelengths, i.e. for microtexture measurements (P D Cenek, 1996). The principle of
using the laser technology to determine the macrotexture or texture depth is shown in Figure 4-33. This method is
only capable of measuring the macrotexture. It can be either vehicle mounted or used on a stationary platform.

Figure 4-33: Laser Sensors for Texture Measurements (Millard, 1993)

Whenever a road surface profile is measured by a laser based system as shown in Figure 4-34, it is divided into
length segments of 100mm, or baselines. Each baseline is further divided into two equal parts of 50mm and the
highest peak in each part determined. The two peak levels are then arithmetically averaged to give the average peak
profile level. The average profile level of the entire 100mm baseline is then calculated and subtracted from the
averaged peak profile level to give the Mean Profile Depth. This calculation is a two dimensional simulation of the
volumetric (three dimensional) sand circle method.

Figure 4-34: Definition of Mean Profile Depth

The PIARC International Harmonisation experiment (Wambold et al., 1995) compared 14 various texture methods
and concluded that very good linear correlations exist between the Mean Profile Depth method and the volumetric
patch method
It is possible to very accurately predict MTD [Mean Texture Depth] as measured by the patch
method from profile measurements. The best devices gave an unexplained variance of only
103

8-10% when porous surfaces are disregarded and around 15% when porous surfaces are

Chapter 4

included.
Every texture measuring device that participated in the PIARC experiment (Y) was correlated with all of the other
devices (X) by linear regression of the form below and reported in Appendix E of the PIARC report (Wambold et
al., 1995).

Y = A. X + B
where:

Eq 4-8

Y = a texture measuring device method such as the volumetric sand patch based mean
texture depth (MTD) in mm
X = is a sensor measured MPD texture depth such as from the WDM or ARAN
profilometers (mm)
A, B = constants for an individual texture profilometer measuring device.

The laser profile method is suitable both for low-speed surveys using portable test equipment and for routine
monitoring where a survey vehicle fitted with a laser can operate in the normal traffic flow without the need for
traffic management. Examples of various commonly used laser profile based equipment that are used in the
southern hemisphere and what operating speed mode they can operate in are:

WDM SCRIM++ (in highway speeds)

WDM T2 Texture Profiler (stationary)

Pavement Management Systems Profilometer (in highway speeds)

ARRB laser Profilometer (walking speeds)

Transit New Zealand s portable laser profilometer (stationary)

Circular Track Meter (stationary companion device to the DF Tester (ASTM E2157-01, 2001).

4.5.4 Outflow Method

This method does not measure the texture depth directly. It measures the ability of the depth and interconnected
nature of the voids in the surface to let water move through the surface (i.e. it is a measure of permeability). The
method is based upon a known volume of water, under a standard head of pressure, which is allowed to disperse
through the gaps between a circular rubber ring and the road surface. The time taken (the outflow time) is
measured. As the water can disperse very quickly in a coarse-textured surfacing, this method is only appropriate for
surfacings with low texture depth. A European standard for this method is currently under development (Austroads,
2005b). The method is especially useful with certain surfacing types, e.g. Open Graded Porous Asphalt (OGPA),
that are designed to allow water to pass into and through the body of the material. In such cases, this is a measure of
the drainage characteristic (permeability) as well as the effect of the texture depth.

4.6

Models of Skid Resistance

4.6.1 Introduction
There are several models that have been developed for the determination of pavement surface friction. Early models
were based on the skid testing devices developed such as the Sideways Force Coefficient device (SCRIM), Braking
Force Coefficient and the Polished Stone Value. The latter, whilst not strictly a model of tyre-road friction, is used
to predict in-service skid resistance in terms of the SCRIM coefficient of friction. These models were developed
104

Chapter 4
from the mid 1930s by the UK Transport Research Laboratory. By the mid 1970s the Penn State Model was
developed in the US, and was later extended by the Rado Model in the 1990s (J. J Henry, 2000). A skid resistance
model that attempted to adjust skid resistance measurements for seasonal and short-term weather related variations
was developed by Hill and Henry (1981). The PIARC World Road Association developed the PIARC model, that
utilises the general shape of the Penn State Model, and was part of an experiment to harmonise friction
measurements from different devices by the use of an International Friction Index - IFI (Wambold et al., 1995). A
more complex mathematical model was developed by Golden (1981) to determine the theoretical friction in the
contact patch. More recent advances in friction modelling have used neural networks (Owusu - Ababio, 1995). The
commonly used models, or the models that have investigated the relationship between the factors that effect skid
resistance, are described in the following section.

4.6.2 Polished Stone Value Empirical Model

As discussed in Section 4.4.2, the Polished Stone Value test was developed by the UK Transport Research
Laboratory (TRL) over many years to predict aggregate performance in terms of its resistance to polishing. The
PSV of an aggregate has been related to its in-field skid resistance as measured by the SCRIM device (SFC) by an
empirical based prediction equation that suggests PSV varies linearly with commercial traffic volumes (Q). This
equation (Eq 4-9) when originally reported by Szatkowski and Hosking (Szatkowski & Hosking, 1972) had a
correlation coefficient (r2) of 0.83, demonstrating that it predicted the variation in PSV reasonably well for those test
conditions.

SFC50 = 0.024 0.0000663.CVD + 0.10.PSV

where:

Eq 4-9

SFC50 = Sideways Force Coefficient as measured by the SCRIM device at a measuring

speed of 50km/hr or 30m/hr
CVD = the number of Commercial Vehicles (CV) per lane per day. In the UK, a CV was
defined as a vehicle exceeding 15kN mass.
PSV = Polished Stone Value

The prediction equation assumes there is a linear loss in SFC with speeds greater than 30m/hr (50km/hr) and a
fractional loss in SFC with texture depths less than 2mm. Calculating the design aggregate PSV for road surfacings
by rearranging Eq 4-9 to Eq 4-10 has been utilised effectively in the UK Design standards current HD 36/99(1999)
for many years as significant reductions in loss of control type crashes have occurred. In terms of crash records, the
UK roads are amongst the safest in Europe and the world.

PSV = SFC .T + 0.00663CVD A + B 2.4

where:

Eq 4-10

PSV = Polished Stone Value

SFC30 = Sideways Force Coefficient as measured by the SCRIM device at a measuring
speed of 30m/hr or 50km/h
CVD = the number of Commercial Vehicles (CV) per lane per day. In the UK a CV was
defined as a vehicle exceeding 15kN mass.
T= a factor depending upon traffic speed and texture depths to convert design SFC to
SFC30.
A = a factor used to reduce the required PSV when using aggregate less than 20mm size.
105

Chapter 4
B = a factor used that increases the required PSV for high stressed sites such as areas
where excessive turning and braking are likely.
However, the original research undertaken by the TRL considers only some of the factors that affect skid resistance
(refer to Chapter 3, Table 3-1). The most notable omission from the work is the effect of aggregate type (Jellie,
2003). Furthermore, Jellie (2003) states the premise upon which the empirically derived equation was developed
has changed. Positively textured hot rolled asphalt (HRA) with 20mm chippings is no longer specified on trunk
roads in the UK. Instead, negatively textured thin surfacings composed of finer aggregate are now the default
specification. The validity of the existing equation must therefore be questioned.
The relationship is an empirical predictor, developed for UK conditions for primarily straight roads. With empirical
research, care must be taken when using the relationships outside the environment within which they were
developed (Haydon, 2005). In preparing Transit New Zealand s predictive specification policy (Transit New
Zealand, 2002), a national State Highway SCRIM survey was undertaken in 1995 (WDM Ltd, 1998). The WDM
study used traffic data and surfacing records to identify traffic level, aggregate quarry source, PSV with site location
and reported a similar relationship existed in New Zealand to that in the UK (Eq. 4-11).

SFC50 = 0.018 0.0000311.CVD + 0.00637.PSV

Eq 4-11

However, this equation had a correlation coefficient (r2) of 0.28, considerably lower than the TRL model. WDM
noted the poor correlation in their model, which improved to explaining 43% of the variability when they included
the chip size grade in the prediction, although this was still significantly lower than UK findings. WDM also
investigated the NZ-SFC predicted by the TRL equation (Eq 4-9) compared to the actual SFC measured and found
that the actual values were substantially lower than the predicted values (refer Figure 4-35). The mean actual value
was only 0.453 compared to a mean predicted value of 0.562.
0.75
598 sites at equilibrium level

SCRIM Measured

0.65

0.55

0.45

0.35

0.25
0.25

0.35

0.45

0.55

0.65

0.75

SCRIM Predicted from TRL Equation

Figure 4-35: Underestimate of NZ Skid Resistance based on UK PSV Equation - sourced from (Haydon,
2005)
The possible reasons for the poor correlation of the NZ equation (Eq 4-11) and the differences from the UK TRL
equation (Eq 4-9) are summarised from (WDM Ltd, 1998) and Haydon (2005) as:

Errors in CVD numbers;

Natural variation in aggregate PSV and aggregate geological properties;

106

Chapter 4
Climatic differences between the two environments, UK study was based in south-east England. New
Zealand most probably would have shorter wet periods and more defined dry summers which may account
for higher polishing in New Zealand;

UK study data was only for straight roads, whereas the NZ data included curves and stressed sections;

Substantial differences in the UK commercial vehicle fleet in the 1970s to NZ s vehicle fleet in the late
1990s including the classification of a commercial vehicle (higher axle weights etc);

The TRL relationship is based on a different surfacing (chipped hot rolled asphalt) from that predominantly
used in New Zealand (chipseal); and

A significantly greater proportion of site category one and two locations on the NZ road network in
comparison to the UK trunk road network due to NZ s constrained topography.

WDM (1998) recommended that an increase of 5 PSV units be incorporated into the NZ skid resistance policy in
comparison to the UK equation to compensate for additional polishing. As such Transit NZ specified in TNZ T/10:
(2002) the following predictor equation:

PSV = 100 ESC50 + 0.00663CVD + 2.6

where:

Eq 4-12

PSV = Polished Stone Value

ESC50 = Sideways Force Coefficient as measured by the SCRIM device corrected for
Mean Summer SCRIM Coefficient and subsequently for Equilibrium Skid Resistance
(yearly variations) at a measuring speed of 50km/h
CVD = the number of Commercial Vehicles (CV) per lane per Day. In NZ, a CV is
defined as a vehicle exceeding 34kN mass.

The Transit NZ T/10: (2002) policy sets guidelines for assigning skid resistance levels (TNZ T/10 Appendix A) and
importantly states:
IL are not minimum standards or intervention levels. Using the IL as the initial basis sets the
actual skid resistance level of any site and the specific site characteristics can be used to raise,
lower or maintain this level.
Example site characteristics that are given as possible reasons for reducing the Investigatory Level are:

Low traffic volumes, a reduction of 0.05 off the IL is suggested when the traffic volumes are lower than 500
vechicles per lane per day; and

Up hill gradients the IL may be reduced by 0.05 if the site is on an up hill gradient of >10%. If in the
down hill direction, the uphill lane could be used for overtaking, then this may negate the justification for
the reduction.

Example site characteristics that are given as possible reasons for increasing the Investigatory Level are:

Multiple events sites with more than one event, then a 0.05 to0.1 increase may be justified;

Superelevation on curves for cases with negative superelevation >5% the IL should be increased by 0.05
or more, whilst geometric improvements are considered;

Road Roughness11 on very rough roads (> 120 NAASRA counts per km12 or IRI13 of approximately 4.9)
it is suggested that an increase of the IL of 0.05 be considered;

11
12
13

The consequences of irregularities in the longitudinal profile of a road with respect to the intended profile.
Roughness as measured by the NAASRA Roughometer, a (NAASRA) Roughness Response Meter.
International Roughness Index usually measured by Laser Profilometer.

107

Chapter 4
Isolated events causing driver surprise e.g. an isolated curve on a high speed and otherwise event-free
road may be increased from 0.05 to 0.1; and

Crashes where two or more loss of control and wet skidding crashes over a maximum of 5 year
period, the IL may be increased by 0.05 whilst consideration is given to other safety initiatives.

However, the Transit NZ T/10 relationship adjustment and guidelines do not seem to take account of:

Predominantly smaller aggregate chip sizes that are used in New Zealand;

The differences in the definition of commercial vehicles between New Zealand (34 kN) and the UK
(15kN); and

The UK equation was based upon straight roads and the extra polishing for braking and turning was
addressed by an addition of +5 PSV units. New Zealand s original study included bends in the analysis
however the adjustment for high stressed areas can only be accounted for in the Appendix A guidelines of
TNZ T/10 reactively if crashes are occurring or the site can be defined as an isolated event causing driver
surprise .

Furthermore, recent research on the TRL equation by Roe et al. (1998) show different aggregates with the same PSV
provide varying skid resistance levels in service. Roe et al. (1998) concluded:
The model on which the UK standard was based (does) not adequately reflect in-service
performance for skid resistance under present day conditions.
Aggregate from the same source delivered a range of skid resistance for the same volume of commercial vehicle
traffic, suggesting the balance between polishing and traffic is not constant (P. G. Roe & S. A. Hartshorne, 1998).
Research undertaken by Cenek et al. (2003b) found Transit NZ s T/10 PSV model predicts in-field measured skid
resistance levels very poorly, with a correlation coefficient (r2) of 0.08. This is significantly worse than the WDM
reported correlation of 0.28. Cenek et al. (2003b) notes that the T/10 model overestimates expected levels of skid
resistance. This is a significant concern as the formulated equation included sites where polishing stresses due to
braking and cornering occurr. Furthermore, the result shows that aggregates are not performing as expected and
therefore safety levels on the road network in terms of surface friction are lower than expected. When Cenek
(2003b) included seal chip grade in his analysis, the correlation coefficient (r2) improved to explaining 35 percent of
the variation, still significantly lower than the TRL s UK original 83 percent.
In summary, that the model does not consider the geological differences of the aggregates is in itself reason to
believe that the model cannot produce the required confidence that is required in terms of aggregate resistance to
polishing. Furthermore, the differences between the application of the equation and the two original studies also
indicate that a better method of prediction of the in-field skid resistance level is required.

4.6.3 The Penn State Model

The Penn State Model (Leu & Henry, 1983) describes the relationship of friction () to slip speed (S) by an
exponential function (Eq 4-13):

= 0 .e
where:

PNG
S
100

Eq 4-13

= coefficient of surface friction

0 = the intercept of friction at zero speed
108

Chapter 4
PNG = the percent normalised gradient (the speed gradient times 100 divided by the
friction) defined by:

PNG =

100 d
.
ds

Eq 4-14

S= Slip Speed (km/h)

It was demonstrated by Leu and Henry (1983) that PNG is constant with speed and therefore Eq 4-13 follows by
rearranging Eq 4-14 and integrating from S=0 to S. Furthermore, it was discovered that PNG is highly correlated
with macrotexture and that 0 can be predicted from microtexture. Later versions of the Penn State model replaced
the term [PNG/100] by a speed constant Sp:

= 0 .e
where:

S
Sp

Eq 4-15

= coefficient of surface friction

0 = the intercept of friction at zero speed (highly correlated with microtexture)
S= slip speed (km/h)
Sp = speed constant, a highly correlated regression model was developed to relate mean
profile depth (macrotexture) to Sp.

Figure 4-36 shows the Penn State Model for two cases that have the same level of friction at a slip speed of 60km/h,
but behave very differently at other speeds because of differences in texture, resulting in different values for PNG
and Sp. The example demonstrates the need for specifying more than a single value, such as the friction at 60km/h,
to describe the skid resistance of a pavement (J J Henry, 2000).

Figure 4-36: Penn State Model for two surfaces with differing textures (J J Henry, 2000)

109

Chapter 4

4.6.4 The Rado Model

As shown in Figure 4-10, the friction varies throughout the braking process and therefore is dependent upon the slip
speed of the vehicle. As a tyre proceeds from free rolling to the locked wheel condition under braking, the friction
increases from zero to a peak value and then decreases to the locked wheel condition (J J Henry, 2000). The rising
portion of the friction slip speed curve is dependent on tyre properties, whereas the portion after the peak is
dependent upon the pavement surface texture properties (macrotexture). Rado (1994) modelled this behaviour as
follows:
2

ln S

(S ) = peak .e
where:

S peak
C

Eq 4-16

(S) = coefficient of surface friction at a given slip speed (S)

peak = peak friction level
S= slip speed (km/h)
Speak = peak slip speed
C = a shape factor related to the harshness of the texture.

Figure 4-37 is a plot of Eq 4-16 with some typical values: peak = 0.6, Speak = 15km/h, C = 0.5 with a forward test
vehicle speed of 120km/h.

Figure 4-37: Rado Model plot with typical values (J J Henry, 2000)
Henry (2000) states the Rado and Penn State Models can be related to actual vehicle braking in emergency
situations (i.e. when a driver brakes heavily and attempts to lock the wheels). When the brake is first applied, the
friction follows the Rado model until the wheels are fully locked and, if braking continues after the wheels are fully
locked, then vehicle speed (which is now equal to the slip speed) decreases and the friction follows back along the
Penn State Model until the vehicle stops. If anti-lock brake systems (ABS) are used, the friction follows the Rado
Model until a predetermined slip percentage is reached. The brakes then release and the friction drops to zero and

110

Chapter 4
then the brake re-engages after the wheels spin up and the cycle is then repeated. With each successive cycle, the
Rado curve is followed for a reduced vehicle speed.

4.6.5 Seasonal and Short-Term Variation Models

As discussed in Section 3.5, Hill and Henry (1981) state that variations in surface friction from day to day,
seemingly due to rainfall pattern and local weather conditions, are superimposed on an annual cycle. Hill and Henry
(1981), as part of a three year study, attempted to determine the parameters that can be used to predict the influence
of seasonal and short-term effects on skid resistance measurements made with a locked wheel tester (SN). The study
included 21 sites in Pennsylvania and 10 sites located in North Carolina and Tennessee (FHWA Region 15). The
Penn State negative exponential model was used as a basis for relating skid resistance to speed and texture as
defined in Eq 3-12. It was found that long term variations (SNOL) in skid resistance for asphalt surfaces could be
considered to follow an exponential relationship whereas the portland cement concrete (PCC) surfaces exhibit a
linear increase in skid resistance with time. Hence, the value of SNO at any time could be expressed as:

SN 0 = SN 0 R + SN 0 L + SN 0 F
where:

Eq 4-17

SN0 = zero speed skid number intercept (a function of pavement microtexture)

SN0R = the short-term residual variation in measured skid number
SN0L = the long-term variation in measured skid number
SN0F = is a measure of SN0 that is independent of short and long-term variations.

For asphalt surfaces:

SN 0 L = SN 0 .e

Eq 4-18

Whereas, for PCC surfaces:

SN 0 L =
where:

SN 0
t
t

Eq 4-19

SN0 = change in SN0 over testing season, a function of aggregate polish susceptibility
= rate at which polishing takes place, a function of average daily traffic (ADT)
t = time in days
t = length of the testing season (days)
SN0F = is a measure of SN0 that is independent of short and long-term variations.

Hill and Henry (1981) found that the variations in PNG (a function of macrotexture) appeared to be random
measurement errors and were of sufficiently small magnitude to allow an average value to be used for each site
without altering significantly the SN0 values subsequently obtained. Therefore, the skid number at 64km/h (SN64)
would contain similar long-term and short term effects, and the value of SN64 after the removal of these effects could
be expressed as:

SN 64 F = SN 64 (SN 0 R + SN 0 L ).e(0.64 PNG )

Eq 4-20

111

Chapter 4
where:

SN64F = the skid number tested at a test speed of 64km/h and after the removal of long
term and short term variations
SN64 = the skid number measured by a locked wheel tester at 64km/h
SN0R = the short-term residual variation in measured skid number
SN0L = the long-term variation in measured skid number
-0.64PNG= a factor for the testing site based upon the macrotexture.

Hill and Henry (1981) performed multiple linear regression analysis on the short term residual variations SN0R as a
function of the effects of rainfall (Dry Spell Factor DSF discussed in Section 3.5) and pavement temperature (Tp).
Rainfall data for the Pennsylvania site were obtained from records by the Pennsylvania State University Weather
station and for the FHWA Region 15 sites from two different weather stations. Data analysis showed that with two
separate regression equations for the two regions, 35% of the short-term variation was explained for the FHWA
Region 15 site and 57% of the short-term variation was explained for the Pennsylvania sites. The regression
equations developed were:
FHWA Region 15

SN 0 R = 3.79 1.17 DSF 0.104 T p

Eq 4-21

SN 0 R = 1.88 0.77 DSF 0.15Tp

Eq 4-22

Pennsylvania sites

where:

SN0R = the short-term residual variation in measured skid number

DSF = dry spell factor refer Eq 3-2

DSF = ln(t R + 1)
where tR = number of days since the last rainfall of 2.5mm or more and an upper limit of
seven days
Tp = the pavement temperature
These equations were then incorporated with Eq 4-20 to determine the value of SN64 after adjustment for seasonal
and short-term effects (SN64F). An example of a plot of the original SN64 and the adjusted (SN64F) value is shown in
Figure 4-38

112

SN64, SN64F

Chapter 4

SN64
SN64F

TIME DAYS FROM 3-6-79

Figure 4-38: A comparison of skid resistance measurements corrected for short-term variations (Hill &
Henry, 1981)
Hill and Henry (1981) state that ideally, if the models explained all of the seasonal and short-term weather effects,
then the SN64F. line should be constant with time. The low correlation coefficients however limit the ability of
regression equations to smooth the data for short-term variations. However, the standard deviations of the
smoothed values were shown to be between 2.04 and 3.06 skid numbers for both regional sites and this should be
compared with the possible variations in SN64 due to measurement and other sources of errors. Hill and Henry
(1981) refer to Meyer et al. (1974) that list a number of factors responsible for errors in locked wheel skid resistance
test measurements (SN) and the average error band associated with each type of error. These include:
Error source factor

Average Tester Error Band (SN)

Speed Holding

1.5

Pavement variability
- Lateral

- Longitudinal

Dynamic wheel-load change

Data evaluation by operator

Hill and Henry (1981) demonstrated that models can be developed to adjust significant seasonal and short-term skid
resistance related variations explaining up to 57% of the measured variation, although this needs to be undertaken as
a minimum on a regional basis to incorporate regional weather conditions. Furthermore, the research concluded that
large variations in skid resistance measurements occur systematically over a short-term period (day to day or week
to week) although a major cause of apparent skid resistance variations is measurement error, particularly lateral
placement of the test tyre.

4.6.6 The PIARC Model and the International Friction Index

In 1992, PIARC undertook The International PIARC Experiment to compare and harmonise texture and skid
resistance measurements (Wambold et al., 1995). This involved 51 different friction and texture measurement
devices from 14 different countries on a total of 54 sites around Belgium and Spain. The PIARC Model adopted the
Penn State model as it was less complex than the Rado model. However, it adequately describes the relationship
113

Chapter 4
between slip speed and friction by a simple exponential function with the gradient of the curve controlled by
macrotexture. The PIARC model shifted the vehicle test speed intercept of the Penn State model to a standard
60km/h. The Penn State model and the Rado model are similar at slip speeds above the peak slip speed and from
this point, are dependent upon pavement texture characteristics. The form of the PIARC model is shown in Eq 423:

F ( S ) = F 60.e
where:

60 S
Sp

Eq 4-23

F(S) = friction coefficient at slip speed (S)

F60 = is the coefficient of friction at 60km/h
S= slip speed (km/h)
Sp = speed constant, determined from macrotexture.

The PIARC harmonisation experiment took measurements at three different vehicle speeds (30, 60 and 90km/h) and
each tester device made two repeated runs at each speed. The linear regression of the logarithm of the friction
coefficient against measurement speed was established for each device and the average values of friction coefficient
and speed constant were plotted to give a golden curve (Wambold et al., 1995). From this, a comparison was
made between the outputs of the different skid resistance measurement devices. The curve can also then be used to
estimate the friction coefficient at any speed.
The International Friction Index (IFI) was developed from this research to enable the harmonisation of
measurements using different testing methods and varying surface textures to be compared on a common scale. The
IFI is comprised of two numbers that describe the skid resistance of a pavement surface over a range of operating
speeds:

Sp - a speed constant, (which characterises the change in friction with change in slip speed); and

F60 - a friction constant (the harmonised estimate of friction at 60km/h).

The speed constant Sp is linearly related to the result of a macrotexture measurement (TX) and transformation
constants have been determined for each type of macrotexture measurement device used in the experiment:

S p = a + b.TX
where:

Eq 4-24

Sp = speed constant
TX= measurement of macrotexture with a specific device / method
a and b= constants for a specific macrotexture measurement device that participated in the
PIARC experiment.

The friction number (F60) is determined from a specific friction measurement device by:

F 60 = A + B.FRS .e
where:

S 60
Sp

+ C.TX

Eq 4-25

F60 = the IFI friction number

FRS = the measurement of friction by a device operating at a slip speed (S)
S= slip speed (km/h)

114

Chapter 4
Sp = speed constant, determined from Eq 3-17
A, B and C = constants determined for each friction measurement device that participated
in the PIARC experiment. Note that C is always zero for friction devices that measure
friction with smooth tread tyres.
The transformation constants developed by PIARC allow measurement of IFI with all common measurement
devices that were included in the experiment including SCRIM, GripTester, ROAR (previously known as OSCAR)
and the Dynamic Friction Tester (DFTester) machines. The ASTM Standard Practice E-1960 (1998) reports the
values of a, b, A and B for the participating devices.
The friction measurement at any slip speed, when combined with macrotexture measurements, enables the
calculation of IFI. An advantage of the IFI is that the value of F60 for a pavement will be the same regardless of the
slip speed which permits the test vehicle to operate at any safe speed. A disadvantage however, is that the
experiment produced a virtual golden curve and is therefore practically impossible to repeat. This means that any
changes to equipment or the development of new friction measurement devices can only be compared to existing
equipment that participated in the original PIARC experiment. Some machines correlated better than others to the
IFI golden curve and therefore care is needed in choosing a device against which to compare new equipment.
As discussed in Henry (2000), PIARC found that the best correlations and prediction of Sp were obtained using the
Mean Profile Depth (MPD) and have recommended that this be used whenever possible. However, good correlation
was also found between volumetric measurements such as Mean Texture Depth (MTD) calculated from the Sand
Circle Test, and this therefore remains an accepted measurement tool when more precise laser measurement
equipment is not available.

4.6.7 Goldens Model

Jellie (2003) summarises a mathematical model that was developed to predict tyre-road friction for cases of linear
braking and free rolling cornering by Golden (1981). Golden s model incorporated many of the known contributory
factors such as the effect of macrotexture, microtexture, drainage, vehicle velocity, tyre slip, contact patch geometry,
tyre pressure and load. The model idealises the road surface as a series of hemispheres representing positive texture
formed by 20mm chippings. Golden s model is complex and was based upon a positively textured hotrolled
asphalt surface with 20mm chippings that is not common in New Zealand nor now typical in the UK or Europe.
Furthermore, many of the assumptions made for the tyreroad contact area were unrealistic. It also does not take
into account the effects of elastohydrodynamic and dynamic hydroplaning and distortion in the contact patch during
braking and cornering. As such the validity of this model for predicting skid resistance on negatively textured thin
surfacings or of varied road surface types and chip sizes is questionable.

4.6.8 Neural Networks to Model Skid Resistance

An area of promising research into skid resistance is the application of neural networks. Neural networks consist of
computational models in the form of interconnected non-linear mesh-like processing elements (PE s) or nodes
capable of mimicking neurons, i.e. the basic PE s of the human brain. Neural networks have gained considerable
attention recently because of their ability to learn and deal with complex problems. They have been used in an
increasingly wide range of disciplines.
OwusuAbabio (1995) used data from 10 asphaltic concrete pavements to examine variation in skid resistance. The
purpose of the analysis was to develop and compare neural network and regression models to predict skid resistance
115

Chapter 4
for assessing future rehabilitation needs. The surface material for all of the pavements in the study was consistent,
therefore it was assumed that any variation in skid resistance was due to external factors. The main variables found
significant in explaining skid resistance using a multiple linear regression model, were the same as those found
significant using a neural network model. The difference was that the neural network was better able to fit models
to the data than the linear regression models. However, one shortcoming of using neural networks is that the steps
taken to arrive at a particular solution cannot be easily extracted from some of the neural network algorithms, which
often does not help in understanding of the relationships between the variables.
Chapter 5 that follows describes the experimental design of the research, including identifying the hypothesis to be
tested and a robust testing methodology and protocol that was developed to ensure confidence in the research
findings.

116

Chapter 5

5 THE EXPERIMENTAL DESIGN

5.1

Introduction

As described in Chapters 2, 3 and 4, after over 70 years of significant research into the factors that determine the
measured skid resistance of a road pavement surface and how this varies with time, the factors, their effects and
their inter-relationships are still little understood. The issues and variables are many and complex. There is,
however, an established and proven relationship between the level of skid resistance and loss of control crashes on
the road network. The effect of not managing skid resistance on the road network to the appropriate level is
significant in terms of the loss of life and serious injury to road users. Significant road safety benefits have been
gained over the last 20 to 30 years by the implementation of skid resistance standards that have raised the skid
resistance levels by the targeting of polish-resistant aggregates on the areas of the road network that have a higher
risk or demand for braking and cornering (an equalisation of risk policy). However, the models and asset
management procedures currently in use are not adequately predicting the variability of skid resistance as measured
on real surfaces. Further research is needed to better understand and model the variability of skid resistance over
time. The research reported herein is aimed at a better understanding of the field of skid resistance research,
including the development of alternative methods, processes and models to better manage the inherent variability of
surface friction.

5.2

The Hypothesis

This research specifically considers the seasonal and short-term variation of skid resistance over time. A review of
the literature has indicated that environmental effects such as temperature and rainfall have a significant effect on
the short-term variation of skid resistance and the longer term seasonal fluctuations. Whilst the occurrence of
rainfall is generally accepted as a reason for short-term variations in skid resistance, the mechanism by which the
variations are produced is not yet sufficiently well understood to permit reliable modelling (P. D. Cenek et al.,
1999). Others have indicated that contamination of surfaces by vehicle oil droppings and tyre detritus could be
eliminated as a cause of the variation observed in skid resistance (J.W.H. Oliver et al., 1988). Rainfall-skid
resistance functions have been developed in the US (Hill & Henry, 1981) for a specific climatic zone where the
climate and general topography is significantly different to that in New Zealand. The relationships that were
developed in the US were predominantly on asphalt mixes or Portland cement concrete surfaces and not on the thin
chipsealed surfaces that are predominant in New Zealand. The causal relationships that have been developed are not
well understood and therefore have not been incorporated into current road asset management practices. Yet, road
asset managers have great difficulty in being able to manage currently specified standards of skid resistance with
naturally sourced aggregates. This is especially so on roads with relatively high traffic volumes and especially high
volumes of heavy commercial vehicles, as well as on road sections with high demand for braking and cornering (e.g.
Transit NZ T/10 (2002) site categories 1 or 2). The effects of temperature variation, the length of time since rain,
combined with the effect of the accumulation of surface contaminants on various aggregate types is therefore still
little understood. Furthermore, current prediction methods of an aggregate s ability to resist polishing (primarily the
PSV test) have been shown to not adequately predict the long term skid resistance level that is obtained in the field
(P D Cenek et al., 2003b; P. G. Roe & S. A. Hartshorne, 1998; D. Woodward et al., 2004).
This research establishes an hypothesis and designs an experiment and testing methodology to test for statistical
significance of the effects of the seasonal and short-term variation of measured in-field skid resistance. The specific
aims of the research are to quantify:
117

Chapter 5
9) Confidence limits of measured skid resistance variation that includes seasonal and shortterm variability of skid resistance on various road surfaces;
10) Variability between measurement devices (SCRIM, GripTester, and the Dynamic Friction
Tester) on various surfaces;
11) Rate of deterioration of the measured skid resistance of newly laid chipsealed surfaces on
straight sections of road in comparison to corners, for various levels of traffic;
12) Effects of temperature on the GripTester and the Dynamic Friction Tester results;
13) Concentration and particle size distribution of the types of contaminants that accumulate on the
road surface and how they vary with time and rainfall;
14) Effects of various washing treatments on the pavement surfacing to determine whether the
presence of detritus in itself reduces measured skid resistance;
15) Effects of wet and dry accelerated polishing on prepared laboratory specimens;
16) Effects of various contaminants on skid resistance due to accelerated polishing on prepared
laboratory specimens.
The specific hypotheses that will be tested for statistical significance are:
17) That skid resistance fluctuates in a band about an equilibrium level , once initial aggregate
polishing has been completed;
18) That the relative confidence limits of the band about the equilibrium level are approximately
0.1 sfc14 units or 0.15 GN15 units from upper to lower limits;
19) That deterioration in measured skid resistance from a newly laid surface is significantly greater
on curves than on straight road sections and increases as the ratio of approach speed to the safe
curve design speed increases;
20) That the time period since the last rain has a significant effect on the short-term deterioration of
measured skid resistance, i.e. that the slope of the short term deterioration line is significantly
different from zero and approximately 0.01 sfc units (or equivalent GN) per day; and
21) That the type and classification of the contaminants that accumulate on the road surface are
significant in determining the magnitude of the short-term variation of measured skid resistance.

5.3

Research Methodology

The research methodology was designed and developed in two stages:

Stage 1: Field Tests regular skid resistance testing at a number of carefully chosen locations using a
continuous friction measurement device with the aim of quantifying the extent of the skid resistance
measurement and, where possible, controlling and/or collecting the dependent variable information at each
site location; and

Stage 2: Laboratory based polishing tests constructing chipseal surfaces and undertaking accelerated
polishing of the surfaces and periodically measuring the coefficient of friction to determine the effects of
the variation in measured skid resistance due to aggregate polishing to an equibrium level . Subsequently,
quantifying the effects of the addition of contaminants and polishing under controlled laboratory
conditions.

The sections that follow discuss the development of a pragmatic research methodology that was designed around the
significant constraints that existed for the researcher.
14

Sideways friction coefficient (sfc) is the coefficient of friction as measured by the SCRIM device.

118

5.4

Field Test Site Characteristics

Chapter 5

Particular attention was given to the selection of test sites. A field testing site matrix was designed to incorporate a
range of the most significant surface friction variables that are common in Auckland and Northland conditions and
that were thought would show variation in skid resistance measurements with time. The number of test sites and the
frequency of testing for the data collection programme were designed with statistical significance in mind. However,
the desirable number of tests required limiting to a feasible level due to the significant practical constraints of the
research programme:

Skid resistance testing in a live traffic environment requires the appropriate management of a significant
number of non-technical skid resistance considerations especially Health and Safety requirements for
temporary traffic control;

High capacity highways (i.e. most state highways closest to the Auckland CBD) require a pilot vehicle in
front of the skid resistance testing device due to their high traffic volumes and therefore were not
considered as feasible seasonal monitoring sites; and

The sites chosen for regular testing were required to be reasonably close to the University of Auckland City
campus and available for regular skid testing measurement.

The field testing site matrix required careful planning to ensure safe testing procedures whilst still including sites
with a range of the following road classifications, surface and loading characteristics (refer to Table 5-1):

Road classification and function (private road, heavy industrial and state highways);

Surface types (chipseal and asphaltic mix);

Aggregate geological type (sedimentary, igneous and artificial);

Traffic volumes and traffic loads (light to high traffic volumes and the rate of Equivalent Standard Axle
loads per day);

Polishing action (predominant direction of traffic loading, longitudinal, transverse and radial); and

Operating speeds (30km/h to 100km/h posted speed limits).

The field testing matrix shown in Table 5-1 was then further categorised and sectioned by road geometry, section
length, surface treatment, the age and source of the aggregate and its resistance to polishing including (refer toTable
5-2):

Horizontal and vertical geometry (straight and curved, flat and sloped);

A range of surface treatments and age (from newly surfaced to a polished equilibrium state and various
types of surface treatments that had no planned periodic maintenance treatments for the duration of the data
collection period);

Macrotexture levels (ranging from low to high);

Microtexture levels the aggregate source and susceptibility to polishing as measured by the Polished
Stone Value (reasonably low to reasonably high).

15

GripNumber (GN) is the coefficient of friction as measured by the GripTester.

119

Chapter 5
Site
No.

Site Name

Road Class

Surface
Type

Geological
Group &
properties

Traffic
Vols

Predominant
Polishing
Action

HCV %

Operating
Speed
(km/h)

UoA Tamaki
Campus Loop
Road

Urban
private

Asphalt
Mix

Igneous
Basalt

Light

Longitudinal

NA

20

Ports of
Auckland
(POAL)

Container
Loading area

Asphalt
Mix

Igneous
Basalt

Heavy

Transverse

Container fork
lifts /
straddlers

30

Kaiwaka
Northland

State
Highway 1

Chipseal

Artificial

8650

Longitudinal
and radial

9.6%

100

Brynderwyn
Northland

State
Highway 1

Chipseal

Sedimentary
Greywacke

8650

Longitudinal
and radial

9.6%

80-100

Hikurangi
Northland

State
Highway 1

Chipseal

Sedimentary
Greywacke

9700

Longitudinal

9.7%

100

Kara Road
Northland

State
Highway 14

Chipseal

Sedimentary
Greywacke

5500

Longitudinal
and radial

5.5%

90-100

Snooks Tatton
Northland

State
Highway 14

Chipseal

Sedimentary
Greywacke

5500

Longitudinal
and radial

5.5%

90-100

Table 5-1: Field Testing Sites - Location and General details

Site
No.

Site Name

Geometry
&
Topography

Section

Curve
Radii

TNZ
T/10
Site
Cat

Surface
Treatment

Seal Date

Macrotexture
TD
(mm)

Aggregate
Source

Approx
PSV

Length

1a

UoA
Tamaki Flat

Straight,
Level

200m

NA

NA

AM Mix 10

2002

0.5

FH Reliable
Way
Quarry

Approx
58

1b

UoA
Tamaki
Sloped

Straight,
7.2%

200m

NA

NA

AM Mix 10

2002

0.5

FH Reliable
Way
Quarry

Approx
58

POAL

Straight,
Level

750m

NA

NA

AM Mix 20

1996

0.40

Lunn Ave
Quarry

Approx
55

Kaiwaka
Northland

Curved,
Rolling

550m

200m

Chipseal
2 Coat 3/5

7/05/2003

1.7

Melter Slag
Glenbrook
Steel Mill

58

4a

Brynderwyn
2003
Northland

Curved,
Rolling

170m

160m

Chipseal
2 Coat G4

24/03/2003

2.5

Bellingham
Larmers Rd
Quarry

58

4b

Brynderwyn
2004
Northland

Curved,
Rolling

170m

160m

Chipseal
Racked in
G2/4

16/03/2004

2.5

Otaika
Quarry

51/52

Hikurangi
Northland

Straight,
Level

600m

NA

Chipseal
Single Coat
G2

20/04/1999

2.3

Otaika
Quarry

51/52

Kara Road
Northland

Curved,
Rolling

660m

150m
160m

Chipseal
Racked in
G3/5

28/01/2003

2.3

Otaika
Quarry

51/52

7a

Snooks
Tatton
Section 1

Curved,
Level

1740m

140m

Chipseal
2 Coat G
3/5

28/01/2003

2.3

Otaika
Quarry

51/52

7b

Snooks
Tatton
Section 2

Straight
Level

1740m

NA

Chipseal
2 Coat G
3/5

24/02/2000

1.9

Otaika
Quarry

51/52

7c

Snooks
Tatton
Section 3

Curved,
Level

1740m

160m

Chipseal
2 Coat G
3/5

28/01/2003

2.0

Otaika
Quarry

51/52

Table 5-2: Field Testing Sites - Surface Material characteristics, properties and Loads

120

Further requirements of the test sites were:

Chapter 5

Test section lengths should be at least 150m long;

Test sections should be homogeneous in terms of texture with no major defects such as cracks, flushing
and potholes;

The test sites should allow for safe and easy turn around near both ends to make repeated skid testing
measurement runs over the same section;

An open safe area in the vicinity of the site was required for the assembly of the test equipment, and where
possible, access to water supply and storage of a trailer with associated equipment whilst testing; and

The test sites must allow for measurements to be carried out in conditions of maximum safety. This
required either low traffic volumes or traffic control and good sight distance visibility along the test
sections.

The research programme had other considerations that required managing, including:

Availability of testing equipment to undertake the research. As this was a new area of research for the
Department, little equipment existed at the outset of the research. All equipment that was needed for the
research required successful research funding applications to be prepared, approved, ordered, shipped
and/or designed and fabricated prior to proceeding with the data collection programme. Some internal
funding for equipment required supplementing from external industry partners to enable purchase of the
new equipment;

Training - once the equipment arrived, initial training, calibration and familiarity with the testing
equipment and procedures was required prior to the programmed research;

Temporary traffic control; stationary tests required a reliance on industry partners for temporary traffic
control and lane closures due to the New Zealand Department of Labour and Transit New Zealand Health
and Safety legislation requirements for working in a live traffic environment;

Support technical staff for skid testing, as a minimum of two people are required for field tests; and

Weather conditions field testing cannot be undertaken in adverse weather conditions where rain water is
physically flowing or ponding on the surface or where the macrotexture is full of water, thereby changing
the tyre-surface contact area and pressure of the measuring wheel.

5.5

Field Test frequency

The literature review has shown that skid resistance changes over short and longer term time-frames, i.e. not only
monthly throughout a seasonal calendar year but, it is suggested, daily and weekly (Hill & Henry, 1981). To
accurately measure and thereby predict the full range of short-term variations in skid resistance would require daily
readings of skid resistance measurements. A research programme designed for skid testing measurement at daily
intervals, for a number of different physical location and sites, and for a duration of a minimum of two years for
yearly seasonal variation was seen to be both practically infeasible as well as unrealistic, due to the high economic
costs in terms of resources (human, capital, equipment and time). A pragmatic approach was therefore adopted to
collect minimum monthly field results (and more often when possible) at the selected sites. It was expected that this
approach would, over a three year data collection period, obtain results that were close to the minimum and
maximum levels measured, but would miss much of the short-term daily variation. However, the relationship of the
short-term variations in measured skid resistance with variables such as temperature, contaminant levels and rainfall
could reasonably be investigated with daily records of rainfall patterns.

121

5.6

Chapter 5

Field Test Equipment and Standard Test conditions

5.6.1 Continuous Skid Resistance (microtexture) Measurements

Continuous friction measurement testing was undertaken using a Type-D GripTester. The GripTester is a 15% fixed
slippage braked in-line wheel tester (refer to Section 4.3.3). The GripTester is towed behind a test vehicle (see
Figure 5-1) that houses a water bag / container and associated automatic pump control equipment (see Figure 5-2).
The pump delivers a computer-set water depth (usually 0.25mm) beneath the measuring tyre by varying the water
pump delivery rate corresponding to the vehicle speed. In some instances, skid resistance measurements have also
been obtained using the hand operated push-mode of the GripTester.

Figure 5-1: University of Auckland GripTester and 4WD Tow Vehicle.

GripTester Water
Delivery pipe

Computer Data
Control Box
Inlet pipe Tank

Water Pump
Outlet pipe
Tank

450 litre
Water Tank

Water Pump

Water Flow rate

control (Gate Valve)

Water Pump
Inlet pipe
Tank

Figure 5-2: GripTester Automatic Watering System and Water Storage Tank in back of testing vehicle.
122

5.6.2 Standard Skid Resistance Field Test Procedures

Chapter 5

Each test site was tested in both directions, classified as Increasing or Decreasing directions based on the Transit
New Zealand linear Route Position (RP) referencing system or by direction (north, east, south, west) for private
internal pavement testing sites. The extents of the test sections were marked on the pavement and these points were
recorded with markers by the computer operator and recorded in the GripTester computer software at the
commencement of each test run. Clear change points such as a change of seal were also marked during surveys,
which allowed confidence in the locational referencing from one test date to another and for the multiple test runs.
The following outlines the standard GripTester test parameters used for monitoring of the test sections:

Target test speed a standard test speed of 50km/h was used for all sites except for the Tamaki site where
it was not possible to travel at this speed, and a test speed of 30km/h was therefore utilised. A 50 km/h test
speed is the same target speed as adopted for the Transit New Zealand national skid resistance surveys
using the SCRIM device.

Variable speeds were also undertaken on some sites to investigate the

relationship of speed, texture and skid resistance.

Water film depth a standard 0.25mm depth of water under the measuring tyre was set for GripTester
surveys. This corresponded to a pump rate of approximately 10.5 litres per minute at a test speed of
50km/h.

Wheel path tested the Left hand Wheel Path (LWP) was tested for all standard test sites (however, a
defined longitudinal wheel track was not applicable at the Ports of Auckland site as most traffic was
transverse in direction at this site). The NZ SCRIM survey vehicle tests both the left and the right wheel
paths in one survey. The results from historical SCRIM surveys generally show the left hand wheel path to
have lower skid resistance than the right hand wheel path. Areas of deficiency have often been identified
by the annual SCRIM device surveys, where the left hand wheel path is well below the TNZ T/10
Investigatory Level (2002), whilst the right hand wheel path remains compliant.

Number of test passes per lane - five test runs per test site were undertaken. The methodology of multiple
runs over each site was adopted in order to evaluate the run to run variability and repeatability of the
GripTester. Furthermore, the adopted scheme allowed consideration of whether multiple runs physically
changed the skid resistance in terms of any surface contaminants being washed off the surface or whether
any further aggregate polishing was taking place. Henry (1996) identified in the PIARC harmonisation
study in 1992 that repeated runs on a surface were a potential source of difference between testing vehicles
as it was feared that repeat tests may polish the aggregate surface.

5.6.3 Stationary Skid Resistance Tests

On testing sites where traffic was either very light (e.g. Tamaki site) and not on public roads (Ports of Auckland site)
or where temporary traffic control and a lane closure was possible, stationary skid resistance tests were also obtained
periodically at defined points using the Dynamic Friction Tester (refer Section 4.3.5). The DF Tester is a stationary
skid resistance device that measures the frictional resistance of three rubber sliders mounted on a 284mm diameter
spinning disk. The disk is usually spun up to a normal testing speed of 80km/h and then lowered onto the test
surface and the frictional resistance is progressively determined from the torque forces generated during the spin
down of the disk. Initial tests showed that on high textured surfaces the initial test speed needed to be reduced to
60km/h, otherwise the DF Tester could catch on large aggregate chips and dynamically move during a test. The DF
Tester has been found to be very stable with time and gives highly reproducible measurements. Accordingly, it has
been chosen as the standard reference in the recently revised IFI ASTM International Standard for static devices
123

Chapter 5
(ASTM E1911-98, 2002), thereby superceding historical reliance on the British Pendulum Tester as a static device.
The DF Tester allowed seasonal and short-term variations to be considered at a specific repeatable point and a
correlation between the GripTester and the Dynamic Friction Tester to be developed. A further advantage of the DF
Tester is that it can also be used in the laboratory. A disadvantage of the DF Tester for in-field testing is that it does
not test skid resistance in the line of traffic movement due to its circular motion and the diameter of the spinning
disk, and it may therefore average out some of the variation that occurs in defined longitudinal wheel paths. Other
continuous friction measurement devices such as the GripTester, ROAR and the SCRIM have shown differences in
measurement across different lines of the traffic lane (P. D Cenek et al., 1996) and the SCRIM device has shown
differences in skid testing measurement when the same wheel path was tested in the opposing direction (J. R
Hosking & Woodford, 1976a). This disadvantage is relevant when comparing the DF Tester results with other inline longitudinal systems (e.g. GripTester, ROAR and SCRIM), but is not significant when the device is being used
as a comparative device at a specific location.

5.6.4 Macrotexture Measurements

As discussed in Section 3.2, skid resistance measurement depends upon both microtexture and macrotexture texture
profiles. However, if the surface has been appropriately constructed, macrotexture levels should not markedly
change over time or due to seasons of the year. If the surface binder and aggregate chips have been appropriately
applied, then the macrotexture levels should gradually and linearly decrease over time as the aggregate surface
slowly abrades away (attrition). If there is a significant and unexpected decrease in macrotexture over a reasonably
short time period (within a year and usually in the summer period), then it is most likely that the binder is flushing
(where excess binder rises over the chip and reduces macrotexture) or bleeding (which occurs in hot weather when
the binder is soft and it adheres to vehicle tyres and can be tracked downstream). Macrotexture levels can also
change reasonably quickly (both up and down) due to loss of aggregate chip through one of the following
processes (Transit New Zealand & Roading New Zealand, 2005):

Stripping, which occurs generally along wheelpaths, in long strips;

Attrition, in which chips are worn away by friction; and

Scabbing, which is chip loss from patches of chipseal.

If the chip loss occurs soon after construction, it is usually related to construction deficiencies and could be due to a
number of factors, including:

Low binder application rate or binder spreading temperature;

Too hard binder penetration grade;

Lack of appropriate volume of temporary cutback diluent (e.g. kerosene) or semi-permanent flux diluent
(e.g. Automotive Gas Oil AGO) and /or adhesion agent that has been applied during construction;

Lack of adhesion between the aggregate stone and the binder due to the seal application being undertaken
during adverse weather conditions or soon after, or a lack of chemical bonding, e.g due to the presence of
detritus / dust on the aggregate;

Excess traffic speeds during or soon after construction.

If the stripping occurs later in the seal life, this is more likely to be due to oxidation of the binder (embrittlement)
where the binder loses its elasticity due to repeated tension / compression wheel load cycles, causing the aggregate
to break away from the binder.
In summary, macrotexture levels when compared to microtexture levels are, if constructed and designed
appropriately, more stable, consistent and more readily predictable over time than microtexture levels. The field
124

Chapter 5
sites for this research were carefully chosen to investigate microtexture changes over time, and therefore the sites
macrotexture levels should not change markedly during the data collection period. Hence, the frequency of testing
macrotexture at each of the test sites did not need to be as regular as the microtexture measurements. Measurements
of macrotexture at the sites were undertaken approximately annually to ensure the surface macrotexture was not
affecting differences in the measured coefficient of surface friction. Macrotexture measurements were conducted on
various occasions at defined points using either the volumetric sand patch method (refer Section 4.5.2) to determine
Mean Texture Depth (MTD) as specified in Transit NZ T/3 Specification (1981), or by laser profilometer Mean
Profile Depth (MPD) method. The State Highway test sites were tested with the MPD laser profilometer method as
measured by the Transit NZ high speed SCRIM data collection programme.

5.7

Field Test Sites

A brief description of each of the seven chosen test sections follows. The test locations were split into two main
locations:

Auckland sites (refer to Figure 5-3) - two urban test sites were chosen within Auckland City. The first site
has predominantly light personal vehicle traffic as the road is an internal private road to the University of
Auckland Tamaki campus. The other site is a Ports of Auckland container loading site that has very heavy
and mostly transverse traffic movements by ship container straddlers and fork lifts. Automatic rainfall
gauge stations on site at Tamaki and nearby for the Ports of Auckland (POAL) site provided rainfall data
for analysis;

Northland sites (refer to Figure 5-4) five test sites were chosen on State Highways with more typical and
standard traffic conditions on chipseal surfaces with varied aggregates, geometrics, polishing conditions
and geographical areas. Manual rain gauges were installed on each of the five sites and these were read by
Works Infrastructure staff in Northland as often as possible, but usually daily. The manual rainfall gauge
station data were also correlated against the closest NIWA automatic rainfall gauge stations in order to
have continual daily rainfall recordings at each of the locations of the Northland sites. The NIWA rainfall
gauge station locations and their proximity to the test sites are shown on Figure 5-4.

125

Chapter 5

Ports of Auckland
Test Site

University of
Auckland Tamaki
Campus Test Site

Figure 5-3: Auckland Field Test Location Sites (source NZMS 262 Map series)

126

Chapter 5

Figure 5-4: Northland Field Test Location Sites (source ARC GIS)

5.7.1 Tamaki Campus (University of Auckland) site

The University of Auckland Tamaki Campus test site is located on an internal 15km/h speed limit campus road.
Speeds are restricted on this site due to the presence of pedestrians and other University campus and road users.
However, no stopping or parking is allowed on the test section and therefore the testing is generally not affected by
other users. The road pavement and surface was newly laid in early 2002 and comprises a 25mm thick TNZ Mix 10
Asphalt Mix. The aggregate was sourced from the Fulton Hogan Reliable Way Quarry that is reported as having a
PSV of 58. The test section was split into two sections approximately 150m each in length, one on level ground
whilst the other is on a grade of 7.2%. The extent of the test sections were painted on the road surface and traffic
cones were placed at points of differences for the placement of software markers whilst undertaking a skid test (refer
to Figure 5-5). Regular, approximately monthly, skid testing with the GripTester began on this site in July 2002.

127

Chapter 5

University of Auckland
Tamaki Campus Internal
Road site.

280mE

260mE
250mE

Skid Testing Line

Start / End points
Timeslice positions
DFTester positions

50mW
60mW

X
X

80mW

120mW

East Direction

10mE

70mE 80mE
320mW

-0.5% Grade
210mE

-1.8% Grade

Start of
New Seal

X
270mW

-7.2% Grade
110mE

X
X
250mW

220mW

West Direction
Figure 5-5: Tamaki Campus (University of Auckland) Test site

5.7.2 Ports of Auckland site

The Ports of Auckland site is located adjacent to a railway loading area at the Fergusson Container Terminal, Ports
of Auckland (known to Ports of Auckland staff as the Rail Grid). The testing strip is a 475m long, straight and level
section of asphaltic concrete (refer to Figure 5-6) with a pavement depth of 1.3m and a top surface layer of 50mm
deep TNZ Mix 20. The test site was clear of all other traffic and pedestrians during testing. This enabled a wide
range of speeds to be used and static tests to be undertaken without risk or inconvenience to other road users. Skid
testing at this site was restricted, due to Ports of Auckland Health and Safety regulations, to periods outside of
normal Port working hours. Testing was normally conducted from 4pm 6:30pm after the rail grid container
loading was completed for the day or on a Sunday. Regular approximately monthly skid testing with the GripTester
began on this site in September 2002. Due to the time of the testing and on occasions during winter months,
considerable temperature variation occured at this site between the first and last skid resistance survey runs.

128

Figure 5-6: Ports of Auckland Rail Grid site (source Auckland City Council aerial photo maps)

129

Chapter 5

Chapter 5

5.7.3 Kaiwaka Site (SH 1N)

The Kaiwaka test site is a 790m section on State Highway 1 some 62 km south of the intersection of State Highways
1 and 14 in Whangarei and immediately south of Kaiwaka township (refer to Figure 5-7). The horizontal geometry
consists of a series of three back to back reverse curves with curve radii of approximately 200m and the vertical
geometry is undulating. The section operates under a 100km/h open road speed limit and the Average Annual Daily
Traffic (AADT) volume is approximately 8650 vpd with 9.6% HCVs. The site is characterised under Transit NZ
T/10 specification (2002) as Site Category 2 as the horizontal curve radii are less than 250m. The site was
resurfaced in April 2003, as a result of the December 2002 SCRIM survey identifying the site as being deficient due
to the measured MSSC being lower than the Investigatory Level (IL) of 0.50, with some sections lower than the
Threshold Level of 0.40. The previous seal consisted of a Grade 5 voidfill with an aggregate PSV of 55. The low
macrotexture that corresponded with a Grade 5 void fill (approximately 0.5mm) was also demonstrated in the Dec
2002 SCRIM texture results and was less than the desirable minimum of 0.7mm texture for high speed highway
sections. The site was first skid tested with the GripTester on the 21st March 2003, prior to a new two-coat Grade
3/5 reseal with an artificial melter slag aggregate (sourced from SteelServ having a Transit NZ published PSV of 58
(2004)) being placed on the 7th May 2003. Regular (approximately monthly) GripTester skid measurements have
been undertaken at this site between March 2003 until June 2005.

Increasing direction

Kaiwaka Site
200m R

Whangarei
Start (00m)

170m

230m
(220m)

260m 360m
440m
Time slice position
Change of Seal
Start / End positions

590m

690m

800m
End (790m)

SH 1

Figure 5-7 : Kaiwaka Slag Site (State Highway One Northland, RP 220/9.27 220/9.87)

5.7.4 Brynderwyn Site (SH 1N)

The Brynderwyn test site is a 490m section on State Highway 1 some 53 km south of the intersection of State
Highways 1 and 14 in Whangarei and immediately south of the intersection of State Highway 1 and State Highway
12 (refer to Figure 5-8). It is located approximately 9 km north of the Kaiwaka site. In the northbound direction,
this section is the first curve that leads into a lower speed environment and hilly to mountainous climbing section of
highway. This short curve operates under a 100km/h open road speed limit and the Average Annual Daily Traffic
(AADT) volume is approximately 8650 vpd with 9.6% HCVs. The 85th percentile operating speed prior to this
curve is assessed as being approximately 90 to 100km/h, the curve leading into an extended section of constrained
horizontal and vertical geometry, with an estimated 85th percentile speed of approximately 70km/h. The horizontal
geometry consists in the northbound direction of a sharp left hand horizontal curve of 160m and the vertical
geometry is slightly rising in the northbound direction. The site is classified under Transit NZ T/10 (2002)
130

Chapter 5
specification as Site Category 2, because the horizontal curve radius is less than 250m. The site has had a
consistently poor crash history primarily due to the less than desirable horizontal geometrics at the site. The site
now has multiple layers of chipseal surfaces in an attempt to keep high friction values, although the positive effects
of a new resurfacing have been shown to quickly reduce due to microtexture polishing. The site was shown to be
consistently deficient in terms of SCRIM Surveys undertaken on the Northland PSMC 002 and consequently the
surface has been resealed twice since the beginning of the PSMC 002 contract. The previous chipseal surface was
constructed with a PSV aggregate of 58 in April 2002 and yet the site failed the SCRIM survey some six months
later. Both Transit NZ and Works Infrastructure have realised that, due to the relatively high approach speeds from
the southern direction, the long term solution to this site is to reduce the demand for friction and increase the factor
of safety by realignment and increasing the horizontal curve radii. Planning for capital works improvements is
currently progressing to this end. The site was first skid tested with the GripTester on the 21st March 2003 prior to a
new two-coat Grade 4/6 reseal, with a Bellingham aggregate having a published PSV of 56 being placed in April
2003.
The results of the regular skid testing (discussed in Section 8.4.1) also showed that the aggregate polished very
quickly, albeit combined with some bitumen flushing and loss of macrotexture to the point of becoming deficient
again by the next summer period. Works Infrastructure staff once again resurfaced this site in 2004 with a racked-in
Grade 2/4 reseal with a locally sourced Otaika sedimentary aggregate having a published PSV of 51/52 (2004).
Regular (approximately monthly) GripTester skid measurements have been undertaken at this site from March 2003
until June 2005.

N
ng
si
a
e
cr
In

c
re
di

End (500m)

160m R

n
tio

300m

490m

SH 1

Time slice position

150m

Start / End positions

Brynderwyn Site

ha
ng
ar
ei

Start (00m)

Change of Seal

Figure 5-8: Brynderwyn South Curve (State Highway One Northland RP 220/0.36 220/0.532)

5.7.5 Hikurangi Site (SH 1N)

The Hikurangi test site is a 720m section on State Highway 1 located 14.5 km north of the intersection of State
Highways 1 and 14 in Whangarei and immediately south of the Hikurangi township (refer to Figure 5-9). The
Hikurangi site was chosen as a control testing site because:

Horizontal geometrics are straight and level;

Seal age was more than five years old (last reseal - 20/04/1999 with a Grade 2 chip from Otaika quarry and
a reported PSV of 51-52); and

Macrotexture levels were stable and no flushing or other surface defects were evident.
131

Chapter 5
This site was also a SCRIM seasonal site where no periodic maintenance was planned in the short to medium term
and where no other skid resistance variation would be expected, other than seasonal environmental variations.
Previous SCRIM surveys had shown that a reaonably consistent level of SFC had been reported and therefore the
surface aggregate could be considered as being in an equilibrium level of polish. The section operates under a
100km/h open road speed limit and the Average Annual Daily Traffic (AADT) volume in 2004 was approximately
9700 vpd with 9.7% HCVs.

Increasing direction
Hikurangi Site

SH 1
To Whangarei

X
160m

X
300m

End (720m)

440m

Start (00m)

Time slice position

Change of Seal

SH 1

Start / End positions

Figure 5-9 : Hikurangi Site (State Highway One Northland RP 144/6.215 144/6.766)

5.7.6 Kara Road Site (SH 14)

The Kara Road site is located on State Highway 14 approximately 8.5 km west of the intersection of State Highway
1 and 14 in Whangarei (refer Figure 5-10). The site consists of several TNZ Site Category 2 horizontal corners with
curve radii ranging from 150 to 160m. The site also had a previous history of SCRIM deficiency. The test section
included a reasonable down grade in the westerly (increasing) direction of -7%. The section also included the
intersection of Kara Road that intersects with State Highway 14 at almost the point of intersection (PI) of the 150m
radius curve. The site was resealed in January 2003 with a racked-in seal using a Grade 3/5 Otaika quarry
aggregate. The reason for the reseal was that the April 2003 SCRIM survey showed some sections of the left wheel
path had deteriorated below the TNZ T/10 Intervention level, although the average between the two wheel paths
remained compliant with TNZ T/10. The section operates under a 100km/h open road speed limit although the 85%
percentile operating speed is estimated to be around 90km/h, due to the constrained horizontal and vertical
geometrics. The Average Annual Daily Traffic (AADT) volume in 2004 on this section of State Highway 14 was
approximately 5500 vpd with 5.5% HCVs. Kara Road has very low traffic volumes. The site has been monitored
(approximately monthly) with GripTester skid measurements from March 2003 until May 2005.

132

Chapter 5
Start (00m)
40m

Kara Road Site

To Whangarei

Time slice position

n
io
ct
Se

Change of Seal

220m

1S
l
ea

150m R

120m

Start / End positions

Incre
asing
direc
tion

450m

520m

2S
eal

End (660m)

600m 650m
Se
ctio
n

160m R

SH 14

Figure 5-10 : Kara Road Site (State Highway 14 Northland RP 0/8.442 to 0/9.099)

5.7.7 Snooks - Tatton Site (SH 14)

The Snooks - Tatton testing site is also located on State Highway 14 approximately 13.9 km west of the intersection
of State Highway 14 and State Highway 1 in Whangarei and 5.5 km west of the Kara Road test site (refer to Figure
5-11). It is the longest test site of all of the test sections (1700m) and it contains three TNZ T10 Site Category 2
corners. The test section is made up of two sections of new seal separated by a 300m section of older seal in
between. The two newly resealed sections were sealed with a two-coat reseal with a Grade 3/5 Otaika quarry
aggregate (PSV 51/52). The traffic volume (AADT) in 2004 is recorded in the RAMM database as being the same
as the Kara Road site, being approximately 5500 vpd with 5.5% HCV. The section operates under a 100 km/h open
road speed limit with a relatively level vertical alignment, and therefore the 85% percentile operating speed is
estimated to be around 100 km/h. The traffic volumes on both Snooks Road and Tatton Road are relatively minor.
The site has been monitored (approximately monthly) with GripTester skid measurements from March 2003 until
May 2005.
New Seal Ends
@ 1720m

1500m

Change of Seal

200m

1650m

Time slice position

Snooks Road

Increasing direction

Se
al

SH 14

620m

700m

920m
1000m

1170m

Se
ct
io
n

520m

3:
Ne
w

350m

140m R
Section 1: New Seal

l
Section 2: Old Sea

160m R
Tatton Road

Snooks / Tatton Site

Figure 5-11 : Snooks - Tatton Site (State Highway 14 Northland RP 0/13.91 to 15/0.647)

133

5.8

Surface Detritus and Skid Resistance

Chapter 5

5.8.1 Introduction
As discussed in Section 3.5.6, it has been thought that the accumulation of finer particle detritus material contributes
towards the polishing of the aggregate, usually coinciding with lower rainfall summer months. Conversely, the
coarse grit contributes to abrasion of the aggregate and, when the finer material has been washed away in wetter
months, helps to rejuvenate skid resistance (Jayawickrama & Thomas, 1998).
Road based detritus has also been reported as being a major contributor to stormwater pollution as the road dusts
contain a variety of suspended and dissolved inorganic and organic contaminants (Sansalone & Buchberger, 1997).
The amount of detritus accumulation on road surfaces is dependent on the traffic, road use, and local environment
conditions. Local environment conditions include such factors as (Onwumere, 2000; Stotz, 1987):

Rainfall frequency and intensity;

Wind;

Preceeding dry days;

Total volume of runoff discharged;

Drainage systems; and

Contributing catchment characteristics, such as the surrounding geology, areas of permeable and
impermeability.

In particular, rainfall volume appears to be the major factor in the removal of total road dust detritus, and rainfall
frequency mainly affects the accumulation of particle-bound contaminants (O'
Riley et al., 2002). In order to
substantiate or disprove the effect of detritus on the variation of measured skid resistance, procedures were
developed to quantify in-field pavement surface detritus that lodge within the surface macrotexture. The main
objectives of the detritus collection and laboratory analysis were the:

Quantification of the detritus that accumulates on the road surface; and

Classification of the detritus in terms of particle size distribution and its material characteristics i.e.
suspended solids (sediments), heavy metals (i.e. copper, lead, zinc and cadmium) organic and petroleum
hydrocarbons).

A secondary objective was, a consideration of the likely source of the detritus and its effects in terms of the
receiving environment in comparison to other international studies.
The expectation was that the longer the period of no rainfall, the greater the volume of detritus and therefore the
greater the effect of finer material polishing the microtexture of the surface, thereby reducing measured skid
resistance.

5.8.2 Detritus Sample Collection Method

A significant constraint on the number of collected samples that were possible, was the requirement to close the road
traffic lane and provide temporary traffic control whilst obtaining samples. The detritus on the road surface was
sampled by a brushing of the target test areas with deionised water and collecting the liquid using a high power
vacuum. The vacuuming technique has been reported as an efficient method of collecting solid particles from the
road surface whilst preserving their physical and chemical characteristics (W.H Ng et al., 2003).

134

Chapter 5
The test site locations were at each of the field sites immediately over the left wheel path (LWP). The test area of
each sampling site was one square metre limited by a 1m x 1m wooden frame. The gaps between the edge of the
frame and the macroxtexture of the road pavement surface were sealed with a foam sponge to prevent leakage of the
sample detritus. The following procedures were developed for the collection and storing of the surface detritus for
subsequent laboratory analysis (refer to Figure 5-12):

Placing a 1m x 1m wooden frame with sponge over the target surface sample area;

Standing on the wooden frame in order to compress the foam sponge into the macrotexture of the pavement
surface, spreading one litre of deionised water evenly over the target area;

With a relatively vigorous but careful action, brushing the sample area with a hard nylon scrubbing brush to
mobilise the detritus within the macrotexture area;

Collecting the mobilised detritus and liquid mixture from within the timber frame by the use of a high
powered wet and dry vacuum cleaner, ensuring that the edges of the sponge on the bottom of the wooden
frame are also vacuumed;

Using an additional 250 mls of deionised water through the vacuum cleaner hose and machine to wash any
detritus sticking to the inner surfaces of the vacuum;

Taking the sampled mixture from the vacuum cleaner and storing it in a labelled plastic container; and

Storing all samples in a temperature controlled-refridgerator until laboratory analysis.

Containers of
deionised water
1m x 1m wooden
frame with sponge
for sample area

Power
generator

Wet & dry vacuum

(2.2 kW)
Detritus sample
storage container

Water measuring
container

Scrubbing brush for

loosening contaminant
detritus

Figure 5-12: Pavement surface detritus sample collection method.

5.8.3 Detritus Laboratory Analysis Methods

The detritus samples obtained by the method described in Section 5.8.2 were then separated into water and sediment
fractions by centrifugation and air-drying. Samples were analysed for suspended solids, particle size distribution,
(PSD), heavy metals (i.e., copper, lead, zinc and cadmium), total petroleum hydrocarbons (TPH) and total organic
carbon (TOC). All laboratory tests were undertaken by the Environmental Engineering Laboratory at the University
of Auckland except for the PSD analysis, which was conducted by Malvern Instruments Ltd.
Suspended solids were determined using the gravimetric method outlined in section #8158 in HACH (1997), and the
heavy metals extraction by the standard nitric acid digestion method outlined in section 3030E in American Public
135

Chapter 5
Health Association et al. (1992). The concentrations of copper, lead, zinc and cadmium were measured by a Varian
SpectrAA Atomic Absorption Spectrometer equipped with an atomiser-burner. The measurement range of the
instrument was 0 to 50 ppb, and as such, the digested samples required dilution using deionised water prior to
testing.
The laboratory testing for TPH was undertaken according to the methodology provided by TPH Working
Group(1998). The samples were mixed with high purity reagent grade n-pentane in a 1:1 volume ratio, vortexed for
5 minutes, and then the extracts were filtered through a 0.2 m nylon filter membrane (Phenomenex AFO-0501)
prior to the TPH determination. The analysis instrument was a gas chromatograph (GC) using a Hewlett-Packard
HP 6890 series GC system equipped with an Alltech Econo-cap EC-1000 column and a flame ionisation detector.
The total organic matter, including TPHs, in the samples was measured using TOC, calculated as the difference of
the measured total carbon and total inorganic carbon values. Both dissolved and insoluble forms of TOC were
analysed by a Shimadzu model TOC-VCSH Total Organic Carbon Analyser with a Solid Sample Module SSM5000A. Specifically, the dissolved form of TOC was quantified based on units of TOC mass per unit of collected
fluid (mg-TOC/L-water), whilst the insoluble form of TOC content was based on units of TOC mass per unit of
associated sediment mass (mg-TOC/g-sediments).

5.9

Laboratory Based Polishing Methodology

5.9.1 Introduction
A primary objective of the research was a better understanding of the factors that cause the significant changes in
skid resistance due to microtexture changes over time. The literature reviewed in Chapters 3 and 4, especially Hill
and Henry (1981), and early confirmation during the first year of field testing of the research clearly demonstrates
that significant short-term variation in skid resistance occurs. This largely unpredictable variation could not be
explained by a yearly seasonal pattern alone and it became clear that other factors were affecting the variation in
measured skid resistance. This initial knowledge led to the formulation of the following research questions:

What causes the variability in skid resistance measurement?

What is the specific mechanism for replenishment of the aggregate chip microtexture and over what period
of time can this occur?

How would a road asset manager know when the chip had reached its worn-out state and could this be
predicted and or improved?

What surfacing treatments can last the desired asset life-cycle?

If these questions could be understood and answered by road asset managers, improved skid resistance asset
management would occur, enabling better decision making and thereby potentially significantly improved road
safety. The literature review and the initial field data collection demonstrated the difficulty in understanding the
complex inter-relationships of skid resistance variables at any point in time. Too many in-field variables existed that
could be neither controlled nor monitored closely enough to be understood. It was therefore apparent that controlled
laboratory-based experiments were required to simulate the aggregates and road surface texture, traffic loading, and
environment conditions.

The experiments needed to simulate certain conditions whilst other variables were

controlled to isolate their effects. This section discusses laboratory-based experiments that were designed to
simulate the rate of recovery of measured skid resistance by the replication of equivalent in-field traffic and
surfacing conditions with the control of various environment factors.

136

Chapter 5

5.9.2 The experimental variables

The experiments required the control and simulation of the effects of the following variables:

Road pavement surfacing types, utilising the same materials used in practice;

Temperature control;

Traffic action simulating heavy commercial vehicle polishing effects;

Contaminant loading;

Rainfall / washing cycles; and

A method of repeatedly testing the coefficient of friction in the laboratory on prepared specimens.

The experiments required laboratory testing equipment and surfacing samples to be constructed that were
compatible with each other. As the Department of Civil and Environmental Engineering had recently jointly
purchased with Pavement Management Services Limited a stationary skid tester (The Dynamic Friction Tester, refer
to Section 4.3.5 and also Figure 5-13 and Figure 5-14), this became the critical factor that determined most of the
other experimental variables.

Figure 5-13: The UoA DFTester on a Prepared

Sample

Figure 5-14: Rubber Sliders and Rotating Disk of

the UoA DFTester

A laboratory experiment was designed to simulate the variation of skid resistance by the change in microtexture
observed for a typical range of chip-seal aggregates used in New Zealand. The aggregates tested were specifically
to include (with others) local aggregates used by Works Infrastructure Limited in the Northland Region on the
PSMC 002 contract. The aggregate chip sample variables that were initially considered for the experimental design
were:

Aggregate source and reported resistance to polishing as measured by the PSV test;

Chip size (i.e. to determine surface macrotexture), and

Traffic loading.

Due to the considerable time involved, both in the preparation of samples and in the accelerated polishing and skid
resistance testing, it was decided to vary only one of the test variables whilst keeping the other variables constant.
As such, the experiment was designed to test various aggregate sources and geological types, from a range of high to
low reported values of resistance to polishing (as measured by the PSV test). A similar aggregate chip grading size
was used, being a TNZ Grade 4 chip that was further sieved through a 9.5mm sieve in an attempt to deliver a similar
macrotexture for the prepared samples. The effect of varying texture on skid resistance measurement have been
well reported by Oliver (2003a), Wilson et al. (2003), the PIARC Harmonisation project (Wambold et al., 1995) and
137

Chapter 5
PIARC (2003) and was therefore seen as a non-essential component of the controlled laboratory experiments. The
simulated traffic loading (57kg over three wheels) was also kept at a constant throughout the accelerated polishing
process.

5.9.3 Laboratory Surface Samples

A stable and robust method of constructing a chip seal surface sample large enough to be polished and then tested by
the DF Tester was required. A number of trial construction methods were attempted before settling upon a surface
sample method prepared by the following steps :
Step 1

Grade 4 aggregate samples were sieved through a 9.5mm sieve, with any flaky aggregate either sieved
out with a slotted sieve and / or rejected by hand.

Step 2

A sheet of glass was prepared with a thin layer of oil and sand (sieved through a 600 m sieve) to hand
place the aggregate chips. The oil and sand was intended to prevent the sand / cement mix (to be later
poured on the aggregate sample to bind the aggregate surface together) from sticking onto the glass.

Step 3

Roughly cubically shaped aggregate stones were selected from the product in Step 1 and then hand
placed onto the glass one by one within a 450mm diameter circle ensuring that each stone was placed
with minimal gaps and with the flattest side of the stone faced towards the glass. This method of
placement was used to ensure that the surface sample had a texture value comparable to fieldspreading and rolling of aggregate chips onto their Average Least Dimension (ALD).

Step 4

Once the hand-placed aggregate chips completely filled the 450 mm diameter circle, the textural
spaces between each aggregate were filled and sprayed evenly with approximately 2 mm depth of sand
that had been sieved through a 600 m sieve. The fine sands were used to prevent the texture being
completely filled by the mortar and avoid the concrete sticking onto the glass. This procedure was
used to simulate the in-field chip seal surface binder reaching approximately 0.7 x ALD after initial
rolling and subsequent traffic action.

Step 5

The surface sample was then bound together by the mixing of a 1:1 mixture of sand and cement that
was reasonably fluid and was poured over the prepared sample, taking care not to move any of the
aggregate chips.

Step 6

Once the mix had semi hardened (approximately one to two days later), the bound sample of aggregate
chips and sand / cement mortar was turned over and the excess sand brushed off the surface. The
sample was then placed with the aggregate chips facing upwards in a mix of 1:1 sand and cement
mortar mix into a prepared 540 x 540mm (internal dimension) timber frame mould (for lateral edge
support) mounted on 18mm thick plywood.

Step 7

The timber frame and surface sample was then labelled and numbered before any polishing or testing
was undertaken. The sample was then left for a minimum of a week for the sand /cement mortar mix
to be largely cured.

Two samples of each aggregate source were constructed, one to be a control sample that was not polished and the
other sample polished by the accelerated polishing machine. Figure 5-15(a) shows a sample being prepared by hand
with the placing of the aggregate chips on a sheet of glass in a mixture of sand and oil. Figure 5-15(b) shows a
sample that is upside down and immediately prior to being fixed in a sand / cement mortar mix. Figure 5-15(c)
shows a completed sample after being set and cured by the sand / mortar mix into the wooden frame and ready for
accelerated polishing and skid resistance testing.

138

Chapter 5

(a) Placing the aggregate chips by

hand.

(b) Sample ready for setting with

sand / cement mortar.

(c) A completed sample mounted in a

wooden frame and ready for testing.

Figure 5-15: Preparation of the Laboratory Surface Samples

Four aggregate sources were chosen from a range of geological types for the building of the laboratory surface
samples:

Moutohora sedimentary greywacke, being the highest reported natural PSV in the North Island (Napier);

Holcim igneous basalt, mid range reported PSV, sourced from Auckland Region (Bombay);

Otaika sedimentary greywacke, lowest reported PSV acceptable for Transit NZ sufacings, locally sourced
from Northland; and

Melter slag, artificial by-product from Glenbrook Steel mill, through SteelServ Limited.

5.9.4 Laboratory-based Texture Measurement

In terms of the laboratory experiment, it was hoped that the method of construction of the surface sample would
prevent changes in macrotexture. The experiment was designed to determine the causal effects of the variation in
microtexture. Therefore, macrotexture measurements were only required at the beginning, prior to polishing, and on
completion of the tests, to ensure that during the experiment macrotexture did not effectively change. Ideally, the
macrotexture could have been tested utilising the Circular Texture Meter (ASTM E2157-01, 2001), which is a
companion apparatus to the DFTester that measures the two dimensional texture of the surface by laser. The
Circular Texture Meter features the same circular diameter as the DFTester and gives Mean Profile Depth (MPD).
However, this apparatus was unavailable to the research and therefore the volumetric texture depth was measured by
the Sand Circle Test as set out in TNZ specification T/03 (1981). This method determines the mean texture depth
(MTD) at the centre of a surface sample by measuring the diameter of the circle formed when a known quantity of
sand (50ml) to a specified grading is spread evenly over the surface. The PIARC Harmonisation experiment
(Wambold et al., 1995) demonstrated that it was possible to very accurately predict MTD as measured by the
volumetric patch method from profile measurements using laser profilometers (MPD). The best devices gave an
unexplained variance of only 8-10% when porous surfaces are disregarded and around 15% when porous surfaces
are included.

5.9.5 An Accelerated Polishing Machine

To enable the simulation of the approximately seasonal skid resistance effect required a method of simulating
heavy vehicle traffic action on prepared chip seal surfacing samples. Furthermore, this variation occurs only after
an equilibrium level of skid resistance (ESR) has been reached. This depends upon the:

Resistance of aggregate to polishing, traditionally measured by the Polished Stone Value (PSV) test, and
139

Chapter 5
Number of passes of heavy commercial vehicles.

An international literature review was undertaken to consider various options for polishing, testing and analysing
pavement surface frictional properties in the laboratory. A laboratory device was required to initially polish a
prepared surface sample to ESR. The device also needed to be compatible with the DF Tester and therefore polish
on a circular track at the same diameter as the DF Tester rotating disk, i.e. 284mm diameter to the centre of the
rubber sliders. Testing methods considered in the review were:

The accelerated polishing machine used in the PSV test in conjunction with the British Pendulum Tester
(AS 1141.40-1999, 1999);

A scuffing wheel apparatus developed by the TRL for determining the resistance to wear by scuffiing of
high friction surfaces at elevated temperatures, Appendix G (Nicholls, 1997);

A road machine apparatus (with standard car tyres) developed by the TRL for determining the resistance
to wear by repeated turning wheels of a high friction surface at low temperatures, refer Appendix H
(Nicholls, 1997); and

National Centre for Asphalt Technology (NCAT) slab polishing machine developed in the US to
specifically wear samples at an appropriate diameter for the DFTester (McDaniel & Coree, 2003).

A number of devices, such as the accelerated polishing machine used in the PSV test, were ruled out due to the
surface sample size required by the use of the DFTester and / or the expense and timeframe required to develop and
assemble the apparatus.
A decision was made to proceed with developing a device similar to the NCAT slab polishing machine method.
This device was developed specifically for use with the DFTester and polished the prepared sample in the same
circular motion as the DFTester. This was seen to be of benefit as it would simulate in-field traffic action that
occurs in the wheel paths. The developers of the NCAT device and Purdue University, who developed a similar
machine, were contacted and permission was gained to design and build an accelerated polishing machine based
upon the principles of the NCAT device. Some modifications were made to the mechanical operation to better
control the loads on the wheels whilst also taking account of the higher macrotexture and irregularities of chip seal
samples common in New Zealand. This is in contrast to the concrete or asphalt mix samples that the NCAT
apparatus was developed for in the USA.
The main features of the accelerated polishing machine developed as part of this research (refer to Figure 5-16) are:

Three castor wheels with rubber tyres, at a tyre pressure of 20 psi and rotating on a diameter of
approximately 284mm;

A loaded wheel assembly weight of approximately 57kg over the three wheels;

A variable drive electric motor and gearing to enable a maximum speed of 47 rpm at a motor speed of 50
Hz.

140

Chapter 5

Lifting frame
Electric motor, gear
box and belt drive

Safety cage

Drive shaft

Load weights

3 rotating pneumatic
castor polishing wheels

Water Delivery
pipes
Test sample

Figure 5-16: Schematic Front Elevation of the Accelerated Polishing Machine

Figure 5-17 and Figure 5-18 show the accelerated polishing machine and the wheel assembly unit that was
manufactured for the accelerated polishing of chip seal surfaces, during testing operation on a prepared sample.

Figure 5-17: The Accelerated Polishing Machine

(left) with DF Tester (right)

Figure 5-18: The Accelerated Polishing Machine

wheel assembly unit in operation

To examine and simulate the approximately seasonal variation of measured skid resistance required two stages of
laboratory testing, being:

Stage 1: Polishing the prepared surface samples to Equilibrium Skid Resistance (ESR) level; and

Stage 2: Simulating the cyclical effects of variation of the summer and winter polishing, rejuvenation of
surface samples through the effects of contaminants, rainfall and vehicle trafficking.

141

Chapter 5

Skid Resistance Value

Initial Roughening

Polishing Phase

STAGE ONE

Equilibrium Phase

STAGE TWO
Seasonal Variations

Winter

Summer
After 1M standard Axles or
approximatley
2 years
Duration
of Polishing

Figure 5-19: Stages of Accelerated Polishing modified from Prowell et al. (2003)

5.9.6 Laboratory Procedures for Polishing and Skid Resistance Measurement

A procedure was developed to polish the prepared surface samples to an equilibrium level whilst measuring the
variation of the coefficient of friction on both the un-polished and the polished sample. The following steps were
used to test the samples during the Stage One accelerated polishing test phase:
Step 1

Record the ambient temperature with a mercury thermometer and the surface temperature with an
infrared thermometer prior to every test of the Dynamic Friction Tester.

Step 2

Lightly clean the prepared samples with water to remove any dust or other detritus.

Step 3

Test the paired samples for the coefficient of friction of both the un-polished and the to-be polished
sample three times each using the DF Tester at an initial starting speed of 60km/h.

Step 4

Polish the sample marked polished for 15 minutes (or more regularly if the change in skid resistance
is rapid) with the accelerated polishing machine with continual watering and no addition of
contaminants.

Step 5

Lightly clean the prepared samples with water to remove any dust or other detritus and test both
samples with the DF Tester.

Step 6

Repeat Step 4 for another 15mins and once again repeat Step 5.

Step 7

Continue polishing the polished sample at the following time intervals or until ESR is reached
(0mins, 15mins, 30mins, 45mins, 60mins, 90mins, 120mins, 3hrs, 4hrs, 5hrs, 6hrs and 7hrs if
required).

Once each of the polished surface samples had clearly reached an equilibrium level for that specific aggregate,
load and polishing action, the samples were ready for the Stage Two polishing phase.

The Stage Two polishing

phase was designed to determine the effect on the variation of the coefficient of friction when certain specific
additives were placed upon the surface with the addition of polishing. This was designed to simulate, in the
laboratory, the effect of traffic, environmental effects and surface detritus on the coefficient of friction on a surface
that had polished to an equilibrium level . The polishing action and the method of wetting and drying of the
surface, and all other known variables (where possible) were held constant to isolate the effect of the specific
additive.
142

Chapter 5
The theory of the Phase Two variation in measured skid resistance (known as the seasonal variation ) as described
by Jayawickrama and Thomas (1998) and discussed in Section 3.5.6, has been that the accumulation of finer particle
detritus material contributes towards the polishing of the aggregates (reducing skid resistance) whereas the coarse
grit contributes to abrasion of the aggregate which helps to rejuvenate skid resistance. Cycles of rainfall wash away
finer material, therefore allowing the grit to rejuvenate skid resistance during wetter periods. During drier periods,
the finer material accumulates, polishing the aggregate and reducing skid resistance.
To test this theory, a number of additives were considered for addition to the laboratory surface samples combined
with accelerated polishing to measure the effect on measured skid resistance. The samples additives that were
chosen were similar to the detritus particle size distribution expected to accumulate in the field and were:

Oedometer clay a soft but well graded material with particle size distribution ranging from approximately
d(0.1)16 = 2 m to d(0.9)17 of 1190

m and a mean size d(0.5)18 of 550

m.

Oedometer clay is

predominantly a soft material (kaolinite) and is strongly anistropic in terms of its properties, with a Mohs
hardness of 2-2.5;

Emery powder - a fine but very hard material with particle size distribution ranging from approximately
d(0.1) = 2.5 m to d(0.9) of 100 m and a mean size d(0.5) of 8.0 m. Emery powder is derived
predominantly from corundum minerals (Al203) with a Mohs hardness of 9. Emery powder is used in the
PSV test machine as a polishing medium;

Leighton Buzzard sand - a coarse and hard material with particle size distribution ranging from
approximately d(0.1) = 600 m to d(0.9) of 1265 m and a mean size d(0.5) of 860 m. Leighton Buzzard
sand is predominantly from quartz minerals with a Mohs hardness of 7.

The Phase Two polishing consisted of measuring the effect of the following variations, in turn, to the Phase One wet
polishing phase:

Dry polishing with the addition of Oedometer clay (10 grams);

Dry polishing with the addition of sieved Oedometer clay (10 grams) with particle sizes of <0.15 mm;

Dry polishing with the addition of sieved Oedometer clay (10 grams) with particle sizes of >1.15 mm;

Dry polishing with the addition of Emery Powder (10 grams);

Dry polishing with the addition of Leighton Buzzard sand (20 grams); and

Wetted surface (damp) polishing with the addition of Oedometer clay (10 grams).

After each polishing variation had been tested, the surface sample was then polished, wet but with no additives, to
try to restore the sample to its equilibrium skid resistance level .
The following Stage Two testing phase procedure was developed to test each of the polished samples for the
variations outlined above. The surface sample was assumed to be at an equilibrium level of skid resistance prior to
this procedure being undertaken.
Step 1

Record the ambient temperature with a mercury thermometer and the surface temperature with an
infrared thermometer prior to every test of the Dynamic Friction Tester.

Step 2

Lightly clean the prepared samples with water to remove any dust or other detritus.

Step 3

Test the paired samples for the coefficient of friction of both the un-polished and the to-be polished
sample three times each using the DF Tester at an initial starting speed of 60km/h.

16

d(0.1) denotes the ten percentile diameter of the sample particle size distribution.
d(0.9) denotes the ninety percentile diameter of the sample particle size distribution.
18
d(0.5) denotes the fifty percentile diameter or mean of the sample particle size distribution.
17

143

Chapter 5
Step 4

Allow the polished sample to dry.

Step 5

Evenly distribute the prescribed amount of additive described above to the polished sample around
the circular polishing diameter of the polishing machine.

Step 6

Dry polish the sample marked polished for 10 minutes with the accelerated polishing machine
without any water addition.

Step 7

Lightly clean the polished sample with water to remove the additive material.

Step 8

Test the paired samples for the coefficient of friction of both the un-polished and the polished sample
three times each using the DF Tester at an initial starting speed of 60km/h.

Step 9

Dry the polished sample either in an oven or overnight and repeat Steps 6-8 for as many times as
required until a new equilibrium level has been established.

Step 10

Wet Polish the sample marked polished for 15 minutes with the accelerated polishing machine with
continual watering and no addition of contaminants.

Step 11

Lightly clean the prepared samples with water to remove any dust or other detritus and test both
samples with the DF Tester.

Step 12

Repeat Step 10 for another 15mins and once again repeat Step 11.

Step 13

Continue wet polishing the polished sample every 15mins until the same or new equilibrium level
is reached.

All four aggregate samples were tested by this procedure and the DF Tester coefficient of friction for each test was
immediately logged in a laboratory test book. Later the test data were downloaded from the DF Tester controller,
processed and further descriptive statistical analyses were undertaken. The results of these tests are shown and
discussed in Chapter 10.
Chapter 6 discusses the skid resistance field results.

144

Chapter 6

6 SKID RESISTANCE FIELD TESTING RESULTS AND DISCUSSION

6.1

Introduction

This chapter discusses the results of skid resistance measurements undertaken at the field sites shown in Section 5.7.
Each of the sites had regular minimum monthly skid resistance measurements undertaken utilising the GripTester
device along with the collection of rainfall and temperature data. Less frequently, additional data such as the
macrotexture levels, traffic volume and composition data, SCRIM SFC data and DF Tester results were obtained
where this was possible. The regular skid resistance measurement results, as with specific device repeatibility and
reliability tests, contaminant before and after tests and their subsequent data analyses are discussed in Chapter 7.

6.2

Standard Skid Resistance Field Testing Procedures

6.2.1 Introduction
Like most skid resistance testing devices, the GripTester and Dynamic Friction Tester used in this research are
sensitive pieces of equipment that rely heavily upon electronic sensor equipment. The sensors measure very small
changes in vertical, horizontal or torque forces (i.e the hysteretic and adhesion forces that combine to develop
surface friction between the rubber / surface contact area). However, the devices are required to be transported to a
site (usually as part of or on a trailer, or in the back of a vehicle), operated in harsh and rough environments, are
towed at high speeds whilst measuring (or have parts that move at high speeds) and require testing under wet surface
conditions. As a consequence, the measuring devices and associated equipment require regular maintenance and
strict calibration procedures to be developed and followed to ensure the integrity of the resulting data. These
operating conditions have an effect on the reliability and repeatibiltity of an individual device to measure a specific
surface coefficient of friction.
There are however, other sampled data integrity criteria that need to be considered when choosing a specific device
and measurement methodology, such as:

Relevance of the method to deliver a result that provides confidence in that result; and

Affordability of the collected sample data.

These data issue criteria are discussed in detail by the author and colleagues in Wilson et al. (2002), Henning et al.
(2000) and Paterson and Scullion (1990) and will not be discussed further here other than when directly related to
the equipment device in question.
In order to help minimise the component of variation of the measured data result due to the specific equipment
devices, standard field testing procedures were developed for both the continuous GripTester skid resistancesurveys
and the stationary Dynamic Friction Tester (DF Tester) tests. The procedures were based on the specific operating
and maintenance procedures supplied by the device manufacturers, the associated equipment setup and other on-site
safety requirements. The sections that follow summarise the procedures that were specifically developed for field
testing using the GripTester and the Dynamic Friction Tester, respectively.
Figure 6-1 below summarises the steps that were developed for the skid testing of the field sites described in Section
5.7 to ensure consistency in procedures for field testing using the GripTester and the DF Tester, to minimise any
variation due to device error.

145

Check, maintain and calibrate

equipment device

Load, transport, set up testing and

associated equipment on site

Undertake Skid resistance Tests at

required locations and sections

Reload equipment and transport to

next site or return to base

Post process
field Data

Analyse
Data

Chapter 6

Figure 6-1: Flow Chart of the Process developed for Skid Resistance Field Tests

6.2.2 GripTester Testing Procedures

The following specific steps were undertaken for skid resistance testing and subsequent data processing using the
GripTester device. The steps refer to the stages (from A to F) in Figure 6-1 above.
Step A11

Check the GripTester for proper operation and undertake any maintenance and lubrication that is
required as per the GripTester Operation and Maintenance Manual (Findlay Irvine Ltd, 2002). This
includes checking the battery, the wheels and tyre operation and alignment of axle, chain, water and
distance sensors.

Step A21

Check and calibrate if necessary the GripTester axle strain gauges for vertical load and horizontal drag
forces (load zero, load gain, drag zero and drag gain).

Step B3

Once the GripTester has been properly calibrated, carefully load the GripTester and associated
equipment onto the trailer and transport the equipment to the site.

Step B42

Set up, connect and test auxiliary equipment including the GripTester, the automatic watering system,
USB box connector and the laptop computer, following the procedures in the GripTester User Manual
(Findlay Irvine Ltd, 2003a). Undertake a Pre-test check to ensure all equipment is properly
communicating and working.

Step B52

Appropriately title the survey header and record ambient and pavement surface temperatures in the
proprietary GripTester software fields as with test operators, environment conditions and proposed
speed.

Step C6

Test the site in both directions at the allocated speed, being careful to ensure that software markers are
placed at the beginning and end points of the survey and at other points of significant interest such as a
change of seal. Save the first survey run in both directions as a warm-up run and remove from
subsequent analysis as the testing tyre temperature needs conditioning to an operating temperature.

Step C7

Check the input data on the computer software and save the run data in the GripTester software.
Repeat steps B5 and C6 for a minimum of 5 runs in each direction and save the file to hard disk.

Step D8

Once the 5 runs have been completed, disconnect all equipment and load all equipment into the
vehicle and trailer and continue to the next site (if applicable) and / or return to base.
146

Chapter 6
Step E8

Download the Microsoft Access database files from the laptop to the University IT computer network
(which is backed up every night) which includes the survey header, run header, raw skid resistance
data and markers table from each run.

Step E9

Post process the data by using the Visual Basic Application (VBA) template developed specifically for
the processing of the GripTester data collected at each of the research sites. The template developed
determines the mean and other statistical information of the five GripTester survey runs for the site
(e.g. mean, 95% confidence limit, standard deviation, covariance and standard error). The coefficient
of friction (GN) is plotted against distance and the average result displayed on the graph over the
designated section lengths identified in Figure 5-5 to Figure 5-11 for each of the test sites. The
average result excludes the initial warm-up run.

Step E10

The average coefficient of friction (GN) for each test section and test date is then transferred and
plotted against time to generate the seasonal variations for each site.

Step F11

Analyse the data with respect to temperature and additional rainfall data (from manual on site rainfall
gauges and / or NIWA automatic rainfall gauge stations). Undertake statistical testing using typical
descriptive and parametric statistical tests.

Notes:
1) Refer to the Findlay Irvine GripTester Operation and Maintenance Manual (Findlay Irvine Ltd, 2002) for specific
information and procedures on the maintenance and calibration of the GripTester.
2) Refer to the Findlay Irvine GripTester User Manual (Findlay Irvine Ltd, 2003a) for specific testing and setup
procedures.

6.2.3 Dynamic Friction Tester Field Procedures

The following specific steps were undertaken for skid resistance testing and subsequent data processing using the
DF Tester device. The steps also refer to the stages (from A to F) in Figure 6-1 above.
Step A1*

Check the DF Tester for proper operation and undertake any maintenance and lubrication that is
required as per the DF Tester Help File (Nippo Sangyo Co. Ltd, 2005a). This includes charging the
battery, ensuring the rubber sliders are in good condition and that there are plenty of spares, and that
all moving parts and the water delivery system are working satisfactorily.

Step B2

Once the DF Tester has been properly checked, carefully load the DF Tester, controller and associated
equipment onto the trailer and/or vehicle and transport the equipment to the site.

Step B3*

If required, set up approved traffic control by means of a temporary lane closure. Place vehicle in
front of test location protecting personnel and test equipment from any errant vehicle. Connect the DF
Tester, controller, auxiliary water system and battery equipment following the procedures in the DF
Tester Help file (Nippo Sangyo Co. Ltd, 2005a). Undertake a pre-test check to ensure all equipment is
properly communicating and working.

Step B4*

Appropriately title the test file on the DF Tester controller with the test date, site location and site
position (e.g. file name 200205HIK01denotes a DF Test undertaken on the 20/02/05 at the Hikurangi
site in Northland and in test position 1). The test positions are identified by the timeslice positions
on Figure 5-5 to Figure 5-11 and the date and time of the test. If the macrotexture is not too high (i.e.
<2.0 mm MTD) then an initial test speed of 80km/h can be used. However, if the MTD > 2mm, the
initial speed should be reduced to 60km/h to prevent movement of the DF Tester during the test.
Record separately the ambient and pavement surface temperatures and any other site condition and / or
relevant environmental factors.

Step C5

Test each of the marked DF Test locations in the Left Wheel Track identified on Figure 5-5 to Figure
5-11 with three DF tests per location. After the completion of each test location, check the condition
of the rubber sliders before continuing with the next test. Replace test rubbers if required.

Step D6

Once the three DF Tests have been completed for each of the test locations, disconnect all equipment
and load all equipment into the vehicle and trailer. Reopen temporary lane closure to normal traffic
and continue to the next site (if applicable) and / or return to base.

Step E7*

Once back at the University, undertake DF Tester data transfer following the procedures in the DF
Tester Help file (Nippo Sangyo Co. Ltd, 2005a) from the DF Tester controller to the University IT
147

computer network (which is backed up every night). Ensure the data have been transferred
appropriately and then delete the test data from the controller.

Chapter 6

Step E8

Post process the DF Tester data by using the Visual Basic Application (VBA) template developed
specifically for the processing of the DF Tester data. The template developed takes the three DF
Tester runs and plots the measured coefficient of friction ( ) against the measured slip speed. The
template also calculates the mean of the three DF Tester tests at 20 km/h, 40 km/h and the average
(between 20 and 40 km/h) slip speeds.

Step E9

Transfer and co-plot against time the average coefficient of friction ( ) for each test location on the
graphs for the GripTester seasonal variations.

Step F10

Analyse the data with respect to temperature and additional rainfall data (from manual on site rainfall
gauges and / or NIWA automatic rainfall gauge stations). Undertake statistical testing using typical
descriptive and parametric statistical tests.

* Note:
Refer to the DF Tester Help File (Nippo Sangyo Co. Ltd, 2005a) for specific information and procedures on the
maintenance and calibration and the testing and setup procedures of the DF Tester.

6.3

Processing of the Field Test results

6.3.1 Introduction
Not all of the raw or processed skid resistance measurement field test results are shown here, as there are too many
results and figures for consideration, and in themselves they will prove little. The summarised test results and
processed data is compiled on a CD-ROM that has been submitted with the thesis. However, an example of the post
processing of data and the analysis that was undertaken to arrive at the summary data is important to demonstrate
that a robust methodology was used and as a verification of the process to arrive at the summary results.
Accordingly, the sections that follow show by means of an example, for both the GripTester and the Dynamic
Friction Tester, how the summarised coefficients of friction data (to be discussed in Section 6.4) were arrived at.

6.3.2 Processing of Raw GripTester Results

The raw access database file for a typical GripTester survey, this one undertaken on the 26th June 2003 at 8:56am, is
shown below in Figure 6-2. This access data file shows the results of a GripTester survey with five runs in each
direction. A Marker table (i.e. Markers1 denote the markers for run number one which, in this case, is in the
increasing direction) as shown in Figure 6-3, gives a running distance from the beginning of the survey start point to
the position where each of the three marker locations was entered by the operator, in this case at 126.29m, 747.05m
and 841.65m from the beginning of the survey. The RawData1 table (i.e. the raw data of run one of the GripTester
survey) is shown in Figure 6-4 and reports three columns. The first column (ID) refers to the running distance from
the beginning of the survey and each number is the running distance in metres divided by 10. The second column
(10MetresAvage) is the reported rolling average GripNumber (GN) over the 10 previous one metre readings. The
third column (Speed) is the average speed at which the GripTester was travelling over the 10 metre section.

148

Chapter 6

Figure 6-2: Screen shot of the raw Access database file structure for a GripTester Survey

Figure 6-3: GripTester Raw Markers Access database

Table

Figure 6-4: GripTester Raw Data Access

database Table

The RunHeader1 table (i.e. Run one header table) displays the GripTester survey screen-prompted inputs that the
operator fills in prior to the survey test commencing. An example GripTester Survey input screen is shown in
Figure 6-5 and the prompts include inputs such as:

The start side of the road e.g. L for Left hand side;

Acceleration length, usually 10m;

Deceleration length, usually 10m;

Target Speed, usually 50 km/h;

Self water film depth, usually 0.25mm;

Run date;

Run Number;

Surface condition, either dry, damp, wet or flooded;

Weather condition, either fine, cloudy, or raining;

Ambient temperature;

Surface Temperature;

Comments;

Operator; and

Number of markers.

149

Chapter 6

Figure 6-5: GripTester Pre-Survey Screen

The SurveyHeader table has summary survey data such as the maximum survey length, site description,
GripTester Number and the Type of GripTester.

6.3.3 Example Analysis of GripTester Results

The raw data when processed are exported to a Microsoft Excel spreadsheet for analysis. The data are exported by
means of a Visual Basic Application (VBA) template developed specifically for the processing of the GripTester
data collected in this research. All of the survey runs can then be placed alongside each other in columns, the
surveys plotted and analysis undertaken. Figure 6-6 shows an example of a plot of a GripTester Survey in relation
to distance on the Kaiwaka Slag site in the increasing direction on the 26th August 2004.
The lower grouping of lines on the y1-axis show the coefficient of friction results as measured by the GripTester
(GripNumber - GN) and the higher grouping of lines on the y2 axis show the test Speed (km/h). The bolder blue
line of the coefficient of friction (GN) on the y1 axis is the mean value of the five GripTester runs excluding the
warmup which is discarded (but still shown) as it is used to condition the testing tyre. The bolder red line on
the y2 axis shows the mean value of the average of the five speed test runs in km/h, and it can be seen that it is very
close to the 50km/h target test speed.
On this site, there is some oscillation around the target test speed as the section is characterised by rolling terrain and
reasonably low radii reverse horizontal curves, and it is therefore difficult for the driver to keep to the target speed.
For the same reason, it is even more difficult to keep the GripTester tyre on the same longitudinal line and in the
middle of the left hand wheel track for consecutive test runs. These explanations would lead one to expect that the
variation between runs (the standard deviation ( ) and the cofficient of variation (CoV)) would be greater on a site
such as this in comparison to a straight and level site. This will be further reported on in Section 7.

150

Chapter 6
GripTester Survey - Kaiwaka Site 26-08-04 (Increasing direction)
1.2

60

Survey Test
Speed (km/h)
50

Warm-up GN
discarded in analysis

0.8

0.6

30

Coefficient of friction
as measured by the
GripTester GN

Irregular result. On-site

inspection revealed severe
flushing in the wheel track
causing loss of macrotexture

0.4

GN Warmup
GN Run 5
Speed Run 1

0.2

GN Run 1
Speed Run 2
Speed Run 3

40

GN Run 2
Speed Run 5
Speed Run 4

GN Run 3
GN Ave
Speed Ave

Speed (km/h)

GripNumber (GN)

1.0

20

GN Run 4
Speed Warmup

0.0

10

0
10

110

210

310

410

510

610

710

810

910

Route Distance (m)

Figure 6-6: Example Plot of the Coefficient of Friction (GN) in relation to Distance (chainage) at Kaiwaka in
Northland on SH 1
The plot shown in Figure 6-6 is useful for initial examination of the test results as it can be clearly seen whether a
test run or survey should be discarded from further analysis. For instance, if another road vehicle (such as a slow
moving farm tractor) held up the survey team and the target speed could not be reached for a part of, or the whole
test run, then the GripNumber results would be invalid. As discussed in Section 3.2.4 research has shown that the
coefficient of friction is dependent upon the slip speed of the measuring tyre and the texture (macrotexture) of the
surface. For this case, one would expect to report a higher GripNumber (GN) than the true result.
To help understand the repeatability and reliability of the GripTester (further discussed in Section 7.2) and a
comparison with other devices, as well as to investigate how the site surfaces change with time, the variation
between test runs over time was investigated by means of their descriptive statistical parameters. The following
descriptive statistical parameters (and their Microsoft Excel function name) were analysed between test runs
(excluding the warm-up run ) and plotted for ease of investigation against distance (chainage) for each site and
survey test dates.

Sample mean (AVERAGE);

Median (MEDIAN);

Maximum (MAX);

Minimum (MIN);

Range (RANGE the difference between the maximum and minimum results);

The sum of squares of the deviations of data points from the sample mean (DEVSQ);

Standard deviation (STDEV);

Sample variance (the standard deviation squared);

95th percentile confidence limits (CONFIDENCE); and

Coefficient of variation (the standard deviation divided by the mean).

151

Chapter 6
An example of a statistical plot featuring most of the above statistical descriptions for the same site and same
GripTester survey date as the example above is shown in Figure 6-7. It can be seen that the variation (i.e. standard
deviation, range and coefficient of variation the lines on the lower part of the graph) in this case is reasonably low
between the survey runs. An examination of Figure 6-6 and Figure 6-7 also shows some other irregular features,
readily discernible on a plot, that can point to the need for further investigative visual field inspections. In this case,
investigations revealed flushing of excess bitumen below the new seal, migrating to the surface. This effectively
drowned the aggregate in bitumen, causing loss of macrotexture and a bitumen coating over the microtexture. The
combined effect resulted in a dramatic loss of measured skid resistance(almost half) as measured by the GripTester,
and demonstrates the importance of ensuring macrotexture levels stay stable over time. Similar flushing effects can
also be seen at the other end of the new seal section. A comparison with a GripTester survey taken some 13.5
months earlier (refer to Figure 6-8), when the new seal was only approximaterly 2.5 months old demonstrates that
whilst the overall GN level was approximately the same, the flushing was not evident. The seal section average
would therefore be more representative and the average coefficient of variation significantly less over the same
sections in the earlier measurements.
Statistical Data Kaiwaka 26-08-04 (Increasing direction)
1.2

Surface
maintenance
(patching)

GripNumber (GN)

1.0

New Seal - approx 15.5 months old

Old Seal

(New seal section length for seasonal averaging)

0.8

0.6

Irregular results. On-site inspection

revealed severe flushing in the
wheel track causing loss of
macrotexture that occurred during
the first summer after construction

0.4

0.2

Little statistical
variation between
test runs

0.0
10

110

210

310

410

510

610

710

810

910

Route Distance (m)

AVE

MEDIAN

MAX

MIN

RANGE

DEVSQ

STDEV

95% Conf

Marks

CoV

Figure 6-7: Example Statistical Plot of the GripTester survey data at Kaiwaka in Northland on SH 1

152

Chapter 6
Statistical Data Kaiwaka 13-06-03 (Increasing direction)
1.2

New Seal - approx 1 month old

GripNumber (GN)

1.0

(New seal section length for seasonal averaging)

0.8

0.6

No significant reductions in skid

resistance as flushing had not yet
occurred (in first winter after
construction)

0.4

0.2

0.0
10

110

210

310

410

510

610

710

810

910

Route Distance (m)

AVE

MEDIAN

MAX

MIN

RANGE

DEVSQ

STDEV

95% Conf

CoV

Figure 6-8: Statistical Plot of the GripTester survey at Kaiwaka when the New Seal was 2.5 months old
Two principal methods have been used to analyse how the test sections perform and vary with time. An initial
investigation of the full test section by route distance can demonstrate the total range of variation at various points
along the road. An example plot is shown in Figure 6-9 for the Kaiwaka site.

Kaiwaka Slag Site Site Averages, Increasing Direction

1.20
1.00
Grip Number (GN)

These GripTester surveys should

be ignored as there was a fault in
the GripTester
0.15 GN variation over 26
months for stable seal sections

Unstable seal sections

due to maintenance
treatments and flushing

0.80
0.60
0.40
GripTester survey prior to
new seal being constructed

0.20

0.43 GN variation over 26

months for flushed section

0.00
0

100

200

300

400

500

600

700

800

900

21-Mar-03
27-May-03
SITE EXTENTS
13-Jun-03
11-Jul-03
25-Jul-03
26-Jun-03
19-Sep-03
8-Oct-03
22-Oct-03
6-Nov-03
12-Dec-03
18-Dec-03
7-Jan-04
15-Jan-04
5-Feb-04
19-Mar-04
30-Apr-04
14_May-04
3-Jun-04
6-Jul-04
31-Jul-04
26-Aug-04
10-Sep-04
13-Oct-04
9-Nov-04
27-Nov-04
2-Feb-05
21-Feb-05
14-Mar-05
18-Apr-05
12-May-05

Route Distance (m)

Figure 6-9: The Variation of GripTester runs at the Kaiwaka site over 26 month period
The plot shown in Figure 6-9 is useful to demonstrate the total range of the variation over the data collection period.
However, it is unable to demonstrate how the section varies over time between the extents. To enable an analysis
and investigation of this variation of measured skid resistance over time, and to see whether there is any relationship
with the seasons of the year (commonly known as the seasonal variation and discussed in Section 3.5.6), requires
an averaging of site representative sections. These sections should desirably behave as homogeneous sections and
153

Chapter 6
can then be monitored for any significant relationships against time, season and other factors such as macrotexture,
temperature and rainfall over the data collection period.
Figure 6-10 shows a plot of the average section GripTester (GN) results (y1 axis) from 200m to 690m at the State
Highway One Kaiwaka site in the linear referencing increasing direction of travel. The figure also shows the
average Mean Profile Depth (MPD) macrotexture measurement in the Left Wheel Path (LWP) over this same
section when it has been surveyed by the WDM SCRIM device. At these times, the coefficient of friction by the
Sideways Force Method (SFC) as measured by the SCRIM device is also plotted for comparative reasons. As this
site was resealed after the first GripTester survey on the 7th May 2003, one would expect to see a gradual
exponential decrease in the measured skid resistance over time (stage one polishing phase). The rate of this
decrease since that new seal construction date should depend upon the aggregate s resistance to polishing.

If the

seal was in the second equilibrium or seasonal variation phase, then the literature has suggested that a sinusoidal
summer-winter pattern could be expected. A sinusoidal pattern has been shown on Figure 6-10 (GN in theory) as a
dotted line with an amplitude of 0.20 GN, purely for a visual representation to determine whether a possibe pattern
exists. As stated above it would not be expected to be seen at this site at least until an equilibrium phase of initial
aggregate polishing had been completed. This could be expected to occur in the second year of the seal s life. It
should also be stated that this site is not a typical surfacing as it was not re-surfaced with a natural aggregate. It was
resurfaced in May 3003 as a trial site with a melter slag artificial aggregate, a by-product of the BHP steel mill in
Waiuku in South Auckland. These factors will be discussed later.

Kaiwaka Slag Site - GN and Texture vs Time (Ave Incr dir)

0.90
Grip Number(GN) / SFC

6.0

GripTester was not working

properly during this period

5.0

0.80
0.70

4.0

0.60
0.50

3.0

0.40

2.0

New Seal at 07/05/2003

TWORS chip size 3/5,
PSV 58

0.30
0.20

1.0

0.10
21/03/2005

21/01/2005

21/11/2004

21/09/2004

21/07/2004

21/05/2004

21/03/2004

21/01/2004

21/11/2003

21/09/2003

21/07/2003

21/05/2003

0.0
21/03/2003

0.00

LWP Macrotexture (mm)

1.00

Date
Grip Number

GN in theory

SCRIM SFC

LH-Ave LWP-Inc

Figure 6-10: The Variation in Average GN, SFC and Macrotexture over Time at the Kaiwaka site over 26
month period

6.3.4 Skid resistanceTimeslice Analysis

As the data collection period was over a three year timeframe, there was concern that averaging the test section
results was masking localised changes that were occuring within the section lengths themselves. Consequently, a
timeslice technique was developed in order to analyse the performance of specific locations on the field test sites
154

Chapter 6
and their polishing mechanisms, as opposed to the overall section averages. The timeslice technique was developed
as a pragmatic compromise between the inaccuracies of the linear referencing and transverse location of the
GripTester meaurement wheel for a single 10m average section and the total section length average that would
develop variances based upon localised differences in the polishing mechanisms. As such, a timeslice is defined as
an average of a 30m section calculated by the mean of the three 10m average sections reported in this case by the
GripTester, one 10m average either side of the timeslice location and including the reported location distance. For
example, a timeslice at location 690m on the Kaiwaka slag site in the increasing direction (refer to Figure 5-7)
would report the average10m GripTester coefficient of friction (GN) for the 10m averaged results of 680m, 690m
and 700m.
The specific locations of time slices were chosen based on the type of downstream comparative analysis that was
required. Some examples of the types of geometrical element comparative analysis that the timeslice technique
made possible are:

New seal performance in comparison to adjacent older seal lengths;

Horizontal curves in comparison to straight road sections;

Turning intersection areas in comparison to straight non-turning traffic areas; and

Down-gradient sections in comparison to level gradient sections.

The results of these types of comparison and analysis are discussed in Section 8.4.

6.3.5 Processing of Dynamic Friction Test Results

The Dynamic Friction Tester (DF Tester) records a measured coefficient of friction for every 0.1km/h interval from
the operator set initial test speed (usually 80km/h for areas with low macrotexture and 60km/h for high
macrotexture) down to 0km/h slip speed. The raw data for a typical DF Tester test location with three runs, was
processed by a Visual Basic Application (VBA) template developed specifically for the processing of the DF Tester
data. The template was developed in a Microsoft Excel spreadsheet and tabulates the three DF Tester runs and plots
the measured coefficient of friction ( ) against the measured slip speed. An example of a typical processed
Dynamic Friction Test result undertaken on the 9th December 2004 at the Hikurangi Site in Northland and at test
position three, is shown in Figure 6-11. The template also calculates the mean of the three test runs at slip speeds of
20 and 40 km/h and the average between 20 and 40km/h.

155

Chapter 6

09 Dec 2004 Hikurangi Position 3 (HIK 03)

1000

Friction Coefficient ( * 1000)

900

Run 1

Changing coefficient of friction as

slip speed reduces to zero

800

Run 2
Run 3

700
600

Initial braking of rubber sliders

as dynamic test load varies

500
Mean of 20-40 km/h

400

CoF @ 20
km/h

300
200

CoF @ 40
km/h

Equivalent slip speed as

GripTester = 7.5km/h

Initial Test Speed, 60km/h

100
0
0

10

20

30

40

50

60

70

80

90

100

Slip Speed (km/h)

Figure 6-11: Typical Processed Dynamic Friction Test Result showing three test runs

6.3.6 Analysis of Dynamic Friction Test Results

The number of static Dynamic Friction tests that were obtained in the field throughout the data collection period was
limited due to the:

Availability of the equipment for the research (the first trial tests with the DFTester device were in
February 2004, more than 1.5 years after skid testing with the GripTester began at the Tamaki site in July
2002); and

Requirement for temporary traffic lane closure and control of live traffic sites for static tests to be
undertaken on the road surface.

The DF Tester is also a stationary device and provides a reading at a specific location. Specific points of interest
and/or positions that were representative of the test section were chosen and, whenever possible, repeat tests at the
same location were obtained. The sites chosen for the DF Test monitoring were:

University of Auckland Tamaki Campus site, 12 positions (6 in either direction) - traffic control was not
required at this site;

Ports of Auckland Rail Grid sites, 6 locations (3 in either direction) traffic control was not required at this
site;

Hikurangi control site in Northland (3 locations in the increasing direction) full traffic control was
required for stationary tests at this location with the DF Tester. Traffic control and management was
undertaken by Works Infrastructure staff from the Northland PSMC 002 Contract staff in Whangarei.

The locations of the DF Tester points at the above three sites are shown on Figure 5-5, Figure 5-6 and Figure 5-9
respectively of Section 5.7 of Chapter 5. The other Northland sites (except for the Brynderwyn South curve site)
had DF Tester measurememts undertaken on only one occasion and in one direction, and for a limited number of
locations as identified below:

156

Chapter 6
Kaiwaka Slag site - 6th July 2004, three locations in the LWT and the RWT in the increasing route position
direction;

Kara Road site - 14th March 2005, 4 locations in LWT and in the increasing route position direction;

Snooks / Tatton Road site 14th March 2005, 4 locations in LWT and in the increasing route position
direction.

The processed results of the DF Tester (refer to Section 6.3.5) allowed an analysis of:

Seasonal variations over time;

The repeatibility of the device in comparison to the GripTester; and

A correlation between the GripTester and the DF Tester.

6.4

Summary Field Site Results

6.4.1 Introduction
This section provides summary results of the seven field test sites described in Section 5.7. A summary figure for
each field site displays the results of the average section GripTester coefficient of friction measurement (GN on y1
axis) and surface temperature (y2 axis) in both traffic directions compared over time (x axis). Where applicable the
figure also displays the average Dynamic Friction Tester coefficients of friction ( ) obtained at the specified test
locations along the section length. A summary table follows each figure displaying descriptive section statistics for
each field site. The raw and processed data is attached with the accompanying CD ROM. Sections that follow will
discuss the analysis of specific aspects of the field results.
Unfortunately, due to a series of problems wth the GripTester skid testing device (including requiring a new signal
processing unit, new measuring axle with horizontal and vertical force transducers, and thirdly due to a slowly
seizing measuring wheel bearing unit) the results over the 2003/2004 summer season on all of the sites cannot be
regarded as credible, and have subsequently been removed from the analysis as data outliers . However, for
completeness of presentation of the data, these measurement values have been included and the problem period
identified on the figures. These resultant problems with the GripTester over the 2003/2004 summer season meant
that, in essence, from March 2004 the GripTester skid testing device was almost a completely new device (except
for the chassis, suspension and driving wheels), an undesirable, but uncontrollable external factor.
Coincidentally or not, during this same period (summer 2003/2004), all of the field test sites skid resistance
measurements increased to a higher level (from before the device problems to after ). It is unknown whether this
change was due to the changes in the device (which is possible, and if so, the amount is significant) or whether the
coefficient of friction increase was real and due to environmental changes external to the device. It was fortunate
that the DF Tester became available during this period, allowing a retrospective analysis that clearly demonstrates
that the GripTester device was having problems with its measurement during this period. However, because the DF
Tester was not available to the research programme in 2003, a direct comparison cannot be made with the previous
year.

6.4.2 Tamaki Campus (University of Auckland) Site

The Tamaki campus section is one of the two Auckland asphalt mix (TNZ Mix 10) surface field sites (as described
in Section 5.7.1). The testing section was divided into two sections for analysis, a flat grade section and a sloped
section on a 7.2% grade. The skid resistance testing was undertaken in both traffic directions (generally identified
as being in the West and East directions) as shown in Figure 5-5. Minimum monthly skid resistance testing was
157

Chapter 6
undertaken at this site, beginning in July 2002. The average skid resistance test result was obtained for each test
date by calculating the sample mean value of test runs one to five (excluding the warm-up run) for the following
sections:

West direction (flat section) running distance 60m to 140m;

West direction (sloped section) running distance 200m to 300m;

East direction (sloped section) running distance 30m to 130m; and

East direction (flat section) running distance 190m to 270m.

Timeslice locations were also chosen, processed and analysed on the Tamaki site, many of which were coincident
with repeated DF Testing locations. The timeslice locations are identified below, with the coincident DF Tester test
locations denoted by the initials DFT and the position location shown in Figure 5-5:

West direction 60m (DFT W1), 80m (DFT W2), 120m, 220m, 250m (DFT W3), and 270m (DFT W4);
and

East direction 70m (DFT E1), 80m (DFT E2), 110m, 210m, 250m (DFT E3) and 260m (DFT E4).

The performance of the Tamaki site flat and sloped sections in terms of its measured coefficient of friction over
time, and as measured by the GripTester and the DF Tester devices, is shown in Figure 6-12 and Figure 6-13. Table
6-1 and Table 6-2 summarise the descriptive section statistics data for the flat and sloping sections respectively.
UoA Tamaki Campus Site (Flat Section, Both Directions)
Skid Tester not working
properly during this period

1.00

100

Water delivery problems causing

higher than expected results

90
80
70

0.80

60
0.60

50
40

0.40

30
20

0.20

Surface Temperature (C)

GripNumber (GN) / DFT ( )

1.20

10
17/11/2005

17/09/2005

17/07/2005

17/05/2005

17/03/2005

17/01/2005

17/11/2004

17/09/2004

17/07/2004

17/05/2004

17/03/2004

17/01/2004

17/11/2003

17/09/2003

17/07/2003

17/05/2003

17/03/2003

17/01/2003

17/11/2002

17/09/2002

0
17/07/2002

0.00

Time
GNE30 R1-R5

GNW30 R1-R5

DFT E20 ( )

DFT W20 ( )

Surface Temperature

Figure 6-12: Tamaki Campus Summary results over time (Flat section, both directions)

158

Chapter 6

Tamaki Campus Site (West Flat Section)

Sample Section
Sect Ave
TIMESLICES
Descriptive Statistics
GN 30
GN T/S
GN T/S
GN T/S
R1-R5
W1 60m W2 80m W3 120m
Sample count
31
30
30
30
Mean
0.80
0.79
0.76
0.83
Maximum
0.93
0.96
0.93
1.01
Minimum
0.71
0.67
0.67
0.73
Range
0.22
0.29
0.26
0.27
Standard Deviation
0.056
0.072
0.067
0.068
95% Confidence Interval
0.020
0.026
0.024
0.024
Variance
0.003
0.005
0.004
0.005
Coefficient of Variation (CoV)
0.070
0.091
0.088
0.082

Ave
Temperature (C)
DFT 20 Ambient
Surface
( )
(Start)
(Start)
11
30
30
0.73
19.4
30.2
0.78
28.4
52.0
0.66
13.5
15.0
0.13
14.9
37.0
0.035
4.4
10.1
0.021
1.6
3.6
0.001
19.6
101.8
0.048
0.228
0.334

Tamaki Campus Site (East Flat Section)

Sample Section
Sect Ave
TIMESLICES
Descriptive Statistics
GN 30
GN T/S
GN T/S
GN T/S
R1-R5
E4 210m E5 260m E6 280m
Sample count
32
26
26
26
Mean
0.79
0.77
0.79
0.77
Maximum
0.93
0.88
0.94
0.90
Minimum
0.70
0.70
0.70
0.69
Range
0.23
0.17
0.23
0.21
Standard Deviation
0.059
0.044
0.057
0.046
95% Confidence Interval
0.021
0.017
0.022
0.018
Variance
0.004
0.002
0.003
0.002
Coefficient of Variation (CoV)
0.075
0.057
0.073
0.060

Temperature (C)
Ave
DFT 20 Ambient
Surface
( )
(Start)
(Start)
15
30
30
0.72
19.4
29.8
0.77
28.4
52.0
0.64
13.5
13.0
0.13
14.9
39.0
0.037
4.4
10.8
0.019
1.6
3.9
0.001
19.6
116.9
0.051
0.228
0.363

Table 6-1: Tamaki Campus Site Descriptive Statistics (Flat Section)

UoA Tamaki Campus Site (Slope Section, Both Directions)
Skid Tester not working
properly during this period

100

Water delivery problems causing

higher than expected results

90

GripNumber (GN) / DFT ( )

1.00

80
70

0.80

60
0.60

50
40

0.40

30
20

0.20

Surface Temperature (C)

1.20

10
17/11/2005

17/09/2005

17/07/2005

17/05/2005

17/03/2005

17/01/2005

17/11/2004

17/09/2004

17/07/2004

17/05/2004

17/03/2004

17/01/2004

17/11/2003

17/09/2003

17/07/2003

17/05/2003

17/03/2003

17/01/2003

17/11/2002

17/09/2002

0
17/07/2002

0.00

Time
GNW30 R1-R5

GNE30 R1-R5

DFT W20 ( )

DFT E20 ( )

Surface Temperature

Figure 6-13: Tamaki Campus Summary results over time (Slope section, both directions)

159

Chapter 6
Tamaki Campus Site (West Slope Section)
Sample Section
Sect Ave
TIMESLICES
Descriptive Statistics
GN 30
GN T/S
GN T/S
GN T/S
R1-R5 W4 220m W5 250m W6 270m
Sample count
31
29
29
29
Mean
0.76
0.75
0.76
0.74
Maximum
0.94
0.90
0.94
0.91
Minimum
0.69
0.67
0.68
0.66
Range
0.25
0.23
0.26
0.25
Standard Deviation
0.059
0.060
0.060
0.058
95% Confidence Interval
0.021
0.022
0.022
0.021
Variance
0.003
0.004
0.004
0.003
Coefficient of Variation (CoV)
0.078
0.080
0.079
0.078

Ave
DFT 20
( )
13
0.71
0.77
0.67
0.10
0.028
0.015
0.001
0.040

Temperature (C)
Ambient
Surface
(Start)
(Start)
30
30
19.4
30.2
28.4
52.0
13.5
15.0
14.9
37.0
4.4
10.1
1.6
3.6
19.6
101.8
0.228
0.334

Tamaki Campus Site (East Slope Section)

Sample Section
Sect Ave
TIMESLICES
Descriptive Statistics
GN 30
GN T/S
GN T/S
GN T/S
R1-R5
E1 70m
E1 80m E3 110m
Sample count
32
26
26
26
Mean
0.79
0.78
0.75
0.78
Maximum
0.92
0.87
0.89
0.90
Minimum
0.69
0.70
0.63
0.70
Range
0.23
0.17
0.26
0.20
Standard Deviation
0.056
0.043
0.061
0.052
95% Confidence Interval
0.019
0.017
0.024
0.020
Variance
0.003
0.002
0.004
0.003
Coefficient of Variation (CoV)
0.072
0.056
0.082
0.067

Ave
DFT 20
( )
14
0.67
0.74
0.64
0.11
0.035
0.019
0.001
0.053

Temperature (C)
Ambient
Surface
(Start)
(Start)
30
30
19.4
29.8
28.4
52.0
13.5
13.0
14.9
39.0
4.4
10.8
1.6
3.9
19.6
116.9
0.228
0.363

Table 6-2: Tamaki Campus Site Descriptive Statistics (Slope Section)

The Tamaki Campus site had been constructed only a few months prior to the skid testing measurement programme
commenced and was therefore expected to demonstrate an initial polishing phase generally following a negative
exponential form to an equilibrium skid resistance level. Thereafter, the expectation from previous research was
that seasonal variations throughout summer and winter seasons would occur. Initial observations of the first year
of data on both sections do demonstrate an initial polishing phase reducing down to a low reading in
February/March of 2003. However, the measured coefficient of friction then increased again over the next winter
period up to almost as high in August 2003, as the first tests in the preceeding winter of July 2002. As this site is an
internal campus road, the traffic during this period was predominantly composed of private cars and only minor
heavy commercial vehicle traffic due to on-site building construction vehicles .
As discussed earlier, the GripTester device coefficient of friction increased after the 2003/2004 summer period in
comparison to the previous year (this was also seen on the other field test sites). It is unknown whether this increase
was real or related to the problems associated with the GripTester device over this period. After this period, the
Tamaki site GN and DFT measurements varied around the total sample mean coefficient of friction (0.79-0.80 GN
and 0.76-0.79 GN for the flat and sloped sections respectively) similar to that tested at the beginning of the testing
period. However, when a water delivery problem occurred in May 2005, it caused, understandably, an increase in
the measured GripTester coefficient of friction (GN) value in comparison to the DF Tester coefficient of friction.
This high level close to the originally tested value was not initially expected as, the initial polishing phase was
thought to polish the aggregate, thereby reducing this level to a lower equilibrium level than the initial testing
results close to the original surfacing date. The results tend to indicate that, as the site is not heavily trafficked, this
initial polishing phase may not occur at all, and only seasonal and short-term variations occur. In support of this,
the highest results were obtained in the 2004 year during the months of July, August and October coinciding with
the expected increase during winter.
160

Chapter 6
A descriptive statistical analysis (refer to Table 6-1 and Table 6-2) of the coefficient of friction, as measured by the
GripTester, for each of the four sections of the Tamaki site shows that there is very little statistical difference
between the sample section means, maximums and minimum, standard deviations, the 95% confidence interval,
variance and coefficient of variation. The number of data sample points was either 31 or 32 and the sample mean
(x) ranged from 0.76 GN on the West slope section to 0.80 GN on the West flat section. The remaining two sections
had a sample mean (x) of 0.79. The flat section on average gave marginally higher means (0.80 and 0.79) than the
sloped section (0.76 and 0.79) for the west and east directions respectively. This can be explained due to the higher
traction forces on the sloped section (7% grade) causing greater aggregate polishing on the sloped section in
comparison to the flat section. The standard deviation was marginally below 0.06 GN on all four sections and the
coefficient of variation (CoV) between GripTester test dates ranged from 7.2% to 7.8% for the section averages.
The 95% confidence interval was also very consistent between the four sections as they ranged between 0.019 and
0.021 meaning for example that there was a 95% probability that the mean coefficient of friction would fall between
0.78 and 0.82 GN for the West flat section.
The DF Tester measured coefficient of friction had a maximum of 15 sample points for the various test point
locations with a sample mean (x) ranging from 0.67 to 0.73 ( ) and a sample standard deviation ( ) ranging from
0.028 to 0.037. A similar pattern was demonstrated with the DF Tester as the GripTester as marginally lower
coefficients of friction were obtained on the sloped section (0.71 and 0.67) in comparison to the flat section (0.73
and 0.72) for the west and east directions respectively. The coefficient of variation (CoV) between the DF Tester
test dates at the 12 test spot locations range from 4.0% to 5.1%, and the mean sample data demonstrated a 95%
confidence interval range of 0.015 to 0.021 ( ) for the various test positions. This means, for example on the West
flat section, that there was a 95% probability that the mean coefficient of friction , as measured by the DF Tester,
would fall between a range of DF Tester ( ) value of 0.71 and 0.75 for the test positions.
The sample standard deviations ( ) and the CoV are lower as measured by the DF Tester in comparison to the
GripTester. This is most probably due to the measuring method differences and the greater likelihood that the DF
Tester position was recorded in exactly the same location. However, the variation from test position to test position
is greater for the DF Tester than the averaged GripTester results as two of the test location positions for the DF
Tester out of the 12 were chosen at the position that was most likely to polish the greatest (as also shown with the
GripTester timeslice analysis for positions 80m in the West direction and 260m in the East direction). This will
have skewed the averaging process and can be confirmed by comparing the CoV for the GripTester at these same
locations; the values are higher by approximately 2% than the section means discussed above.

6.4.3 Ports of Auckland Site

The Ports of Auckland site is the second Auckland asphalt mix surface (TNZ Mix 20) and is described in Section
5.7.2. The skid resistance testing was undertaken on a staight and level section adjacent to the Ports of Auckland
Rail Grid site where ship containers are loaded onto rail cars (refer to Figure 5-6).

The rail grid site has

predominantly very heavy fork lift and container straddler traffic. The principal traffic direction is predominantly
transverse to the direction of skid measurement testing, although a significant amount of sharp turning movements
also occurs along the rail grid site. The site was tested in both the western and eastern traffic direction, as shown in
Figure 5-6. Two straight white painted lines, approximately 3.5m apart and for approximately 475m along the Rail
Grid site, provided a reasonably good locational reference guide for the skid testing line. Whilst there was no
defined wheel path or traffic lane, the left wheel path (LWP) position was defined for the purposes of this study and
tested in both directions, allowing a separation of the test lines. Minimum monthly skid resistance testing with the
161

Chapter 6
GripTester device was undertaken at this site beginning in September 2002 and predominantly between the times of
4:00pm to 7:00pm, when access was allowed to the site. As no other vehicles used the testing location during the
skid resistance measurements, other aspects of skid resistance could be investigated at various times (e.g. the
relationship between texture, speed and skid resistance and a device s sensitivity to temperature).
Average skid resistance test results were obtained for each test date after post processing by calculating the sample
mean value of test runs one to five (excluding the warm-up run) for the following sections:

West direction running distance 220m to 420m; and

East direction running distance 180m to 380m.

Six DF Tester test positions were also marked and repeatedly tested on the same date as the GripTester
measurements from the date when the DF Tester device became available (February 2004). The test locations are
shown on Figure 5-6 and are listed below in each travel direction:

West direction 160m, 190m and 210m; and

East direction 240m, 270m and 290m.

The performance of the Ports of Auckland site in terms of its measured coefficient of friction over time and as
measured by the GripTester and the DF Tester devices is shown in Figure 6-14 and Figure 6-15. Table 6-3
summarises the descriptive section statistics data.
Ports of Auckland Rail Grid Site (West Direction)
90

15/07/2005

0
15/05/2005

0.00
15/03/2005

10
15/01/2005

0.10
15/11/2004

20

15/09/2004

0.20

15/07/2004

30

15/05/2004

0.30

15/03/2004

40

15/01/2004

50

0.40

15/11/2003

0.50

15/09/2003

60

15/07/2003

0.60

15/05/2003

70

15/03/2003

0.70

15/01/2003

80

15/11/2002

0.80

15/09/2002

GripNumber (GN) / DFT ( )

100

Skid Tester not working

properly during this period

0.90

Surface Temperature (C)

1.00

Time
GN50 R1-R5

GN20

GN30

GN70

GN90

DFT20 ( )

Surface Temperature

Figure 6-14: Ports of Auckland site Summary results over time (West direction)

162

Chapter 6

Ports of Auckland (West Direction)

Sample Section
Sect Ave
Descriptive Statistics
GN 20
R1-R5
Sample count
13
Mean
0.69
Maximum
0.74
Minimum
0.64
Range
0.10
Standard Deviation
0.034
95% Confidence Interval
0.019
Variance
0.001
Coefficient of Variation (CoV)
0.050

Sect Ave
GN 30
R1-R5
15
0.67
0.72
0.62
0.10
0.032
0.016
0.001
0.047

Sect Ave
GN 50
R1-R5
24
0.61
0.74
0.53
0.22
0.064
0.026
0.004
0.106

Sect Ave
GN 50
After
4
0.69
0.71
0.63
0.08
0.038
0.038
0.001
0.056

Sect Ave
GN 70
R1-R5
13
0.47
0.52
0.42
0.10
0.030
0.016
0.001
0.063

Sect Ave
GN 90
R1-R5
8
0.44
0.47
0.40
0.07
0.024
0.017
0.001
0.056

Ave
DFT 20
( )
14
0.489
0.551
0.412
0.139
0.041
0.022
0.002
0.084

Temperature (C)
Ambient
Surface
(Start)
(Start)
26
26
18.4
25.0
29.0
41.0
10.5
12.0
18.5
29.0
5.3
9.1
2.0
3.5
27.6
83.3
0.285
0.366

Ports of Auckland (East Direction)

Sample Section
Sect Ave
Descriptive Statistics
GN 20
R1-R5
Sample count
13
Mean
0.74
Maximum
0.79
Minimum
0.66
Range
0.13
Standard Deviation
0.040
95% Confidence Interval
0.022
Variance
0.002
Coefficient of Variation (CoV)
0.054

Sect Ave
GN 30
R1-R5
14
0.70
0.75
0.65
0.10
0.032
0.017
0.001
0.046

Sect Ave
GN 50
R1-R5
23
0.63
0.77
0.52
0.25
0.070
0.029
0.005
0.112

Sect Ave
GN 50
After
4
0.71
0.75
0.65
0.10
0.045
0.044
0.002
0.063

Sect Ave
GN 70
R1-R5
13
0.49
0.56
0.41
0.15
0.045
0.025
0.002
0.093

Sect Ave
GN 90
R1-R5
9
0.46
0.50
0.41
0.09
0.036
0.023
0.001
0.078

Ave
DFT 20
( )
14
0.550
0.618
0.431
0.187
0.058
0.031
0.003
0.106

Temperature (C)
Ambient
Surface
(Start)
(Start)
26
26
18.4
25.0
29.0
41.0
10.5
12.0
18.5
29.0
5.3
9.1
2.0
3.5
27.6
83.3
0.285
0.366

Table 6-3: Ports of Auckland field site Summary Statistics (both directions)
Ports of Auckland Rail Grid Site (East Direction)
0.90

90

0.80

80

15/07/2005

15/05/2005

15/03/2005

15/01/2005

0
15/11/2004

10
15/09/2004

0.10
0.00
15/07/2004

20

15/05/2004

0.20

15/03/2004

30

15/01/2004

0.30

15/11/2003

40

15/09/2003

0.40

15/07/2003

50

15/05/2003

0.50

15/03/2003

60

15/01/2003

70

15/11/2002

0.70
0.60

15/09/2002

GripNumber (GN) / DFT ( )

100

Skid Tester not working

properly during this period

Surface Temperature (C)

1.00

Time
GN50 R1-R5

GN20

GN30

GN70

GN90

DFT20 ( )

Surface Temperature

Figure 6-15: Ports of Auckland site Summary results over time (East direction)
The asphalt mix surface at the Ports of Auckland site was constructed in 1996 and was therefore expected to be well
past the initial polishing phase and very clearly at an equilibrium skid resistance level , if such a phenomenon
exists. Any variations in measured skid resistance levels were therefore expected to be due to seasonal and shortterm environmental effects. The effects of varying the measurement test speed from 20 km/h to 90km/h was
investigated for the first 15 months of the data collection period at this site. After this date a standard 50 km/h
measurement speed was utilised and DF Tester tests commenced. The first 15 months of data (refer to Figure 6-14
and Figure 6-15) clearly demonstrated the effect of varying the device measurement speed and thereby the slip
speed of the testing tyre on an asphalt mix surface, i.e. that the coefficient of friction reduces as the speed increases.
163

Chapter 6
The 90km/h test speed resulted in an almost 40% reduction in measured skid resistance when compared to the skid
resistance measured at 20km/h.
Careful visual observations of the GripTester measurements over the first and last year at the measuring speed of
50km/h at the Ports of Auckland site (refer to Figure 6-14 and Figure 6-15) indicate that measured skid resistance at
this site does demonstrate a sinusoidal pattern with the lowest measured coefficient of friction in the months of
February / March in both years and the highest in approximately August to November. The results indicate that the
phase minimum may actually coincide with late summer (March) and the phase maximum in spring (October),
indicating that a clear seasonal variation effect occurs at this site.
As discussed, regarding the Tamaki Campus site, a similar increase in the GripTester device coefficient of friction
also occurred at the Ports of Auckland site after the 2003 / 2004 summer period, in comparison to the previous year.
It is unknown whether this increase was real or related to the problems associated with the GripTester device over
this period. The results also show that the measured skid resistance in the west direction is slightly lower than in the
east direction for both the GripTester and the Dynamic Friction Tester device.
A descriptive statistical analysis (refer to Table 6-3), with the identified outliers removed, showed that the Ports of
Auckland site, as measured by the GripTester device, with a maximum of 24 test surveys for the 50km/h testing
speed, had very little difference in section means (x) by test direction. The sample mean (x) was 0.61 GN and 0.63
GN for the western and eastern directions respectively. The standard deviation of the averaged GripTester section
survey results was 0.064 GN (west) and 0.070 GN (east) and the coefficient of variation (CoV) was 10.6% (west) to
11.2% (east) respectively.

The data resulted in a 95% confidence interval that ranged from 0.026 to 0.029

respectively, meaning that there was a 95% probability that the mean GN value, for example in the west direction,
would fall between the values of 0.58 GN and 0.64 GN at a 50km/h testing speed.
The DF Tester-measured coefficient of friction had a maximum of 14 test date points for the six test point locations
(three in either direction as shown in Figure 5-6) and resulted in a sample mean (x) of 0.49 ( ) in the west direction
and 0.55 ( ) in the east direction. The sample standard deviation ( ) was 0.041 (west) and 0.058 (east) and the
coefficient of Variation (CoV) between the DF Tester test dates at the 14 test spot locations was 8.4% (west) and
10.6% (east). The sample data demonstrate a 95% confidence interval of 0.022 and 0.031 respectively, meaning that
there was a 95% probability that the DF Tester ( ) value will, for example in the west direction, fall between the DF
Tester ( ) values of 0.47 and 0.51.
As discussed earlier on the Tamaki site, the sample standard deviations ( ) and the CoV are lower for the DF Tester
in comparison to the GripTester. This would be expected, due to the differences in measuring method (continuous
in comparison to stationary) and because the variability due to the operator and transverse measuring position for the
GripTester is removed from the DF Tester. The total variation in measured skid resistance at the Ports of Auckland
site is greater than the Tamaki site (10.9% CoV c.f 7.4% CoV) as would be expected, due to the considerably greater
proportion and movements of heavy commercial traffic at the Ports of Auckland site causing greater polishing
effects.

6.4.4 Kaiwaka Slag Site

The Kaiwaka slag test site is the first of five Northland State Highway test sites and its physical attributes are
described in Section 5.7.3 and shown in Figure 5-7. The site is on State Highway 1 and is made up of two reverse
horizontal curves of radii of approximately 200m and is therefore categorised under TNZ T/10 (Transit New
Zealand, 2002) specification as a Site Category 2 section. The site was resurfaced in April 2003 with a two coat
164

Chapter 6
chipseal utilising a TNZ Grade 3/5 artificial steel melter slag aggregate with a reported PSV of 58. Skid resistance
measurements with the GripTester began in the month prior to the new surfacing being constructed. Annual
Average Daily Traffic Volumes (AADT) are recorded in the RAMM database as being 8650 vpd with a heavy
commercial vehicle (HCV) composition of 9.6%. The section operates under a 100km/h posted speed limit. The
site was tested approximately monthly between March 2003 and May 2005 in both the increasing (southbound) and
decreasing (northbound) linear directions. A defined longitudinal wheel path and therefore regular tyre / aggregate
contact area and polishing mechanism occurred at this site. However, as the horizontal curve radii of the two curves
are reasonably low (radius = 200m) and there are reasonably wide sealed shoulders, the transverse wheel tracking
width would be expected to be wider than on a straight section of road.
Average skid resistance test results were obtained for each test date after post processing by calculating the sample
mean value of test runs one to five (excluding the warm-up run) for the following sections:

Increasing direction running distance 220m to 710m; and

Decreasing direction running distance 170m to 760m.

Timeslice locations were also chosen, processed and analysed on the Kaiwaka site. The timeslice locations are
identified below and shown in Figure 5-7:

Increasing direction 170m, 230m, 260m, 360m, 440m, 590m, 690m and 800m;

Decreasing direction 100m, 210m, 310m, 460m, 540m, 640m, 670m, and 730m.

The performance of the Kaiwaka slag site in terms of its measured coefficient of friction over time and as measured
by the GripTester device is shown in Figure 6-16. Table 6-4 summarises the descriptive section statistics data.
Kaiwaka Slag Site (Both Directions)
0.90

90

0.80

80

0.70

70

0.60

60

0.50

50
New seal at 07/05/2003,
TWORS, chip size 3/5 PSV 58

0.40

40

21/03/2005

21/04/2005

21/01/2005

21/02/2005

21/12/2004

21/10/2004

21/11/2004

21/08/2004

21/09/2004

21/07/2004

21/05/2004

21/06/2004

21/04/2004

21/02/2004

21/03/2004

21/12/2003

21/01/2004

21/11/2003

0
21/09/2003

0.00
21/10/2003

10
21/07/2003

0.10
21/08/2003

20

21/06/2003

0.20

21/04/2003

30

21/05/2003

0.30

21/03/2003

GripNumber (GN)

100

Skid Tester not working

properly during this period

Surface Temperature (C)

1.00

Time
GN R1-R5 Decr

GN R1-R5 Inc

Surface Temperature

Figure 6-16: Kaiwaka Slag Site Summary Results over Time (Both directions)

165

Chapter 6

Kaiwaka Slag Site (Both Directions)

Sample Section
Descriptive Statistics

Sect Ave Sect Ave Sect Ave Sect Ave

Timeslice
Temperature (C)
GN Mean GN Mean GN Mean GN Mean GN Mean GN Mean Ambient
Surface
W/Up INC R1-R5 INC W/Up DEC R1-R5 DEC 800m INC 100m DEC (Start)
(Start)
Sample count
27
28
27
28
27
27
28
28
Mean
0.69
0.70
0.69
0.72
0.58
0.55
16.4
16.6
Maximum
0.82
0.80
0.83
0.82
0.71
0.66
23.9
30.0
Minimum
0.54
0.55
0.60
0.62
0.49
0.45
10.0
7.0
Range
0.29
0.25
0.23
0.20
0.23
0.21
13.9
23.0
Standard Deviation
0.074
0.057
0.059
0.055
0.058
0.057
3.8
5.7
95% Confidence Interval
0.028
0.021
0.022
0.020
0.022
0.021
1.4
2.1
Variance
0.006
0.003
0.004
0.003
0.003
0.003
14.1
33.0
Coefficient of Variation (CoV)
0.108
0.082
0.086
0.076
0.100
0.103
0.229
0.346

Table 6-4: Kaiwaka Slag site Summary Statistics (both directions)

The Kaiwaka slag site was not expected to display seasonal type variations over time as the site was newly sealed.
It was thought that an initial polishing phase would, over the first year, reduce the measured skid resistance from an
initial value down to an equlibrium level. However, as can be seen from Figure 6-16, the first approximately 18
months of skid resistance tests showed little microtexture polishing with the new chip seal utilising the melter slag
aggregate. As was consistent with other testing sites, there was a slight increase in GN values in the second year of
testing that coincided with replacing significant parts of the GripTester device in the summer between the two years
of data. There is also visible evidence (refer to Figure 6-16) that an approximately seasonal sinusoidal variation
effect was displayed at the site, resulting in a lower measured coefficient of friction in March and April in
comparison to higher results in September and October.
Statistical analysis of the processed data (refer to Table 6-4), with identified outliers removed, shows that the
averaged Kaiwaka slag site survey results with a maximum of 28 test surveys, had a sample mean (x) of 0.70 GN
and 0.72 GN (the warmup runs were not included in this analysis) for the increasing and decreasing directions
respectively. The standard deviation of the averaged GripTester section survey results (excluding the warmup runs
but including the first skid resistance measurement on the old seal surface) was 0.057 GN and 0.055 GN for the
increasing and decreasing directions respectively. The coefficient of variation (CoV) of the section averages ranged
from 8.3% for the increasing direction to 7.6% for the decreasing direction. The data resulted in a 95% confidence
value that was 0.021 for the increasing direction and 0.020 in the decreasing direction. The implication is that for
the increasing direction, for example, there was a 95% probability that the mean GN value would fall between the
values of 0.68 GN and 0.72 GN at a 50km/h testing speed. An analysis of the warmup runs in comparison to the
subsequent test runs in either direction results in a higher recorded standard deviation, 95% confidence interval,
variance and coefficient of variation for either direction. This indicates that the first run of the GripTester is more
susceptible to changes due to environmental factors such as the conditioning of the measuring tyre and possibly the
level of detritus on the surface.
The averaged GripTester device section results show that, in general, the measured skid resistance values in each
direction is very similar. There is some indication from Figure 6-16 that the measured skid resistance is beginning
to reduce (polish) to a lower level, and it is unknown as yet whether this will increase again to a level close to the
initial measured value with the new seal surface. A close observation of the data and on site inspections shows that
the new seal section length is now quite variable in the longitudinal direction. This is due to a number of areas that
are now flushed and/or bleeding with excess bitumen that has migrated from lower chip seal layers. This will
certainly produce lower measured skid resistance.

166

Chapter 6
The total variation in measured skid resistance at the Kaiwaka slag site is similar to the Ports of Auckland site, but
higher than the Tamaki site. However, if the first test result (on the old seal) was removed from the subsequent
analysis, the total coefficient of variation (CoV) would reduce to similar section values of approximately 7.0%.

6.4.5 Brynderwyn South Curve Site

The Brynderwyn South curve site is the second of five Northland State Highway test sites and its physical attributes
are described in Section 5.7.4 and shown in Figure 5-8. The site is on State Highway 1 and is made up of one
relatively sharp horizontal curve of 160m and is therefore categorised under TNZ T/10 specification (Transit New
Zealand, 2002) as a Site Category 2 section.

The horizontal curve geometry is inconsistent with the speed

environment and relatively high approach speeds, and therefore the demand for friction often is close to or exceeds
the available supply. This means that the polishing forces are high on the corner section, thereby requiring regular
maintenance treatment intervention. The treatment in the last few years has consisted of Works Infrastructure
Limited (the network maintenance contractors on behalf of Transit New Zealand) applying a new chip seal surface
in March 2003 and April 2004 as shown in Figure 6-17 and during the data collection period, with the aim of
increasing the skid resistance levels.
Skid resistance measurements with the GripTester began in the month prior to the first new surfacing being
constructed. Annual Average Daily Traffic Volumes (AADT) are recorded in the RAMM database as being 8650
vpd with a heavy commercial vehicle (HCV) composition of 9.6%. The section operates under a 100km/h posted
speed limit. The site was tested approximately monthly between March 2003 and May 2005 in both the increasing
(southbound) and decreasing (northbound) directions. A defined longitudinal wheel path and therefore regular tyre /
aggregate contact area and polishing mechanism occurred at this site. However, as the horizontal curve radii of the
curves are low (radius=160m) with reasonably wide sealed shoulders, the transverse wheel tracking width would be
expected to be wider than on a straight section of road.
Average skid resistance test results were obtained for each test date after post processing by calculating the sample
mean value of test runs one to five (excluding the warm-up run) for the following sections:

Increasing direction running distance 230m to 410m; and

Decreasing direction running distance 210m to 400m.

Timeslice locations were also chosen, processed and analysed on the Brynderwyn site. The timeslice locations are
identified below and shown in Figure 5-8:

Increasing direction 150m, 300m and 490m; and

Decreasing direction 110m, 200m and 350m.

The performance of the Brynderwyn south curve site in terms of its measured coefficient of friction over time and as
measured by the GripTester device is shown in Figure 6-17. Table 6-5 summarises the descriptive section statistics
data.

167

Chapter 6
Brynderwyn South Curve Site (Both Directions)
1.00

100
Skid Tester not working
properly during this period

80

0.70

70

0.60

60

0.50

50
New seal, 16-Mar-2004,
RACKR, chip size:2/4, PSV:53

New seal, 24-Mar-2003,

TWORS, chip size:4/6, PSV:56

0.40

40

21/04/2005

21/02/2005

21/03/2005

21/01/2005

21/11/2004

21/12/2004

21/10/2004

21/08/2004

21/09/2004

21/07/2004

21/05/2004

21/06/2004

21/04/2004

21/02/2004

21/03/2004

21/12/2003

21/01/2004

21/11/2003

21/09/2003

0
21/10/2003

10
21/08/2003

0.10
0.00
21/06/2003

20

21/07/2003

0.20

21/05/2003

30

21/03/2003

0.30

Surface Temperature (C)

90

0.80

21/04/2003

GripNumber (GN)

0.90

Time
GN R1-R5 Dec

GN R1-R5 Inc

Surface Temperature

Figure 6-17: Brynderwyn South Curve Site Summary results over time (Both directions)

Brynderwyn South Curve Site (Both Directions)

Sample Section
Descriptive Statistics

Sect Ave Sect Ave Sect Ave Sect Ave

Timeslice
Temperature (C)
GN Mean GN Mean GN Mean GN Mean GN Mean GN Mean Ambient
Surface
W/Up INC R1-R5 INC W/Up DEC R1-R5 DEC 490m INC 110m DEC (Start)
(Start)
Sample count
28
29
28
29
26
28
29
29
Mean
0.60
0.63
0.54
0.57
0.57
0.55
18.9
21.9
Maximum
0.73
0.83
0.68
0.70
0.64
0.69
27.0
38.0
Minimum
0.45
0.44
0.39
0.44
0.49
0.45
11.5
9.0
Range
0.28
0.39
0.29
0.25
0.16
0.24
15.5
29.0
Standard Deviation
0.074
0.082
0.067
0.058
0.041
0.054
4.2
7.6
95% Confidence Interval
0.027
0.030
0.025
0.021
0.016
0.020
1.5
2.8
Variance
0.005
0.007
0.005
0.003
0.002
0.003
17.3
57.6
Coefficient of Variation (CoV)
0.123
0.132
0.126
0.102
0.073
0.098
0.220
0.346

Table 6-5: Brynderwyn South curve site Summary Statistics (both directions)
Due to a high degree of transverse and longitudinal polishing on the sharp curve, the site has a high demand for
friction largely due to the high operating speeds in comparison to the safe design speed. This has led to relatively
frequent overlaying of chipseal surfaces with aggregate that cannot naturally withstand the high polishing demands
required of it. Subsequently, it was not expected that the site would display seasonal type variations over time, but
remain in a typical initial negative exponential polishing phase from an initial value down to an equlibrium level.
The length of time interval required for this polishing to occur was investigated as part of the skid resistance testing
programme.
Statistical analysis of the processed data (refer to Table 6-5), with identified outliers removed, shows that the
Brynderwyn site with a maximum of 29 test surveys, had a sample mean (x) of 0.63 GN and 0.57GN (the warmup
runs were not included in this analysis) for the increasing and decreasing directions, respectively. The standard
deviation of the averaged GripTester section survey results (including the previous old seal surfaces) was 0.082 GN
and 0.058 GN and the Coefficient of Variation (CoV) was 13.2% and 10.2% for the increasing and decreasing
directions, respectively. These statistical variation descriptors would be expected to be higher than on a site that was
in it is second seasonal variation phase after initial polishing was completed such as the Hikurangi site. The data
resulted in a 95% confidence value that was 0.030 (increasing) and 0.025 (decreasing). Thus, for example, there
168

Chapter 6
was a 95% probability that the mean GN value would fall between the values of 0.60 GN and 0.66 GN at a 50km/h
testing speed, for the increasing direction.
The averaged GripTester device section results show (refer to Figure 6-17) that, generally, the measured skid
resistance in the increasing direction was higher than in the decreasing direction by up to 0.06 GN. This is
especially apparent for the first seal constructed in March 2003. The differences could possibly be attributed to a
probable lower approach operating speed in the increasing direction in comparison to higher approach speeds (and
therefore higher polishing forces) in the decreasing direction. This also aligns with Land Transport New Zealand
crash statistics that have reported a higher crash rate in the decreasing direction than in the increasing direction.
Figure 6-17 also shows an attempt to visually demonstrate the averaged initial polishing phase from the measured
skid resistance data and the polishing factors involved.

Of significance, is how quickly the newly surfaced

aggregate polishes to a level that is very similar to that prior to the surface construction. The first two-coat surface
seal lasted only approximately three months before the skid resistance reduced to a level as low or lower than the
previous seal. The aggregate used from Bellingham s quarry had a reported PSV of 56, and the results show that it
was not polish-resistant enough to prevent the polishing of the microtexture. An analysis/calculation of the TNZ
T/10 PSV equation as an established method predicts that an aggregate PSV of approximately 60 should have been
used before the addition of any PSV stress points for this TNZ T/10 Site Category 2 curve. The previous history of
known aggregate polishing and the volume of HCV could be utilised to determine how many additional PSV points
should have been added to the calculated PSV equation above to determine an apropriate aggregate to use.
Unfortunately, the problems that arose with the GripTester in the summer of 2003 / 2004 meant it was not known
what occurred in the summer months where the lowest results were expected. The expectation was (and the last few
data points tended to indicate this trend) that the measured coefficient of friction would reduce further in the summer
months. In fact, on-site visual surveys during this period showed noticeable flushing in the increasing wheel track
direction during these hotter summer months due to multiple layers of unstable seals. The SCRIM survey on the
18th December 2003 adequately demonstrated this effect of losing macrotexture by returning an SFC result of 0.29
in the increasing direction (well below the Threshold Level (TL) of 0.40 SFC required for the site) and a 0.53 SFC
in the decreasing direction. This effect can be seen with the last few credible GripTester reported results in Figure
6-17 where the increasing coefficient of friction (GN) sharply decreases to the same level as the decreasing
direction. The results after this point, however, unfortunately coincide with the period where problems occurred
with the GripTester and therefore the results are not credible.
The surface was again resurfaced in March 2004, this time with a racked-in seal with a PSV of 52, with much closer
attention to temporary traffic control to ensure low speeds and proper embedment of the chip during construction.
The initial effect of the treatment and then its rate of reduction over the ensuing months is again evident on Figure
6-17, although the rate of polishing is somewhat less than the previous seal. This is surprising, as it was expected
that a lower reported aggregate PSV would polish at a faster rate than a higher reported PSV. However, by the
following summer, some 9-10 months later, the skid resistance values had reduced to a level where the SCRIM and
GripTester surveys in December 2004, and then later in May 2005, showed it was right on the boundary of the
Threshold Level (TL)19, the trigger level for determining priority for treatment. However, the SCRIM and
GripTester surveys had shown a slight increase in the surveyed results from the December 2004 surveys to the May
2005 surveys, indicating that perhaps this was the equilibrium level of polishing at the site, and that seasonal
fluctuations were now occurring around this level and that further reductions would not occur. However, according
to the TNZ T/10 specification, this again triggered priority for treatment as the measured coefficient of friction

169

Chapter 6
levels was well below the Investigatory Level 20, and on the border of the Threshold Level , which is set at 0.1
SFC below the Investigatory Level.
The total variation in measured skid resistance at the Brynderwyn site (approximately 13% over the 2 year period) is
higher than the other reported field sites. This is relatively high, as would be expected, due to the high polishing
demand in the initial polishing phases from the low radii curve and the high operating speeds.

6.4.6 Hikurangi Site

The Hikurangi site was chosen as a control seasonal site and is the third of five Northland State Highway test sites.
Its physical attributes are described in Section 5.7.5 and shown in Figure 5-9. The site is on State Highway 1, north
of Whangarei, immediately south of the Hikurangi township and is comprised of a straight and level section with a
chipseal surface that is more than five years old. The macrotexture levels have also been stable over time and the
site had also been used as a SCRIM seasonal site over the previous few years.
Skid resistance measurements with the GripTester began in March 2003 and continued through to May 2005 in both
the increasing and decreasing directions. Annual Average Daily Traffic Volumes (AADT) are recorded in the
RAMM database as being 9700 vpd with a heavy commercial vehicle (HCV) composition of 9.7%. The section
operates under a 100km/h posted speed limit.

A defined longitudinal wheel path, and therefore regular

tyre/aggregate contact area and polishing mechanism, occurred at this site and therefore any variation in measured
skid resistance due to transverse location of the measuring wheel was expected to be minimal. Average skid
resistance test results were obtained for each test date after post processing, by calculating the sample mean value of
test runs one to five (excluding the warm-up run) for the following sections:

Increasing direction running distance 160m to 450m; and

Decreasing direction running distance 250m to 550m.

The performance of the Hikurangi site in terms of its measured coefficient of friction over time and as measured by
the GripTester device is shown in Figure 6-18. Table 6-6 summarises the descriptive section statistics data.

19
20

The Threshold Level (TL) is currently set at 0.1 below the IL and is the trigger level for determining priority for treatment.
The Investigatory Level (IL) is the level of skid resistance at or below which a site investigation is to be undertaken, and the
information used as a priority for programming treatment.

170

Chapter 6
Hikurangi Control Site (Both Directions)

0.70

90
80

0.60

70

0.50

60
Example of short-term variation
of 0.17GN over one month

0.40

50
40

0.30

30

0.20

20

21/04/2005

21/03/2005

21/02/2005

21/01/2005

21/12/2004

21/11/2004

21/10/2004

21/09/2004

21/08/2004

21/07/2004

21/06/2004

21/05/2004

21/04/2004

21/03/2004

21/02/2004

21/01/2004

21/12/2003

21/11/2003

21/10/2003

21/09/2003

21/08/2003

21/07/2003

0
21/06/2003

0.00
21/05/2003

10
21/04/2003

0.10

21/03/2003

GripNumber (GN) / DFT ( )

100

Skid Tester not working

properly during this period

Surface Temperature (C)

0.80

Time
GN R1-R5 Decr

GN R1-R5 Inc

DFT20 ( ) Inc

Surface Temperature

Figure 6-18: Hikurangi Control Site Summary results over time (Both directions)
Hikurangi Control Site (Both Directions)
Sample Section
Sect Ave Sect Ave Sect Ave Sect Ave
Ave
GN Mean GN Mean GN Mean GN Mean
Descriptive Statistics
DFT 20
( )
W/Up INC R1-R5 INC W/Up DEC R1-R5 DEC
Sample count
32
32
32
32
8
Mean
0.47
0.53
0.51
0.56
0.62
Maximum
0.61
0.72
0.64
0.73
0.73
Minimum
0.34
0.41
0.42
0.46
0.55
Range
0.27
0.31
0.21
0.27
0.17
Standard Deviation
0.068
0.069
0.063
0.066
0.069
95% Confidence Interval
0.024
0.024
0.022
0.023
0.048
Variance
0.005
0.005
0.004
0.004
0.005
Coefficient of Variation (CoV)
0.146
0.130
0.124
0.117
0.110

Temperature (C)
Ambient
Surface
(Start)
(Start)
32
32
21.7
26.2
34.5
44.0
11.5
14.0
23.0
30.0
5.3
8.9
1.8
3.1
27.6
78.6
0.243
0.339

Table 6-6: Hikurangi site Summary Statistics (both directions)

The Hikurangi site was expected to display seasonal type variations over time as the site was stable and over five
years in age. The site was chosen on the basis that if a sinusoidal seasonal pattern of skid resistance variation did
exist, then this site was expected to show this effect. However, as can be seen from the section averaged results in
Figure 6-18, the skid resistance levels were surprisingly variable and clearly not predictable, although initial
aggregate polishing phases were plainly completed prior to the data collection period. As was consistent with other
testing sites, there was an increase in GN values in the second year of testing that coincided with replacing
significant parts of the GripTester device in the summer between the two years of data. There is also some
evidence (refer to Figure 6-18) that an approximately seasonal variation effect was displayed at the site, resulting in
a lower measured coefficient of friction in March and April in comparison to higher results in August, September
and October. The averaged GripTester device section results also show that generally, the measured skid resistance
in the increasing direction is marginally lower than in the decreasing direction.
Of surprise was how much the measured coefficient of friction varied over such a short period of time. Figure 6-18
shows the results of the skid resistance monitoring which shows that more than 30% variation can occur (a
difference of 0.17 GN or approx 0.13 SFC) within one month. A check of periodic macrotexture measurements
171

Chapter 6
over the two year survey period shows that this remained reasonably constant and could therefore not explain the
differences in skid resistance measurement. It must therefore be concluded that this skid resistance variation was
due either to contamination of the surface and/or real microtexture changes during this period. The amount of this
variation is of considerable concern to road network managers as sites that have been surveyed and passed, can fail
within weeks, and sometimes even days, later.
Statistical analysis of the processed data (refer to Table 6-6), with the identified outliers removed, shows that the
Hikurangi site with a maximum of 32 test surveys, had a sample mean (x) of 0.53 GN and 0.56 GN for the
increasing and decreasing directions respectively. The standard deviation of the averaged GripTester section survey
results was 0.069 GN (increasing) and 0.066 GN (decreasing) and the coefficient of variation (CoV) was 13.0% and
11.7% for the western and eastern directions respectively. The data resulted in a 95% confidence value that was
0.024 and 0.023, meaning that there was a 95% probability that the mean GN value, for example, would fall between
the values of 0.50 GN and 0.55 GN in the increasing direction at a 50km/h testing speed.
The total variation in measured skid resistance at the Hikurangi site is greater than at all of the other field test sites.
This was an unexpected result as the Hikurangi site should only be displaying seasonal variations after an initial
equilibrium level has been reached. All other Northland sites (Brynderwyn, Kaiwaka, Kara and SnooksTatton) had
sections that had new chip-seal surfaces, and it was expected that these sites would have demonstrated greater
standard deviations and coefficients of friction over the data collection period due to inclusion of initial polishing
phases. If anything, the Hikurangi site results have shown that skid resistance has in total increased over the testing
period.
The Hikurangi site was also tested whenever traffic control was possible with the DF Tester at three defined
locations. Only eight different test survey dates were possible during the survey data period and only in the
increasing direction. The DF Tester test data resulted in a sample mean (x) ranging from 0.62 ( ) and a sample
standard deviation ( ) of 0.069. The coefficient of variation (CoV) between the DF Tester test dates at the eight test
spot locations was 11.0% and this corroborates with the GripTester data and other DF Tester field sites in the
Auckland region, that demonstrated that this site had considerably more variation than other field test sites. The
results of the data analysis show that there is a 95% probability that the mean DF Tester ( ) result will fall between
the values of 0.57 and 0.67.
A comparison was made between the Hikurangi site variation data (up to 14.7% CoV) and the middle section on the
SnooksTatton Road site, Section 2 (discussed in section 6.4.8) with an increasing direction CoV of 8.1%). The
sites were similar in that they were sections that were at least six years old and that should only display seasonal
variations . This confirms that the Hikurangi site is unusually high in terms of its measured variation in skid
resistance. Whilst it is difficult to understand exactly what is causing the greater degree of skid resistance variation
at this site, a possible site specific explanation could be that there is a lime quarry north of the site. It is possible that
some of the variability in measured skid resistance is due to either contamination and/or variable traffic loading /
skid resistance / rainfall rejuvenation cycles.

6.4.7 Kara Road Site

The Kara Road test site is the fourth of five Northland State Highway test sites. Its physical attributes are described
in Section 5.7.6 and shown in Figure 5-10. The site is on State Highway 14 and is approximately 8.5 km west of
Whangarei. It consists of a series of reverse horizontal curves of radii as low as 150m to 160m and is therefore
categorised under TNZ T/10 (Transit New Zealand, 2002) specification as a TNZ Site Category 2 section. The site
was resurfaced in January 2003 with a racked-in Grade 3/5 chipseal surface with an aggregate sourced from Otaika
172

Chapter 6
quarry with a reported PSV of 51/52. Skid resistance measurements with the GripTester began in March 2003.
Annual Average Daily Traffic Volumes (AADT) are recorded in the RAMM database as being 5500 vpd with a
heavy commercial vehicle (HCV) composition of 5.5%. The section operates under a 100km/h posted speed limit.
The site was tested approximately monthly between March 2003 and April 2005 in both the increasing (southbound)
and decreasing (northbound) directions. A defined longitudinal wheel path and therefore regular tyre/aggregate
contact area and polishing mechanism occurred at this site. However, as the horizontal curve radii of the reverse
curves are reasonably constrained, the transverse wheel tracking width would be expected to be somewhat wider
than on a straight section of road.
The field test site was split into two sections, one predominantly on a flat grade and the other predominantly on a
grade of 7.2%. Average skid resistance test results were obtained for each test date after post processing by
calculating the sample mean value of test runs one to five (excluding the warm-up run) for the following sections:

Section 1 (Increasing direction) running distance 50m to 170m;

Section 2 (Increasing direction) running distance 210m to 570m;

Section 1 (Decreasing direction) running distance 50m to 460m;

Section 2 (Decreasing direction) running distance 470m to 590m.

Timeslice locations were also chosen, processed and analysed on the Kara Road site. The timeslice locations are
identified below and shown on Figure 5-10:

Increasing direction 40m, 120m, 220m, 450m, 520m, 600m, and 650m;

Decreasing direction 30m, 110m, 180m 410m, 450m, 570m and 610m.

The performance of the Kara Road site in terms of its measured coefficient of friction over time and as measured by
the GripTester device is shown in Figure 6-19. Table 6-7 summarises the descriptive section statistics data.
Kara Road S1 & S2 (Both Directions)
1.00

100
Skid Tester not working
properly during this period

0.80

80
70

0.60

60

0.50

50

0.40

40

31/03/2005

28/02/2005

31/01/2005

31/12/2004

30/11/2004

31/10/2004

30/09/2004

31/08/2004

31/07/2004

30/06/2004

31/05/2004

30/04/2004

31/03/2004

29/02/2004

31/01/2004

31/12/2003

30/11/2003

31/10/2003

0
30/09/2003

0.00
31/08/2003

10
31/07/2003

0.10
30/06/2003

20

31/05/2003

30

30/04/2003

0.30
0.20

Surface Temperature (C)

90

0.70

31/03/2003

GripNumber (GN)

0.90

Time
GN R1-R5 S1 Inc

GN R1-R5 S2 Inc

GN R1-R5 S1 Dec

GN r1-R5 S2 Dec

Surface Temperature

Figure 6-19: Kara Road Site Summary results over time (Both directions)

173

Chapter 6

Kara Road Site (Both Directions)

Sample Section
Descriptive Statistics
Sample count
Mean
Maximum
Minimum
Range
Standard Deviation
95% Confidence Interval
Variance
Coefficient of Variation (CoV)

Sect Ave
GN Mean
S1 INC
26
0.67
0.78
0.60
0.18
0.043
0.016
0.002
0.063

Sect Ave
GN Mean
S2 INC
26
0.67
0.78
0.61
0.18
0.044
0.017
0.002
0.065

Sect Ave
GN Mean
S1 DEC
26
0.64
0.72
0.55
0.17
0.043
0.016
0.002
0.067

Sect Ave
GN Mean
S2 DEC
26
0.63
0.72
0.51
0.21
0.046
0.018
0.002
0.073

Timeslice
Temperature (C)
GN Mean GN Mean Ambient
Surface
40m INC 650m DEC (Start)
(Start)
26
25
26
26
0.66
0.65
20.1
26.4
0.79
0.76
27.0
41.0
0.55
0.55
10.8
10.0
0.24
0.21
16.3
31.0
0.058
0.058
4.3
9.0
0.022
0.023
1.6
3.4
0.003
0.003
18.1
80.3
0.089
0.089
0.211
0.340

Table 6-7: Kara Road site Summary Statistics (both directions)

The Kara Road site was not expected to display seasonal type variations over time as the site had recently been
sealed prior to skid resistance monitoring. It was thought that an initial polishing phase would continue, for the first
year or so, as HCV volumes were approximately one third that of State Highway 1 HCV volumes. Hence, the
measured skid resistance would decrease from an initial value down to an equilibrium level. This decreasing effect
is evident in Figure 6-19, where the skid resistance levels for all section averages generally reduce over the data
collection period, although at a lesser rate than on the Brynderwyn South curve site which has three times the
amount of HCVs. As was consistent with other testing sites, there was an increase in GN values in the second year
of testing that coincided with replacing significant parts of the GripTester device in the 2003/2004 summer between
the two years of data. There is also visible evidence from the second year of data (refer to Figure 6-19) that a
seasonal variation effect occurs at the site, resulting in a lower measured coefficient of friction in March and April in
comparison to higher results in September and October.
Statistical analysis of the processed data (refer to Table 6-7), with the identified outliers removed, shows that the
Kara Road site with a maximum of 26 test surveys, had a sample mean (x) of 0.67 GN for both sections 1 and 2 of
the increasing direction. In the decreasing direction, a mean (x) of 0.64 GN and 0.63 GN was obtained for Sections
1 and 2, respectively (the warmup runs were not included in this analysis). The standard deviations for all sections
and directions were very consistent, the results ranging between 0.043 GN and 0.046 GN. The coefficient of
variation (CoV) ranged from 6.3% to 7.3%. The data resulted in a 95% confidence value that ranged from 0.016 to
0.018, meaning that there was a 95% probability that the mean GN value would fall, for exampe, between the values
of 0.65 GN and 0.69 GN on Section 1 in the increasing direction at a 50km/h testing speed.
The averaged GripTester device section results show that generally the measured skid resistance in the increasing
direction is marginally higher than in the decreasing direction. The total variation in measured skid resistance at the
Kara Road site is the lowest of the chip seal surfaces in Northland, however not as low as the asphalt mix surface at
the Ports of Auckland.

6.4.8 Snooks Tatton Site

The Snooks - Tatton Road field test site is the last of the Northland State Highway test sites. Its physical attributes
are described in Section 5.7.7 and shown in Figure 5-11. The site is the longest field site (1700m) and is separated
into three sections, two sections that were resealed in Jan 2003, a few months prior to skid resistance monitoring
commenced at the site, separated by a middle 300m section that remained with the old seal surface which was
constructed in February 2000. The site is also on State Highway 14 and is approximately 5.5 kms further west than
the Kara Road field site and 13.9kms west of Whangarei. It consists of a series of reverse horizontal curves of radii
as low as 140m to 160m and is therefore also categorised under TNZ T/10 (Transit New Zealand, 2002)
174

Chapter 6
specification as a TNZ Site Category 2 section. The two newly sealed sections were resurfaced with a two-coat
Grade 3/5 chipseal with aggregate sourced from Otaika quarry with a reported PSV of 51/52. Skid resistance
measurements with the GripTester began in March 2003. Annual Average Daily Traffic Volumes (AADT) are
recorded in the RAMM database as being 5500 vpd with a heavy commercial vehicle (HCV) composition of 5.5%.
The section operates under a 100km/h posted speed limit and is level. The site was tested approximately monthly
between March 2003 and April 2005 in both the increasing (southbound) and decreasing (northbound) directions. A
defined longitudinal wheel path and therefore regular tyre/aggregate contact area and polishing mechanism occurred
at this site. However, as the horizontal curve radii of the reverse curves are reasonably constrained, the transverse
wheel tracking width would be expected to be somewhat wider than on a straight section of road.
As discussed above, the field test site was split into three sections separated by the change in seals. The average
skid resistance test results were obtained for each test date after post processing, by calculating the sample mean
value of test runs one to five (excluding the warm-up run) for the following sections:

Section 1 (Increasing direction) running distance 120m to 590m;

Section 2 (Increasing direction) running distance 610m to 930m;

Section 3 (Increasing direction) running distance 950m to 1690m;

Section 1 (Decreasing direction) running distance 140m to 850m;

Section 3 (Decreasing direction) running distance 1220 to 1690m.

The section 2 in the decreasing direction was not analysed as this section had a few recent routine maintenance
patches which had considerably modified the chipseal skid resistance performance. Timeslice locations were also
chosen, processed and analysed on the Snook-Tatton site. The timeslice locations are identified below and the
increasing direction is shown on Figure 5-11):

Increasing direction 80m, 300m, 630m, 1100m, 1280m, 1450m, 1600m and 1740m;

Decreasing direction 60m, 200m, 350m 520m, 700m, 1170, 1500 and 1720m.

The performance of the Snooks Tatton Road skid resistance site in terms of its measured coefficient of friction
over time and as measured by the GripTester device is shown in Figure 6-20. Table 6-8 summarises the descriptive
section statistics data.
Snooks - Tatton Road S1, S2, S3 (Both Directions)
0.90

90

31/03/2005

28/02/2005

31/01/2005

31/12/2004

30/11/2004

31/10/2004

30/09/2004

31/08/2004

31/07/2004

30/06/2004

0
31/05/2004

10

0.00
30/04/2004

0.10
31/03/2004

20

29/02/2004

30

0.20

31/01/2004

0.30

31/12/2003

40

30/11/2003

0.40

31/10/2003

50

30/09/2003

60

0.50

31/08/2003

0.60

31/07/2003

70

30/06/2003

0.70

31/05/2003

80

30/04/2003

0.80

31/03/2003

GripNumber (GN)

100

Skid Tester not working

properly during this period

Surface Temperature (C)

1.00

Time
GN R1-R5 S1 Inc

GN R1-R5 S2 Inc

GN R1-R5 S3 Inc

GN R1-R5 S1 Dec

GN R1-R5 S3 Dec

Surface Temperature

Figure 6-20: Snooks Tatton Site Summary results over time (Both directions)
175

Chapter 6
Snooks - Tatton Road Site (Both Directions)
Sample Section
Sect Ave Sect Ave Sect Ave Sect Ave Sect Ave Timeslice Temperature (C)
GN Mean GN Mean GN Mean GN Mean GN Mean GN Mean Ambient
Descriptive Statistics
Surface
S1 INC
S2 INC
S3 INC
S1 DEC
S3 DEC 700m DEC (Start)
(Start)
Sample count
24
24
24
24
24
20
24
24
Mean
0.69
0.57
0.70
0.67
0.69
0.56
20.4
24.6
Maximum
0.80
0.66
0.81
0.75
0.78
0.69
48.0
47.0
Minimum
0.61
0.47
0.64
0.60
0.61
0.48
9.0
12.0
Range
0.18
0.18
0.17
0.16
0.17
0.21
39.0
35.0
Standard Deviation
0.051
0.046
0.046
0.042
0.045
0.057
7.5
9.5
95% Confidence Interval
0.021
0.019
0.018
0.017
0.018
0.025
3.0
3.8
Variance
0.003
0.002
0.002
0.002
0.002
0.003
56.3
90.8
Coefficient of Variation (CoV)
0.074
0.082
0.065
0.063
0.065
0.101
0.368
0.388

Table 6-8: Snooks - Tatton Road site Summary Statistics (both directions)
The Snooks-Tatton Road site (sections one and three) was not expected to display seasonal type variations over
time, as the site had been sealed in Jan 2003, a few months prior to skid resistance monitoring. However, Section 2
in the increasing direction was over three years old and was expected to demonstrate seasonal variations, as the seal
would be at an equilibrium level. As with the Kara Road site, the number of HCVs was approximately one third
that of State Highway 1 at the Kaiwaka, Brynderwyn and Hikurangi sites. Figure 6-20 shows that the individual
sections behaved very similarly over the data collection period. However, the newly chip-sealed surface was clearly
but slowly reducing and thereby converging towards the middle Section 2 skid resistance levels. As was consistent
with other testing sites, there was an increase in GN values in the second year of testing that coincided with
replacing significant parts of the GripTester device in the 2003/2004 summer between the two years of data. There
is also visible evidence from the second year of data (refer to Figure 6-20) that a seasonal variation effect occurs at
the site, resulting in a lower measured coefficient of friction in March and April in comparison to higher results in
September and October. The averaged GripTester device section results show that, generally, the measured skid
resistance in the increasing direction is marginally higher than in the decreasing direction.
Statistical analysis of the processed data (refer to Table 6-8), with the identified outliers removed, shows that the
two newly surfaced Snooks-Tatton Road site sections, with a maximum of 24 test surveys, had a sample mean (x)
ranging from 0.67 GN to 0.70 GN (the warmup runs were not included in this analysis). The middle older seal
section had a sample mean (x) of 0.57. The standard deviations of the averaged GripTester section survey results
for all sections ranged between 0.042 GN and 0.051 GN and the Coefficient of Variation (CoV) ranged from 6.3% to
8.2%. The data resulted in a 95% confidence value that ranged from 0.017 to 0.021, meaning that there was a 95%
probability that the mean GN value would fall between the values of 0.65GN and 0.72 GN at a 50km/h testing speed
in either direction.

6.5

Aggregate PSV and Measured Skid Resistance

The skid resistance field sites in the Northland Region were selected for differing purposes and/or characteristics.
One of the characteristics which varies for each of the sites is the actual surface treatment that was utilised after a
surface treatment was required (refer to Table 5-2).
Figure 6-21 shows the initial skid resistance measured after each of the surfaces were treated and how skid
resistance performed over repeated traffic loads as a function of the Transit NZ Published Polished Stone Value
(PSV). The figure demonstrates a number of important points:

That a higher published PSV aggregate does not necessarily lead to a high initial skid resistance value as
measured by the GripTester;

176

Chapter 6
The published PSV of the aggregate does not necessarily determine the level of equilibrium skid resistance
(however, the sites do have varying loading and geometric elements which will effect the final level). This
result tends to corroborate with research by Cenek et al. (2003b) that states there is very little relationship
between the PSV of the aggregate and the in-field skid resistance as measured by SCRIM network surveys
(even on straight and level sections of road);

The highest initial Grip Number (GN) is recorded at the Brynderwyn Curve. This is the only site of the five
that was treated during the monitoring period with a Grade 2 and 4 chip size. The other sites were treated
with Grade 3 and 5 chip. This result is not unexpected as the larger chips with higher macrotexture and the
same microtexture are expected to produce the highest initial skid resistance;

For similar traffic loadings and curvature (eg Kara Road and Snooks Tatton) approximately the same range
of skid resistance was lost over the same trafficking period;

The above losses in skid resistance for each of the sites must be placed within the context of the control site
which over the same two year period increased by approximately 0.1 GN for the highest traffic loading;

Whilst the iron making melter slag from the Glenbrook Steel Mill has the highest published PSV of 58
compared with a PSV of 52 and 53 for the other sites, its initial Grip Number of 0.7 is the second lowest of
the results.

However, when one looks at the skid resistance performance of the slag over time in

comparison to the other aggregates and sites, even with much higher loading, its performance is
significantly better.

Grip Number versus PSV of Chip

GripNumber (GN)

RACKR GR24 (PSV 53)

0.8
Kara Rd
HCV load =
22.3x10^5

0.6
Hikurangi
HCV load =
70.7x10^5

Kaiw aka Slag

HCV load =
54.6x10^5
Snooks-Tatton
HCV load =
22.3x10^5
Brynderw yn 03
HCV load =
17.7x10^5

0.4
50

TWORS GR35 (PSV 52)

Kara 03
RACKR GR35 (PSV 52)
Snooks - Tatton 03
TWORS GR 46 (PSV 56)
Brynderwn 04
SC G2 (PSV 52)

Brynderw yn 04
HCV load =
28.5x10^5

55
Transit NZ
Published PSV
PSV

Brynderwyn 03
SLAG (PSV 58)
Kaiwaka SLAG 03

60

Hikurang 99

Figure 6-21: Measured Skid resistance over Time in Comparison to Published PSV for Northland Sites
From the above analysis it is indicated that the published Transit NZ Polished Stone Value (PSV) of the aggregate in
itself, without taking other factors into consideration, cannot be reliably used as a predictor of the the initial skid
resistance of the aggregate and or the level of equilibrium skid resistance after polishing. Other methods of
predicting how aggregates will perform over time and under specific traffic, geometric and braking stresses are
required for road asset managers to appropriate good decision making.

6.5.1 Field Testing Summary Discussion

Seven field sites were regularly tested (approximately monthly) with the GripTester device, for over two years and
three of the sites were tested with the Dynamic Friction Tester for up to 14 months. Table 6-9 summarises the
177

Chapter 6
results from the two skid resistance devices and the seven field sites over approximately two years of data
collection.
Whilst it was originally thought that the extent of the variation in the GripTester results would be less for surfaces
that had reached an equilibrium level of polishing (experiencing only seasonal and or short-term variations) in
comparison to sites that were still in the initial polishing phase from a new surface level, it is surprising that the
results have not shown this. In fact, the results have shown that in terms of the coefficient of variation (CoV), the
reverse has occurred, and greater variation occurs on the sites that are supposedly at a stable equilibrium level of
polishing. On closer examination of the Table 6-9 results, the coefficient of variation has increased on these sites
because the mean skid resistance coefficient of friction value is lower, however the standard deviations of the results
have remained reasonably consistent between the sites (although ranging from 0.043 on the roads with lower HCV
volumes (SH 14) to 0.069 on roads with approximately three times the volume of HCV s (SH 1)).
The asphalt mix surfaces in the Auckland Region (Tamaki Campus site with light traffic conditions, and the Ports of
Auckland site with heavy traffic conditions) had a coefficient of variation (CoV) of GripTester results of 7.4% and
10.9%, and a 95% confidence interval of 0.020 GN and 0.028 GN respectively. It should be noted that there was
also a difference between the sites in terms of testing speed which could account for some of the additional variation
on the Ports of Auckland site that was tested at the higher speed.
Field Test Site
Ave of both directions 1

No. of
Data
Points

CoF
Mean
range

Ave
Standard
Deviation

95%
Conf
Interval

Variance

Coeff of
Variati
on

Comments on
Polished state

Tamaki Campus GT30 2

31-32

0.76-0.80

0.057

0.020

0.003

7.4%

Nearly new seal

Tamaki Campus DFT20 3

11

0.67-0.73

0.034

0.019

0.001

4.8%

Nearly new seal

Ports of Auckland GT50

23-24

0.61-0.63

0.067

0.028

0.004

10.9%

Equilibrium

Ports of Auckland DFT20

14

0.49-0.55

0.050

0.027

0.003

9.5%

Equilibrium

Kaiwaka Slag Site GT50

28

0.70-0.72

0.056

0.021

0.003

7.9%

New seal layers

Brynderwyn South GT50

29

0.63-0.57

0.070

0.026

0.005

11.7%

New seal layers

Hikurangi GT50

32

0.53-0.56

0.068

0.024

0.005

12.4%

Equilibrium

Hikurangi DFT20

0.62

0.069

0.048

0.005

11%

Equilibrium

Kara Road S1 GT50

26

0.67-0.64

0.043

0.016

0.002

6.5%

New seal layers

Kara Road S2 GT50

26

0.67-0.63

0.045

0.018

0.002

6.9%

New seal layers

Snooks-Tatton S1 GT50

24

0.69-0.67

0.047

0.019

0.003

6.9%

New seal layers

Snooks-Tatton S2 GT50
Increasing direction

24

0.57

0.046

0.019

0.002

8.2%

Equilibrium

Snooks-Tatton S3 GT50

24

0.70-0.69

0.046

0.018

0.002

6.5%

New seal layers

Notes:
1)
All sites are summarised for both directions of travel unless otherwise specified.
2)
GT30 denotes GripTester CoF tested at 30km/h.
3)
DFT20 denotes Dynamic Friction Tester CoF at a slip speed of 20km/h.
Table 6-9: Summary Coefficient of Friction Statistics of the Field Sites
The CoV reported from the GripTester results of the chip seal sites in the Northland region also seem to follow the
heavy commercial vehicle loadings at the site more than the polishing phase of the aggregate. For instance, the CoV
178

Chapter 6
of the Kaiwaka, Brynderwyn and Hikurangi sites (all on State Highway 1) were 7.9%, 11.7% and 12.4%
respectively with approximately three times higher commercial vehicle volumes than the Kara Road and Snooks
Tatton sites (on State Highway 14) with a range of 6.5% to 8.2% CoV. The Kaiwaka site CoV results were lower
than the other two State Highway 1 results and it is surmised that this is a result of being surfaced with an artificial
melter slag aggregate that was more resistant to polishing and therefore variation in measured results, than natural
aggregates used on the other sites. The 95% confidence values of the site means generally followed the same trend
of CoV results for the Northland sites with higher ranges for the three State Highway 1 results (0.021, 0.026 and
0.024 GN for the Kaiwaka, Brynderwyn and Hikurangi sites) in comparison to the State Highway 14 results that
ranged from 0.016 to 0.019 GN.
Also of significance, is the relatively narrow range and consistency between all of the sites of the standard deviation,
95% confidence interval and the variance of the mean data values over the data collection period.
It is difficult to establish any significant trends for the coefficient of friction results utilising the Dynamic Friction
Tester as only three sites were tested, they were less regular and the number of data points is fewer. However, it is
pleasing to see that the DF Tester results demonstrated the same relativity between sites in terms of the CoV as the
GripTester for the Tamaki, Ports of Auckland and Hikurangi sites (e.g. DFT20 CoV=4.8% c.f. GT50 CoV=7.4% for
the Tamaki site, DFT20 CoV=9.5% c.f. GT50 CoV=10.9% for the Ports of Auckland site and DFT20 CoV=11.0%
c.f. GT50 CoV=12.4% for the Hikurangi site). This gives some confidence that the two skid resistance devices are
independently measuring the same effects at each site.
Further discussion on the repeatibility of the skid resistance testing devices that have been used in this research and
those that are commonly used in New Zealand, is presented in Chapter 7.

179

Chapter 7

7 REPEATABILITY OF SKID RESISTANCE TESTING DEVICES

7.1

Introduction

The repeatability of a skid resistance measuring device is dependent upon a number of important factors:

The method of measurement (i.e. dynamic and continuous or static and stationary test);

The reliability of the device in terms of electronic sensors;

The testing line (i.e. longitudinal in-line or circular);

The environmental factors (i.e. level of detritus, temperature effects);

Linear referencing methods and post processing of data;

The driver operator s ability, with continuously towed measurement methods, to test the same line of
testing.

These factors are very difficult to separate from each other and, combined together, determine the variability of the
device. A device s sensitivity to each of these factors is very important in terms of understanding the differences
between the various devices.

The following section discusses the repeatability of the GripTester device in

comparison to the Dynamic Friction Tester device. These are the two skid resistance devices primarily used in this
research.

7.2

GripTester

To quantify the variability of the GripTester device, a series of multiple GripTester device runs was undertaken on
the Hikurangi control site in Northland on the 9th December 2004. Previous research by the author (Wilson et al.,
2003) had identified that the first run (the warmup-run) was usually lower than subsequent runs, and that this goes
against established research regarding the operating temperature of the tyre. The developers of the GripTester
(Findlay Irvine Ltd, 2002) had stated that, as long as the GripTester measuring tyre had been conditioned for
approximately 500m in normal testing mode, then temperature and water/rubber testing tyre conditioning effects
were not a significant issue. Hence, unlike other devices, the GripTester would not require a temperature correction
factor. However, when the skid testing procedures were developed for this research, the methodology included
always undertaking one run on each site as a warm-up run which would be rejected in the post-processing analysis.
The reported result was the average of the subsequent five runs.
To consider the effect of this procedure in the context of understanding the repeatability of the GripTester device, a
series of three test runs was undertaken on the same left wheel track at the Hikurangi site in Northland. The ambient
and surface temperatures varied for the three test runs by up to 26C in surface temperature, and the results of post
processing of the data are shown in Figure 7-1. The results clearly show that on this site and on this day the
GripTester run values (irrespective of the recorded temperature range) increased from run to run and approximately
at the same rate. The results also show a surprising and significant difference between the maximum and minimum
recorded GripTester values on the site during the same day, giving a recorded 0.15GN difference. Perhaps more
surprising is that the run values continued to increase in the 13:44pm series run, up to run number eight, before they
levelled off. It was initially thought that maybe the differences between the runs could be explained by the
temperature differences. However, this was not the case, as the lowest recorded temperature gave results that fell
between the other two runs of higher temperatures. A second hypothesis was that the variance between runs could
be explained by a slight cleaning of the pavement between subsequent test runs (recorded rainfall records show that
it had been two days since the last rainfall and four days since greater than five mm of rainfall had fallen).
However, for the hypothesis to be valid, the next series of runs (only a few hours later) would still retain an
180

Chapter 7
increased value due to the cleaner pavement , thereby starting from a higher GN value. It is highly unlikely that the
traffic in the interim period between series runs could have explained the subsequent drop in GN value. Therefore,
the hypothesis is invalid.
Grip Tester Run to Run Repeatability, 9th Dec 2004
Hikurangi site (Increasing Direction)
0.70

Grip Number (GT)

0.65
0.60

Total possible GN
variation in one day
(0.15GN)

Ave Run 1-Run 5

variation over one day
(0.03GN)

0.55

Run 1 GN variation over

one day (0.04GN)
07:20am - Ambient Temp=13.5C
Surface Temp = 18C
11:40am - Ambient Temp=20C
Surface Temp = 38C
13:44pm - Ambient Temp=22.5C
Surface Temp = 44C

0.50
0.45

Warm-up run

0.40
0

10

R1-R5
Average

GripTester Run
07:20am
13:44pm

11:40am
07:20 R1-R5 Average

Figure 7-1: Multiple runs of the GripTester on the same day and same site.
In summary, the primary reason for the significant variation between GripTester device test run results on the same
day and the same wheel track position is inconclusive. The variation cannot be completely explained solely by the
temperature nor solely by a cleaning effect on the pavement. It is unclear at this point whether a combination of
variables could explain this effect. However, what is worth noting is that when one considers the difference
between the series runs, ignoring the warmup runs and the subsequent runs after run numbers six to ten, the
measured variability significantly reduces. As shown in Figure 7-1, the difference between the maximum and
minimum values of the first runs was 0.04 GN and the difference between the averages for runs one to five was a
little less at 0.03 GN. The minimal difference between the means of runs one to five, and the exclusion of the warmup run, gives confidence that the procedure developed for this research is robust.
In order to understand better, the variability of the five runs that make up the average GN test run, an example boxplot has been shown in Figure 7-2 for the Kaiwaka slag field site that shows not only the average of the GripTester
runs (without the warm-up run included) but the maximum and minimum values and the average of the warmup run
itsef. This plot is useful as it demonstrates that the variation between successive GripTester runs generally increases
with the age of the seal. This is shown in Figure 7-3 for the Kaiwaka Slag site where the coefficient of variation is
plotted against the age of the seal surface in days. A reasonably good exponential equation can be fitted to the data
points returning a correlation coefficient (R2) of 0.58. This effect of a greater variation between GripTester runs can
be expected as the seal surface deteriorates at different rates, becoming less homogeneous over time, especially in
the transverse direction, due to the vehicle traffic polishing that occurs in defined wheel tracks. However, an
increased variation or loss of homogeneity also occurs in the longitudinal direction. If the driver was perfectly
repeatable in terms of the line of the GripTester measuring wheel, this variation would be less significant although
would still increase with time.
181

Chapter 7
A closer examination of the results in Figure 7-3 shows that, in addition to the relationship between the age of the
seal and the coefficient of variation, there is also an annual cyclical variation in the coefficient of variation (CoV)
that is superimposed on the age of the seal. This annual seasonal cycle indicates a peak of the greatest variation
between GripTester runs in February / March which drops again in April (although not to the same level as in the
previous year). Reanalysing the exponential relationships for each individual year shows that the correlation
coefficents (R2) can be improved from 0.58 for all data combined to 0.88 and 0.59 for Years 1 and 2 respectively, as
shown in Figure 7-3 and Figure 7-4. The relationship is strong for the first year and weakens with each successive
year.
Kaiwaka Increasing direction - GN Ave, Min and Max
1.00

Skid tester was not working

properly during this period

Grip Number (GN)

0.90
0.80
0.70
0.60

New Seal

0.50

GN Mean of all runs - 1st run

Max Value
Min Value
GN Ave_Warmup Run

0.40

21/04/2005

21/05/2005

21/03/2005

21/01/2005

21/02/2005

21/11/2004

21/12/2004

21/09/2004

21/10/2004

21/08/2004

21/06/2004

21/07/2004

21/04/2004

21/05/2004

21/03/2004

21/01/2004

21/02/2004

21/11/2003

21/12/2003

21/09/2003

21/10/2003

21/08/2003

21/06/2003

21/07/2003

21/05/2003

21/03/2003

21/04/2003

0.30

Date

Figure 7-2: Box-plot of GripNumber (GN) showing Mean, Minimum and Maximum GripTester values in
comparison to the Age of the Seal, Kaiwaka (Increasing Direction)

0.10

February /
March Year 1

0.09
0.08
0.07

February /
March Year 2

Old Seal CoV

0.06

y = 0.0067e 0.1384x
R2 = 0.8748

0.05
0.04

New Seal

0.03

0.0498x

y = 0.0133e
2
R = 0.5809

0.02
0.01

736

677

637

581

552

492

451

393

359

274

245

219

168

135

69

37

0.00
-47

Coefficient of Variation (CoV)

Kaiwaka Slag Site - GN Coefficient of Variation (CoV) and Seal Age -(Incr Direction)

Seal Age (days)

Old Seal CoV

Yr 1 CoV

Expon. (Yr 1 CoV)

Expon. (Old Seal CoV)

Figure 7-3: GripTester CoV between runs in relation to Age of Seal, Kaiwaka (Increasing Direction)

182

Chapter 7

0.10
0.09

April - May

Feb - March

Sept - Oct

0.08

0.0356x

y = 0.0218e
2
R = 0.5926

0.07
0.06

Year 2 CoV

0.05
0.04
0.03

0.0733x

y = 0.0058e
2
R = 0.8838

0.02

Year 1 CoV

0.01

364

339

320

300

272

264

235

215

201

175

134

116

97

76

59

42

0.00
0

GN Coefficient of Variation (CoV)

Kaiwaka Slag Site - GN Coefficient of Variation (CoV) and Days since March of
Each year -(Incr Direction)

No. of Days Since March each year

Yr 1 CoV

Yr 2 CoV

Yr 3 CoV

Expon. (Yr 1 CoV)

Expon. (Yr 2 CoV)

Figure 7-4: GripTester Coefficient of Variation between runs in relation to Number of Days since March of
each year, Kaiwaka (Increasing Direction)
This effect is an interesting one as it is an indicator of a variable that is significant in terms of modifying the
repeatability of the GripTester device as measured by the coefficient of variation (CoV) between repeated test runs.
The example shown above applies only to the Kaiwaka slag site. Further analysis was required of all of the
Northland sites combined to see whether this trend was significant for all of the sites together. The results of this
combined Northland site data analysis are shown in Figure 7-5 for each calendar month of the year. The spread of
data is shown as well as the mean CoV for each calendar month.
All Northland Sites - GN Coefficient of Variation (CoV) and
Calendar month of Year

KSS Inc
KSS Decr

0.12

Bryn Inc

Coefficient of Variation (CoV)

Bryn Decr

0.10

HIK Inc
HIK Decr

0.08

Kara Inc S1
Kara Inc S2

0.06

Kara Decr S1

0.060

Kara Decr S2
0.046

0.045

0.04

0.045
0.036

0.035

0.037

ST Inc S1

0.038

ST Inc S2

0.031

0.030

0.028

0.02

0.046

ST Inc S3
ST Decr S1

0.00
0

10

11

12

Calendar Month of Year

13

ST Decr S3
Ave CoV

Figure 7-5: GripTester Coefficient of Variation between runs for all Northland sites
The data clearly demonstrate that the GripTester CoV is least in the winter months (June, July and August) with an
approximate CoV of 3.0%, and subsequently greater in the summer months (December, January and February) with
a respective CoV of 5.0%. It is also apparent that the Hikurangi site which, due to the age of the seal, is clearly
183

Chapter 7
supposed to be at an equilibrium level , has consistently the greatest amount of variation between GripTester test
runs in comparison to the other Northland test sites. The lowest site CoV values tend to be the Kara Road and
Snooks - Tatton sites that have approximately one third of the HCV volumes. This may be a significant factor in
explaining the run-to-run variability between the sites.
It may be concluded that an environmental effect that includes temperture, rainfall and traffic volumes (load) exists
which affects the run-to-run variability of the GripTester device, that is, the variation is not merely a function of the
device itself.

7.3

Dynamic Friction Tester

As the DF Tester is a static device, and repeat runs of the device can be undertaken at the exact same position, the
number of variables in determing the repeatability of the device is reduced in comparison to other devices (i.e.
differences due to the transverse position of the testing tyre and the driver test speed are removed). The expectation
would therefore be that repeat testing on the same position with the DF Tester would demonstrate significantly less
variation, in comparison to repeat continuous friction measurement devices, such as obtained with the GripTester,
over a section of road. From preliminary testing with both devices, it was clear that the DF Tester is a more
repeatable device in terms of run-to-run variation in comparison to the GripTester. This helped to determine the DF
Tester methodology that was used, that is the use of three repeat runs in the same position for the DF Tester,
compared to the five test runs undertaken after a warm-up run with the GripTester device.
Figure 7-6 shows a comparison of the coefficent of variation (CoV) of the DF Tester between three test runs and the
GripTester with five test runs on the Hikurangi increasing direction section by calendar month of the year. The
average CoV for the GripTester for this section for the whole year is 6.3%. This is considerably higher than the
average of all Northland sites together (as shown in Figure 7-5 and on ) demonstrating that the Hikurangi site
features greater variation in the GripTester device than other Northland sites. As this was the only site that was
regularly tested with the DF Tester (although only on seven different dates), it was not possible to see whether this
same effect was evident with the DF Tester, which has demonstrated a much lower CoV of 1.6% for the Hikurangi
increasing direction site. A comparison can be made with the Tamaki Campus and Ports of Auckland sites although
both of theses sites are asphalt mix surfaces in contrast to all of the Northland sites that are chipseal surfaces. The
average DFT20 CoV for the Tamaki Campus site (refer to Figure 7-7) and Ports of Auckland site (refer to Figure
7-8) is 2.0% and 3.3%, respectively which is a lot higher than the Hikurangi site of 1.6%.

184

Chapter 7
Hikurangi Site (Increasing direction) - GripTester and DF Tester Coefficient of
Variation (CoV) by Calendar month of Year
0.12
0.11

Coefficient of Variation (CoV)

0.10
0.09
0.08
0.07
0.06

0.060

0.05
0.046

0.045

0.04

0.046

0.045
0.036

0.035

0.03

0.038

0.037
0.031

0.030

0.028

0.02
0.016

0.01
0.00
0

5
6
7
8
Calendar Month of Year

Hikurangi DFT Inc CoV

Hikurangi GN Inc CoV

10

All Northland sites GN Ave CoV

Hikurangi GN Inc Ave CoV

11

12

13

DFT Ave CoV

Figure 7-6: A comparison of run to run variability of the GripTester and DF Tester devices

1.00

0.10

0.90

0.09

DFT20 Number ( )

0.80

0.08

0.812
0.757

0.70
0.677

0.60

0.772

0.750

0.728

0.712

0.683

0.664

0.701

0.700

0.07
0.06

0.50

0.05

0.40

0.04

0.30

0.03
0.025

0.20

0.022

0.021

0.10

0.023

0.019

0.017

0.023
0.016

0.019 0.020

0.02

Coefficient of Variation (CoV)

Tamaki Campus Site, DFT CoV and Calendar Month of year,

Asphalt Mix Surfaces (Both Directions)

0.01

0.012

0.00

0.00
0

10

11

12

Calendar Month of Year

Tamaki West Flat DFT20
Tamaki East Slope DFT20
Tamaki West Slope CoV
Ave CoV

Tamaki West Slope DFT20

Ave DFT20
Tamaki East Slope CoV

Tamaki East Flat DFT20

Tamaki West Flat CoV
Tamaki East Flat CoV

Figure 7-7: Tamaki Campus DFT 20 and CoV by calendar month of the year.

185

Chapter 7

POAL Site, DFT CoV and Calendar Month of year, Asphalt Mix
Surfaces (Both Directions)

0.70

0.10

DFT20 Number ( )

0.557

0.50

0.574

0.544

0.514

0.509

0.480

0.456

0.40

0.08

0.565

0.07

0.492

0.06

0.428

0.05
0.048

0.048

0.30
0.20

0.028

0.037

0.034

0.030

0.027

0.04
0.035

0.03

0.033

0.027

0.02

0.10

Coefficient of Variation (CoV)

0.09

0.60

0.01

0.00

0.00
0

10

11

12

Calendar Month of Year

POAL West DFT20

POAL East DFT20

POAL East CoV

POAL Ave CoV

POAL Ave DFT20

POAL West CoV

Figure 7-8: Ports of Auckland DFT 20 and CoV by calendar month of the year.
This was a somewhat unexpected result, as it was expected that the DF Tester would display a higher CoV on the
chipseal surfaces in comparison to the usually more homogeneous asphalt mix surfaces, although the differing
traffic loading conditions may explain some of these differences. It is also clear from Figure 7-6, Figure 7-7 and
Figure 7-8 for the Hikurangi, Tamaki Campus and the Ports of Auckland sites respectively, that the relationship
most clearly evident with the GripTester device (shown in Figure 7-5 for the Northland sites) showing a low GN
CoV for the winter months and higher GN CoV during the summer months, is not repeated with the DF Tester
device. Although the number of data points is less, the DFT20 data indicate that the measured variation between
repeat DF Tester runs (whilst less in amplitude) is much more random and less predictable over the calendar months
of the year. This reported difference in the test results of the two devices is most likely to be due to differences in
their respective action; the DF Tester is a spinning disk which, when lowered onto the surface, repeatedly spins on
the same position until the kinetic energy has been lost. As discussed previously, the reported coefficient of friction
varies with the slip speed of the spinning disk. What has been reported in this research is the coefficient of friction
at a slip speed of 20km/h (DFT 20), whereas the test starts at an initial speed of 80km/h or 60km/h for asphalt mix
and chipseal surfaces respectively. This means that, not only have the rubber sliders had time to be conditioned
during a test run, but also the test surface has been conditioned by repeat spins of the disk on the exact same location
prior to the test result being reported.
In contrast to the DF Tester, the GripTester device makes one pass over the surface, and therefore the reported result
is the measured coefficient of friction (GN) of the surface at that specific point in time. Any subsequent runs with
the GripTester can thus demonstrate any differences that are occurring to the surface itself, due to washing or
cleaning of the surface, which would therefore increase the reported coefficent of variation between test runs, as
shown in Figure 7-1.

Therefore, a greater variation between subsequent test runs would be expected in the

GripTester device than in the DF Tester device, as the surface is conditioned prior to taking the reading of the
reported coefficient of friction in the latter case. Considering the fact that the DF Tester removes variance between
186

Chapter 7
test runs due to locational referencing in comparison to the GripTester, and that the DFTester is circular in motion
and the GripTester is longitudinal (better reflecting longitudinal traffic loading), it is not surprising that these testing
device s methodological differences result in differences in the reported coefficient of variation (CoV) results.
Clearly the GripTester method better simulates the coefficient of friction that is available to the road user at a
specific point in time with specific surface and prevailing environmental conditions. However, the DF Tester better
reports the coefficient of friction due to the microtexture condition of the surface with the short term conditioning
and environmental effects of the surface removed.
Figure 7-9 shows the normalised DFT20 results (refer to Section 8.2.1 for a description of the Normalisation
process) to a result of DFT20

=1.0 for the various sections of the two Auckland sites combined and the coefficient

of variation (CoV) for subsequent test runs plotted against the test calendar months of the year. Whilst the CoV
results (as discussed above) do not demonstrate any distinct relationship with calendar month of year, the
normalised DFT20 results do somewhat indicate that a seasonal pattern occurs with higher skid resistance occurring
in the winter months and lower results in the summer. The seasonal pattern results will be discussed in more detail
in Chapter 8.

Combined Auckland Sites, DFT CoV and Calendar Month of year,

Asphalt Mix Surfaces (Both Directions)
1.20

0.10

Normalised DFT20 ( )

0.08
1.036

1.035

1.00

1.057
1.032

0.958

0.07

1.025

1.002

1.000

0.990

0.954

0.06

0.982

0.90

0.05
0.04

0.80
0.029

0.026

0.022

0.70

0.023

0.020

0.031

0.026

0.03

0.029
0.024

0.023

0.02

0.017

Coefficient of Variation (CoV)

0.09
1.10

0.01

0.60

0.00
0

10

11

12

Calendar Month of Year

Norm DFT20

Ave Norm DFT20

CoV

Ave CoV

Figure 7-9: Normalised DFT20 results and CoV by Calendar month of the year for Auckland Sites.

7.4

Temperature Effects on Skid resistance measurement devices

7.4.1 Introduction
As discussed in Section 3.5.5, variations in the ambient and surface temperature during skid resistance
measurements can affect the coefficient of friction. Some skid resistance measurement devices are potentially more
susceptible to temperature changes than others, leading to the derivation of temperature correction factors for some
skid testing devices (e.g. the SCRIM device and the British Pendulum Tester). Neither the GripTester nor the
Dynamic Friction Tester have published temperature correction factors.
187

Chapter 7
The following sections discuss the results of temperature analysis undertaken for the GripTester and the Dynamic
Friction Tester, respectively.

7.4.2 GripTester
Ideally, temperature effects should be analysed by testing under controlled conditions in which all other known
variables are held constant, whilst varying the dependent variable (temperature). This type of analysis can be
undertaken in a controlled laboratory environment with static devices such as the DF Tester. However, it is very
difficult to undertake such analysis with a CFME device such as the GripTester that needs to be used on a live road
environment with associated temperature variations. Choosing months of the calendar year that have significant
daily extremes of ambient temperatures is difficult, as the daily temperature cycles still occur over significant hourly
time periods. This time variation allows other possible changes in variables, such as rain precipitation, traffic flow
and contaminants.
An attempt to undertake a controlled experiment with the GripTester (as explained above) was made on the
Hikurangi site in Northland on the 9th December 2004. The experiment incorporated repeat tests of the GripTester
on the same surface and on the same day, but with variation in the recorded ambient and surface temperatures. The
results of these test are shown in Figure 7-1 and discussed in Section 7.2. The results of this experiment, which
included a 26C surface temperature range, demonstrated that the variation measured on the site over the day could
not be directly attributed solely to surface temperature variation.
To further analyse temperature effects, a normalised bi-variate analysis was undertaken of temperature effects on the
GripTester for all of the seasonal test dates in the Auckland and Northland field test sites. The coefficient of friction
obtained by the GripTester was normalised (refer to Section 8.2.1 for a description of the Normalisation process) to
a result of 1.0 GripNumber for all sites grouped by common surfacing type. Each data point was an average of the
test runs (excluding the warm-up run) for that specific site and section on a specific day and recorded ambient and
surface temperature.

The sites and sections used in the analysis were sections that should have been at an

equilibrium level of polishing. It is known that temperature is only one of many variables that affect the coefficient
of friction. However, if it was a significant factor, one would expect to see some underlying relationship from a bivariate analysis. Figure 7-10 and Figure 7-11 show the relationship between the recorded normalised GripNumber
and surface temperature for asphalt mix surfaces from the Auckland sites and chip seal surfaces from the Northland
sites respectively.

188

Chapter 7

Normalised GN and Surface Temperature

Auckland Sites, Asphalt Mix (Both Directions)

1.20
1.15

y = -0.0016x + 1.0454

Normalised GN

1.10

R = 0.043

1.05
1.00
0.95
0.90
0.85
0.80
0.0

10.0

20.0

30.0

40.0

50.0

60.0

Surface Temperature (C)

POAL West
Tamaki East Flat
Linear (All Data)

Tamaki West Flat

Tamaki East Slope

Tamaki West Slope

POAL East

Figure 7-10: Normalised GripTester results and recorded Surface Temperature on Asphalt Mix surfaces

Normalised GN and Surface Temperature

Northland Sites, Chip Seal Surfaces (Both Directions)

1.20
1.15

Normalised GN

1.10
1.05
1.00
0.95
y = 0.0001x + 0.9998
2
R = 0.0002

0.90
0.85
0.80
0.0

10.0

20.0

30.0

40.0

50.0

60.0

Surface Temperature (C)

KSS T/S 800m Inc
KARA T/S 40m Inc
BRYN T/S 110m Dec

BRYN T/S 490m Inc

S&T S2 Ave Inc
HIK Sect Ave Dec

HIK Sect Ave Inc

KSS T/S 100 Dec
Linear (All Data)

Figure 7-11: Normalised GripTester results and recorded Surface Temperature on Northland chip seal Sites
The above figures clearly show that the variation of the coefficient of friction as recorded by the GripTester cannot
be attributed to the surface temperature alone. Similarly, as shown in Figure 7-12 and Figure 7-13, the recorded
ambient temperature by itself explains very little of the measured variability of the coefficient of friction as
measured by the GripTester or it is hidden by other variables. .

189

Chapter 7

Normalised GN and Ambient Temperature

Auckland Sites, Asphalt Mix (Both Directions)

1.20
1.15

Normalised GN

1.10
1.05
1.00
0.95

y = -0.0002x + 1.0029
2
R = 0.0002

0.90
0.85
0.80
0.0

10.0

20.0

30.0

Ambient Temperature (C)

POAL West
Tamaki East Flat
Linear (All Data)

Tamaki West Flat

Tamaki East Slope

Tamaki West Slope

POAL East

Figure 7-12: Normalised GripTester results and recorded Ambient Temperature on Asphalt Mix surfaces

Normalised GN and Ambient Temperature

Northland Sites, Chip Seal Surfaces (Both Directions)

1.20
1.15

Normalised GN

1.10
1.05
1.00
0.95
y = -0.0013x + 1.027

0.90

R = 0.0106

0.85
0.80
0.0

10.0

20.0

30.0

40.0

Ambient Temperature (C)

KSS T/S 800m Inc
KARA T/S 40m Inc
BRYN T/S 110m Dec

BRYN T/S 490m Inc

S&T S2 Ave Inc
HIK Sect Ave Dec

HIK Sect Ave Inc

KSS T/S 100 Dec
Linear (All Data)

Figure 7-13: Normalised GripTester results and recorded Ambient Temperature on Northland chip seal Sites
An additional temperature analysis was undertaken to determine whether the U-shaped GripTester coefficient of
variation (CoV) by calendar month of the year (shown in Figure 7-5 for Northland chip-seal sites) could be
explained by the variation in surface temperature. This would explain the higher variation in run-to-run variability
of the GripTester in predominantly summer and correspondingly warmer months of the year. Figure 7-14 shows the
relationship between recorded surface temperature and the CoV of the repeated GripTester runs for all the Northland
190

Chapter 7
sites with chip seal surfaces. The relationship indicates that temperature by itself explains only up to 10% of the
reported variation in repeated GripTester runs.
In summary, it is concluded that the observed increase in GripNumber with repeated runs (as shown in Figure 7-1)
on the same section of road cannnot reasonably be attributed to the varying surface temperature or the warming /
conditioning of the GripTester measuring tyre alone. It is conjectured that this observed increase, after initial
conditioning of the tyre, is partly a function of repeated washing and/or cleaning of the pavement surface. This is
reflected in the greater increase in recorded coefficient of friction for repeated GripTester runs in summer months
than in predominantly wetter winter months as shown in Figure 7-5.
All Northland Sites - GN Coefficient of Variation (CoV) and Surface Temperature

KSS Inc
KSS Decr

0.14

Bryn Inc
Bryn Decr

0.12

Coefficient of Variation (CoV)

HIK Inc
HIK Decr

0.10

Kara Inc S1
Kara Inc S2

0.08
y = 0.0005x + 0.0288

Kara Decr S1

R = 0.056

0.06

Kara Decr S2
ST Inc S1
ST Inc S2

0.04

ST Inc S3
0.02

ST Decr S1
ST Decr S3

0.00
0

10

15

20

25

30

35

40

45

50

Linear (All
Data)

Surface Temperature (C)

Figure 7-14: GripTester Coefficient of Variation and Surface Temperature of Northland sites.

7.4.3 Dynamic Friction Tester

An analysis of the variation of the normalised coefficient of friction measured by the DF Tester with respect to
surface temperature was undertaken for the three sites where reasonably regular DF Tester results were obtained (i.e.
the Tamaki Campus, Ports of Auckland and the Hikurangi sites). The results of this analysis are shown in Figure
7-15. It should be remembered that the Tamaki Campus and the Ports of Auckland sites are both asphalt mix
surfaces and the Hikurangi site is a chip seal surface. Consequently, a direct comparison between these sites for
normalised DF Tester results is less applicable. It is shown for comparative reasons only.
As with the variation of the measured GripNumber (CoF measured by the GripTester), the coefficient of friction
measured by the DF Tester (DFT20 ) cannot be directly explained by the variation in recorded surface temperature.
The results of both the GripTester and the DF Tester on the two asphalt mix surfaces tend to indicate that, as the
asphalt surface temperature increases, the normalised coefficient of friction decreases. This may not be a reflection
of the effects of temperature, but a consequence of the seasonal variation effects due to the differences in polishing
of the aggregates with finer contaminants. Conversely, in winter this finer material is washed away, leaving larger
191

Chapter 7
sized grit particles that abrade and therefore rejuvenate levels of skid resistance. However, on high macrotextured
chip seal surfaces, the coefficient of friction does not decrease with increasing surface temperature; and the results
remain completely random. When ambient temperatures are used, the relationships become more random and the
effect of decreasing coefficient of friction with increasing temperature is less evident and tends to be completely
random for both surface types.
Normalised DFT20 POAL, Tamaki and Hikurangi Sites
1.20
1.15
y = -0.002x + 1.0603
2
R = 0.1069

Normalised DFT20 ( )

1.10
1.05
1.00
0.95
0.90

Normalised POAL & Tamaki AM DFT20

Hikurangi DFT20 Chip Seal
Linear (Normalised POAL & Tamaki AM DFT20)

0.85
0.80
0

10

15

20

25

30

35

40

45

50

Surface Temperature (C)

Figure 7-15: Surface Temperature effects on the DF Tester device.

The surface temperature was also compared against the GripTester and DF Tester coefficients of variation for a
specific site, to determine whether it could explain any of the variation between subsequent test runs of the two skid
resistance devices. The resulting device cofficient of variation (CoV) was analysed for the Hikurangi increasing
direction site. Figure 7-16 shows this relationship which clearly shows that surface temperature by itself does not
explain either the GripTester repeated run-to-run variation or the DF Tester DFT20 coefficient of variation for the
Hikurangi site. The surface temperature by itself explains only up to 14% of the reported variation for both devices.
Consequently, surface temperature in itself is not a good predictor of the run-to-run variability of skid resistance
measurement devices.

192

Chapter 7
Hikurangi Increasing Direction, Northland
GripTester and DF Tester CoV and Surface Temperature
0.12

Coefficient of Variation (CoV)

0.10
y = 0.0002x + 0.0564
2

R = 0.0074

0.08

0.06

0.04

0.02

y = 0.0007x - 0.0072
2
R = 0.1373

0.00
0

10

15

20

25

30

35

40

45

50

Surface Temperature (C)

HIK Inc GN CoV
Linear (HIK Inc GN CoV)

HIK Inc DFT20 CoV

Linear (HIK Inc DFT20 CoV)

Figure 7-16: A comparison of the effects of Surface Temperature on the GripTester and DF Tester devices.
As the DF Tester was able to be used in controlled laboratory experiments, it was decided to undertake a series of
DF Tester tests on prepared samples (the results of which are discussed in Chapter 10), controlling all other
variables other than temperature.

Prepared surface samples allowed control of the surface and test water

temperature as the samples could be placed in temperature controlled refrigeration and oven units to determine the
sensitivity of the DF Tester device to variations in surface and test water temperature. At the same time, the effect
of changing the DF Tester rubbers from worn rubber sliders to new rubbers was examined. Figure 7-17 shows the
results of these tests. The numbered positions of the figure are consistent for each sample tested and relate to the
following variations in the testing methodology:
Position 1:

Test sample and water used for testing both refrigerated to 4C prior to testing - New rubbers.

Position 2:

Test sample refrigerated to 4C. Normal tap water temperature used for testing (approx 17-

17.5C). New rubbers.

Position 3:

Test sample tested when sample surface, air temperature and water temperature all range between

18.5 and 20C - Old rubbers.

Position 4:

Test sample tested when sample surface, air temperature and water temperature all ranged

between 19 and 20C - New rubbers. Sample 5 was the first sample that was used with a new set of rubber sliders.
Position 5:

Test sample tested when sample surface had been heated in an oven up to 55C and the air

temperature and water temperature ranged between 19 and 20C - New rubbers.

193

Chapter 7

Effects of Temperature and Change of Rubbers on DF Tester

Dynamic Friction Test Number ()

0.80

0.75
0.70

0.65

Note for the same surface sample, that CoF ( )

is insensitive to change in surface temperature

0.60
0.55
0.50
Initial change of DF Tester rubbers from
old rubber to new caused significant
decrease in CoF ( ). This conditioning of
a new rubber however recovered by the
time of the next test.

0.45
0.40
0.35
0.30
0

10

15

20

25

30

35

40

45

50

55

60

Surface Temperature (C)

Sample 2

Sample 3

Sample 4

Sample 5

Figure 7-17: The effects of Temperature and Change of Rubber Sliders on DF Tester Results
The results of the temperature trials using the DF Tester are shown in Figure 7-17 and demonstrate that:

The measured coefficient of friction is very insensitive to change in surface and/or air temperature for the
normal ranges that could be expected in New Zealand (difference between positions 1,2,4 and 5);

Using very cold water temperatures for the skid testing increases the measured coefficient of friction only
marginally (difference between positions 1 and 2);

Changing the DF Tester rubbers significantly reduces the measured coefficient of friction ( ) of the first
few test results (refer to Sample 5). However, this effect quickly wears off as the rubber sliders roughen
and become conditioned (difference between positions 3 and 4 of Samples 2,3 and 4).

7.4.4 Summary of Temperature Effects

Research from Oliver (1980) using a BPT device, Hill and Henry (1981) using a Locked Wheel Tester device and
Hosking (1992) using a SCRIM device, has shown that the measured coefficient of friction tends to decrease with
increasing air temperature. This trend of decreasing coefficient of friction with increasing ambient or surface
temperature does not explain the variation seen from month to month for normalised GripTester device
measurements as shown in Figure 7-10 and Figure 7-11 for surface temperature results and Figure 7-12 and Figure
7-13 for ambient temperature results on asphalt mix and chip seals sites, respectively. The results are mostly
random with a coefficient of determination (R2) of less than 5%.
Temperature tests were undertaken on one Northland field site (Hikurangi control site) on one day to determine
whether any variation of the coeffiecient of friction could be explained by temperature changes experienced over the
day. The surface temperature on the day had a range of 26 C as shown in Figure 7-1. However, as discussed in
Section 7.2, the change in the coefficient of friction could not be explained by the change in temperature alone.
These findings agree with correspondence from the developers of the GripTester (Findlay Irvine Ltd, 2003b) who
found that, once the measuring tyre has been adequately conditioned , then the measured coefficient of friction is
194

Chapter 7
insensitive to changes in ambient and surface temperature. The change in the coefficient of friction must therefore
be due to a combination of factors that include temperature, or to factors that are other than temperature.
The analysis of the normalised Dynamic Friction Tests with surface temperature for the asphalt mix sites in the
Auckland Region (refer to Figure 7-15) indicates a possible decreasing trend with increasing skid resistance.
However, it is still mostly random with a coefficient of determination (R2) of approximately 11%. The results of a
temperature controlled experiment using the DF Tester on prepared laboratory examples that had been polished to
an equlibrium level of polishing (shown in Figure 7-17) show that the DF Tester is also very insenstive to changes in
temperature over a range from 4C to 54C. It is believed that this is partly due to the method of testing (i.e. a
circular grinding of the rubber sliders on the same position allows conditioning of the rubber slider throughout the
test). Furthermore, the synthetic rubber test slider compounds used are less sensitive to variation in temperature
than natural rubbers. The results of these controlled laboratory tests are useful as they prove that the trend indicated
in the field tests (refer to Figure 7-15) should not be explained by the variation in ambient or surface temperature. It
must therefore be concluded that the indication of a trend of decreasing measured coefficient of friction with
increasing temperature is due to real seasonal effects in the surface aggregate microtexture itself. This aligns with
the established theory that, during summer months, the aggregate microtexture is polished, thereby reducing
microtexture, and during winter seasons, the microtexture is rejuvenated by abrasion to a higher level. This effect is
therefore not related to temperature differences for both the GripTester and the Dynamic Friction Tester, although
temperature could still be significant in the case of other devices, especially those that use natural rubbers (e.g. the
UK based BPT).
Results of controlled laboratory tests with the DF Tester, with varied water temperature from 4C to 17.5C (refer to
Figure 7-15), agreed with previous research which showed that the temperature of the water used in the skid
resistance tests has negligible effect on the measured coefficients of friction.
It is important to note that, whilst the GripTester and the DF Tester are not significantly affected by varying
temperatures, the road user s vehicles could be. Research has shown that tyre temperatures tend to be proportional
to air and pavement temperatures, therefore increased temperatures in summer months will adversely effect the
frictional properties of vehicle tyres, leading to a loss in hysteresis and therefore a decrease in the available skid
resistance in summer months. This is in addition to real decreases in skid resistance which result from the
microtextural polishing of the aggregate itself, indicating that road users are at a higher risk in summer than during
winter months in wet conditions.
There is also an indication that the coefficient of variation for the GripTester and the DF Tester increases with
increasing temperature (refer to Figure 7-14 and Figure 7-16). This indication reinforces the findings above that a
washing or cleaning effect may help to increase the coefficient of friction between successive test runs. However,
this is an indirect effect, as it has little do with the temperature itself.

7.5

Correlation of Skid Resistance Devices

7.5.1 Introduction
As has been discussed in Section 4.6.6, the PIARC World Road Association undertook an experiment in 1992 to
compare and harmonise texture and skid resistance measurements (Wambold et al., 1995). The experiment included
51 different friction and texture measurement devices from 14 different countries on a total of 54 sites around
Belgium and Spain. The devices used in this research (the GripTester, the Dynamic Friction Tester and the SCRIM
device as part of the Transit New Zealand network surveys) were all part of the 1992 PIARC experiment, which
195

Chapter 7
allow a comparison to be made between the devices for New Zealand surfaces in comparison to the PIARC
experiment. As has been discussed in Section 4.3 and shown in Figure 4-9, there are significant differences between
the devices in terms of the method of testing, the angle of the tyre or sliders, the size of the measuring tyre and or
sliders, the loads placed on the measuring tyre and the tyre pressures. However, all devices should demonstrate a
reasonable ability to predict relative changes of the coefficient of friction within a road network. It was therefore
hoped that all of the devices used in this research would follow similar trends (i.e. if the SCRIM device
demonstrated a low result on a particular section of road so would the GripTester and the Dynamic Friction Tester).
The following section discusses the results for the three devices.

7.5.2 Correlation of the GripTester with the DF Tester

The GripTester (a CFME device), records the coefficient of friction every one metre of longitudinal distance
travelled. In its normal reporting mode, it reports the mean result for every 10m. In comparison, the DF Tester is a
static device (as is the BPT device) and therefore tests at a specific locational reference point. If a comparison is
made with a DF Tester test result at a specific point with an averaged 10m section (or greater) from the GripTester
device, then depending upon how homogeneous the section length is, significant variations in any correlation results
can occur. For this reason, it was decided to compare results for the two devices to determine how well the results
correlate when tested with the same number of test points and spacing intervals (i.e. 1m spacings).
The GripTester software was modified to enable reporting at 1m intervals in comparison to the normal 10m mode.
The tests were undertaken at the Tamaki Campus site on an asphalt mix (TNZ Mix 14) on the 11th Jan 2005 when
the surface was approximately three years old. A 120m section of pavement was used as the test section. It
included a middle section with significantly lower skid resistance, due to turning vehicle movements entering and
exiting a car park area on the internal University of Auckland Tamaki Campus road. The testing line was in the
western direction in the left wheel track along the flat grade (-0.5%) section (refer to Figure 5-5). The GripTester
survey speed was undertaken at 30km/h, giving a slip speed of 4.5km/h, whilst the DF Tester ( ) results were
compared at a slip speed of 30km/h (DFT30). As the DF Tester has proved to be a very repeatable device when
tested in the same position (refer to section 7.3), only one DF Tester result was recorded at each running distance
location. The DF Tester test locations were spaced at 5m intervals for the first 49m, a further 35 tests were
undertaken at 1m spacings and a final 7 tests were again taken at 5m intervals. Four test runs of the GripTester
(GN) were undertaken in each of the two reporting modes (10m and 1m). A comparison with the DF Tester of the
mean of the four test runs for each of the two GripTester test modes (10m and 1m) is shown in Figure 7-18. Figure
7-18 clearly shows that, even though the devices are different in terms of their testing methodology (GripTester is a
braked in-line tester whereas the DF Tester has a circular diameter of three rubber sliders), both devices in tandem
identify similar changes in the coefficient of friction. These are real microtexture differences due to the additional
polishing effects of high turning movements at the intersection of the adjacent carpark.

196

Chapter 7
0.90

Correlation of GripTester and DFTester (Tamaki Campus Site - Aspahalt Mix)

Ave GN in Normal 10m mode

GripTester (GN30) & DFT30 ()

0.85

Ave GN in 1m mode

0.80
0.75
0.70

Location of intersection with

carpark. High turning
movements causing additional
polishing of microtexture

0.65

0.60
0.55

10 DFT tests at 5m intervals

35 DFT tests at 1m intervals

7 DFT tests at 5m intervals

120

115

110

105

100

95

90

85

80

75

70

65

60

55

50

45

40

35

30

25

20

15

10

0.50

Distance from Start (m)

DFT30

Ave GN - 1m mode

Ave GN (10m mode)

Figure 7-18: Comparison of the GripTester and DF Tester results on an Asphalt Mix
Figure 7-19 shows the results of a linear correlation of the two devices over the section of 1m tests for the
GripTester and the DF Tester, with a correlation coefficient (R2) of 0.79 with 35 data points. This reasonably high
correlation coefficient gives some confidence that, even though the two devices have significantly different
equipment methodology, both devices are sensitive enough to report real differences in microtextural changes in the
aggregate surface. This finding, whilst only undertaken on an asphalt surfacing gives confidence that the DF Tester
could be used in the laboratory and could be correlated against in-field results of the GripTester on similar surfaces.

A Correlation of GripTester and DFTester for 1m interval (Tamaki Campus)

0.90

GripTester CoF (GN)

0.85

y = 0.9531x + 0.0913
2
R = 0.7919

0.80
0.75
y=x
0.70
0.65
0.60
0.55
0.50
0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

DFTester CoF ()

Figure 7-19: Correlation between the GripTester and DF Tester at 1m interval spacings
197

Chapter 7
A comparison of the 10m and 1m GripTester reporting modes shows that due to the averaging process, that the 10m
averaging mode will not report the lowest or highest coefficient of friction areas. The areas of lowest skid resistance
and localised high surface polishing stresses are of primary importantance to the road asset manager. It is therefore
feasible (especially in urban areas) with either tight horizontal geometry or areas of high braking or acceleration,
that 10m averaged CFME device surveys will miss localised problem areas.
This is not likely to be a significant problem on rural State highways where fewer localised problems occur at less
than 10m intervals occur. However asset managers need to be aware of the potential for the effect in specific
locations that may cause very localised loss of control type crashes to occur. An area of two to three metres of low
skid resistance at a critical point where the demand for friction is greatest (e.g. on the minimum horizontal curve
radius with marginal design speed, the approach to an intersection or an at-grade rail crossing or one-way bridge)
can be enough to become a critical factor in causing road crashes. However, it is also realised that in terms of data
collection and database management to enable appropriate decision making, 1m surveys or even 5m surveys would
mostly not be justifiable, as it would significantly increase the required data management issues and cost of data
processing. Furthermore, as most non-Transit NZ managed road controlling authorities in New Zealand do not
undertake any skid resistance testing, 10m averaged network surveys would still be a significant and appropriate
improvement.
Less regular testing occurred with the DF Tester (due to requirements for temporary traffic control) at the other field
testing sites during the data collection period, including the Ports of Auckland, Kaiwaka, Hikurangi, Kara Road and
Snooks Tatton sites. A correlation analysis undertaken on these sites between the 10m averaged GripNumber (GN)
and the DF Tester reported coefficient of friction at a slip speed of 20km/h (DFT20). The results of a linear
correlation between the two devices is shown in Figure 7-20, separated for the Ports of Auckland and Tamaki sites
with an Asphalt Mix and a low macrotexture, and the Northland chip seal sites with medium to high macrotexture.

GripTester Device Correlated with Dynamic Friction Tester device

(known outliers removed)
1.00
0.95

GripNumber (GN50)

0.90

GN50 Northland sites

GN30 1m mode
Linear (GN 50 POAL)

GN 50 POAL
Linear (GN50 Northland sites)
Linear (GN30 1m mode)

Tamaki Asphalt
Mix, low MacroTX
y = 0.9531x + 0.0913
2
R = 0.7919

0.85
0.80
0.75
0.70

y=x

Northland Chip Seal,

high MacroTX

0.65
0.60

POAL Asphalt
Mix, low MacroTX

0.55
0.50
0.45
0.40
0.40

y = 0.4085x + 0.3502

y = 0.647x + 0.0633
2
R = 0.5188
0.45

0.50

0.55

R = 0.4147
0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

Dynamic Friction Tester (DFT20)

Figure 7-20: Correlation between the GripTester and DF Tester for all sites
The Ports of Auckland asphalt mix site and the Northland chip seal surface sites were all tested at a vehicle testing
speed of 50km/h (i.e. 7.5km/h slip speed). The GripNumbers (GN) recorded were either an average of the total test
198

Chapter 7
section (Hikurangi and the Ports of Auckland sites), or the average of a 30m timeslice (Kara Road and Snooks
Tatton sites). The 30m timeslices were undertaken on the Kara Road and Snooks Tatton sites as these sites were
less homogeneous over longer sections and the sections included geometric elements which meant that longer
section lengths were not applicable. The Ports of Auckland site reported a linear correlation coefficient (R2) of 0.41
for an asphalt mix with 20 data points, and the Northland sites a linear correlation coefficient (R2) of 0.52 with 16
data points.

The Tamaki site was measured at a lower vehicle testing speed of 30km/h (producing higher

GripNumbers) and on one date, with 1m spaced intervals between results (as explained above and shown in Figure
7-20). The lower correlation coefficients for the Ports of Auckland and the Northland sites in comparison to the
Tamaki site are thought to be largely due to the variation that occurs with the 10m averaging process in comparison
to the 1m spaced correlation exercise.

7.5.3 Correlation of GripTester and SCRIM

Given that the SCRIM and the GripTester devices have significantly different modes of operation (i.e the SCRIM is
a Sideways-force device and the GripTester is a braked in-line fixed slippage device) it is not suprising that the
reported means of the devices for the same sections of road also differ significantly. However, one would expect
that, on the same section of road, the results obtained from the two devices would be similar in shape and visually
apparent to the observer that the same section of road had been tested. Example GripTester and SCRIM surveys for
the Snooks-Tatton Northland site on the 14th April 2003 are shown in Figure 7-21. The figure clearly shows that
both the GripTester and SCRIM follow the same pattern with a higher coefficient of friction for the two newly
sealed sections (Sections 1 and 3) and a significantly lower level of skid resistance for the middle older seal (Section
2). A commonly used approximation method for correlating the devices is also shown in Figure 7-21.
Snooks Tatton Road, Increasing Direction
SFC and Grip Number (14 April 2003)
1.00
Section 1, New Seal

SCRIM SFC and GripNumber (GN)

0.90

Section 2, Old Seal

Section 3, New Seal

Approximation GN=SFC/ 0.78GN

0.80
Grip Number LWT
0.70
0.60
0.50
0.40
0.30

TNZ Investigatory Levels

(IL) / Site Categories

TNZ Site Category 2

(IL=0.50)
TNZ Site Category 4
(IL=0.40)

SCRIM Coefficient LWT

0.20
0.10

GripNumber LWT (GN50)

SCRIM SFC LWT

TNZ Site Category IL

Approximation SFC=0.78GN

560

610

460

510

360

410

260

310

160

210

60

110

10

14960

14910

14860

14810

14760

14710

14660

14610

14560

14510

14460

14410

14360

14310

14260

14210

14160

14110

14060

14010

13960

13910

0.00

TNZ Route Position (Running Distance)

Figure 7-21: Example GripTester and SCRIM survey for the Snooks Tatton Northland site
The figure also demonstrates Transit New Zealand s T/10 Investigatory Level (IL) policy for skid resistance
standards where the various geometric sections have differing Investigatory Levels (e.g. the higher levels of IL
shown are due to horizontal curves with radii less than 250m, thereby categorised under Transit New Zealand s T/10
199

Chapter 7
policy as Site Category 2 with an IL of 0.50). The Investigatory Levels are reported in terms of SFC and if a
surveyed section has a lower recorded SFC measurement than the appropriate IL site category standard, then the
section should be investigated further to determine if and when some form of treatment should be applied to increase
the measured skid resistance level.
Correlations between devices have been investigated in a large international experiment undertaken by the PIARC
World Road Association (Wambold et al., 1995) with the developemnt of the International Friction Index IFI (F60
and Sp). This was discussed in Section 4.6.6 and the application of the equations to New Zealand roads by Cenek
and Jamieson (2000).
In their study of 1m averaged skid resistance values for 10km on State Highway 58 and a short section of State
Highway 2 that incorporated a wide variation in seal type and condition, and in road geometry (curvature,
gradient and crossfall) Cenek and Jamieson (2000) reported that:
the degree of correlation between the SCRIM, GripTester and Norsemeter ROAR was not
significantly improved through the application of IFI and the inclusion of road geometry and
road surface condition variables, with about 50% of the total variation still not being able to
be explained .
They suggested that for the application of Transit New Zealand s T/10 policy, the following relationships be used
for converting GripNumber (GN) and Norsemeter ROAR Mu ( ) when measured at 34% fixed slip readings to
equivalent SCRIM SFC values to a confidence level of 95%:
GripTester equivalent

SFC ( 0.08) = 0.42GN + 0.17

R 2 = 0.46

Norsemeter ROAR equivalent

SFC ( 0.08) = 0.55 + 0.12

R 2 = 0.53

The relationship developed by Cenek and Jamieson (2000) that explained 46% of the variation between the
GripTester and the SCRIM, has been compared with the data collected on the Northland sites over the data
collection period used in this research (refer to Figure 7-22). Whenever a SCRIM network survey or seasonal site
was tested by the SCRIM device, a corresponding GripTester survey was undertaken as close as practically possible
in terms of time and site conditions for all five of the Northland sites. Owing to the contractual obligations of WDM
Ltd, who perform the SCRIM surveys for Transit New Zealand and Works Infrastructure Ltd on the PSMC02 State
highway network, combined with the physical distance from Auckland to the Northland sites, the two-device tests
were at best measured on the same day and within a few hours of each other. This meant that the environmental test
conditions between the two test devices could have varied in terms of temperature, rainfall and traffic conditions.
Additionally, the data points for the comparison in this research were averaged section results for the SCRIM and
GripTester reported coefficients of friction, whereas the Cenek relationship was based on every 1m, reported at the
same time and conditions, and from only two sites. The averaged section lengths for the Northland sites were those
described in Section 5.7.

200

Chapter 7

1.00
0.90

GripTester Device Correlated with SCRIM device

Northland Sites
Cenek, 2000 correlation
SFC = 0.42GN + 0.17
y=x

GripNumber (GN)

0.80
0.70
0.60

y = 0.8253x + 0.2959
2
R = 0.4485

0.50
0.40
0.30
0.20
0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

SCRIM (SFC)

Figure 7-22: A Comparison of Correlation relationships for the SCRIM and GripTester
As can be clearly seen in Figure 7-22, the linear correlation relationship from this research can explain
approximately 40% of the data variation and is quite different from the relationship developed by Cenek and
Jamieson (2000). Both relationships have been developed from data from less than desirable ranges of recorded
coefficients of friction (especially the data from this research that had significantly fewer data points, even though
the data points were section averages and therefore could be weighted higher than the approximately 10,000 1m data
points for the Cenek and Jamieson equation). The real correlation relationship would also be far more robust if data
points could have been surveyed from both ends of the skid resistance data, desirably down to 0.3 GripNumber (GN)
and up to 1.0 GN, and correspondingly to 0.20 SCRIM SFC and up to 0.80 SFC. With more data in these data
boundary regions, the slope of the equation line could change significantly and it may be shown that the relationship
is not simply linear. From observation and experience of other lower skid resistance sites, it is thought that the slope
of the equation line developed from data in this research is not steep enough and that the Cenek and Jamieson
equation may be too steep.

7.5.4 Correlation of GripTester and SCRIM by IFI

As discussed in Section 4.6.6, the PIARC experiment developed the International Friction Index (IFI) that enabled
the harmonisation of measurements using different testing devices and varying surface textures to be compared on a
common scale. The IFI is comprised of two numbers that describe the skid resistance of a pavement surface over a
range of operating speeds:

Sp - a speed constant, (which characterises the change in friction with change in slip speed); refer to Eq 7-1
and

F60 - a friction constant (the harmonised estimate of friction at 60km/h), refer to Eq 7-2.

The speed constant Sp is linearly related to the result of a macrotexture measurement (TX) and transformation
constants (a and b) that were determined for each type of macrotexture measurement device used in the PIARC
experiment (e.g. Sand Circle tests and laser or sensor measured texture devices). The friction number (F60) is the
resultant harmonised friction result determined from a specific friction measurement device that participated in the
201

Chapter 7
PIARC experiment. The F60 friction number is transformed with device specific constants (A,B and if applicable
C) to transform the device specific coefficient of friction (FRS) to a harmonised IFI Gold Friction curve (F60).
The FRS component is the coefficient of friction obtained using the various skid resistance measurement devices.
As each friction device is more or less sensitive to the speed constant (Sp) and therefore macrotexture, the
harmonised F60 friction value incorporates the Sp component and therefore attempts to account for differences in the
surface macrotexture.
An analysis was undertaken to determine whether the variation of the measured coefficient of friction could be
reduced between the friction results obtained on the Northland sites as measured by the SCRIM, GripTester and the
DF Tester, by harmonising the results to the International Friction Index (F60). The macrotexture measurements
were obtained from the SCRIM MPD surveys that were undertaken at the same time as the SCRIM SFC surveys.
The transformation coefficients for the Sp component and the typical friction devices used in New Zealand, and
other well known locked wheel devices are shown below in Table 7-1.
Macrotexture
Measurement Device
ASTM E-965 (Sand Circle Test) MTD
SCRIM SMTD
Friction Measurement Device

S p = a + b.TX

IFI Speed Number

a

-11.59813

113.632463

25.8322

139.6801

IFI Friction Number

F 60 = A + B.FRS .e

S 60
Sp

+ C.TX

GripTester

0.082091

0.910403

SCRIM

0.032581

0.871726

Norsemeter OSCAR @20% slip

0.119172

0.6431185

ASTM E274 Locked Wheel Tester

0.119172

0.6431185

French LCPC Skid Trailer

0.002257

1.007617

DF Tester and Sp from MPD

0.081136

0.731584

PSRT and Sp from MPD

0.056262

0.0075635

Table 7-1: PIARC IFI Transformation Coefficients (Wambold et al., 1995)

An example of the use of the International Friction Index in terms of the F60 friction component and the Speed
Number Sp for the Kaiwaka slag site in Northland is shown in Figure 7-23. The figure shows the transformed
monthly (approximately) GripTester results to IFI F60 for the section averages, with the latter parameter plotted
against the IFI F60 results transformed from the SCRIM device. The Speed Number Sp has also been plotted,
calculated from macrotexture measurements obtained from the SCRIM macrotexture MPD surveys collected at the
same time as the SCRIM SFC survey. The transformed SCRIM F60 results are shown whenever they were
measured (on the Northland PSMC 02 contract this was a minimum of two full network surveys per calendar year).
A close visual observation of Figure 7-23 clearly shows that the transformed results from the GripTester and the
SCRIM to IFI F60 are reasonably good in that both devices are giving reasonable predictions of the relative changes
in the skid resistance level.
202

Chapter 7

Kaiwaka Slag Site -IFI F60 and SP vs Time (Ave Incr dir)
0.80

600

0.70

500
400

0.50

300

0.40
0.30

New Seal at 07/05/2003

TWORS chip size 3/5,
PSV 58

0.20

IFI SP

IFI F60

0.60

200
100

0.10
21/03/2005

21/01/2005

21/11/2004

21/09/2004

21/07/2004

21/05/2004

21/03/2004

21/01/2004

21/11/2003

21/09/2003

21/05/2003

21/07/2003

0
21/03/2003

0.00

Date
IFI F60 (GN)

IFI, F60 (SFC)

IFI, Sp Act

Figure 7-23: International Friction Index (IFI) F60 and Sp for Kaiwaka Slag site against time
The IFI F60 GripTester results and the IFI F60 SCRIM results were compared for all of the Northland averaged
section sites together to determine whether a better correlation could be obtained by including the Sp component.
The latter allows differences in device sensitivity to macrotexture to be taken into account. This was to enable a
comparison with the direct measurement result of GN and SFC shown in Figure 7-22 which yielded a linear
coefficient of determination (R2) of 0.45. The results of the correlated IFI F60 friction component from the two
devices is shown in Figure 7-24 which indicates that the coefficient of determination increased slightly from 0.45 to
0.48 (i.e. 3%).
International Friction Index (F60) - Northland Sites
0.75

IFI F60 from GripTester

0.70
0.65
0.60
0.55

y = 0.6979x + 0.2638
2

R = 0.4847

0.50
0.45
0.40
0.35
0.30
0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

IFI F60 from SCRIM

Figure 7-24: IFI F60 correlation between the GripTester and SCRIM devices
Thus, it is evident that the results of this research between the SCRIM device and the GripTester device agree
reasonably well with the statement by Cenek and Jamieson (2000) that the degree of correlation was not
significantly improved through the application of IFI, with about 50% of the total variation still not being able to be
explained.
203

Chapter 7

7.5.5 Correlation of GripTester and Dynamic Friction Tester by IFI

A similar analysis was undertaken using the IFI transformation factors for a correlation between the GripTester and
the Dynamic Friction Tester devices. This was executed for the Northland chipseal sites whenever tests with the
devices were completed on the same sites and at the same time. The result of the correlation are shown in Figure
7-25. These results show a relatively low coefficient of determination (R2=0.29) where just under 30% of the
variation can be explained by the IFI transformation process, whereas a direct correlation of the two-device explain
approximately 52% of the variation, as shown in Figure 7-20.
IFI F60 - Northland Sites
0.75
y=x

IFI F60 from DF Tester ( )

0.70
0.65
y = 0.2853x + 0.308

0.60

R = 0.2896

0.55
0.50
0.45
0.40
0.35
0.30
0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

IFI F60 from GripTester

Figure 7-25: IFI F60 Correlation between the GripTester and Dynamic Friction Tester devices
An IFI F60 correlation is not displayed for the Auckland asphalt mix, as non-detectable differences were reported
between the macrotexture results at the two sites over the data collection period. If macrotexture is constant over
time, then the Speed number Sp is also constant and the variation of IFI F60 is based solely upon the the skid
resistance measurement device results, leading to the same correlation coefficient equation as the direct comparison,
shown in Figure 7-20.

7.6

Christchurch Surface Friction Device Correlation Trials

7.6.1 Introduction
An International Conference on Surface Friction of Roads and Runways was held in May 2005 in Christchurch New
Zealand, hosted by Transit New Zealand and WDM Ltd. Two days of field trials were undertaken on 5th and 6th
May 2005 immediately following the conference, with the objective of gaining further understanding of the
relationship between the various commonly used skid resistance and texture test devices. Special surfaces were laid
on an existing road to provide test sections that would allow further analysis of data obtained from surfacing types
commonly used on the New Zealand State Highway network. The author of this research contributed in helping to
organise and present at the technical conference (Wilson & Dunn, 2005a; Wilson & Kirk, 2005), inviting
international speakers and by participating in the field tests with the University of Auckland s GripTester and
Dynamic Friction Tester. Some 11 devices designed to measure the coefficient of friction and/or texture were used
in the trials. These included:

1 x SCRIM + device owned and operated by WDM Ltd collecting both SFC and Texture;
204

Chapter 7
5 x GripTester devices (including Mk 1 and Mk2 versions) from Findlay Irvine Ltd (the developers), Fulton
Hogan NZ, Transport South Australia, Auckland International Airport Ltd and Auckland University;

1 x ROAR device owned and operated by Pavement Management Systems;

1 x Portable British Pendulum Tester (PBPT) owned and operated by Fulton Hogan;

1 x Dynamic Friction Tester owned and operated by Auckland University;

1 x Vericom skid tester owned and operated by NZ Police;

1 x Skidman skid tester operated by John Bullas of the University of Southampton, UK and NZ Police;

1 x Mini-texture Meter owned and operated by Fulton Hogan;

Sand Circle Tests operated by Fulton Hogan;

1 x TNZ Laser profilometer owned and operated by Transit NZ; and

1 x T2GO device owned and operated by NZ Safe Zone.

A draft report of the results of the skid resistance test device trials has been published by Austroads (2005a).

7.6.2 Main Trial Test Site

The trial site was situated on South Eyre Road, Eyreton (near Clarkville) in the Waimakariri District Council, South
Island, New Zealand, approximately 20 kilometres north of Christchurch. The surfacings that were constructed
specifically for the main trials were chosen to represent different material types provided across road networks in
New Zealand. These included the existing Grade 3 chipseal surfacing and five new surfacings, purpose-laid during
April 2005 and shown in Figure 7-26. The new surfacings, each 100m long, comprised a Grade 6 chipseal, which
was laid across the full width of the site at its eastern end, and four newly-laid asphalt mixtures. On the westbound
lane, two asphalt mixtures were laid: an Ultra Thin Asphalt (UTA) and a Stone Mastic Asphalt (SMA). On the
eastbound lane, adjacent to the UTA and SMA, a dense asphalt concrete with 16mm coarse aggregate (AC16) was
laid. Part of this length was left smooth (with very low texture depth) and the remainder was provided with
transverse sawn grooves (AC16(G)) .

Figure 7-26: Schematic diagram of Christchurch Trial Test site (Austroads, 2005a)
Most of the surface sections were new (less than one month old) and existing traffic volumes on the road were low.
Therefore, the surfaces were mostly in the initial polishing phase where successive skid resistance tests by multiple
devices could have modified or influenced the polished state of the surface. For this reason, a comparison of the
results on the different surfaces should be treated with caution (Austroads, 2005a). However, a significant benefit of
new sufaces is that the section lengths should be more homogeneous both longitudinally and transversely and
therefore, a comparison or correlation between devices (especially between CFME devices and static devices) tested
on the same day and under the same conditions should be most appropriate. Furthermore, as full temporary traffic
control was in place and wheel track guide lines were marked on the road surface, the transverse locational
positioning between the devices was more accurate than normal in-field skid resistance testing. This comparative

205

Chapter 7
device analysis was not reported on by Austroads (2005a). However, the summary results obtained from the report
are further analysed and reported on here.

7.6.3 Reported Results

The analysis of the data collected in the Christchurch field trials was limited.

The following points were

summarised from the Austroads (2005a) report in relation to the various participating surface friction devices:

From an assessment of the between-run standard deviations and the coefficients of variation of the various
devices (with three test runs) on the differing surface sections, it was reported that the repeatability of all of
the devices was generally good and the levels were broadly consistent with experience of the authors from
other larger international exercises;

Each device has its own advantages and disadvantages;

Analysis demonstrated that different devices using different principles give different results and, whilst
they will often reflect the general friction condition of the road, the responses can be quite different from
each device;

The GripTesters, for example, respond differently to different textures when compared with the SCRIM
device and the ROAR which appeared to have been more variable on uneven surfaces;

One of the five GripTesters participating in the field trials gave noticeably higher results than the other four
similar devices. This suggested that there may have been a systematic fault with the machine or, possibly,
although less likely, that its driver consistently followed a different test line;

The measurement results were heavily influenced by the freshness of the new seals and therefore
comparison of the different surfaces should be treated with caution;

The low speed skid resistance levels were generally high on all of the sections, reflecting the new surfaces
and lightly-trafficked route.

7.6.4 Further Data Analysis

The Austroads draft report (2005a) reported on the results of the individual devices that partipated in the trials.
However, it did not statistically analyse or correlate the devices in comparison with each other. As there was a
reasonably good range of low friction surfaces to high friction surfaces as well as low texture to high textured sites,
reasonable correlation equations could be expected.

As the Dynamic Friction Tester has been shown to be very

repeatable, both in this research and by others (Henry 2000; Henry et al. 2000c; Henry 2003), coupled with the fact
that the pavement surface sections in the field trials were relatively new and therefore relatively homogeneous, it
was chosen as the most appropriate static device with which to compare the results of each of the other devices on
the trial surfaces.
Figure 7-27 displays the average section results of the five GripTesters with the Dynamic Friction Tester results on
the Christchurch test surfaces. All of the coefficients of determination (R2) are very good as they all explain over
90% of the variation. However, it can also be seen that one device (GripTester C) was measuring significantly
higher (approximately 0.12 to 0.20 GN higher) than the other devices. GripTester D also seemed to measure
considerably higher (approximately 0.07 GN higher) than GripTesters A, B and E on surfaces with high friction, and
more importantly, GripTester E reports higher (approx 0.10 GN higher) than GripTesters A, B and D on surfaces of
low surface friction. The consequences of measuring and reporting skid resistance incorrectly and specifically too
high, on surfaces with low skid resistance are, in terms of road safety indicators, more critical. Another correlation
exercise was undertaken at Sydney Airport, Australia in 2003 with five GripTesters which also reported one
206

Chapter 7
GripTester device that gave results significantly different from the other four. The author also participated and
helped process and analyse the results of this exercise (Dardano, 2005; Wilson & Dunn, 2004).
Both correlation exercises demonstrate the need for regular correlation and calibration trials, as the owner/operators
of the devices usually have no way of knowing whether their device is measuring correctly. This finding is not
something especially specific to the GripTester device as, in both correlation trials, the GripTester was the only
device with several participating. The same issue is prevalent for other devices, although some of the other devices
such as the UK based SCRIM devices go through annual correlation trials. This is recommended for all devices that
are commonly used. As reported in Dardano (2005) and Wilson and Dunn (2004), there needs to be more regular
and specific procedures developed for the Australasian region.
Distinction should also be made between correlation exercises and calibration of the devices as they are different
issues. For example, the GripTesters have set calibration procedures and, prior to both trials discussed above, the
devices were calibrated within the manufacturers guidelines. Yet it is only when a comparison of the results of
surveys is undertaken on the same surfaces at the same time and under the same conditions, that it becomes apparent
that a device is not measuring correctly.

GripT ester Device Correlated with Dynamic Friction T ester device

1.00
0.95

y = 1.0552x + 0.0094
R2 = 0.9784

0.90
GripNumber (GN50)

0.85

y = 0.7924x + 0.2862
R2 = 0.9302

0.80

y = 0.7147x + 0.257
R2 = 0.9038

0.75

y = 0.8264x + 0.1311
R2 = 0.9283

0.70
0.65
y = 0.8575x + 0.1129
R2 = 0.9889

0.60
0.55

GTA
GTD
Linear (GTC)
Linear (GTB)

0.50
y=x

0.45

GTB
GTE
Linear (GTA)

GTC
Linear (GTD)
Linear (GTE)

0.40
0.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

Dynamic Friction Tester (DFT20)

Figure 7-27: GripTester results as compared with the Dynamic Friction Tester (Christchurch trials)
A comparison of the other test devices at the Christchurch trials was also made against the Dynamic Friction Tester
and the results of this analysis are shown in Figure 7-28 (the GripTester shown is the University of Auckland
device). These results demonstrate that some devices correlated with the Dynamic Friction Tester better than others,
namely (in order of highest coefficient of determination), the GripTester, the Portable British Pendulum Tester, the
ROAR and the SCRIM device.

207

Chapter 7

Various devices correlated with Dynamic Friction T ester

Christchurch Correlation trials
1.00
y = 1.0552x + 0.0094
R2 = 0.9784

Coefficient of Friction (CoF)

0.95
0.90

y=x

0.85
0.80
0.75
0.70

y = 0.5047x + 0.3648
R2 = 0.7192

y = 0.61x + 0.2393
R2 = 0.7464

0.65
0.60
0.55

y = 0.3114x + 0.3773
R2 = 0.3995

0.50
0.45
0.40
0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

Dynamic Friction Tester (DFT20)

SCRIM50

GripTester GN50

ROAR

PSRT

Linear (SCRIM50)

Linear (ROAR)

Linear (GripTester GN50 )

Linear (PSRT)

Figure 7-28: All devices compared with the Dynamic Friction Tester (Christchurch trials)
Figure 7-28 clearly shows that the GripTester and the Dynamic Friction Tester are more sensitive to changes of skid
resistance and texture (the slope of the gradient is very close to 1) on the road surface than the ROAR and especially
the SCRIM (the least sensitive). Further analysis of the Portable British Pendulum Tester results (PBPT, a static
device) demonstrated that, if the results of the AC16 Grooved surface were removed from the analysis, then the
gradient of the equation line and the coefficient of determination between the PBPT and the Dynamic Friction
Tester were very similar to the values for the GripTester, with an R2 of 0.98. A correlation analysis could also be
undertaken between any of the devices that participated in the field trials. Table 7-2 shows the coefficient of
determination (R2) results of this analysis.
Whilst the true skid resistance of the road is not known, it is clear from the table that on these trial surfaces the
Dynamic Friction Tester results correlate very well with the GripTester and the Portable British Pendulum Tester
(when surfaces are not grooved), reasonably well with the ROAR and less with the SCRIM device. The other
devices explained less than 30% of the variation and the T2GO device (a new device used in push mode) showed no
correlation at all. Figure 7-29 gives an example of a between-device correlation with the SCRIM device and three
GripTesters for the Christchurch trial sites. Up to approximately 50% of the variation between the devices can be
attributed directly to the devices themselves.

208

Chapter 7
Surface
Friction
Device (R2)

DF Tester
(DFT20)

DF Tester

British
Pendulum
(PBPT)

SCRIM

GripTester

ROAR

Vericom

Skidman

T2GO

NA

PBPT

0.75 1

NA

SCRIM

0.40

0.39

NA

GripTester

0.95 2

0.84

0.46

NA

ROAR

0.72

0.81

0.60

0.79 2

NA

Vericom

0.21 3

0.38

0.08

0.26 2

0.43

NA

Skidman

0.26

0.31

0.07

0.32 2

0.40

0.99

NA

T2GO

0.02

0.0

0.0

0.0 2

0.06

0.07

0.03

NA

Notes:
1)
This value increases to R2 = 0.98 if the Grooved AC16 surface result is removed.
2)
The values quoted are the average of the 5 participating GripTesters (Range 0.90-0.99)
3)
Values quoted are with ABS brakes off. With ABS brakes on, R2 = 0.002
Table 7-2: Coefficient of Determination (R2) between surface friction devices from Christchurch trials

SCRIM Correlated with GripTester - Christchurch trials

1.00

GripTester (GN50)

0.95
0.90

y = 1.0933x + 0.099

0.85

R = 0.5133

0.80
0.75

y = 1.2713x - 0.0218

0.70

R = 0.3446

0.65

y= x

0.60
0.55

y = 1.2186x - 0.0247

0.50
0.45

R = 0.4898

0.40
0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

SCRIM (SFC50)
GT D

GT E

GT B

Linear (GT D)

Linear (GT E)

Linear (GT B)

Figure 7-29: GripTester correlated against SCRIM on Christchurch test trial sites
Figure 7-30 shows a correlation of the Dynamic Friction Tester with the Vericom and SkidMan. These are both invehicle systems that determine the coefficient of friction by the rate of deceleration during hard braking. The
Vericom system was used with non-ABS braking as well as with ABS braking. Figure 7-30 demonstrates the effect
of ABS braking systems. These keep the wheel braking oscillating around the peak slip speed that corresponds to a
peak coefficient of friction, which is significantly higher than when there is no ABS braking system. With no ABS
braking systems, the wheels are fully locked (100% slip speed) and the coefficient of friction significantly reduces.
However, the results shown in Figure 7-30 from the Christchurch trials are unusual as they show almost no gradient
209

Chapter 7
of the equation line. If true, this would indicate that it does not matter what the skid resistance is, because under
extreme braking there will be relatively little difference in the reported coefficient of friction, with or without ABS
brakes. This we know intuitively, to be untrue and therefore some explanation is required.

In-car Decelerometer Systems Correlated with

Dynamic Friction Tester device
Coefficient of friction (CoF)

1.10
Grade 6 Chip seal
surfaces

y = -0.0182x + 0.9023

1.00

R = 0.0015

0.90

y = 0.2159x + 0.4901

0.80

R = 0.2575

0.70
0.60

y = 0.2001x + 0.4979

y=x

0.50
0.40
0.40

0.45

0.50

0.55

R = 0.2088

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

Dynamic Friction Tester (DFT20)

SkidMan ABS Off Car 1

Vericom ABS Off Car 1

Vericom ABS On Car 2

Linear (SkidMan ABS Off Car 1)

Linear (Vericom ABS Off Car 1)

Linear (Vericom ABS On Car 2)

Figure 7-30: Dynamic Friction Tester and In-vehicle Decelerometers

On closer examination of the data and the corresponding test surfaces, it was noted that the surface which produced
the highest results for the DF Tester and all other commonly used devices (the Grade 6 chipseal) produced almost
the lowest results with the in-vehicle decelerometer dry braking systems. This is postulated to be a combined effect
of the very new nature of the seal surfaces that were being tested and the more destructive nature of a dry locked
wheel test. Whilst the normal skid resistance devices did not visibly change the nature of the seal, it was clear that
the braked wheel tests (especially the locked wheel non-ABS brake systems) caused visible damage to the seal
surface, especially on some of the seal sections such as the Grade 6 chipseal. It is possible that on the Grade 6
chipseals, due largely to the small size of the aggregate chips, fully locking the wheels in a dry sliding test broke the
adhesion of the relatively soft and viscous bitumen, and the aggregate and the seal chips were carried on a shear
plane at the bitumen aggregate interface. This would give significantly lower coefficients of friction than if the chip
seal had a reasonable period of time to age and harden. It was also visually apparent that some of the new surfaces
had aggregates that were still coated in bitumen and that, in a dry locked-wheel sliding context, the bitumen could
be heated and mobilised underneath the tyre. This would mask the aggregate microtexture/adhesion component of
skid resistance and reduce the measured coefficient of friction and provide an explanation for the much lower than
expected result (thereby flattening the gradient line of best fit to almost zero). This would also significantly affect
the coefficient of determination between these devices and other devices that were less destructive in their nature,
and that were light enough not to adversely change the surface during the test.

7.6.5 IFI Transformations

The test trials allowed the transformation of the reported device results in terms of their specific testing
methodology, to the International Friction Index (F60, Sp). The equations incorporate a Speed Number (Sp) variable,
that has proven to be highly correlated with the macrotexture of the surface, which is then used in conjunction with
210

Chapter 7
transformation constants for each device to standardise the skid resistance measurement to a 60km/h friction
coefficient (F60). The transformation exercise will help to determine whether the IFI transformation process is
appropriate for typical New Zealand surfacing types and whether the surface section coefficient of variation will
reduce, and correspondingly the device coefficients of determination increase, when using the IFI F60 to harmonise
between the devices. Figure 7-31a and Figure 7-31c show the results for the test trials for the five commonly used
devices in New Zealand for the various test surfaces in their originally reported form of coefficient of friction for the
westbound and eastbound directions respectively. Figure 7-31b and Figure 7-31d show the IFI F60 transformed
results for the corresponding directions of travel.
Surface Friction results from each of the trial participants

Transformed IFI F60 results from each of the trial

participants
1.10

Grade 6 Chipseal

1.00

1.00

0.90

0.90

0.80

0.80

Grade 3 Chipseal

0.70

Grade 3 Chipseal

IFI F60

Coefficient of Friction (CoF)

1.10

0.60

Grade 6 Chipseal
Grade 3 Chipseal

0.70

Grade 3 Chipseal

0.60
0.50

0.50
SMA

0.40

0.40

UTA

0.30

0.30

0.20

0.20
0

100

200

SCRIM50

300
400
500
600
700
Running Distance (Westbound)
GripTester GN50

ROAR

800

PBPT

900

SMA
0

1000

200

SCRIM50

DFT20

(a) Raw device CoF (Westbound)

300
400
500
600
700
Running Distance (Westbound)
GripTester GN50

ROAR

800

PBPT

900

1000

DFT20

(b) IFI F60 Transformed CoF (Westbound)

Transformed IFI F60 results from each of the trial
participants

Surface Friction results from each of the trial participants

1.10

1.10

Grade 6 Chipseal

1.00

1.00
Grade 6 Chipseal

0.90

0.90
Grade 3 Chipseal

0.80

0.80

Grade 3 Chipseal

0.70

IFI F60

Coefficient of Friction (CoF)

100

UTA

AC16G

0.60

AC16

0.50

0.40

0.40

0.30

0.30

0.20

0.20

100

200

SCRIM50

300
400
500
600
700
Running Distance (Eastbound)
GripTester GN50

(c) Raw device CoF (Eastbound)

ROAR

PBPT

800

900
DFT20

1000

Grade 3 Chipseal

0.60

0.50

Grade 3 Chipseal

0.70

AC16G
AC16

100

200

SCRIM50

300
400
500
600
700
Running Distance (Eastbound)
GripTester GN50

ROAR

PBPT

800

900
DFT20

(b) IFI F60 Transformed CoF (Eastbound)

Figure 7-31: Christchurch Trial surfaces Raw CoF and Transformed IFI F60
It was expected that if the IFI transformation equations for each device were appropriate for New Zealand
conditions, the variation between the IFI F60 results for each of the surfaces and devices would be relatively small.
However, it is visually apparent that this is not so, and in some surfaces the variation between the devices increases
when using the IFI F60 transformation in comparison to the individually reported device coefficients of friction (e.g.
the Grade 6 chipseal CoV increases from 13% to 16% between devices and on the SMA and UTA surfaces the CoV
increases from 12% to 20%). Whilst the Grade 6 chipseal surface is not very common overseas, the SMA and UTA
type surfaces are more common. Results could be affected differently with the various devices due to the newness
of the surfaces.

211

1000

Chapter 7
A closer examination of Figure 7-31b and Figure 7-31d shows that the increased Coefficient of Variation between
the devices is largely due to the GripTester and SCRIM device results. However, the GripTester results are an
average of the four devices that correlated well together (excluding GripTester C) and whilst they are in step with
the other devices, show significantly higher IFI F60 results than these. This can be clearly seen in Figure 7-32
where the results of each of the transformed IFI F60 results are correlated bi-variately against the IFI F60 results of
the Dynamic Friction Tester. This analysis shows that, whilst a very good coefficient of determination (R2=0.97)
exists between the DF Tester IFI F60 results and the GripTester IFI F60 results, the transformation coefficients do
not appropriately modify the results for these types of surfaces. This is also true for the SCRIM device, the
coefficient of determination (R2=0.52) explains only just over 50% of the variation between the two devices.
However, for these test surfaces, an adjustment to the IFI equation coefficients for the GripTester and SCRIM
results could adequately modify the results to better correlate together, although some caution is required, as the data
set is much more limited than the original PIARC experiment from which the coefficients were developed. This
may also be an anomaly due to the newness of the surfaces.

DF Tester IFI F60 compared with other devices IFI F60 Christchurch test trials

Coefficient of Friction (IFI F60)

0.90
y = 0.5672x + 0.2979
R2 = 0.5159

0.85
0.80
0.75
0.70

y = 1.1126x + 0.0718
R2 = 0.9718

0.65
0.60
0.55
0.50

y=x

0.45
0.40
0.35
0.30
0.30

y = 0.4909x + 0.2527
R2 = 0.7669

y = 0.7045x + 0.1179
R2 = 0.7658

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

Dynamic Friction Tester IFI F60

SCRIM50

GripTester GN50

ROAR

PBPT

Linear (SCRIM50)

Linear (ROAR)

Linear (GripTester GN50 )

Linear (PBPT)

Figure 7-32: Correlation of the DF Tester device with other devices using IFI F60 Transformations
This call for caution in modifying the transformation constants is reinforced when the results of the correlation trials
undertaken at Sydney Airport in 2003 on various test surfaces (Dardano (2005)) are considered. That correlation
exercise had most of the same GripTester devices participating and the results demonstrated a very good correlation
of IFI F60 with the Dynamic Friction Tester and GripTester (R2=0.96) and SCRIM21 (R2=0.98) devices as shown in
Figure 7-33. Interestingly however, the ROAR which had one common device in both trials did not correlate well
in the Sydney trials but did in the Christchurch trials.

21

It should be noted that the SCRIM devices used in the Australian correlation trials were different from the WDM Ltd
Christchurch SCRIM device that applies a 0.78 multiplication factor. The Australian devices do not.

212

Chapter 7
Australian Friction Workshop - August 2003
DF Tester vs CFME
F60 Comparison
0.90
0.80

Friction Result (F60)

0.70

ROAR
y = 0.9227x + 0.1882
R2 = 0.7392

0.60
0.50
0.40
0.30
0.20

GT
y = 1.0278x - 0.0248
R2 = 0.9637

0.10
0.00
0.00

0.10

0.20

0.30

0.40

0.50

SCRIM
y = 1.0947x - 0.0569
R2 = 0.9753

0.60

0.70

0.80

DF Tester Results (F60)

ROAR

GT

SCRIM

Linear (SCRIM)

Linear (GT)

Linear (ROAR)

Figure 7-33: IFI F60 Correlation Results of the Australian Friction Workshop Between the DF Tester,
SCRIM, GripTester and ROAR - Dardano (2005)

7.7

Summary Device Discussion

As has been shown in this chapter, comparing the results of one skid resistance device with those of another device
is a difficult and complex task. There are many factors and variables that need to be considered, including:

The methodology and operating mode of the device;

The sensitivity of the device to temperature and other environmental factors;

The rubber/surface contact area and the differences between devices in vertical loads, slip speed, compound
and hardness of the rubber, the tyre pressure (if applicable) and size of measuring wheel or rubber slider;
and

The surface age and macrotexture.

A number of international and local correlation equations have been developed from previous correlation trials and
research. Nevertheless, this research has shown that, at a local level, these equations still cannot explain the real and
significant variations that do occur in the field between devices which can also be experienced by road users.
Additional research is required to more predictably determine real changes in skid resistance and to enable better
correlation between devices that take account of differences experienced in the field.

213

Chapter 8

8 SEASONAL AND SHORT-TERM VARIATIONS

8.1

Introduction

A number of researchers have shown, that after initial phases of aggregate polishing, skid resistance varies
throughout the year in a seasonal or cyclical annual pattern, with generally low skid resistance in the summer and
high skid resistance in wetter winter periods (P. D. Cenek et al., 1999; Hill & Henry, 1981; J.R. Hosking, 1976;
J.W.H. Oliver et al., 1988; Rogers & Gargett, 1991). The way in which skid resistance varies over time has been
idealised in model form by Prowell et al. (2003). It is shown in an idealised form in Figure 3-9 and discussed in
Section 3.5.6. Existing models which have been developed to predict changes in skid resistance are discussed in
Section 4.6.5 This predictable effect has not been universally agreed upon. As Oliver et al. (1988) reported, in
some climates (e.g. Queensland, Australia) the seasonal effect was not evident. They also showed that the results
from year to year in other states of Australia could vary considerably and were not predictable.
Rogers and Gargett (1991) concluded from research in the UK that the seasonal variation was due to the combined
effects of traffic and weather. When roads are dry, the polishing effect of traffic tends to dominate, but when they
are wet for prolonged periods, they recover some of their former harshness. The scale of the seasonal effects
depended largely upon the geological history and petrography of the aggregates.
A secondary and additional short-term effect was reported by Bird and Scott (1936) in the UK who found that skid
resistance levels varied significantly with changing weather conditions. They reported that, as rain fell, there was an
immediate drop in skid resistance from that prior to a rainfall event. As more rain fell, the skid resistance level
increased as contaminants were washed away and then it increased again as the pavement dried out. This effect was
shown by Bennis and De Witt (2003), in Figure 3-14.
Hill and Henry (1981) stated that large variations in skid resistance from day-to-day or week-to-week, seemingly
occur due to the rainfall pattern and local weather conditions that are superimposed on an annual cycle. They
demonstrated that models could be developed in the US to adjust significant seasonal and short-term skid resistancerelated variations and explain up to 57% of the measured variation. They also found that a major cause of apparent
skid resistance variation is measurement error, particularly the lateral placement of the test tyre.
This chapter presents an analysis of the field test site results for seasonal, aggregate weathering and/or polishing22
effects due to variances in geometric elements, and the short-term environmental-based variation of skid resistance
measurements that were collected at the field test sites with the GripTester over an approximately 2.5 year period.

8.2

Seasonal Variation Results

8.2.1 Introduction
As discussed in Section 5.6, standardised field testing procedures were developed to ensure a consistency of test
methodology and experimental approach. The sites used in this analysis were sites that had a weathered surface,
meaning that the surface should be acting within the equilibrium skid resistance phase. The sites were also
geometrically benign in that they were reasonably straight and almost level in grade. Therefore, any changes in skid
resistance were expected to be due only to the seasonal and or short-term effect. This analysis investigated these
effects in the field by:

214

Chapter 8
Measuring the extent or range of seasonal variation of the measured skid resistance by minimum monthly
skid resistance testing at sites or timeslice locations where it was thought that an equilibrium level of skid
resistance existed (specifically the Tamaki Campus site, the Ports of Auckland site, the Hikurangi site,
Section 2 of the Snooks-Tatoon site and timeslices taken prior to new seals at the Brynderwyn site, Kara
Road site and the Kaiwaka slag site);

Recording manual rainfall pluviometer records, desirably daily but often less regularly, at each of the five
sites in the Northland region;

Obtaining automatic NIWA rainfall records as close to all of the sites as possible (including the Ports of
Auckland and the Tamaki Campus sites) and, where applicable, correlating these against the manual
rainfall gauge data to enable interpolation for days that manual rainfall information was not available;

Collecting on-site data such as air and surface temperature during skid resistance testing; and

Whenever possible, collecting surface detritis samples for subsequent environmental laboratory analysis.

The variation at each of the field sites is individually described in Section 6.4. However, for a comparative analysis,
a greater number of data points is desirable to determine whether there are common trends occurring across similar
field sites and across the data collection period. To enable this analysis, a normalisation of the individual field
section GripTester coefficient of friction results to a common level (e.g. 1.0 GN) was initially required to remove
the differences in terms of the polished state of the aggregate. It is recognised that differences in the geological
makeup of the surface aggregate may affect the extent of variation demonstrated for the respective aggregates.
However, the normalisation process was expected to help reveal these effects. The normalisation process allows a
combination of similar field sites, providing a greater number of data points, thereby ensuring greater confidence in
any statistical relationships observed in the measured skid resistance data. The normalisation process consisted of
modifying each field recorded mean test section result (GNj) for each test date (u) to a common Normalised

GripTester result (NGNu), based upon the sample mean result ( x ) for the field site being shifted linearly to a
normalised mean section result for the data collection period of 1.0:

NGN u = 1 x + GN j
where:

Eq 8-1

NGNu = the Normalised GripNumber test result for test date u

x = the surface section mean GripNumber result for the sample test period where:

x=

xiju
i

nmk

i: where i=1 to 5 reported GripNumber (GN) test runs for a specific 10m section;
j: where j =1 to m reported GripNumber (GNi) test runs for the section distance;
u: where u=1 to k reported GripNumber (GNj) for the number of test dates.
GNj = the mean section GripNumber for test date u

22

The word polishing is used loosely in this chapter to describe the process of aggregate grinding under traffic with
contaminants. It is not used specifically to mean polishing that results in a reduction of surface friction, as in some cases
(depending upon the contaminants) an increase in skid resistance will result.

215

Chapter 8
Each data point from each field site section was modified by this process. The following sections give the analysis
of these combined normalised results.

8.2.2 Auckland Asphalt Mix Sites

A descriptive statistical analysis of the Tamaki Campus site is given in Section 6.4.2 and shown in Figure 6-12 and
Figure 6-13 for the flat and sloped sections respectively. Similarly, the Ports of Auckland site is discussed in
Section 6.4.3 and the results are shown in Figure 6-14 and Figure 6-15 for the increasing and decreasing directions
respectively. As both the Tamaki Campus and Ports of Auckland surfaces are of similar asphalt mix type surfaces
and have been shown to be both acting in an equilibrium phase the various site section results from the two similar
field sites have been normalised by the process explained above.
Figure 8-1 shows the GripTester (GN) results of the two Auckland Asphalt mix sites over the three year data
collection period. The asphalt mix surface has polished to different average section levels, due primarily to the
significantly higher traffic loads at the Ports of Auckland site in comparison to the relatively low traffic loads at the
Tamaki Campus site. It should also be noted that the Tamaki Campus site skid resistance testing was undertaken at
a vehicle testing speed of 30km/h, but the Ports of Auckland site testing speed was 50km/h. It is apprarent in Figure
8-1 that in the majority of cases when one site is reading high or low the other site is similarly high or low.

Auckland Sites - Asphalt Mix Surfaces, Seasonal Variation (2 Sites )

Both Directions
1.20

Skid Tester not w orking

properly during this period

Water delivery problems causing

higher than expected results

GripNumber (GN)

1.00

0.80

0.60

0.40

0.20

Tamaki Flat West GN30

Tamaki West Slope GN30

Tamaki East Flat GN30

Tamaki East Slope GN30

POAL West GN 50

POAL East GN 50

7-Jul-05

18-May-05

29-Mar-05

7-Feb-05

19-Dec-04

30-Oct-04

10-Sep-04

22-Jul-04

2-Jun-04

13-Apr-04

23-Feb-04

4-Jan-04

15-Nov-03

26-Sep-03

7-Aug-03

18-Jun-03

29-Apr-03

10-Mar-03

19-Jan-03

30-Nov-02

11-Oct-02

22-Aug-02

3-Jul-02

0.00

Time

Figure 8-1: Seasonal Variation of Auckland Asphalt Mix Surfaces over 3 Year Period
When the GripTester (GN) results for each section of the two asphalt mix surfaces in Auckland are normalised
(NGN), and the known and identified outliers removed, a comparison can be made, by seasonal calendar months, to
determine whether significant monthly trends are visually apparent from the data. Figure 8-2 is a plot of this
relationship and Table 8-1 gives the statistical data by calendar month of the year. The average monthly results for
the asphalt mix surfaces are not conclusive. The data for each month are scattered. However, it is apparent that
higher results do occur in the winter months, in comparison to the summer months. The lowest normalised average
216

Chapter 8
monthly result is February with a reported NGN value of 0.94, a coefficient of variation of 2.5% and a 95%
confidence interval of 0.019. The descriptive statistical analysis therefore demonstrates that there was a 95%
probability that the Normalised GN value in the month of February would fall between the values of 0.92 and 0.96.
Conversely, for the month of August, an NGN value of 1.05 was calculated (the highest normalised average monthly
result) with a coefficient of variation of 2.9%. The 95% confidence interval is also 0.019 which means that there is
a 95% probability that the Normalised GN value in the month of August would fall between the values of 1.03 and
1.07. These two months are separated by a six month time period, indicating that possibly seasonal variations do
exist, but these are offset with the effects of the polishing occurring late in summer and vice versa in winter where
the rejuvenation is greatest in late winter. The results either side of these calendar months are significantly more
variable, with no distinct patterns and coefficients of variation increasing in the months of March, May and June to
8.4%, 9.3% and 8.0%, respectively. In these months, the 95% confidence intervals ranges between 0.036 and
0.060 which means that there there is no confidence at the 95% level that the result will be either higher or lower
than the normalised average monthly data.
It should be noted that the number of data points varies for each month and ranges from as low as 6 to 22, the lowest
month being February. It should also be taken into account that the Ports of Auckland site is not a typical
longitudinal wheel path polishing site and whilst the loads are significant, they are mostly transverse in nature.
Furthermore, the Tamaki Campus site is a very lightly trafficked internal campus road and therefore would not be
expected to demonstrate typical urban or State highway polishing mechanisms. This may explain the more random
nature of the results that display little predictable pattern by month of the year. It could also explain the greater
variation between the reported results. The Northland State highway results (discussed in the next section) are
expected to give more predictable patterns.

Auckland Sites, Normalised GN - Asphalt Mix Surfaces

(Both Directions)

1.20
1.15

Normalised NGN

1.10
1.053

1.05
1.012

1.032

1.00

1.028

1.013

1.008

0.994

1.014

1.006

0.980

0.95

0.962

0.936

0.90
0.85
0.80
0

Tamaki Flat W30

Tamaki Flat E30
Normalised Month Average

5
6
7
8
Calendar Month of Year
Tamaki Slope W30
Tamaki Slope E30

10

11

12

13

POAL Norm W50

POAL Norm E50

Figure 8-2: The Normalised GripTester Results (NGN) for the Auckland Asphalt Mix Sites by Month of Year

217

Chapter 8

All Auckland Sites (Both directions) Normalised NGN results by Months of the Year
Sample Section
Descriptive Statistics

Jan
Month
1
Sample Count
17
Mean
0.99
Maximum
1.16
Minimum
0.95
Range
0.21
Standard Deviation
0.047
95% Confidence Interval
0.023
Variance
0.002
Coefficient of Variation (CoV) 0.048

Feb
Month
2
6
0.94
0.96
0.91
0.06
0.023
0.019
0.001
0.025

Mar
Month
3
8
1.03
1.12
0.89
0.22
0.087
0.060
0.007
0.084

April
Month
4
16
0.96
1.05
0.91
0.14
0.041
0.020
0.002
0.042

May
Month
5
10
1.01
1.13
0.90
0.23
0.094
0.059
0.009
0.093

June
Month
6
20
1.01
1.18
0.91
0.27
0.080
0.036
0.007
0.080

July
Month
7
17
1.03
1.18
0.97
0.20
0.054
0.026
0.003
0.053

August Sept
Month Month
8
9
10
22
1.05
0.98
1.11
1.09
1.02
0.92
0.08
0.17
0.031 0.043
0.019 0.018
0.001 0.002
0.029 0.044

Oct
Month
10
18
1.01
1.15
0.93
0.22
0.064
0.030
0.004
0.063

Nov
Month
11
22
1.01
1.08
0.97
0.11
0.027
0.011
0.001
0.027

Dec
Month
12
9
1.01
1.14
0.93
0.20
0.076
0.050
0.006
0.075

Table 8-1: Normalised GripTester results for Auckland Sites by Calendar Month of Year

8.2.3 Northland State Highway Site Results

A descriptive statistical analysis is reported for each of the Northland State Highway field sites in Sections 6.4.4 to
Section 6.4.8 for the Kaiwaka Slag site, Brynderwyn South curve, Hikurangi site, Kara Road site and the SnooksTatton Roads site, respectively. The time-averaged results for these sites are also shown in Figure 6-16 to Figure
6-20, respectively. All of these sites are of similar chip seal type surfaces (although the Kaiwaka site utilises an
artificial aggregate). Furthermore, each of the sites has sections that have been tested over the two year data
collection period, that should have been acting in an equilibrium skid resistance phase and that can be appropriately
analysed for seasonal variations. Some of the sections chosen for this analysis are averages for a reasonable section
length and others are averages of 30m timeslices as identified below:

Kaiwaka Slag site (adjacent old seal), 30m timeslices at T/S location: 800m in increasing direction and T/S
location 100m in decreasing direction;

Brynderwyn South curve site (adjacent old seal), 30m timeslices at T/S location 490m: in increasing
direction and T/S location 110m: in increasing direction;

Hikurangi site, full section averages for both increasing and decreasing directions;

Kara Road site (adjacent old seal), 30m timeslice at T/S location: 40m increasing direction;

Snooks-Tatton Section 2, middle old seal section between two adjacent new seals in increasing direction.

Figure 8-3 and Figure 8-4 show the GripTester (GN) results for the Northland chip seal surfaces for those increasing
and decreasing directions over the two year data collection period selected for the seasonal analysis.

An

examination of Figure 8-3 and Figure 8-4 shows that all of the sites are acting in an equilibrium polished phase as
the variation is relatively random about the section average from the start, and through to the end periods of data
collection. There are no obvious areas of deterioration due to polishing other than normal seasonal type variations.
It is also apparent that some regional seasonal pattern effects occur between the sites because similar patterns of
high and low recorded results mostly coincide between the sites. It is also apparent that there are localised
differences between the sites as well, meaning that localised factors are effecting the measured skid resistance.

218

Chapter 8
Northland PSMC 02 Sites Seasonal Variation (5 Sites ) - Incr Dir
1.00
Skid Tester not working
properly during this period

0.90
0.80

GripNumber (GN)

0.70
0.60
0.50
0.40
0.30
0.20
Hik Inc GN

0.10

S&T Inc S2 GN

KSS T/S 800 Inc

Bryn T/S 490m Inc

Kara T/S 40m Inc

18-May-05

29-Mar-05

7-Feb-05

19-Dec-04

30-Oct-04

10-Sep-04

22-Jul-04

2-Jun-04

13-Apr-04

23-Feb-04

4-Jan-04

15-Nov-03

26-Sep-03

7-Aug-03

18-Jun-03

29-Apr-03

10-Mar-03

0.00

Time

Figure 8-3: Seasonal Variation of Northland Chipseal Surfaces over 2 year Period, (Increasing direction)
Northland PSMC 02 Sites Seasonal Variation (3 Sites ) - Decr Dir
1.00
Skid Tester not working
properly during this period

0.90
0.80

GripNumber (GN)

0.70
0.60
0.50
0.40
0.30
0.20
Hik Decr GN

0.10

KSS T/S 100 Dec

Bryn T/S 110 Decr

18-May-05

29-Mar-05

7-Feb-05

19-Dec-04

30-Oct-04

10-Sep-04

22-Jul-04

2-Jun-04

13-Apr-04

23-Feb-04

4-Jan-04

15-Nov-03

26-Sep-03

7-Aug-03

18-Jun-03

29-Apr-03

10-Mar-03

0.00

Time

Figure 8-4: Seasonal variation of Northland Chipseal Surfaces over 2 year period, (Decreasing direction)
As with the Auckland sites, the GripTester results for the Northland chipseal sites were normalised. The known and
identified outliers were removed and a comparison made by seasonal calendar months of the year to determine
whether significant monthly trends are apparent from the data. Figure 8-5 is a plot of this relationship with the
average normalised NGN. Table 8-2 gives the statistical data by calendar month of the year. Whilst the data still
demonstrates a significant range and variation of normalised GripNumber results especially for some calendar
months (e.g. May, June and July), there does seem to be a more distinct pattern in terms of low normalised skid
resistance values in late summer (e.g. April and May) and high normalised skid resistance for mid to late late winter
(July and August).
219

Chapter 8
The lowest normalised average monthly result for the Northland sites is April with a reported NGN value of 0.96, a
coefficient of variation of 4.7% and a 95% confidence interval of 0.017. The descriptive statistical analysis
therefore demonstrates that there was a 95% probability that the NGN value in the month of April would fall
between the values of 0.94 and 0.98. By comparison, for the month of August, a reported NGN value of 1.04 has
been calculated (the highest normalised average monthly result) with a reported coefficient of variation of 2.2%.
The 95% confidence interval is also 0.017 which means that there was a 95% probability that the NGN value in the
month of August would fall between the values of 1.02 and 1.06. The 95% confidence limits are plotted on Figure
8-5 which shows that significant data fall outside this range.
These results are consistent with some of the normalised indicators for the Auckland asphalt mix surfaces (discussed
above), whereby low measured skid resistance results are reported late in the summer (March, April) and higher
measured skid resistance is apparent in July and August.

All Northland Sites, Normalised NGN - Chip Seal Surfaces

(Both Directions)

1.20
1.15

Normalised NGN

1.10
1.05
1.04

1.00

1.02

1.02

0.95

1.00

1.00

0.99

0.98

1.03

1.02

1.00

0.99

0.96

0.90
0.85
0.80
0

KSS T/S 800m Inc

S&T S2 Ave Inc
Norm Ave GN

5
6
7
8
Calendar Month of Year

BRYN T/S 490m Inc

KSS T/S 100m Dec
95% Confidence Limits

HIK Sect Ave Inc

BRYN T/S 110m Dec

10

11

12

13

Kara T/S 40m Inc

HIK Sect Ave Dec

Figure 8-5: The Normalised GripTester Results (NGN) for the Northland Chipseal sites by Month of Year
The results either side of these calendar months are more variable (e.g. January, May, June and October) with
coefficients of variation increasing to 7.5%, 8.7% and 6.7% and 6.7% respectively. In these months, the 95%
confidence intervals range between 0.027 and 0.051 which means that there there is no confidence at the 95%
level that the result will be either higher or lower than the normalised average monthly data. It should be noted that
the number of data points vary for each month, ranging from 8 to 27, the lowest month being January.

220

Chapter 8
All Northland Sites (Both directions) Normalised NGN results by Months of the Year
Sample Section
Descriptive Statistics

Jan
Month
1
Sample Count
8
Mean
0.98
Maximum
1.11
Minimum
0.90
Range
0.21
Standard Deviation
0.074
95% Confidence Interval
0.051
Variance
0.005
Coefficient of Variation (CoV) 0.075

Feb
Mar
Month Month
2
3
14
27
1.02
0.99
1.06
1.08
0.97
0.90
0.09
0.19
0.030 0.047
0.016 0.018
0.001 0.002
0.030 0.047

April
Month
4
26
0.96
1.07
0.88
0.19
0.045
0.017
0.002
0.047

May
Month
5
21
1.00
1.19
0.90
0.29
0.087
0.037
0.009
0.087

June
Month
6
20
1.02
1.14
0.90
0.24
0.068
0.030
0.005
0.067

July
Month
7
26
1.02
1.11
0.92
0.19
0.045
0.017
0.002
0.045

August Sept
Month Month
8
9
7
15
1.04
1.00
1.07
1.05
1.01
0.94
0.07
0.11
0.023 0.031
0.017 0.015
0.001 0.001
0.022 0.031

Oct
Month
10
24
1.00
1.14
0.91
0.23
0.067
0.027
0.004
0.067

Nov
Month
11
24
0.99
1.05
0.90
0.15
0.043
0.017
0.002
0.044

Dec
Month
12
14
1.03
1.08
0.99
0.09
0.027
0.014
0.001
0.026

Table 8-2: Normalised GripTester results for Northland Sites by Calendar Month of the Year
Further anlysis was undertaken to test the hypothesis that sorting the dates into the four quarters of the year, would
confirm the seasonal effect across quarters, thereby demonstrating lower NGN results in the summer quarters and
higher results in the winter quarters. The results of this analysis are shown in Figure 8-6 which suggests that lower
normalised skid resistance results occur in the first two quarters (0.98 and 0.95 NGN respectively) and higher values
occur in the third and fourth quarters (1.02 and 1.0 NGN).

All Northland Sites, Normalised NGN - Chip Seal Surfaces

(Both Directions)

1.20
1.15

Normalised NGN

1.10
1.05
1.016

1.00

1.012

0.984
0.953

0.95
0.90
0.85
0.80
0

1
KSS T/S 800m Inc
KARA T/S 40m Inc
BRYN T/S 110m Dec

2
3
Calendar Quarter of Year
BRYN T/S 490m Inc
S&T S2 Ave Inc
HIK Sect Ave Dec

HIK Sect Ave Inc

KSS T/S 100m Dec
Norm Ave GN

Figure 8-6: The Normalised GripTester Results (NGN) for the Northland Chipseal sites by Quarter of Year

8.2.4 Seasonal Variation Summary discussion

The analysis described above determines the average section Normalised GripNumber (NGN) results for a number
of field test sections, including the comparison of these results for calendar months and quarters of the year. It
suggests that a seasonal polishing/rejuvenation cyclical effect does exist. However, the scatter of the results and the
variation within the months or quarters of the year also demonstrates that other factors are involved, needing to be
accounted for in future prediction methods. These factors (e.g. rainfall and surface contaminants) can cause
significant short-term variations in the measured coefficient of friction and are considered in the next section.
221

8.3

Short-Term Variations

Chapter 8

8.3.1 Introduction
As discussed in Sections 3.5.7 and 4.6.5, Hill and Henry (1981) state that large variations in skid resistance from
day-to-day or week-to-week, seemingly occur due to rainfall patterns and local weather conditions, these being
superimposed on an annual cycle. They developed a model that expresses the value of skid resistance at any time
(SNO) as a function of the short-term residual variation in measured skid resistance (SN0R), the long-term residual
variation in measured skid resistance (SNOL) and the measure of (SNO) that is independent of short and long-term
variations (refer to Eq 8-2 in Section 4.6.5). They also found differences between surface types for the value of long
term variations in skid resistance, e.g. a negative exponential relationship for asphalt surfaces and a linear increase
in skid resistance with time for Portland Cement Concrete (PCC) surfaces. From this, they demonstrated that
models could be developed in the US to adjust for significant seasonal and short-term skid resistance-related
variations, explaining up to 57% of the measured variation.
Cenek et al. (2003a) confirmed the rainfall/skid resistance relationship from skid testing results in the Wellington
region of New Zealand. The results demonstrated that the time lapse since the last rainfall can have a significant
short-term negative effect on skid resistance. The measured variation (using a GripTester, i.e.GN), apparently due
to the combined effects of pollutants and rainfall, demonstrated a reduction in measured skid resistance of at least
10% over 8-11 days since the last rainfall (refer to Figure 8-7). The four data points shown were each the average
result of six different sites on separate days, and indicate the trend. Wilson et al. (2003) and Wilson and Dunn
(2004), tentatively indicated a similar trend from measurements at the Hikurangi site.23 with both the GripTester and
the SCRIM device at different times.
An observation in the Cenek et. al (2003a) research study was:
Experience and common sense indicates that [skid resistance] cannot continue to decrease at
the same rate indefinitely if there is a prolonged period without rain. It is likely that the rate of
reduction is greatest within the first 2 weeks after rain. The data in [Figure 8-7] supports
this view with some indication that the rate of reduction decreases as time goes on a
reasonable hypothesis might be that the rate of reduction is logarithmic. For example, a
10% reduction in skid resistance occurs after 10 days whereas a 20% reduction occurs after
100 days In conclusion, the study does show a clear effect on skid resistance caused by
previous rainfall history. However, the magnitude and form of this effect is not proven.

23

Which has since proven to have the highest range of skid resistance variation of the Northland sites refer to Section 6.5.1 and
specifically Table 6-9.

222

Chapter 8
Each point was the
average of six different
sites on separate days.

Figure 8-7:Measured Decline in Skid Resistance due to Environmental Effects (P.D. Cenek & Davies, 2004)
The corresponding increase in measured skid resistance, apparently due to washing and traffic action during a
rainfall event was also reported by Cenek et al. (2003a; 2004) from eight successive runs using the GripTester (GN)
(refer Figure 8-8).

The overall measured increase (approx 10%) is, as would be expected, in the same order of

variation as the decrease in measured skid resistance due to the combined effect of pollutants and time lapse since
the last rainfall.

These measured short-term variations indicate that skid resistance does change over very short

time intervals. Such effects are reported in the following sections.

Figure 8-8: Measured increase in skid resistance during rainfall event (Cenek et al., 2004)

8.3.2 Rainfall Data

Rainfall data were collected for this research at the chosen field sites by a combination of two methods:

By the recording (where possible) of daily rainfall records by Works Infrastructure staff in Northland from
manual pluviometers installed in a protected position adjacent to the field sites; and from

The closest National Institute of Water and Atmosphere (NIWA) automatic rainfall gauge or Auckland City
Council operated automatic rainfall gauge stations.
223

Chapter 8
For the Tamaki Campus and the Ports of Auckland sites, rainfall data from the closest automatic rainfall gauge
stations were used. The Tamaki Campus site had a full automatic weather station on site. For the Ports of Auckland
site, data from the Auckland City Council-operated Albert Park rainfall gauge station were used.
Due to the distance between the Northland field sites and to the closest NIWA rainfall gauge stations, manually
recorded pluviometer data were used to correlate automatically recorded NIWA data with the on-site rainfall data at
the Northland sites. The closest NIWA rain gauge stations used for the Northland sites and their associated
proximity to the five field sites are shown in Figure 5-4. A correlation analysis of the manually recorded data from
the 1st January 2003 to the 10th November 2003 was used for each of the five Northland sites to determine which
NIWA rainfall gauge stations best represented the on-site rainfall conditions. In some cases this was not the closest
gauge station. It was found that in some cases, using the recorded data from two rain gauge stations gave a better
correlation than with just one gauge station. The best correlated results obtained by this analysis were then used to
modify the NIWA rainfall recorded data to the on-site rainfall data used in subsequent analysis. The results of this
analysis and the chosen gauges for the rainfall analysis with measured skid resistance are shown in Table 8-3 for the
five Northland sites.
Northland Field
Site

NIWA Gauges and

Weighting for site

Kaiwaka Slag

A54821

Brynderwyn
South

NIWA Gauge
Name

Distance to
Site

Coefficient of
Determ(R2)

Respective Linear
Modification factor 2

Mangapai

39.2kms

0.90

0.8044

A54821

Mangapai

39.2kms

Not available 1

0.8044

Hikurangi

A54623 (48%) &

A54737 (52%)

Ruatangata &
Whangarei Aero

8.6 kms &

18.4 kms

0.81 & 0.87

1.2034 & 1.4541

Kara Road

A54623 (52%) &

A54821 (48%)

Ruatangata &
Mangapai

9.0 kms &

13.0kms

0.82 & 0.75

1.0817 & 1.2723

Snooks Tatton

A54623 (50%) &

A54821 (50%)

Ruatangata &
Mangapai

9.0 kms &

13.0kms

0.80 & 0.81

0.9855 & 1.2734

Notes:
1)
A problem with the manual rainfall gauge at this site meant that it was decided to use the adjacent Kaiwaka
data which was 10kms south of the Brynderwyn South site.
2)
The linear modification factors are used respectively to multiply the rainfall gauge station data that gave the
highest correlation with the manual rainfall data recorded at the site.
Table 8-3: Correlation of NIWA Rainfall Gauges with Manually Recorded Data
The rainfall data for all sites were recorded as the total depth in millimetres (mm) of rainfall per day. The intensity
of rain could therefore not be taken into account, although this is expected to have some effect on measured skid
resistance. There are also some significant differences in the methods of recording the rainfall data that could affect
the subsequent analysis. The NIWA rainfall gauge stations use a 24hour clock that sums the rainfall for the previous
day until 6:00 am the next morning. The manual rain gauges can only determine the total depth of rain since the last
reading (at best over a 24 hour period). The time of this reading was not always consistent, although most readings
were undertaken early to mid morning. The timing of the recorded rainfall data and the timing of the skid resistance
test becomes important when analysing the previous rainfall history and when deciding whether rain occurred prior
to a test at the site or after a period of rain. At times this was difficult to determine, as rainfall patterns in Northland
are often irregular and, on many occasions, the data collection team would leave Auckland in fine weather, and by
the time of arrival at the site (up to 2.5 hours later), a number of showers had occurred. Furthermore, the five sites
were tested during the day between approximately 9:00am and 5:00pm before returning to Auckland, which could
have resulted in quite different testing conditions during that day. On a few occasions when repeat tests were
224

Chapter 8
undertaken on the same site in the morning and then later in the afternoon, significantly differrent results were
recorded.
The following section discusses a number of rain functions that were used to characterise the amount of rain that had
occurred at the field sites over a certain previous length of time. This was required to determine what effect the
previous rainfall history had in terms of affecting measured skid resistance.

8.3.3 Rainfall Functions

Various rain functions were used to compare the effect of the rainfall history prior to a measured coefficient of
friction. Hill and Henry (1981) used a Dry Spell Factor (DSF) featuring a logarithmic function of the number of
days plus one since the rainfall was greater than 2.5mm, for a maximum period of the previous 7 days. They also
compared the Dry Spell Factor results to a Weighted Rain Function (WRF) that had been developed earlier by Dahir
et al. (1979). The WRF is a sum of the total for each day of the amount of rain on a given day divided by the
number of days preceding the skid resistance test, up to a maximum of the previous five days. These rain functions
are discussed in Section 3.5.7 and were assessed in this research along with others. However, it was not known how
long an effect rain or the lack of rain, lasts on the measured pavement skid resistance. Therefore, in this research, a
range of rainfall periods was trialled. As manual rainfall pluviometers were used at the site, the effect of the
intensity of rainfall was not addressed even though it is recognised that this may have a significant effect on the
washing/cleaning of detritus off the road surface, thereby affecting the measured coefficient of friction. A range of
previous rainfall periods was used to compare with the measured skid resistance (most commonly 5 days, 7 days and
20 days). The rain function that produced the highest correlation was then used for further analysis. The rain
functions that were trialled, and the periods of rainfall history, prior to a skid resistance test, that were analysed, are:

Days Since Last Rainfall (DSLR) > x mm of rainfall. This simple rain function measure records the
number of days prior to the skid resistance test until x mm of rainfall occurred. It is a relatively crude
measure that does not take into account how much total rainfall that has occurred over previous periods
prior to the test. Rainfall of greater than 1mm, 2mm, and 5mm were analysed.

Dry Spell Factor (DSF), where x was trialled for 1mm, 2mm and 5mm of rainfall.

DSF = ln(t R + 1)
where:

DSF = Dry Spell Factor

tR = the number of days since the last rainfall of x mm or more (up to a maximum of 7
days, hence 0<tR<7)

Weighted Rain Function (WRF), where the period of previous rainfall in days (n) was trialled for 5, 7 and
20 days.

WRF =

n
i =1

where:

Ri
i
WRF = Weighted Rain Function
Ri = Rainfall in mm on the ith day prior to the test
i = number of days prior to the test ranging from n = 5, 7 or 20 days respectively

The following sections discuss the results of the analysis using the above rain functions, firstly with the GripTester
device and secondly, with the SCRIM device for the Hikurangi site in Northland. The Hikurangi site had an old but
225

Chapter 8
stable seal surface, and it was expected to be in an equilibrium state , other than exhibiting short-term and seasonal
variations. For this reason, it was used to trial differing rain functions against the measured values of skid resistance
to determine whether rainfall can be correlated against the measured coefficient of friction. The Hikurangi site has
also been shown to have the greatest variation of all the Northland sites. This variation is hypothesised to be related
to the amount of detritus/contaminants that builds up on the road surface, thereby helping to polish/abrade the
surface, which either reduces or rejuvenates the microtexture.

8.3.4 Effects of Rainfall on GripTester Device Measurements

The effect of the previous rainfall history was examined in relation to the variation in measured skid resistance as
measured by the GripTester device. Previous research (refer to Section 4.6.5) had shown that the greater the fine
and dry period prior to a skid resistance measurement, the lower the measured coefficient of friction. Figure 8-9 (a
to f) shows the analysis of the measured coefficient of friction by the GripTester at the Hikurangi site for the
increasing and decreasing directions, respectively, in relation to the number of days since the last rainfall, was
greater than 1mm, 2mm or 5mm. The analysis of the first year of results is shown separately from the second year,
because the two respective years have significantly different results (as seen in the figures). As discussed earlier, it
is unknown whether the differences from the first year of results (2003) are real recorded differences, in comparison
to 2004 / 2005 data, because the problems with the GripTester occurred, during the summer of 2003/2004. The
device was rebuilt during this period. However, if measured skid resistance does depend upon the length of the
period of no rain, then a decreasing (negative) trend of skid resistance with increase in the number of days since last
rainfall would be expected within both years data.
Hikurangi Days since last rainfall >1mm vs GN
(Decr dir)

0.75

0.75

0.70

0.70

0.65

y = -0.0043x + 0.5856
R2 = 0.0673

0.60

Grip Number (GN)

Grip Number (GN)

Hikurangi Days since last rainfall >1mm vs GN

(Incr dir)

0.55
0.50
0.45
y = 0.0097x + 0.4445
R2 = 0.1833

0.40
0.35

y = -0.0039x + 0.6185
R2 = 0.0729

0.65
0.60
0.55
0.50
0.45

y = 0.0066x + 0.4853
R2 = 0.154

0.40
0.35
0.30

0.30
0

10

12

Days Since Last Rainfall >1mm

14

10

12

Days Since Last Rainfall >1mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(a) Hikurangi DSLR > 1mm vs GN (Increasing direction)

(b) Hikurangi DSLR > 1mm vs GN (Decreasing direction)

226

14

Chapter 8
Hikurangi Days since last rainfall >2mm vs GN
(Decr dir)

0.75

0.75

0.70

0.70

0.65

Grip Number (GN)

Grip Number (GN)

Hikurangi Days since last rainfall >2mm vs GN

(Incr dir)

y = -0.0052x + 0.5923
R2 = 0.1864

0.60
0.55
0.50
0.45

y = 0.0006x + 0.4643
R2 = 0.0016

0.40
0.35

y = -0.0045x + 0.6233
R2 = 0.1778

0.65
0.60
0.55
0.50
y = 0.0009x + 0.4972
R2 = 0.0069

0.45
0.40
0.35
0.30

0.30
0

10

15

20

25

Days Since Last Rainfall >2mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(c) Hikurangi DSLR > 2mm vs GN (Increasing direction)

0.70

0.65

Grip Number (GN)

Grip Number (GN)

0.75

0.70

y = -0.0047x + 0.6063
R2 = 0.388

0.50
0.45

y = 0.0036x + 0.4494
R2 = 0.0651

0.40

20

25

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

Hikurangi Days since last rainfall >5mm vs GN

(Decr dir)

0.75

0.55

15

(d) Hikurangi DSLR > 2mm vs GN (Decreasing direction)

Hikurangi Days since last rainfall >5mm vs GN

(Incr dir)

0.60

10

Days Since Last Rainfall >2mm

y = -0.0037x + 0.6324
R2 = 0.306

0.65
0.60
0.55
0.50
0.45

y = -0.0011x + 0.5053
R2 = 0.0107

0.40
0.35

0.35

0.30

0.30
0

10

15

20

25

30

35

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(e) Hikurangi DSLR > 5mm vs GN (Increasing direction)

10

15

20

25

30

Days Since Last Rainfall >5mm

Days Since Last Rainfall >5mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(f) Hikurangi DSLR > 5mm vs GN (Decreasing direction)

Figure 8-9: Days Since Last Rainfall and GripNumber (Hikurangi site) for Various Depths of Rainfall
Whilst all of the figures show a decreasing trend for the 2004/2005 years, it is not a highly correlated trend, with the
coefficient of determination (R2) ranging from 0.07 to 0.39.

The outliers at both ends of the data have a

significantly greater effect on the coefficient of determination than data points in the middle. This effect affects the
trends in all graphs. The highest correlations were derived when the days since last rainfall greater than 5mm was
used with the coefficient of determination (R2) recorded as 0.39 and 0.31 (for the 2004/2005 data), respectively, for
the increasing and decreasing directions. The rate of decrease in measured coefficient of friction between all of the
graphs (i.e. for the DSLR >1, 2 and 5mm) is relatively consistent for the 2004/2005 year, which has a value of
approximately -0.004 GN per day. It is surprising that a possible effect exists up to 35 days for the greater than 5
mm depth of rain. If there were a lot more data points between 5 and 25 days since last rainfall, then as suggested
by Cenek et al. (2003b; 2004), it could become apparent that the relationship is not linear, but logarithmic, and that
little real change occurs after approximately 15 to 20 days. Cenek et al. (2003b; 2004) have suggested, using a
much smaller data set, that a 10% loss of skid resistance can occur within the first two weeks of no rainfall.
An analysis similar to that using the DSLR rain function was undertaken on the Hikurangi site and raw rainfall data
using the logarithmic Dry Spell Factor (DSF) rain function described in Section 8.3.3. The results of this analysis
are shown in Figure 8-10(a to f). These graphs show the measured coefficient of friction as recorded by the
GripTester, in both directions at the Hikurangi site, in relation to the natural logarithm of the time (in days) prior to a
227

35

Chapter 8
measured skid resistance test where the rainfall was greater than 1mm, 2mm and 5mm of rain, respectively. As
discussed above, it was expected that a logarithmic rainfall function would be a better fit to the data. However this
is not so for the data for the 2004/2005 year, as the range of coefficents of determination (R2) was significantly
lower (0.02 to 0.14) using the DSF in comparison to the DSLR rain function. Again, the 2003 year data showed
either random results with no gradient, or slightly positive results, which is contrary to what was expected and
shown by others. The Dry Spell Factor developed by Hill and Henry (1981) was limited to a maximum of only the
previous seven days rainfall history. As shown in the DSLR analysis, the rainfall history further back than 7 days
could continue to affect the skid resistance result. Subsequently, periods of 15 days and 20 days were analysed for
depths of rainfall greater than 5mm (the highest correlated result). However, the coefficient of determination (R2) in
the increasing direction increased only marginally, by approximately 0.03 to 0.17, but in the decreasing direction
reduced by approximately 0.045 to 0.048. Therefore, it can be concluded that for this data set, the DSF rain function
explains very little of the measured variation of surface friction.
Hikurangi Dry Spell Factor (DSF) >1mm vs GN
(Decr dir)

0.75

0.80

0.70

0.75

0.60
0.55
0.50
0.45
0.40

y = 0.023x + 0.4432
R2 = 0.1587

0.35
0.30
0.00

0.50

1.00

1.50

y = -0.008x + 0.6096
R2 = 0.0173

0.70

y = -0.0096x + 0.5741
R2 = 0.0168

0.65

Grip Number (GN)

Grip Number (GN)

Hikurangi Dry Spell Factor (DSF) >1mm vs GN

(Incr dir)

2.00

0.65
0.60
0.55
0.50
0.45

y = 0.0194x + 0.4806
R2 = 0.2045

0.40
0.35
0.30
0.00

2.50

0.50

Dry Spell Factor (DSF) >1mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(a) Hikurangi DSF > 1mm vs GN (Increasing direction)

2.00

2.50

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

Hikurangi Dry Spell Factor (DSF) >2mm vs GN

(Decr dir)
0.80

0.75

0.75

0.70

0.60
0.55
0.50
0.45
0.40

y = -0.0023x + 0.4691
R2 = 0.0014

0.35
0.50

1.00

1.50

y = -0.008x + 0.6096
R2 = 0.0173

0.70

y = -0.0096x + 0.5741
R2 = 0.0168

0.65

Grip Number (GN)

Grip Number (GN)

1.50

(b) Hikurangi DSF > 1mm vs GN (Decreasing direction)

Hikurangi Dry Spell Factor (DSF) >2mm vs GN

(Incr dir)

0.30
0.00

1.00

Dry Spell Factor (DSF) >1mm

2.00

Dry Spell Factor (DSF) >2mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(c) Hikurangi DSF > 2mm vs GN (Increasing direction)

0.65
0.60
0.55
0.50
0.45

y = 0.0138x + 0.4832
R2 = 0.0933

0.40
0.35

2.50

0.30
0.00

0.50

1.00

1.50

2.00

2.50

Dry Spell Factor (DSF) >2mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(d) Hikurangi DSF > 2mm vs GN (Decreasing direction)

228

Chapter 8
Hikurangi Dry Spell Factor (DSF) >5 mm vs GN
(Decr dir)

Hikurangi Dry Spell Factor (DSF) >5 mm vs GN

(Incr dir)
0.80

0.75

0.75

0.60
0.55
0.50
0.45
0.40

y = -0.0009x + 0.4677
R2 = 0.0002

0.35
0.30
0.00

0.50

1.00

1.50

y = -0.0237x + 0.6397
R2 = 0.0929

0.70

y = -0.0355x + 0.6226
R2 = 0.1405

0.65

Grip Number (GN)

Grip Number (GN)

0.70

2.00

Dry Spell Factor (DSF) >5mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(e) Hikurangi DSF > 5mm vs GN (Increasing direction)

0.65
0.60
0.55
0.50
0.45
0.40

y = 0.0003x + 0.4997
R2 = 4E-05

0.35

2.50

0.30
0.00

0.50

1.00

1.50

2.00

2.50

Dry Spell Factor (DSF) >5mm

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(f) Hikurangi DSF > 5mm vs GN (Decreasing direction)

Figure 8-10: Dry Spell Factor (DSF) and GripNumber (Hikurangi site) for various depths of rainfall
An analysis similar to that using the DSLR and DSF rain functions was undertaken on the Hikurangi site and raw
rainfall data using the linear Weighted Rain Function (WRF) described in Section 8.3.3. The results of this analysis
are shown in Figure 8-11(g to h) for the increasing and decreasing directions respectively and for rainfall periods of
5 days, 7 days, 10 days and 20 days. These graphs show the measured coefficient of friction as recorded by the
GripTester, in both directions at the Hikurangi site, in relation to the sum of each days rainfall divided by the
number of days prior to the measured skid resistance test. The rainfall period scenarios analysed were for 5 days, 7
days, 10 days and 20 days prior to a skid resistance test. For this rain function, it was expected that a positive trend
would occur, meaning that the greater amount of rainfall that has occurred prior to the test, the higher the measured
skid resistance value.
The Weighted Rain Function (WRF) in this research had significantly higher coefficents of determination (R2) than
the 2004/2005 skid resistance data results obtained with the GripTester and analysed using the DSLR or DSF rain
functions. The WRF coefficient of determination (R2) ranged from 0.40 to 0.61. However, the correlation is still
poor for the 2003 year, explaining only up to 15% of the variation. The scatter of data points and especially the
sensitivity to change in the outliers, would significantly affect the resultant coefficient of determination. This result
is contrary to the research undertaken by Hill and Henry (1981) with a locked wheel tester device. They determined
that the DSF better explained the variation in skid resistance results in comparison to the WRF developed by Dahir
et al. (1979).

229

Chapter 8
Hikurangi Weighted Rain Function (WRF05) vs GN
(Decr dir)

Hikurangi Weighted Rain Function (WRF05) vs GN

(Inc dir)
0.75

0.75

0.70

0.65

Grip Number (GN)

Grip Number (GN)

0.70
y = 0.0031x + 0.5403
R2 = 0.3971

0.60
0.55
0.50
0.45

y = 0.0005x + 0.4612
R2 = 0.026

0.40

0.65

y = 0.0029x + 0.5744
R2 = 0.408

0.60
0.55
0.50

y = 0.0008x + 0.4922
R2 = 0.1182

0.45
0.40
0.35

0.35

0.30

0.30
0

10

20

30

40

50

60

10

Weighted Rain Function (WRF) over 5days

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(a) Hikurangi WRF for 5 days vs GN (Increasing direction)

40

50

60

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

Hikurangi Weighted Rain Function (WRF07) vs GN

(Decr dir)
0.75

0.75

0.70

0.70
0.65

y = 0.0033x + 0.529
R2 = 0.4892

0.60

Grip Number (GN)

Grip Number (GN)

30

(b) Hikurangi WRF for 5 days vs GN (Decreasing direction)

Hikurangi Weighted Rain Function (WRF07) vs GN

(Inc dir)

0.55
0.50
0.45

y = 0.0002x + 0.4632
R2 = 0.0068

0.40

y = 0.0033x + 0.5618
R2 = 0.569

0.65
0.60
0.55
0.50

y = 0.0007x + 0.4894
R2 = 0.1488

0.45
0.40
0.35

0.35

0.30

0.30
0

10

20

30

40

50

60

70

80

10

20

30

40

50

60

70

80

Weighted Rain Function (WRF) over 7days

Weighted Rain Function (WRF) over 7days

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(c) Hikurangi WRF for 7 days vs GN (Increasing direction)

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(d) Hikurangi WRF for 7 days vs GN (Decreasing direction)

Hikurangi Weighted Rain Function (WRF10) vs GN
(Decr dir)

Hikurangi Weighted Rain Function (WRF10) vs GN

(Inc dir)
0.75

0.75
y = 0.0028x + 0.5201
R2 = 0.4292

0.65

0.70

Grip Number (GN)

0.70

Grip Number (GN)

20

Weighted Rain Function (WRF) over 5 days

0.60
y = 0.0001x + 0.463
R2 = 0.0056

0.55
0.50
0.45
0.40

y = 0.0031x + 0.5479
R2 = 0.6149

0.65
0.60
0.55
0.50

y = 0.0006x + 0.4865
R2 = 0.183

0.45
0.40
0.35

0.35

0.30

0.30
0

10

20

30

40

50

60

70

80

Weighted Rain Function (WRF) over 10days

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(e) Hikurangi WRF for 10 days vs GN (Increasing direction)

90

10

20

30

40

50

60

70

80

Weighted Rain Function (WRF) over 10days

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(f) Hikurangi WRF for10 days vs GN (Decreasing direction)

230

90

Chapter 8
Hikurangi Weighted Rain Function (WRF20) vs GN
(Decr dir)

Hikurangi Weighted Rain Function (WRF20) vs GN

(Inc dir)
0.75

0.75
y = 0.0015x + 0.5196
R2 = 0.3971

0.65
0.60
0.55

y = -0.0003x + 0.4816
R2 = 0.0384

0.50

y = 0.0015x + 0.5501
R2 = 0.5088

0.70

Grip Number (GN)

Grip Number (GN)

0.70

0.45
0.40

0.65
0.60

y = 0.0001x + 0.4925
R2 = 0.0173

0.55
0.50
0.45
0.40
0.35

0.35

0.30

0.30
0.0

20.0

40.0

60.0

80.0

100.0

120.0

0.0

20.0

Weighted Rain Function (WRF) over 20days

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(g) Hikurangi WRF for 20 days vs GN (Increasing direction)

40.0

60.0

80.0

100.0

120.0

Weighted Rain Function (WRF) over 20days

Ave GN -Yr 2003

Ave GN Yr 2004-05

Linear (Ave GN -Yr 2003)

Linear (Ave GN Yr 2004-05)

(h) Hikurangi WRF for20 days vs GN (Decreasing direction)

Figure 8-11: Weighted Rain Function (WRF) and GripNumber (Hikurangi site) for Various Rainfall Periods
The graphs in Figure 8-11 also show that the WRF better explains the variation in the measured skid resistance data
when either 7 days or 10 days rainfall period are recorded and analysed in comparison to the 5 day or 20 day period.
It can be concluded therefore, that for this data set, site and skid resistance device (GripTester), the WRF is the best
rain function to use as it explains up to 60% of the variation of measured surface friction. The DSLR and the DSF
rain functions explain little of the measured variation of skid resistance as measured by the GripTester device. The
next section discusses a similar analysis with the SCRIM device results.

8.3.5 Effects of Rainfall on SCRIM Device Measurements

The effect of the previous rainfall history was also examined in relation to the variation in measured skid resistance
as measured by the SCRIM device. The number of data points is considerably less than the GripTester results due
to the less frequent survey measurements that were undertaken during the data collection period. WDM Ltd
undertake the SCRIM skid resistance network surveys in the Northland region both for Transit New Zealand and for
Works Infrastructure, the network managers. Furthermore, the Hikurangi site was, at the outset of the research, a
seasonal control section, which required a minimum of three surveys per summer period. Unfortunately, the
decision was made mid-way through the data collection period to relocate the Transit New Zealand seasonal site to
another location. This meant that the three seasonal SCRIM skid resistance tests per summer were not continued at
this site throughout the full data collection period and only the two SCRIM network survey results could be analysed
(i.e for Transit NZ and Works Infrastructure PSMC 002 network surveys).
Figure 8-12 (a to f) shows the analysis of the measured coefficient of friction by the SCRIM device at the Hikurangi
site for the increasing and decreasing directions in relation to the DSLR and DSF rain functions for greater than
1mm, 2mm and 5mm of recorded rain depth. As with the GripTester analysis, if the hypothesis that the greater the
period of no rain, the lower the measured skid resistance is true, then a decreasing (negative) trend of skid resistance
with higher value of DSLR and DSF would be expected.

231

Chapter 8
Hikurangi Dry Spell Factor (DSF>1mm)
vs SFC (Both directions)

0.60

0.60

0.55

0.55

SCRIM (SFC)

SCRIM (SFC)

Hikurangi Days Since Last Rainfall (DSLR >1mm)

vs SFC (Both directions)

y = -0.0027x + 0.4613
R2 = 0.1199

0.50
0.45
0.40

y = -0.0039x + 0.4399
R2 = 0.4126

0.35

y = -0.0194x + 0.4697
R2 = 0.2495

0.50
0.45
0.40

y = -0.0219x + 0.4459
R2 = 0.5167

0.35
0.30

0.30
0

10

12

14

16

18

0.5

Days Since Last Rainfall (DSLR) > 1mm

Increasing

Decreasing

Linear (Increasing)

Increasing

Linear (Decreasing)

(a) Hikurangi DSLR > 1mm vs SFC (Both directions)

0.55

0.55

SCRIM (SFC)

SCRIM (SFC)

0.60

y = -0.0027x + 0.4615
R2 = 0.1227

0.45
0.40

10

12

Decreasing

Linear (Increasing)

14

16

y = -0.0191x + 0.4696
R2 = 0.2481

0.40
y = -0.0218x + 0.446
R2 = 0.5276

0.5

Increasing

Linear (Decreasing)

2.5

Decreasing

Linear (Increasing)

Linear (Decreasing)

Hikurangi Dry Spell Factor (DSF>5mm)

vs SFC (Both directions)
0.60

0.55

0.55

SCRIM (SFC)

SCRIM (SFC)

1.5

(d) Hikurangi DSF > 2mm vs SFC (Both directions)

0.60

y = -0.0019x + 0.4635
R2 = 0.1611

0.40
y = -0.0013x + 0.4351
R2 = 0.1134

0.35

Dry Spell Factor (DSF) > 2mm

Hikurangi Days Since Last Rainfall (DSLR >5mm)

vs SFC (Both directions)

0.45

Linear (Decreasing)

0.45

18

(c) Hikurangi DSLR > 2mm vs SFC (Both directions)

0.50

Linear (Increasing)

0.50

Days Since Last Rainfall (DSLR) > 2mm

Increasing

2.5

0.30

0.30
2

Decreasing

0.35

y = -0.004x + 0.4404
R2 = 0.4369
0

Hikurangi Dry Spell Factor (DSF>2mm)

vs SFC (Both directions)

0.60

0.35

1.5

(b) Hikurangi DSF > 1mm vs SFC (Both directions)

Hikurangi Days Since Last Rainfall (DSLR >2mm)

vs SFC (Both directions)

0.50

Dry Spell Factor (DSF) > 1mm

y = -0.0295x + 0.4875
R2 = 0.4945

0.50
0.45
0.40

y = -0.0121x + 0.4425
R2 = 0.1309

0.35
0.30

0.30
0

10

15

20

25

30

0.5

Days Since Last Rainfall (DSLR) > 5mm

Increasing

Decreasing

Linear (Increasing)

Linear (Decreasing)

(e) Hikurangi DSLR > 5mm vs SFC (Both directions)

1.5

2.5

Dry Spell Factor (DSF) > 5mm

Increasing

Decreasing

Linear (Increasing)

Linear (Decreasing)

(f) Hikurangi DSF > 5mm vs SFC (Both directions)

Figure 8-12: DSLR and DSF Analysis with the SCRIM device (Hikurangi site)
As can be seen from Figure 8-12 (a-f), both the DSLR and the DSF rain functions demonstrate the expected negative
trend of decreasing measured skid resistance with increasing dry spells prior to the skid resistance test. The linear
rate of decrease is approximately -0.003 SFC per day of no rain. However, as discussed for the GripTester device
232

Chapter 8
analysis, with more data, this would be expected to be closer to a negative logarithmic function, and to feature
greater decrease initially. The rate is reasonably consistent among the different depths of rainfall, ranging from 0.0013 to -0.040 SFC. The increasing direction gives significantly better correlation results than the decreasing
direction for the greater than 1mm and 2mm analysis results. Although, the opposite trend occurs for the rain depth
greater than 5mm. The coefficients of determination (R2) range from 0.11 (very low) to 0.43 (reasonable) for the
DSLR rain function and for the various directions of test and depths of rain, with the highest correlation obtained for
the greater than 2mm rain depth. These correlations are also higher than the results obtained using the GripTester
device for the 1mm and 2mm rain depth analysis except for the greater than 5mm analysis, which featured similar
coefficients of determination (R2=0.38).
As expected the results of the logarithmic DSF analysis with the SCRIM device (unlike the GripTester device) are
better than those from the DSLR rain function and demonstrate a stronger correlation between the SCRIM and the
DSF rain function in comparison to the GripTester and the DSF function. The coefficients of detemination (R2)
range from 0.24 (reasonably low) for the decreasing direction to 0.53 (reasonable) for the increasing direction when
the >2mm rain depth analysis is undertaken. This compares to a maximum (R2) value of 0.14 with the GripTester
with the >5mm rain depth. Further analysis was undertaken to determine whether increasing the previous rainfall
period from 7 days to 10 and 15 days for the DSF rain function would result in an increase in the coefficient of
determination (R2) of the resultant equation. However, the analysis did not increase the observed (R2) value and
therefore it is concluded that the 7 day period is the most appropriate period for the DSF rain function based on skid
resistance measurements with the SCRIM device.
An analysis similar to both the DSLR and DSF rain functions was undertaken on the Hikurangi site and raw rainfall
data using the linear Weighted Rain Function (WRF) described in Section 8.3.3 for the SCRIM device. The results
of this analysis are shown in Figure 8-13 (a to d) for rainfall period scenarios of 5 days, 7 days, 10 days and 20 day,s
respectively. These graphs show the measured coefficient of friction as recorded by the SCRIM device in both
directions at the Hikurangi site, in relation to the sum of each days rainfall divided by the number of days prior to
the measured skid resistance test. For this rain function, it was expected that a positive trend would occur, meaning
that the greater amount of rainfall that has occurred prior to the test, the higher the measured skid resistance will be.
Application of the Weighted Rain Function (WRF) to the SCRIM data gave (R2) values that ranged from 0.31 to
0.37 (refer to Figure 8-13a to d) in the increasing direction and from 0.24 to 0.34 in the decreasing direction for the
various rainfall period scenarios. These coefficients of determination (R2) are lower than those for GripTester
device, which showed values of approximately 0.60 for the 7 day and 10 day previous rainfall periods for the 2004 /
2005 data collection period. However, the results of the GripTester device with the WRF function for the 2003 year
data are still poor, explaining only 15% of the variation.
The results of the SCRIM device agree better with the results of Hill and Henry (1981) who used a locked wheel
tester device. They determined that the DSF better explained the variation in skid resistance results in comparison
to the WRF developed by Dahir et al. (1979), whereas the opposite trend was noted for the GripTester device.

233

Chapter 8
Hikurangi Weighted Rain Function (WRF07)
vs SFC (Both directions)

0.60

0.60

0.55

0.55

y = 0.0011x + 0.4351
R2 = 0.2405

0.50

SCRIM (SFC)

SCRIM (SFC)

Hikurangi Weighted Rain Function (WRF05)

vs SFC (Both directions)

0.45
y = 0.0011x + 0.4101
R2 = 0.3703

0.40

y = 0.001x + 0.43
R2 = 0.313

0.50
0.45

y = 0.0011x + 0.4127
R2 = 0.3396

0.40
0.35

0.35

0.30

0.30
0

10

20

30

40

50

60

10

Weighted Rain Function (WRF) over 5days

Increasing

Decreasing

Linear (Increasing)

Increasing

Linear (Decreasing)

(a) Hikurangi WRF05 days vs SFC (Both directions)

30

40

50

60

70

Decreasing

Linear (Increasing)

Linear (Decreasing)

(b) Hikurangi WRF07 days vs SFC (Both directions)

Hikurangi Weighted Rain Function (WRF20)
vs SFC (Both directions)

Hikurangi Weighted Rain Function (WRF10)

vs SFC (Both directions)
0.60

0.60

0.55

0.55
y = 0.001x + 0.43
R2 = 0.313

0.50

SCRIM (SFC)

SCRIM (SFC)

20

Weighted Rain Function (WRF) over 7days

0.45
y = 0.0009x + 0.4078
R2 = 0.365

0.40

y = 0.0006x + 0.4279
R2 = 0.3401

0.50
0.45
y = 0.0005x + 0.4063
R2 = 0.3739

0.40
0.35

0.35

0.30

0.30
0

10

20

30

40

50

60

70

20

Weighted Rain Function (WRF) over 10days

Increasing

Decreasing

Linear (Increasing)

Linear (Decreasing)

(c) Hikurangi WRF 10days vs SFC (Both directions)

40

60

80

100

120

Weighted Rain Function (WRF) over 20days

Increasing

Decreasing

Linear (Increasing)

Linear (Decreasing)

(d) Hikurangi WRF 20 days vs SFC (Both directions)

Figure 8-13: Weighted Rain Function (WRF) and SCRIM SFC (Hikurangi site) for Various Rainfall Periods
These differences may be due to having too few data points with the SCRIM device or a possible real difference in
terms of the device operating mode, and especially the differences in slip speed (Griptester = 15%, SCRIM =34%
and Locked Wheel Tester =100%) and the sensitivity of the device to external changes in measured surface friction.
As discussed in Section 4.3 and summarised in Section 7.7, the devices have significant variations in their modes of
operation.

8.3.6 Normalised GripTester Results for Northland Sites and Effects of Rainfall
The normalised GripTester device results obtained from field sites or timeslice locations which were predicted to be
acting in a stable equilibrium level of polishing and that were used as part of the seasonal analysis described in
Section 8.2.3 were also analysed against the three rain functions discussed above. The results are shown in Figure
8-14, Figure 8-15 and Figure 8-16 for the rain depth scenario, or previous rainfall period scenario, that indicated the
highest coefficient of determination (R2).
When the results are normalised from different field sites, even when carefully chosen to be in the equilibrium
polishing phase , the bi-variate coefficients of determination (R2) become significantly lower than the results
obtained just from the Hikurangi field site. The rain functions used (eg. DSLR>x mm, DSF> x mm and WRF for x
234

140

Chapter 8
days) explain, at best, only 5.5% of the observed variation. It is clear from this analysis that other localised site and
or environmental factors are also affecting the measured variation. These other factors are expected to include, but
not be limited to:

Temperature;

Localised contaminants / detritus;

Geological makeup of aggregate; and

Traffic loading conditions (i.e. the number of heavy commercial vehicles- HCV): two of the field sites on
State Highway 14 (Kara Rd and Snooks Tatton) had approximately 1/3rd of the HCVs as the State Highway
1 sites (Kaiwaka slag, Brynderwyn and Hikurangi).

Normalised NGN and Rainfall (Both Directions)

1.20

Normalised NGN

1.15
1.10
1.05
1.00
0.95
0.90

y = -0.0032x + 1.0144
2
R = 0.0511

0.85
0.80
0

10

12

14

16

18

20

22

Days Since Last Rainfall (DSLR) >2mm

KSS T/S 800m Inc
KARA T/S 40m Inc
BRYN T/S 100m Dec

BRYN T/S 490m Inc

S&T S2 Ave Inc
HIK Sect Ave Dec

HIK Sect Ave Inc

KSS T/S 100m Dec
Linear (All Data)

Figure 8-14: Days Since Last Rainfall (DSLR>5mm) and Normalised GripNumber (NGN) for Northland sites

235

Chapter 8

Normalised NGN and Rainfall (Both Directions)

1.20

Normalised NGN

1.15
1.10
1.05
1.00
0.95
0.90
y = -0.0213x + 1.035
R2 = 0.0549

0.85
0.80
0.0

0.2

0.4

0.6
0.8
1.0
1.2
1.4
1.6
Dry Spell Factor (DSF) >5mm

KSS T/S 800m Inc

KARA T/S 40m Inc
BRYN T/S 100m Dec

BRYN T/S 490m Inc

S&T S2 Ave Inc
HIK Sect Ave Dec

1.8

2.0

2.2

HIK Sect Ave Inc

KSS T/S 100m Dec
Linear (All Data)

Figure 8-15: Dry Spell Factor (DSF>5mm) and Normalised GripNumber (NGN) for Northland sites

Normalised NGN and Rainfall (Both Directions)

1.20

Normalised NGN

1.15

y = 0.0004x + 0.9937
R2 = 0.0375

1.10
1.05
1.00
0.95
0.90
0.85
0.80
0

20

40

60

80

100

120

140

Weighted Rain Function for 7 days (WRF07)

KSS T/S 800m Inc
KARA T/S 40m Inc
BRYN T/S 100m Dec

BRYN T/S 490m Inc

S&T S2 Ave Inc
HIK Sect Ave Dec

HIK Sect Ave Inc

KSS T/S 100m Dec
Linear (All Data)

Figure 8-16: Weighted Rain Function (WRF07) and Normalised GripNumber (NGN) for Northland sites
Although the coefficients of determination (R2) of the normalised GripTester device results are poor, Figure 8-14,
Figure 8-15 and Figure 8-16 show the expected direction of the slope of the equation line and confirm the Hikurangi
site trend observations. For example, for the DSLR > 2mm rain depth and the DSF > 5mm rain depth analysis, both
equations are negative, implying that the greater the period of no rainfall, the lower the measured skid resistance will
236

Chapter 8
be. Alternatively, the greater the amount of weighted rain depth over a given period prior to a skid resistance
measurement (as characterised by the WRF function), the higher the measured skid resistance will be. The slope of
these equation lines is likely to be highly dependent upon the traffic loading, the type of surface aggregate and its
susceptibility to polishing and/or abrasion and the likely contaminant agents (detritus / contaminants) that build up
in the aggregate macrotexture of the surface. These factors will vary from site to site.

8.4

Effects of Geometric Elements on Skid Resistance

8.4.1 Introduction
The various geometrical sections of a new surfacing will deteriorate differently due to the mechanisms of aggregate
polishing such as:

Sideways thrust on horizontal curves;

Braking areas on down grades;

Turning movements at intersections; and

Traffic loads that are placed upon the surface.

Therefore, to report the skid resistance average for the total section length, would mask changes at various sections
due to varying loading requirements.
The monitoring of skid resistance for over two years gave an opportunity to undertake timeslice analysis through
different sections of the road and to compare how the slices performed against each other. The skid resistance
monitoring methodology consisted of testing short sections before and after new surface seals, which allowed for
30m comparisons to be made between the wearing of:

New straight surfaces and adjacent older straight sections;

New horizontal curve sections in comparison to new straight sections; and

New level graded section in comparison to -6% vertical grade.

8.4.2 Timeslice Analysis for Geometric Elements on the Kara Road Field Site
Figure 5-10 (in Section 5.7.6) shows an aerial photograph of the Kara Road skid resistance monitoring site on State
Highway 14 that was used to demonstrate the effect of varying rates of polishing due to changes in surfaces or
geometric elements. The 30m timeslice locations are shown in Figure 5-10. These were selected to enable the
comparison of the specific geometric elements listed above, as measured by the GripTester device. Figure 8-17
compares timeslice 120m with timeslice 40m and 600m with 650m, respectively, in the increasing direction. This
comparison is expressed as the ratio of the new sealed section performance to the older adjacent seal sections on
straight sections of road, over a period of approximately 820 days since the sections were sealed.
The data, clearly demonstrate that a significant negative logarithmic relationship exists between skid resistance and
seal age with a coefficient of determination (R2) of 0.90 and 0.82 for the two timeslice comparisons using 24 data
points. The trend lines demonstrate that the new racked-in Grade 3/5 chip seal that was placed on the 28/01/2003
with an Otaika Quarry chip (PSV 52):

Initially increased the measured skid resistance by 15 to 20% above the adjacent older seal sections;

Performed for 13 to 17 months before the new seal deteriorated to the point where it was the same as the
seal either side of the new seal;

Continued to deteriorate to approximately 5 to 10% lower than the older adjacent seal sections (although
this appears now to have levelled off).
237

Chapter 8

Ratio of GN (New Seal to Old Seal)

Timeslice - New Seal vs Old Seal (straight section)

1.4
y = -0.08Ln(x) + 1.51
2
R = 0.82

1.2
1.0
0.8

y = -0.10Ln(x) + 1.61
2
R = 0.90

0.6
120m/40m

0.4

600m/650m
Log. (600m/650m)

0.2

Log. (120m/40m)

0.0
0

100

200

300

400

500

600

700

800

900

No. of Days since new seal construction

Figure 8-17: Performance of the New Seal in Comparison to the Old Adjacent Seal (Kara Road)
Figure 8-18 compares timeslice 220m with timeslice 120m and 520m with 450m, respectively, in the increasing
direction. Thus, it is a comparison on the basis of the new sealed section performance on geometrically straight
sections to that of a horizontal curve over a period since the sections were resealed. The horizontal curves were
150m and 160m radii curves at locations 120m and 450m respectively. The second curve at running distance 450m
occurs on a downhill gradient and therefore operating speeds were likely to be higher than the curve at section 120m
which was also located at an intersection to a minor rural access road.
The data demonstrate that a negative logarithmic relationship exists with a coefficient of determination (R2) of 0.70
and 0.54 for the two timeslice comparisons using 24 data points. The trend lines demonstrate that the new curved
sections of the racked in Grade 3/5 chip seal that was placed on the 28/01/2003 with an Otaika Quarry chip (PSV
52):

Initially began at the same level of skid resistance as adjacent straight sections;

Steadily deteriorated faster than their adjacent straight sections and then levelled off some 12 to 16 months
after the new seal surface; and that

The level that they deteriorated to was approximately 5% to 15% lower than the adjacent straight sections.

238

Chapter 8

Timeslice - New Seal (straight vs curved section)

Ratio of GN (New Seal to Old Seal)

1.4
1.2

y = -0.0213Ln(x) + 1.0851
R2 = 0.5450

1.0
0.8
0.6

y = -0.0728Ln(x) + 1.3502
R2 = 0.7041

220m/120m
520m/450m

0.4

Log. (520m/450m)

0.2

Log. (220m/120m)

0.0
0

100

200

300

400

500

600

700

800

900

No. of Days since new seal construction

Figure 8-18: Performance of a Straight with Curved Geometrics (Kara Road) New Seal
Figure 8-19 compares timeslice 120m on a straight and reasonably level gradient with a timeslice at 600m that is on
a straight section, but -6% downgrade, in the increasing direction. It was expected that, due to additional braking
forces on the down gradient, the surface may polish more quickly than on the level gradient. However, as can be
seen from Figure 8-19, the two sections performed reasonably consistently, and if anything, showed that the level
section deteriorated marginally faster than the -6% down gradient section. It is conjectured that the down gradient
was not steep enough or curvilinear enough for heavy braking forces to cause higher degrees of microtexture
polishing on this section, in comparison to a level gradient.

Timeslice - New Seal (Level vs -6% gradient)

Ratio of GN (New Seal to Old Seal)

1.4
1.2
1.0
0.8

y = -0.0201Ln(x) + 1.0992
R2 = 0.3210

0.6
120m/600m

0.4

Log. (120m/600m T)

0.2
0.0
0

100

200

300

400

500

600

700

800

900

No. of Days since new seal construction

Figure 8-19: Performance of a Level section in comparison with -6% Gradient (Kara Road) - New Seal
239

8.4.3 Summary of Timeslice Analysis on Geometric Elements

Chapter 8

The timeslice analyses of carefully chosen sections (rather than taking the average seal extent result for
geometrically different sections), give a better understanding of how the seal sections perform over time. The
graphs show a 30m average for the timeslice positions and also a ratio of two chosen timeslices. The latter gives a
comparison of how the sections perform relative to each other.
Timeslice analysis of various geometric sections of the Kara Road field site was undertaken with curve radii that
range from 150m to 160m. The surface was treated in March 2003 with a racked-in Grade 3/5 seal utilising Otaika
aggregate (with a PSV of 52) and a traffic loading of approximately 300 HCVs per day. This analysis has shown
that for the Kara Road Road site in Northland:

Significant negative logarithmic deterioration profiles exist when comparing adjacent 30m sections for
various geometrical features (R2 range from 0.54 to 0.90 for 24 data points);

Initially measured skid resistance (GN) increased by up to 15% to 20% above adjacent older seals;

Skid resistance for straight sections lasted for approximately 13 to 17 months before decreasing below
adjacent seal sections which appear to have then levelled off some 5% to 10% below the adjacent older seal
sections;

Newly sealed curved sections with radii in the range of 150m to 160m have deteriorated in terms of
measured skid resistance faster than new straight sections and have taken 12 to 16 months to level off 5%
to 15% below the straight sections;

A comparison of a level straight section of road with a -6% downgrade did not demonstrate any significant
difference between the measured skid resistance of the two sections.

8.5

Washing Trials

8.5.1 Introduction
Due to the increasing evidence of the effects of contamination on measured skid resistance results, a washing trial
was undertaken at the Hikurangi Control Site in Northland in early July 2003. The purpose of this trial was to
evaluate whether the cleaning of the road surface would result in an improvement in skid resistance, and to identify
if any of the treatments applied were more effective than the others. A 300m control section was established at one
end of the test section identified as Section 1. Several washing test sections were established, each 150 long. These
were treated using the following methods:

Section 2.

Gentle washing (two passes with a water cart)

Section 3.

Washing and Brooming (two passes with a water cart followed by a nylon rotary broom)

Section 4.

Frimokar NZ Ltd high water blasting process (operating at relatively low pressure of

approximately 2000psi).
Two sets of benchmarking tests were undertaken, one prior to the installation of traffic control and one after it had
been installed. The treated sections were not open to traffic between treatment and testing. Testing of the site was
undertaken using both the GripTester and British Pendulum Tester, with texture being determined by the Sand
Circle test method. The results of the trial are discussed in the following section.

240

Chapter 8

8.5.2 GripTester Results

Figure 8-20(a and b) shows the GripTester results of the benchmarking runs in the increasing and decreasing
directions, respectively. Reasonable repeatability and trending is seen along the length of the sites with each
successive run gradually increasing. The test sections were all located along the left hand wheel path of the
decreasing direction, as shown in Figure 8-20c. No treatment was undertaken in the increasing direction, and this
lane was left open to carry traffic in a modified arrangement in both directions.
The effects of roughness on the 10m averages of skid resistance produced by the GripTester are demonstrated in
Figure 8-20a by the spike seen at 100m, while the effect of bleeding can also be seen by the trough located at 750m.
These sections were deleted from the analysis results so as not to skew the averaging process.
Hikurangi
SeasonalSite,
Site, Washing
Washing Trial,
11/7/03,
Hikurangi
Seasonal
Trial,
11/7/03
Increasing
- Benchmark
Testing
IncreasingDirection
Direction
- Final Testing

Hikurangi Seasonal Site, Washing Trial, 11/7/03,

Increasing Direction - Benchmark Testing
1.0

0.9

0.9

0.8

0.8

0.7

0.7

GripNumber (GN)

GripNumber (GN)

1.0

0.6
0.5
0.4
0.3
0.2

GN run1
GN run4
AVE (-1) Inc

0.1

GN run2
GN run5

GN run3
SITE EXTENTS

200

0.5
0.4
0.3
0.2

400

600

800

1000

GN run2
GN run5
GN run6

GN run3
SITE EXTENTS

0.0

1200

Running Distance

200

400

600

800

1000

1200

Running Distance

(a) Hikurangi Pre-Washing (Increasing direction)

(b) Hikurangi Post-Washing trial -(Increasing direction)

Hikurangi Seasonal Site, Washing Trial, 11/7/03,
Decreasing direction - Final Testing

Hikurangi Seasonal Site, Washing Trial, 11/7/03,

Decreasing direction - Benchmark Testing

1.0

1.0

0.9

0.9
0.8

0.8

0.7

0.7

GripNumber (GN)

GripNumber (GN)

GN run1
GN run4
AVE (-1) Inc

0.1

0.0
0

0.6

0.6
0.5

Section 2

0.4

Section 1

0.3
GN run1
GN run4
AVE (-1) Dec

0.2
0.1

Section 4
Section 3

GN run2
SITE EXTENTS

GN run3
GN run5

0.6
0.5
Section 2

0.4

Section 1

0.3
0.2

GN run1
GN run4
AVE (-1) Dec

0.1

0.0
0

200

400

600

800

Running Distance

(c) Hikurangi Pre-Washing (Decreasing direction)

1000

1200

Section 4
Section 3

GN run2
SITE EXTENTS
GN run6

GN run3
GN run5

0.0
0

200

400

600

800

Running Distance

1000

1200

(d) Hikurangi Post-Washing trial (Decreasing direction)

Figure 8-20: Results of Washing Trials on Hikurangi Site, Northland (Pre and Post Washing)
The results shown in Figure 8-20(b and d) represent the testing after the treatments had been applied and the traffic
closures had been removed. Six test runs were undertaken in each direction. Analysis of the results shown in Figure
8-20d for the treated wheel path showed that no significant increase in friction levels measured by the GripTester
was apparent due to the cleaning. This is more clearly shown in Figure 8-21 where the average of the various trial
sections are compared against each other, both before and after the washing treatments. It can be seen that all of the
241

Chapter 8
sections (including the control section where no treatment was applied) reduced slightly in the measured skid
resistance as measured by the GripTester. This difference is negligible and less than the repeatibility of the
measurement device itself. It can therefore be concluded that none of the washing treatments had any effect on the
measured skid resistance as measured by the GripTester.
Average GripNumber for LHWP Before and After Treatment
0.8
0.7

Benchmarking Before Washing

After Washing Treatment

Ave GripNumber (GN)

0.6
0.5
0.4
0.3
0.2
0.1
0.0
Conditioning

Control

Wetted

Broomed

Frimo

Washing Section

Figure 8-21: Before and After Results of the Washing Trials (decreasing direction)
The results for the increasing direction (that had no washing treatments applied) and that remained open to traffic
were unexpected. Figure 8-20b shows the results from the six runs that were undertaken. These show a much larger
spread (standard deviation) between the consecutive GripTester runs and poor trending along the length of the test
site as well as a drop in the average coefficient of friction as measured by the GripTester. Whilst the increasing
trend of results is still present, it appears as though the road is not in a uniform skid resistance condition. These
results have been interpreted as showing the effects of contamination of the road surface. In order to have room to
treat and test the left-hand wheel path in the decreasing direction, traffic was moved closer to the shoulder. In
places, it was observed that vehicles travelled onto the unsealed shoulder, resulting in mud being splashed and
tracked onto the pavement surface. Normal cleaning processes, such as effects of traffic and water interaction, did
not occur, as all traffic was restricted to low speeds. The testing of this wheel path illustrates the type of results a
contaminated pavement with detritus would be expected to exhibit, that is a reduction in the measured skid
resistance.

8.5.3 British Pendulum Results

All of the treated sections were also tested using a British Pendulum Tester, before and after treatment, including the
untreated control section. Three test points were used in all sections with the exception of the Frimokar treated
section which was tested at five locations, with two repeat tests undertaken after treatment. These results were
corrected for temperature as per normal device practice.
The results indicated no significant change in the measured British Pendulum Number (BPN) for the Control,
Washed or Broomed sections. The only change identified occurred in the Frimokar section where an increase of 5
BPN (approximately 0.05 GN) was recorded on the first repeat test after the cleaning occurred. However, the re-test
revealed a smaller improvement of only 2 BPN.
It is unclear whether these results suggest that the Frimokar offers an improved washing mechanism in comparison
to the other treatments or whether the Frimokar treatment may be rejuvenating microtexture. As the cleaning was
242

Chapter 8
only undertaken at low pressure, it is considered unlikely to be the latter, although this possibility cannot be totally
discounted. For whatever reason, it seems that the positive effect is temporary, as most of the positive effect was
lost very quickly. A visual inspection of the sections after treatment identified that material was removed from the
Frimokar section much more effectively than the other treatments, as would have been expected, particularly in the
case of accumulated material at the bottom of the macrotexture. The macrotexture results (which were only
available for the Frimokar section due to rainfall preventing testing on the other sections) revealed an increase of
0.4mm average MTD, obtained from the sand circle tests from this treatment.

8.5.4 Washing Trial Conclusions

The washing trial revealed some interesting, although inconclusive results, regarding the effectiveness of a range of
washing treatments that can be applied for improving skid resistance levels. None of the washing treatments
produced any improvement as measured by the GripTester device. The British Pendulum tester also showed no
improvement between the benchmarked levels and the post-washed treated sections on all sections except the
Frimokar section. The Frimokar section results with the British Pendulum Tester indicated an improvement in skid
resistance after the treatment was applied (5 BPN) although this improvement was very temporary as a further test a
short time later, resulted in a reduced improvement of 2 BPN.
It is suggested that the inconclusiveness of the results can partly be attributed to the clean state of the pavement
prior to the trial being undertaken. Although, observation and inspection of the water blasted material that was
collected by vacuum into the tank of the Frimokar vehicle, calls into question the relativity of the word clean .
High rainfall was experienced (accumulative rainfall of approximately 100mm over the previous two weeks) prior to
the washing trial, with 6.5 mm falling over the course of the day that the trial was conducted. This is borne out by
the comparison of the skid resistance levels for the normal monitoring of the site as shown on Figure 6-18 and the
test results either side of those for the 11th July 2003. The skid resistance measurements on this washing trial date
resulted in one of the highest levels of measured skid resistance. It is concluded that the relatively clean state of the
pavement initially, negated some of the expected increase in skid resistance due to the varying levels of pavement
washing .
Other spot washing trials were undertaken with the Dynamic Friction Tester, on surfaces that were initially tested
and then retested after waterblasting the surface with relatively low pressure (approximately 2000psi). Sites trialed
included, the Tamaki Campus, Ports of Auckland and Hikurangi sites and after significant dry periods. The results
did not show marked improvements in the skid resistance results following washing of the surface.

It is

hypothesised that washing of the surface by itself does not improve the skid resistance. However, the combined
effect of the pavement surface being wet with the kneading and abrasion effect of traffic loads on the wet surface
does help to re-juvenate the surface.

The controlled laboratory based experiments discussed in Chapter 10

investigates this effect further.

243

Chapter 9

9 DETRITUS RESULTS AND DISCUSSION

9.1

Introduction

As discussed in Section 3.5.6, it has been thought that the accumulation of finer particle detritus material contributes
towards the polishing of the aggregate, usually coinciding with lower rainfall summer months. Conversely, the
coarse grit contributes to abrasion of the aggregate and, when the finer material has been washed away in wetter
months, helps to rejuvenate skid resistance (Jayawickrama & Thomas, 1998).
This chapter discusses the results of the environmental analysis on the surface detritus samples that were collected
from one metre squared test positions at the skid testing field test sites. The methodology used for the detritus
sample collection and the detritus laboratory analysis methods is discussed in Sections 5.8.2 and 5.8.3 respectively.
The samples were analysed primarily in the Environmental Engineering Laboratory (Department of Civil and
Environmmental Engineering, The University of Auckland) for suspended solids, heavy metals (i.e copper, lead,
zinc and cadmium), total petroleum hydrocarbons (TPH) and total organic carbon. Particle size distribution (PSD)
tests were also undertaken by Malvern Instruments Ltd.

9.2

Sampling Site Details

Samples were obtained primarily from four of the skid resistance test sites in 2005. The test site characteristics are
discussed in Section 5.7. Traffic conditions ranged from primarily light private car usage to moderately heavy state
highway traffic volume conditions.

The roadside vegetation alongside the field test sites varied from urban

conditions to rural rolling terrain with predominantly farmland or light agricultural activities. Three of the field sites
were located in rural areas in the Northland region and the fourth was at the University of Auckland Tamaki
Campus site. The field site details, characteristics and the sample collection locations are shown in Table 9-1.
Sampling
location

Site details

Surrounding
Environment

Traffic
speed,
km/hr

Traffic
Volume
ADT

Heavy
commercial
vehicles

Sample
collection
date

T1, T2, T3, T4

Tamaki campus,
University of
Auckland

Urban, flat
terrain

30

08 Mar 2005

K1, K2, K3, K4

Kara Road, State

Highway 14

Rural
rolling terrain

100

5510

5.5%

14 Mar 2005

S1, S2, S3, S4

Snook/Tatton Titoki
State Highway 14

Rural
flat terrain

100

5510

5.5%

14 Mar 2005

H1, H2, H3

Hikurangi,
State Highway 1

Rural
flat terrain

100

9700

9.7%

18 Apr 2005

H1 , H2 , H3 *

Same site as above

As above

As Above

9700

9.7%

12 May 2005

*Samples H1 , H2 and H3 were collected at the same location at different days as H1, H2 and H3, respectively.
Table 9-1: Detritus Sampling Details (2005)
In addition, comparative analysis was undertaken on the results of a preliminary study completed in 2004 when the
sampling methods were being trialed. The preliminary study included an analysis of the samples for suspended
solids, PSD and heavy metals. The details of the sites are shown in Table 9-2.

244

Chapter 9
Location Details

Surrounding
Environment

Traffic
Speed,
km/h

Traffic
Volume

Sample Collection
Date

TeAtatu
Motorway off Ramp

Urban

100

Heavy

12 Jan 2004

Kaiwaka Slag Site

Rural

100

Heavy

06 Jul 2004

Ports of Auckland

Urban

50

Heavy

21 Sep 2004

Tamaki Campus,
University of Auckland

Urban
flat terrain

30

Light

22 Sep 2004

Table 9-2: Preliminary Study Detritus Sampling Details (2004)

In order to evaluate the effects of the previous rainfall history on the sample characteristics, particularly for the
comparisons of the samples collected at the Tamaki and Hikurangi field sites that were collected on different days,
the preceding 15-day rainfall depth data was determined. The rainfall data were recorded from either manual
rainfall gauges installed at the site (i.e for the Northland rural sites) and/or obtained from the closest NIWA
automatic rainfall gauge station data. The summary rainfall data are given in Table 9-3.
Days

Tamaki
(2004)

Tamaki
(2005)

Kara
Road

Snook/
Tatton

Hikurangi
(Apr 05)

Hikurangi
(May 05)

0.50

0.49

-1

1.51

0.56

0.74

-2

1.51

0.80

2.23

2.77

-3

2.52

2.12

-4

3.52

0.49

-5

0.50

3.59

-6

7.04

1.70

0.16

14.03

-7

0.50

2.12

0.49

5.66

-8

5.03

3.27

7.63

-9

9.55

1.97

-10

13.37

0.25

1.48

-11

0.21

1.72

-12

5.45

0.64

-13

20.71

0.74

-14

2.44

0.49

-15

2.61

Table 9-3: Preceding 15-day Rainfall Depth (mm)

245

9.3

Chapter 9

Suspended Solids and Particle Size Distribution

The event mean sediment loads, calculated using the observed sediment concentration and the one litre volume of
fluid collected at each site, are presented in Figure 9-1. At the Hikurangi and Tamaki sites duplicate samples were
collected on different days, which is an indication of the variation in sediment load that could be expected over time.
A maximum of four samples were collected at each site as shown in Table 9-1. During any sampling event the
observed variation in the sediment load, measured as suspended solids, was less than 25 %. However, it is apparent,
particularly at the Hikurangi site, that local conditions, vehicle commodities and environmental changes (e.g. rainfall
and wind speed) can significantly affect the amount of road sediment accumulating on road surfaces. On the other
hand, although the number of preceding dry days was expected to be one of the significant factors in determining the
sediment load, the limited data in Table 9-3 meant that a correlation was unable to be determined.
Hikurangi in May, 2005
Hikurangi in Apr, 2005
Snook/T atton in Mar, 2005
Kara Road in Mar, 2005
T amaki in Mar. 2005
T amaki in Sep. 2004
Ports of Auckland in Sep. 2004
Kaiwaka Slag in Jul. 2004
T eAtatu in Jan. 2004
0

4
6
8
10
Suspe nde d Solid Load, mg/m2

12

14

16

Figure 9-1: Mean Sediment Load at Each Site (mg/m2)

Similar to the sediment load, the PSD values for the sediments obtained during individual sampling events were
surprisingly consistent, as shown in Figure 9-2.

Accordingly, the mean values of the particle characteristics

obtained for the replicate samples were adopted for the analysis (shown in Table 9-4). It is worth mentioning that
the means of the particle characteristics in all sampling events were also similar. These observations suggest that
the PSD did not change significantly with time of sampling and/or location. This fact is likely to be influenced by
the sampling methodology and protocol. The methodology was based upon obtaining the majority of sediments at
source, and that originated from the weathering of the road surface, which gave similar particle sizes from one
location to another. The weathering of the aggregates were also thought to be related to the geological makeup of
the aggregates themselves. Yet the results have shown surprisingly little variation between samples.

246

Chapter 9
100

Percent Finer, %

90
80

T eAtatu in Jan. 2004

70

Kaiwaka Slag in Jul. 2004

60

Ports of Auckland in Sep. 2004

T amaki in Sep. 2004

50

T amaki in Mar. 2005

40

Kara Road in Mar, 2005

30

Snook/T atton in Mar, 2005

20

Hikurangi in Apr, 2005

10

Hikurangi in May, 2005

0
0.001

0.01

0.1
Particle Diameter, mm

10

Figure 9-2: Mean Particle Size Distributions of Solid Samples

Mean Particle Size, m

Date

Sample Location

Mean Specific
Surface Area,
m2/g

d(0.1)

d(0.5)

d(0.9)

Jan. 04

TeAtatu Site

0.371

7.968

22.976

67.341

Jul. 04

Kaiwaka Slag Site

0.544

5.026

16.529

54.529

Sep. 04

Ports of Auckland

0.617

4.095

14.565

250.589

Sep. 04

Tamaki Site

0.384

7.153

25.484

120.674

Mar. 2005

Tamaki Site

0.338

8.041

32.034

105.415

Mar. 2005

Kara Road Site

0.410

6.249

27.634

125.544

Mar. 2005

Snook/Tatton Site

0.443

5.578

24.752

114.131

Apr. 2005

Hikurangi Site

0.488

5.388

20.794

91.185

May. 2005

Hikurangi Site

3.017

15.541

79.678

Overall Mean

0.449

5.835

22.257

112.121

Standard Deviation

0.164

1.606

5.587

54.046

Table 9-4: Mean Particle Size Distribution Data of Each Site

The mean PSD of sediments for the replicate samples collected at each site are presented in Figure 9-2, which shows
that the sediment sizes at all locations were similarly distributed. The sediment particles ranged from 0.001mm to
approximately 1 mm a range that is somewhat smaller to the 0.001mm to 10 mm range that is reported for
sediments from previous stormwater based PSD research as shown in Figure 9-3.

247

Chapter 9

Figure 9-3: Distributions of road runoff sediment sizes from literature - Modified from Walker et al. (1999)
and Ng et al. (2003)

9.4

Heavy Metals

The results of the analysis of heavy metal concentrations for the one litre samples collected at various locations are
presented in Figure 9-4. The metal concentrations measured from the samples collected from the field test sites are
relatively low in comparison to the literature. This is presumably due to low levels of contamination at the selected
sampling locations, associated with low/moderate levels of traffic volumes and the rural surroundings. Previous
research has predominantly been in more dense urban contexts where there are large areas of heavy braking and
traffic volumes are significantly higher than the sites used for skid resistance testing in this research.
The concentrations of cadmium were particularly low (below 5g /m2). The results of earlier samples collected in
2004 in other urban areas (given in Table 9-2), which included a high traffic volume motorway off-ramp on State
Highway 16, had similar ranges of copper and zinc concentrations. As would be expected, the Ports of Auckland
site , which has very heavy industrial ship container straddlers and fork lift traffic which load containers on to the
immediately adjacent railway grid tracks, had significantly higher concentrations of heavy metals. Concentrations at
the Ports of Auckland site were significantly higher than the average values for the other sites, 17 times higher for
copper and 8 times higher for lead concentrations.
5

Cu,mg/L

Zn,mg/L

Pb,mg/L

Cd,m/L

0
T1

T2

T3

T4

K1

K2

K3

K4

S1

S2

S3

S4

H1

H2

Conc. of Cd, m/L.

Conc. of Cu, Zn & Pb, mg/L

H3

Site

Figure 9-4: Metal Concentration of Each Sample

248

Chapter 9
Attempts to correlate the metals to sediment concentrations showed moderate correlations (R2 of 0.49 to 0.76)
between copper, zinc, lead and the associated suspended solids, suggesting that the metals may have been sedimentbound. Cadmium, on the other hand, showed no correlation (R2 below 0.04) with suspended solids or with the other
metals. This trend is contrary to that reported by Yim and Nau (1987) who observed strong correlations between
cadmium and copper as well as between cadmium and zinc. The weak correlation from this research is likely to be
due to the negligible amounts of cadmium existing in all of the collected samples.

9.5

Petroleum Hydrocarbons and Organic Carbon

As discussed in section 5.8.3, the detection limit for the Hewlett-Packard HP 6890 series GC system was
approximately 0.001% (by volume) or 10 mg/L of the 1:1 petrol-diesel mixture in n-pentane. The Total Petroleum
Hydrocarbons (TPH) amount in each of the samples was close to the method detection limit (i.e., below 10 mg/L),
indicating that little or no TPH existed in all samples.
To separate the contribution of dissolved and particulate organic matter, the Total Organic Carbon (TOC) values
were quantified separately for the sediment-associated (mg-TOC/g-sediments) and dissolved (mg-TOC/L-water)
fractions of the one litre samples collected at the different sites. The sediment-associated total carbon and inorganic
carbon values are presented in Figure 9-5, which shows that the inorganic carbon concentration was generally
negligible and the total carbon concentrations ranged from 110 to 240 mg/g. The small amounts of inorganic carbon
on the Hikurangi site (H1-H3 sites) are believed to be due to calcium carbonate (CaCO3) formed by the reaction
between atmospheric carbon dioxide (CO2) and lime. This is presumably spilled from trucks carrying lime on State
Highway 1, as a lime quarry is located reasonably close and north of the Hikurangi site.
300
Insoluble total carbon

250

Insoluble inorganic carbon

mg/g

200
150
100
50
0
T1

T2

T3

T4

K1 K2

K3

K4

S1

S2

S3

S4

H1 H2

H3 H1' H2' H3'

Site

Figure 9-5: Carbon Load in Sediments (mg/g)

The dissolved TOC values for the sampled liquid, along with the sediment associated TOC values converted to
mg/L-water, are presented in Figure 9-6. It can be clearly seen from Figure 9-5 and Figure 9-6 that the organic
carbon is primarily associated with particulate matter. The two sets of samples collected from the Hikurangi site
show that when TOC is expressed on the basis of the volume of sample collected, it can give a very different picture
from that when the TOC is expressed on the basis of sediment mass in samples. In addition, it is interesting to note
that there is a consistent trend for sediment load, insoluble inorganic carbon and sediment-associated TOC
concentration at the Hikurangi site. Although it is difficult to predict the variability of these measured parameters
based on limited data, it is believed that the variations are due to external sources. Furthermore, as the TPH levels
measured were negligible, the organic carbon has more likely resulted from roadside vegetation rather than from
other sources.

249

Chapter 9
2500
Dissolved TOC Conc.

Conc., mg/L

2000

Sediment-associated TOC Conc.

1500
1000
500
0
T1

T2

T3

T4

K1

K2

K3 K4

S1 S2
Site

S3

S4

H1

H2

H3 H1' H2' H3'

Figure 9-6: TOC Concentration of each sample (mg/L)

9.6

Summary of Detritus Analysis

In this analysis, 29 samples of detritus were collected from seven field sites between January 2004 and May 2005.
The samples were analysed for sediment load and particle distribution, heavy metals, TPH and TOC. The sample
results revealed that the sediment loads ranged from 1.5 to 14.4 mg/m2 in response to local conditions (e.g., traffic
volume, vehicle composition, and site setting) and environmental factors (e.g. rainfall and wind speed). Samples
collected during the same sampling event showed little variability, whereas samples collected at the same site on
different days varied significantly. Although the number of preceding dry days was expected to be one of the
primary factors influencing the amount of road sediment accumulating on the road surfaces, due to a limited number
of data points at the same sites, a correlation could not feasibly be considered.
The sediment size distribution remained relatively similar, with particles ranging from 0.001 to 1 mm, at different
locations and time of sampling. The mean particle size d(0.5) was 21.5 5.8 m, while the average d(0.1) was 5.7
1.6 m. These results give somewhat finer distributions than those reported for stormwater sediments researched
elsewhere. All metals collected from all sites, excluding one exception (the Ports of Auckland site) had low metal
concentrations (Cd 1-4 g/L, Cu 0.4-1 mg/L, Pb 0.3-2 mg/L, and Zn 1-3 mg/L). The samples collected from the
Ports of Auckland site, reported 17-fold higher levels of zinc and 8-fold higher levels of lead were found compared
to the average results observed for the other sites. This is understandable, given the very heavy industrial traffic
nature of this site. The majority of the copper, zinc and lead measured appear to be sediment-bound while cadmium
was primarily dissolved. The TPH concentrations were close to the detection limit of the analysis (i.e. 10 mg/L).
Although, little or no TPH could be detected in the samples, TOC values suggest the presence of non-TPH organic
matter, primarily in the form of particulate matter.
Previous research had shown that rainfall volume appears to be the major factor in the removal of total road dust
detritus, and rainfall frequency mainly affects the accumulation of particle-bound contaminants (O'
Riley et al.,
2002). Unfortunately, due to the limitations in sampling the road surface based detritus, insufficient samples were
obtained at the same sites to enable a proper assessment of any relationship between road-based sediment and
rainfall volume and frequency.
In summary, analysis of the detritus results has demonstrated that the variations of sediment and pollutants were
found to be site specific, and no unique differences between predominantly urban and rural settings could be
detected, other than for very heavy industrial sites. The analysis also demonstrated that:

Samples collected at the same time and same location showed relatively little variability;

250

Chapter 9
Samples collected at the same location but at different times could vary significantly, indicating that site
specific environmental factors are significant and could be highly correlated with the variation in measured
skid resistance;

In comparison to previous research, the particle size distribution of the collected samples is somewhat finer,
but very consistent for all samples and sites, and provides confidence in determining particle size
distributions for the controlled laboratory analysis methodology described in Section 5.9. and the results of
the analysis in Chapter 10;

In comparison to previous research, the sediment contained low concentrations of heavy metals except for
the Ports of Auckland site, where (as would be expected) significantly higher concentrations were found.
These relatively low concentrations of metals at the highway did indicate (from a preliminary analysis) that
the heavy metal concentration does not play a significant role in the variation in measured skid resistance.
However, more data are required to confirm this preliminary indication.

Chapter 10 will discuss the results of controlled laboratory experiments developed to simulate in-field polishing of
surfacing aggregates and the effect of various contaminants being placed upon prepared surface samples.

251

Chapter 10

10 LABORATORY POLISHING TEST RESULTS AND DISCUSSION

10.1 Introduction
As discussed in Section 5.9 a primary objective of the research was a better understanding of the factors that cause
the significant changes in skid resistance due to microtexture changes. As has been shown in Section 8.3, research
by Hill and Henry (1981) demonstrated that significant short-term variations in skid resistance exist over relatively
short periods of time. This largely unpredictable variation can not be explained by yearly seasonal patterns alone,
nor by the effects of temperature or rainfall changes (as discussed in Sections 6.5.1, 7.4 and 8.3.4 respectively). It
therefore became clear that other factors, hitherto not accounted for were affecting the variation in measured skid
resistance.
The literature review and the initial field data collection demonstrate the difficulty in understanding the complex
inter-relationships of skid resistance variables at any point in time. Too many in-field variables exist that can
neither be controlled nor monitored closely enough to be understood. It was apparent that controlled laboratorybased experiments were required to simulate the aggregates and road surface texture, traffic loading, and
environment conditions. The experiments would need to simulate certain conditions whilst other variables were
controlled to isolate their effects. This section discusses the results of the controlled laboratory-based experiments
that were designed and developed to simulate cycles of polishing and rejuvenation of surface aggregates. The
methodology that was used is discussed in Section 5.9.
Two surface samples for each aggregate type were constructed by hand placing the individual aggregate chips onto a
thin layer of sand on a plane of oiled glass. The aggregate chips were then bonded together in a mix of sand and
cement mortar. Once hardened, the mix was turned over, the sand brushed away and the mould set into a wooden
frame and cemented in place. One of each surface sample was left as a Master sample that remained un-polished
whilst the other sample was polished with the accelerated polishing machine developed and manufactured
specifically for this research. To examine and simulate the approximately seasonal and short-term variations of
measured skid resistance, required two stages of laboratory testing, namely:

Stage 1: Polishing the prepared surface samples to Equilibrium Skid Resistance (ESR) level; and

Stage 2: Simulating the cyclical effects of polishing/rejuvenation of the surface samples thereby decreasing
and increasing measured surface friction.

Some geological descriptions of the chosen aggregates are presented first. Secondly, the results of the controlled
laboratory experiments and the aggregate polishing that was undertaken are given. Some explanations of the
geological mechanisms of polishing and abrasion that affect measured skid resistance are also presented.

10.2 Geological Properties of the Aggregates

10.2.1 Introduction
The selected samples of sealing aggregates used in the controlled laboratory experiments were chosen on the basis
of the aggregates being commonly used in the Auckland and Northland Regions and/or having varying geological
properties. A range of low, medium and high reported PSV aggregates was also seen as being desirable for the test
matrix. Four aggregate sources were chosen from a range of geological types for the building of the laboratory
surface samples:

Moutohora sedimentary greywacke, being the highest reported natural PSV in the North Island (Napier);
252

Chapter 10
Holcim igneous basalt, mid to low range reported PSV, sourced from Auckland Region (Bombay);

Otaika sedimentary greywacke, lowest reported PSV acceptable for Transit NZ surfacings, locally sourced
from Whangarei, Northland; and

Melter slag, artificial by-product from Glenbrook Steel mill, through SteelServ Limited.

Professor Black (2005b), from the Department of Geology in the University of Auckland, supplied geological
descriptions of the four sample aggregates used in the laboratory experiments from specific tests undertaken in the
University of Auckland s Geology Laboratories. These descriptions and explanations are given in the following
sections.

10.2.2 Moutohora Greywacke

Greywacke sandstone chips from two quarries were tested in the laboratory experiment: Moutohora quarry from the
Matawai District north of Gisborne and Otaika quarry near Whangarei, Northland. These two sediments are
different in age and nature. The Cretaceous Moutohora sandstone, is more uniform in its properties, has a coarser
grainsize, and is inherently less lithified (i.e. it is more weakly metamorphosed) than the older late Triassic / Jurassic
greywacke from Otaika quarry near Whangarei. The Otaika greywacke is a fine grained sandstone variable in
grainsize (ranging through to siltstone) and strongly lithified. These differences influence the aggregates polishing
properties.
The Moutohora greywacke aggregate is part of the Raukumara Series from the Cretaceous geological period (from
Matawai District in Gisborne). The sample sealing chips showed a variety of colours indicating different degrees of
weathering. However, they all appeared to be coarse well sorted, uniform grain size sandstones as shown in Figure
10-1a.

(a)

Moutohora Chips (diameter field of view 3.5cm)

(b)

Microphoto of Typical Moutohora Sandstone Thinsection

(diameter of field of view 2mm)

Figure 10-1: Moutohora Sandstone Chips

In thinsection (refer to Figure 10-1b) the Moutohora chips are seen to be well sorted medium to coarse grainsupported sandstones. The grains are dominantly quartz, including a variety of polygranular quartz grains and
feldspar (plagioclase and k-feldspar). Lithic clasts are about 35% of the sand grain clasts and include sediment and
volcaniclastic debris. Grain shapes were subangular to subrounded. Detrital mineral grains include hornblende,
minor chlorite and biotite. The matrix constitutes about 10-20% of the sandstone and X-ray diffraction shows it

253

Chapter 10
consists largely of chlorite and illite. Some chips also have calcite cement. The clast-supported nature of the
sandstone, the variety of grains and their angular to subangular shape are clearly seen in Figure 10-1b.

10.2.3 Otaika Greywacke

The Otaika greywacke aggregate is from Whangarei and is from the Jurassic geological period. The Otaika chips
are dark grey in colour and massive. There is a range of grainsizes in the chips which vary from sand to siltstone.
The grain size of the sandstones is notably finer than that of the Moutohora sandstones.
In thinsections the Otaika sandstones show angular to subangular quartz grains, albitised feldspar and abundant
lithic debris (refer to Figure 10-2). The feldspars and lithic debris are more highly metamorphosed, and thus the
individual grains are less obvious in microphotos of the Otaika sample, compared with the Moutohoro sandstone,
although the amount of matrix in the two sandstones is roughly equivalent. White veins of prehnite are evident
cutting some grains and prehnite also occurs in the matrix where it replaces grains in the rock. Many chips contain
lenses or layers of siltstone.

Figure 10-2: Microphoto of Typical Otaika Greywacke Sandstone Thinsection(diameter field of view is 2mm)
Figure 10-2 shows the angular grain-supported sandstone composed of quartz, albitised feldspar and lithic grains top
and centre right, and siltstone lens bottom centre left.

10.2.4 Holcim Basalt

The Holcim basalt is from a quarry near Pokeno, South Auckland and is a completely crystalline porphyritic basalt.
In hand specimen the basalt samples are grey with brown nodular areas (phenocrysts) about 1mm in diameter and
have small vesicles. In thinsection (refer to Figure 10-3), the basalt is seen to contain phenocrysts of augite,
sometimes with cores clouded with iron oxides and overgrowths of augite. There is minor olivine in the rock (about
10%) as small crystals. The dominant component is plagioclase which is always in the matrix of the rock occurring
as needles enclosing iron oxides and augite. No glass was observed.

254

Chapter 10

Figure 10-3: Microphoto of Typical Holcim Basalt Thinsection (diameter field of view is 2mm)
Most of the large crystals are pyroxene, with a few smaller lighter coloured olivines (refer to Figure 10-3). Black
iron oxides show minor oxidation. The colourless matrix is an interlocking matte of plagioclase feldspar.

10.2.5 Melter Slag

The melter slag is an artificial aggregate that is an iron making by-product of the New Zealand Steel Mill in
Glenbrook. It is available through SteelServ Ltd (2005) and is part of a number of iron and steel by-products which
form aggregates that can be used in the New Zealand Roading Industry (refer to Figure 10-4). The Kaiwaka South
field site in Northland was a trial site for this aggregate to enable a consideration of its performance over time.

Iron making
Melter slag

New
NewZealand
ZealandSteel
Steel
(Glenbrook)
(Glenbrook)

Pacific
PacificSteel
Steel
(Otahuhu)
(Otahuhu)

Irons
Ironsand/Coal/Limestone
and/Coal/Limestone

Steel
SteelScrap
Scrap

Iron
IronMaking
Making
oror
Melter
MelterAggregate
Aggregate

Sub-base/Base
Sub-base/Basecourse
course
Drainage/surfacings
Drainage/surfacings

Steel
Steelmaking
making
oror
KOBM
KOBMaggregate
aggregate

Road
Roadstabilisation
stabilisation

Blend
Blend

Farm
Farmtracks
tracks
Temporary
TemporaryRoads
Roads

EAF
EAF
Steel
Steelmaking
making
Aggregate
Aggregate

Surfacing
SurfacingAggregates
Aggregates

Figure 10-4: Iron and Steel Making Aggregates in New Zealand (SteelServ Ltd & Bourke, 2005)
The melter slag chips are black with some reddish-brown patches. They appear metallic, but with reddish-brown
patches on surfaces and they contain many gas vesicles. Figure 10-5 shows an example of a chip thinsectioned that
contains large cubic crystals of blue-green magnesium (Mg) rich spinel mantled with a red-brown pseudobrookite (a
Ti-rich oxide). The matrix of the slag showed a typical quench texture consisting of interlocking dendrites of
extremely small grains of perovskite (CaTiO3) enclosed in pyroxene (colourless material). Many chips have
255

Chapter 10
vesicles and show brown oxidation effects. Figure 10-5 shows large crystals in the centre field view that are
magnesium - spinel surrounded by pseudobrookite.

Figure 10-5: Microphoto of Typical Melter Slag Thinsection (diameter field of view is 2mm)

10.3 Polishing Aggregates to Equilibrium Levels (Stage 1)

10.3.1 Introduction
A methodological procedure was developed and described in Section 5.9.6 to polish the prepared surface samples to
an equilibrium level whilst periodically measuring the variation of the coefficient of friction on both the unpolished and the polished sample with the Dynamic Friction Tester. Table 10-1 summarises, for each of the test
samples built (polished and un-polished samples), the aggregate source, the reported PSV from Transit New Zealand
surfacing specifications, the actual laboratory tested PSV, the geological grouping and aggregate properties and the
measured macrotexture of the surfacing sample.
All four paired aggregate samples were tested by the procedure set out in Section 5.9.6. Table 10-2 shows the
results of the mean of three test coefficients of friction () for an average slip speed of between 20 and 40km/h as
measured by the DF Tester for the first 6 hours of polishing or until an equilibrium level had been reached.

256

Chapter 10
1

Sample
Number

Aggregate
Source

TNZ
Reported
PSV

Actual
Sample
PSV

Geological
Grouping

Geological
Properties

Macrotexture TD
(mm)

Comment /
Descript

Moutohora

65

63

Sedimentary

Greywacke

Prototype

Moutohora

65

63

Sedimentary

Greywacke

1.0

Polished

Moutohora

65

63

Sedimentary

Greywacke

1.0

Un-polished

Holcim

56

52

Igneous

Basalt

0.8

Polished

Holcim

56

52

Igneous

Basalt

0.9

Un-polished

Otaika

51

52

Sedimentary

Greywacke

0.9

Polished

Otaika

51

52

Sedimentary

Greywacke

1.0

Un-polished

Melter Slag

58

55

NA

NA

1.1

Polished

Melter Slag

58

55

NA

NA

1.3

Un-polished

Notes:
1)

These PSV results are the actual results of a PSV test taken from a sample of each of the four sealing
aggregates. The tests were undertaken by TelArc-approved Works Infrastructure Testing Laboratories,
Auckland to BS EN 1097-8 (2000) specification.

Table 10-1: Laboratory Test Sample Details

Polishing
Hours
0.00
0.25
0.50
0.75
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
6.00

Moutohora PSV 65
S2-Pol
S3-UnP
0.87
0.90

0.77

0.87

0.68

Holcim PSV 56
S4-Pol
S5-UnP
0.60
0.61
0.65
0.57
0.60
0.59
0.57
0.58
0.53
0.58
0.50
0.58
0.46
0.56

Otaika PSV 51
S6-Pol
S7-UnP
0.51
0.59
0.58
0.55
0.53
0.50
0.51
0.49
0.50
0.48
0.47
0.45
0.45
0.44

0.62

0.83

0.44

0.55

0.43

0.44

0.54

0.81

0.41

0.54

0.42

0.43

0.39

0.56

0.52
0.52

0.78

Melter slag PSV 58

S8-Pol
S9-UnP
0.90
0.87
0.87
0.86
0.86
0.85
0.82
0.81

0.83
0.82

0.76

0.82

0.75, 0.72 0.79, 0.80

0.71

0.78

Table 10-2: Stage 1 Wet Polishing Results until 'Equilibrium Level'is reached
Macro photographs were also taken through a microscope of the surface of individual chips for each of the
laboratory- prepared aggregate samples used for accelerated polishing (i.e. Samples 2, 4, 6 and 8), both before and
after the accelerated polishing had been undertaken to an equilibrium level . This was done to determine whether
the acclerated polishing could produce microtextural changes on the aggregate surface that could be seen through a
microscope and that could offer an explanation for the lower reported values of measured skid resistance as tested
by the Dynamic Friction Tester. The following sections and figures show the results of the polishing and skid
resistance testing and associated microscopic photographs for each of the four paired laboratory samples.

257

Chapter 10

10.3.2 Moutohora Greywacke Laboratory Sample

Figure 10-6 compares the coefficient of friction of the paired sample results for the polished and unpolished samples
of the Moutohora greywacke aggregate with a Transit NZ published PSV of 65. The sample that was used in this
research was tested by Works Infrastructure testing laboratory and returned an actual PSV value of 63. The results
of the accelerated polishing and subsequent skid resistance testing, utilising the Dynamic Friction Tester show:

A high initial measured coefficient of friction of approximately DFT ( )=0.87 for both the unpolished and
polished samples was achieved;

The polished sample took approximately 6.5 hours of polishing by the accelerated polishing machine to
level off to an Equilibrium Skid Resistance (ESR) level;

The coefficient of friction reduced by 46% from initial measurements to ESR for the polished sample;

The coefficient of friction for the unpolished control sample reduced by approximately 20% with
approximately 33 tests (3x11) of the DFTester;

The polished sample ended up approximately 30% lower (DFT ( ) = 0.24), in measured skid resistance,
than the unpolished sample (using the accelerated polishing machine);

A highly significant coefficient of determination (R2=0.99) polynomial prediction equation can be fitted to
the polished sample data points.

Moutohora Greywacke (PSV 65) - Samples 2 and 3

1.00

Sample 2
Expon. (Sample 3)

0.90

Sample 3
Poly. (Sample 2)

Un-polished Sample
(Control)

DFT ( )

0.80
0.70

y = 0.8828e-0.0319x
R2 = 0.8894

0.60
Polished Sample

0.50

y = 0.0078x 2 - 0.1099x + 0.8695

R2 = 0.9861

0.40
0.30
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

Polishing time (hours)

Figure 10-6: Stage 1 Polishing of Moutohora Greywacke Aggregate
Macro photographs were taken through a microscope of a sample of the surface of a Moutohora aggregate chip
before and after the accelerated polishing had been undertaken to an equilibrium level on the polished laboratory
sample (Sample 2). Examples of the before and after photos are shown in Figure 10-7a and b respectively.

258

Chapter 10

(a)

Unpolished State (approx. field of view 3.5x 2mm)

(b)

Polished State (approx. field of view 3.5x 2mm)

Figure 10-7: Moutohora Greywacke Sandstone Chips (Unpolished and Polished states)
Figure 10-7a shows the angular nature of the grains in the unpolished surface as is the variety of clasts in this
sandstone. In the polished sample (refer to Figure 10-7b) the rounded surfaces on the lithic grain surfaces and the
clear quartz grains are notably compared with the more angular grains in the unpolished sample (Figure 10-7a).

10.3.3 Holcim Basalt Laboratory Sample

Figure 10-8 compares the paired sample results of the polished and unpolished samples for the Holcim basalt
aggregate with a Transit NZ published PSV of 56. The sample that was used in this research was tested by Works
Infrastructure testing laboratory and returned an actual PSV value of 52. The results of the accelerated polishing
and subsequent skid resistance testing utilising the Dynamic Friction Tester show:

An initial measured coefficient of friction of approximately DFT ( )=0.60 was achieved for both samples;

The polished sample took approximately five hours to level off to ESR;

The coefficient of friction reduced by 39% from initial measurements to ESR for the polished sample;

The coefficient of friction for the unpolished control sample reduced by approximately 8% with
approximately 30 tests (3x10) of the DFTester;

A significant coefficient of determination (R2 = 0.95) polynomial prediction equation can be fitted to the
polished sample data points.

259

Chapter 10

Holcim Basalt (PSV 56) - Samples 4 and 5

1.00
sample 4
Poly. (sample 4)

0.90

sample 5
Poly. (sample 5)

0.80

DFT ( )

Un-polished Sample (Control)

0.70

y = 0.0028x 2 - 0.0205x + 0.5927

R2 = 0.5971

0.60
0.50
Polished Sample

0.40

y = 0.009x 2 - 0.092x + 0.620

R2 = 0.951

0.30
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

Polishing time (hours)

Figure 10-8: Stage 1 Polishing of Holcim Basalt Aggregate
Macro photographs were taken through a microscope of a sample of the surface of a Holcim basalt aggregate chip
before and after the accelerated polishing had been undertaken to an equilibrium level on the polished laboratory
sample (Sample 4). Examples of the before and after photos are shown in Figure 10-9a and b, respectively.

(a)

Unpolished State (approx. field of view 3.5x 2mm)

(b)

Polished State (approx. field of view 3.5x 2mm)

Figure 10-9: Holcim Basalt Chips (Unpolished and Polished States)

Figure 10-9a and b of the Holcim Basalt chips clearly shows the vesicles and the dark grains which are pyroxene
and oxide crystals. The polished chip surface (Figure 10-9b) retains its roughness although individual mineral
grains show minor rounding.

260

Chapter 10

10.3.4 Otaika Greywacke Laboratory Sample

Figure 10-10 compares the paired sample results of the polished and unpolished samples for the Otaika greywacke
aggregate (a locally sourced aggregate in the Northland Region of NZ) with a Transit NZ published PSV of 51.
However, the sample that was used in this research was tested by Works Infrastructure testing laboratory and
returned an actual PSV value of 52. It should be noted that the Otaika greywacke aggregate, in terms of PSV, is the
lowest quality aggregate specified in the Transit NZ list of suppliers of surfacing aggregates (Transit NZ, 2004).
The results of the accelerated polishing and subsequent skid resistance testing utilising the Dynamic Friction Tester
show:

An initial average measured coefficient of friction of approximately DFT( )=0.55 for both samples
although the samples demonstrated a DFT ( )=0.08 difference in initial measured coefficient of friction
between the two samples (this however relatively quickly came together);

Interestingly, for this aggregate, both the polished and unpolished samples performed approximately the
same in terms of the deterioration of measured coefficient of friction;

The polished sample took approximately four hours to level off to ESR;

The coefficient of friction reduced by 24% from initial measurements to ESR for the polished sample;

The coefficient of friction for the unpolished control sample also reduced by approximately 23% with
approximately 30 tests (3x10) of the DFTester;

The polishing action of the DFTester (3x8=24 tests) for this lower PSV specified aggregate is surprisingly,
more aggressive initially than the accelerated polishing machine with two hours of polishing;

A significant coefficient of determination (R2 = 0.82) polynomial prediction equation can be fitted to the
polished sample data points.

Otaika Greywacke (PSV 51) - Samples 6 and 7

1.00

Sample #6
Poly. (Sample #6)

0.90

DFT ( )

0.80

Sample#7
Poly. (Sample#7)

Un-polished Sample
(Control)

0.70

y = 0.0175x 2 - 0.1x + 0.5626

R2 = 0.9093

0.60

Polished Sample
y = 0.007x 2 - 0.0604x + 0.5485
R2 = 0.8228

0.50
0.40
0.30
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

Polishing time (hours)

Figure 10-10: Stage 1 Polishing of Otaika Greywacke
Macro photographs were taken through a microscope of a sample of the surface of the Otaika greywacke sandstone
aggregate chip before and after the accelerated polishing had been undertaken to an equilibrium level on the
261

Chapter 10
polished laboratory sample (Sample 6). Examples of the before and after photos are shown in Figure 10-11a and b
respectively.

(a)

Unpolished State (approx. field of view 3.5x 2mm)

(b)

Polished State (approx. field of view 3.5x 2mm)

Figure 10-11: Otaika Greywacke Sandstone Chips (Unpolished and Polished states)
Figure 10-11a) shows the unpolished Otaika greywacke sandstone surface with generally fine grained but uniform
and irregular surface.

Figure 10-11b) shows the same aggregate surface after accelerated polishing showing

smearing and rounding of the matrix and grains.

10.3.5 Melter Slag Laboratory Sample

Figure 10-12 compares the paired sample results of the polished and unpolished samples for the Melter slag artificial
aggregate (a by-product from iron-making at the Glenbrook NZ steel mill). It has a Transit NZ published PSV of
58. However, the sample that was used in this research was tested by Works Infrastructure testing laboratory and
returned an actual PSV value of 55. The results of the laboratory accelerated polishing and DFTester friction tests
on the artificial melter slag show:

The initial level of SR for the slag (actual PSV 55) is similar to the Moutohora natural aggregate (actual
PSV 63);

The percentage reduction in measured skid resistance from the initial level of skid resistance is significantly
less than the natural aggregates (16% c.f. 46% for the Moutohora greywacke; 39% for the Holcim basalt
and 24% for the Otaika greywacke);

The time to polish the slag to its equilibrium polishing level with the accelerated polishing machine is
approximately the same as the Holcim Basalt aggregate (4-4.5 hours);

The melter slag significantly out-performed all of the natural aggregates in terms of resistance to polishing
including the Moutohora (actual PSV 63), which had a significantly greater PSV.

This confirms the field based results at the Kaiwaka Slag field site (refer to Figure 6-16 and Table 6-4) that was
resealed with melter slag aggregate. The performance of the melter slag is discussed in Section 6.4.4 and reported in
Wilson and Kirk (2005) who concluded that melter slag placed on high stressed highway corners in Northland has
similar long lasting performance.

262

Chapter 10

Melter Slag (PSV 58) - Samples 8 and 9

1.00
Un-polished Sample (Control)
0.90

y = 0.0025x 2 - 0.0284x + 0.8615

R2 = 0.9686

DFT ( )

0.80
0.70

Polished Sample
y = 0.0058x 2 - 0.0653x + 0.8873
R2 = 0.9817

0.60
0.50

Sample 9
Poly. (Sample 8)

0.40

Sample 8
Poly. (Sample 9)

0.30
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

Polishing time (hours)

Figure 10-12: Stage 1 Polishing of Melter Slag Aggregate
Macro photographs were taken through a microscope of a sample of the surface of the Melter slag aggregate chip
before and after the accelerated polishing had been undertaken to an equilibrium level on the polished laboratory
sample (Sample 8). Examples of the before and after photos are shown in Figure 10-13a and b, respectively.

(a)

Unpolished State (approx. field of view 3.5x 2mm)

(b)

Polished State (approx. field of view 3.5x 2mm)

Figure 10-13: Melter Slag Chips (Unpolished and Polished States)

Figure 10-13a and b show that the general surface of the melter slag is rough with vesicles and other irregularities.
In the polished state (refer to Figure 10-13b) the surface retains its irregularities in terms of relief although the
metallic oxides are clearly rounded by the polishing. The brown colour is probably pseudobrookite rather than rust.

263

10.3.6 Summary of Stage 1 Accelerated Polishing

Chapter 10

The combined summary results of the three natural aggregate samples and the melter slag artificial aggregate are
shown in Figure 10-14 and reported in Wilson and Dunn (2005b). All surface aggregates were polished using the
accelerated polishing machine developed at the University of Auckland (as shown in Figure 4-28) to an ESR level
(Stage 1). They were regulary tested for their measured skid resistance using the DF Tester. The results indicate
some interesting comparative results namely:

The actual PSV of the aggregate sample generally predicts the ranking order of the initial level of skid
resistance of natural aggregates prior to any accelerated polishing;

The percentage reduction in measured skid resistance, from the initial level of skid resistance to ESR,
reduces as the aggregate PSV reduces (from 46% for the Moutohora PSV 63 to 24% for the Otaika PSV
52);

Generally, the lower the PSV of the aggregate, the faster the aggregate polishes to its ESR level (from 7
hours of accelerated polishing for Moutohora PSV 63 to 2.5 hours for Otaika PSV of 52);

There is very little difference in the final level of ESR as measured by the DF Tester obtained for the three
natural aggregate samples (DFT()=0.47 for Moutohora PSV 63, DFT()=0.39 for Holcim basalt PSV 52
and DFT()=0.43 for Otaika PSV 52);

The greywacke sandstone and the basalt aggregates polish by different mechanisms, and therefore have
different deterioration rates, under the same accelerated polishing loads. Therefore the time to a polished
state and the ranking order of the final level of ESR for the natural samples may not be the same as the PSV
ranking. For example, the basalt sample polished to a greater extent than the Otaika greywacke even
though the actual measured PSV is the same.

The results of the Stage 1 laboratory accelerated polishing and DF Tester friction tests on the artificial melter slag
demonstrates some very promising results as this material has not deteriorated at the same rate or in the same
manner as the natural aggregates. The main findings are:

Whilst the actual tested PSV of the melter slag was a lot lower (PSV=0.55) than the Moutohora natural
aggregate (PSV=63) the initial level of measured skid resistance DFT()=0.87 was very similar;

The percentage reduction in measured skid resistance from the initial level of skid resistance to ESR is
significantly less for the melter slag than the natural aggregates of higher or lower measured PSV (16% c.f.
46%, 39% and 24% for the Moutohora, Holcim and Otaika aggregates);

The time to polish the slag to its equilibrium polishing level is approximately the same as the natural
aggregate of similar PSV (4-4.5 hours);

The melter slag significantly out-performed all of the natural aggregates in terms of being resistant to
polishing, including the Moutohora (PSV 65) which had a significantly greater PSV.

Trial tests sites with the melter slag (e.g. the Kaiwaka slag site in Northland) have demonstrated similar
measured skid resistance to polishing. The rate of decrease even in high stressed areas, is significantly
lower over reasonably long life cycles, and outperforms highly specified natural aggregates.

264

Chapter 10
Dynamic Friction Test Results ( ) Compared with Time in Accelerated Polishing hours
1.00

0.90

DF Tester ( )

0.80

0.70

0.60

0.50

0.40

0.30
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

Time in Polishing (hours)

S2-Moutohora PSV 63 Polished
S5-Holcim PSV 52 Unpolished
S8-Slag PSV 55 Polished

S3-Moutohora PSV 63 Unpolished

S6-Otaika PSV 52 Polished
S8-Slag PSV 55 Un-Polished

S4-Holcim PSV 52 Polished

S7-Otaika PSV 52 Unpolished

Figure 10-14: Summary of Accelerated Polishing of all Laboratory Samples to ESR (Stage 1)

10.4 Geological Properties and the Aggregates Resistance to Polishing

10.4.1 Introduction
Previous studies have indicated that the nature of the minerals present and the material s microtexture play major
roles in an aggregate s Polished Stone Value (PSV) and/or its resistance to polishing by traffic (Smith & Collis,
2001). The relative hardness of individual minerals has been particularly identified as an important factor in
determining the polishing properties of the material (Neville, 1974).
Natural rocks (and in this case also slags) are heterogeneous materials composed of several different minerals which
frequently have different grainsizes and shapes as well as different chemical and physical properties. Further, there
are differences in the nature and strength of the cement that bonds the constituent mineral grains. Volcanic rocks
and slags which have crystallised and were quenched at high temperatures have the individual mineral grains welded
together or cemented by glass. On the other hand, the grains in sedimentary rocks are bound together by a matrix
which is usually dominated by clay, if poorly lithified, or by silica cement if strongly lithified.
Studies that have been undertaken of the grinding/polishing behaviours of various rocks and minerals (Attaway,
2005; Evans et al., 2003; Golini & Jacobs, 1991; Lampropoulos et al., 1997; Lampropoulos et al., 1996; Sinkankas,
1999) have clearly shown that :

Mixtures of hard and soft minerals will not polish well and thus the rock surface will retain its roughness;

The harder the mineral, the less effect abrading/grinding has on the surface of the mineral;

The depth of scratch when an abrasive medium is used correlates with the hardness difference between the
abrasive and abraded material (the greater the difference, the deeper the scratch);

Finer grain sizes of abrasive material remove more material from surfaces;

265

Chapter 10
The removal rate of material is influenced by the environment (i.e. the fluid in contact with the abrading
material); water as a lubricant aids brittle fracture;

Surface stress increases with decreasing grit size of abrading/polishing media;

Polishing of samples with very fine but relatively harder or softer abrasive material causes surface
deformation (work hardening); this occurs irrespective of whether the polishing medium is harder or softer
than the material polished; and

The lapping/grinding/polishing process involves both brittle fracture and ductile deformation of the surface.

Abrasive wear is the grinding/scratching (surface roughening) process that takes place when material is removed
from the surface by harder abrasive grains. The abrasive grains hammer surface asperities, knocking small particles
(brittle failure) out of the sample surface and causing strong local deformation in the grain that has been hammered.
Empirical studies of ground surfaces have shown that for a depth of cut less than 1m only material flow is
observed around the cuts. As the depth of the cut increases cracking and material flow occurs. At depths of 10m
or more, scratches are associated with fracturing in the material, and large scale chipping and crushing has been
observed which weakens and roughens the overall surface that is, deep scratching results in mechanical grinding
of the surface. Under constant conditions the depth of the cut (scratch) will be related to the hardness difference
between the cutting and cut materials. Adjacent to the cracks and fractures are areas of strained / deformed material.
Thus, in all modes of grinding, there is a deformed subsurface layer caused by the buildup of stress (shear or work
hardening).

According to Golini and Jacobs (1991): the magnitude and depth of permanent deformation

accompanying the brittle process depends on the abrasive size with smaller abrasives producing a higher overall
percentage of deformation caused by plastic flow .
Polishing of relatively larger crystals, either by fine micron sized harder particles or by softer materials (such as
clay), is a chemomechanical process that alters the surface properties of the material being polished.

The

grinding/polishing process induces a layer of compressive stress on the surface of the sample. The compressive
stress alters the surface properties of the material, producing a surface layer along which smoothing of the surface
by ductile deformation occurs.

10.4.2 Constituents of the Aggregate Samples

Table 10-3 shows the major mineral constituents of the four aggregates that were examined, listed in order of their
Moh s hardness number (i.e. their order of resistance to abrasion/scratching), together with other diagnostic
properties that are relevant to the minerals polishing and failure/fracture properties. However, it should be noted
that Moh s hardness scale is not linear, and its relationship to the hardness number determined by indentation
methods (such as the Vickers hardness number) is log / linear.
Many minerals, particularly silicates and oxides, are brittle at ambient temperatures, that is they do not have the
capacity to store energy in the form of strain, and react to stress by fracturing. Brittle minerals that are strongly
anisotropic in terms of their physical properties and which have crystal structures which contain planes of different
atomic densities and bonding, will break along closely spaced planar surfaces which reflect planes of weakness in
the crystal lattice (cleavage). If the mineral has a highly uniform internal crystal structure it will break in a random
manner (fracture). Some hard minerals which lack cleavage, break along randomly located fracture surfaces with
curved concavities (conchoidal fractures). The nature of the fracture and the presence or absence of cleavages are
diagnostic properties of individual mineral species, since they reflect the fundamental crystal structure.

266

Chapter 10
Mineral

Formula

Spinel

MgAl2O4

Quartz

SiO2

Olivine

Mg2SiO4

Prehnite

Ca2Al2Si3O10(OH)2

Plagioclase
feldspar

CaAl2Si2O6

Pseudobrookite

FeTiO5

Ilmenite

Mohs scale
Hardness
7.5 - 8

Fracture Properties

Aggregate
Occurrence

No cleavages

Slag

Conchoidal fracture

Greywacke

Conchoidal fracture

Basalt

6.5

One good cleavage

Greywacke

Two sets of perfect to

good cleavages

Basalt and
Greywackes

No cleavage

Slag

FeTiO3

No cleavage

Slag

Magnetite

Fe3O4

5.5 - 6

No cleavage

Basalt

Perovskite

CaTiO3

5.5

Poor cleavage

Slag

Pyroxene

Ca (Fe,Mg) SiO3

5-6

Two sets of good

cleavages

Basalt and Slag

6 6.5

Table 10-3: Major Mineral Constituents of the Aggregates used in the Laboratory Experiments (Black,
2005b)
Abrasion in the brittle mode (the case for all the samples considered) is a fracture process in which the surface is
reduced by fracture and chipping (Golini & Jacobs, 1991). In the abrasive grinding environment, minerals that have
two sets of good to excellent cleavages, such as pyroxenes and feldspars, will break along those planes. The
intersections of the cleavage planes provide points of weakness causing surface pitting and the intersecting cleavage
systems also allow small rectangular blocks of the crystal to be plucked out of the rock surface. In this manner, the
surface of the rock is mechanically eroded, retreats and becomes smoother. Hard silicates, such as olivine and
quartz which do not have cleavage planes, tend to pluck out of the rock as entire crystals which then scratch the
surface of other minerals as continued polishing eliminates them from the surface.
The metallic oxides, on the other hand, lack cleavages and have polishing hardness that is they retain their relief
and stand proud (i.e. have a higher surface level) with respect to the silicate minerals. Thus, the presence of
abundant metal oxides increases a material s polishing resistance.

10.4.3 A Geological Interpretation of the Polishing of the Sample Aggregates

Black (2005b) states that the petrography of the aggregates that have been tested in this research, and their
microtextures, allows an interpretation of the results in terms of the measured skid resistance obtained with the
Dynamic Friction Tester.
The Melter Slag, which has an inherently irregular surface with many vesicles, is composed dominantly of metal
oxides. Spinel, which is a substantial component of the slag, is an exceptionally hard mineral (Moh s scale 7.5 8)
and the other Titanium oxides (pseudobrookite and ilmenite) are also very cohesive and hard. None of these metal
oxides have cleavages. The micro-texture of the rock is formed by interlocking large crystals of the metal oxides
which provide cohesion to the material. Since they polish hard, these oxides effectively protect the smaller
perovskite and pyroxene crystals located between them from abrasion in the polishing process, while still providing
a surface relief that has rough microtexture and therefore high skid resistance. Thus, it is not surprising that the slag
267

Chapter 10
has a reasonably high initial skid resistance and out-performs, over time, the tested natural aggregates in terms of
skid resistance.
The Holcim Basalt sample is composed dominantly of the silicate minerals, plagioclase feldspar and augite. These
minerals have similar hardness and both have two very good cleavage sets that intersect at approximately 90, which
allow the mineral grains to be physically broken down. The only hard silicate mineral present (olivine) in the
Holcim basalt constitutes less than 10% of the rock and occurs as small crystals interlocked with the other silicate
minerals. The magnetite also occurs as small crystals integrated in the rock matrix and has a similar hardness to the
dominant silicate minerals. Thus the polishing and abrasion of the basalt surface occurs relatively evenly and there
is less change between the before and after polishing phases. However, as there is less initial harshness in the
microtexture, a low measured polished skid resistance results.
Greywacke Sandstone Chips from two quarries were tested: Moutohora Quarry from the Matawai District north of
Gisborne and the Otaika Quarry from Whangarei. These two greywacke sediments are different in age, degree of
metamorphism (lithification) and nature.
The Moutohora Sandstone is from the Cretaceous period and is more uniform in its properties. It has a coarser
grainsize, and is inherently less lithified than the older late Triassic / Jurassic greywacke from Otaika quarry near
Whangarei. The different colours of the chips (greenish, brownish and grey, refer to Figure 10-1), indicates
different degrees of oxidation, together with a lack of veining, that shows the Moutohora greywacke still retains
some permeability with respect to water.
The Otaika Greywacke is a finer grained sandstone, variable in grainsize (ranging through to siltstone) and
strongly lithified by low grade metamorphism.

Metamorphic prehnite is found in the matrix of the Otaika

greywacke and in white veins cutting the rock.

The ratio of the major X-ray diffraction peaks for quartz to plagioclase in bulk Moutohora greywacke sample is
10:11 compared to 10:6 in the Otaika greywacke. It is not possible at this stage to determine the absolute amounts
of quartz and feldspar present in the two greywackes (the Moutohoro sample contains minor potassium feldspar as
well as plagioclase feldspar), but the relative proportions and peak heights (intensities) of the quartz and feldspar in
the diffractogrammes indicate that the Moutohora sample contains approximately 40% more feldspar than the
Otaika greywacke. Judging from visual estimates of quartz and feldspar in the thinsections, much of the excess
quartz in the Otaika samples must be in the recrystallised metamorphic cement/matrix of the rock.
These differences in degree of lithification, mineral content and grainsize provide explanations for the different
polishing and skid resistance properties of the two greywackes.

10.5 The Results of Polishing with Contaminants (Stage 2)

10.5.1 Introduction
Once each of the polished surface samples had clearly reached an equilibrium level for that specific aggregate,
load and polishing action (Stage One), the samples were ready for the Stage Two polishing phase. The Stage Two
laboratory polishing phase required the simulation of the approximately seasonal variation of measured skid
resistance. This required simulating the cyclical effects of variation of the summer and winter polishing, that is,
rejuvenation of surface samples through the effects of contaminants, rainfall and vehicle trafficking. Specific
procedures were developed and are reported in Section 5.9.6 to determine the effect on the variation of the
coefficient of friction, when specific additives were placed upon the surface, followed by accelerated polishing to
268

Chapter 10
simulate traffic loads. The polishing action and the method of wetting and drying of the surface, and all other
known variables were held constant (where possible) to isolate the effect of the additive.
The theory of the Phase Two variation in measured skid resistance (known as the seasonal variation ) as described
by Jayawickrama and Thomas (1998) and discussed in Section 3.5.6, has been that the accumulation of finer particle
detritus material contributes towards the polishing of the aggregates (reducing skid resistance), whereas the coarse
grit contributes to abrasion of the aggregate which helps to rejuvenate skid resistance. Cycles of rainfall wash away
finer material, thereby allowing the grit to rejuvenate skid resistance during wetter periods. During drier periods, the
finer material accumulates, polishing the aggregate and reducing skid resistance.
The Stage Two polishing phase and procedures were developed to test this theory by the addition of a number of
additives to the laboratory surface samples combined with accelerated polishing to measure the effect on measured
skid resistance.

10.5.2 Contaminant Additives

The sample additives that were chosen were similar to the detritus particle size distribution expected to accumulate
in the field and are described in Section 5.9.6. They are summarised as:

Oedometer clay a soft but well graded material that is predominatly kaolinite and is strongly anistropic in
terms of its properties, with a Moh s hardness of 2-2.5;

Emery powder - a fine but very hard material derived predominantly of corundum (Al203) with a Moh s
hardness of 9;

Leighton Buzzard sand - a coarse and hard material that is predominantly from quartz with a Moh s
hardness of 7.

The Phase Two Accelerated Polishing (AP) consisted of measuring the effect of the following variations in turn, to
the Phase One wet polishing phase:

Dry polishing with the addition of Oedometer clay (10 grams);

Dry polishing with the addition of sieved Oedometer clay (10 grams) with particle sizes of <0.15 mm;

Dry polishing with the addition of sieved Oedometer clay (10 grams) with particle sizes of >1.15 mm;

Dry polishing with the addition of Emery Powder (10 grams);

Dry polishing with the addition of Leighton Buzzard sand (20 grams); and

Wetted surface (damp) polishing with the addition of Oedometer clay (10 grams).

After each polishing variation had been tested, the surface sample was then polished, wet but with no additives, to
try to restore the sample to its equilibrium skid resistance level . The following sections discuss the results in terms
of each of the four laboratory samples.

10.5.3 Moutohora Greywacke

As discussed in Section 10.3.2, the Moutohora sample that was wet polished using the Accelerated Polishing
Machine (APM), reduced the Stage 1 accelerated polishing from an initial value of DFT()=0.87 to an equilibrium
wet polished level of DFT()=0.47. The Stage 2 polishing phase included the accelerated polishing of the sample
(in wet, dry and damp conditions) with the addition of either oedometer clay, emery powder or Leighton Buzzard
sand. Figure 10-15 shows the results of the two stages of polishing for the two Moutohora greywacke laboratory
samples (one with accelerated polishing and the other with no polishing). Some increases and decreases were
expected that could be considered as being seasonal in their extent. However, the extent of the variation that could
269

Chapter 10
occur with the addition of contaminants and accelerated polishing was surprising. Measured skid resistance levels
were reached that were very close to the initial measured coefficient of friction prior to any accelerated polishing,
after only 10 minutes of accelerated polishing with Leighton Buzzard sand. An increase of DFT()=0.20) occurred
after an ESR level of approximately DFT()=0.50 had been established, after a total of 650 minutes of accelerated
polishing (as shown in Figure 10-15).

Moutohora Greywacke Aggregate DFT( ) Stage 1 and 2 Polishing

1.0
0.9

Unpolished Sample

DF Tester ( )

0.8
0.7
0.6
0.5

Polished Sample

0.4
0.3

Stage 1: Polishing to ESR

Stage 2: Polishing with Additives

0.2

See Figure 10-16

0.1
0.0
0

50

100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

850

Time (mins)
S2 - Moutohora PSV 63 Polished Sample

S3 - Moutohora PSV 63 Unpolished Sample

Figure 10-15: Moutohora Greywacke Laboratory Sample (Stages 1 and 2 Polishing Phases)
These significant variations require explanation sequentially, in terms of which additive was used, the effect of the
additive in terms of the measured coefficient of friction with the DF Tester () and then what may have caused the
variation.

The explanation for the Moutohora greywacke sample is visually displayed in Figure 10-16.

chronological explanation and its effect in terms of measured coefficient of friction from the beginning of Stage 2
polishing follows:

The addition of 10 grams of Oedometer clay and accelerated polishing for 10 minutes with no water
increased the DFT() value by 0.10 to 0.58, a percentage increase of 23.4%;

Further additions of 10 grams of Oedometer clay and samples that had been seived to be less than 0.15mm
Particle Size Distribution (PSD) did not further modify the measured DFT() result obtained from the first
oedometer clay sample;

The polished sample was then wet polished with no additives for 15 minutes and the subsequent measured
DFT() returned to a level similar (but slightly higher) to that determined as being the previous ESR level
at the end of the Stage 1 polishing phase, DFT()=0.53;

An addition of 10 grams of oedometer clay material that was retained on a 1.15mm sieve with dry
accelerated polishing also increased the DFT(), back to 0.58;

The addition of five 10 gram doses (5 x 10 grams) of emery powder every 10 minutes of dry accelerated
polishing resulted in a minor decrease in measured DFT() to 0.55;

Wet polishing for 30 minutes with no contaminants then reduced the measured DFT() to 0.48, almost
exactly the same level that was obtained at the end of the Stage 1 polishing;
270

Chapter 10
A series of 4 x 20 grams of Leighton Buzzard sand was then added to the sample every 10 minutes for

accelerated polishing and a measured DFT() result was obtained. The initial 10 minutes of polishing with
the Leighton Buzzard sand produced a DFT() rise of 0.18 to 0.66; a significant increase. Subsequent
additions of the sand and 10 minutes of polishing produced only minor further increases in measured
DFT() in comparison to the first;

The next dry polishing phase comprised the addition of 3 x 10 grams of emery powder every 10 minutes
after the Leighton Buzzard sand accelerated polishing phase was completed. A measured DFT() was not
obtained until the end of the 30 minute period that resulted in a signficant decrease in measured surface
friction, back to the levels of the ESR level measured at the end of Stage 1. The resultant value was a
measured DFT() of 0.49.

A further 30 minutes of wet polishing with no contaminants was then undertaken, which resulted in a slight
decrease in measured DFT() to a level of 0.45, a little lower than the initial ESR level obtained.

The full PSD of the oedometer clay was then added in 3 x 10 gram batches with the surface being very
lightly sprayed with water to keep the surface in a damp condition. The measured skid resistance level was
increased, this time signficantly higher than for the same dry polishing additive undertaken previously, with
a measured DFT() of 0.67 being obtained;

A further 30 minutes of wet polishing with no contaminants produced a characteristic decrease in measured
DFT() of 0.06 to 0.61;

A repeated cycle of 2 x 20 grams of Leighton Buzzard sand for 10 minutes each with accelerated polishing,
produced an increase in the measured DFT() value to 0.67 similar to that in the first cycle and;

A final 3 x 10 grams of emery powder every 10 minutes with dry accelerated polishing, produced a
measured DFT() value of 0.41, the lowest obtained for the Moutohora sample.

Moutohora Aggregate DFT( ) and Addition of Contaminants

(Stage 2 Polishing) over Time (mins)
0.8

Three 10g of Emery Powder for threex10 mins

Two 20g of Leighton Buzzard Sand for twox10 mins

Wet Polishing for 15 minutes

Wet Polishing for 15 minutes

Wet polishing with 10g of Oedometer Clay

Wet polishing with 10g of Oedometer

Wet polishing with 10g of Oedometer

Wet Polishing for 15 minutes

Wet Polishing for 15 minutes

Three 10g of Emery Powder for three x 10 mins

Added 20g of Leighton Buzz Sand

Added 20g of Leighton Buzz Sand

Added 20g of Leighton Buzz Sand

Added 20g of Leighton Buzz Sand

Wet Polishing for 30 minutes

Added Emery Powder

Added Emery Powder

Added Emery Powder

Added Emery Powder

Added Emery Powder

Oedometer Clay - Sieve >1.15mm

0.2

Wet Polishing for 15 minutes

0.3

Added Oedometer Clay

0.4

Oedometer Clay - Sieve <0.15mm

0.5
Added Oedometer Clay

DF Tester ( )

0.6

1.3

0.8

0.3

Differences in DFT ( )

1.8

0.7

0.1

350

340

330

320

310

300

290

280

270

260

250

240

230

220

210

200

190

180

170

160

150

140

130

120

110

100

90

80

70

60

50

40

30

20

-0.2
10

Time (min)
S2-Moutohora PSV 63 Polished

S3-Moutohora PSV 63 Unpolished

Differences

Figure 10-16: Moutohora Greywacke Laboratory Samples (Stage 2 Polishing with Addition of Contaminants)
271

Chapter 10

10.5.4 Holcim Basalt

As discussed in Section 10.3.3, the Holcim basalt sample that was wet polished using the Accelerated Polishing
Machine (APM) initially roughened up to DFT()=0.63 (from an initial value of DFT()=0.60). It then decreased
due to polishing, to an equilibrium wet polished level of DFT()=0.39 (Stage 1 accelerated polishing). The Stage 2
polishing phase included the accelerated polishing of the sample (in wet, dry and damp conditions) with the addition
of either oedometer clay, emery powder or Leighton Buzzard sand. Figure 10-17 shows the results of the two stages
of polishing for the two Holcim basalt laboratory samples (one with accelerated polishing and the other with no
polishing). Measured skid resistance levels were reached on the Stage 2 contaminant testing phase that were even
higher than the initial measured coefficient of friction prior to any accelerated polishing. After only 10 minutes of
accelerated polishing with Leighton Buzzard sand, an increase of DFT()=0.35 occurred after an approximately
DFT() level of 0.31 had been established after 370 minutes of accelerated polishing. Furthermore, when emery
powder was added for 10 minutes of accelerated polishing, the DFT() reduced by 0.33 from the peak increase level
(DFT()=0.66) gained with the Leighton Buzzard sand.

Holcim Basalt Aggregate DFT( ) Stage 1 and 2 Polishing

0.7
Unpolished Sample

DF Tester ( )

0.6
0.5
0.4

Polished Sample

0.3
0.2

Stage 2: Polishing with Additives

Stage 1: Polishing to ESR

0.1

See Figure 10-18

0.0
0

50

100

150

200

250

300

350

400

450

500

550

Time (mins)
S4 - Holcim PSV 52 Polished Sample

S5 - Holcim PS 52 Unpolished Sample

Figure 10-17: Holcim Basalt Laboratory Sample (Stages 1 and 2 Polishing Phases)
These significant variations require some explanation sequentially as to what additive was given, the effect of the
additive in terms of the measured coefficient of friction with the DF Tester () and then what may have caused the
variation to occur. The explanation for the Holcim basalt sample is visually displayed in Figure 10-18 and a
chronological explanation and its effect in terms of measured coefficient of friction, from the beginning of Stage 2
polishing follows:

The addition of 10 grams of Oedometer clay material that was retained on a 1.15mm sieve, with dry
accelerated polishing for 10 minutes increased the DFT() value by by 0.02 to 0.41, a percentage increase
of 5.1%;

The addition of 3 x 10grams of emery powder every 10 minutes of dry accelerated polishing resulted in a
decrease in measured DFT() to 0.33;
272

Chapter 10
The polished sample was then wet polished with no additives for 30 minutes and the subsequent measured
DFT() did not significantly alter from the level reached previously.
DFT()=0.32, 0.01 lower than the level reached with the emery powder.

It remained at a level of
This level was however,

significantly lower than the level determined as being the ESR level previously DFT()=0.39 at the end of
the Stage 1 polishing phase;

A series of 2 x 20 grams of Leighton Buzzard sand was then added to the sample every 10 minutes for
accelerated polishing and a measured DFT() result was obtained. The initial 10minutes of polishing with
the Leighton Buzzard sand produced a DFT() increase of 0.15 to 0.46; and the second 10 minutes a
further DFT() of 0.19 to 0.65, both significant increases;

The next dry polishing phase comprised the addition of 3 x 10 grams of emery powder every 10minutes
after the Leighton Buzzard sand accelerated polishing phase was completed. A measured DFT() was not
obtained until the end of the 30 minute period that resulted in a signficant decrease in measured skid
resistance to lower levels than the recorded ESR level, measured at the end of Stage 1. The resultant value
was a measured DFT() of 0.23 and a significant loss of DFT() 0.43.

A further 30 minutes of wet polishing with no contaminants was then undertaken which resulted in a slight
increase (that was against the commonly observed trend) in measured DFT() to a level of 0.26, however
still lower than the original wet polished ESR value, obtained at the end of the Stage 1 polishing phase;

The full PSD of the oedometer clay was then added in 3 x 10 gram batches with the surface being very
lightly sprayed with water to keep the surface in a damp condition. The measured skid resistance level
after the first 10 minutes of accelerated polishing increased to a level of DFT()=0.41, that was 0.15 higher
than that measured previously and (restablishing the same level as ESR at the end of the Stage 1 phase).
Two further 10 minutes of accelerated polishing with additional skid resistance measurements did not
further alter the result;

A further 30 minutes of wet polishing with no contaminants added, produced a characteristic decrease in
measured DFT() of 0.07 to 0.34;

A repeated cycle of 2 x 20 grams of Leighton Buzzard sand for 10 minutes each with accelerated polishing
produced an expected increase (as with the earlier cycle) in the measured DFT() value to 0.44, so did not
attain the same level of measured skid resistance DFT() as obtained in the first cycle;

A final 3 x 10 grams of emery powder every 10 minutes with dry accelerated polishing produced a reduced
measured DFT() value of 0.22, a reduction of DFT()=0.22 from the previous level of 0.44. This is the
lowest value obtained for the Holcim basalt sample and any of the other three laboratory surface samples.

273

Chapter 10
Holcim Basalt DFT( ) and Addition of Contaminants
(Stage 2 Polishing) over Time (mins)
0.7

5.0
4.5

0.6

4.0

2.5
2.0
Three 10g of Emery Powder
for three 10 minutes

Two 20g of Leighton Buzzard

Sand for two 10 minutes

Wet Polishing for 15 minutes

Wet Polishing for 15 minutes

Wet polishing with 10g of Oedometer Clay

Wet polishing with 10g of Oedometer Clay

Wet polishing with 10g of Oedometer Clay

Wet Polishing for 15

mins

Wet Polishing for 15 mins

Three 10g of Emery Powder for

three 10 mins

Added 20g of Leighton Buzz Sand

Wet Polishing for 30 minutes

0.1

Added Emery Powder

0.2

Added Emery Powder

0.3

Added Emery Powder

Added Oedometer Clay - Sieve >1.15mm

DF Tester ( )

3.0
0.4

1.5

Differences in DFT ( )

3.5

0.5

1.0
0.5
0.0

260

250

240

230

220

200

210

190

180

170

160

140

150

130

120

110

90

100

80

70

60

50

40

30

20

10

-0.5
0

Time (mins)
S4-Holcim PSV 52 Polished

S5-Holcim PSV 52 Unpolished

Differences

Figure 10-18: Holcim Basalt Laboratory Samples (Stage 2 Polishing with Addition of Contaminants)

10.5.5 Otaika Greywacke

As discussed in Section 10.3.4 the Otaika greywacke sample that was wet polished using the Accelerated Polishing
Machine (APM) initially roughened up to DFT()=0.58 (from an initial value of DFT()=0.49). It then decreased
due to polishing, to an equilibrium wet polished level of DFT()=0.42. The Stage 2 polishing phase included the
accelerated polishing of the sample (in wet, dry and damp conditions) with the addition of either oedometer clay,
emery powder or Leighton Buzzard sand. Figure 10-19 shows the results of the two stages of polishing for the two
Otaika greywacke laboratory samples (one with accelerated polishing and the other with no polishing). Measured
skid resistance levels were reached on the Stage 2 contaminant testing phase that were almost as high as the initial
measured coefficient of friction prior to any accelerated polishing. After only 10 minutes of accelerated polishing
with Leighton Buzzard sand, an increase of DFT()=0.20 occurred after a DFT() level of approximately 0.33 had
been established following 355 minutes of accelerated polishing (as shown Figure 10-19). Furthermore, when
emery powder was added for 10 minutes of accelerated polishing, the DFT() reduced by 0.27 from the peak
increase level (DFT()=0.53) gained with the Leighton Buzzard sand.

274

Chapter 10

Otaika Greywacke Aggregate DFT( ) Stage 1 and 2 Polishing

0.7
Unpolished Sample

DF Tester ( )

0.6
0.5
0.4

Polished Sample

0.3
0.2

Stage 1: Polishing to ESR

Stage 2: Polishing with Additives

0.1

See Figure 10-20

0.0
0

50

100

150

200

250

300

350

400

450

500

Time (mins)
S6 - Otaika PSV 52 Polished Sample

S7 - Otaika PSV 52 Unpolished Sample

Figure 10-19: Otaika Greywacke Laboratory Sample (Stages 1 and 2 Polishing Phases)
These significant variations require some explanation sequentially in terms of what additive was used, the effect of
the additive in terms of the measured coefficient of friction with the DF Tester () and then what may have caused
the variation to occur. The explanation for the Otaika greywacke sample is visually displayed in Figure 10-20. A
chronological explanation and its effect in terms of measured coefficient of friction from the beginning of Stage 2
polishing follows:

The addition of 10 grams of Oedometer clay material that was either the full PSD of the sample, the
material retained on a 1.15mm sieve, or the material that was sieved through a 0.15mm seive with dry
accelerated polishing for 10 minutes, increased the DFT() value by 0.01/0.02 to 0.43, a percentage
increase of 5.1%;

The addition of 3 x 10 grams of emery powder every 10 minutes of dry accelerated polishing resulted in a
decrease in measured DFT() to 0.34;

The polished sample was then wet polished with no additives for 30 minutes and the subsequent measured
DFT() did not significantly alter from the level reached previously, remaining at a level of DFT()=0.34.
This level, was significantly lower than the level identified previously as the ESR level, DFT()=0.42 (at
the end of the Stage 1 polishing phase);

A series of 2 x 20 grams of Leighton Buzzard sand was then added to the sample every 10 minutes for
accelerated polishing and a measured DFT() result was obtained. The initial 10 minutes of polishing with
the Leighton Buzzard sand produced a DFT() increase of 0.20 to 0.53. The second 10 minutes resulted in
a slight decrease in DFT() of 0.01 to 0.52. Overall, this is a significant increase from the ESR level that
was previously observed;

The next dry polishing phase comprised the addition of 3 x 10 grams of emery powder every 10 minutes
after the Leighton Buzzard sand accelerated polishing phase was completed. A measured DFT() was not
obtained until the end of the 30 minute period that resulted in a signficant decrease in measured skid

275

Chapter 10
resistance to lower levels than the recorded ESR level measured at the end of Stage 1. The resultant value
was a measured DFT() of 0.26 and a significant reduction of DFT() of 0.26.

A further 30 minutes of wet polishing with no contaminants was then undertaken. This resulted in a slight
increase (that was against the commonly seen trend) in measured DFT() to a level of 0.27, still lower than
the original wet polished ESR value obtained at the end of the Stage 1 polishing phase;

The full PSD of the oedometer clay was then added in 3 x 10 gram doses, with the surface being very
lightly sprayed with water to keep the surface in a damp condition. The measured skid resistance level
after the first 10 minutes of accelerated polishing increased to a level of DFT()=0.45 that was 0.18 higher
than that measured previously and also the same level as the ESR at the end of the Stage 1 phase. Two
further 10 minutes of accelerated polishing with additional skid resistance measurements increased the
measured DFT() at a much slower rate to 0.49;

A further 30 minutes of wet polishing with no contaminants added, produced a characteristic decrease in
measured DFT() of 0.06 to 0.43;

A repeated cycle of 2 x 20 grams of Leighton Buzzard sand for 10 minutes each with accelerated polishing
produced an expected increase (as with the earlier cycle) in the measured DFT() value to 0.45, but not to
the same level of measured skid resistance DFT() as obtained in the first cycle;

A final 3 x 10 grams of emery powder every 10 minutes with dry accelerated polishing, produced a reduced
measured DFT() value of 0.37, a reduction of DFT()=0.08.
Otaika Aggregate DFT( ) and Addition of Contaminants
(Stage 2 Polishing) over Time (mins)

0.6

3.0

1.5

Three 20g of Emery

Powder for three 10 minutes

Two 20g of Leighton Buzzard Sand

for two 10 minutes

Wet Polishing for 15 minutes

Wet Polishing for 15 minutes

Wet polishing with 10g of Oedometer Clay

Wet polishing with 10g of Oedometer Clay

Wet Polishing for 15

minutes

Three 20g of Emery Powder for three 10 minutes

0.1

Added 20g of Leighton Buzz Sand

Added 20g of Leighton Buzz Sand

Wet Polishing for 30 minutes

Added Emery Powder

Added Emery Powder

Added Emery Powder

Oedometer Clay - Sieve >1.15mm

Wet Polishing for 15 minutes

Oedometer Clay - Sieve <0.15mm

0.2

Added Oedometer Clay

Added Oedometer Clay

DF Tester ( )

0.3

Wet polishing with

10g of Oedometer Clay

2.0
0.4

1.0

0.5

Differences in DFT( )

2.5

0.5

0.0

290

280

270

260

250

240

230

220

210

200

190

180

170

160

150

140

130

120

110

100

90

80

70

60

50

40

30

20

10

-0.5
0

Time (mins)
S6-Otaika PSV 52 Polished

S7-Otaika PSV 52 Unpolished

Difference

Figure 10-20: Otaika Greywacke Laboratory Samples (Stage 2 Polishing with Addition of Contaminants)

10.5.6 Melter Slag

As discussed in Section 10.3.5, the Melter slag sample that was wet polished using the Accelerated Polishing
Machine (APM) reduced the Stage 1 accelerated polishing from an initial value of DFT()=0.90 to an equilibrium
wet polished level of DFT()=0.71. The Stage 2 polishing phase included the accelerated polishing of the sample
(in wet, dry and damp conditions) with the addition of either oedometer clay, emery powder or Leighton Buzzard
276

Chapter 10
sand. Figure 10-21 shows the results of the two stages of polishing for the two Melter slag laboratory samples (one
with accelerated polishing and the other with no polishing). Initially, it appeared that the melter slag was performng
very well as it was retaining its high initial skid resistance measurement, with the addition of contaminants and
accelerated polishing. However, after the first cycle of Leighton Buzzard sand and emery powder, a trend of
increasing loss of measured skid resistance began. This decreasing trend continued, increasing at a faster rate which
raises some concerns in terms of the longevity of skid resistance of melter slag. After 10 minutes of accelerated
polishing with Leighton Buzzard sand, an increase of DFT()=0.08 occurred after a DFT() level of approximately
0.62 had been established following 340 minutes of accelerated polishing (as shown in Figure 10-21). Furthermore,
when emery powder was added, prior to 10 minutes of accelerated polishing, the DFT() reduced by 0.22 from the
peak increase (DFT()=0.73) gained with the Leighton Buzzard sand.

Melter Slag Aggregate DFT( ) Stage 1 and 2 Polishing

1.0
Unpolished Sample

0.9

DF Tester ( )

0.8
0.7
Polished Sample

0.6
0.5
0.4
0.3

Stage 1: Polishing to ESR

0.2

Stage 2: Polishing with Additives

0.1

See Figure 10-22

0.0
0

50

100

150

200

250

300

350

400

450

500

Time (mins)
S8 - Slag PSV 55 Polished Sample

S9 - Slag PSV 55 Unpolished Sample

Figure 10-21: Melter Slag Laboratory Sample (Stages 1 and 2 Polishing Phases)
These significant variations require some explanation sequentially in terms of what additive was used, the effect of
the additive in terms of the measured coefficient of friction with the DF Tester () and then what may have caused
the variation to occur. A diagrammatical explanation for the Melter slag sample is shown in Figure 10-22. A
chronological explanation and its effect in terms of measured coefficient of friction from the beginning of Stage 2
polishing follows:

The addition of 10 grams of Oedometer clay material that had been sieved and retained on a 1.15mm sieve
combined with dry accelerated polishing for 10 minutes did not significantly alter the DFT() value
previously obtained by wet polishing to an ESR level of 0.74;

The addition of an initial 3 x 10 grams of emery powder every 10 minutes of dry accelerated polishing
decreased the measured DFT() to 0.68;

The polished sample was then wet polished with no additives for 30 minutes and the subsequent measured
DFT() decreased further to 0.63;

The addition of a series of 2 x 20 grams of Leighton Buzzard sand to the sample every 10 minutes during
accelerated polishing, resulted in an increase in the measured DFT() to 0.71. The initial 10 minutes of
277

Chapter 10
polishing with the Leighton Buzzard sand produced a DFT() increase of 0.06 to 0.69; and the second 10
minutes resulted in a slight increase in DFT() of 0.02 to 0.71;

The next dry polishing phase comprised the addition of 3 x 10 grams of emery powder every 10 minutes
after the Leighton Buzzard sand accelerated polishing phase was completed. A measured DFT() was not
obtained until the end of the 30 minute period. This resulted in a decrease in measured skid resistance to
significantly lower levels than the recorded ESR level measured at the end of Stage 1. The value decreased
significantly by a DFT() of 0.20 resulting in a measured DFT() of 0.51;

A further 30 minutes of wet polishing with no contaminants was then undertaken which resulted in little
change in the measured DFT(), remaining significantly lower than the original wet polished ESR value
obtained at the end of the Stage 1 polishing phase of DFT()=0.74;

The full PSD of the oedometer clay was then added in 3 x 10 gram batches with the surface being very
lightly sprayed with water to keep the surface in a damp condition. The measured skid resistance level
after the first 10 minutes of accelerated polishing increased to a level of DFT()=0.58. Two further 10
minutes of accelerated polishing with additional skid resistance measurements did not significantly alter the
measured DFT() further;

A further 30 minutes of wet polishing with no contaminants added produced a characteristic decrease in
measured DFT() of 0.07 to 0.51;

A repeated cycle of 2 x 20 grams of Leighton Buzzard sand for 10 minutes each with accelerated polishing
produced an expected increase (as with the earlier cycle) in the measured DFT() value to 0.61, but not to
the same level of measured skid resistance DFT() as obtained in the first cycle;

A final 3 x 10 grams of emery powder every 10 minutes with dry accelerated polishing produced a
significant further reduction of DFT()=0.22 in measured DFT() value to 0.39.
Melter Slag Aggregate DFT( ) and Addition of Contaminants
(Stage 2 Polishing) over Time (mins)

0.9

2.5

0.8
2.0

Differences in DFT ( )

Three 10g of Emery Powder

for three 10 minutes

Wet Polishing for 15 minutes

Wet Polishing for 15 minutes

Two 20g of Leighton Buzz Sand for

two 10mins

1.5
Wet Polishing for 15 minutes

Wet polishing with 10g of Oedometer

Wet polishing with 10g of Oedometer Clay

Wet Polishing for 15

minutes

Wet Polishing for 15

minutes

Three 10g of Emery

Powder for three 10 minutes

Wet polishing with 10g of

Oedometer Clay

0.2

Added 20g of Leighton Buzz Sand

0.3

Wet Polishing for 30 minutes

0.4

Added Emery Powder

0.5

Added Emery Powder

Added Oedometer Clay - Sieve >1.15mm

DF Tester ( )

0.6

Added Emery Powder

0.7

1.0

0.5

0.1

270

280

260

250

240

230

220

210

200

180

190

170

160

150

140

130

120

110

100

90

80

70

60

50

40

30

20

10

0.0
0

Time (mins)
S8-Slag PSV 55 Polished

S9-Slag PSV 55 Unpolished

Difference

Figure 10-22: Melter Slag Laboratory Samples (Stage 2 Polishing with Addition of Contaminants)

278

Chapter 10

10.5.7 Summary of Stage 2 Accelerated Polishing with Contaminants

The effect of adding various contaminants (oedometer clay, Leighton Buzzard sand and emery powder) to the four
laboratory prepared samples, consequent with accelerated polishing, was measured by the Dynamic Friction Tester
(DFT()) device. This stage is named the Stage 2 polishing phase and the results are shown in Table 10-4.
Description of
Contaminants and
Accelerated Polishing

Moutohora
Greywacke
PSV 63
DFT()

%Diff1

Holcim
Basalt
PSV 52
DFT()

%Diff1

0.60

Otaika
Greywacke
PSV 52
DFT()

%Diff1

0.58

Melter
Slag
PSV 55
DFT()

%Diff1

Initial Skid Resistance DFT ()

0.87

0.90

ESR Polished DFT()

After Wet Polishing 2

0.47

46.0%

0.39

35.0%

0.42

27.6%

0.71

18.4%

Oedometer Clay DFT ()

Dry Polishing 3

0.58

-23.4%

0.41

-5.1%

0.43

-2.4%

0.72

-1.4%

Emery Powder DFT ()

Dry Polishing

0.55

5.2%

0.33

19.5%

0.36

16.3%

0.67

6.9%

Wet Polishing for 30mins

(no contaminants)

0.48

12.7%

0.32

3.0%

0.34

5.6%

0.63

6.0%

Leighton Buzzard sand DFT ()

Dry Polishing-Cycle 1

0.70

-45.8%

0.66

-106.3%

0.54

-8.8%

0.71

-12.7%

Emery Powder (3x10gms every

10mins) before DFT()-Cycle 1

0.49

30.0%

0.23

65.2%

0.26

51.9%

0.51

28.2%

Wet Polishing for 30mins

(no contaminants)

0.45

8.2%

0.26

-13.0%

0.26

0%

0.51

0%

Oedometer Clay DFT ()

Damp Polishing

0.67

-48.9%

0.41

-50.0%

0.48

-84.6%

0.58

-13.7%

Wet Polishing for 45 mins

(no contaminants)

0.60

10.5%

0.34

12.8%

0.43

10.4%

0.51

12.1%

Leighton Buzzard sand DFT ()

Dry Polishing-Cycle 2

0.67

-11.7%

0.44

-29.4%

0.45

-4.7%

0.61

-19.6%

Emery Powder (3x10gms every

10mins) before DFT()-Cycle 2

0.41

38.8%

0.22

50.0%

0.37

17.8%

0.39

36.1%

Notes:
1) The values in this column are the percentage difference in the result from the prior-measured coefficient of
friction condition result. This measures the change in the coefficient of friction DFT() due to the addition
of the contaminant under the action of the accelerated polishing.
2) End of Stage 1 Accelerated Polishing and skid resistance measurement DFT() results.
3) Beginning of Stage 2 Polishing and skid resistance measurement DFT() results.
Table 10-4: DFT() and % Difference in Results of the Stage 2 Addition of Contaminants and Accelerated
Polishing
The DFT() results of each polished laboratory samples for the value at which the DFT() levelled off and/or the
polishing time was (for pragmatic reasons) stopped, are shown graphically in Figure 10-23. Each progressive series
of bar charts given in Figure 10-23 shows the effects of the addition of contaminants and whether accelerated
polishing was undertaken in the dry, wet or damp conditions in the order shown in the bar charts (other trials
showing less effect are not shown; they are described in detail in the previous section).
279

Chapter 10

DFT ( ) of Laboratory Samples

(Addition of Contaminants - Stage 2 polishing)
1.0
0.90

Skid Resistance DFT ( )

0.9
0.8

0.78

0.74

0.74 0.73

0.7
0.58

0.6
0.5

0.71

0.66
0.60

0.58

0.52

0.49

0.46

0.42 0.43

0.39

0.4
0.3

0.51

0.43

0.26

0.23

0.2
0.1
0.0
Moutohora - Greywacke

Otaika - Greywacke

Holcim - Basalt

Melter Slag

Laboratory Samples
Initial DFT () Before polishing
Oedometer Clay DFT () Dry Polishing
Emery Powder DFT () Dry Polishing-Cycle 1

ESR Polished DFT () After Wet polishing

Leighton Buzzard Sand DFT () Dry Polishing-Cycle 1

Figure 10-23: Skid Resistance DFT () and the Effects of the Addition of Contaminants on the 4 Lab Samples
All surface aggregates were initially polished using the Accelerated Polishing Machine (APM) developed at the
University of Auckland (as shown in Figure 4-28) to an ESR level (Stage 1). They were periodically tested
throughout the polishing process for their skid resistance using the DF Tester DFT(). The Stage 2 polishing phase
also used the APM with the addition of contaminants and accelerated polishing, that resulted in significant changes
in the measured coefficient of friction for the four laboratory prepared samples. Figure 10-24 shows the real
changes in measured skid resistance in terms of DFT() and also the percentage difference in the result from that of
the initially tested surface aggregate, prior to any accelerated polishing. The data analysis and Figure 10-23 and
Figure 10-24 show the folowing for the four laboratory samples:

The three natural aggregates (whilst beginning at different levels) decreased in measured skid resistance to
an ESR level due only to wet polishing, by 28% to 41%. Only small differences pertained in the actual
DFT() values at ESR (0.39-0.47) between the three natural aggregate samples, even though their initial
values featured much greater differences;

The melter slag had the highest initial position value DFT()=0.90, this decreased by the smallest amount
(18%) to an ESR level of DFT()=0.71;

The addition of oedometer clay (soft and well graded to fine and coarse PSD) with dry accelerated
polishing, on the three natural aggregates, generally increased the measured skid resistance DFT() to a
small degree. The highest increase in DFT() was recorded for the Moutohora aggregate with an increase
of DFT() of 0.12. Very minor increases in DFT() were recorded for the Holcim basalt and Otaika
greywacke aggregates;

The addition of oedometer clay and dry accelerated polishing on the melter slag aggregate made little
change (although slightly reduced) of the skid resistance DFT() value;

The Leighton Buzzard sand (hard and coarse material) when added to the samples, with dry polishing with
the accelerated polishing machine, significantly increased the measured skid resistance DFT() on all three
natural aggregates.

Furthermore, the Leighton Buzzard sand raised the skid resistance DFT() to,
280

Chapter 10
significantly higher levels than the initially recorded value of ESR from wet polishing. The Moutohora and
Holcim aggregates, increased the skid resistance DFT() almost back to the initial measured values prior to
any polishing. The Otaika aggregate sample when polishing with Leighton Buzzard sand increased the
DFT() to a value that was higher than the initially tested value;

Accelerated dry polishing with Leighton Buzzard sand on the melter slag aggregate also increased the value
of measured skid resistance DFT(), but to a lower level than that reached at ESR;

The addition of emery powder (hard and fine) significantly decreased the measured skid resistance DFT()
on all laboratory samples from the prior level which existed after the Leighton Buzzard sand had coarsened
the microtexture. The Moutohora DFT() level decreased to approximately the same level as that obtained
at ESR, but no further. The measured skid resistance DFT() of the Otaika aggregate decreased to almost a
third of its measured skid resistance when roughened by the Leighton Buzzard sand and to a level
significantly lower than the ESR level reached with wet polishing. The Holcim basalt and the melter slag
aggregate DFT() level also decreased significantly with the addition of emery powder to a level
significantly lower than that reached at ESR;

Subsequent cycles of adding Leighton Buzzard sand and emery powder to the Moutohora, Holcim Basalt,
and the Otaika greywacke aggregates with accelerated polishing indicates that a historical memory of an
ESR level may exist with a natural aggregate, as the measured skid resistance DFT() oscillates up and
down around this value, but when wet polished the skid resistance DFT() tends towards returning to the
ESR level;

The results of continued cycles of the addition of contaminants and accelerated polishing on the melter slag
were surprising after this aggregate had initially performed well. The results from this research indicate
that once the crystalline structure is disrupted, an overall trend of decreasing measured skid resistance can
occur and that the melter slag does not recover from this level. The initial benefits of the melter slag under
on-going cyclical polishing with contaminants, resulted in the melter slag DFT() being reduced to a value
similar to that for the Moutohora greywacke aggregate (DFT()=0.39). It is currently unknown whether
this would continue to decrease further, although a linear extrapolation of the stage 2 results would indicate
so;

Polishing with emery powder produced the lowest recorded skid resistance value (DFT()) for all of the
laboratory sample aggregates.

The lowest recorded skid resistance DFT() value for each of the

aggregates, which included the second cycle of accelerated polishing with emery powder are:
Moutohora=0.41, Holcim=0.23,Otaika=0.26 and Melter slag=0.39).

281

DFT ( ) - Addition of Contaminants, (Stage 2 polishing) and Percentage

Differences of Laboratory Samples
2.0

100%

1.4

80%
60%
41%

37%

35%

28% 26%

26%

1.2

57%
28%
13%

5%

1.0
0.8
0.6

0.78

0.46

0.66

0.58
0.49

0.42 0.43

0.4
0.2

0.60

0.73

0.52

0.43

40%
20%
0%

0.74

0.39
0.23

21%

18% 19%

0.90

-14%

0.74
0.58

60%

43%

0.71
0.51

-20%
-40%

% Difference

Skid Resist. DFT ()

1.8
1.6

Chapter 10

-60%

0.26

-80%

0.0

-100%
Moutohora - Greywacke

Otaika - Greywacke

Holcim - Basalt

Melter Slag

Aggregate Samples (Polished)

Initial DFT () Before polishing

ESR Polished DFT () After Wet polishing

Oedometer Clay DFT () Dry Polishing

Emery Powder DFT () Dry Polishing

Leighton Buzzard Sand DFT () Damp Polishing

Wet Polishing % Difference

Oedometer Clay (Dry Polishing) % Difference

Emery Powder (Dry Polishing) % Difference

Leighton Buzzard Sand (Damp Polishing) % Difference

Figure 10-24: Skid Resistance DFT () and the Effects of the Addition of Contaminants and the % Difference
from the Initial SR value for the 4 Lab samples

10.5.8 Macro Photographs of the Effects of Polishing with Contaminants

Macro photographs were taken through a microscope of the same aggregate chip and position for each of the four
polished laboratory samples (or as close as was possible to the same position). Initially, a photograph was taken
immediately after the addition of 4 doses of 20 grams of Leighton Buzzard sand and dry accelerated polishing for 10
minutes for each dose. This position (Photo A) coincides with the time when the surface was rejuvenated in terms
of skid resistance and measured DFT() was high. The second photograph (Photo B) was taken after 30 minutes of
polishing with emery powder and coincides with the time when the surface was polished in terms of skid resistance
and when measured DFT() was low.
A comparison of a sample of the Moutohora greywacke aggregate after each state of polishing is shown in Figure
10-25 (a and b) Photos A and B. It is apparent that significant differences are evident in the microtexture of the
aggregate surface. Figure 10-25b is a lot less harsh than the rougher and uneven surface of Photo A shown in Figure
10-25a. As the Moutohora aggregate consists of a matrix of variable grain sizes and hardness of minerals, the
coarse and hard quartz minerals of the Leighton Buzzard sand (Photo A) have resulted in the grinding down and the
plucking out of grains from the surface, thereby producing a rough, harsh and uneven surface microtexture, leading
to a relatively high measured skid resistance of DFT()=0.74. The various grain sizes are much more rounded and
polished in Photo B (Figure 10-25b after the effect of polishing with emery powder, which is very fine and
comprises hard corundum minerals. It is clear that reasonable skid resistance can be attained on this polished
surface (Photo B) as confirmed by the skid resistance measurement of DFT()=0.49.

282

Chapter 10

(a)

Photo A After Leighton Buzzard sand polishing

(b)

Photo B - After Emery Powder polishing

Figure 10-25: Moutohora - After Polishing with Various Contaminants (approx. field of view 3.5x 2mm)
A comparison of a sample of the Holcm basalt aggregate chip photos is shown in Figure 10-26(a and b). It is
apparent that differences exist in the microtexture of the aggregate surface. Figure 10-26b is a lot more polished on
the top grain surfaces than the rougher Figure 10-26a surface in Photo A. The surface in Photo A has been ground
and scratched with the abrasive Leighton Buzzard sand leaving a harsher surface. However, the grain matrix has
stayed reasonably intact, as both the polishing contaminants (Leighton Buzzard sand) and the basalt minerals are
predominantly quartz minerals of similar hardness. The measured skid resistance after polishing with the Leighton
Buzzard sand (refer to Figure 10-26a - Photo A) is DFT()=0.52. This is a lot better than that shown in Figure
10-26b) - Photo B, but is still a lot lower than the Moutohora greywacke shown in Figure 10-25a. The polishing of
the basalt surface with emery powder has caused the highest protruding surface grains to begin to flow in some
cases, and therefore the measured skid resistance at this point is very low, DFT()=0.26.

(a)

Photo A - After Leighton Buzzard sand polishing

(b)

Photo B - After Emery Powder polishing

Figure 10-26: Holcim basalt - After Polishing with Various Contaminants (approx. field of view 3.5x 2mm))
A comparison of a sample of the Otaika greywacke aggregate chip photos is shown in Figure 10-27 (a and b). It can
be clearly seen that the Otaika grain size is much smaller and more uniform in size than the Moutohora aggregate
shown in Figure 10-25. The two photos show the same aggregate chip surface after polishing with Leighton
Buzzard sand (Photo A) and polishing with emery powder (Photo B). When comparing the two photographs
283

Chapter 10
significant differences can be seen in the microtexture of the aggregate surface where Photo A shows a more harsh
surface (although not very rough and uneven) and Photo B demonstrates severe polishing of the grain matrix with
the surface beginning to flow. The differences in the photographs can explain the large difference in the measured
skid resistance between the surfaces, although the Photo A skid resistance level is still significantly lower than the
Moutohora greywacke value after the same type and amount of polishing with Leighton Buzzard sand. This is due
to the less rough and uneven surface of the Otaika aggregate. The measured skid resistance of Figure 10-27a
Photo A after Leighton Buzzard sand polishing is DFT()=0.66 and for Figure 10-27b Photo B with the emery
powder, is DFT()=0.23.

(a)

Photo A - After Leighton Buzzard sand polishing

(b)

Photo B - After Emery Powder polishing

Figure 10-27: Otaika Greywacke - After Polishing with Various Contaminants (approx. field of view 3.5x
2mm))
A comparison of a sample of the Melter slag aggregate chip photos is shown in Figure 10-28 (a and b). There is
much less visual difference between the two photographs after polishing with Leighton Buzzard sand (Figure 10-28a
and with emery powder in Figure 10-28b compared with the other aggregate samples. More polishing of the grain
top surfaces in Photo B is apparent, although the overall surface is still very rough and harsh. This explains why the
measured skid resistance is still relatively high DFT()=0.51. As would be expected from a visual inspection of
Photo A, the skid resistance measurement after Leighton Buzzard sand, was relatively high in comparison to the
other natural aggregate samples DFT()=0.71, although this has reduced somewhat from the initial skid resistance
measurement that was obtained prior to any polishing, which resulted in a DFT()=0.90.
However, as discussed above, a second cycle of polishing with Leighton Buzzard sand and emery powder (after the
photos taken above) has shown a trend of continuing reduction in measured skid resistance for the melter slag
similar to that of the Moutohora greywacke aggregate DFT()= 0.39. This decreasing trend is of some concern and
it is currently unknown whether this would continue to decrease further, although extrapolating results would
indicate so.

284

Chapter 10

(a)

Photo A - After Leighton Buzzard sand polishing

(b)

Photo B - After Emery Powder polishing

Figure 10-28: Melter Slag - After Polishing with various contaminants (approx. field of view 3.5x 2mm))

10.5.9 A Geological Explanation of the Stage 2 Polishing Experiments

Black (2005a) considers that, in simple terms there are three processes that effect change in the measured skid
resistance of the laboratory samples:

Polishing surfaces with small soft material - in this case, the oedometer clay (kaolin with Moh s Hardness
(H)=2 to 2.5), which is softer than all the minerals in the samples and thus cannot scratch them. Clay
minerals are flexible, and when damp, form cohesive masses which would be likely to fill existing
depressions between asperities when polishing takes place in the damp (as distinct from wet) state. In wet
polishing however, the clay material will be washed out of depressions;

Scratching of surfaces by small hard grains - in this case with emery powder (H=9), which is harder than
all the minerals present in the samples; and

Scratching of surfaces by large hard grains - in this case by the Leighton Buzzard sand which is almost
pure quartz (H=7) and will scratch anything softer, forming deep scratches and removing much of the
work-hardened surface produced by polishing with the emery and oedometer clay.

The scratching (mechanical) effect can be seen clearly in the behaviour of the four samples when quartz sand
(Leighton Buzzard) is added to them. The greatest effect is seen in the Holcim basalt, composed dominantly of
feldspar, magnetite and pyroxene, which, with a Hardness range of 5 to 6, are all softer than quartz and thus will be
scratched by quartz (H=7). The minimum effect is seen in the Melter slag sample, which contains abundant spinel
that is harder than quartz, and therefore will not be scratched by quartz. Consequently, the application of Leighton
Buzzard quartz does not revive the surface, with results remaining lower than the original DFT() values.
The Moutohora aggregate starts with a very high DFT() or Polished Stone Value (PSV) and examination of the
aggregate surfaces shows very good relief, the quartz, feldspar and other grains standing out clearly on the surfaces
of the rock. However, probably because of the high feldspar content and the very low level of metamorphism (and
hence lithification), the measured skid resistance decreases rapidly with polishing and the final skid resistance for
the Moutohora aggregate is similar to that of the other natural aggregates.

285

Chapter 10
The Otaika greywacke, because of its relatively higher level of lithification and hardness, and higher quartz content
in the rock matrix, has a more uniform surface texture and a low initial measured skid resistance and consequent
PSV, and shows less change with polishing.
Surface memory: The grinding/polishing process induces a layer of compressive stress on the surface of the
samples resulting in work hardened surfaces. It is this change in the structure of the surface of the samples which
creates the memory effect seen in the experiments. While the application of coarse quartz grains will cause new
scratching and thus rejuvenate the surface to some degree, the work hardened surface clearly persists in areas
surrounding the new scratches and thus the skid resistance of the chips drops quickly to a level close to that which
existed before the application of the quartz sand.

10.6 Summary of Controlled Laboratory Experiments

The controlled laboratory experiments have successfully demonstrated that significant variations in skid resistance
can be simulated by polishing samples that were prepared in the laboratory. The results of a two-stage polishing
process that consisted of accelerated wet polishing to an equilibrium skid resistance level (Stage 1 polishing) and
the addition of various contaminant additives to the accelerated polishing process (Stage 2 polishing) demonstrated
results and degrees of variation similar to those shown at field sites with normal heavy traffic conditions.
A summary of the main findings of the controlled laboratory experiments follows under the following categories:
Polishing to an Equilibrium Level:

The accelerated wet polishing of aggregates (without any contaminant additions) can polish natural
aggregates to an equilibrium skid resistance level (ESR);

The level of polish achieved at ESR does not specifically relate to the Polished Stone Value (PSV) results.

The Addition of Contaminants with Accelerated Polishing:

Significant variations (both increases and decreases from ESR) in measured skid resistance can be
simulated in the laboratory by the addition of various contaminants;

The significant variation in measured skid resistance observed in the field is intrinsically related to the
geological properties of the aggregates themselves and the contaminants that end up in the macrotexture of
the surfacing.

Geological Properties of Aggregates (Grain Size and Hardness):

Contaminants that are fine in size (less than 10m) and that consist of hard minerals (e.g. emery powder)
polish the aggregate surfaces, thereby reducing the measured skid resistance;

The amount of decrease in measured skid resistance that is achieved on a specific aggregate depends upon
the difference in hardness and the geological makeup of the contaminants and the aggregate surface being
polished (the greater the difference, the greater the decrease in skid resistance);

Contaminants that are coarse (greater than 1mm) and consist of hard minerals (e.g. Leighton Buzzard sand)
abrade, grind and scour the aggregate surface, thereby increasing measured skid resistance;

The amount of increase in measured skid resistance that is achieved on a specific aggregate also depends
upon the difference in hardness and the geological makeup of the contaminants, as well as the aggregate
surface being polished (the greater the difference the greater the increase in skid resistance);

286

Chapter 10
Contaminants that are soft in mineral content in comparison to the surfacing aggregate (e.g. oedometer
clay), irrespective of particle size, have been shown to increase the measured skid resistance only slightly
when dry. However, when polishing occurs under the action of trafficking or accelerated polishing, in
damp conditions (although, not so wet that the contaminants are washed away) the measured skid
resistance can increase on aggregates that have a softer grain matrix and a variable particle size. This effect
remains unexplained.

287

Chapter 11

11 RESEARCH SUMMARY
11.1 Overview
The research reported here has highlighted that there are many independent variables that determine the skid
resistance of a section of road pavement at a particular time. Not all of these have been uncovered, at this point.
Skid resistance is not a stable quantity and this research has shown that it can change markedly in just a few hours,
the change depending not only on the surfacing aggregate petrology, but also on the degree of wetness, (heavy)
traffic, and the properties of the contaminant/detritus. Recognition of these factors will need to be strengthened in
the operational and management arena, and it may well be that skid resistance can be targeted only as a band or
range of values at any location, because of its inherent variability.

11.2 Research Aims and Objectives

The overall objective of the research detailed in this thesis is the advancement of understanding of the seasonal and
short-term variation of skid resistance. Better understanding will have significant implications for Road Controlling
Authorities that are often struggling to come to terms with which skid resistance policy to adopt or, in the case of
Transit New Zealand, how to best manage skid resistance within the context of their current TNZ T/10 skid
resistance policy. The research uses a combination of fieldwork and laboratory experiments to address the specific
aims of the research and to quantify the following:
1)

Confidence limits of measured skid resistance variation that includes seasonal and short-term variability
of skid resistance on various road surfaces;

2)

Variability between measurement devices (SCRIM, GripTester, and the Dynamic Friction Tester) on
various surfaces;

3)

Rate of deterioration of the measured skid resistance of newly laid chipsealed surfaces on straight sections
of road in comparison to corners, for various levels of traffic;

4)

Effects of temperature on the GripTester and the Dynamic Friction Tester results;

5)

Concentration and particle size distribution of the types of contaminants that accumulate on the road
surface and how they vary with time and rainfall;

6)

Effects of various washing treatments on the pavement surfacing to determine whether the presence of
detritus in itself reduces measured skid resistance;

7)

Effects of wet and dry accelerated polishing on prepared laboratory specimens;

8)

Effects of various contaminants on skid resistance due to accelerated polishing on prepared laboratory
specimens.

The specific hypotheses tested are:

1)

That skid resistance fluctuates in a band about an equilibrium level , once initial aggregate polishing has
been completed;

2)

That the relative confidence limits of the band about the equilibrium level are approximately 0.1 sfc units
or 0.15 GN units from upper to lower limits;

3)

That deterioration in measured skid resistance from a newly laid surface is significantly greater on curves
than on straight road sections and increases as the ratio of approach speed to the safe curve design speed
increases;

288

4)

Chapter 11
That the time period since the last rain has a significant effect on the short-term deterioration of measured
skid resistance, i.e. that the slope of the short term deterioration line is significantly different from zero and
approximately 0.01 sfc units (or equivalent GN) per day; and

5)

That the type and classification of the contaminants that accumulate on the road surface are significant in
determining the magnitude of the short-term variation of measured skid resistance.

The following sections provide the conclusions of the research in respect to the above research aims, objectives and
hypotheses.

11.3 Seasonal and Short-Term Variations of Measured Skid Resistance

11.3.1 Outline
Seven field sites were regularly tested with the GripTester over a two to three year period. Two sites were asphalt
mix surfaces in Auckland and five sites were chipseal surfaces on State Highways in the Northland Region.
Conclusions addressing the specific study aims and objectives with regard to the seasonal and short term variations
in measured skid resistance are addressed below.
Research Aim 1: Quantify the confidence limits of measured skid resistance variation that
include seasonal and short-term variability of skid resistance on various road surfaces.

11.3.2 Variations in the Mean Value

Auckland Asphalt Mix Sites
Two Auckland asphalt mix sites were measured approximately monthly for almost three years. The two sites
(Tamaki Campus site and the Ports of Auckland site), were considered to be acting in an equilibrium skid resistance
phase. The Ports of Auckland surface was over seven years old, and the Tamaki Campus surface approximately one
year old at the beginning of the data collection period. Both sites exhibited (especially in the second year of testing
the Tamaki Campus site) signs of acting in an equilibrium skid resistance phase. GripTester measurements from
both sites were normalised together and analysed. The results of the analysis demonstrated no clear noticeable or
predictable patterns of the variation of skid resistance measurement over time. This is partially due to the very
different and unpredictable trafficking actions at the two sites (heavy transverse trafficking at the Ports of Auckland
site and very light traffic at the Tamaki Campus site).
A summary statistical description of the skid resistance measurements at each of the sites, including the average
95% percentile confidence interval, the standard deviation ( ) and the coefficient of variation (CoV) of the site mean
measured skid resistance (x), for the two sites, and for the GripTester and Dynamic Friction Tester devices
respectively, are:

Tamaki Campus asphalt mix site using the GripTester (32 data points):

x 0.020 GN 30

CoV = 7.4%

Tamaki Campus asphalt mix site using the DF Tester (11 data points):

x 0.019 DFT 20

= 0.057
= 0.034

CoV = 4.8%

Ports of Auckland asphalt mix site using the GripTester (24 data points with heavy vehicle traffic
movements ):

x 0.028 GN 50

= 0.067

CoV = 10.9%
289

Chapter 11
Ports of Auckland asphalt mix site using the DF Tester (14 data points with heavy vehicle traffic
movements):

x 0.027 DFT 20

= 0.050

CoV = 9.5%

It can be concluded, that the mean skid resistance value for the asphalt mix sites remains relatively consistent over
time, after an equilibrium level has been establised, as the 95% confidence interval is less than 0.06GN between
sites.

Furthermore, the GripTester and the DF Tester device measurements are consistent in that they both

demonstrate that the Ports of Auckland site with heavy traffic conditions demonstrates a significantly greater 95%
confidence interval, standard deviation and coefficient of variation in comparison to the Tamaki Camus site with
light traffic.
Northland State Highway Chipseal Sites
Skid resistance measurements were undertaken approximately monthly for over two years using the GripTester
device on five chipseal sites in Northland, and the Dynamic Friction Tester when possible on the Hikurangi site.
Two of the Northland two-lane State highway chip seal surface sites, due to the age of the seal, were expected to be
acting within an equilibrium skid resistance phase (i.e. Hikurangi site and the Snooks Tatton - Section 2 site). The
HCV traffic volumes for the Hikurangi and Snooks Tatton site were approximately 950 vpd and 300 vpd,
respectively. The Hikurangi and Snooks Tatton site skid resistance results were normalised together with stable
30m timeslice sections from the Kaiwaka, Brynderwyn and Kara Road sites. The normalised results were analysed
and no distinct seasonal patterns were visually apparent. The average 95% percentile confidence interval, the
standard deviation ( ) and the coefficient of variation (CoV) of the mean value of the Hikurangi and Snooks Tatton
(Section 2) sites over the data collection period were:

Hikurangi chipseal site using the GripTester (32 data points):

x 0.024 GN 50

CoV = 12.4%

Hikurangi chipseal site using the DF Tester (8 data points):

x 0.048 DFT 20

= 0.068
= 0.069

CoV = 11.0%

Snooks Tatton Section 2 chipseal site using the GripTester (24 data points with a HCV traffic loading that
is approximately one third of the Hikurangi site):

x 0.019 GN 50

= 0.046

CoV = 8.2%

When the sites that were known to be in the initial polishing phase (due to a newly laid seal; i.e. Kaiwaka Slag,
Brynderwyn, Kara Road and Snooks Tatton Sections 1 and 3 sites) were analysed, and compared against those that
were supposed to be acting in an equilibrium level phase , there was very little difference in the measured 95%
confidence intervals of the site mean. This was surprising. It was initially expected that the extents of the variation
would be less for surfaces that had reached an equlibrium level of polishing (experiencing only seasonal and or
short-term variations) in comparison to sites that were still in the initial polishing phase from a new surface level.
However, the results have not shown this. In fact, the results show that in terms of the 95% confidence interval, the
standard deviation ( ) and the coefficient of variation (CoV) of the mean value, the reverse has occurred. As shown
below, the greater variation occurs on the sites that are supposedly at a stable equilbrium level of polishing.

Kaiwaka slag chipseal site using the GripTester and sealed with an iron making melter slag (28 data
points):

290

x 0.021 GN 50

Chapter 11

CoV = 7.9%

Brynderwyn South curve chipseal site using the GripTester (29 data points):

x 0.026 GN 50

= 0.056
= 0.070

CoV = 11.7%

Kara Road S1 and S2 chipseal sites using the GripTester. HCV traffic loading is approximately one third
of the Kaiwaka, Brynderwyn and Hikurangi sites (26 data points) respectively for S1 and S2 sections:

x 0.016, 0.018 GN 50

= 0.043, 0.045

CoV = 6.5%, 6.9%

Snooks Tatton S1 and S3 chipseal sites using the GripTester. HCV traffic loading is approximately one
third of the Kaiwaka, Brynderwyn and Hikurangi sites (24 data points) respectively for S1 and S3 sections:

x 0.019, 0.018 GN 50

= 0.047, 0.046

CoV = 6.9%, 6.5%

It can be concluded that the mean measured skid resistance result (GN50) for the Northland chipseal sites remain
relatively consistent over time, as does the standard deviation ( ). This result is true both after an equilibrium level
of skid resistance has been established, and during the initial polishing phases. The 95% confidence interval for the
site mean is less than 0.05GN when the sites have high HCV content (SH 1 sites with approximately 900-950 HCVs
per day) and less than 0.04 GN between sites with approximately one third of the HCV content (SH 14 sites with
approximately 300 HCVs per day). The only exception is the Kaiwaka melter slag site on SH1 that performed better
and more consistently than the other SH1 sites with similar HCV volumes, its performance being similar to the sites
with one third of the HCV volumes.
These conclusions must, however, be put into context with the aggregates that have been used in the field and their
resistance to polishing. As has been shown in the controlled laboratory experiments, the lower PSV aggregates
(such as the Otaika aggregate which is the aggregate used on all of the Northland field sites except the Kaiwaka
melter slag site) lose a lesser percentage of their initial value of skid resistance than those aggregates with a higher
PSV (such as the Moutohora aggregate). Furthermore, the lower PSV aggregates can vary significantly with the
addition of contaminants and can therefore mask any variation expected in the initial polishing phases.
One can therefore conclude from the field results of the five Northland chipseal sites and, to some extent, the two
Auckland asphalt mix sites, that the variation in the measured skid resistance as described by the 95% confidence
interval range, the standard deviation ( ) and the coefficient of variation (CoV) increases with increasing percentage
of heavy commercial vehicles irrespective of the polishing phase of the aggregate.
The following two hypotheses were developed as part of the experimental design:
Hypothesis 1: That skid resistance fluctuates in a band about an equilibrium level , once
initial aggregate polishing has been completed.
The above analysis has demonstrated that this hypothesis is true.
Hypothesis 2: That the relative confidence limits of the band about the equilibrium level are
approximately 0.1 sfc units or 0.15 GN units from upper to lower limits.
The analysis has shown that this hypothesis is not true as the confidence interval around the true mean site value to
the 95% probability level is surprisingly small, being approximately 0.06 GN for natural surfacing aggregates in the
field. However, the variation shown from calendar month to month (discussed below) demonstrates that significant
variation does exist on a monthly basis and also that distinct seasonal patterns can and do occur.

291

11.3.3 Seasonal Variations by Calendar Months of Year

Chapter 11

Auckland Asphalt Mix Sites

An analysis of the normalised GN results for the Auckland asphalt mix sites (Ports of Auckland and Tamaki
Campus sites) by calendar month of the year resulted in the lowest monthly results in February (95% normalised
GripNumber confidence interval from 0.92 to 0.96) and the highest monthly results in August (95% normalised
GripNumber confidence interval from 1.03 and 1.07). These two monthly periods are separated by six months, with
the lowest measured values in mid to late summer and the highest mean measured values in August (late winter)
with a 0.12 GN difference in the mean result range.
The variations between other months on the Auckland asphalt mix sites had no distinct patterns. The months either
side of February and August were highly variable and the data demonstrated no confidence at the 95% level that the
result in those months would be higher or lower than the normalised mean monthly data. The within month
normalised NGN value (the difference between the maximum and minimum normalised monthly result) ranged from
0.06 GN (recorded in February) to 0.27GN (recorded in June), with an average of 0.18 GN for each month. This
clearly demonstrates the significant and largely unpredictable variation that can occur in measured skid resistance
within months of the year. Interestingly, the months of February (the lowest monthly mean result) and August (the
highest monthly mean result) had the smallest range of result, 0.06 GN and 0.08 GN respectively. This demonstrates
that for these two sites, there is some confidence that a measured coefficient of friction during those months would
be close to the minimum or to the maximum result, respectively. However, it should also be noted that these two
months had only six and ten data points, compared to others that had up to 22 data points.
Northland State Highway Chipseal Sites
An analysis of the normalised NGN results for the Northland chipseal sites by calendar month of the year resulted in
a much more distinct seasonal pattern. The lowest mean monthly measured values occurred in April (late summer),
with a 95% confidence interval from 0.94 and 0.98 (an average of 0.96) and the highest mean measured values in
August (late winter) with a 95% confidence interval between 1.02 and 1.06. The difference in the mean monthly
result between February and August was 0.08 GN. Once again, the results either side of these months proved to be
much more variable.
The within month normalised NGN range for the Northland chipseal sites (the difference between the maximum and
minimum normalised monthly result) ranged from 0.07 GN (recorded in August) to 0.29 GN (recorded in May),
with an average normalised range of 0.17 GN. This clearly demonstrates the significant and largely unpredictable
variation that can occur in measured skid resistance within months of the year. The months of February, August
(the highest monthly mean result) and December had the smallest range of monthly results: 0.07, 0.09 and 0.09 GN,
respectively. The normalised mean monthly results of the Northland chipseal sites also indicate a sinusoidal pattern,
although unexpectedly, with a period of approximately 6 months rather than the expected 12 month period.
However, the pattern is indicative only and is thought to be partly due to the inherent and somewhat variable nature
of skid resistance and its measurement, and perhaps unusual rainfall events in the summer data collection periods.
More data would be required to confirm whether these findings are repeatable.
The data analysis has clearly shown that for Northland chipseal sites, testing in the month of April would produce a
monthly result that is, at better than 95% probability, significantly lower than the 12 monthly average for the site.
Similarly, a skid resistance measurement obtained in August would have a greater than 95% probability of being
significantly higher than the 12 monthly average for the site.
292

Chapter 11
It can be concluded from this research that whilst there is a clear seasonal variation in measured skid resistance (i.e.
lower results occur in late summer and higher results occur in late winter), there are still significant variations that
occur within calendar months of the year. This indicates that there are other short-term variable factors that need to
be taken into account. The conclusions from this aspect of the research follow.

11.3.4 Short-Term Variations related to Rainfall

The research considered that rainfall and detritus accumulation are significant factors in the variation of the skid
resistance measurements. Rainfall records were kept on all of the sites and these were analysed with the measured
GripTester results. Three rainfall functions were considered:

Days Since Last Rainfall (DSLR) > x mm, where rainfall greater than x =1, 2 and 5mm was considered;

Dry Spell Factor (DSF) where 1, 2 and 5mm of rainfall was considered; and

Weighted Rain Function (WRF) over a period of 5, 7, 10 or 20 days.

These rainfall functions were analysed against the measured coefficients of friction for both the GripTester and
SCRIM device. The number of data points was significantly less for the SCRIM device.
The GripTester device measurements were analysed for the Hikurangi control site, a site that was clearly at an
equilibrium skid resistance level. The results demonstrated strongest positively sloped linear correlation with
WRF for a period of either 7 or 10 days rainfall. The results were statistically analysed and the WRF function was
able to explain up to 60% of the variation of the 2004/2005 measured skid resistance results when the data were split
into annual groups. The slope of the equation line demonstrated an approximate increase of 0.003 x WRF (mm)
over a 7 day rainfall period prior to a measured skid resistance result. Conversely, the 2003 measured GripNumber
data did not demonstrate a positively sloped equation, as the results were inexplicably more variable. The DSLR
and DSF rain functions explained little of the measured variation of the skid resistance as measured by the
GripTester device.
The SCRIM device results on the Hikurangi site (in contrast to the GripTester results) indicated that a negative trend
of decreasing measured skid resistance existed, with increasing dry spells prior to a test measurement, for both the
DSLR and DSF rain functions. The result of the negative logarithmic DSF rain function analysis were better than
the DSLR rain function and much better than the GripTester correlations with these rain functions. A coefficient of
determination (R2) of 0.25 and 0.53 was obtained for the increasing and decreasing directions respectively. The best
DSF correlation occurred when >2 mm rainfall depth and a previous rainfall period of 7 days was used. The slope
of the equation in terms of the SCRIM coefficient of friction (sfc) demonstrated an approximate decrease of 0.02 x
DSF >2mm over a 7 day rainfall period prior to a measured skid resistance result. However, the number of data
points was significantly less than that of the GripTester measurements.
The WRF function, when correlated against the SCRIM device measurements produced a lower coefficient of
determination (R2) result (maximum of 0.37) in comparison to the DSF function, and the GripTester (GN)
correlations for the 2004 / 2005 skid resistance measurements.
The SCRIM results however, compared well with the results of Hill and Henry (1981), who concluded that the DSF
rain function was better than the WRF. This seeming anomaly between the results of the SCRIM machine and the
GripTester results may be related to the testing device method and especially the differences in percent slippage of
the devices (the GripTester=15%, the SCRIM=34% and the Locked Wheel Tester used by Hill and Henry
(1981)=100%).

293

Chapter 11
When the normalised results for various Northland sites or timeslice sections were combined, the coefficients of
determination (R2) became significantly lower than when the Hikurangi site alone was used. At best, only 5.5% of
the measured variation could be explained by any of the rain functions with the normalised data.
It is concluded that other localised site and/or environmental factors also affect the measured variation. The other
factors are expected to be:

Temperature;

Localised contaminants / detritus;

Geological makeup of the aggregates; and

Traffic loading conditions.

Whilst the coefficients of determination are low for the normalised NGN results, the expected slope of the equations
was, as expected, negative for the DSLR and DSF and positive for the WRF rain functions. The slopes of these
equations are expected to depend upon the traffic loading, the susceptibility of the aggregate to polishing and the
polishing agents / contaminants.
In summary, a hypothesis was developed as part of the experimental design stating:
Hypothesis 4: That the time period since the last rain has a significant effect on the short-term
deterioration of measured skid resistance, i.e. that the slope of the short term deterioration line
is significantly different from zero and of the magnitude of approximately 0.01 sfc units (or
equivalent GN) per day.
The rainfall function analysis has indicated that the hypothesis is true, because the WRF explained up to 60% of the
variation in measured skid resistance with the GripTester device at the Hikurangi site for the 2004/2005 data.
However, whilst the slope of the trend could be confirmed when the five Northland sites were normalised, only
5.5% of the variation was explained due to the WRF alone. When analysing the SCRIM results, the DSF rainfall
function explained up to 53% of the variation. The analysis indicated that a prior rainfall history of 7 days gave the
best results and an indication of the slope of the equation was:

an increase of 0.003 x WRF (mm) over a 7 day rainfall period in terms of the GripTester device; and

a decrease of 0.02 x DSF >2mm over a 7 day rainfall period in terms of the SCRIM device.

However, there are other factors that need to be taken into account in the analysis, as the prior rainfall history cannot
fully explain the variation in measured skid resistance.

11.4 Skid Testing Devices

11.4.1 Outline
In this research three skid resistance measurement devices were used to monitor the variation in skid resistance on
the field sites and within the laboratory. Two were CFME devices (SCRIM and GripTester) and the third was a
static device (DF Tester). Conclusions addressing the specific study aims and objectives with regard to the skid
resistance devices are addressed below.
Research Aim 2: Quantify the variability between measurement devices (SCRIM, GripTester,
and the Dynamic Friction Tester) on various surfaces.
Research Aim 4: Quantify the effects of temperature on the GripTester and the Dynamic
Friction Tester results.
294

11.4.2 Comparative Device Conclusions

Chapter 11

Comparing the results of one skid resistance device with those of another is a difficult and complex task as the real
value of skid resistance for a given section is unknown. Although all devices report a coefficient of friction (CoF)
the task of determining which devices are measuring better than others is difficult, as there are many factors and
variables that need to be considered. These include:

The methodology and operating mode of the device;

The sensitivity of the device to temperature and other environmental factors such as detritus;

The rubber/surface contact area and the differences between devices in vertical loads, slip speed, compound
and hardness of the rubber, the tyre pressure (if applicable) and size of measuring wheel or rubber slider;
and

The road surface age and macrotexture.

Conclusions from this research with regard to the GripTester device are:

The CoV of measurements recorded by the GripTester device on a new, smooth and clean chipseal surface,
with good repeatability in the transverse wheel path, can be as low as 1.2%;

The repeatability of the GripTester, in terms of the measured CoV between multiple runs on the same day,
on normally trafficked highway chipseal sections, varies throughout the calendar year and with the age of
the seal;

The CoV of the GripTester is least in the winter months of the year (June, July and August) with an
approximate mean CoV of 3.0% and is greatest in the summer months (December, January and February)
with a respective mean CoV of 5.0%;

An environmental effect exists which is related to the contaminants / cleanness of the surface, which affects
the measured run-to-run variability of the GripTester device. The observed variation is not merely a
function of the device itself;

The variation of CoV of the GripTester by month of the year is not due to the variation in temperature;

The GripTester is relatively insensitive to variation in temperature as long as the tyre has been properly
conditioned prior to testing; and

Indications are that the CoV of the GripTester device increases at sites that have increasing volumes of
HCVs.

Conclusions from this research with regard to the Dynamic Friction Tester (DF Tester) device are:

The mean CoV of the DF Tester for asphalt mix surfaces is 1.6%. The device is therefore more repeatable
than the GripTester. This is partly due to the device being static, so that differences in terms of the location
referencing are removed;

In terms of the repeatability of the DF Tester, the measured Coefficient of Variation (CoV) between
multiple test runs on typical Auckland asphalt mix sites does not vary by calendar month of the year;

The operation of the DF Tester with multiple spins on the same diameter track largely removes any
environmental effects which can affect other devices such as the GripTester, ROAR and SCRIM device
which perform one run over the surface for a test result;

The DF Tester-measured coefficient of friction (CoF) is insensitive to changes in surface, water and/or
ambient temperature between the ranges of 0 and 50 Celsius;

New rubber sliders on the DF Tester need to be conditioned with approximately two test runs to roughen
the new rubber surfaces, as initial test runs give an artificially low test result.
295

Chapter 11

11.4.3 The Correlation of Devices

Conclusions from this research with regard to the correlation of various skid resistance measurement devices are:

When correlating the GripTester device with the DF Tester device on an asphalt mix surface in like modes
(i.e. with the GripTester reporting in 1m mode and the DF Tester undertaking spot tests every 1m), a high
coefficient of determination (R2=0.79) is achieved, with the slope of the equation being very close to 1.0
(35 data points);

When correlations are undertaken with the GripTester in normal 10m rolling average and averaged over
normal chipseal section lengths in comparison to three test locations with the DF Tester, the coefficient of
determination reduces to approximately R2 =0.50;

On newly laid homogeneous surfaces that were specifically constructed for correlation trials in
Christchurch, with varying levels of microtexture and macrotexture, a very high coefficient of
determination, R2=0.98, was obtained between the GripTester and DF Tester devices, with the slope of the
equation also being very close to 1.0;

The results of the Christchurch correlation trials and a comparison between the results of the Dynamic
Friction Tester and the ROAR device show a good coefficient of determination of R2=0.72;

The results of the Christchurch correlation trials and a comparison between the results of the Dynamic
Friction Tester and the Portable British Pendulum Tester (PBPT) also show a good coefficient of
determination of R2=0.75. This increased to 0.98 when the grooved AC16 surface that adversely affected
the PBPT device was removed from the analysis;

A correlation between the SCRIM device and the GripTester device on the Northland sites (as close as
practically possible in terms of time, and mostly within the same day) resulted in a coefficient of
determination of R2=0.45. The regression equation however, did not compare well with the relationship
developed by Cenek and Jamieson (2000). Nevertheless, the range of the measured skid resistance results
in this research was relatively low, and therefore the regression equation could significantly change, with
more data at the high and low ends of measured skid resistance levels;

A correlation between the SCRIM device and the GripTester device for the Christchurch correlation trials,
under much more controlled conditions, more homogeneous surface conditions, various surface types and a
wider range of measured skid resistance, also resulted in an average coefficient of determination of
R2=0.46;

The Christchurch correlation trials also clearly showed that the SCRIM device is a lot less sensitive to
changes in skid resistance than the GripTester;

The relatively sensitive GripTester is more adversely affected by road roughness than the SCRIM device;

Harmonising the results from various devices to the International Friction Index (IFI F60 and Sp), whilst
desirable in theory, as it incorporates macrotexture effects on measured skid resistance, did not practically
improve the coefficients of determination (R2) determined in this research.

Adjustments to the

transformation coefficients, especially for the SCRIM device and the GripTester for New Zealand
conditions, may help to reduce the observed variation between the devices, although a much larger
correlation data set would be required than obtained from the Christchurch trials.
In summary, a number of international and local correlation equations have been developed for the commonly used
devices in New Zealand from previous correlation trials and research e.g.:

the PIARC International Harmonisation Experiment reported by Wambold et al. (1995);

296

Chapter 11

New Zealand research by Cenek and Jamieson (2000); and

Australasian correlation exercise reported by Dardano (2005).

This research has shown that, on a New Zealand highway network, these equations do not explain the significant
and real variations that can occur between devices. Additional research is required to more predictably determine
changes in measured skid resistance on the highway and to enable better correlation between measuring devices.

11.5 Timeslice Analysis and Influence of Geometry on Geometric Elements

Research Aim 3: Quantify the rate of deterioration of the measured skid resistance of newly
laid chipsealed surfaces on straight sections of road in comparison to corners, for various
levels of traffic;
By undertaking timeslice analyses of carefully chosen chipseal sections on the Northland state highway sites (rather
than taking the average seal extent result for geometrically different sections), a better understanding of how
different seal sections perform over time can be developed. Timeslice analyses of various geometric sections of the
Kara Road field were undertaken with straights and with curve radii that ranged down to 150 to 160m. The
following conclusions are made with regard to this site and specific geometric elements:

Significant negative logarithmic deterioration profiles occur when comparing adjacent 30m timeslice
sections as a ratio of one measured section skid resistance value to another for various geometrical features
(R2 range from 0.54 to 0.90 for 24 data points);

Initially measured skid resistance (GN) was improved by resealing with predominantly local aggregates
(Otaika aggregate) up to 20% above values for the adjacent older seals;

Skid resistance for a newly surfaced straight section lasted for approximately 13 to 17 months before
decreasing below adjacent older seal sections and then appeared to level off some 5 to 10% below the
adjacent older seal sections;

The rate of deterioration on the straight sections, between a new and old surface, ranged between
-0.08 to -0.10 Ln(x) - where x is the number of days since the new seal construction;

The rate of deterioration of newly sealed curved sections of radii in the range of 150m to 160m was faster
than new straight sections deterioration levelled off at 5 to 15% below the straight sections after 12 to 16
months;

The rate of deterioration between a new straight section and a new curved section ranged between
-0.021 and -0.073 Ln(x) - where x is the number of days since the new seal construction;

A comparison of a level section of road with a 6% downgrade did not demonstrate any marked difference
between the measured skid resistance of the two sections.

In summary, a hypothesis was developed as part of the experimental design stating:

Hypothesis 3: That deterioration in measured skid resistance from a newly laid surface is
significantly greater on curves than on straight road sections and increases as the ratio of
approach speed to the safe curve design speed increases.
The timeslice analysis of various 30m sections on the Kara Road site in Northland and other sites has indicated this
to be true. However, further research is required to determine whether the rate of deterioration is related to an
increasing ratio of the approach speed to the safe curve design speed.

297

11.6 Washing Trials

Chapter 11

Research Aim 6: Quantify the effects of various washing treatments on the pavement
surfacing, to determine whether detritus in itself reduces measured skid resistance.
A washing trial was undertaken on the Hikurangi site in Northland using three methods; gentle washing, washing
and brooming, and waterblasting at relatively low pressure using the Frimokar NZ Ltd hydrotexturising machine.
The washing trial revealed some interesting, although inconclusive results, regarding the effectiveness of a range of
washing treatments that can be applied for improving skid resistance levels. The conclusions of the washing trials
are:

None of the washing treatments produced any improvement as measured by the GripTester device;

The British Pendulum Tester results also showed no improvement between the benchmarked levels and the
post-washed treated sections on all sections, except for the Frimokar treatment section;

After the Frimokar treatment, the British Pendulum Tester indicated an improvement in skid resistance of 5
BPN although this improvement was very temporary, as a further test a short time later showed it had
reduced improvement to 2 BPN;

The relatively clean state of the washed pavement negated some of the expected increase in skid resistance.

Washing trials were also undertaken with the Dynamic Friction Tester, on surfaces that were initially tested and then
retested after waterblasting. The results from the Tamaki Campus, Ports of Auckland and Hikurangi sites (and after
significant dry periods), also did not show marked improvements in the measured skid resistance results. It is
hypothesised that washing of the surface does not improve the measured skid resistance, although the combined
effect of the pavement surface being wet with the kneading and abrasion effect of traffic loads on the wet surface
does help to re-juvenate the surface in terms of measured skid resistance. This hypothesis was tested as part of the
accelerated polishing of laboratory prepared samples. The conclusions of this analysis are discussed in Section 11.8.

11.7 Samples of Detritus Collected off the Road

Research Aim 5: Quantify the concentration and particle size distribution of the types of
contaminants that accumulate on the road surface and how they vary with time and rainfall.
In this research, 29 samples of detritus were collected from seven field sites between January 2004 and May 2005.
The samples were analysed for sediment load and particle distribution, heavy metals, TPH and TOC. Conclusions
from the sample analysis and results are:

Sediment loads ranged from 1.5 to 14.4 mg/m2 depending upon local conditions (e.g., traffic volume,
vehicle composition and site setting) and environmental factors (e.g. rainfall and wind speed);

Samples collected during the same sampling event showed little variability, whereas samples collected at
the same site on different days varied significantly;

The sediment size distribution remained relatively similar, with particles ranging from 0.001 to 1 mm, at
different locations and time of sampling;

The mean particle size d(0.5) was 21.5 5.8 m, while the average d(0.1) was 5.7 1.6 m. These results
give somewhat finer distributions than those reported for stormwater sediments researched elsewhere;

All metals collected from all sites, excluding one exception (the Ports of Auckland site), had low metal
concentrations (Cd 1-4 g/L, Cu 0.4-1 mg/L, Pb 0.3-2 mg/L, and Zn 1-3 mg/L);

298

Chapter 11
The samples collected from the Ports of Auckland site reported both 17-fold higher levels of Zinc and 8fold higher levels of Lead over and above the average results observed for the other sites. This is
understandable given the very heavy industrial traffic nature of this site;

The majority of the Copper, Zinc and Lead measured in the samples appear to be sediment-bound while
Cadmium was primarily dissolved; and

The TPH concentrations were close to the detection limit of the analysis (i.e. 10 mg/L). Although little or
no TPH could be detected in the samples, TOC values suggest the presence of non-TPH organic matter,
primarily in the form of particulate matter.

Previous research shows that rainfall volume appears to be the major factor in the removal of total road dust detritus,
and rainfall frequency mainly affects the accumulation of particle-bound contaminants (O'
Riley et al., 2002).
Unfortunately, due to the limitations in being able to sample the road surface based detritus, insufficient samples
were obtained at the same sites to enable a proper assessment of any relationship between road based sediment and
rainfall volume and frequency.
In summary, the analysis of the detritus results demonstrated that the variation of sediment and pollutants was site
specific and no differences between predominantly urban and rural settings could be detected other than for very
heavy industrial sites. The analysis also demonstrated that:

Samples collected at the same time and same location show relatively little variability;

Samples collected at the same location but at different times could vary significantly, indicating that site
specific environmental factors are significant and could be highly correlated with the variation in measured
skid resistance;

In comparison to previous research, the particle size distribution of the collected samples is somewhat finer,
but very consistent for all samples and sites, and provides confidence in determining particle size
distributions for the controlled laboratory analysis;

In comparison to previous research, the sediment contained low concentrations of heavy metals except for
the Ports of Auckland where (as would be expected) significantly higher concentrations were found. The
relatively low concentrations of metals on the highway sites indicate that the heavy metal concentration
does not play a significant role in the variation in measured skid resistance. However, more data are
required to confirm this preliminary indication.

11.8 Accelerated Polishing Experiments

11.8.1 Outline
Controlled laboratory experiments were required to simulate the observed variation of skid resistance in the field on
prepared samples.

The experiments could then investigate the effects of accelerated polishing, rainfall and

contaminants on measured skid resistance. The experiments were undertaken in two stages. The first stage
consisted of accelerated wet polishing to an equilibrium skid resistance (ESR) level (Stage 1 polishing). The
second stage incorporated the addition of various contaminant to the accelerated polishing process (Stage 2
polishing). The results have demonstrated similar results and degrees of variation to those shown at field sites with
normal heavy traffic conditions.
Conclusions addressing the specific study aims and objectives (outlined below) with regard to the laboratory based
accelerated polishing experiments are:

299

Chapter 11
Research Aim 7: Quantify the effects of wet and dry accelerated polishing on prepared
laboratory specimens.
Research Aim 8: Quantify the effects of various contaminants on skid resistance due to
accelerated polishing on prepared laboratory specimens.

11.8.2 Stage 1 Accelerated Polishing

The conclusions regarding the Stage 1 accelerated polishing of the three natural aggregate samples and the melter
slag artificial aggregate using the machine developed at the University of Auckland are:

The actual PSV of the aggregate sample generally predicts the ranking order of the initial level of skid
resistance of natural aggregates prior to any accelerated polishing;

The percentage reduction in the measured skid resistance from the initial level of skid resistance to ESR
reduces as the aggregate PSV reduces (from 46% for the Moutohora PSV 63 to 24% for the Otaika PSV
52);

Generally, the lower the PSV of the aggregate, the faster the aggregate polishes to its ESR level (from 7
hours of accelerated polishing for Moutohora PSV 63 to 2.5 hours for Otaika PSV of 52);

There is very little difference in the final level of ESR as measured by the DF Tester obtained for the three
natural aggregate samples (DFT()=0.47 for Moutohora PSV 63, DFT()=0.39 for Holcim basalt PSV 52
and DFT()=0.43 for Otaika PSV 52);

The greywacke sandstone and the basalt aggregates polish by different mechanisms and therefore the
deterioration rates vary under the same accelerated polishing loads; and

The time to a polished state and the ranking order of the final level of ESR for the natural samples may not
be the same as the PSV ranking (the basalt sample polished to a greater extent than the Otaika greywackes,
even though similar PSVs were measured).

The results of the laboratory accelerated polishing and DF Tester friction tests on the artificial melter slag initially
demonstrated very promising results as that material did not deteriorate at the same rate or in the same manner as the
natural aggregates. The main conclusions of the melter slag aggregate for the Stage 1 polishing to ESR are:

Whilst the actual tested PSV of the melter slag was lower (PSV=55) than that for the Moutohora natural
aggregate (PSV=63), the initial level of measured skid resistance DFT()=0.90 was very similar;

The percentage reduction in measured skid resistance from the initial level of skid resistance to ESR is
significantly less for the melter slag than for the natural aggregates of higher or lower measured PSV (16%
c.f. 46%, 39% and 24% for the Moutohora, Holcim and Otaika aggregates);

The time to reach its equilibrium polishing level is approximately the same as those for natural aggregates
of similar PSV (4-4.5 hours);

The melter slag significantly out-performed all of the natural aggregates in terms of being resistant to
polishing, including the Moutohora (PSV=63), which had a significantly greater PSV for the Stage 1
polishing phase.

The Kaiwaka slag site in Northland was resurfaced with the Melter slag material and used as a field test site. The
field based skid resistance measurements demonstrated similar long-lasting performance as the Stage 1 controlled
laboratory based polishing. The rate of decrease even in high stressed TNZ Site Category 2 corners is significantly
lower over the data collection period than field sites that were resealed with natural aggregates.

300

Chapter 11

11.8.3 Stage 2 Accelerated Polishing

The Stage 2 polishing phase included extended accelerated polishing with the addition of contaminants (oedometer
clay, Leighton Buzzard sand and emery powder) on the four laboratory prepared samples. The effects of the
addition of contaminants and accelerated polishing were measured by the Dynamic Friction Tester and by
comparing micro photographs of the surface of the aggregates. Significant differences were measured and observed,
proving that the methodology and equipment that was developed achieved its intended purpose.

Specific

conclusions from the Stage 2 polishing and subsequent analysis are:

The addition of Oedometer clay (soft and well graded to fine and coarse PSD) in whatever PSD and with
dry polishing, generally increased the measured CoF to a small degree. The highest increase with the
addition of oedometer clay was recorded for the Moutohora aggregate with a DFT() of 0.12. Very minor
increases were recorded for the Holcim basalt and Otaika greywacke aggregates;

The addition of Oedometer clay and dry accelerated polishing made little change to the results for the
Melter slag aggregate. Slight decreases in the measured CoF were observed in some cases;

The Leighton Buzzard sand (hard and coarse material), when added to the sample and dry polished with the
Accelerated Polishing Machine, significantly increased the measured skid resistance DFT() on all natural
aggregates. Furthermore, it raised the skid resistance DFT() in most cases, to significantly higher levels
than the initially recorded value of ESR from wet polishing.

For the Moutohora and Holcim aggregates, the skid resistance DFT() value reverted almost back to the
initial measured values prior to any polishing.

For the Otaika aggregate sample, the polishing with

Leighton Buzzard sand increased the DFT() higher than the initially tested value;

Accelerated dry polishing with Leighton Buzzard sand on the melter slag aggregate increased the value of
measured skid resistance, but not to a level higher than that reached at ESR;

The addition of emery powder (hard and fine) significantly decreased the measured CoF on all laboratory
samples from the level prior, which existed was after the Leighton Buzzard sand had coarsened the
microtexture. The Moutohora DFT() level decreased down to approximately the same level as that
obtained at ESR. The Otaika aggregate decreased almost to a third of its measured CoF when roughened
by the Leighton Buzzard sand and to a level significantly lower than the ESR level reached with wet
polishing. The Holcim basalt DFT() level also decreased significantly with the addition of emery powder
to a level significantly lower than that reached at ESR.

The melter slag aggregate also reduced

significantly with the addition of the emery powder and accelerated polishing;

Subsequent cycles of adding Leighton Buzzard sand and emery powder to the Moutohora, Holcim Basalt,
and the Otaika greywacke indicates that an historical memory of ESR level may exist with a natural
aggregate as the measured CoF fluctuates about this value, but when wet polished, has a tendency to revert
to the ESR level;

The results of continued cycles of the addition of contaminants and accelerated polishing on the melter slag
demonstrates an overall trend of gradual decreasing measured CoF;

The initial benefits of the melter slag under on-going cyclical polishing with contaminants, resulted in the
melter slag DFT() being reduced to that similar to the Moutohora greywacke aggregate (0.39). It is
currently unknown whether this would continue to decrease further, although a linear extrapolation of the
Stage 2 polishing results would indicate so;

Polishing with emery powder produced the lowest recorded skid resistance value (DFT() for all of the
laboratory sample aggregates. The lowest recorded DFT() values for each of the aggregates, which
301

Chapter 11
included a second cycle of accelerated polishing with Leighton Buzzard sand and emery powder, are
Moutohora=0.41, Holcim=0.23,Otaika=0.26 and Melter slag=0.39.

11.8.4 Summary Conclusions of the Accelerated Polishing Laboratory Experiments

Summary conclusions of the two stage controlled laboratory experiments using accelerated wet polishing to an ESR
level (Stage 1) and accelerated polishing with the addition of contaminants (Stage 2) are:

The significant variation in measured skid resistance observed in the field is intrinsically related to the
geological properties of the aggregates themselves and the contaminants that end up on surfacings;

The accelerated wet polishing of aggregates (without any contaminant additions) can polish natural
aggregates to an equilibrium skid resistance level (ESR);

The level of polish achieved at ESR does not appear to relate to the Polished Stone Value (PSV) results;

Significant variations (both increases and decreases from ESR) in measured skid resistance can be
simulated in the laboratory by the addition of various contaminants;

Contaminants that are fine in size (less than 10m) and that consist of hard minerals (e.g. emery powder)
polish the aggregate surfaces thereby reducing the measured skid resistance;

The amount of decrease in measured skid resistance that is achieved on a specific aggregate depends upon
the difference in hardness and the geological makeup of the contaminants and the aggregate surface being
polished (the greater the difference, the more the decrease in skid resistance);

Contaminants that are coarse (greater than 1mm) and consist of hard minerals (e.g. Leighton Buzzard sand)
abrade, grind and scour the aggregate surface, thereby increasing measured skid resistance;

The amount of increase in measured skid resistance that is achieved on a specific aggregate depends upon
the difference in hardness and the geological makeup of the contaminants and the aggregate surface being
polished (the greater the difference, the more the increase in skid resistance);

Contaminants that are soft in mineral content in comparison to the surfacing aggregate (e.g. oedometer
clay), irrespective of particle size, were shown to increase only slightly the measured skid resistance when
dry. However, when polishing under the action of trafficking or accelerated laboratory polishing becomes
damp (however, not so wet that the contaminants are washed away), the measured skid resistance can
increase on aggregates that have a softer grain matrix and a variable particle size.

In summary, a hypothesis was developed as part of the experimental design stating:

Hypothesis 5: That the type and classification of the contaminants that accumulate on the road
surface are significant in determining the magnitude of the short-term variation of measured
skid resistance.
The laboratory based experiments combined with the geological descriptions, classifications and analysis has clearly
demonstrated this hypothesis to be true.

11.8.5 An Analysis of the Performance of the Polished Stone Value Test

The field site testing and analysis of a sample of State Highway sites in Northland has compared the initial skid
resistance, measured after each of the surfaces were treated and measurements later in its seal life. How the sites
performed in relation to measured skid resistance over repeated traffic loads as a function of the Transit NZ
Published Polished Stone Value (PSV) was determined. The conclusions of the analysis are:

302

Chapter 11
A higher published PSV aggregate does not necessarily lead to a high initial skid resistance value as
measured by the GripTester;

The published PSV of the aggregate does not necessarily determine the level of equilibrium skid resistance
(however, the sites do have varying traffic loading and geometric elements which will affect the final
level). This result tends to corroborate with research by Cenek et al. (2003b) which states that there is very
little relationship between the PSV of the aggregate and the in-field skid resistance as measured by SCRIM
network surveys;

The highest initial skid resistance was recorded where the site was treated during the monitoring period
with a Grade 2 and 4 chip size. The other sites were treated with Grade 3 and 5 chip. This result is not
unexpected, as the larger chips with higher macrotexture are expected to produce higher initial skid
resistance;

For similar traffic loadings and curvature, approximately the same range of skid resistance was lost over the
same period;

Whilst the melter slag from the Glenbrook Steel Mill has the highest published PSV of 58 compared with a
PSV of 52 and 53 for the other sites, its initial skid resistance is the second lowest. However, the
performance of the slag over time in comparison to the other aggregates and sites, even with much higher
loading, is significantly better.

From the above analysis, it is concluded that the published Transit NZ Polished Stone Value (PSV) of the aggregate,
without taking other factors into consideration, cannot be reliably used as a predictor of the initial skid resistance of
the aggregate nor of the level of equilibrium skid resistance after polishing. Other methods, such as the use of the
accelerated polishing machine developed in this research are required for road asset managers to ensure good
decision making. This will enable better predictions of how aggregates will perform over time and under specific
traffic, geometric and braking stresses.

303

Chapter 12

12 FINAL CONCLUSIONS
12.1 Overview
This study has investigated the factors that are known to affect the seasonal variation of skid resistance over time.
An understanding of these factors and how they inter-relate is required for effective road surfacing and safety
management. Policies and standards that move towards effective long term skid resistance management must
recognise that current measurement techniques (as used in this study) have demonstrated an inherent and significant
variability. This amount of variability poses significant difficulties for Road Controlling Authorities to manage as
the level of measured skid resistance currently seems to have an unpredicatable component. When this is combined
with the variability in driver behaviour, vehicle type, tyre tread, depth, condition and hardness, it is evident that the
performance of an individual vehicle at a given site and time is very difficult to predict.
The overall conclusions from this study can be grouped under the following three main areas:

In-field Skid Resistance Variation:

The short-term variation in measured skid resistance can be up to 30% over very short time-periods (days rather than
months) meaning that predictive models are unreliable.

However, the measured variation in the mean of

homogeneous sections of surfacing over a three year data collection period was found to be relatively small (e.g.the
95% confidence interval <0.06GN for the sites in this study). This result was found to be irrespective of whether the
sites were asphalt mix types or chipseals, what the PSV of the surfacing aggregate was, or whether the surfacing
sections were thought to be during the initial polishing phase or after a theoretical equilibrium level of skid
resistance had been established. The in-field results demonstrated that the higher the total number of heavy
commercial vehicles (HCV) per day at a site generally the higher the 95% confidence interval for the site mean, the
standard deviation and the coefficient of variation (CoV), except for the site that was surfaced with the artificial
melter slag. The melter slag performed better and more consistently than sites with similar HCV volumes and also
did not lose as much skid resistance from the beginning of the surface life to the end of study measurement period in
comparison to other natural aggregates. This better performance was also reflected in the controlled laboratory
experiments where the loss of skid resistance from a new surface to a polished surface was much less in the melter
slag samples compared to the natural aggregates.
The research has demonstrated that there is a clear seasonal variation effect in measured skid resistance, i.e. lower
skid resistance results in late summer (April) and higher results occur in late winter (August) to the 95% confidence
level. However this effect is not an obvious nor predictable sinusoidal shape and there are still significant variations
that occur within calendar months of the year. This indicates that there are other short term variable factors that
need to be taken into account.
Rainfall function analysis using the Weighted Rain Function (WRF) over the previous seven day period at best
explained up to 60% of the variation in skid resistance with the GripTester device at the control seasonal site
(Hikurangi site) for the 2004/2005 data, but much less at other sites. When analysing the SCRIM results with the
Dry Spell Factor (DSF) up to 53% of the variation could be explained.
Other significant factors in the short-term variation of skid resistance are; the ambient temperature depending upon
the testing device, the amount of detritus on the road surface, its associated particle size and mineral hardness and
the petrology of the surfacing aggregate.
304

Chapter 12

Skid Testing Devices:

The various skid resistance measurement devices (e.g. SCRIM, GripTester, ROAR, Dynamic Friction Tester) are
reasonably repeatable devices, however transformation equations between devices (e.g. the IFI) do not currently
provide good correlations for New Zealand highway surface conditions.
The GripTester is a more responsive device than the SCRIM device and better correlates with the Dynamic Friction
Tester and the British Pendulum Tester, however as it is significantly lighter than the SCRIM device it is more
sensitive to road roughness.

The coefficient of variation (CoV) of the GripTester during winter months is

significantly less in the winter months (June, July and August - CoV = approximately 3.0%) than in the summer
months (December, January and February CoV = 5.0%). This indicates that an environmental effect exists related
to the contaminants / cleanness of the surface that affects the run to run variability of the device.
The Dynamic Friction Tester is a very repeatible device (CoV = 1.6% on asphalt mix surfaces) and has proven to be
reliable for static skid resistance measurement tests and is ideal to use both in controlled laboratory experiments and
in the field (especially for calibration / correlation of CFME devices). However, due to the repeat testing action of
the device itself, initial skid resistance reductions due to contaminants / short term environmental effects are
averaged out whereas in-line continuous friction testers such as the GripTester (and normal road users) will reflect
a result that is due to an initially contaminated surface with one pass of the device or vehicle. Furthermore, due to
the circular line of measurement of the Dynamic Friction Tester, any in-field longitudinal vehicle wheel path
differences are also averaged out thereby producing a different result to in-line CFME devices.

Laboratory Tests:
This research confers with Cenek et al (2003b) that the Polished Stone Value test method does not reliably predict
the in-service long-term level of skid resistance.
Research in this study has however shown, by the development of a new accelerated polishing machine and test
method, that significant variations in the measured skid resistance of chip seal samples can be simulated in the
laboratory. The addition of hard but fine grained contaminants (e.g. emery powder) onto a surface that undergoes
accelerated polishing, will significantly reduce the measured skid resistance as it will polish the surface. The
addition of coarse but hard contaminants (e.g. Leighton Buzzard sand) will scratch and abrade the surface under
accelerated trafficking and will significantly improve the measured skid resistance of the surface.
In summary, a geological interpretation of the aggregate properties and the types of contaminants that end up on the
surface is required before an assessment can be made regarding the performance of the aggregate with regard to
trafficking. Additionally, the seasonal and short term variation of skid resistance is currently not able to be
adequately predicted, as it is unknown how the multiple factors that affect skid resistance are interrelated.

12.2 Relevance of the Research

The research reported here has highlighted that there are many independent variables that determine the skid
resistance of a section of road pavement at a particular time. Not all of these have been uncovered, at this point.
Skid resistance is not a stable quantity and this research has shown that it can change markedly in just a few hours,
the change depending not only on the surfacing aggregate petrology, but also on the degree of wetness, (heavy)
traffic loading, and the properties of the contaminant/detritus.

305

Chapter 12
The relevance of the research to industry can be categorised into three areas based upon the organisations varying
roles and perspectives in terms of skid resistance policy, management and operation.

These three areas are

identified below:

Road Controlling Authorities

Road Controlling Authorities, such as Transit NZ, have a desire to construct, manage, operate and maintain a road
network such that it is safe, efficient, economic and sustainable in the long term. To achieve this requires the
development of policies, standards, and performance indicators that are not only measurable (reliably and
repeatably), but achievable (affordably and practically).
This research has shown that two of the current cornerstone factors influencing consideration of the topic, namely
Polished Stone Value and Seasonal Correction, that form the basis behind current policy standards (Transit NZ
T/10, 2002 and other international standards) appear to have little relevance in New Zealand due to the inherent
variability of skid resistance. This is especially so when a great proportion of a Road Controlling Authorities
network is measured at best only once a year. The skid resistance surveys can and should be used as a network level
tool to determine whether the overall trend performance of a network is deteriorating or increasing. However, it
must be recognised that at a project level, further investigations are required to determine whether intervention
treatments are required.

Depending upon both the geological properties of the aggregates at a site and the

environmental conditions, the measured level could vary significantly from the true mean. It is recommended that
continuance of their use, at least at the current level of emphasis and policy, be reviewed.
Recognition of these factors will need to be strengthened in the operational and management arena. It may well be
that skid resistance can be targeted only as a band or range of values at any location, because of its inherent
variability.

Road Asset Managers:

As with any Pavement Management System (e.g. HDM 4, NZ dTIMS), condition data (as well as inventory data)
forms the core of all analyses performed. Data collection is also one of the most expensive items in an overall
pavement management system. All of the components of the system interact with the data provided by the user and
predictive modelling is used to predict what the future condition will be. Therefore the success of the long-term
maintenance planning (in the case of skid resistance, surface treatments) is a direct function of the
comprehensiveness and the accuracy of the data.
Data forms the core of all pavement management systems, and largely determines whether the output from the
system is good or bad . Skid resistance measurement devices such as the GripTester, SCRIM, ROAR, and the
Dynamic Friction Tester measure skid resistance objectively and reasonably economically using high speed data
collection systems and this is much better than subjective testing methods. However, it is equally important to know
what the natural variation is over time, to allow a proper understanding to enable appropriate decision making. This
research has shown that the natural and unpredictable variations of skid resistance are much greater than previously
understood. It, therefore, questions the fundamental basis behind skid resistance policy that relies heavily on largely
one network survey per year and does not take into account the factors that effect skid resistance variation.
Furthermore, this research has demonstrated that the PSV test also does not well predict the in-field level of skid
resistance and as such its continued use should be reviewed. The research described herein has demonstrated that it
is not feasible nor appropriate at this point that models be used for the prediction of skid resistance in New Zealand
as the natural variation is currently large and unpredictable.
306

Chapter 12
This study has shown that artificial melter slag has significantly out-performed natural aggregates both in the
laboratory and in the field. The application of artificial melter slag (and potentially other artificial aggregates) has
significant safety and economic benefits where the surface friction demands are relatively high (e.g. Transit New
Zealand Site Category 2 areas) and where the transportation costs do not out-weigh the longer lasting surface
friction benefits.
From a road safety point of view, road managers also need to be able to develop further means of communicating
information to road users: this includes continuing education of drivers, as well as an improved awareness of
environmental conditions leading to sudden change in skid resistance. Improved real time recording of weather and
environmental conditions and the use of variable message signs at key sites on a road network are suggested for
future improvements in road safety.

Crash Investigators:
Current practices for crash investigators includes measuring a coefficient of friction (by various skid testing
methods) including locked wheel tests with tyres of similar tread depth and patterns to those in a crash. This
coefficient of friction (once obtained) is then used to back calculate probable operating speeds of the vehicles
involved in the crash by using equations of momentum. The relevance of this research is that unless the tests were
undertaken almost immediately after the crash occurred and in the same environmental conditions, the measured
coefficient of friction could significantly vary from that involved in the crash. This in turn means that the estimated
speed of the vehicles prior to the crash could also vary significantly. This possible level of innacuracy in the
measurement and estimation of vehicle speeds poses some significant issues for legal proceedings in crash
reconstruction work.

12.3

Further Research

This research has provided considerable insight into some of the factors that affect the seasonal and short-term
variation of measured skid resistance on typical New Zealand sufacings. There is however, still much research to do
and the following gives recommendations for continuing this research work grouped under three categories.

Policy and Risk Management:

Develop risk management procedures and skid resistance policy that incorporates not only the measured skid
resistance of the surface, but other issues such as geometric, topographical, numbers of heavy vehicles, traffic
operating speeds and the consequence of a crash at that specific location.

Quantify the effects of the daily variation of skid resistance with rainfall and contaminants.

Skid Testing Devices:

Quantify the effects of road roughness on measured skid resistance by various devices.

Develop better IFI transformation coefficients for the correlation of various devices for typical New Zealand
surfacings and textures.

Laboratory Tests:

Evaluate the effects of accelerated polishing and the addition of contaminants on a wider range of New
Zealand aggregates.

Evaluate the effects of other contaminants in conjunction with accelerated polishing.

307

Chapter 12
Evaluate the effect of varying the load on the accelerated polishing machine and to correlate these effects
with field results of the same aggregates.

Evaluate the effects of accelerated polishing on different textured laboratory samples by using a range of chip
sizes.

Evaluate the effects of continued accelerated polishing with contaminants, on the melter slag and other
proprietary artificial aggregates.

308

13 REFERENCES
Allbert, B. J., & Walker, J. C. (1965). Tyre to wet road friction at high speeds. Paper presented at the Institution of
Mechanical Engineers.
American Public Health Association (APHA), American Water Works Association (AWWA), & Water
Environment Federation (WEF). (1992). Standard Methods for the Examination of Water and Wastewater
(18th Edition ed.). Washington, D.C.
AS 1141.40-1999. (1999). Methods for sampling and testing aggregates - polished aggregate friction value - vertical
road machine. In Australian Standards: Standards Australia.
AS 1141.41-1999. (1999). Methods for sampling and testing aggregates - polished aggregate friction value horizontal bed machine. In Australian Standards: Standards Australia.
ASTM D3319-00 (2000). Standard Practice for Accelerated Polishing of Aggregates Using the British Wheel. In C.
D04.51 (Ed.), Book of ASTM Standards (Vol. 04.03). West Conshohocken, Pa, USA: ASTM International.
ASTM E274-97. Standard Test Method for Skid Resistance of Paved Surfaces Using a Full-Scale Tire. In Book of
ASTM Standards (Vol. Book of Standards Volume 04.03). West Conshohocken, Pa, USA: ASTM
International.
ASTM E274-97. (1997). Standard Test Method for Skid Resistance of Paved Surfaces Using a Full-Scale Tire. In C.
E17.21 (Ed.), Book of ASTM Standards (Vol. Book of Standards Volume 04.03). West Conshohocken, Pa,
USA: ASTM International.
ASTM E501-94. (2000). Standard Specification for Standard Rib Tire for Pavement Skid-Resistance Tests. In C.
E17.24 (Ed.), Book of ASTM Standards (Vol. 04.03). West Conshohocken, Pa, USA: ASTM International.
ASTM E965. (2001). Standard Test Method for Measuring Pavement Macrotexture Depth Using a Volumetric
Technique. In C. E17.23 (Ed.), Book of ASTM Standards (Vol. 04.03). West Conshohocken, Pa, USA:
ASTM International.
ASTM E1845-01. (2001). Standard Practice for Calculating Pavement Macrotexture Mean Profile Depth. In C.
E17.23 (Ed.), Book of ASTM Standards (Vol. E1845-01). West Conshohocken, Pa, USA: ASTM
International.
ASTM E1911-98. (2002). Standard Test Method for Measuring Paved Surface Frictional Properties Using the
Dynamic Friction Tester. In C. E17.21 (Ed.), Book of ASTM Standards (Vol. 04.03). West Conshohocken,
Pa, USA: ASTM International.
ASTM E1960-98. (1998). Standard Practice for Calculating International Friction Index of a Pavement Surface. In
C. E17.21 (Ed.), Book of ASTM Standards (Vol. 04.03). West Conshohocken, Pa, USA: ASTM
International.
ASTM E2157-01. (2001). Standard Test Method for Measuring Pavement Macrotexture Properties Using the
Circular Track Meter. In C. E17.23 (Ed.), Book of ASTM Standards (Vol. 04.03). West Conshohocken, Pa,
USA: ASTM International.
ASTM E 274-97. (1997). Standard Test Method for Skid Resistance of Paved Surfaces Using a Full-Scale Tire. In
C. E17.21 (Ed.), Book of ASTM Standards (Vol. Book of Standards Volume 04.03). West Conshohocken,
Pa, USA: ASTM International.
ASTM International. (2001). E 1859-97 Friction Coefficient Measurements Between Tire and Pavement Using a
Variable Slip Technique. In C. E17.21 (Ed.), Book of ASTM Standards (Vol. Book of Standards Volume
04.03). West Conshohocken, Pa, USA: ASTM International.
ASTM International. (2003). E303-93 Standard Test Method for Measuring Surface Frictional Properties Using the
British Pendulum Tester. In C. E17.21 (Ed.), Book of ASTM Standards (Vol. Book of Standards Volume
04.03). West Conshohocken, Pa, USA: ASTM International.
Attaway, S. W. (2005). The mystery of gemstone polish. Lapidary Journal, 58, 12p.
Austroads. (2003). AP-G63/03:Guide to the selection of Road Surfacings(2nd ed.).
Austroads. (2005a). DRAFT International Conference On Road Surface Friction (2005) Christchurch, New
Zealand. Report On Skid Resistance Test Device Trials, 5 and 6 May 2005 (No. DRAFT - Report No
unknown). Sydney, Australia: Austroads.
Austroads. (2005b). Guide to the Management of Road Surface Skid Resistance. Sydney, Australia: Austroads
Incorporated.
309

Bean, D., & Pidwerbesky, B. (1994). Does the polished stone value test assure skid resistance? Road and Transport
Research, 3(2), 84-92.
Bells, M. (2006). The History of Pneumatic Devices. Retrieved [Accessed 10 January 2006]
Bennis, T. A., & De Witt, L. B. (2003, 5-7 August 2003). PIARC State-Of-The-Art on Friction and IFI. Paper
presented at the 1st Annual Australian Runway and Roads Friction Testing Workshop, Sydney, Australia.
Bird, G., & Scott, W. J. O. (1936). Measurement of the non-skid properties of road surfaces (No. Ministry of
Transport Road Research Bulletin No.1). London, UK: Department of Scientific and Industrial Research,
Ministry of Transport Road Research.
Black, P. (2005a). Explaining the Polishing Experiments. In D. J. Wilson (Ed.) (pp. Geological notes provided as an
explanation of the Polishing Experiments). Auckland: Professor Philippa Black, Department of Geology,
The University of Auckland.
Black, P. (2005b). Untitled. In D. J. Wilson (Ed.) (pp. Geological notes provided as background to aggregate'
s
geological properties). Auckland: Professor Philippa Black.
Bowmonk Ltd. (2005). Bowmonk AFM2. Retrieved 5 Dec 2005, 2005, from http://www.bowmonk.com/
Breuer, B., Bachmann, T. H., Ernesti, S., & Stcker, J. (1994). Methods and instruments for on-board measurement
of tyre / road-friction. Paper presented at the SAE Motor Sports Engineering Conference, Dearborn, USA.
British Standards Institute. (2002). BS 7976-2: Pendulum testers - Part 2: Method of operation. In British Standards.
London: British Standards Institute.
BS 598-105. (2000). Sampling and examination of bituminous mixtures of roads and other paved areas - part 105:
Methods of test for the determination of texture depth. In British Standards. London, UK: British Standards
Institution (BSI).
BS 812 Part 114. (1989). Methods for determination of the Poilshed Stone Value (PSV). In British Standards.
London, UK: British Standards Institution (BSI).
BS 7976-1. (2002). Pendulum testers - Part 1: Specification. In British Standards. London, UK: British Standards
Institute (BSI).
BS EN 1097-1. (1996). Tests for mechanical and physical properties of aggregates - Part 1: Determination of the
resistance to wear (mirco-Deval). In British Standards. London, UK: British Standards Institute (BSI).
BS EN 1097-8. (2000). Tests for mechanical and physical properties of aggregates - Part 2: Determination of the
polished stone value. In British Standards. London, UK: British Standards Institute (BSI).
Bullas, J. C. (2005, 1-4 May). Slippery when DRY? - Low dry friction and binder-rich road surfaces. Paper
presented at the International Conference, Surface Friction, Roads and Runways - Improving Safety
through Assessment and Design, Christchurch, New Zealand.
Cairney, P. (1997). Skid resistance and crashes - a review of the literature (No. 311). Vermont South, Victoria,
Australia: ARRB Transport Research.
Cairney, P. (1997). Skid resistance and crashes : a review of the literature (No. Research report ARR 311).
Vermont South, Victoria, Australia: ARRB Transport Research.
Cenek, P. D. (1996). Review of Instruments for Measuring the Texture Profile of Road Surfaces (No. Research
Report No 63). Wellington, NZ: Transit New Zealand.
Cenek, P. D. (1998). Skid resistance : the influence of alluvial aggregate size & shape (No. Research Report No.
119). Wellington, N.Z.: Transfund New Zealand.
Cenek, P. D., Alabaster, D. J., & Davies, R. B. (1999). Seasonal and weather normalisation of skid resistance
measurements (No. Research Report 139). Wellington, N.Z.: Transfund New Zealand.
Cenek, P. D., Brown, D. N., Jamieson, N. J., & Stewart, P. F. (1996, 3-4 September). Operational Limitations of
Skid Resistance Testers Used in Australia. Paper presented at the 3rd International Symposium on
Pavement Surface Characteristics, Christchurch, NZ.
Cenek, P. D., Carpenter, P., Jamieson, N. J., & Stewart, P. F. (2003a, Aug 2003). Lesons Learnt from Conducting
Friction Surveys of Runways and Roads. Paper presented at the 1st Annual Australian Runway and Roads
Friction Testing Workshop, Sydney, Australia.
Cenek, P. D., Carpenter, P., Jamieson, N. J., & Stewart, P. F. (2003b). Prediction of Skid Resistance Performance of
Chipseal Roads (No. Research Report No 256). Wellington, New Zealand: Transfund New Zealand.
Cenek, P. D., & Davies, R. B. (2004). Crash Risk Relationships for Improved Safety Management of Roads. Proc
Towards Sustainable Land Transport.
310

Cenek, P. D., & Jamieson, N. J. (2000). Correlation of Skid Resistance Measuring Devices Under Normal State
Highway Survey Conditions (No. Central Laboratories Report 00-529282.00). Lower Hutt, New Zealand:
Opus International Consultants Central Laboratories.
Cenek, P. D., Jamieson, N. J., Locke, N. J., & Stewart, P. F. (1998). Selection of Cost Effective Skid Resistance
Restoration Treatments (No. 141). Wellington, N.Z.: Transfund New Zealand.
Cenek, P. D., Jamieson, N. J., & Towler, J. I. (2000, 9-11 October). The Influence of Texture Depth on Skidding
Resistance. Paper presented at the The New Zealand Land Transport Symposium, Rotorua, NZ.
Cook, D., Donbavand, J., & Kennedy, C. K. (2004, November). A review of Skid Resistance Investigatory Levels for
New Zealand. Paper presented at the Towards Sustainable Land Transport Conference, Wellington.
Dahir, S. H., & Henry, J. J. (1979). Seasonal and Short-Term Variations in Skid Resistance. Transportation
Research Record(715), 69-76.
Dahir, S. H., Henry, J. J., & Meyer, W. E. (1979). Seasonal Skid Resistance Variations (No. Final Report).
Harrisburg, USA: Pennsylvania Department of Transportation.
Dardano, J. (2005, 1-4th May). Results of Physical Workshop. 1st Australia Runway and Roads Friction Testing
Workshop. Paper presented at the International Conference, Surface Friction, Roads and Runways Improving Safety through Assessment and Design, Christchurch, New Zealand.
Dickinson, E. J. (1989). Effect of climate on the seasonal variation of pavement skid resistance. Australian Road
Research Board, 19(2), 129-144.
Donbavand, J. (1989). Skidding Resistance of Road Surfaces-Implications for New Zealand. Road Research Unit
Bulletin 81 ISSN 0549-0030, Road Research Unit.
Donbavand, J. (2004, 6th April). Skid Resistance Principles for Road Networks. Paper presented at the One day
workshop on Skid Resistance Principles for Road Networks, Auckland, NZ.
Donges, E. (1978). A Cybernetic Two-Level Model of the Human Steering Behavior of the Motor Vehicle. Journal
for Traffic Safety (Zeitschrift fuer Verkehrsicherheit), 3, Germany, 98-112.
Donges, E. (1982). Aspects of Active Safety when Steering Passenger Cars. Automobile -Industry (AutomobilIndustrie), 2, Germany, 183-190.
Douglas, R. A., Woodward, W. D. H., & Woodside, A. R. (2000). Road contact stresses and forces under tires with
low inflation pressure. Canadian Journal of Civil Engineering, 27(6), 1248-1258.
Dravitzki, V. K., & Wood, C. W. B. (2001). Temporary effects of CMA on skid resistance. (No. Central
LAboratories Report: 01-527596.00). Lower Hutt, New Zealand: Opus Central Laboratories.
Dravitzki, V. K., Wood, C. W. B., & Potter, S. M. (2003). Road surfaces and loss of skid resistance caused by frost
and thin ice in New Zealand (No. Research Report 244). Wellington, N.Z.: Transfund New Zealand.
Dynatest. (2005). Dynatest 1295 Pavement Friction Tester.
http://www.dynatest.com/hardware/pavementfriction.htm

Retrieved 5 Dec 2005, 2005, from

Eisenberg, D. (2003). The mixed effects of precipitation on traffic crashes. Accident Analysis and Prevention.
Evans, C. J., Paul, E., Dornfeld, D., Lucca, D. A., Byrne, G., Tricard, M., et al. (2003). Material removal
mechanisms in lapping and polishing. STC "G" Keynote CIRP Annals, 52(2), 59pp.
Findlay Irvine Ltd. (2002). The GripTester Operations and Maintenance Manual (No. Ref No 292-462.doc and Ref
No 292-452.doc). Penicuik, Midlothian, Scotland: Findlay Irvine Ltd.
Findlay Irvine Ltd. (2003a). AIRBASE GripTester Survey Software for Airports: User Manual (No. Doc 292-570
Rev 1.0). Penicuik, Midlothian, Scotland: Findlay Irvine Ltd.
Findlay Irvine Ltd. (2003b). Temperature effects on the GripTester device. In D. J. Wilson (Ed.). Email
correspondence.
Findlay Irvine Ltd. (2005). Highways - GripTesterMK2 Skid Resistance Tester. Retrieved 5 Dec 2005, 2005, from
http://www.findlayirvine.com/products/prod_det.php?sectorId=16&productId=106
Forster, S. W. (1989). Pavement microtexture and its relation to skid resistance. Transportation Research Record,
12(15), 151-164.
Fox, J. C., Good, M. C., & Joubert, P. N. (1979). Collisions with Utility Poles. Melbourne, Australia: University of
Melbourne, Report prepared for The Department of Transport, Australia.
Fricke, L. B. (1990). Traffic Accident Reconstruction (Vol. Vol 2). Evanston, Illinois: Northwestern University
Traffic Institute.
311

Giles, C. G. (1957). The Skidding resistance of roads and the requirements of modern traffic. Paper presented at the
Proceedings of the UK Institution of Civil Engineers.
Giles, C. G., & Sabey, B. E. (1959). A Note on the Problem of "Seasonal Variation" in Skid Resistance. Proc., First
International Skid Prevention Conference, Part II.
Giles, C. G., Sabey, B. E., & Cardew, K. H. F. (1962). Development and performance of the portable skid resistance
tester (No. Technical Paper No 66). Crowthorne, UK: Department of Industrial and Scientific Research,
Road Research Laboratory.
Golden, J. M. (1981). A theory of wet road-tyre friction. Wear, Vol.71, pp. 307-331.
Golini, D., & Jacobs, S. D. (1991). Physics of loose abrasive microgrinding. Applied Optics, 30(19), 2761-2777.
Gothie, M. (2002). Could adherence of a road surface be used to identify the areas where maintenance is needed?
Bearing capacity of roads, railways and airfields.
Gothi, M. (2000). The contribution to road safety of pavement surface characteristics [Apport la securite routiere
des caracteristiques de surface des chaussees]. Bulletin des Laboratoires des Ponts et Chaussees(224), 512.
Gray, J. E., & Renninger, F. A. (1963). Limestone and dolomite sands in skid resistant portland cement mortar
surfaces. (No. Research Report). Washington: National Crushed Stone Association.
Gray, J. E., & Renninger, F. A. (1965). Skid resistant properties of carbonate aggregates (No. Research Report).
Washington: National Crushed Stone Association.
HACH. (1997). HACH Water Analysis Handbook - ISO9001 Certified. Loveland, Colorado, U.S.A: HACH
Company,.
Hartley, A. (1974). Mechanical properties of road surface aggregates. Quarterly Journal of Engineering Geology,
Vol 7, pp. 69-100.
Harwood, D. W., Blackburn, R. R., St.John, A. D., & Sharp, M. C. (1977). Evaluation of accident rate - skid number
relationships for a nationwide sample of highway sections. Transportation Research Record, TRR No 623,
pp.142-150.
Harwood, D. W., Mason, J. M., Kulakowski, B. T., & Fitzpatrick, K. (1989). Truck characteristics for use in
highway design and operation (No. Research Report Vol 1 FHWA RD-89-226): US Federal Highway
Administration.
Hatherly, L. W., & Young, A. E. (1977). Paper presented at the 2nd International Skid Resistance Conference,
Columbus.
Hawkes, J. R., & Hosking, J. R. (1972). British Arenaceous Rocks for Skid-Resistant Road Surfacings. Crowthorne,
UK: Transport and Road Research Laboratory.
Haydon, M. (2005). Risk-Based Assessment of Skid Resistance (Project submitted to fulfil requirements for Project
Unit 661). Melbourne, Australia: Centre for Pavement Engineering Education.
Hegmon, R. R. (1978). Seasonal Variations in Pavement Skid Resistance - Are These Real? Public Roads, 42(2),
55-62.
Henning, T., Pradhan, N. K., & Wilson, D. J. (2000). dTIMS Data Issues (No. DT/2000/01). Auckland, New
Zealand: Road Information Management Systems.
Henry, J. J. (1996, 3-4 Sept). Overview of the International PIARC experiment to compare and harmonise texture
and skid resistance measurements: the International Friction Index. Paper presented at the Third
International symposium on Pavement Surface Characteristics, Christchurch, NZ.
Henry, J. J. (2000). Evaluation of Pavement Friction Characteristics (No. NCHRP Synthesis 291): Transportation
Research Board, National Research Council.
Henry, J. J. (2000). Evaluation of pavement friction characteristics (No. Synthesis of Highway Practice No. 291).
Washington, D.C.: National Academy Press, National Cooperative Highway Research Program, National
Research Council (U.S.). Transportation Research Board, American Association of State Highway and
Transportation Officials.
Henry, J. J. (2003, 5-7 August 2003). Measurement of Wet Runway Friction and the International Friction Index.
Paper presented at the 1st Annual Australian Runway and Roads Friction Testing Workshop, Sydney,
Australia.
Henry, J. J., H. Abe, S. Kameyama, A, Tamai, A, Kasahara, K, Saito. (2000). Determination of the international
friction index (IFI) using the circular texture meter (CTM) and the dynamic friction tester (DFT). PIARC,
No. 109.01.06.B2000(IV symposium on surface characteristics,).
312

Henry, J. J., & Saito, K. (1983). Skid-Resistance Measurements with Blank and Ribbed Test Tires and Their
Relationship to Pavement Texture. Transportation Research Record, 946, 38-43.
Highways Agency. (1999). HA 36/99. Surfacing Materials for New and Maintenance Construction. In Design
Manual for Roads and Bridges (Vol. Vol 7, Section 5). London, UK: Highways Agency.
Highways Agency. (2001). Design Manual for Roads and Bridges (Vol. Volume 7 - Pavement Design and
Maintenance). London, UK: Highways Agency.
Highways Agency. (2004). HD 28/04. Skid Resistance. In Design Manual for Roads and Bridges (Vol. Volume 7).
London, UK: Highways Agency.
Hill, B. J., & Henry, J. J. (1981). Short-Term, Weather-Related Skid Resistance Variations. Transportation
Research Record, 836, 76-81.
Homburger, W. S., Kell, J. H., & Perkins, D. D. (2001). Fundamentals of Traffic Engineering (15 ed.): Computer
Center of the University of California at Berkeley.
Hosking, J. R. (1970). Synthetic aggregates of high resistance to polishing: Part 1 - gritty aggregates (No. Report
LR 350). Crowthorne, UK: Road Research Laboratory.
Hosking, J. R. (1973). The effect of Aggregate on the Skidding Resistance of Bituminous Surfacings; Factors other
than Resistance to Polishing. 1973.
Hosking, J. R. (1976). Aggregates for Skid - Resistant Roads. (No. TRRL Rep). Crowthorne, UK: Transport Road
Research Labarotory.
Hosking, J. R. (1986). Relationship between skidding resistance and accident frequency: estimates based on
seasonal variation. (No. Research Report Number 76.). Crowthorne, UK: Transport and Road Research
Laboratory.
Hosking, J. R., & Woodford, G. C. (1976a). Measurement of skidding resistance Part III. Factors affecting SCRIM
measurements (No. TRRL Laboratory Report LR739.). Crowthorne, Berkshire, England: Transport and
Road Research Laboratory.
Hosking, J. R., & Woodford, G. C. (1976b). Measurement of skidding resistance: Part I. Guide to the use of SCRIM.
(No. TRRL Report LR 737). Crowthorne, UK: Department of the Environment, Transport and Road
Research Laboratory.
Hosking, R. (1992). Road Aggregates and Skidding (No. TRL State of the Art Review No. 4). Crowthorne, UK:
Transport Research Laboratory.
Hudson, K. C., & Mumm, P. (2003). Six years of calcined bauxite. Paper presented at the 21st ARRB Transport
Research Conference, Cairns, Australia.
Huschek, S. (1990). The influence of road surface roughness on tire noise generation in the Federal Republic of
Germany. (Vol. STP 1031): ASTM Special Publication.
Ingenium, New Zealand National Asset Management Steering Group, & Institute of Public Works Engineering
Australia. (2002). International Infrastructure Management Manual (Australia/New Zealand ed. ed.).
Wellington, New Zealand: Ingenium.
Jayawickrama, P. W., & Thomas, B. (1998). Correction of field skid measurements for seasonal variations in Texas.
Transportation Research Record(1639), 147-154.
Jellie, J. H. (2003). A study of Factors Affecting Skid Resistance Characteristics. Unpublished Thesis (DPhil),
University of Ulster, Jordanstown, Northern Ireland.
Karmel, N., & Gartshorne, T. (1982). Ontario'
s wet pavement accident reduction program. In Pavement Surface
Characteristics and Materials (STP 763 ed.). West Conshohocken, Pennsylvania: American Society for
Testing Materials.
Kennedy, C. (2005, 1-4 May). Allocating skid resistance investigatory levels on the basis of risk analysis. Paper
presented at the International Conference, Surface Friction, Roads and Runways - Improving Safety
through Assessment and Design, Christchurch, NZ.
Kennedy, C. K., Young, A. E., & Butter, I. C. (1990). Measurement of skidding resistance and surface Texture and
use of results in the United Kingdom: ASTM STP 1031.
Kinnear, D. G., Lainson, L. N., & Penn, H. G. (1984, 27-31 August). An evaluation of four pavement treatments in
Sydney in relation to crash reduction, skid resistance and cost. Paper presented at the 12th Australian Road
Research Board Conference, Victoria, Australia.
Knill, C. D. (1960). Petrographical aspects of the polishing of natural roadstones. Journal of Applied Chemistry,
vol.10, pp. 28-35.
313

Kokkalis, A. G. (1998). Prediction of skid resistance from texture measurements. Proceedings of the Institution of
Civil Engineers: Transport, 129(2), 85-93.
Lamm, R., Psarianos, B., & Mailaender, T. (1999). Highway Design and Traffic Safety Engineering Handbook: McGraw Hill.
Lampropoulos, J. C., Xu, S., & Fang, T. (1997). Loose abrasive lapping hardness of optical glasses and its
interpretation. Applied Optics, 36(7), 1501-1516.
Lampropoulos, J. C., Xu, S., Fang, T., & Golini, D. (1996). Twyman effect mechanics in grinding and
microgrinding. Applied Optics, 35(28), 5704-5713.
Land Transport New Zealand. (2004). A companion document to the Austroads Guide to Traffic Engineering
Practice. Part 4. Treatment of crash locations. Wellington: Land Transport New Zealand.
Lay, M. G. (1998). Handbook of road technology (2 vols). (3rd edition).
Lees, G., & Kennedy, C. K. (1975). Quality, shape and degradation of aggregates. Quarterly Journal of Engineering
Geology, Vol 8, pp. 193-209.
Leu, M. C., & Henry, J. J. (1983). Prection of Skid Resistance as a Function of Speed from Pavement Texture.
Transportation Research Record, Vol 946.
Lippman, S. A. (1985). Effects of tire structure and operating conditions on the distribution of stress between the
tread and the road. In M. G. Pottinger & T. J. Yager (Eds.), The tire pavement interface, ASTM STP 929
(Vol. ASTM STP 929, pp. pp. 91-109). Philadelphia: American Society for Testing and Materials.
Lister, N. W., & Nunn, D. E. (1968). Contact areas of commercial vehicle tyres (No. RRL Report LR 172).
Crowthorne, UK: Road Research Laboratory.
Liu, G. X. (1993). The area and stresses of contact between tyres and road surface and their effects on road surface.
Unpublished Thesis, (DPhil), University of Ulster, Jordanstown, Northern Ireland.
Maclean, D. J., & Shergold, F. A. (1958). The polishing of roadstone in relation to the resistance to skidding of
bituminous road surfacings. (No. Road Research Technical Paper No. 43). Crowthorne, London, UK: Road
Research Laboratory.
McCullough, B. V., & Hankins, K. D. (1966). Skid resistance guidelines for surface improvements on Texas
highways. Highway Research Record, No.131.
McDaniel, R. S., & Coree, B. J. (2003). Identification of Laboratory Techniques to Optimize Superpave HMA
Surface Friction Characteristics. Phase I: Final Report (No. SQDH 2003 - 6, HL 2003-19). Indiana, USA:
North Central Superpave Centre, Purdue University.
Meyer, W. E., Hegmon, R. R., & Gillespie, T. D. (1974). Locked-Wheel Pavement Skid Tester Correlation and
Calibration Techniques (No. Report 151): National Cooperative Highway Research Program.
Millard, R. (1993). Road Building in the Tropics. Crowthorne, London, UK: Transport Research Laboratory.
Miller, M. M., & Johnston, H. D. (1973). Effects of resistance to skidding on: surface dressing on an elevated
section of the M4 motorway. (No. Laboratory Report 542). Crowthorne, London, UK: Transport Road
Research Laboratory.
Ministry of Transport. (1970). Report of the committee of highway maintenance. London, UK: HMSO.
Neville, G. (1974). A Study of the Mechanism of Polishing of Roadstones by Traffic (No. LR 621): Transport and
Road Research Laboratory.
New Zealand Government. (2002). Local Government Bill. In Public Act. Wellington, NZ: New Zealand
Government.
New Zealand Government. (2003). Land Transport Management Act. In Public Act. Wellington, NZ: New Zealand
Government.
Ng, W. H., Buckeridge, J. S., & Ockleston, G. (2003, 14-16 May). Distribution of Heavy Metal Contaminants in
Road Sediments. Paper presented at the 3rd South Pacific Stormwater Conference, Auckland, NZ.
Ng, W. H., J.S, B., & G., O. (2003, 14-16 May). Distribution of Heavy Metal Contaminants in Road Sediments.
Paper presented at the 3rd South Pacific Stormwater Conference, Auckland, New Zealand.
Nicholls, J. C. (1997). Laboratory Tests on High Friction Surfaces for Highways (No. TRL Report 176).
Crowthorne, Berkshire, England: Transport Research Laboratory.
Nippo Sangyo Co. Ltd. (2005a). New DF Tester Help File (No. New DF Tester Help.doc). Tokyo, Japan: Nippo
Sangyo Co. Ltd.
314

Nippo Sangyo Co. Ltd. (2005b). Product Guide - ASTM : E 1911-98.

http://www.nippou.com/en/products/dft.html

Retrieved 5 Dec 2005, 2005, from

O'
Riley, A., Pandey, S., Langdon, A., & Wilkins, A. (2002). Characterisation of Road Runoff Contaminants from
New Zealand Roads, & Effect of Rainfall Events (No. 228): Transfund New Zealand.
Ogden, K. W., & Taylor, S. Y. (1996). Traffic Engineering and Management: Department of Civil Engineering,
Monash Department, Clayton.
Oliver, J. W. H. (1980). Temperature Correction of Skid Resistance Values obtained with the British Portable Skid
Resistance Tester (Internal Report No. ARRB Internal Report AIR314-2). Melbourne, Victoria, Australia:
Australian Road Research Board (ARRB).
Oliver, J. W. H. (2003a, Aug). Factors Affecting the Correlation of Skid Testing Machines and a Proposed
Correlation Framework. Paper presented at the 1st Annual Australian Runway and Roads Friction Testing
Workshop, Sydney, Australia.
Oliver, J. W. H. (2003b). Skid Resistance Presentation. Auckland, New Zealand: REAAA Roadshow.
Oliver, J. W. H., Tredrea, P. F., & Pratt, D. N. (1988). Seasonal Variation of Skid Resistance in Australia (No.
Special Report No 37). Melbourne, Australia: Australian Road Research Board.
Oliver, J. W. H., & Witt, H. P. (1978). Detailed Procedures for Calibration of the British Portable Skid Resistance
Tester (Technical Manuals No. ARRB Technical Manual No.6). Melbourne, Victoria, Australia: Australian
Road Research Board.
Onwumere, G. C. (2000). An evaluation of Highway Stormwater Runoff Quality in the C.V.R.D. Unpublished PhD
Thesis, University of British Columbia, Vancouver, Canada.
Owusu - Ababio, S. (1995). Modelling skid resistance for flexible pavements: a comparison between regression and
neural network models. Transportation Research Record, no.1501, pp.60-71.
Paterson, W. D. O., & Scullion, T. (1990). Information Systems for Road Management: Draft Guidelines on System
Design and Data Issues (No. World Bank Technical Paper INU 77). Washington, D.C.: Infrastructure and
Urban Development Department, The World Bank.
Pavement Management Services. (2005). Pavement Management Services - Testing Services.
2005, 2005, from http://www.pavement.com.au/testing_services.html#

Retrieved 5 Dec

Pazzaglia, R., & Nelson, G. S. (1993). United States of America Patent No. 5,245,856.
Perry, M. J. (1996). A study of the factors influencing the polishing characteristics of gritstone aggregate.
Unpublished Thesis, (DPhil), University of Ulster, Jordanstown, Northern Ireland.
Peterson, R. F., Eckert, C. F., & Carr, C. I. (1974). Tread Compound Effects in Tire Traction. In The Physics of Tire
Traction: Theory and Experiment (pp. p. 223). New York - London: Plenum Press.
PIARC World Road Association. (2003). Road Safety Manual: R2ute market.
Pradhan, N. K., Henning, T., & Wilson, D. J. (2001). Development of Pavement Deterioration Modelling in New
Zealand - A Review of Achievements (No. DT/2000/18). New Zealand: Road Information Management
Steering (RIMS) Group.
Pradhan, N. K., Wilson, D. J., & Dunn, R. C. M. (2003). Training for a pavement management system: experiences
from New Zealand ARRB Road & Transport Research Journal, 12(1), 12.
Prowell, B. D., Xie, H., Cooley, A. L., Powell, R. B., & Hanson, D. (2003, 12-16 January). Relationships between
pavement friction and material properties at NCAT test track. Paper presented at the 82nd Annual meeting
of the Transportation Research Board, Washington, DC.
Rado, Z. (1994). Analysis of Texture Profiles (No. PTI Report 9510). Pennsylvania, USA: Pennsylvania
Transportation Institute, State College.
Rice, J. M. (1977). Seasonal Variations in Pavement Skid Resistance. Public Roads, 40(4), 160-166.
Road Research Laboratory. (1960). Instructions for using the portable skid resistance tester (No. HMSO Road Note
No. 27). Crowthorne, London, UK: Department of Scientific and Industrial Research.
Roe, P., & Hartshorne, S. A. (1998). The Polished Stone Value of Aggregates and In-service Skidding Resistance
(No. 322): Transport Research Laboratory.
Roe, P., Parry, A. R., & Viner, H. E. (1998). High and Low Speed Skidding Resistance: The influence of Texture
Depth (No. TRL Report 367). Crowthorne, London, UK: Transport Research Laboratory.
Roe, P. G., & Hartshorne, S. A. (1998). The polished stone value of aggregates and in-service skidding resistance
(No. TRL report 322). Crowthorne, Berkshire: Transport Research Laboratory.
315

Roe, P. G., & Sinhal, R. (2005a, 1-4 May). How do you compare? Correlation and calibration of skid resistance
and road surface friction measurement devices. Paper presented at the International Conference, Surface
Friction, Roads and Runways - Improving Safety through Assessment and Design, Christchurch, New
Zealand.
Roe, P. G., & Sinhal, R. (2005b, 1-4 May). Recent Developments to the SCRIM Measurement Technique in the UK.
Paper presented at the International Conference, Surface Friction, Roads and Runways - Improving Safety
through Assessment and Design, Christchurch, New Zealand.
Roe, P. G., Webster, D. C., & West, G. (1991). The relation between the surface texture of roads and accidents.
(No. TRL Report 296). Crowthorne, London, UK: Transport Research Laboratory (TRL).
Rogers, M. P., & Gargett, T. (1991). Skidding resistance standard for the National Road Network. Highways and
Transportation, 38(4).
Rubber Manufacturers Association. (2006). Whats inside a tyre. Retrieved Accessed 10 January 2006 -, 2006
Sabey, B. E. (1980). Road Safety and Value for Money. (No. SR 581). Crowthorne, UK: Transport and Road
Research Laboratory.
Salt, G. F. (1977). Research on Skid Resistance at the Transport and Road Research Laboratory (1927-1977).
Transportation Research Record, 622, 26-38.
Salt, G. F., & Szatkowski, W. S. (1973). A Guide to Levels of Skidding Resistance for Roads (No. TRL Report).
Sandberg, U. (1999). Influence of Road Surface Texture on Traffic Characteristics. Nordic Road & Transport
Research, Volume 2 No2.
Sansalone, J. J., & Buchberger, S. G. (1997). Partitioning and First Flush of Metals in Urban Roadway Storm Water.
Journal of Environmental Engineering, 123, pp. 134-143.
Schlosser, L. H. M. (1977). Traffic accidents and road surface skidding resistance. Transportation Research Record,
623, pp.11-20.
Segel, L. (1977). Opening paper (Vol. 621). Columbus, Ohio: Transportation Research Record.
Shupe, J. W., & Lounsbury, R. W. (1959). Polishing characteristics of mineral aggregates. Paper presented at the
First International Skid Prevention Conference.
Siegfried, F. (1998). The study of contact characteristics between tyre and road surface. Unpublished Thesis
(DPhil), University of Ulster, Jordanstown, Northern Ireland.
Sinhal, R. (2005, 1-4 May). The implementation of a skid policy to provide the required friction demand on the main
road network in the United Kingdom. Paper presented at the Internation Conference Surface Friction Roads
and Runways - Improving safety through assessment and design, Christchurch, New Zealand.
Sinkankas, J. (1999). What is Polishing. Lapidary Journal, 52(11), 51-56.
Smith, M. R., & Collis, L. (2001). Aggregates: Sand, gravel and crushed rock aggregates for construction purposes.
London: The Geological Society.
Speyer, A. G., & Kelley, J. D. (1977). Effect of tire construction variables on passenger tire wet traction.
Transportation Research Record, 621, p. 99.
SteelServ Ltd, & Bourke, B. (2005). Iron and Steel Making Aggregates in NZ. In D. J. Wilson (Ed.) (pp. Personal
emails). Auckland.
Stotz, G. (1987). Investigations of the Surface Water Runoff from Federal Highways in the FRG. Science of the
Total Environment, 59, pp. 329-337.
Szatkowski, W. S., & Hosking, J. R. (1972). The Effect of Traffic and Agregate on the Skidding Resistance of
Bituminous Surfacing (No. LR 504). Crowthorne, London, UK: Transport and Road Research Laboratory.
The Wheel and Tyre Bible. (2006). Tyre Constructions.
longhurst.com/carbibles/tyre_bible.html

Retrieved 17 March 2006, from http://www.chris-

Tielking, J. T., & Roberts, F. L. (1987). Tire contact pressure and its effect on pavement strain. ASCE Journal of
Transportation Engineering, 113(1), pp. 56-71.
TPH Working Group. (1998). Characterisation of C6 to C35 Petroleum Hydrocarbons in Environmental Samples,
Revision 5.1. from http://www.aehs.com/publications/catalog/contents/Direct%20Method.PDF
Transit New Zealand. (1981). T/03 - Standard test procedure for measurement of Texture by the Sand Circle
method. In Transit New Zealand Specifications. Wellington, NZ: Transit New Zealand.

316

Transit New Zealand. (2002). T/10 - Specification for Skid Resistance Investigation and Treatment Selection. In
Transit New Zealand Specifications. Wellington, NZ: Transit New Zealand.
Transit New Zealand. (n.d.). T/2 - Specification for calibration of equipment. In Transit New Zealand Specifications.
Wellington, NZ: Transit New Zealand.
Transit New Zealand, & Roading New Zealand. (2005). Chipsealing in New Zealand. Wellington, New Zealand:
Transit New Zealand, Road Controlling Authorities, Roading New Zealand.
Transit NZ, & Roading NZ. (2005). Chipsealing in New Zealand. Wellington, New Zealand.
Transit NZ M/06 - PSV Chart. (2004). Suppliers of Surfacing Aggregate - Polished Stone Values. In Transit New
Zealand Specifications. Wellington, NZ: Transit New Zealand.
Treat, J. R. (1980). A Study of pre-crash factors involved in traffic accidents. Research Review 10(6) and 11(1).
Turnkey Instruments Ltd. (2005). Skidman.
instruments.com/gb/skidman.html
Tyre

Retrieved 5 Dec 2005, 2005, from http://www.turnkey-

Talk.
(2006).
Tyre
Constructions.
Retrieved
10
http://www.orbweb.net/autoshop247/tyre_talk/2_tyre_construction.html

Tyres-Online. (2005). The benefits of silica in tyre design.

online.co.uk/technology/silica.asp

January

2006,

from

Retrieved 5 Dec 2005, from www.tyres-

Veith, A. G. (1983). Tires-roads-rainfall-vehicles: The traction connection. In W. E. Meyer & W. J. D. Walter

(Eds.), Frictional interaction of tire and pavement: a symposium / sponsored by ASTM Committees E-17 on
Traveled Surface Characterstics and F-9 on Tires (Vol. Special Technical Publication 793, pp. pp3-40).
Akron-Fairlawn, Ohio: ASTM.
Viner, H. E., Sinhal, R., & Parry, A. R. (2005, 1-4 May). Linking road traffic accidents with skid resistance - recent
UK developments. Paper presented at the International Conference, Surface Friction, Roads and Runways Improving Safety through Assessment and Design, Christchurch, NZ.
VTI. (1981). Road Icing on different pavement structures: Investigations at Test Field Linkoping 1976 over the
period 1976-1980 (No. VTI Report 216A). Linkoping, Sweden: VTI.
Walker, T. A., Allison, R. A., Wong, T. H. F., & Wootton, R. M. (1999). Removal of Suspended Solids and
Associated Pollutants by a CDS Gross Pollutant Trap (No. Report 99/2): Cooperative Research Centre for
Catchment Hydrology.
Wallman, C. G., & Astrom, H. (2001). Friction measurement methods and the correlation between road friction and
traffic safety. VTI Report 911A. Swedish National Road and Transport Research Institute, Linkoping,
Sweden.
Walsh, I. (2000). Out of the skid pan. Surveyor, pp 12-15.
Wambold, J. C., Antle, C. E., Henry, J. J., & Rado, Z. (1995). International PIARC Experiment to Compare and
Harmonize Texture and Skid Resistance Measurements. Madrid, Spain: PIARC: PIARC Technical
Commitee on Surface Characteristics C.1, AIPCR -01.04.T-1995.
WDM Ltd. (1998). Investigation into the Relationship between Aggregate Polished Stone Value and Wet Skid
Resistance (No. Project PR3-0154). Wellington, NZ: Report prepared for Transit New Zealand.
Wehner, B. (1965). Results of Skid-Resistance Measurements and Traffic Safety. Road and Interstate (Strasse and
Autobahn), Vol 8, pp. 261-268.
Williams, A. R., & Evans, M. S. (1983). Influence of Tread Wear Irregularity on Wet Friction Performance of Tires.
In W. E. Meyer & W. J. D. Walter (Eds.), Frictional interaction of tire and pavement: a symposium /
sponsored by ASTM Committees E-17 on Traveled Surface Characterstics and F-9 on Tires (Vol. ASTM
Special Technical Publication 793, pp. 41-64). Akron-Fairlawn, Ohio: American Society for Testing and
Materials.
Wilson, D. J., & Dunn, R. C. M. (2004, 1-3rd March). The Changing Skid Resistance of Roads. Paper presented at
the Association of South Pacific and Australasian Collission Investigators (ASPACI 2004) Conference,
Auckland, New Zealand.
Wilson, D. J., & Dunn, R. C. M. (2005a, 1-4 May). Polishing aggregates to Equilibrium Skid Resistance. Paper
presented at the International Conference, Surface Friction, Roads and Runways - Improving Safety
through Assessment and Design, Christchurch, New Zealand.
Wilson, D. J., & Dunn, R. C. M. (2005b). Polishing aggregates to Equilibrium Skid Resistance. ARRB Road &
Transport Research Journal, 14(2), 17.
317

Wilson, D. J., Findlay, M., & Dunn, M. (2003, Aug). Variability of Highway Pavement Skid Resistance - A Case
Study in Northland . Paper presented at the 1st Annual Australian Runway and Roads Friction Testing
Workshop, Sydney, Australia.
Wilson, D. J., & Hutchison, D. (2004, 21-24 Nov). Sustainable Road Management Practices and Skid Resistance
Policy. Paper presented at the Towards Sustainable Land Transport, Wellington, New Zealand.
Wilson, D. J., & Kirk, G. (2005, 1-4th May). Lessons to Learn from Skid Resistance Monitoring in the Northland
Region. Paper presented at the International Conference, Surface Friction, Roads and Runways - Improving
Safety through Assessment and Design, Christchurch, New Zealand.
Wilson, D. J., Pradhan, N. K., & Henning, T. F. P. (2002). The Appropriateness of Data in a Pavement Management
System - The New Zealand Experience. ARRB Road & Transport Research Journal, Volume 11(No 2), pp.
14-28.
Woodside, A., & Woodward, D. (2002). Chapter 18 - Wet Skid Resistance. In C. A. O'
Flaherty (Ed.), Highways The Location, Design, Construction and Maintenance of Pavements (4th Edition ed., pp. pp. 479-499).
Oxford, UK: Butterworth Heinemann.
Woodside, A. R. (1981). A study of the characteristics of roadstones with particular reference to polishing and
skidding resistance. Unpublished Thesis (MPhil), University of Ulster, Jordanstown, Northern Ireland.
Woodward, D., Woodside, A., & Jellie, J. (2004, May 26-28). Improved prediction of aggregate skid resistance
using modified PSV tests. Paper presented at the Eighth International Conference on Applications of
Advanced Technologies in Transportaion Engineering, Beijing, China.
Woodward, W. D. H. (1995). Laboratory prediction of surfacing aggregate performance. Unpublished Thesis
(DPhil), University of Ulster, Jordanstown, Northern Ireland.
Woodward, W. D. H., Woodside, A. R., & Jellie, J. H. (2002, 21-22 November). Development of early life skid
resistance for high stone content asphalt mixes. Paper presented at the 3rd International conference
bituminous mixtures and pavements, Thessaloniki, Greece.
Woodward, W. D. H., Woodside, A. R., & Jellie, J. H. (2005, 1-4 May). Early and mid life SMA skid resistance.
Paper presented at the International Conference, Surface Friction, Roads and Runways - Improving Safety
through Assessment and Design, Christchurch, New Zealand.
Wright, P. H., & Dixon, K. (2004). Highway Engineering (Seventh Edition ed.). USA: John Wiley and Sons, Inc.
Yim, W. W. S., & Nau, P. S. (1987). Distribution of Lead, Zinc, Copper and Cadmium in Dust from Selected Urban
Areas of Hong Kong. Hong Kong Engineer, 15(1), 7-14.
Zipkes, E. (1944). Die Reibungskennziffer als Kriterium Beurteilung von Strassenbelagen Eidgenosissche
Technische Hockschule Zurich. Zurich: Verlag A.G. Gebr. Leeman and Co.

318